StackGenVis: Alignment of Data, Algorithms, and Models for Stacking Ensemble Learning Using Performance Metrics
https://doi.org/10.1109/TVCG.2020.3030352
You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
31825 lines
1.0 MiB
31825 lines
1.0 MiB
(function (global, factory) { |
|
typeof exports === 'object' && typeof module !== 'undefined' ? module.exports = factory() : |
|
typeof define === 'function' && define.amd ? define(factory) : |
|
(global = global || self, global.prettier = factory()); |
|
}(this, (function () { 'use strict'; |
|
|
|
var name = "prettier"; |
|
var version = "1.19.1"; |
|
var description = "Prettier is an opinionated code formatter"; |
|
var bin = { |
|
prettier: "./bin/prettier.js" |
|
}; |
|
var repository = "prettier/prettier"; |
|
var homepage = "https://prettier.io"; |
|
var author = "James Long"; |
|
var license = "MIT"; |
|
var main = "./index.js"; |
|
var engines = { |
|
node: ">=8" |
|
}; |
|
var dependencies = { |
|
"@angular/compiler": "8.2.13", |
|
"@babel/code-frame": "7.5.5", |
|
"@babel/parser": "7.7.3", |
|
"@glimmer/syntax": "0.41.0", |
|
"@iarna/toml": "2.2.3", |
|
"@typescript-eslint/typescript-estree": "2.6.1", |
|
"angular-estree-parser": "1.1.5", |
|
"angular-html-parser": "1.3.0", |
|
camelcase: "5.3.1", |
|
chalk: "2.4.2", |
|
"cjk-regex": "2.0.0", |
|
cosmiconfig: "5.2.1", |
|
dashify: "2.0.0", |
|
dedent: "0.7.0", |
|
diff: "4.0.1", |
|
editorconfig: "0.15.3", |
|
"editorconfig-to-prettier": "0.1.1", |
|
"escape-string-regexp": "1.0.5", |
|
esutils: "2.0.3", |
|
"find-parent-dir": "0.3.0", |
|
"find-project-root": "1.1.1", |
|
"flow-parser": "0.111.3", |
|
"get-stream": "4.1.0", |
|
globby: "6.1.0", |
|
graphql: "14.5.8", |
|
"html-element-attributes": "2.2.0", |
|
"html-styles": "1.0.0", |
|
"html-tag-names": "1.1.4", |
|
ignore: "4.0.6", |
|
"is-ci": "2.0.0", |
|
"jest-docblock": "24.9.0", |
|
"json-stable-stringify": "1.0.1", |
|
leven: "3.1.0", |
|
"lines-and-columns": "1.1.6", |
|
"linguist-languages": "7.6.0", |
|
"lodash.uniqby": "4.7.0", |
|
mem: "5.1.1", |
|
minimatch: "3.0.4", |
|
minimist: "1.2.0", |
|
"n-readlines": "1.0.0", |
|
"normalize-path": "3.0.0", |
|
"parse-srcset": "ikatyang/parse-srcset#54eb9c1cb21db5c62b4d0e275d7249516df6f0ee", |
|
"postcss-less": "2.0.0", |
|
"postcss-media-query-parser": "0.2.3", |
|
"postcss-scss": "2.0.0", |
|
"postcss-selector-parser": "2.2.3", |
|
"postcss-values-parser": "1.5.0", |
|
"regexp-util": "1.2.2", |
|
"remark-math": "1.0.6", |
|
"remark-parse": "5.0.0", |
|
resolve: "1.12.0", |
|
semver: "6.3.0", |
|
"string-width": "4.1.0", |
|
typescript: "3.7.2", |
|
"unicode-regex": "3.0.0", |
|
unified: "8.4.1", |
|
vnopts: "1.0.2", |
|
"yaml-unist-parser": "1.1.1" |
|
}; |
|
var devDependencies = { |
|
"@babel/core": "7.7.2", |
|
"@babel/preset-env": "7.7.1", |
|
"@rollup/plugin-alias": "2.2.0", |
|
"@rollup/plugin-replace": "2.2.1", |
|
"babel-loader": "8.0.6", |
|
benchmark: "2.1.4", |
|
"builtin-modules": "3.1.0", |
|
codecov: "3.6.1", |
|
"cross-env": "6.0.3", |
|
eslint: "6.6.0", |
|
"eslint-config-prettier": "6.5.0", |
|
"eslint-formatter-friendly": "7.0.0", |
|
"eslint-plugin-import": "2.18.2", |
|
"eslint-plugin-prettier": "3.1.1", |
|
"eslint-plugin-react": "7.16.0", |
|
execa: "3.2.0", |
|
jest: "23.3.0", |
|
"jest-junit": "9.0.0", |
|
"jest-snapshot-serializer-ansi": "1.0.0", |
|
"jest-snapshot-serializer-raw": "1.1.0", |
|
"jest-watch-typeahead": "0.4.0", |
|
mkdirp: "0.5.1", |
|
prettier: "1.19.0", |
|
prettylint: "1.0.0", |
|
rimraf: "3.0.0", |
|
rollup: "1.26.3", |
|
"rollup-plugin-babel": "4.3.3", |
|
"rollup-plugin-commonjs": "10.1.0", |
|
"rollup-plugin-json": "4.0.0", |
|
"rollup-plugin-node-globals": "1.4.0", |
|
"rollup-plugin-node-resolve": "5.2.0", |
|
"rollup-plugin-terser": "5.1.2", |
|
shelljs: "0.8.3", |
|
"snapshot-diff": "0.4.0", |
|
"strip-ansi": "5.2.0", |
|
"synchronous-promise": "2.0.10", |
|
tempy: "0.2.1", |
|
"terser-webpack-plugin": "2.2.1", |
|
webpack: "4.41.2" |
|
}; |
|
var scripts = { |
|
prepublishOnly: "echo \"Error: must publish from dist/\" && exit 1", |
|
"prepare-release": "yarn && yarn build && yarn test:dist", |
|
test: "jest", |
|
"test:dist": "node ./scripts/test-dist.js", |
|
"test-integration": "jest tests_integration", |
|
"perf-repeat": "yarn && yarn build && cross-env NODE_ENV=production node ./dist/bin-prettier.js --debug-repeat ${PERF_REPEAT:-1000} --loglevel debug ${PERF_FILE:-./index.js} > /dev/null", |
|
"perf-repeat-inspect": "yarn && yarn build && cross-env NODE_ENV=production node --inspect-brk ./dist/bin-prettier.js --debug-repeat ${PERF_REPEAT:-1000} --loglevel debug ${PERF_FILE:-./index.js} > /dev/null", |
|
"perf-benchmark": "yarn && yarn build && cross-env NODE_ENV=production node ./dist/bin-prettier.js --debug-benchmark --loglevel debug ${PERF_FILE:-./index.js} > /dev/null", |
|
"check-types": "tsc", |
|
lint: "cross-env EFF_NO_LINK_RULES=true eslint . --format friendly", |
|
"lint-docs": "prettylint {.,docs,website,website/blog}/*.md", |
|
"lint-dist": "eslint --no-eslintrc --no-ignore --env=browser \"dist/!(bin-prettier|index|third-party).js\"", |
|
build: "node --max-old-space-size=3072 ./scripts/build/build.js", |
|
"build-docs": "node ./scripts/build-docs.js", |
|
"check-deps": "node ./scripts/check-deps.js", |
|
spellcheck: "npx -p cspell@4.0.31 cspell {bin,scripts,src}/**/*.js {docs,website/blog,changelog_unreleased}/**/*.md" |
|
}; |
|
var _package = { |
|
name: name, |
|
version: version, |
|
description: description, |
|
bin: bin, |
|
repository: repository, |
|
homepage: homepage, |
|
author: author, |
|
license: license, |
|
main: main, |
|
engines: engines, |
|
dependencies: dependencies, |
|
devDependencies: devDependencies, |
|
scripts: scripts |
|
}; |
|
|
|
var _package$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name, |
|
version: version, |
|
description: description, |
|
bin: bin, |
|
repository: repository, |
|
homepage: homepage, |
|
author: author, |
|
license: license, |
|
main: main, |
|
engines: engines, |
|
dependencies: dependencies, |
|
devDependencies: devDependencies, |
|
scripts: scripts, |
|
'default': _package |
|
}); |
|
|
|
function _typeof(obj) { |
|
if (typeof Symbol === "function" && typeof Symbol.iterator === "symbol") { |
|
_typeof = function (obj) { |
|
return typeof obj; |
|
}; |
|
} else { |
|
_typeof = function (obj) { |
|
return obj && typeof Symbol === "function" && obj.constructor === Symbol && obj !== Symbol.prototype ? "symbol" : typeof obj; |
|
}; |
|
} |
|
|
|
return _typeof(obj); |
|
} |
|
|
|
function _classCallCheck(instance, Constructor) { |
|
if (!(instance instanceof Constructor)) { |
|
throw new TypeError("Cannot call a class as a function"); |
|
} |
|
} |
|
|
|
function _defineProperties(target, props) { |
|
for (var i = 0; i < props.length; i++) { |
|
var descriptor = props[i]; |
|
descriptor.enumerable = descriptor.enumerable || false; |
|
descriptor.configurable = true; |
|
if ("value" in descriptor) descriptor.writable = true; |
|
Object.defineProperty(target, descriptor.key, descriptor); |
|
} |
|
} |
|
|
|
function _createClass(Constructor, protoProps, staticProps) { |
|
if (protoProps) _defineProperties(Constructor.prototype, protoProps); |
|
if (staticProps) _defineProperties(Constructor, staticProps); |
|
return Constructor; |
|
} |
|
|
|
function _defineProperty(obj, key, value) { |
|
if (key in obj) { |
|
Object.defineProperty(obj, key, { |
|
value: value, |
|
enumerable: true, |
|
configurable: true, |
|
writable: true |
|
}); |
|
} else { |
|
obj[key] = value; |
|
} |
|
|
|
return obj; |
|
} |
|
|
|
function _inherits(subClass, superClass) { |
|
if (typeof superClass !== "function" && superClass !== null) { |
|
throw new TypeError("Super expression must either be null or a function"); |
|
} |
|
|
|
subClass.prototype = Object.create(superClass && superClass.prototype, { |
|
constructor: { |
|
value: subClass, |
|
writable: true, |
|
configurable: true |
|
} |
|
}); |
|
if (superClass) _setPrototypeOf(subClass, superClass); |
|
} |
|
|
|
function _getPrototypeOf(o) { |
|
_getPrototypeOf = Object.setPrototypeOf ? Object.getPrototypeOf : function _getPrototypeOf(o) { |
|
return o.__proto__ || Object.getPrototypeOf(o); |
|
}; |
|
return _getPrototypeOf(o); |
|
} |
|
|
|
function _setPrototypeOf(o, p) { |
|
_setPrototypeOf = Object.setPrototypeOf || function _setPrototypeOf(o, p) { |
|
o.__proto__ = p; |
|
return o; |
|
}; |
|
|
|
return _setPrototypeOf(o, p); |
|
} |
|
|
|
function isNativeReflectConstruct() { |
|
if (typeof Reflect === "undefined" || !Reflect.construct) return false; |
|
if (Reflect.construct.sham) return false; |
|
if (typeof Proxy === "function") return true; |
|
|
|
try { |
|
Date.prototype.toString.call(Reflect.construct(Date, [], function () {})); |
|
return true; |
|
} catch (e) { |
|
return false; |
|
} |
|
} |
|
|
|
function _construct(Parent, args, Class) { |
|
if (isNativeReflectConstruct()) { |
|
_construct = Reflect.construct; |
|
} else { |
|
_construct = function _construct(Parent, args, Class) { |
|
var a = [null]; |
|
a.push.apply(a, args); |
|
var Constructor = Function.bind.apply(Parent, a); |
|
var instance = new Constructor(); |
|
if (Class) _setPrototypeOf(instance, Class.prototype); |
|
return instance; |
|
}; |
|
} |
|
|
|
return _construct.apply(null, arguments); |
|
} |
|
|
|
function _isNativeFunction(fn) { |
|
return Function.toString.call(fn).indexOf("[native code]") !== -1; |
|
} |
|
|
|
function _wrapNativeSuper(Class) { |
|
var _cache = typeof Map === "function" ? new Map() : undefined; |
|
|
|
_wrapNativeSuper = function _wrapNativeSuper(Class) { |
|
if (Class === null || !_isNativeFunction(Class)) return Class; |
|
|
|
if (typeof Class !== "function") { |
|
throw new TypeError("Super expression must either be null or a function"); |
|
} |
|
|
|
if (typeof _cache !== "undefined") { |
|
if (_cache.has(Class)) return _cache.get(Class); |
|
|
|
_cache.set(Class, Wrapper); |
|
} |
|
|
|
function Wrapper() { |
|
return _construct(Class, arguments, _getPrototypeOf(this).constructor); |
|
} |
|
|
|
Wrapper.prototype = Object.create(Class.prototype, { |
|
constructor: { |
|
value: Wrapper, |
|
enumerable: false, |
|
writable: true, |
|
configurable: true |
|
} |
|
}); |
|
return _setPrototypeOf(Wrapper, Class); |
|
}; |
|
|
|
return _wrapNativeSuper(Class); |
|
} |
|
|
|
function _assertThisInitialized(self) { |
|
if (self === void 0) { |
|
throw new ReferenceError("this hasn't been initialised - super() hasn't been called"); |
|
} |
|
|
|
return self; |
|
} |
|
|
|
function _possibleConstructorReturn(self, call) { |
|
if (call && (typeof call === "object" || typeof call === "function")) { |
|
return call; |
|
} |
|
|
|
return _assertThisInitialized(self); |
|
} |
|
|
|
function _superPropBase(object, property) { |
|
while (!Object.prototype.hasOwnProperty.call(object, property)) { |
|
object = _getPrototypeOf(object); |
|
if (object === null) break; |
|
} |
|
|
|
return object; |
|
} |
|
|
|
function _get(target, property, receiver) { |
|
if (typeof Reflect !== "undefined" && Reflect.get) { |
|
_get = Reflect.get; |
|
} else { |
|
_get = function _get(target, property, receiver) { |
|
var base = _superPropBase(target, property); |
|
|
|
if (!base) return; |
|
var desc = Object.getOwnPropertyDescriptor(base, property); |
|
|
|
if (desc.get) { |
|
return desc.get.call(receiver); |
|
} |
|
|
|
return desc.value; |
|
}; |
|
} |
|
|
|
return _get(target, property, receiver || target); |
|
} |
|
|
|
function _taggedTemplateLiteral(strings, raw) { |
|
if (!raw) { |
|
raw = strings.slice(0); |
|
} |
|
|
|
return Object.freeze(Object.defineProperties(strings, { |
|
raw: { |
|
value: Object.freeze(raw) |
|
} |
|
})); |
|
} |
|
|
|
function _slicedToArray(arr, i) { |
|
return _arrayWithHoles(arr) || _iterableToArrayLimit(arr, i) || _nonIterableRest(); |
|
} |
|
|
|
function _toConsumableArray(arr) { |
|
return _arrayWithoutHoles(arr) || _iterableToArray(arr) || _nonIterableSpread(); |
|
} |
|
|
|
function _arrayWithoutHoles(arr) { |
|
if (Array.isArray(arr)) { |
|
for (var i = 0, arr2 = new Array(arr.length); i < arr.length; i++) arr2[i] = arr[i]; |
|
|
|
return arr2; |
|
} |
|
} |
|
|
|
function _arrayWithHoles(arr) { |
|
if (Array.isArray(arr)) return arr; |
|
} |
|
|
|
function _iterableToArray(iter) { |
|
if (Symbol.iterator in Object(iter) || Object.prototype.toString.call(iter) === "[object Arguments]") return Array.from(iter); |
|
} |
|
|
|
function _iterableToArrayLimit(arr, i) { |
|
if (!(Symbol.iterator in Object(arr) || Object.prototype.toString.call(arr) === "[object Arguments]")) { |
|
return; |
|
} |
|
|
|
var _arr = []; |
|
var _n = true; |
|
var _d = false; |
|
var _e = undefined; |
|
|
|
try { |
|
for (var _i = arr[Symbol.iterator](), _s; !(_n = (_s = _i.next()).done); _n = true) { |
|
_arr.push(_s.value); |
|
|
|
if (i && _arr.length === i) break; |
|
} |
|
} catch (err) { |
|
_d = true; |
|
_e = err; |
|
} finally { |
|
try { |
|
if (!_n && _i["return"] != null) _i["return"](); |
|
} finally { |
|
if (_d) throw _e; |
|
} |
|
} |
|
|
|
return _arr; |
|
} |
|
|
|
function _nonIterableSpread() { |
|
throw new TypeError("Invalid attempt to spread non-iterable instance"); |
|
} |
|
|
|
function _nonIterableRest() { |
|
throw new TypeError("Invalid attempt to destructure non-iterable instance"); |
|
} |
|
|
|
function Diff() {} |
|
|
|
Diff.prototype = { |
|
diff: function diff(oldString, newString) { |
|
var options = arguments.length > 2 && arguments[2] !== undefined ? arguments[2] : {}; |
|
var callback = options.callback; |
|
|
|
if (typeof options === 'function') { |
|
callback = options; |
|
options = {}; |
|
} |
|
|
|
this.options = options; |
|
var self = this; |
|
|
|
function done(value) { |
|
if (callback) { |
|
setTimeout(function () { |
|
callback(undefined, value); |
|
}, 0); |
|
return true; |
|
} else { |
|
return value; |
|
} |
|
} // Allow subclasses to massage the input prior to running |
|
|
|
|
|
oldString = this.castInput(oldString); |
|
newString = this.castInput(newString); |
|
oldString = this.removeEmpty(this.tokenize(oldString)); |
|
newString = this.removeEmpty(this.tokenize(newString)); |
|
var newLen = newString.length, |
|
oldLen = oldString.length; |
|
var editLength = 1; |
|
var maxEditLength = newLen + oldLen; |
|
var bestPath = [{ |
|
newPos: -1, |
|
components: [] |
|
}]; // Seed editLength = 0, i.e. the content starts with the same values |
|
|
|
var oldPos = this.extractCommon(bestPath[0], newString, oldString, 0); |
|
|
|
if (bestPath[0].newPos + 1 >= newLen && oldPos + 1 >= oldLen) { |
|
// Identity per the equality and tokenizer |
|
return done([{ |
|
value: this.join(newString), |
|
count: newString.length |
|
}]); |
|
} // Main worker method. checks all permutations of a given edit length for acceptance. |
|
|
|
|
|
function execEditLength() { |
|
for (var diagonalPath = -1 * editLength; diagonalPath <= editLength; diagonalPath += 2) { |
|
var basePath = void 0; |
|
|
|
var addPath = bestPath[diagonalPath - 1], |
|
removePath = bestPath[diagonalPath + 1], |
|
_oldPos = (removePath ? removePath.newPos : 0) - diagonalPath; |
|
|
|
if (addPath) { |
|
// No one else is going to attempt to use this value, clear it |
|
bestPath[diagonalPath - 1] = undefined; |
|
} |
|
|
|
var canAdd = addPath && addPath.newPos + 1 < newLen, |
|
canRemove = removePath && 0 <= _oldPos && _oldPos < oldLen; |
|
|
|
if (!canAdd && !canRemove) { |
|
// If this path is a terminal then prune |
|
bestPath[diagonalPath] = undefined; |
|
continue; |
|
} // Select the diagonal that we want to branch from. We select the prior |
|
// path whose position in the new string is the farthest from the origin |
|
// and does not pass the bounds of the diff graph |
|
|
|
|
|
if (!canAdd || canRemove && addPath.newPos < removePath.newPos) { |
|
basePath = clonePath(removePath); |
|
self.pushComponent(basePath.components, undefined, true); |
|
} else { |
|
basePath = addPath; // No need to clone, we've pulled it from the list |
|
|
|
basePath.newPos++; |
|
self.pushComponent(basePath.components, true, undefined); |
|
} |
|
|
|
_oldPos = self.extractCommon(basePath, newString, oldString, diagonalPath); // If we have hit the end of both strings, then we are done |
|
|
|
if (basePath.newPos + 1 >= newLen && _oldPos + 1 >= oldLen) { |
|
return done(buildValues(self, basePath.components, newString, oldString, self.useLongestToken)); |
|
} else { |
|
// Otherwise track this path as a potential candidate and continue. |
|
bestPath[diagonalPath] = basePath; |
|
} |
|
} |
|
|
|
editLength++; |
|
} // Performs the length of edit iteration. Is a bit fugly as this has to support the |
|
// sync and async mode which is never fun. Loops over execEditLength until a value |
|
// is produced. |
|
|
|
|
|
if (callback) { |
|
(function exec() { |
|
setTimeout(function () { |
|
// This should not happen, but we want to be safe. |
|
|
|
/* istanbul ignore next */ |
|
if (editLength > maxEditLength) { |
|
return callback(); |
|
} |
|
|
|
if (!execEditLength()) { |
|
exec(); |
|
} |
|
}, 0); |
|
})(); |
|
} else { |
|
while (editLength <= maxEditLength) { |
|
var ret = execEditLength(); |
|
|
|
if (ret) { |
|
return ret; |
|
} |
|
} |
|
} |
|
}, |
|
pushComponent: function pushComponent(components, added, removed) { |
|
var last = components[components.length - 1]; |
|
|
|
if (last && last.added === added && last.removed === removed) { |
|
// We need to clone here as the component clone operation is just |
|
// as shallow array clone |
|
components[components.length - 1] = { |
|
count: last.count + 1, |
|
added: added, |
|
removed: removed |
|
}; |
|
} else { |
|
components.push({ |
|
count: 1, |
|
added: added, |
|
removed: removed |
|
}); |
|
} |
|
}, |
|
extractCommon: function extractCommon(basePath, newString, oldString, diagonalPath) { |
|
var newLen = newString.length, |
|
oldLen = oldString.length, |
|
newPos = basePath.newPos, |
|
oldPos = newPos - diagonalPath, |
|
commonCount = 0; |
|
|
|
while (newPos + 1 < newLen && oldPos + 1 < oldLen && this.equals(newString[newPos + 1], oldString[oldPos + 1])) { |
|
newPos++; |
|
oldPos++; |
|
commonCount++; |
|
} |
|
|
|
if (commonCount) { |
|
basePath.components.push({ |
|
count: commonCount |
|
}); |
|
} |
|
|
|
basePath.newPos = newPos; |
|
return oldPos; |
|
}, |
|
equals: function equals(left, right) { |
|
if (this.options.comparator) { |
|
return this.options.comparator(left, right); |
|
} else { |
|
return left === right || this.options.ignoreCase && left.toLowerCase() === right.toLowerCase(); |
|
} |
|
}, |
|
removeEmpty: function removeEmpty(array) { |
|
var ret = []; |
|
|
|
for (var i = 0; i < array.length; i++) { |
|
if (array[i]) { |
|
ret.push(array[i]); |
|
} |
|
} |
|
|
|
return ret; |
|
}, |
|
castInput: function castInput(value) { |
|
return value; |
|
}, |
|
tokenize: function tokenize(value) { |
|
return value.split(''); |
|
}, |
|
join: function join(chars) { |
|
return chars.join(''); |
|
} |
|
}; |
|
|
|
function buildValues(diff, components, newString, oldString, useLongestToken) { |
|
var componentPos = 0, |
|
componentLen = components.length, |
|
newPos = 0, |
|
oldPos = 0; |
|
|
|
for (; componentPos < componentLen; componentPos++) { |
|
var component = components[componentPos]; |
|
|
|
if (!component.removed) { |
|
if (!component.added && useLongestToken) { |
|
var value = newString.slice(newPos, newPos + component.count); |
|
value = value.map(function (value, i) { |
|
var oldValue = oldString[oldPos + i]; |
|
return oldValue.length > value.length ? oldValue : value; |
|
}); |
|
component.value = diff.join(value); |
|
} else { |
|
component.value = diff.join(newString.slice(newPos, newPos + component.count)); |
|
} |
|
|
|
newPos += component.count; // Common case |
|
|
|
if (!component.added) { |
|
oldPos += component.count; |
|
} |
|
} else { |
|
component.value = diff.join(oldString.slice(oldPos, oldPos + component.count)); |
|
oldPos += component.count; // Reverse add and remove so removes are output first to match common convention |
|
// The diffing algorithm is tied to add then remove output and this is the simplest |
|
// route to get the desired output with minimal overhead. |
|
|
|
if (componentPos && components[componentPos - 1].added) { |
|
var tmp = components[componentPos - 1]; |
|
components[componentPos - 1] = components[componentPos]; |
|
components[componentPos] = tmp; |
|
} |
|
} |
|
} // Special case handle for when one terminal is ignored (i.e. whitespace). |
|
// For this case we merge the terminal into the prior string and drop the change. |
|
// This is only available for string mode. |
|
|
|
|
|
var lastComponent = components[componentLen - 1]; |
|
|
|
if (componentLen > 1 && typeof lastComponent.value === 'string' && (lastComponent.added || lastComponent.removed) && diff.equals('', lastComponent.value)) { |
|
components[componentLen - 2].value += lastComponent.value; |
|
components.pop(); |
|
} |
|
|
|
return components; |
|
} |
|
|
|
function clonePath(path) { |
|
return { |
|
newPos: path.newPos, |
|
components: path.components.slice(0) |
|
}; |
|
} |
|
|
|
var characterDiff = new Diff(); |
|
|
|
function diffChars(oldStr, newStr, options) { |
|
return characterDiff.diff(oldStr, newStr, options); |
|
} |
|
|
|
function generateOptions(options, defaults) { |
|
if (typeof options === 'function') { |
|
defaults.callback = options; |
|
} else if (options) { |
|
for (var name in options) { |
|
/* istanbul ignore else */ |
|
if (options.hasOwnProperty(name)) { |
|
defaults[name] = options[name]; |
|
} |
|
} |
|
} |
|
|
|
return defaults; |
|
} // |
|
// Ranges and exceptions: |
|
// Latin-1 Supplement, 0080–00FF |
|
// - U+00D7 × Multiplication sign |
|
// - U+00F7 ÷ Division sign |
|
// Latin Extended-A, 0100–017F |
|
// Latin Extended-B, 0180–024F |
|
// IPA Extensions, 0250–02AF |
|
// Spacing Modifier Letters, 02B0–02FF |
|
// - U+02C7 ˇ ˇ Caron |
|
// - U+02D8 ˘ ˘ Breve |
|
// - U+02D9 ˙ ˙ Dot Above |
|
// - U+02DA ˚ ˚ Ring Above |
|
// - U+02DB ˛ ˛ Ogonek |
|
// - U+02DC ˜ ˜ Small Tilde |
|
// - U+02DD ˝ ˝ Double Acute Accent |
|
// Latin Extended Additional, 1E00–1EFF |
|
|
|
|
|
var extendedWordChars = /^[A-Za-z\xC0-\u02C6\u02C8-\u02D7\u02DE-\u02FF\u1E00-\u1EFF]+$/; |
|
var reWhitespace = /\S/; |
|
var wordDiff = new Diff(); |
|
|
|
wordDiff.equals = function (left, right) { |
|
if (this.options.ignoreCase) { |
|
left = left.toLowerCase(); |
|
right = right.toLowerCase(); |
|
} |
|
|
|
return left === right || this.options.ignoreWhitespace && !reWhitespace.test(left) && !reWhitespace.test(right); |
|
}; |
|
|
|
wordDiff.tokenize = function (value) { |
|
var tokens = value.split(/(\s+|[()[\]{}'"]|\b)/); // Join the boundary splits that we do not consider to be boundaries. This is primarily the extended Latin character set. |
|
|
|
for (var i = 0; i < tokens.length - 1; i++) { |
|
// If we have an empty string in the next field and we have only word chars before and after, merge |
|
if (!tokens[i + 1] && tokens[i + 2] && extendedWordChars.test(tokens[i]) && extendedWordChars.test(tokens[i + 2])) { |
|
tokens[i] += tokens[i + 2]; |
|
tokens.splice(i + 1, 2); |
|
i--; |
|
} |
|
} |
|
|
|
return tokens; |
|
}; |
|
|
|
function diffWords(oldStr, newStr, options) { |
|
options = generateOptions(options, { |
|
ignoreWhitespace: true |
|
}); |
|
return wordDiff.diff(oldStr, newStr, options); |
|
} |
|
|
|
function diffWordsWithSpace(oldStr, newStr, options) { |
|
return wordDiff.diff(oldStr, newStr, options); |
|
} |
|
|
|
var lineDiff = new Diff(); |
|
|
|
lineDiff.tokenize = function (value) { |
|
var retLines = [], |
|
linesAndNewlines = value.split(/(\n|\r\n)/); // Ignore the final empty token that occurs if the string ends with a new line |
|
|
|
if (!linesAndNewlines[linesAndNewlines.length - 1]) { |
|
linesAndNewlines.pop(); |
|
} // Merge the content and line separators into single tokens |
|
|
|
|
|
for (var i = 0; i < linesAndNewlines.length; i++) { |
|
var line = linesAndNewlines[i]; |
|
|
|
if (i % 2 && !this.options.newlineIsToken) { |
|
retLines[retLines.length - 1] += line; |
|
} else { |
|
if (this.options.ignoreWhitespace) { |
|
line = line.trim(); |
|
} |
|
|
|
retLines.push(line); |
|
} |
|
} |
|
|
|
return retLines; |
|
}; |
|
|
|
function diffLines(oldStr, newStr, callback) { |
|
return lineDiff.diff(oldStr, newStr, callback); |
|
} |
|
|
|
function diffTrimmedLines(oldStr, newStr, callback) { |
|
var options = generateOptions(callback, { |
|
ignoreWhitespace: true |
|
}); |
|
return lineDiff.diff(oldStr, newStr, options); |
|
} |
|
|
|
var sentenceDiff = new Diff(); |
|
|
|
sentenceDiff.tokenize = function (value) { |
|
return value.split(/(\S.+?[.!?])(?=\s+|$)/); |
|
}; |
|
|
|
function diffSentences(oldStr, newStr, callback) { |
|
return sentenceDiff.diff(oldStr, newStr, callback); |
|
} |
|
|
|
var cssDiff = new Diff(); |
|
|
|
cssDiff.tokenize = function (value) { |
|
return value.split(/([{}:;,]|\s+)/); |
|
}; |
|
|
|
function diffCss(oldStr, newStr, callback) { |
|
return cssDiff.diff(oldStr, newStr, callback); |
|
} |
|
|
|
function _typeof$1(obj) { |
|
if (typeof Symbol === "function" && _typeof(Symbol.iterator) === "symbol") { |
|
_typeof$1 = function _typeof$1(obj) { |
|
return _typeof(obj); |
|
}; |
|
} else { |
|
_typeof$1 = function _typeof$1(obj) { |
|
return obj && typeof Symbol === "function" && obj.constructor === Symbol && obj !== Symbol.prototype ? "symbol" : _typeof(obj); |
|
}; |
|
} |
|
|
|
return _typeof$1(obj); |
|
} |
|
|
|
function _toConsumableArray$1(arr) { |
|
return _arrayWithoutHoles$1(arr) || _iterableToArray$1(arr) || _nonIterableSpread$1(); |
|
} |
|
|
|
function _arrayWithoutHoles$1(arr) { |
|
if (Array.isArray(arr)) { |
|
for (var i = 0, arr2 = new Array(arr.length); i < arr.length; i++) { |
|
arr2[i] = arr[i]; |
|
} |
|
|
|
return arr2; |
|
} |
|
} |
|
|
|
function _iterableToArray$1(iter) { |
|
if (Symbol.iterator in Object(iter) || Object.prototype.toString.call(iter) === "[object Arguments]") return Array.from(iter); |
|
} |
|
|
|
function _nonIterableSpread$1() { |
|
throw new TypeError("Invalid attempt to spread non-iterable instance"); |
|
} |
|
|
|
var objectPrototypeToString = Object.prototype.toString; |
|
var jsonDiff = new Diff(); // Discriminate between two lines of pretty-printed, serialized JSON where one of them has a |
|
// dangling comma and the other doesn't. Turns out including the dangling comma yields the nicest output: |
|
|
|
jsonDiff.useLongestToken = true; |
|
jsonDiff.tokenize = lineDiff.tokenize; |
|
|
|
jsonDiff.castInput = function (value) { |
|
var _this$options = this.options, |
|
undefinedReplacement = _this$options.undefinedReplacement, |
|
_this$options$stringi = _this$options.stringifyReplacer, |
|
stringifyReplacer = _this$options$stringi === void 0 ? function (k, v) { |
|
return typeof v === 'undefined' ? undefinedReplacement : v; |
|
} : _this$options$stringi; |
|
return typeof value === 'string' ? value : JSON.stringify(canonicalize(value, null, null, stringifyReplacer), stringifyReplacer, ' '); |
|
}; |
|
|
|
jsonDiff.equals = function (left, right) { |
|
return Diff.prototype.equals.call(jsonDiff, left.replace(/,([\r\n])/g, '$1'), right.replace(/,([\r\n])/g, '$1')); |
|
}; |
|
|
|
function diffJson(oldObj, newObj, options) { |
|
return jsonDiff.diff(oldObj, newObj, options); |
|
} // This function handles the presence of circular references by bailing out when encountering an |
|
// object that is already on the "stack" of items being processed. Accepts an optional replacer |
|
|
|
|
|
function canonicalize(obj, stack, replacementStack, replacer, key) { |
|
stack = stack || []; |
|
replacementStack = replacementStack || []; |
|
|
|
if (replacer) { |
|
obj = replacer(key, obj); |
|
} |
|
|
|
var i; |
|
|
|
for (i = 0; i < stack.length; i += 1) { |
|
if (stack[i] === obj) { |
|
return replacementStack[i]; |
|
} |
|
} |
|
|
|
var canonicalizedObj; |
|
|
|
if ('[object Array]' === objectPrototypeToString.call(obj)) { |
|
stack.push(obj); |
|
canonicalizedObj = new Array(obj.length); |
|
replacementStack.push(canonicalizedObj); |
|
|
|
for (i = 0; i < obj.length; i += 1) { |
|
canonicalizedObj[i] = canonicalize(obj[i], stack, replacementStack, replacer, key); |
|
} |
|
|
|
stack.pop(); |
|
replacementStack.pop(); |
|
return canonicalizedObj; |
|
} |
|
|
|
if (obj && obj.toJSON) { |
|
obj = obj.toJSON(); |
|
} |
|
|
|
if (_typeof$1(obj) === 'object' && obj !== null) { |
|
stack.push(obj); |
|
canonicalizedObj = {}; |
|
replacementStack.push(canonicalizedObj); |
|
|
|
var sortedKeys = [], |
|
_key; |
|
|
|
for (_key in obj) { |
|
/* istanbul ignore else */ |
|
if (obj.hasOwnProperty(_key)) { |
|
sortedKeys.push(_key); |
|
} |
|
} |
|
|
|
sortedKeys.sort(); |
|
|
|
for (i = 0; i < sortedKeys.length; i += 1) { |
|
_key = sortedKeys[i]; |
|
canonicalizedObj[_key] = canonicalize(obj[_key], stack, replacementStack, replacer, _key); |
|
} |
|
|
|
stack.pop(); |
|
replacementStack.pop(); |
|
} else { |
|
canonicalizedObj = obj; |
|
} |
|
|
|
return canonicalizedObj; |
|
} |
|
|
|
var arrayDiff = new Diff(); |
|
|
|
arrayDiff.tokenize = function (value) { |
|
return value.slice(); |
|
}; |
|
|
|
arrayDiff.join = arrayDiff.removeEmpty = function (value) { |
|
return value; |
|
}; |
|
|
|
function diffArrays(oldArr, newArr, callback) { |
|
return arrayDiff.diff(oldArr, newArr, callback); |
|
} |
|
|
|
function parsePatch(uniDiff) { |
|
var options = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {}; |
|
var diffstr = uniDiff.split(/\r\n|[\n\v\f\r\x85]/), |
|
delimiters = uniDiff.match(/\r\n|[\n\v\f\r\x85]/g) || [], |
|
list = [], |
|
i = 0; |
|
|
|
function parseIndex() { |
|
var index = {}; |
|
list.push(index); // Parse diff metadata |
|
|
|
while (i < diffstr.length) { |
|
var line = diffstr[i]; // File header found, end parsing diff metadata |
|
|
|
if (/^(\-\-\-|\+\+\+|@@)\s/.test(line)) { |
|
break; |
|
} // Diff index |
|
|
|
|
|
var header = /^(?:Index:|diff(?: -r \w+)+)\s+(.+?)\s*$/.exec(line); |
|
|
|
if (header) { |
|
index.index = header[1]; |
|
} |
|
|
|
i++; |
|
} // Parse file headers if they are defined. Unified diff requires them, but |
|
// there's no technical issues to have an isolated hunk without file header |
|
|
|
|
|
parseFileHeader(index); |
|
parseFileHeader(index); // Parse hunks |
|
|
|
index.hunks = []; |
|
|
|
while (i < diffstr.length) { |
|
var _line = diffstr[i]; |
|
|
|
if (/^(Index:|diff|\-\-\-|\+\+\+)\s/.test(_line)) { |
|
break; |
|
} else if (/^@@/.test(_line)) { |
|
index.hunks.push(parseHunk()); |
|
} else if (_line && options.strict) { |
|
// Ignore unexpected content unless in strict mode |
|
throw new Error('Unknown line ' + (i + 1) + ' ' + JSON.stringify(_line)); |
|
} else { |
|
i++; |
|
} |
|
} |
|
} // Parses the --- and +++ headers, if none are found, no lines |
|
// are consumed. |
|
|
|
|
|
function parseFileHeader(index) { |
|
var fileHeader = /^(---|\+\+\+)\s+(.*)$/.exec(diffstr[i]); |
|
|
|
if (fileHeader) { |
|
var keyPrefix = fileHeader[1] === '---' ? 'old' : 'new'; |
|
var data = fileHeader[2].split('\t', 2); |
|
var fileName = data[0].replace(/\\\\/g, '\\'); |
|
|
|
if (/^".*"$/.test(fileName)) { |
|
fileName = fileName.substr(1, fileName.length - 2); |
|
} |
|
|
|
index[keyPrefix + 'FileName'] = fileName; |
|
index[keyPrefix + 'Header'] = (data[1] || '').trim(); |
|
i++; |
|
} |
|
} // Parses a hunk |
|
// This assumes that we are at the start of a hunk. |
|
|
|
|
|
function parseHunk() { |
|
var chunkHeaderIndex = i, |
|
chunkHeaderLine = diffstr[i++], |
|
chunkHeader = chunkHeaderLine.split(/@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@/); |
|
var hunk = { |
|
oldStart: +chunkHeader[1], |
|
oldLines: +chunkHeader[2] || 1, |
|
newStart: +chunkHeader[3], |
|
newLines: +chunkHeader[4] || 1, |
|
lines: [], |
|
linedelimiters: [] |
|
}; |
|
var addCount = 0, |
|
removeCount = 0; |
|
|
|
for (; i < diffstr.length; i++) { |
|
// Lines starting with '---' could be mistaken for the "remove line" operation |
|
// But they could be the header for the next file. Therefore prune such cases out. |
|
if (diffstr[i].indexOf('--- ') === 0 && i + 2 < diffstr.length && diffstr[i + 1].indexOf('+++ ') === 0 && diffstr[i + 2].indexOf('@@') === 0) { |
|
break; |
|
} |
|
|
|
var operation = diffstr[i].length == 0 && i != diffstr.length - 1 ? ' ' : diffstr[i][0]; |
|
|
|
if (operation === '+' || operation === '-' || operation === ' ' || operation === '\\') { |
|
hunk.lines.push(diffstr[i]); |
|
hunk.linedelimiters.push(delimiters[i] || '\n'); |
|
|
|
if (operation === '+') { |
|
addCount++; |
|
} else if (operation === '-') { |
|
removeCount++; |
|
} else if (operation === ' ') { |
|
addCount++; |
|
removeCount++; |
|
} |
|
} else { |
|
break; |
|
} |
|
} // Handle the empty block count case |
|
|
|
|
|
if (!addCount && hunk.newLines === 1) { |
|
hunk.newLines = 0; |
|
} |
|
|
|
if (!removeCount && hunk.oldLines === 1) { |
|
hunk.oldLines = 0; |
|
} // Perform optional sanity checking |
|
|
|
|
|
if (options.strict) { |
|
if (addCount !== hunk.newLines) { |
|
throw new Error('Added line count did not match for hunk at line ' + (chunkHeaderIndex + 1)); |
|
} |
|
|
|
if (removeCount !== hunk.oldLines) { |
|
throw new Error('Removed line count did not match for hunk at line ' + (chunkHeaderIndex + 1)); |
|
} |
|
} |
|
|
|
return hunk; |
|
} |
|
|
|
while (i < diffstr.length) { |
|
parseIndex(); |
|
} |
|
|
|
return list; |
|
} // Iterator that traverses in the range of [min, max], stepping |
|
// by distance from a given start position. I.e. for [0, 4], with |
|
// start of 2, this will iterate 2, 3, 1, 4, 0. |
|
|
|
|
|
function distanceIterator(start, minLine, maxLine) { |
|
var wantForward = true, |
|
backwardExhausted = false, |
|
forwardExhausted = false, |
|
localOffset = 1; |
|
return function iterator() { |
|
if (wantForward && !forwardExhausted) { |
|
if (backwardExhausted) { |
|
localOffset++; |
|
} else { |
|
wantForward = false; |
|
} // Check if trying to fit beyond text length, and if not, check it fits |
|
// after offset location (or desired location on first iteration) |
|
|
|
|
|
if (start + localOffset <= maxLine) { |
|
return localOffset; |
|
} |
|
|
|
forwardExhausted = true; |
|
} |
|
|
|
if (!backwardExhausted) { |
|
if (!forwardExhausted) { |
|
wantForward = true; |
|
} // Check if trying to fit before text beginning, and if not, check it fits |
|
// before offset location |
|
|
|
|
|
if (minLine <= start - localOffset) { |
|
return -localOffset++; |
|
} |
|
|
|
backwardExhausted = true; |
|
return iterator(); |
|
} // We tried to fit hunk before text beginning and beyond text length, then |
|
// hunk can't fit on the text. Return undefined |
|
|
|
}; |
|
} |
|
|
|
function applyPatch(source, uniDiff) { |
|
var options = arguments.length > 2 && arguments[2] !== undefined ? arguments[2] : {}; |
|
|
|
if (typeof uniDiff === 'string') { |
|
uniDiff = parsePatch(uniDiff); |
|
} |
|
|
|
if (Array.isArray(uniDiff)) { |
|
if (uniDiff.length > 1) { |
|
throw new Error('applyPatch only works with a single input.'); |
|
} |
|
|
|
uniDiff = uniDiff[0]; |
|
} // Apply the diff to the input |
|
|
|
|
|
var lines = source.split(/\r\n|[\n\v\f\r\x85]/), |
|
delimiters = source.match(/\r\n|[\n\v\f\r\x85]/g) || [], |
|
hunks = uniDiff.hunks, |
|
compareLine = options.compareLine || function (lineNumber, line, operation, patchContent) { |
|
return line === patchContent; |
|
}, |
|
errorCount = 0, |
|
fuzzFactor = options.fuzzFactor || 0, |
|
minLine = 0, |
|
offset = 0, |
|
removeEOFNL, |
|
addEOFNL; |
|
/** |
|
* Checks if the hunk exactly fits on the provided location |
|
*/ |
|
|
|
|
|
function hunkFits(hunk, toPos) { |
|
for (var j = 0; j < hunk.lines.length; j++) { |
|
var line = hunk.lines[j], |
|
operation = line.length > 0 ? line[0] : ' ', |
|
content = line.length > 0 ? line.substr(1) : line; |
|
|
|
if (operation === ' ' || operation === '-') { |
|
// Context sanity check |
|
if (!compareLine(toPos + 1, lines[toPos], operation, content)) { |
|
errorCount++; |
|
|
|
if (errorCount > fuzzFactor) { |
|
return false; |
|
} |
|
} |
|
|
|
toPos++; |
|
} |
|
} |
|
|
|
return true; |
|
} // Search best fit offsets for each hunk based on the previous ones |
|
|
|
|
|
for (var i = 0; i < hunks.length; i++) { |
|
var hunk = hunks[i], |
|
maxLine = lines.length - hunk.oldLines, |
|
localOffset = 0, |
|
toPos = offset + hunk.oldStart - 1; |
|
var iterator = distanceIterator(toPos, minLine, maxLine); |
|
|
|
for (; localOffset !== undefined; localOffset = iterator()) { |
|
if (hunkFits(hunk, toPos + localOffset)) { |
|
hunk.offset = offset += localOffset; |
|
break; |
|
} |
|
} |
|
|
|
if (localOffset === undefined) { |
|
return false; |
|
} // Set lower text limit to end of the current hunk, so next ones don't try |
|
// to fit over already patched text |
|
|
|
|
|
minLine = hunk.offset + hunk.oldStart + hunk.oldLines; |
|
} // Apply patch hunks |
|
|
|
|
|
var diffOffset = 0; |
|
|
|
for (var _i = 0; _i < hunks.length; _i++) { |
|
var _hunk = hunks[_i], |
|
_toPos = _hunk.oldStart + _hunk.offset + diffOffset - 1; |
|
|
|
diffOffset += _hunk.newLines - _hunk.oldLines; |
|
|
|
if (_toPos < 0) { |
|
// Creating a new file |
|
_toPos = 0; |
|
} |
|
|
|
for (var j = 0; j < _hunk.lines.length; j++) { |
|
var line = _hunk.lines[j], |
|
operation = line.length > 0 ? line[0] : ' ', |
|
content = line.length > 0 ? line.substr(1) : line, |
|
delimiter = _hunk.linedelimiters[j]; |
|
|
|
if (operation === ' ') { |
|
_toPos++; |
|
} else if (operation === '-') { |
|
lines.splice(_toPos, 1); |
|
delimiters.splice(_toPos, 1); |
|
/* istanbul ignore else */ |
|
} else if (operation === '+') { |
|
lines.splice(_toPos, 0, content); |
|
delimiters.splice(_toPos, 0, delimiter); |
|
_toPos++; |
|
} else if (operation === '\\') { |
|
var previousOperation = _hunk.lines[j - 1] ? _hunk.lines[j - 1][0] : null; |
|
|
|
if (previousOperation === '+') { |
|
removeEOFNL = true; |
|
} else if (previousOperation === '-') { |
|
addEOFNL = true; |
|
} |
|
} |
|
} |
|
} // Handle EOFNL insertion/removal |
|
|
|
|
|
if (removeEOFNL) { |
|
while (!lines[lines.length - 1]) { |
|
lines.pop(); |
|
delimiters.pop(); |
|
} |
|
} else if (addEOFNL) { |
|
lines.push(''); |
|
delimiters.push('\n'); |
|
} |
|
|
|
for (var _k = 0; _k < lines.length - 1; _k++) { |
|
lines[_k] = lines[_k] + delimiters[_k]; |
|
} |
|
|
|
return lines.join(''); |
|
} // Wrapper that supports multiple file patches via callbacks. |
|
|
|
|
|
function applyPatches(uniDiff, options) { |
|
if (typeof uniDiff === 'string') { |
|
uniDiff = parsePatch(uniDiff); |
|
} |
|
|
|
var currentIndex = 0; |
|
|
|
function processIndex() { |
|
var index = uniDiff[currentIndex++]; |
|
|
|
if (!index) { |
|
return options.complete(); |
|
} |
|
|
|
options.loadFile(index, function (err, data) { |
|
if (err) { |
|
return options.complete(err); |
|
} |
|
|
|
var updatedContent = applyPatch(data, index, options); |
|
options.patched(index, updatedContent, function (err) { |
|
if (err) { |
|
return options.complete(err); |
|
} |
|
|
|
processIndex(); |
|
}); |
|
}); |
|
} |
|
|
|
processIndex(); |
|
} |
|
|
|
function structuredPatch(oldFileName, newFileName, oldStr, newStr, oldHeader, newHeader, options) { |
|
if (!options) { |
|
options = {}; |
|
} |
|
|
|
if (typeof options.context === 'undefined') { |
|
options.context = 4; |
|
} |
|
|
|
var diff = diffLines(oldStr, newStr, options); |
|
diff.push({ |
|
value: '', |
|
lines: [] |
|
}); // Append an empty value to make cleanup easier |
|
|
|
function contextLines(lines) { |
|
return lines.map(function (entry) { |
|
return ' ' + entry; |
|
}); |
|
} |
|
|
|
var hunks = []; |
|
var oldRangeStart = 0, |
|
newRangeStart = 0, |
|
curRange = [], |
|
oldLine = 1, |
|
newLine = 1; |
|
|
|
var _loop = function _loop(i) { |
|
var current = diff[i], |
|
lines = current.lines || current.value.replace(/\n$/, '').split('\n'); |
|
current.lines = lines; |
|
|
|
if (current.added || current.removed) { |
|
var _curRange; // If we have previous context, start with that |
|
|
|
|
|
if (!oldRangeStart) { |
|
var prev = diff[i - 1]; |
|
oldRangeStart = oldLine; |
|
newRangeStart = newLine; |
|
|
|
if (prev) { |
|
curRange = options.context > 0 ? contextLines(prev.lines.slice(-options.context)) : []; |
|
oldRangeStart -= curRange.length; |
|
newRangeStart -= curRange.length; |
|
} |
|
} // Output our changes |
|
|
|
|
|
(_curRange = curRange).push.apply(_curRange, _toConsumableArray$1(lines.map(function (entry) { |
|
return (current.added ? '+' : '-') + entry; |
|
}))); // Track the updated file position |
|
|
|
|
|
if (current.added) { |
|
newLine += lines.length; |
|
} else { |
|
oldLine += lines.length; |
|
} |
|
} else { |
|
// Identical context lines. Track line changes |
|
if (oldRangeStart) { |
|
// Close out any changes that have been output (or join overlapping) |
|
if (lines.length <= options.context * 2 && i < diff.length - 2) { |
|
var _curRange2; // Overlapping |
|
|
|
|
|
(_curRange2 = curRange).push.apply(_curRange2, _toConsumableArray$1(contextLines(lines))); |
|
} else { |
|
var _curRange3; // end the range and output |
|
|
|
|
|
var contextSize = Math.min(lines.length, options.context); |
|
|
|
(_curRange3 = curRange).push.apply(_curRange3, _toConsumableArray$1(contextLines(lines.slice(0, contextSize)))); |
|
|
|
var hunk = { |
|
oldStart: oldRangeStart, |
|
oldLines: oldLine - oldRangeStart + contextSize, |
|
newStart: newRangeStart, |
|
newLines: newLine - newRangeStart + contextSize, |
|
lines: curRange |
|
}; |
|
|
|
if (i >= diff.length - 2 && lines.length <= options.context) { |
|
// EOF is inside this hunk |
|
var oldEOFNewline = /\n$/.test(oldStr); |
|
var newEOFNewline = /\n$/.test(newStr); |
|
var noNlBeforeAdds = lines.length == 0 && curRange.length > hunk.oldLines; |
|
|
|
if (!oldEOFNewline && noNlBeforeAdds) { |
|
// special case: old has no eol and no trailing context; no-nl can end up before adds |
|
curRange.splice(hunk.oldLines, 0, '\\ No newline at end of file'); |
|
} |
|
|
|
if (!oldEOFNewline && !noNlBeforeAdds || !newEOFNewline) { |
|
curRange.push('\\ No newline at end of file'); |
|
} |
|
} |
|
|
|
hunks.push(hunk); |
|
oldRangeStart = 0; |
|
newRangeStart = 0; |
|
curRange = []; |
|
} |
|
} |
|
|
|
oldLine += lines.length; |
|
newLine += lines.length; |
|
} |
|
}; |
|
|
|
for (var i = 0; i < diff.length; i++) { |
|
_loop(i); |
|
} |
|
|
|
return { |
|
oldFileName: oldFileName, |
|
newFileName: newFileName, |
|
oldHeader: oldHeader, |
|
newHeader: newHeader, |
|
hunks: hunks |
|
}; |
|
} |
|
|
|
function createTwoFilesPatch(oldFileName, newFileName, oldStr, newStr, oldHeader, newHeader, options) { |
|
var diff = structuredPatch(oldFileName, newFileName, oldStr, newStr, oldHeader, newHeader, options); |
|
var ret = []; |
|
|
|
if (oldFileName == newFileName) { |
|
ret.push('Index: ' + oldFileName); |
|
} |
|
|
|
ret.push('==================================================================='); |
|
ret.push('--- ' + diff.oldFileName + (typeof diff.oldHeader === 'undefined' ? '' : '\t' + diff.oldHeader)); |
|
ret.push('+++ ' + diff.newFileName + (typeof diff.newHeader === 'undefined' ? '' : '\t' + diff.newHeader)); |
|
|
|
for (var i = 0; i < diff.hunks.length; i++) { |
|
var hunk = diff.hunks[i]; |
|
ret.push('@@ -' + hunk.oldStart + ',' + hunk.oldLines + ' +' + hunk.newStart + ',' + hunk.newLines + ' @@'); |
|
ret.push.apply(ret, hunk.lines); |
|
} |
|
|
|
return ret.join('\n') + '\n'; |
|
} |
|
|
|
function createPatch(fileName, oldStr, newStr, oldHeader, newHeader, options) { |
|
return createTwoFilesPatch(fileName, fileName, oldStr, newStr, oldHeader, newHeader, options); |
|
} |
|
|
|
function arrayEqual(a, b) { |
|
if (a.length !== b.length) { |
|
return false; |
|
} |
|
|
|
return arrayStartsWith(a, b); |
|
} |
|
|
|
function arrayStartsWith(array, start) { |
|
if (start.length > array.length) { |
|
return false; |
|
} |
|
|
|
for (var i = 0; i < start.length; i++) { |
|
if (start[i] !== array[i]) { |
|
return false; |
|
} |
|
} |
|
|
|
return true; |
|
} |
|
|
|
function calcLineCount(hunk) { |
|
var _calcOldNewLineCount = calcOldNewLineCount(hunk.lines), |
|
oldLines = _calcOldNewLineCount.oldLines, |
|
newLines = _calcOldNewLineCount.newLines; |
|
|
|
if (oldLines !== undefined) { |
|
hunk.oldLines = oldLines; |
|
} else { |
|
delete hunk.oldLines; |
|
} |
|
|
|
if (newLines !== undefined) { |
|
hunk.newLines = newLines; |
|
} else { |
|
delete hunk.newLines; |
|
} |
|
} |
|
|
|
function merge(mine, theirs, base) { |
|
mine = loadPatch(mine, base); |
|
theirs = loadPatch(theirs, base); |
|
var ret = {}; // For index we just let it pass through as it doesn't have any necessary meaning. |
|
// Leaving sanity checks on this to the API consumer that may know more about the |
|
// meaning in their own context. |
|
|
|
if (mine.index || theirs.index) { |
|
ret.index = mine.index || theirs.index; |
|
} |
|
|
|
if (mine.newFileName || theirs.newFileName) { |
|
if (!fileNameChanged(mine)) { |
|
// No header or no change in ours, use theirs (and ours if theirs does not exist) |
|
ret.oldFileName = theirs.oldFileName || mine.oldFileName; |
|
ret.newFileName = theirs.newFileName || mine.newFileName; |
|
ret.oldHeader = theirs.oldHeader || mine.oldHeader; |
|
ret.newHeader = theirs.newHeader || mine.newHeader; |
|
} else if (!fileNameChanged(theirs)) { |
|
// No header or no change in theirs, use ours |
|
ret.oldFileName = mine.oldFileName; |
|
ret.newFileName = mine.newFileName; |
|
ret.oldHeader = mine.oldHeader; |
|
ret.newHeader = mine.newHeader; |
|
} else { |
|
// Both changed... figure it out |
|
ret.oldFileName = selectField(ret, mine.oldFileName, theirs.oldFileName); |
|
ret.newFileName = selectField(ret, mine.newFileName, theirs.newFileName); |
|
ret.oldHeader = selectField(ret, mine.oldHeader, theirs.oldHeader); |
|
ret.newHeader = selectField(ret, mine.newHeader, theirs.newHeader); |
|
} |
|
} |
|
|
|
ret.hunks = []; |
|
var mineIndex = 0, |
|
theirsIndex = 0, |
|
mineOffset = 0, |
|
theirsOffset = 0; |
|
|
|
while (mineIndex < mine.hunks.length || theirsIndex < theirs.hunks.length) { |
|
var mineCurrent = mine.hunks[mineIndex] || { |
|
oldStart: Infinity |
|
}, |
|
theirsCurrent = theirs.hunks[theirsIndex] || { |
|
oldStart: Infinity |
|
}; |
|
|
|
if (hunkBefore(mineCurrent, theirsCurrent)) { |
|
// This patch does not overlap with any of the others, yay. |
|
ret.hunks.push(cloneHunk(mineCurrent, mineOffset)); |
|
mineIndex++; |
|
theirsOffset += mineCurrent.newLines - mineCurrent.oldLines; |
|
} else if (hunkBefore(theirsCurrent, mineCurrent)) { |
|
// This patch does not overlap with any of the others, yay. |
|
ret.hunks.push(cloneHunk(theirsCurrent, theirsOffset)); |
|
theirsIndex++; |
|
mineOffset += theirsCurrent.newLines - theirsCurrent.oldLines; |
|
} else { |
|
// Overlap, merge as best we can |
|
var mergedHunk = { |
|
oldStart: Math.min(mineCurrent.oldStart, theirsCurrent.oldStart), |
|
oldLines: 0, |
|
newStart: Math.min(mineCurrent.newStart + mineOffset, theirsCurrent.oldStart + theirsOffset), |
|
newLines: 0, |
|
lines: [] |
|
}; |
|
mergeLines(mergedHunk, mineCurrent.oldStart, mineCurrent.lines, theirsCurrent.oldStart, theirsCurrent.lines); |
|
theirsIndex++; |
|
mineIndex++; |
|
ret.hunks.push(mergedHunk); |
|
} |
|
} |
|
|
|
return ret; |
|
} |
|
|
|
function loadPatch(param, base) { |
|
if (typeof param === 'string') { |
|
if (/^@@/m.test(param) || /^Index:/m.test(param)) { |
|
return parsePatch(param)[0]; |
|
} |
|
|
|
if (!base) { |
|
throw new Error('Must provide a base reference or pass in a patch'); |
|
} |
|
|
|
return structuredPatch(undefined, undefined, base, param); |
|
} |
|
|
|
return param; |
|
} |
|
|
|
function fileNameChanged(patch) { |
|
return patch.newFileName && patch.newFileName !== patch.oldFileName; |
|
} |
|
|
|
function selectField(index, mine, theirs) { |
|
if (mine === theirs) { |
|
return mine; |
|
} else { |
|
index.conflict = true; |
|
return { |
|
mine: mine, |
|
theirs: theirs |
|
}; |
|
} |
|
} |
|
|
|
function hunkBefore(test, check) { |
|
return test.oldStart < check.oldStart && test.oldStart + test.oldLines < check.oldStart; |
|
} |
|
|
|
function cloneHunk(hunk, offset) { |
|
return { |
|
oldStart: hunk.oldStart, |
|
oldLines: hunk.oldLines, |
|
newStart: hunk.newStart + offset, |
|
newLines: hunk.newLines, |
|
lines: hunk.lines |
|
}; |
|
} |
|
|
|
function mergeLines(hunk, mineOffset, mineLines, theirOffset, theirLines) { |
|
// This will generally result in a conflicted hunk, but there are cases where the context |
|
// is the only overlap where we can successfully merge the content here. |
|
var mine = { |
|
offset: mineOffset, |
|
lines: mineLines, |
|
index: 0 |
|
}, |
|
their = { |
|
offset: theirOffset, |
|
lines: theirLines, |
|
index: 0 |
|
}; // Handle any leading content |
|
|
|
insertLeading(hunk, mine, their); |
|
insertLeading(hunk, their, mine); // Now in the overlap content. Scan through and select the best changes from each. |
|
|
|
while (mine.index < mine.lines.length && their.index < their.lines.length) { |
|
var mineCurrent = mine.lines[mine.index], |
|
theirCurrent = their.lines[their.index]; |
|
|
|
if ((mineCurrent[0] === '-' || mineCurrent[0] === '+') && (theirCurrent[0] === '-' || theirCurrent[0] === '+')) { |
|
// Both modified ... |
|
mutualChange(hunk, mine, their); |
|
} else if (mineCurrent[0] === '+' && theirCurrent[0] === ' ') { |
|
var _hunk$lines; // Mine inserted |
|
|
|
|
|
(_hunk$lines = hunk.lines).push.apply(_hunk$lines, _toConsumableArray$1(collectChange(mine))); |
|
} else if (theirCurrent[0] === '+' && mineCurrent[0] === ' ') { |
|
var _hunk$lines2; // Theirs inserted |
|
|
|
|
|
(_hunk$lines2 = hunk.lines).push.apply(_hunk$lines2, _toConsumableArray$1(collectChange(their))); |
|
} else if (mineCurrent[0] === '-' && theirCurrent[0] === ' ') { |
|
// Mine removed or edited |
|
removal(hunk, mine, their); |
|
} else if (theirCurrent[0] === '-' && mineCurrent[0] === ' ') { |
|
// Their removed or edited |
|
removal(hunk, their, mine, true); |
|
} else if (mineCurrent === theirCurrent) { |
|
// Context identity |
|
hunk.lines.push(mineCurrent); |
|
mine.index++; |
|
their.index++; |
|
} else { |
|
// Context mismatch |
|
conflict(hunk, collectChange(mine), collectChange(their)); |
|
} |
|
} // Now push anything that may be remaining |
|
|
|
|
|
insertTrailing(hunk, mine); |
|
insertTrailing(hunk, their); |
|
calcLineCount(hunk); |
|
} |
|
|
|
function mutualChange(hunk, mine, their) { |
|
var myChanges = collectChange(mine), |
|
theirChanges = collectChange(their); |
|
|
|
if (allRemoves(myChanges) && allRemoves(theirChanges)) { |
|
// Special case for remove changes that are supersets of one another |
|
if (arrayStartsWith(myChanges, theirChanges) && skipRemoveSuperset(their, myChanges, myChanges.length - theirChanges.length)) { |
|
var _hunk$lines3; |
|
|
|
(_hunk$lines3 = hunk.lines).push.apply(_hunk$lines3, _toConsumableArray$1(myChanges)); |
|
|
|
return; |
|
} else if (arrayStartsWith(theirChanges, myChanges) && skipRemoveSuperset(mine, theirChanges, theirChanges.length - myChanges.length)) { |
|
var _hunk$lines4; |
|
|
|
(_hunk$lines4 = hunk.lines).push.apply(_hunk$lines4, _toConsumableArray$1(theirChanges)); |
|
|
|
return; |
|
} |
|
} else if (arrayEqual(myChanges, theirChanges)) { |
|
var _hunk$lines5; |
|
|
|
(_hunk$lines5 = hunk.lines).push.apply(_hunk$lines5, _toConsumableArray$1(myChanges)); |
|
|
|
return; |
|
} |
|
|
|
conflict(hunk, myChanges, theirChanges); |
|
} |
|
|
|
function removal(hunk, mine, their, swap) { |
|
var myChanges = collectChange(mine), |
|
theirChanges = collectContext(their, myChanges); |
|
|
|
if (theirChanges.merged) { |
|
var _hunk$lines6; |
|
|
|
(_hunk$lines6 = hunk.lines).push.apply(_hunk$lines6, _toConsumableArray$1(theirChanges.merged)); |
|
} else { |
|
conflict(hunk, swap ? theirChanges : myChanges, swap ? myChanges : theirChanges); |
|
} |
|
} |
|
|
|
function conflict(hunk, mine, their) { |
|
hunk.conflict = true; |
|
hunk.lines.push({ |
|
conflict: true, |
|
mine: mine, |
|
theirs: their |
|
}); |
|
} |
|
|
|
function insertLeading(hunk, insert, their) { |
|
while (insert.offset < their.offset && insert.index < insert.lines.length) { |
|
var line = insert.lines[insert.index++]; |
|
hunk.lines.push(line); |
|
insert.offset++; |
|
} |
|
} |
|
|
|
function insertTrailing(hunk, insert) { |
|
while (insert.index < insert.lines.length) { |
|
var line = insert.lines[insert.index++]; |
|
hunk.lines.push(line); |
|
} |
|
} |
|
|
|
function collectChange(state) { |
|
var ret = [], |
|
operation = state.lines[state.index][0]; |
|
|
|
while (state.index < state.lines.length) { |
|
var line = state.lines[state.index]; // Group additions that are immediately after subtractions and treat them as one "atomic" modify change. |
|
|
|
if (operation === '-' && line[0] === '+') { |
|
operation = '+'; |
|
} |
|
|
|
if (operation === line[0]) { |
|
ret.push(line); |
|
state.index++; |
|
} else { |
|
break; |
|
} |
|
} |
|
|
|
return ret; |
|
} |
|
|
|
function collectContext(state, matchChanges) { |
|
var changes = [], |
|
merged = [], |
|
matchIndex = 0, |
|
contextChanges = false, |
|
conflicted = false; |
|
|
|
while (matchIndex < matchChanges.length && state.index < state.lines.length) { |
|
var change = state.lines[state.index], |
|
match = matchChanges[matchIndex]; // Once we've hit our add, then we are done |
|
|
|
if (match[0] === '+') { |
|
break; |
|
} |
|
|
|
contextChanges = contextChanges || change[0] !== ' '; |
|
merged.push(match); |
|
matchIndex++; // Consume any additions in the other block as a conflict to attempt |
|
// to pull in the remaining context after this |
|
|
|
if (change[0] === '+') { |
|
conflicted = true; |
|
|
|
while (change[0] === '+') { |
|
changes.push(change); |
|
change = state.lines[++state.index]; |
|
} |
|
} |
|
|
|
if (match.substr(1) === change.substr(1)) { |
|
changes.push(change); |
|
state.index++; |
|
} else { |
|
conflicted = true; |
|
} |
|
} |
|
|
|
if ((matchChanges[matchIndex] || '')[0] === '+' && contextChanges) { |
|
conflicted = true; |
|
} |
|
|
|
if (conflicted) { |
|
return changes; |
|
} |
|
|
|
while (matchIndex < matchChanges.length) { |
|
merged.push(matchChanges[matchIndex++]); |
|
} |
|
|
|
return { |
|
merged: merged, |
|
changes: changes |
|
}; |
|
} |
|
|
|
function allRemoves(changes) { |
|
return changes.reduce(function (prev, change) { |
|
return prev && change[0] === '-'; |
|
}, true); |
|
} |
|
|
|
function skipRemoveSuperset(state, removeChanges, delta) { |
|
for (var i = 0; i < delta; i++) { |
|
var changeContent = removeChanges[removeChanges.length - delta + i].substr(1); |
|
|
|
if (state.lines[state.index + i] !== ' ' + changeContent) { |
|
return false; |
|
} |
|
} |
|
|
|
state.index += delta; |
|
return true; |
|
} |
|
|
|
function calcOldNewLineCount(lines) { |
|
var oldLines = 0; |
|
var newLines = 0; |
|
lines.forEach(function (line) { |
|
if (typeof line !== 'string') { |
|
var myCount = calcOldNewLineCount(line.mine); |
|
var theirCount = calcOldNewLineCount(line.theirs); |
|
|
|
if (oldLines !== undefined) { |
|
if (myCount.oldLines === theirCount.oldLines) { |
|
oldLines += myCount.oldLines; |
|
} else { |
|
oldLines = undefined; |
|
} |
|
} |
|
|
|
if (newLines !== undefined) { |
|
if (myCount.newLines === theirCount.newLines) { |
|
newLines += myCount.newLines; |
|
} else { |
|
newLines = undefined; |
|
} |
|
} |
|
} else { |
|
if (newLines !== undefined && (line[0] === '+' || line[0] === ' ')) { |
|
newLines++; |
|
} |
|
|
|
if (oldLines !== undefined && (line[0] === '-' || line[0] === ' ')) { |
|
oldLines++; |
|
} |
|
} |
|
}); |
|
return { |
|
oldLines: oldLines, |
|
newLines: newLines |
|
}; |
|
} // See: http://code.google.com/p/google-diff-match-patch/wiki/API |
|
|
|
|
|
function convertChangesToDMP(changes) { |
|
var ret = [], |
|
change, |
|
operation; |
|
|
|
for (var i = 0; i < changes.length; i++) { |
|
change = changes[i]; |
|
|
|
if (change.added) { |
|
operation = 1; |
|
} else if (change.removed) { |
|
operation = -1; |
|
} else { |
|
operation = 0; |
|
} |
|
|
|
ret.push([operation, change.value]); |
|
} |
|
|
|
return ret; |
|
} |
|
|
|
function convertChangesToXML(changes) { |
|
var ret = []; |
|
|
|
for (var i = 0; i < changes.length; i++) { |
|
var change = changes[i]; |
|
|
|
if (change.added) { |
|
ret.push('<ins>'); |
|
} else if (change.removed) { |
|
ret.push('<del>'); |
|
} |
|
|
|
ret.push(escapeHTML(change.value)); |
|
|
|
if (change.added) { |
|
ret.push('</ins>'); |
|
} else if (change.removed) { |
|
ret.push('</del>'); |
|
} |
|
} |
|
|
|
return ret.join(''); |
|
} |
|
|
|
function escapeHTML(s) { |
|
var n = s; |
|
n = n.replace(/&/g, '&'); |
|
n = n.replace(/</g, '<'); |
|
n = n.replace(/>/g, '>'); |
|
n = n.replace(/"/g, '"'); |
|
return n; |
|
} |
|
|
|
var index_es6 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
Diff: Diff, |
|
diffChars: diffChars, |
|
diffWords: diffWords, |
|
diffWordsWithSpace: diffWordsWithSpace, |
|
diffLines: diffLines, |
|
diffTrimmedLines: diffTrimmedLines, |
|
diffSentences: diffSentences, |
|
diffCss: diffCss, |
|
diffJson: diffJson, |
|
diffArrays: diffArrays, |
|
structuredPatch: structuredPatch, |
|
createTwoFilesPatch: createTwoFilesPatch, |
|
createPatch: createPatch, |
|
applyPatch: applyPatch, |
|
applyPatches: applyPatches, |
|
parsePatch: parsePatch, |
|
merge: merge, |
|
convertChangesToDMP: convertChangesToDMP, |
|
convertChangesToXML: convertChangesToXML, |
|
canonicalize: canonicalize |
|
}); |
|
|
|
var _shim_fs = {}; |
|
|
|
var _shim_fs$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
'default': _shim_fs |
|
}); |
|
|
|
/*! |
|
* normalize-path <https://github.com/jonschlinkert/normalize-path> |
|
* |
|
* Copyright (c) 2014-2018, Jon Schlinkert. |
|
* Released under the MIT License. |
|
*/ |
|
var normalizePath = function normalizePath(path, stripTrailing) { |
|
if (typeof path !== 'string') { |
|
throw new TypeError('expected path to be a string'); |
|
} |
|
|
|
if (path === '\\' || path === '/') return '/'; |
|
var len = path.length; |
|
if (len <= 1) return path; // ensure that win32 namespaces has two leading slashes, so that the path is |
|
// handled properly by the win32 version of path.parse() after being normalized |
|
// https://msdn.microsoft.com/library/windows/desktop/aa365247(v=vs.85).aspx#namespaces |
|
|
|
var prefix = ''; |
|
|
|
if (len > 4 && path[3] === '\\') { |
|
var ch = path[2]; |
|
|
|
if ((ch === '?' || ch === '.') && path.slice(0, 2) === '\\\\') { |
|
path = path.slice(2); |
|
prefix = '//'; |
|
} |
|
} |
|
|
|
var segs = path.split(/[/\\]+/); |
|
|
|
if (stripTrailing !== false && segs[segs.length - 1] === '') { |
|
segs.pop(); |
|
} |
|
|
|
return prefix + segs.join('/'); |
|
}; |
|
|
|
var global$1 = typeof global !== "undefined" ? global : typeof self !== "undefined" ? self : typeof window !== "undefined" ? window : {}; |
|
|
|
var lookup = []; |
|
var revLookup = []; |
|
var Arr = typeof Uint8Array !== 'undefined' ? Uint8Array : Array; |
|
var inited = false; |
|
|
|
function init() { |
|
inited = true; |
|
var code = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/'; |
|
|
|
for (var i = 0, len = code.length; i < len; ++i) { |
|
lookup[i] = code[i]; |
|
revLookup[code.charCodeAt(i)] = i; |
|
} |
|
|
|
revLookup['-'.charCodeAt(0)] = 62; |
|
revLookup['_'.charCodeAt(0)] = 63; |
|
} |
|
|
|
function toByteArray(b64) { |
|
if (!inited) { |
|
init(); |
|
} |
|
|
|
var i, j, l, tmp, placeHolders, arr; |
|
var len = b64.length; |
|
|
|
if (len % 4 > 0) { |
|
throw new Error('Invalid string. Length must be a multiple of 4'); |
|
} // the number of equal signs (place holders) |
|
// if there are two placeholders, than the two characters before it |
|
// represent one byte |
|
// if there is only one, then the three characters before it represent 2 bytes |
|
// this is just a cheap hack to not do indexOf twice |
|
|
|
|
|
placeHolders = b64[len - 2] === '=' ? 2 : b64[len - 1] === '=' ? 1 : 0; // base64 is 4/3 + up to two characters of the original data |
|
|
|
arr = new Arr(len * 3 / 4 - placeHolders); // if there are placeholders, only get up to the last complete 4 chars |
|
|
|
l = placeHolders > 0 ? len - 4 : len; |
|
var L = 0; |
|
|
|
for (i = 0, j = 0; i < l; i += 4, j += 3) { |
|
tmp = revLookup[b64.charCodeAt(i)] << 18 | revLookup[b64.charCodeAt(i + 1)] << 12 | revLookup[b64.charCodeAt(i + 2)] << 6 | revLookup[b64.charCodeAt(i + 3)]; |
|
arr[L++] = tmp >> 16 & 0xFF; |
|
arr[L++] = tmp >> 8 & 0xFF; |
|
arr[L++] = tmp & 0xFF; |
|
} |
|
|
|
if (placeHolders === 2) { |
|
tmp = revLookup[b64.charCodeAt(i)] << 2 | revLookup[b64.charCodeAt(i + 1)] >> 4; |
|
arr[L++] = tmp & 0xFF; |
|
} else if (placeHolders === 1) { |
|
tmp = revLookup[b64.charCodeAt(i)] << 10 | revLookup[b64.charCodeAt(i + 1)] << 4 | revLookup[b64.charCodeAt(i + 2)] >> 2; |
|
arr[L++] = tmp >> 8 & 0xFF; |
|
arr[L++] = tmp & 0xFF; |
|
} |
|
|
|
return arr; |
|
} |
|
|
|
function tripletToBase64(num) { |
|
return lookup[num >> 18 & 0x3F] + lookup[num >> 12 & 0x3F] + lookup[num >> 6 & 0x3F] + lookup[num & 0x3F]; |
|
} |
|
|
|
function encodeChunk(uint8, start, end) { |
|
var tmp; |
|
var output = []; |
|
|
|
for (var i = start; i < end; i += 3) { |
|
tmp = (uint8[i] << 16) + (uint8[i + 1] << 8) + uint8[i + 2]; |
|
output.push(tripletToBase64(tmp)); |
|
} |
|
|
|
return output.join(''); |
|
} |
|
|
|
function fromByteArray(uint8) { |
|
if (!inited) { |
|
init(); |
|
} |
|
|
|
var tmp; |
|
var len = uint8.length; |
|
var extraBytes = len % 3; // if we have 1 byte left, pad 2 bytes |
|
|
|
var output = ''; |
|
var parts = []; |
|
var maxChunkLength = 16383; // must be multiple of 3 |
|
// go through the array every three bytes, we'll deal with trailing stuff later |
|
|
|
for (var i = 0, len2 = len - extraBytes; i < len2; i += maxChunkLength) { |
|
parts.push(encodeChunk(uint8, i, i + maxChunkLength > len2 ? len2 : i + maxChunkLength)); |
|
} // pad the end with zeros, but make sure to not forget the extra bytes |
|
|
|
|
|
if (extraBytes === 1) { |
|
tmp = uint8[len - 1]; |
|
output += lookup[tmp >> 2]; |
|
output += lookup[tmp << 4 & 0x3F]; |
|
output += '=='; |
|
} else if (extraBytes === 2) { |
|
tmp = (uint8[len - 2] << 8) + uint8[len - 1]; |
|
output += lookup[tmp >> 10]; |
|
output += lookup[tmp >> 4 & 0x3F]; |
|
output += lookup[tmp << 2 & 0x3F]; |
|
output += '='; |
|
} |
|
|
|
parts.push(output); |
|
return parts.join(''); |
|
} |
|
|
|
function read(buffer, offset, isLE, mLen, nBytes) { |
|
var e, m; |
|
var eLen = nBytes * 8 - mLen - 1; |
|
var eMax = (1 << eLen) - 1; |
|
var eBias = eMax >> 1; |
|
var nBits = -7; |
|
var i = isLE ? nBytes - 1 : 0; |
|
var d = isLE ? -1 : 1; |
|
var s = buffer[offset + i]; |
|
i += d; |
|
e = s & (1 << -nBits) - 1; |
|
s >>= -nBits; |
|
nBits += eLen; |
|
|
|
for (; nBits > 0; e = e * 256 + buffer[offset + i], i += d, nBits -= 8) {} |
|
|
|
m = e & (1 << -nBits) - 1; |
|
e >>= -nBits; |
|
nBits += mLen; |
|
|
|
for (; nBits > 0; m = m * 256 + buffer[offset + i], i += d, nBits -= 8) {} |
|
|
|
if (e === 0) { |
|
e = 1 - eBias; |
|
} else if (e === eMax) { |
|
return m ? NaN : (s ? -1 : 1) * Infinity; |
|
} else { |
|
m = m + Math.pow(2, mLen); |
|
e = e - eBias; |
|
} |
|
|
|
return (s ? -1 : 1) * m * Math.pow(2, e - mLen); |
|
} |
|
function write(buffer, value, offset, isLE, mLen, nBytes) { |
|
var e, m, c; |
|
var eLen = nBytes * 8 - mLen - 1; |
|
var eMax = (1 << eLen) - 1; |
|
var eBias = eMax >> 1; |
|
var rt = mLen === 23 ? Math.pow(2, -24) - Math.pow(2, -77) : 0; |
|
var i = isLE ? 0 : nBytes - 1; |
|
var d = isLE ? 1 : -1; |
|
var s = value < 0 || value === 0 && 1 / value < 0 ? 1 : 0; |
|
value = Math.abs(value); |
|
|
|
if (isNaN(value) || value === Infinity) { |
|
m = isNaN(value) ? 1 : 0; |
|
e = eMax; |
|
} else { |
|
e = Math.floor(Math.log(value) / Math.LN2); |
|
|
|
if (value * (c = Math.pow(2, -e)) < 1) { |
|
e--; |
|
c *= 2; |
|
} |
|
|
|
if (e + eBias >= 1) { |
|
value += rt / c; |
|
} else { |
|
value += rt * Math.pow(2, 1 - eBias); |
|
} |
|
|
|
if (value * c >= 2) { |
|
e++; |
|
c /= 2; |
|
} |
|
|
|
if (e + eBias >= eMax) { |
|
m = 0; |
|
e = eMax; |
|
} else if (e + eBias >= 1) { |
|
m = (value * c - 1) * Math.pow(2, mLen); |
|
e = e + eBias; |
|
} else { |
|
m = value * Math.pow(2, eBias - 1) * Math.pow(2, mLen); |
|
e = 0; |
|
} |
|
} |
|
|
|
for (; mLen >= 8; buffer[offset + i] = m & 0xff, i += d, m /= 256, mLen -= 8) {} |
|
|
|
e = e << mLen | m; |
|
eLen += mLen; |
|
|
|
for (; eLen > 0; buffer[offset + i] = e & 0xff, i += d, e /= 256, eLen -= 8) {} |
|
|
|
buffer[offset + i - d] |= s * 128; |
|
} |
|
|
|
var toString = {}.toString; |
|
var isArray = Array.isArray || function (arr) { |
|
return toString.call(arr) == '[object Array]'; |
|
}; |
|
|
|
var INSPECT_MAX_BYTES = 50; |
|
/** |
|
* If `Buffer.TYPED_ARRAY_SUPPORT`: |
|
* === true Use Uint8Array implementation (fastest) |
|
* === false Use Object implementation (most compatible, even IE6) |
|
* |
|
* Browsers that support typed arrays are IE 10+, Firefox 4+, Chrome 7+, Safari 5.1+, |
|
* Opera 11.6+, iOS 4.2+. |
|
* |
|
* Due to various browser bugs, sometimes the Object implementation will be used even |
|
* when the browser supports typed arrays. |
|
* |
|
* Note: |
|
* |
|
* - Firefox 4-29 lacks support for adding new properties to `Uint8Array` instances, |
|
* See: https://bugzilla.mozilla.org/show_bug.cgi?id=695438. |
|
* |
|
* - Chrome 9-10 is missing the `TypedArray.prototype.subarray` function. |
|
* |
|
* - IE10 has a broken `TypedArray.prototype.subarray` function which returns arrays of |
|
* incorrect length in some situations. |
|
|
|
* We detect these buggy browsers and set `Buffer.TYPED_ARRAY_SUPPORT` to `false` so they |
|
* get the Object implementation, which is slower but behaves correctly. |
|
*/ |
|
|
|
Buffer.TYPED_ARRAY_SUPPORT = global$1.TYPED_ARRAY_SUPPORT !== undefined ? global$1.TYPED_ARRAY_SUPPORT : true; |
|
|
|
function kMaxLength() { |
|
return Buffer.TYPED_ARRAY_SUPPORT ? 0x7fffffff : 0x3fffffff; |
|
} |
|
|
|
function createBuffer(that, length) { |
|
if (kMaxLength() < length) { |
|
throw new RangeError('Invalid typed array length'); |
|
} |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
// Return an augmented `Uint8Array` instance, for best performance |
|
that = new Uint8Array(length); |
|
that.__proto__ = Buffer.prototype; |
|
} else { |
|
// Fallback: Return an object instance of the Buffer class |
|
if (that === null) { |
|
that = new Buffer(length); |
|
} |
|
|
|
that.length = length; |
|
} |
|
|
|
return that; |
|
} |
|
/** |
|
* The Buffer constructor returns instances of `Uint8Array` that have their |
|
* prototype changed to `Buffer.prototype`. Furthermore, `Buffer` is a subclass of |
|
* `Uint8Array`, so the returned instances will have all the node `Buffer` methods |
|
* and the `Uint8Array` methods. Square bracket notation works as expected -- it |
|
* returns a single octet. |
|
* |
|
* The `Uint8Array` prototype remains unmodified. |
|
*/ |
|
|
|
|
|
function Buffer(arg, encodingOrOffset, length) { |
|
if (!Buffer.TYPED_ARRAY_SUPPORT && !(this instanceof Buffer)) { |
|
return new Buffer(arg, encodingOrOffset, length); |
|
} // Common case. |
|
|
|
|
|
if (typeof arg === 'number') { |
|
if (typeof encodingOrOffset === 'string') { |
|
throw new Error('If encoding is specified then the first argument must be a string'); |
|
} |
|
|
|
return allocUnsafe(this, arg); |
|
} |
|
|
|
return from(this, arg, encodingOrOffset, length); |
|
} |
|
Buffer.poolSize = 8192; // not used by this implementation |
|
// TODO: Legacy, not needed anymore. Remove in next major version. |
|
|
|
Buffer._augment = function (arr) { |
|
arr.__proto__ = Buffer.prototype; |
|
return arr; |
|
}; |
|
|
|
function from(that, value, encodingOrOffset, length) { |
|
if (typeof value === 'number') { |
|
throw new TypeError('"value" argument must not be a number'); |
|
} |
|
|
|
if (typeof ArrayBuffer !== 'undefined' && value instanceof ArrayBuffer) { |
|
return fromArrayBuffer(that, value, encodingOrOffset, length); |
|
} |
|
|
|
if (typeof value === 'string') { |
|
return fromString(that, value, encodingOrOffset); |
|
} |
|
|
|
return fromObject(that, value); |
|
} |
|
/** |
|
* Functionally equivalent to Buffer(arg, encoding) but throws a TypeError |
|
* if value is a number. |
|
* Buffer.from(str[, encoding]) |
|
* Buffer.from(array) |
|
* Buffer.from(buffer) |
|
* Buffer.from(arrayBuffer[, byteOffset[, length]]) |
|
**/ |
|
|
|
|
|
Buffer.from = function (value, encodingOrOffset, length) { |
|
return from(null, value, encodingOrOffset, length); |
|
}; |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
Buffer.prototype.__proto__ = Uint8Array.prototype; |
|
Buffer.__proto__ = Uint8Array; |
|
} |
|
|
|
function assertSize(size) { |
|
if (typeof size !== 'number') { |
|
throw new TypeError('"size" argument must be a number'); |
|
} else if (size < 0) { |
|
throw new RangeError('"size" argument must not be negative'); |
|
} |
|
} |
|
|
|
function alloc(that, size, fill, encoding) { |
|
assertSize(size); |
|
|
|
if (size <= 0) { |
|
return createBuffer(that, size); |
|
} |
|
|
|
if (fill !== undefined) { |
|
// Only pay attention to encoding if it's a string. This |
|
// prevents accidentally sending in a number that would |
|
// be interpretted as a start offset. |
|
return typeof encoding === 'string' ? createBuffer(that, size).fill(fill, encoding) : createBuffer(that, size).fill(fill); |
|
} |
|
|
|
return createBuffer(that, size); |
|
} |
|
/** |
|
* Creates a new filled Buffer instance. |
|
* alloc(size[, fill[, encoding]]) |
|
**/ |
|
|
|
|
|
Buffer.alloc = function (size, fill, encoding) { |
|
return alloc(null, size, fill, encoding); |
|
}; |
|
|
|
function allocUnsafe(that, size) { |
|
assertSize(size); |
|
that = createBuffer(that, size < 0 ? 0 : checked(size) | 0); |
|
|
|
if (!Buffer.TYPED_ARRAY_SUPPORT) { |
|
for (var i = 0; i < size; ++i) { |
|
that[i] = 0; |
|
} |
|
} |
|
|
|
return that; |
|
} |
|
/** |
|
* Equivalent to Buffer(num), by default creates a non-zero-filled Buffer instance. |
|
* */ |
|
|
|
|
|
Buffer.allocUnsafe = function (size) { |
|
return allocUnsafe(null, size); |
|
}; |
|
/** |
|
* Equivalent to SlowBuffer(num), by default creates a non-zero-filled Buffer instance. |
|
*/ |
|
|
|
|
|
Buffer.allocUnsafeSlow = function (size) { |
|
return allocUnsafe(null, size); |
|
}; |
|
|
|
function fromString(that, string, encoding) { |
|
if (typeof encoding !== 'string' || encoding === '') { |
|
encoding = 'utf8'; |
|
} |
|
|
|
if (!Buffer.isEncoding(encoding)) { |
|
throw new TypeError('"encoding" must be a valid string encoding'); |
|
} |
|
|
|
var length = byteLength(string, encoding) | 0; |
|
that = createBuffer(that, length); |
|
var actual = that.write(string, encoding); |
|
|
|
if (actual !== length) { |
|
// Writing a hex string, for example, that contains invalid characters will |
|
// cause everything after the first invalid character to be ignored. (e.g. |
|
// 'abxxcd' will be treated as 'ab') |
|
that = that.slice(0, actual); |
|
} |
|
|
|
return that; |
|
} |
|
|
|
function fromArrayLike(that, array) { |
|
var length = array.length < 0 ? 0 : checked(array.length) | 0; |
|
that = createBuffer(that, length); |
|
|
|
for (var i = 0; i < length; i += 1) { |
|
that[i] = array[i] & 255; |
|
} |
|
|
|
return that; |
|
} |
|
|
|
function fromArrayBuffer(that, array, byteOffset, length) { |
|
array.byteLength; // this throws if `array` is not a valid ArrayBuffer |
|
|
|
if (byteOffset < 0 || array.byteLength < byteOffset) { |
|
throw new RangeError('\'offset\' is out of bounds'); |
|
} |
|
|
|
if (array.byteLength < byteOffset + (length || 0)) { |
|
throw new RangeError('\'length\' is out of bounds'); |
|
} |
|
|
|
if (byteOffset === undefined && length === undefined) { |
|
array = new Uint8Array(array); |
|
} else if (length === undefined) { |
|
array = new Uint8Array(array, byteOffset); |
|
} else { |
|
array = new Uint8Array(array, byteOffset, length); |
|
} |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
// Return an augmented `Uint8Array` instance, for best performance |
|
that = array; |
|
that.__proto__ = Buffer.prototype; |
|
} else { |
|
// Fallback: Return an object instance of the Buffer class |
|
that = fromArrayLike(that, array); |
|
} |
|
|
|
return that; |
|
} |
|
|
|
function fromObject(that, obj) { |
|
if (internalIsBuffer(obj)) { |
|
var len = checked(obj.length) | 0; |
|
that = createBuffer(that, len); |
|
|
|
if (that.length === 0) { |
|
return that; |
|
} |
|
|
|
obj.copy(that, 0, 0, len); |
|
return that; |
|
} |
|
|
|
if (obj) { |
|
if (typeof ArrayBuffer !== 'undefined' && obj.buffer instanceof ArrayBuffer || 'length' in obj) { |
|
if (typeof obj.length !== 'number' || isnan(obj.length)) { |
|
return createBuffer(that, 0); |
|
} |
|
|
|
return fromArrayLike(that, obj); |
|
} |
|
|
|
if (obj.type === 'Buffer' && isArray(obj.data)) { |
|
return fromArrayLike(that, obj.data); |
|
} |
|
} |
|
|
|
throw new TypeError('First argument must be a string, Buffer, ArrayBuffer, Array, or array-like object.'); |
|
} |
|
|
|
function checked(length) { |
|
// Note: cannot use `length < kMaxLength()` here because that fails when |
|
// length is NaN (which is otherwise coerced to zero.) |
|
if (length >= kMaxLength()) { |
|
throw new RangeError('Attempt to allocate Buffer larger than maximum ' + 'size: 0x' + kMaxLength().toString(16) + ' bytes'); |
|
} |
|
|
|
return length | 0; |
|
} |
|
Buffer.isBuffer = isBuffer; |
|
|
|
function internalIsBuffer(b) { |
|
return !!(b != null && b._isBuffer); |
|
} |
|
|
|
Buffer.compare = function compare(a, b) { |
|
if (!internalIsBuffer(a) || !internalIsBuffer(b)) { |
|
throw new TypeError('Arguments must be Buffers'); |
|
} |
|
|
|
if (a === b) return 0; |
|
var x = a.length; |
|
var y = b.length; |
|
|
|
for (var i = 0, len = Math.min(x, y); i < len; ++i) { |
|
if (a[i] !== b[i]) { |
|
x = a[i]; |
|
y = b[i]; |
|
break; |
|
} |
|
} |
|
|
|
if (x < y) return -1; |
|
if (y < x) return 1; |
|
return 0; |
|
}; |
|
|
|
Buffer.isEncoding = function isEncoding(encoding) { |
|
switch (String(encoding).toLowerCase()) { |
|
case 'hex': |
|
case 'utf8': |
|
case 'utf-8': |
|
case 'ascii': |
|
case 'latin1': |
|
case 'binary': |
|
case 'base64': |
|
case 'ucs2': |
|
case 'ucs-2': |
|
case 'utf16le': |
|
case 'utf-16le': |
|
return true; |
|
|
|
default: |
|
return false; |
|
} |
|
}; |
|
|
|
Buffer.concat = function concat(list, length) { |
|
if (!isArray(list)) { |
|
throw new TypeError('"list" argument must be an Array of Buffers'); |
|
} |
|
|
|
if (list.length === 0) { |
|
return Buffer.alloc(0); |
|
} |
|
|
|
var i; |
|
|
|
if (length === undefined) { |
|
length = 0; |
|
|
|
for (i = 0; i < list.length; ++i) { |
|
length += list[i].length; |
|
} |
|
} |
|
|
|
var buffer = Buffer.allocUnsafe(length); |
|
var pos = 0; |
|
|
|
for (i = 0; i < list.length; ++i) { |
|
var buf = list[i]; |
|
|
|
if (!internalIsBuffer(buf)) { |
|
throw new TypeError('"list" argument must be an Array of Buffers'); |
|
} |
|
|
|
buf.copy(buffer, pos); |
|
pos += buf.length; |
|
} |
|
|
|
return buffer; |
|
}; |
|
|
|
function byteLength(string, encoding) { |
|
if (internalIsBuffer(string)) { |
|
return string.length; |
|
} |
|
|
|
if (typeof ArrayBuffer !== 'undefined' && typeof ArrayBuffer.isView === 'function' && (ArrayBuffer.isView(string) || string instanceof ArrayBuffer)) { |
|
return string.byteLength; |
|
} |
|
|
|
if (typeof string !== 'string') { |
|
string = '' + string; |
|
} |
|
|
|
var len = string.length; |
|
if (len === 0) return 0; // Use a for loop to avoid recursion |
|
|
|
var loweredCase = false; |
|
|
|
for (;;) { |
|
switch (encoding) { |
|
case 'ascii': |
|
case 'latin1': |
|
case 'binary': |
|
return len; |
|
|
|
case 'utf8': |
|
case 'utf-8': |
|
case undefined: |
|
return utf8ToBytes(string).length; |
|
|
|
case 'ucs2': |
|
case 'ucs-2': |
|
case 'utf16le': |
|
case 'utf-16le': |
|
return len * 2; |
|
|
|
case 'hex': |
|
return len >>> 1; |
|
|
|
case 'base64': |
|
return base64ToBytes(string).length; |
|
|
|
default: |
|
if (loweredCase) return utf8ToBytes(string).length; // assume utf8 |
|
|
|
encoding = ('' + encoding).toLowerCase(); |
|
loweredCase = true; |
|
} |
|
} |
|
} |
|
|
|
Buffer.byteLength = byteLength; |
|
|
|
function slowToString(encoding, start, end) { |
|
var loweredCase = false; // No need to verify that "this.length <= MAX_UINT32" since it's a read-only |
|
// property of a typed array. |
|
// This behaves neither like String nor Uint8Array in that we set start/end |
|
// to their upper/lower bounds if the value passed is out of range. |
|
// undefined is handled specially as per ECMA-262 6th Edition, |
|
// Section 13.3.3.7 Runtime Semantics: KeyedBindingInitialization. |
|
|
|
if (start === undefined || start < 0) { |
|
start = 0; |
|
} // Return early if start > this.length. Done here to prevent potential uint32 |
|
// coercion fail below. |
|
|
|
|
|
if (start > this.length) { |
|
return ''; |
|
} |
|
|
|
if (end === undefined || end > this.length) { |
|
end = this.length; |
|
} |
|
|
|
if (end <= 0) { |
|
return ''; |
|
} // Force coersion to uint32. This will also coerce falsey/NaN values to 0. |
|
|
|
|
|
end >>>= 0; |
|
start >>>= 0; |
|
|
|
if (end <= start) { |
|
return ''; |
|
} |
|
|
|
if (!encoding) encoding = 'utf8'; |
|
|
|
while (true) { |
|
switch (encoding) { |
|
case 'hex': |
|
return hexSlice(this, start, end); |
|
|
|
case 'utf8': |
|
case 'utf-8': |
|
return utf8Slice(this, start, end); |
|
|
|
case 'ascii': |
|
return asciiSlice(this, start, end); |
|
|
|
case 'latin1': |
|
case 'binary': |
|
return latin1Slice(this, start, end); |
|
|
|
case 'base64': |
|
return base64Slice(this, start, end); |
|
|
|
case 'ucs2': |
|
case 'ucs-2': |
|
case 'utf16le': |
|
case 'utf-16le': |
|
return utf16leSlice(this, start, end); |
|
|
|
default: |
|
if (loweredCase) throw new TypeError('Unknown encoding: ' + encoding); |
|
encoding = (encoding + '').toLowerCase(); |
|
loweredCase = true; |
|
} |
|
} |
|
} // The property is used by `Buffer.isBuffer` and `is-buffer` (in Safari 5-7) to detect |
|
// Buffer instances. |
|
|
|
|
|
Buffer.prototype._isBuffer = true; |
|
|
|
function swap(b, n, m) { |
|
var i = b[n]; |
|
b[n] = b[m]; |
|
b[m] = i; |
|
} |
|
|
|
Buffer.prototype.swap16 = function swap16() { |
|
var len = this.length; |
|
|
|
if (len % 2 !== 0) { |
|
throw new RangeError('Buffer size must be a multiple of 16-bits'); |
|
} |
|
|
|
for (var i = 0; i < len; i += 2) { |
|
swap(this, i, i + 1); |
|
} |
|
|
|
return this; |
|
}; |
|
|
|
Buffer.prototype.swap32 = function swap32() { |
|
var len = this.length; |
|
|
|
if (len % 4 !== 0) { |
|
throw new RangeError('Buffer size must be a multiple of 32-bits'); |
|
} |
|
|
|
for (var i = 0; i < len; i += 4) { |
|
swap(this, i, i + 3); |
|
swap(this, i + 1, i + 2); |
|
} |
|
|
|
return this; |
|
}; |
|
|
|
Buffer.prototype.swap64 = function swap64() { |
|
var len = this.length; |
|
|
|
if (len % 8 !== 0) { |
|
throw new RangeError('Buffer size must be a multiple of 64-bits'); |
|
} |
|
|
|
for (var i = 0; i < len; i += 8) { |
|
swap(this, i, i + 7); |
|
swap(this, i + 1, i + 6); |
|
swap(this, i + 2, i + 5); |
|
swap(this, i + 3, i + 4); |
|
} |
|
|
|
return this; |
|
}; |
|
|
|
Buffer.prototype.toString = function toString() { |
|
var length = this.length | 0; |
|
if (length === 0) return ''; |
|
if (arguments.length === 0) return utf8Slice(this, 0, length); |
|
return slowToString.apply(this, arguments); |
|
}; |
|
|
|
Buffer.prototype.equals = function equals(b) { |
|
if (!internalIsBuffer(b)) throw new TypeError('Argument must be a Buffer'); |
|
if (this === b) return true; |
|
return Buffer.compare(this, b) === 0; |
|
}; |
|
|
|
Buffer.prototype.inspect = function inspect() { |
|
var str = ''; |
|
var max = INSPECT_MAX_BYTES; |
|
|
|
if (this.length > 0) { |
|
str = this.toString('hex', 0, max).match(/.{2}/g).join(' '); |
|
if (this.length > max) str += ' ... '; |
|
} |
|
|
|
return '<Buffer ' + str + '>'; |
|
}; |
|
|
|
Buffer.prototype.compare = function compare(target, start, end, thisStart, thisEnd) { |
|
if (!internalIsBuffer(target)) { |
|
throw new TypeError('Argument must be a Buffer'); |
|
} |
|
|
|
if (start === undefined) { |
|
start = 0; |
|
} |
|
|
|
if (end === undefined) { |
|
end = target ? target.length : 0; |
|
} |
|
|
|
if (thisStart === undefined) { |
|
thisStart = 0; |
|
} |
|
|
|
if (thisEnd === undefined) { |
|
thisEnd = this.length; |
|
} |
|
|
|
if (start < 0 || end > target.length || thisStart < 0 || thisEnd > this.length) { |
|
throw new RangeError('out of range index'); |
|
} |
|
|
|
if (thisStart >= thisEnd && start >= end) { |
|
return 0; |
|
} |
|
|
|
if (thisStart >= thisEnd) { |
|
return -1; |
|
} |
|
|
|
if (start >= end) { |
|
return 1; |
|
} |
|
|
|
start >>>= 0; |
|
end >>>= 0; |
|
thisStart >>>= 0; |
|
thisEnd >>>= 0; |
|
if (this === target) return 0; |
|
var x = thisEnd - thisStart; |
|
var y = end - start; |
|
var len = Math.min(x, y); |
|
var thisCopy = this.slice(thisStart, thisEnd); |
|
var targetCopy = target.slice(start, end); |
|
|
|
for (var i = 0; i < len; ++i) { |
|
if (thisCopy[i] !== targetCopy[i]) { |
|
x = thisCopy[i]; |
|
y = targetCopy[i]; |
|
break; |
|
} |
|
} |
|
|
|
if (x < y) return -1; |
|
if (y < x) return 1; |
|
return 0; |
|
}; // Finds either the first index of `val` in `buffer` at offset >= `byteOffset`, |
|
// OR the last index of `val` in `buffer` at offset <= `byteOffset`. |
|
// |
|
// Arguments: |
|
// - buffer - a Buffer to search |
|
// - val - a string, Buffer, or number |
|
// - byteOffset - an index into `buffer`; will be clamped to an int32 |
|
// - encoding - an optional encoding, relevant is val is a string |
|
// - dir - true for indexOf, false for lastIndexOf |
|
|
|
|
|
function bidirectionalIndexOf(buffer, val, byteOffset, encoding, dir) { |
|
// Empty buffer means no match |
|
if (buffer.length === 0) return -1; // Normalize byteOffset |
|
|
|
if (typeof byteOffset === 'string') { |
|
encoding = byteOffset; |
|
byteOffset = 0; |
|
} else if (byteOffset > 0x7fffffff) { |
|
byteOffset = 0x7fffffff; |
|
} else if (byteOffset < -0x80000000) { |
|
byteOffset = -0x80000000; |
|
} |
|
|
|
byteOffset = +byteOffset; // Coerce to Number. |
|
|
|
if (isNaN(byteOffset)) { |
|
// byteOffset: it it's undefined, null, NaN, "foo", etc, search whole buffer |
|
byteOffset = dir ? 0 : buffer.length - 1; |
|
} // Normalize byteOffset: negative offsets start from the end of the buffer |
|
|
|
|
|
if (byteOffset < 0) byteOffset = buffer.length + byteOffset; |
|
|
|
if (byteOffset >= buffer.length) { |
|
if (dir) return -1;else byteOffset = buffer.length - 1; |
|
} else if (byteOffset < 0) { |
|
if (dir) byteOffset = 0;else return -1; |
|
} // Normalize val |
|
|
|
|
|
if (typeof val === 'string') { |
|
val = Buffer.from(val, encoding); |
|
} // Finally, search either indexOf (if dir is true) or lastIndexOf |
|
|
|
|
|
if (internalIsBuffer(val)) { |
|
// Special case: looking for empty string/buffer always fails |
|
if (val.length === 0) { |
|
return -1; |
|
} |
|
|
|
return arrayIndexOf(buffer, val, byteOffset, encoding, dir); |
|
} else if (typeof val === 'number') { |
|
val = val & 0xFF; // Search for a byte value [0-255] |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT && typeof Uint8Array.prototype.indexOf === 'function') { |
|
if (dir) { |
|
return Uint8Array.prototype.indexOf.call(buffer, val, byteOffset); |
|
} else { |
|
return Uint8Array.prototype.lastIndexOf.call(buffer, val, byteOffset); |
|
} |
|
} |
|
|
|
return arrayIndexOf(buffer, [val], byteOffset, encoding, dir); |
|
} |
|
|
|
throw new TypeError('val must be string, number or Buffer'); |
|
} |
|
|
|
function arrayIndexOf(arr, val, byteOffset, encoding, dir) { |
|
var indexSize = 1; |
|
var arrLength = arr.length; |
|
var valLength = val.length; |
|
|
|
if (encoding !== undefined) { |
|
encoding = String(encoding).toLowerCase(); |
|
|
|
if (encoding === 'ucs2' || encoding === 'ucs-2' || encoding === 'utf16le' || encoding === 'utf-16le') { |
|
if (arr.length < 2 || val.length < 2) { |
|
return -1; |
|
} |
|
|
|
indexSize = 2; |
|
arrLength /= 2; |
|
valLength /= 2; |
|
byteOffset /= 2; |
|
} |
|
} |
|
|
|
function read(buf, i) { |
|
if (indexSize === 1) { |
|
return buf[i]; |
|
} else { |
|
return buf.readUInt16BE(i * indexSize); |
|
} |
|
} |
|
|
|
var i; |
|
|
|
if (dir) { |
|
var foundIndex = -1; |
|
|
|
for (i = byteOffset; i < arrLength; i++) { |
|
if (read(arr, i) === read(val, foundIndex === -1 ? 0 : i - foundIndex)) { |
|
if (foundIndex === -1) foundIndex = i; |
|
if (i - foundIndex + 1 === valLength) return foundIndex * indexSize; |
|
} else { |
|
if (foundIndex !== -1) i -= i - foundIndex; |
|
foundIndex = -1; |
|
} |
|
} |
|
} else { |
|
if (byteOffset + valLength > arrLength) byteOffset = arrLength - valLength; |
|
|
|
for (i = byteOffset; i >= 0; i--) { |
|
var found = true; |
|
|
|
for (var j = 0; j < valLength; j++) { |
|
if (read(arr, i + j) !== read(val, j)) { |
|
found = false; |
|
break; |
|
} |
|
} |
|
|
|
if (found) return i; |
|
} |
|
} |
|
|
|
return -1; |
|
} |
|
|
|
Buffer.prototype.includes = function includes(val, byteOffset, encoding) { |
|
return this.indexOf(val, byteOffset, encoding) !== -1; |
|
}; |
|
|
|
Buffer.prototype.indexOf = function indexOf(val, byteOffset, encoding) { |
|
return bidirectionalIndexOf(this, val, byteOffset, encoding, true); |
|
}; |
|
|
|
Buffer.prototype.lastIndexOf = function lastIndexOf(val, byteOffset, encoding) { |
|
return bidirectionalIndexOf(this, val, byteOffset, encoding, false); |
|
}; |
|
|
|
function hexWrite(buf, string, offset, length) { |
|
offset = Number(offset) || 0; |
|
var remaining = buf.length - offset; |
|
|
|
if (!length) { |
|
length = remaining; |
|
} else { |
|
length = Number(length); |
|
|
|
if (length > remaining) { |
|
length = remaining; |
|
} |
|
} // must be an even number of digits |
|
|
|
|
|
var strLen = string.length; |
|
if (strLen % 2 !== 0) throw new TypeError('Invalid hex string'); |
|
|
|
if (length > strLen / 2) { |
|
length = strLen / 2; |
|
} |
|
|
|
for (var i = 0; i < length; ++i) { |
|
var parsed = parseInt(string.substr(i * 2, 2), 16); |
|
if (isNaN(parsed)) return i; |
|
buf[offset + i] = parsed; |
|
} |
|
|
|
return i; |
|
} |
|
|
|
function utf8Write(buf, string, offset, length) { |
|
return blitBuffer(utf8ToBytes(string, buf.length - offset), buf, offset, length); |
|
} |
|
|
|
function asciiWrite(buf, string, offset, length) { |
|
return blitBuffer(asciiToBytes(string), buf, offset, length); |
|
} |
|
|
|
function latin1Write(buf, string, offset, length) { |
|
return asciiWrite(buf, string, offset, length); |
|
} |
|
|
|
function base64Write(buf, string, offset, length) { |
|
return blitBuffer(base64ToBytes(string), buf, offset, length); |
|
} |
|
|
|
function ucs2Write(buf, string, offset, length) { |
|
return blitBuffer(utf16leToBytes(string, buf.length - offset), buf, offset, length); |
|
} |
|
|
|
Buffer.prototype.write = function write(string, offset, length, encoding) { |
|
// Buffer#write(string) |
|
if (offset === undefined) { |
|
encoding = 'utf8'; |
|
length = this.length; |
|
offset = 0; // Buffer#write(string, encoding) |
|
} else if (length === undefined && typeof offset === 'string') { |
|
encoding = offset; |
|
length = this.length; |
|
offset = 0; // Buffer#write(string, offset[, length][, encoding]) |
|
} else if (isFinite(offset)) { |
|
offset = offset | 0; |
|
|
|
if (isFinite(length)) { |
|
length = length | 0; |
|
if (encoding === undefined) encoding = 'utf8'; |
|
} else { |
|
encoding = length; |
|
length = undefined; |
|
} // legacy write(string, encoding, offset, length) - remove in v0.13 |
|
|
|
} else { |
|
throw new Error('Buffer.write(string, encoding, offset[, length]) is no longer supported'); |
|
} |
|
|
|
var remaining = this.length - offset; |
|
if (length === undefined || length > remaining) length = remaining; |
|
|
|
if (string.length > 0 && (length < 0 || offset < 0) || offset > this.length) { |
|
throw new RangeError('Attempt to write outside buffer bounds'); |
|
} |
|
|
|
if (!encoding) encoding = 'utf8'; |
|
var loweredCase = false; |
|
|
|
for (;;) { |
|
switch (encoding) { |
|
case 'hex': |
|
return hexWrite(this, string, offset, length); |
|
|
|
case 'utf8': |
|
case 'utf-8': |
|
return utf8Write(this, string, offset, length); |
|
|
|
case 'ascii': |
|
return asciiWrite(this, string, offset, length); |
|
|
|
case 'latin1': |
|
case 'binary': |
|
return latin1Write(this, string, offset, length); |
|
|
|
case 'base64': |
|
// Warning: maxLength not taken into account in base64Write |
|
return base64Write(this, string, offset, length); |
|
|
|
case 'ucs2': |
|
case 'ucs-2': |
|
case 'utf16le': |
|
case 'utf-16le': |
|
return ucs2Write(this, string, offset, length); |
|
|
|
default: |
|
if (loweredCase) throw new TypeError('Unknown encoding: ' + encoding); |
|
encoding = ('' + encoding).toLowerCase(); |
|
loweredCase = true; |
|
} |
|
} |
|
}; |
|
|
|
Buffer.prototype.toJSON = function toJSON() { |
|
return { |
|
type: 'Buffer', |
|
data: Array.prototype.slice.call(this._arr || this, 0) |
|
}; |
|
}; |
|
|
|
function base64Slice(buf, start, end) { |
|
if (start === 0 && end === buf.length) { |
|
return fromByteArray(buf); |
|
} else { |
|
return fromByteArray(buf.slice(start, end)); |
|
} |
|
} |
|
|
|
function utf8Slice(buf, start, end) { |
|
end = Math.min(buf.length, end); |
|
var res = []; |
|
var i = start; |
|
|
|
while (i < end) { |
|
var firstByte = buf[i]; |
|
var codePoint = null; |
|
var bytesPerSequence = firstByte > 0xEF ? 4 : firstByte > 0xDF ? 3 : firstByte > 0xBF ? 2 : 1; |
|
|
|
if (i + bytesPerSequence <= end) { |
|
var secondByte, thirdByte, fourthByte, tempCodePoint; |
|
|
|
switch (bytesPerSequence) { |
|
case 1: |
|
if (firstByte < 0x80) { |
|
codePoint = firstByte; |
|
} |
|
|
|
break; |
|
|
|
case 2: |
|
secondByte = buf[i + 1]; |
|
|
|
if ((secondByte & 0xC0) === 0x80) { |
|
tempCodePoint = (firstByte & 0x1F) << 0x6 | secondByte & 0x3F; |
|
|
|
if (tempCodePoint > 0x7F) { |
|
codePoint = tempCodePoint; |
|
} |
|
} |
|
|
|
break; |
|
|
|
case 3: |
|
secondByte = buf[i + 1]; |
|
thirdByte = buf[i + 2]; |
|
|
|
if ((secondByte & 0xC0) === 0x80 && (thirdByte & 0xC0) === 0x80) { |
|
tempCodePoint = (firstByte & 0xF) << 0xC | (secondByte & 0x3F) << 0x6 | thirdByte & 0x3F; |
|
|
|
if (tempCodePoint > 0x7FF && (tempCodePoint < 0xD800 || tempCodePoint > 0xDFFF)) { |
|
codePoint = tempCodePoint; |
|
} |
|
} |
|
|
|
break; |
|
|
|
case 4: |
|
secondByte = buf[i + 1]; |
|
thirdByte = buf[i + 2]; |
|
fourthByte = buf[i + 3]; |
|
|
|
if ((secondByte & 0xC0) === 0x80 && (thirdByte & 0xC0) === 0x80 && (fourthByte & 0xC0) === 0x80) { |
|
tempCodePoint = (firstByte & 0xF) << 0x12 | (secondByte & 0x3F) << 0xC | (thirdByte & 0x3F) << 0x6 | fourthByte & 0x3F; |
|
|
|
if (tempCodePoint > 0xFFFF && tempCodePoint < 0x110000) { |
|
codePoint = tempCodePoint; |
|
} |
|
} |
|
|
|
} |
|
} |
|
|
|
if (codePoint === null) { |
|
// we did not generate a valid codePoint so insert a |
|
// replacement char (U+FFFD) and advance only 1 byte |
|
codePoint = 0xFFFD; |
|
bytesPerSequence = 1; |
|
} else if (codePoint > 0xFFFF) { |
|
// encode to utf16 (surrogate pair dance) |
|
codePoint -= 0x10000; |
|
res.push(codePoint >>> 10 & 0x3FF | 0xD800); |
|
codePoint = 0xDC00 | codePoint & 0x3FF; |
|
} |
|
|
|
res.push(codePoint); |
|
i += bytesPerSequence; |
|
} |
|
|
|
return decodeCodePointsArray(res); |
|
} // Based on http://stackoverflow.com/a/22747272/680742, the browser with |
|
// the lowest limit is Chrome, with 0x10000 args. |
|
// We go 1 magnitude less, for safety |
|
|
|
|
|
var MAX_ARGUMENTS_LENGTH = 0x1000; |
|
|
|
function decodeCodePointsArray(codePoints) { |
|
var len = codePoints.length; |
|
|
|
if (len <= MAX_ARGUMENTS_LENGTH) { |
|
return String.fromCharCode.apply(String, codePoints); // avoid extra slice() |
|
} // Decode in chunks to avoid "call stack size exceeded". |
|
|
|
|
|
var res = ''; |
|
var i = 0; |
|
|
|
while (i < len) { |
|
res += String.fromCharCode.apply(String, codePoints.slice(i, i += MAX_ARGUMENTS_LENGTH)); |
|
} |
|
|
|
return res; |
|
} |
|
|
|
function asciiSlice(buf, start, end) { |
|
var ret = ''; |
|
end = Math.min(buf.length, end); |
|
|
|
for (var i = start; i < end; ++i) { |
|
ret += String.fromCharCode(buf[i] & 0x7F); |
|
} |
|
|
|
return ret; |
|
} |
|
|
|
function latin1Slice(buf, start, end) { |
|
var ret = ''; |
|
end = Math.min(buf.length, end); |
|
|
|
for (var i = start; i < end; ++i) { |
|
ret += String.fromCharCode(buf[i]); |
|
} |
|
|
|
return ret; |
|
} |
|
|
|
function hexSlice(buf, start, end) { |
|
var len = buf.length; |
|
if (!start || start < 0) start = 0; |
|
if (!end || end < 0 || end > len) end = len; |
|
var out = ''; |
|
|
|
for (var i = start; i < end; ++i) { |
|
out += toHex(buf[i]); |
|
} |
|
|
|
return out; |
|
} |
|
|
|
function utf16leSlice(buf, start, end) { |
|
var bytes = buf.slice(start, end); |
|
var res = ''; |
|
|
|
for (var i = 0; i < bytes.length; i += 2) { |
|
res += String.fromCharCode(bytes[i] + bytes[i + 1] * 256); |
|
} |
|
|
|
return res; |
|
} |
|
|
|
Buffer.prototype.slice = function slice(start, end) { |
|
var len = this.length; |
|
start = ~~start; |
|
end = end === undefined ? len : ~~end; |
|
|
|
if (start < 0) { |
|
start += len; |
|
if (start < 0) start = 0; |
|
} else if (start > len) { |
|
start = len; |
|
} |
|
|
|
if (end < 0) { |
|
end += len; |
|
if (end < 0) end = 0; |
|
} else if (end > len) { |
|
end = len; |
|
} |
|
|
|
if (end < start) end = start; |
|
var newBuf; |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
newBuf = this.subarray(start, end); |
|
newBuf.__proto__ = Buffer.prototype; |
|
} else { |
|
var sliceLen = end - start; |
|
newBuf = new Buffer(sliceLen, undefined); |
|
|
|
for (var i = 0; i < sliceLen; ++i) { |
|
newBuf[i] = this[i + start]; |
|
} |
|
} |
|
|
|
return newBuf; |
|
}; |
|
/* |
|
* Need to make sure that buffer isn't trying to write out of bounds. |
|
*/ |
|
|
|
|
|
function checkOffset(offset, ext, length) { |
|
if (offset % 1 !== 0 || offset < 0) throw new RangeError('offset is not uint'); |
|
if (offset + ext > length) throw new RangeError('Trying to access beyond buffer length'); |
|
} |
|
|
|
Buffer.prototype.readUIntLE = function readUIntLE(offset, byteLength, noAssert) { |
|
offset = offset | 0; |
|
byteLength = byteLength | 0; |
|
if (!noAssert) checkOffset(offset, byteLength, this.length); |
|
var val = this[offset]; |
|
var mul = 1; |
|
var i = 0; |
|
|
|
while (++i < byteLength && (mul *= 0x100)) { |
|
val += this[offset + i] * mul; |
|
} |
|
|
|
return val; |
|
}; |
|
|
|
Buffer.prototype.readUIntBE = function readUIntBE(offset, byteLength, noAssert) { |
|
offset = offset | 0; |
|
byteLength = byteLength | 0; |
|
|
|
if (!noAssert) { |
|
checkOffset(offset, byteLength, this.length); |
|
} |
|
|
|
var val = this[offset + --byteLength]; |
|
var mul = 1; |
|
|
|
while (byteLength > 0 && (mul *= 0x100)) { |
|
val += this[offset + --byteLength] * mul; |
|
} |
|
|
|
return val; |
|
}; |
|
|
|
Buffer.prototype.readUInt8 = function readUInt8(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 1, this.length); |
|
return this[offset]; |
|
}; |
|
|
|
Buffer.prototype.readUInt16LE = function readUInt16LE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 2, this.length); |
|
return this[offset] | this[offset + 1] << 8; |
|
}; |
|
|
|
Buffer.prototype.readUInt16BE = function readUInt16BE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 2, this.length); |
|
return this[offset] << 8 | this[offset + 1]; |
|
}; |
|
|
|
Buffer.prototype.readUInt32LE = function readUInt32LE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 4, this.length); |
|
return (this[offset] | this[offset + 1] << 8 | this[offset + 2] << 16) + this[offset + 3] * 0x1000000; |
|
}; |
|
|
|
Buffer.prototype.readUInt32BE = function readUInt32BE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 4, this.length); |
|
return this[offset] * 0x1000000 + (this[offset + 1] << 16 | this[offset + 2] << 8 | this[offset + 3]); |
|
}; |
|
|
|
Buffer.prototype.readIntLE = function readIntLE(offset, byteLength, noAssert) { |
|
offset = offset | 0; |
|
byteLength = byteLength | 0; |
|
if (!noAssert) checkOffset(offset, byteLength, this.length); |
|
var val = this[offset]; |
|
var mul = 1; |
|
var i = 0; |
|
|
|
while (++i < byteLength && (mul *= 0x100)) { |
|
val += this[offset + i] * mul; |
|
} |
|
|
|
mul *= 0x80; |
|
if (val >= mul) val -= Math.pow(2, 8 * byteLength); |
|
return val; |
|
}; |
|
|
|
Buffer.prototype.readIntBE = function readIntBE(offset, byteLength, noAssert) { |
|
offset = offset | 0; |
|
byteLength = byteLength | 0; |
|
if (!noAssert) checkOffset(offset, byteLength, this.length); |
|
var i = byteLength; |
|
var mul = 1; |
|
var val = this[offset + --i]; |
|
|
|
while (i > 0 && (mul *= 0x100)) { |
|
val += this[offset + --i] * mul; |
|
} |
|
|
|
mul *= 0x80; |
|
if (val >= mul) val -= Math.pow(2, 8 * byteLength); |
|
return val; |
|
}; |
|
|
|
Buffer.prototype.readInt8 = function readInt8(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 1, this.length); |
|
if (!(this[offset] & 0x80)) return this[offset]; |
|
return (0xff - this[offset] + 1) * -1; |
|
}; |
|
|
|
Buffer.prototype.readInt16LE = function readInt16LE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 2, this.length); |
|
var val = this[offset] | this[offset + 1] << 8; |
|
return val & 0x8000 ? val | 0xFFFF0000 : val; |
|
}; |
|
|
|
Buffer.prototype.readInt16BE = function readInt16BE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 2, this.length); |
|
var val = this[offset + 1] | this[offset] << 8; |
|
return val & 0x8000 ? val | 0xFFFF0000 : val; |
|
}; |
|
|
|
Buffer.prototype.readInt32LE = function readInt32LE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 4, this.length); |
|
return this[offset] | this[offset + 1] << 8 | this[offset + 2] << 16 | this[offset + 3] << 24; |
|
}; |
|
|
|
Buffer.prototype.readInt32BE = function readInt32BE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 4, this.length); |
|
return this[offset] << 24 | this[offset + 1] << 16 | this[offset + 2] << 8 | this[offset + 3]; |
|
}; |
|
|
|
Buffer.prototype.readFloatLE = function readFloatLE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 4, this.length); |
|
return read(this, offset, true, 23, 4); |
|
}; |
|
|
|
Buffer.prototype.readFloatBE = function readFloatBE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 4, this.length); |
|
return read(this, offset, false, 23, 4); |
|
}; |
|
|
|
Buffer.prototype.readDoubleLE = function readDoubleLE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 8, this.length); |
|
return read(this, offset, true, 52, 8); |
|
}; |
|
|
|
Buffer.prototype.readDoubleBE = function readDoubleBE(offset, noAssert) { |
|
if (!noAssert) checkOffset(offset, 8, this.length); |
|
return read(this, offset, false, 52, 8); |
|
}; |
|
|
|
function checkInt(buf, value, offset, ext, max, min) { |
|
if (!internalIsBuffer(buf)) throw new TypeError('"buffer" argument must be a Buffer instance'); |
|
if (value > max || value < min) throw new RangeError('"value" argument is out of bounds'); |
|
if (offset + ext > buf.length) throw new RangeError('Index out of range'); |
|
} |
|
|
|
Buffer.prototype.writeUIntLE = function writeUIntLE(value, offset, byteLength, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
byteLength = byteLength | 0; |
|
|
|
if (!noAssert) { |
|
var maxBytes = Math.pow(2, 8 * byteLength) - 1; |
|
checkInt(this, value, offset, byteLength, maxBytes, 0); |
|
} |
|
|
|
var mul = 1; |
|
var i = 0; |
|
this[offset] = value & 0xFF; |
|
|
|
while (++i < byteLength && (mul *= 0x100)) { |
|
this[offset + i] = value / mul & 0xFF; |
|
} |
|
|
|
return offset + byteLength; |
|
}; |
|
|
|
Buffer.prototype.writeUIntBE = function writeUIntBE(value, offset, byteLength, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
byteLength = byteLength | 0; |
|
|
|
if (!noAssert) { |
|
var maxBytes = Math.pow(2, 8 * byteLength) - 1; |
|
checkInt(this, value, offset, byteLength, maxBytes, 0); |
|
} |
|
|
|
var i = byteLength - 1; |
|
var mul = 1; |
|
this[offset + i] = value & 0xFF; |
|
|
|
while (--i >= 0 && (mul *= 0x100)) { |
|
this[offset + i] = value / mul & 0xFF; |
|
} |
|
|
|
return offset + byteLength; |
|
}; |
|
|
|
Buffer.prototype.writeUInt8 = function writeUInt8(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 1, 0xff, 0); |
|
if (!Buffer.TYPED_ARRAY_SUPPORT) value = Math.floor(value); |
|
this[offset] = value & 0xff; |
|
return offset + 1; |
|
}; |
|
|
|
function objectWriteUInt16(buf, value, offset, littleEndian) { |
|
if (value < 0) value = 0xffff + value + 1; |
|
|
|
for (var i = 0, j = Math.min(buf.length - offset, 2); i < j; ++i) { |
|
buf[offset + i] = (value & 0xff << 8 * (littleEndian ? i : 1 - i)) >>> (littleEndian ? i : 1 - i) * 8; |
|
} |
|
} |
|
|
|
Buffer.prototype.writeUInt16LE = function writeUInt16LE(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 2, 0xffff, 0); |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
this[offset] = value & 0xff; |
|
this[offset + 1] = value >>> 8; |
|
} else { |
|
objectWriteUInt16(this, value, offset, true); |
|
} |
|
|
|
return offset + 2; |
|
}; |
|
|
|
Buffer.prototype.writeUInt16BE = function writeUInt16BE(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 2, 0xffff, 0); |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
this[offset] = value >>> 8; |
|
this[offset + 1] = value & 0xff; |
|
} else { |
|
objectWriteUInt16(this, value, offset, false); |
|
} |
|
|
|
return offset + 2; |
|
}; |
|
|
|
function objectWriteUInt32(buf, value, offset, littleEndian) { |
|
if (value < 0) value = 0xffffffff + value + 1; |
|
|
|
for (var i = 0, j = Math.min(buf.length - offset, 4); i < j; ++i) { |
|
buf[offset + i] = value >>> (littleEndian ? i : 3 - i) * 8 & 0xff; |
|
} |
|
} |
|
|
|
Buffer.prototype.writeUInt32LE = function writeUInt32LE(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 4, 0xffffffff, 0); |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
this[offset + 3] = value >>> 24; |
|
this[offset + 2] = value >>> 16; |
|
this[offset + 1] = value >>> 8; |
|
this[offset] = value & 0xff; |
|
} else { |
|
objectWriteUInt32(this, value, offset, true); |
|
} |
|
|
|
return offset + 4; |
|
}; |
|
|
|
Buffer.prototype.writeUInt32BE = function writeUInt32BE(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 4, 0xffffffff, 0); |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
this[offset] = value >>> 24; |
|
this[offset + 1] = value >>> 16; |
|
this[offset + 2] = value >>> 8; |
|
this[offset + 3] = value & 0xff; |
|
} else { |
|
objectWriteUInt32(this, value, offset, false); |
|
} |
|
|
|
return offset + 4; |
|
}; |
|
|
|
Buffer.prototype.writeIntLE = function writeIntLE(value, offset, byteLength, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
|
|
if (!noAssert) { |
|
var limit = Math.pow(2, 8 * byteLength - 1); |
|
checkInt(this, value, offset, byteLength, limit - 1, -limit); |
|
} |
|
|
|
var i = 0; |
|
var mul = 1; |
|
var sub = 0; |
|
this[offset] = value & 0xFF; |
|
|
|
while (++i < byteLength && (mul *= 0x100)) { |
|
if (value < 0 && sub === 0 && this[offset + i - 1] !== 0) { |
|
sub = 1; |
|
} |
|
|
|
this[offset + i] = (value / mul >> 0) - sub & 0xFF; |
|
} |
|
|
|
return offset + byteLength; |
|
}; |
|
|
|
Buffer.prototype.writeIntBE = function writeIntBE(value, offset, byteLength, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
|
|
if (!noAssert) { |
|
var limit = Math.pow(2, 8 * byteLength - 1); |
|
checkInt(this, value, offset, byteLength, limit - 1, -limit); |
|
} |
|
|
|
var i = byteLength - 1; |
|
var mul = 1; |
|
var sub = 0; |
|
this[offset + i] = value & 0xFF; |
|
|
|
while (--i >= 0 && (mul *= 0x100)) { |
|
if (value < 0 && sub === 0 && this[offset + i + 1] !== 0) { |
|
sub = 1; |
|
} |
|
|
|
this[offset + i] = (value / mul >> 0) - sub & 0xFF; |
|
} |
|
|
|
return offset + byteLength; |
|
}; |
|
|
|
Buffer.prototype.writeInt8 = function writeInt8(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 1, 0x7f, -0x80); |
|
if (!Buffer.TYPED_ARRAY_SUPPORT) value = Math.floor(value); |
|
if (value < 0) value = 0xff + value + 1; |
|
this[offset] = value & 0xff; |
|
return offset + 1; |
|
}; |
|
|
|
Buffer.prototype.writeInt16LE = function writeInt16LE(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 2, 0x7fff, -0x8000); |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
this[offset] = value & 0xff; |
|
this[offset + 1] = value >>> 8; |
|
} else { |
|
objectWriteUInt16(this, value, offset, true); |
|
} |
|
|
|
return offset + 2; |
|
}; |
|
|
|
Buffer.prototype.writeInt16BE = function writeInt16BE(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 2, 0x7fff, -0x8000); |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
this[offset] = value >>> 8; |
|
this[offset + 1] = value & 0xff; |
|
} else { |
|
objectWriteUInt16(this, value, offset, false); |
|
} |
|
|
|
return offset + 2; |
|
}; |
|
|
|
Buffer.prototype.writeInt32LE = function writeInt32LE(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 4, 0x7fffffff, -0x80000000); |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
this[offset] = value & 0xff; |
|
this[offset + 1] = value >>> 8; |
|
this[offset + 2] = value >>> 16; |
|
this[offset + 3] = value >>> 24; |
|
} else { |
|
objectWriteUInt32(this, value, offset, true); |
|
} |
|
|
|
return offset + 4; |
|
}; |
|
|
|
Buffer.prototype.writeInt32BE = function writeInt32BE(value, offset, noAssert) { |
|
value = +value; |
|
offset = offset | 0; |
|
if (!noAssert) checkInt(this, value, offset, 4, 0x7fffffff, -0x80000000); |
|
if (value < 0) value = 0xffffffff + value + 1; |
|
|
|
if (Buffer.TYPED_ARRAY_SUPPORT) { |
|
this[offset] = value >>> 24; |
|
this[offset + 1] = value >>> 16; |
|
this[offset + 2] = value >>> 8; |
|
this[offset + 3] = value & 0xff; |
|
} else { |
|
objectWriteUInt32(this, value, offset, false); |
|
} |
|
|
|
return offset + 4; |
|
}; |
|
|
|
function checkIEEE754(buf, value, offset, ext, max, min) { |
|
if (offset + ext > buf.length) throw new RangeError('Index out of range'); |
|
if (offset < 0) throw new RangeError('Index out of range'); |
|
} |
|
|
|
function writeFloat(buf, value, offset, littleEndian, noAssert) { |
|
if (!noAssert) { |
|
checkIEEE754(buf, value, offset, 4); |
|
} |
|
|
|
write(buf, value, offset, littleEndian, 23, 4); |
|
return offset + 4; |
|
} |
|
|
|
Buffer.prototype.writeFloatLE = function writeFloatLE(value, offset, noAssert) { |
|
return writeFloat(this, value, offset, true, noAssert); |
|
}; |
|
|
|
Buffer.prototype.writeFloatBE = function writeFloatBE(value, offset, noAssert) { |
|
return writeFloat(this, value, offset, false, noAssert); |
|
}; |
|
|
|
function writeDouble(buf, value, offset, littleEndian, noAssert) { |
|
if (!noAssert) { |
|
checkIEEE754(buf, value, offset, 8); |
|
} |
|
|
|
write(buf, value, offset, littleEndian, 52, 8); |
|
return offset + 8; |
|
} |
|
|
|
Buffer.prototype.writeDoubleLE = function writeDoubleLE(value, offset, noAssert) { |
|
return writeDouble(this, value, offset, true, noAssert); |
|
}; |
|
|
|
Buffer.prototype.writeDoubleBE = function writeDoubleBE(value, offset, noAssert) { |
|
return writeDouble(this, value, offset, false, noAssert); |
|
}; // copy(targetBuffer, targetStart=0, sourceStart=0, sourceEnd=buffer.length) |
|
|
|
|
|
Buffer.prototype.copy = function copy(target, targetStart, start, end) { |
|
if (!start) start = 0; |
|
if (!end && end !== 0) end = this.length; |
|
if (targetStart >= target.length) targetStart = target.length; |
|
if (!targetStart) targetStart = 0; |
|
if (end > 0 && end < start) end = start; // Copy 0 bytes; we're done |
|
|
|
if (end === start) return 0; |
|
if (target.length === 0 || this.length === 0) return 0; // Fatal error conditions |
|
|
|
if (targetStart < 0) { |
|
throw new RangeError('targetStart out of bounds'); |
|
} |
|
|
|
if (start < 0 || start >= this.length) throw new RangeError('sourceStart out of bounds'); |
|
if (end < 0) throw new RangeError('sourceEnd out of bounds'); // Are we oob? |
|
|
|
if (end > this.length) end = this.length; |
|
|
|
if (target.length - targetStart < end - start) { |
|
end = target.length - targetStart + start; |
|
} |
|
|
|
var len = end - start; |
|
var i; |
|
|
|
if (this === target && start < targetStart && targetStart < end) { |
|
// descending copy from end |
|
for (i = len - 1; i >= 0; --i) { |
|
target[i + targetStart] = this[i + start]; |
|
} |
|
} else if (len < 1000 || !Buffer.TYPED_ARRAY_SUPPORT) { |
|
// ascending copy from start |
|
for (i = 0; i < len; ++i) { |
|
target[i + targetStart] = this[i + start]; |
|
} |
|
} else { |
|
Uint8Array.prototype.set.call(target, this.subarray(start, start + len), targetStart); |
|
} |
|
|
|
return len; |
|
}; // Usage: |
|
// buffer.fill(number[, offset[, end]]) |
|
// buffer.fill(buffer[, offset[, end]]) |
|
// buffer.fill(string[, offset[, end]][, encoding]) |
|
|
|
|
|
Buffer.prototype.fill = function fill(val, start, end, encoding) { |
|
// Handle string cases: |
|
if (typeof val === 'string') { |
|
if (typeof start === 'string') { |
|
encoding = start; |
|
start = 0; |
|
end = this.length; |
|
} else if (typeof end === 'string') { |
|
encoding = end; |
|
end = this.length; |
|
} |
|
|
|
if (val.length === 1) { |
|
var code = val.charCodeAt(0); |
|
|
|
if (code < 256) { |
|
val = code; |
|
} |
|
} |
|
|
|
if (encoding !== undefined && typeof encoding !== 'string') { |
|
throw new TypeError('encoding must be a string'); |
|
} |
|
|
|
if (typeof encoding === 'string' && !Buffer.isEncoding(encoding)) { |
|
throw new TypeError('Unknown encoding: ' + encoding); |
|
} |
|
} else if (typeof val === 'number') { |
|
val = val & 255; |
|
} // Invalid ranges are not set to a default, so can range check early. |
|
|
|
|
|
if (start < 0 || this.length < start || this.length < end) { |
|
throw new RangeError('Out of range index'); |
|
} |
|
|
|
if (end <= start) { |
|
return this; |
|
} |
|
|
|
start = start >>> 0; |
|
end = end === undefined ? this.length : end >>> 0; |
|
if (!val) val = 0; |
|
var i; |
|
|
|
if (typeof val === 'number') { |
|
for (i = start; i < end; ++i) { |
|
this[i] = val; |
|
} |
|
} else { |
|
var bytes = internalIsBuffer(val) ? val : utf8ToBytes(new Buffer(val, encoding).toString()); |
|
var len = bytes.length; |
|
|
|
for (i = 0; i < end - start; ++i) { |
|
this[i + start] = bytes[i % len]; |
|
} |
|
} |
|
|
|
return this; |
|
}; // HELPER FUNCTIONS |
|
// ================ |
|
|
|
|
|
var INVALID_BASE64_RE = /[^+\/0-9A-Za-z-_]/g; |
|
|
|
function base64clean(str) { |
|
// Node strips out invalid characters like \n and \t from the string, base64-js does not |
|
str = stringtrim(str).replace(INVALID_BASE64_RE, ''); // Node converts strings with length < 2 to '' |
|
|
|
if (str.length < 2) return ''; // Node allows for non-padded base64 strings (missing trailing ===), base64-js does not |
|
|
|
while (str.length % 4 !== 0) { |
|
str = str + '='; |
|
} |
|
|
|
return str; |
|
} |
|
|
|
function stringtrim(str) { |
|
if (str.trim) return str.trim(); |
|
return str.replace(/^\s+|\s+$/g, ''); |
|
} |
|
|
|
function toHex(n) { |
|
if (n < 16) return '0' + n.toString(16); |
|
return n.toString(16); |
|
} |
|
|
|
function utf8ToBytes(string, units) { |
|
units = units || Infinity; |
|
var codePoint; |
|
var length = string.length; |
|
var leadSurrogate = null; |
|
var bytes = []; |
|
|
|
for (var i = 0; i < length; ++i) { |
|
codePoint = string.charCodeAt(i); // is surrogate component |
|
|
|
if (codePoint > 0xD7FF && codePoint < 0xE000) { |
|
// last char was a lead |
|
if (!leadSurrogate) { |
|
// no lead yet |
|
if (codePoint > 0xDBFF) { |
|
// unexpected trail |
|
if ((units -= 3) > -1) bytes.push(0xEF, 0xBF, 0xBD); |
|
continue; |
|
} else if (i + 1 === length) { |
|
// unpaired lead |
|
if ((units -= 3) > -1) bytes.push(0xEF, 0xBF, 0xBD); |
|
continue; |
|
} // valid lead |
|
|
|
|
|
leadSurrogate = codePoint; |
|
continue; |
|
} // 2 leads in a row |
|
|
|
|
|
if (codePoint < 0xDC00) { |
|
if ((units -= 3) > -1) bytes.push(0xEF, 0xBF, 0xBD); |
|
leadSurrogate = codePoint; |
|
continue; |
|
} // valid surrogate pair |
|
|
|
|
|
codePoint = (leadSurrogate - 0xD800 << 10 | codePoint - 0xDC00) + 0x10000; |
|
} else if (leadSurrogate) { |
|
// valid bmp char, but last char was a lead |
|
if ((units -= 3) > -1) bytes.push(0xEF, 0xBF, 0xBD); |
|
} |
|
|
|
leadSurrogate = null; // encode utf8 |
|
|
|
if (codePoint < 0x80) { |
|
if ((units -= 1) < 0) break; |
|
bytes.push(codePoint); |
|
} else if (codePoint < 0x800) { |
|
if ((units -= 2) < 0) break; |
|
bytes.push(codePoint >> 0x6 | 0xC0, codePoint & 0x3F | 0x80); |
|
} else if (codePoint < 0x10000) { |
|
if ((units -= 3) < 0) break; |
|
bytes.push(codePoint >> 0xC | 0xE0, codePoint >> 0x6 & 0x3F | 0x80, codePoint & 0x3F | 0x80); |
|
} else if (codePoint < 0x110000) { |
|
if ((units -= 4) < 0) break; |
|
bytes.push(codePoint >> 0x12 | 0xF0, codePoint >> 0xC & 0x3F | 0x80, codePoint >> 0x6 & 0x3F | 0x80, codePoint & 0x3F | 0x80); |
|
} else { |
|
throw new Error('Invalid code point'); |
|
} |
|
} |
|
|
|
return bytes; |
|
} |
|
|
|
function asciiToBytes(str) { |
|
var byteArray = []; |
|
|
|
for (var i = 0; i < str.length; ++i) { |
|
// Node's code seems to be doing this and not & 0x7F.. |
|
byteArray.push(str.charCodeAt(i) & 0xFF); |
|
} |
|
|
|
return byteArray; |
|
} |
|
|
|
function utf16leToBytes(str, units) { |
|
var c, hi, lo; |
|
var byteArray = []; |
|
|
|
for (var i = 0; i < str.length; ++i) { |
|
if ((units -= 2) < 0) break; |
|
c = str.charCodeAt(i); |
|
hi = c >> 8; |
|
lo = c % 256; |
|
byteArray.push(lo); |
|
byteArray.push(hi); |
|
} |
|
|
|
return byteArray; |
|
} |
|
|
|
function base64ToBytes(str) { |
|
return toByteArray(base64clean(str)); |
|
} |
|
|
|
function blitBuffer(src, dst, offset, length) { |
|
for (var i = 0; i < length; ++i) { |
|
if (i + offset >= dst.length || i >= src.length) break; |
|
dst[i + offset] = src[i]; |
|
} |
|
|
|
return i; |
|
} |
|
|
|
function isnan(val) { |
|
return val !== val; // eslint-disable-line no-self-compare |
|
} // the following is from is-buffer, also by Feross Aboukhadijeh and with same lisence |
|
// The _isBuffer check is for Safari 5-7 support, because it's missing |
|
// Object.prototype.constructor. Remove this eventually |
|
|
|
|
|
function isBuffer(obj) { |
|
return obj != null && (!!obj._isBuffer || isFastBuffer(obj) || isSlowBuffer(obj)); |
|
} |
|
|
|
function isFastBuffer(obj) { |
|
return !!obj.constructor && typeof obj.constructor.isBuffer === 'function' && obj.constructor.isBuffer(obj); |
|
} // For Node v0.10 support. Remove this eventually. |
|
|
|
|
|
function isSlowBuffer(obj) { |
|
return typeof obj.readFloatLE === 'function' && typeof obj.slice === 'function' && isFastBuffer(obj.slice(0, 0)); |
|
} |
|
|
|
var commonjsGlobal = typeof globalThis !== 'undefined' ? globalThis : typeof window !== 'undefined' ? window : typeof global !== 'undefined' ? global : typeof self !== 'undefined' ? self : {}; |
|
|
|
function unwrapExports (x) { |
|
return x && x.__esModule && Object.prototype.hasOwnProperty.call(x, 'default') ? x['default'] : x; |
|
} |
|
|
|
function createCommonjsModule(fn, module) { |
|
return module = { exports: {} }, fn(module, module.exports), module.exports; |
|
} |
|
|
|
function getCjsExportFromNamespace (n) { |
|
return n && n['default'] || n; |
|
} |
|
|
|
var fs = getCjsExportFromNamespace(_shim_fs$1); |
|
|
|
/** |
|
* @class |
|
*/ |
|
|
|
|
|
var LineByLine = |
|
/*#__PURE__*/ |
|
function () { |
|
function LineByLine(file, options) { |
|
_classCallCheck(this, LineByLine); |
|
|
|
options = options || {}; |
|
if (!options.readChunk) options.readChunk = 1024; |
|
|
|
if (!options.newLineCharacter) { |
|
options.newLineCharacter = 0x0a; //linux line ending |
|
} else { |
|
options.newLineCharacter = options.newLineCharacter.charCodeAt(0); |
|
} |
|
|
|
if (typeof file === 'number') { |
|
this.fd = file; |
|
} else { |
|
this.fd = fs.openSync(file, 'r'); |
|
} |
|
|
|
this.options = options; |
|
this.newLineCharacter = options.newLineCharacter; |
|
this.reset(); |
|
} |
|
|
|
_createClass(LineByLine, [{ |
|
key: "_searchInBuffer", |
|
value: function _searchInBuffer(buffer, hexNeedle) { |
|
var found = -1; |
|
|
|
for (var i = 0; i <= buffer.length; i++) { |
|
var b_byte = buffer[i]; |
|
|
|
if (b_byte === hexNeedle) { |
|
found = i; |
|
break; |
|
} |
|
} |
|
|
|
return found; |
|
} |
|
}, { |
|
key: "reset", |
|
value: function reset() { |
|
this.eofReached = false; |
|
this.linesCache = []; |
|
this.fdPosition = 0; |
|
} |
|
}, { |
|
key: "close", |
|
value: function close() { |
|
fs.closeSync(this.fd); |
|
this.fd = null; |
|
} |
|
}, { |
|
key: "_extractLines", |
|
value: function _extractLines(buffer) { |
|
var line; |
|
var lines = []; |
|
var bufferPosition = 0; |
|
var lastNewLineBufferPosition = 0; |
|
|
|
while (true) { |
|
var bufferPositionValue = buffer[bufferPosition++]; |
|
|
|
if (bufferPositionValue === this.newLineCharacter) { |
|
line = buffer.slice(lastNewLineBufferPosition, bufferPosition); |
|
lines.push(line); |
|
lastNewLineBufferPosition = bufferPosition; |
|
} else if (!bufferPositionValue) { |
|
break; |
|
} |
|
} |
|
|
|
var leftovers = buffer.slice(lastNewLineBufferPosition, bufferPosition); |
|
|
|
if (leftovers.length) { |
|
lines.push(leftovers); |
|
} |
|
|
|
return lines; |
|
} |
|
}, { |
|
key: "_readChunk", |
|
value: function _readChunk(lineLeftovers) { |
|
var totalBytesRead = 0; |
|
var bytesRead; |
|
var buffers = []; |
|
|
|
do { |
|
var readBuffer = new Buffer(this.options.readChunk); |
|
bytesRead = fs.readSync(this.fd, readBuffer, 0, this.options.readChunk, this.fdPosition); |
|
totalBytesRead = totalBytesRead + bytesRead; |
|
this.fdPosition = this.fdPosition + bytesRead; |
|
buffers.push(readBuffer); |
|
} while (bytesRead && this._searchInBuffer(buffers[buffers.length - 1], this.options.newLineCharacter) === -1); |
|
|
|
var bufferData = Buffer.concat(buffers); |
|
|
|
if (bytesRead < this.options.readChunk) { |
|
this.eofReached = true; |
|
bufferData = bufferData.slice(0, totalBytesRead); |
|
} |
|
|
|
if (totalBytesRead) { |
|
this.linesCache = this._extractLines(bufferData); |
|
|
|
if (lineLeftovers) { |
|
this.linesCache[0] = Buffer.concat([lineLeftovers, this.linesCache[0]]); |
|
} |
|
} |
|
|
|
return totalBytesRead; |
|
} |
|
}, { |
|
key: "next", |
|
value: function next() { |
|
if (!this.fd) return false; |
|
var line = false; |
|
|
|
if (this.eofReached && this.linesCache.length === 0) { |
|
return line; |
|
} |
|
|
|
var bytesRead; |
|
|
|
if (!this.linesCache.length) { |
|
bytesRead = this._readChunk(); |
|
} |
|
|
|
if (this.linesCache.length) { |
|
line = this.linesCache.shift(); |
|
var lastLineCharacter = line[line.length - 1]; |
|
|
|
if (lastLineCharacter !== 0x0a) { |
|
bytesRead = this._readChunk(line); |
|
|
|
if (bytesRead) { |
|
line = this.linesCache.shift(); |
|
} |
|
} |
|
} |
|
|
|
if (this.eofReached && this.linesCache.length === 0) { |
|
this.close(); |
|
} |
|
|
|
if (line && line[line.length - 1] === this.newLineCharacter) { |
|
line = line.slice(0, line.length - 1); |
|
} |
|
|
|
return line; |
|
} |
|
}]); |
|
|
|
return LineByLine; |
|
}(); |
|
|
|
var readlines = LineByLine; |
|
|
|
var ConfigError = |
|
/*#__PURE__*/ |
|
function (_Error) { |
|
_inherits(ConfigError, _Error); |
|
|
|
function ConfigError() { |
|
_classCallCheck(this, ConfigError); |
|
|
|
return _possibleConstructorReturn(this, _getPrototypeOf(ConfigError).apply(this, arguments)); |
|
} |
|
|
|
return ConfigError; |
|
}(_wrapNativeSuper(Error)); |
|
|
|
var DebugError = |
|
/*#__PURE__*/ |
|
function (_Error2) { |
|
_inherits(DebugError, _Error2); |
|
|
|
function DebugError() { |
|
_classCallCheck(this, DebugError); |
|
|
|
return _possibleConstructorReturn(this, _getPrototypeOf(DebugError).apply(this, arguments)); |
|
} |
|
|
|
return DebugError; |
|
}(_wrapNativeSuper(Error)); |
|
|
|
var UndefinedParserError = |
|
/*#__PURE__*/ |
|
function (_Error3) { |
|
_inherits(UndefinedParserError, _Error3); |
|
|
|
function UndefinedParserError() { |
|
_classCallCheck(this, UndefinedParserError); |
|
|
|
return _possibleConstructorReturn(this, _getPrototypeOf(UndefinedParserError).apply(this, arguments)); |
|
} |
|
|
|
return UndefinedParserError; |
|
}(_wrapNativeSuper(Error)); |
|
|
|
var errors = { |
|
ConfigError: ConfigError, |
|
DebugError: DebugError, |
|
UndefinedParserError: UndefinedParserError |
|
}; |
|
|
|
// based off https://github.com/defunctzombie/node-process/blob/master/browser.js |
|
|
|
function defaultSetTimout() { |
|
throw new Error('setTimeout has not been defined'); |
|
} |
|
|
|
function defaultClearTimeout() { |
|
throw new Error('clearTimeout has not been defined'); |
|
} |
|
|
|
var cachedSetTimeout = defaultSetTimout; |
|
var cachedClearTimeout = defaultClearTimeout; |
|
|
|
if (typeof global$1.setTimeout === 'function') { |
|
cachedSetTimeout = setTimeout; |
|
} |
|
|
|
if (typeof global$1.clearTimeout === 'function') { |
|
cachedClearTimeout = clearTimeout; |
|
} |
|
|
|
function runTimeout(fun) { |
|
if (cachedSetTimeout === setTimeout) { |
|
//normal enviroments in sane situations |
|
return setTimeout(fun, 0); |
|
} // if setTimeout wasn't available but was latter defined |
|
|
|
|
|
if ((cachedSetTimeout === defaultSetTimout || !cachedSetTimeout) && setTimeout) { |
|
cachedSetTimeout = setTimeout; |
|
return setTimeout(fun, 0); |
|
} |
|
|
|
try { |
|
// when when somebody has screwed with setTimeout but no I.E. maddness |
|
return cachedSetTimeout(fun, 0); |
|
} catch (e) { |
|
try { |
|
// When we are in I.E. but the script has been evaled so I.E. doesn't trust the global object when called normally |
|
return cachedSetTimeout.call(null, fun, 0); |
|
} catch (e) { |
|
// same as above but when it's a version of I.E. that must have the global object for 'this', hopfully our context correct otherwise it will throw a global error |
|
return cachedSetTimeout.call(this, fun, 0); |
|
} |
|
} |
|
} |
|
|
|
function runClearTimeout(marker) { |
|
if (cachedClearTimeout === clearTimeout) { |
|
//normal enviroments in sane situations |
|
return clearTimeout(marker); |
|
} // if clearTimeout wasn't available but was latter defined |
|
|
|
|
|
if ((cachedClearTimeout === defaultClearTimeout || !cachedClearTimeout) && clearTimeout) { |
|
cachedClearTimeout = clearTimeout; |
|
return clearTimeout(marker); |
|
} |
|
|
|
try { |
|
// when when somebody has screwed with setTimeout but no I.E. maddness |
|
return cachedClearTimeout(marker); |
|
} catch (e) { |
|
try { |
|
// When we are in I.E. but the script has been evaled so I.E. doesn't trust the global object when called normally |
|
return cachedClearTimeout.call(null, marker); |
|
} catch (e) { |
|
// same as above but when it's a version of I.E. that must have the global object for 'this', hopfully our context correct otherwise it will throw a global error. |
|
// Some versions of I.E. have different rules for clearTimeout vs setTimeout |
|
return cachedClearTimeout.call(this, marker); |
|
} |
|
} |
|
} |
|
|
|
var queue = []; |
|
var draining = false; |
|
var currentQueue; |
|
var queueIndex = -1; |
|
|
|
function cleanUpNextTick() { |
|
if (!draining || !currentQueue) { |
|
return; |
|
} |
|
|
|
draining = false; |
|
|
|
if (currentQueue.length) { |
|
queue = currentQueue.concat(queue); |
|
} else { |
|
queueIndex = -1; |
|
} |
|
|
|
if (queue.length) { |
|
drainQueue(); |
|
} |
|
} |
|
|
|
function drainQueue() { |
|
if (draining) { |
|
return; |
|
} |
|
|
|
var timeout = runTimeout(cleanUpNextTick); |
|
draining = true; |
|
var len = queue.length; |
|
|
|
while (len) { |
|
currentQueue = queue; |
|
queue = []; |
|
|
|
while (++queueIndex < len) { |
|
if (currentQueue) { |
|
currentQueue[queueIndex].run(); |
|
} |
|
} |
|
|
|
queueIndex = -1; |
|
len = queue.length; |
|
} |
|
|
|
currentQueue = null; |
|
draining = false; |
|
runClearTimeout(timeout); |
|
} |
|
|
|
function nextTick(fun) { |
|
var args = new Array(arguments.length - 1); |
|
|
|
if (arguments.length > 1) { |
|
for (var i = 1; i < arguments.length; i++) { |
|
args[i - 1] = arguments[i]; |
|
} |
|
} |
|
|
|
queue.push(new Item(fun, args)); |
|
|
|
if (queue.length === 1 && !draining) { |
|
runTimeout(drainQueue); |
|
} |
|
} // v8 likes predictible objects |
|
|
|
function Item(fun, array) { |
|
this.fun = fun; |
|
this.array = array; |
|
} |
|
|
|
Item.prototype.run = function () { |
|
this.fun.apply(null, this.array); |
|
}; |
|
|
|
var title = 'browser'; |
|
var platform = 'browser'; |
|
var browser = true; |
|
var env = {}; |
|
var argv = []; |
|
var version$1 = ''; // empty string to avoid regexp issues |
|
|
|
var versions = {}; |
|
var release = {}; |
|
var config = {}; |
|
|
|
function noop() {} |
|
|
|
var on = noop; |
|
var addListener = noop; |
|
var once = noop; |
|
var off = noop; |
|
var removeListener = noop; |
|
var removeAllListeners = noop; |
|
var emit = noop; |
|
function binding(name) { |
|
throw new Error('process.binding is not supported'); |
|
} |
|
function cwd() { |
|
return '/'; |
|
} |
|
function chdir(dir) { |
|
throw new Error('process.chdir is not supported'); |
|
} |
|
function umask() { |
|
return 0; |
|
} // from https://github.com/kumavis/browser-process-hrtime/blob/master/index.js |
|
|
|
var performance = global$1.performance || {}; |
|
|
|
var performanceNow = performance.now || performance.mozNow || performance.msNow || performance.oNow || performance.webkitNow || function () { |
|
return new Date().getTime(); |
|
}; // generate timestamp or delta |
|
// see http://nodejs.org/api/process.html#process_process_hrtime |
|
|
|
|
|
function hrtime(previousTimestamp) { |
|
var clocktime = performanceNow.call(performance) * 1e-3; |
|
var seconds = Math.floor(clocktime); |
|
var nanoseconds = Math.floor(clocktime % 1 * 1e9); |
|
|
|
if (previousTimestamp) { |
|
seconds = seconds - previousTimestamp[0]; |
|
nanoseconds = nanoseconds - previousTimestamp[1]; |
|
|
|
if (nanoseconds < 0) { |
|
seconds--; |
|
nanoseconds += 1e9; |
|
} |
|
} |
|
|
|
return [seconds, nanoseconds]; |
|
} |
|
var startTime = new Date(); |
|
function uptime() { |
|
var currentTime = new Date(); |
|
var dif = currentTime - startTime; |
|
return dif / 1000; |
|
} |
|
var process = { |
|
nextTick: nextTick, |
|
title: title, |
|
browser: browser, |
|
env: env, |
|
argv: argv, |
|
version: version$1, |
|
versions: versions, |
|
on: on, |
|
addListener: addListener, |
|
once: once, |
|
off: off, |
|
removeListener: removeListener, |
|
removeAllListeners: removeAllListeners, |
|
emit: emit, |
|
binding: binding, |
|
cwd: cwd, |
|
chdir: chdir, |
|
umask: umask, |
|
hrtime: hrtime, |
|
platform: platform, |
|
release: release, |
|
config: config, |
|
uptime: uptime |
|
}; |
|
|
|
var semver = createCommonjsModule(function (module, exports) { |
|
exports = module.exports = SemVer; |
|
var debug; |
|
/* istanbul ignore next */ |
|
|
|
if (_typeof(process) === 'object' && process.env && process.env.NODE_DEBUG && /\bsemver\b/i.test(process.env.NODE_DEBUG)) { |
|
debug = function debug() { |
|
var args = Array.prototype.slice.call(arguments, 0); |
|
args.unshift('SEMVER'); |
|
console.log.apply(console, args); |
|
}; |
|
} else { |
|
debug = function debug() {}; |
|
} // Note: this is the semver.org version of the spec that it implements |
|
// Not necessarily the package version of this code. |
|
|
|
|
|
exports.SEMVER_SPEC_VERSION = '2.0.0'; |
|
var MAX_LENGTH = 256; |
|
var MAX_SAFE_INTEGER = Number.MAX_SAFE_INTEGER || |
|
/* istanbul ignore next */ |
|
9007199254740991; // Max safe segment length for coercion. |
|
|
|
var MAX_SAFE_COMPONENT_LENGTH = 16; // The actual regexps go on exports.re |
|
|
|
var re = exports.re = []; |
|
var src = exports.src = []; |
|
var t = exports.tokens = {}; |
|
var R = 0; |
|
|
|
function tok(n) { |
|
t[n] = R++; |
|
} // The following Regular Expressions can be used for tokenizing, |
|
// validating, and parsing SemVer version strings. |
|
// ## Numeric Identifier |
|
// A single `0`, or a non-zero digit followed by zero or more digits. |
|
|
|
|
|
tok('NUMERICIDENTIFIER'); |
|
src[t.NUMERICIDENTIFIER] = '0|[1-9]\\d*'; |
|
tok('NUMERICIDENTIFIERLOOSE'); |
|
src[t.NUMERICIDENTIFIERLOOSE] = '[0-9]+'; // ## Non-numeric Identifier |
|
// Zero or more digits, followed by a letter or hyphen, and then zero or |
|
// more letters, digits, or hyphens. |
|
|
|
tok('NONNUMERICIDENTIFIER'); |
|
src[t.NONNUMERICIDENTIFIER] = '\\d*[a-zA-Z-][a-zA-Z0-9-]*'; // ## Main Version |
|
// Three dot-separated numeric identifiers. |
|
|
|
tok('MAINVERSION'); |
|
src[t.MAINVERSION] = '(' + src[t.NUMERICIDENTIFIER] + ')\\.' + '(' + src[t.NUMERICIDENTIFIER] + ')\\.' + '(' + src[t.NUMERICIDENTIFIER] + ')'; |
|
tok('MAINVERSIONLOOSE'); |
|
src[t.MAINVERSIONLOOSE] = '(' + src[t.NUMERICIDENTIFIERLOOSE] + ')\\.' + '(' + src[t.NUMERICIDENTIFIERLOOSE] + ')\\.' + '(' + src[t.NUMERICIDENTIFIERLOOSE] + ')'; // ## Pre-release Version Identifier |
|
// A numeric identifier, or a non-numeric identifier. |
|
|
|
tok('PRERELEASEIDENTIFIER'); |
|
src[t.PRERELEASEIDENTIFIER] = '(?:' + src[t.NUMERICIDENTIFIER] + '|' + src[t.NONNUMERICIDENTIFIER] + ')'; |
|
tok('PRERELEASEIDENTIFIERLOOSE'); |
|
src[t.PRERELEASEIDENTIFIERLOOSE] = '(?:' + src[t.NUMERICIDENTIFIERLOOSE] + '|' + src[t.NONNUMERICIDENTIFIER] + ')'; // ## Pre-release Version |
|
// Hyphen, followed by one or more dot-separated pre-release version |
|
// identifiers. |
|
|
|
tok('PRERELEASE'); |
|
src[t.PRERELEASE] = '(?:-(' + src[t.PRERELEASEIDENTIFIER] + '(?:\\.' + src[t.PRERELEASEIDENTIFIER] + ')*))'; |
|
tok('PRERELEASELOOSE'); |
|
src[t.PRERELEASELOOSE] = '(?:-?(' + src[t.PRERELEASEIDENTIFIERLOOSE] + '(?:\\.' + src[t.PRERELEASEIDENTIFIERLOOSE] + ')*))'; // ## Build Metadata Identifier |
|
// Any combination of digits, letters, or hyphens. |
|
|
|
tok('BUILDIDENTIFIER'); |
|
src[t.BUILDIDENTIFIER] = '[0-9A-Za-z-]+'; // ## Build Metadata |
|
// Plus sign, followed by one or more period-separated build metadata |
|
// identifiers. |
|
|
|
tok('BUILD'); |
|
src[t.BUILD] = '(?:\\+(' + src[t.BUILDIDENTIFIER] + '(?:\\.' + src[t.BUILDIDENTIFIER] + ')*))'; // ## Full Version String |
|
// A main version, followed optionally by a pre-release version and |
|
// build metadata. |
|
// Note that the only major, minor, patch, and pre-release sections of |
|
// the version string are capturing groups. The build metadata is not a |
|
// capturing group, because it should not ever be used in version |
|
// comparison. |
|
|
|
tok('FULL'); |
|
tok('FULLPLAIN'); |
|
src[t.FULLPLAIN] = 'v?' + src[t.MAINVERSION] + src[t.PRERELEASE] + '?' + src[t.BUILD] + '?'; |
|
src[t.FULL] = '^' + src[t.FULLPLAIN] + '$'; // like full, but allows v1.2.3 and =1.2.3, which people do sometimes. |
|
// also, 1.0.0alpha1 (prerelease without the hyphen) which is pretty |
|
// common in the npm registry. |
|
|
|
tok('LOOSEPLAIN'); |
|
src[t.LOOSEPLAIN] = '[v=\\s]*' + src[t.MAINVERSIONLOOSE] + src[t.PRERELEASELOOSE] + '?' + src[t.BUILD] + '?'; |
|
tok('LOOSE'); |
|
src[t.LOOSE] = '^' + src[t.LOOSEPLAIN] + '$'; |
|
tok('GTLT'); |
|
src[t.GTLT] = '((?:<|>)?=?)'; // Something like "2.*" or "1.2.x". |
|
// Note that "x.x" is a valid xRange identifer, meaning "any version" |
|
// Only the first item is strictly required. |
|
|
|
tok('XRANGEIDENTIFIERLOOSE'); |
|
src[t.XRANGEIDENTIFIERLOOSE] = src[t.NUMERICIDENTIFIERLOOSE] + '|x|X|\\*'; |
|
tok('XRANGEIDENTIFIER'); |
|
src[t.XRANGEIDENTIFIER] = src[t.NUMERICIDENTIFIER] + '|x|X|\\*'; |
|
tok('XRANGEPLAIN'); |
|
src[t.XRANGEPLAIN] = '[v=\\s]*(' + src[t.XRANGEIDENTIFIER] + ')' + '(?:\\.(' + src[t.XRANGEIDENTIFIER] + ')' + '(?:\\.(' + src[t.XRANGEIDENTIFIER] + ')' + '(?:' + src[t.PRERELEASE] + ')?' + src[t.BUILD] + '?' + ')?)?'; |
|
tok('XRANGEPLAINLOOSE'); |
|
src[t.XRANGEPLAINLOOSE] = '[v=\\s]*(' + src[t.XRANGEIDENTIFIERLOOSE] + ')' + '(?:\\.(' + src[t.XRANGEIDENTIFIERLOOSE] + ')' + '(?:\\.(' + src[t.XRANGEIDENTIFIERLOOSE] + ')' + '(?:' + src[t.PRERELEASELOOSE] + ')?' + src[t.BUILD] + '?' + ')?)?'; |
|
tok('XRANGE'); |
|
src[t.XRANGE] = '^' + src[t.GTLT] + '\\s*' + src[t.XRANGEPLAIN] + '$'; |
|
tok('XRANGELOOSE'); |
|
src[t.XRANGELOOSE] = '^' + src[t.GTLT] + '\\s*' + src[t.XRANGEPLAINLOOSE] + '$'; // Coercion. |
|
// Extract anything that could conceivably be a part of a valid semver |
|
|
|
tok('COERCE'); |
|
src[t.COERCE] = '(^|[^\\d])' + '(\\d{1,' + MAX_SAFE_COMPONENT_LENGTH + '})' + '(?:\\.(\\d{1,' + MAX_SAFE_COMPONENT_LENGTH + '}))?' + '(?:\\.(\\d{1,' + MAX_SAFE_COMPONENT_LENGTH + '}))?' + '(?:$|[^\\d])'; |
|
tok('COERCERTL'); |
|
re[t.COERCERTL] = new RegExp(src[t.COERCE], 'g'); // Tilde ranges. |
|
// Meaning is "reasonably at or greater than" |
|
|
|
tok('LONETILDE'); |
|
src[t.LONETILDE] = '(?:~>?)'; |
|
tok('TILDETRIM'); |
|
src[t.TILDETRIM] = '(\\s*)' + src[t.LONETILDE] + '\\s+'; |
|
re[t.TILDETRIM] = new RegExp(src[t.TILDETRIM], 'g'); |
|
var tildeTrimReplace = '$1~'; |
|
tok('TILDE'); |
|
src[t.TILDE] = '^' + src[t.LONETILDE] + src[t.XRANGEPLAIN] + '$'; |
|
tok('TILDELOOSE'); |
|
src[t.TILDELOOSE] = '^' + src[t.LONETILDE] + src[t.XRANGEPLAINLOOSE] + '$'; // Caret ranges. |
|
// Meaning is "at least and backwards compatible with" |
|
|
|
tok('LONECARET'); |
|
src[t.LONECARET] = '(?:\\^)'; |
|
tok('CARETTRIM'); |
|
src[t.CARETTRIM] = '(\\s*)' + src[t.LONECARET] + '\\s+'; |
|
re[t.CARETTRIM] = new RegExp(src[t.CARETTRIM], 'g'); |
|
var caretTrimReplace = '$1^'; |
|
tok('CARET'); |
|
src[t.CARET] = '^' + src[t.LONECARET] + src[t.XRANGEPLAIN] + '$'; |
|
tok('CARETLOOSE'); |
|
src[t.CARETLOOSE] = '^' + src[t.LONECARET] + src[t.XRANGEPLAINLOOSE] + '$'; // A simple gt/lt/eq thing, or just "" to indicate "any version" |
|
|
|
tok('COMPARATORLOOSE'); |
|
src[t.COMPARATORLOOSE] = '^' + src[t.GTLT] + '\\s*(' + src[t.LOOSEPLAIN] + ')$|^$'; |
|
tok('COMPARATOR'); |
|
src[t.COMPARATOR] = '^' + src[t.GTLT] + '\\s*(' + src[t.FULLPLAIN] + ')$|^$'; // An expression to strip any whitespace between the gtlt and the thing |
|
// it modifies, so that `> 1.2.3` ==> `>1.2.3` |
|
|
|
tok('COMPARATORTRIM'); |
|
src[t.COMPARATORTRIM] = '(\\s*)' + src[t.GTLT] + '\\s*(' + src[t.LOOSEPLAIN] + '|' + src[t.XRANGEPLAIN] + ')'; // this one has to use the /g flag |
|
|
|
re[t.COMPARATORTRIM] = new RegExp(src[t.COMPARATORTRIM], 'g'); |
|
var comparatorTrimReplace = '$1$2$3'; // Something like `1.2.3 - 1.2.4` |
|
// Note that these all use the loose form, because they'll be |
|
// checked against either the strict or loose comparator form |
|
// later. |
|
|
|
tok('HYPHENRANGE'); |
|
src[t.HYPHENRANGE] = '^\\s*(' + src[t.XRANGEPLAIN] + ')' + '\\s+-\\s+' + '(' + src[t.XRANGEPLAIN] + ')' + '\\s*$'; |
|
tok('HYPHENRANGELOOSE'); |
|
src[t.HYPHENRANGELOOSE] = '^\\s*(' + src[t.XRANGEPLAINLOOSE] + ')' + '\\s+-\\s+' + '(' + src[t.XRANGEPLAINLOOSE] + ')' + '\\s*$'; // Star ranges basically just allow anything at all. |
|
|
|
tok('STAR'); |
|
src[t.STAR] = '(<|>)?=?\\s*\\*'; // Compile to actual regexp objects. |
|
// All are flag-free, unless they were created above with a flag. |
|
|
|
for (var i = 0; i < R; i++) { |
|
debug(i, src[i]); |
|
|
|
if (!re[i]) { |
|
re[i] = new RegExp(src[i]); |
|
} |
|
} |
|
|
|
exports.parse = parse; |
|
|
|
function parse(version, options) { |
|
if (!options || _typeof(options) !== 'object') { |
|
options = { |
|
loose: !!options, |
|
includePrerelease: false |
|
}; |
|
} |
|
|
|
if (version instanceof SemVer) { |
|
return version; |
|
} |
|
|
|
if (typeof version !== 'string') { |
|
return null; |
|
} |
|
|
|
if (version.length > MAX_LENGTH) { |
|
return null; |
|
} |
|
|
|
var r = options.loose ? re[t.LOOSE] : re[t.FULL]; |
|
|
|
if (!r.test(version)) { |
|
return null; |
|
} |
|
|
|
try { |
|
return new SemVer(version, options); |
|
} catch (er) { |
|
return null; |
|
} |
|
} |
|
|
|
exports.valid = valid; |
|
|
|
function valid(version, options) { |
|
var v = parse(version, options); |
|
return v ? v.version : null; |
|
} |
|
|
|
exports.clean = clean; |
|
|
|
function clean(version, options) { |
|
var s = parse(version.trim().replace(/^[=v]+/, ''), options); |
|
return s ? s.version : null; |
|
} |
|
|
|
exports.SemVer = SemVer; |
|
|
|
function SemVer(version, options) { |
|
if (!options || _typeof(options) !== 'object') { |
|
options = { |
|
loose: !!options, |
|
includePrerelease: false |
|
}; |
|
} |
|
|
|
if (version instanceof SemVer) { |
|
if (version.loose === options.loose) { |
|
return version; |
|
} else { |
|
version = version.version; |
|
} |
|
} else if (typeof version !== 'string') { |
|
throw new TypeError('Invalid Version: ' + version); |
|
} |
|
|
|
if (version.length > MAX_LENGTH) { |
|
throw new TypeError('version is longer than ' + MAX_LENGTH + ' characters'); |
|
} |
|
|
|
if (!(this instanceof SemVer)) { |
|
return new SemVer(version, options); |
|
} |
|
|
|
debug('SemVer', version, options); |
|
this.options = options; |
|
this.loose = !!options.loose; |
|
var m = version.trim().match(options.loose ? re[t.LOOSE] : re[t.FULL]); |
|
|
|
if (!m) { |
|
throw new TypeError('Invalid Version: ' + version); |
|
} |
|
|
|
this.raw = version; // these are actually numbers |
|
|
|
this.major = +m[1]; |
|
this.minor = +m[2]; |
|
this.patch = +m[3]; |
|
|
|
if (this.major > MAX_SAFE_INTEGER || this.major < 0) { |
|
throw new TypeError('Invalid major version'); |
|
} |
|
|
|
if (this.minor > MAX_SAFE_INTEGER || this.minor < 0) { |
|
throw new TypeError('Invalid minor version'); |
|
} |
|
|
|
if (this.patch > MAX_SAFE_INTEGER || this.patch < 0) { |
|
throw new TypeError('Invalid patch version'); |
|
} // numberify any prerelease numeric ids |
|
|
|
|
|
if (!m[4]) { |
|
this.prerelease = []; |
|
} else { |
|
this.prerelease = m[4].split('.').map(function (id) { |
|
if (/^[0-9]+$/.test(id)) { |
|
var num = +id; |
|
|
|
if (num >= 0 && num < MAX_SAFE_INTEGER) { |
|
return num; |
|
} |
|
} |
|
|
|
return id; |
|
}); |
|
} |
|
|
|
this.build = m[5] ? m[5].split('.') : []; |
|
this.format(); |
|
} |
|
|
|
SemVer.prototype.format = function () { |
|
this.version = this.major + '.' + this.minor + '.' + this.patch; |
|
|
|
if (this.prerelease.length) { |
|
this.version += '-' + this.prerelease.join('.'); |
|
} |
|
|
|
return this.version; |
|
}; |
|
|
|
SemVer.prototype.toString = function () { |
|
return this.version; |
|
}; |
|
|
|
SemVer.prototype.compare = function (other) { |
|
debug('SemVer.compare', this.version, this.options, other); |
|
|
|
if (!(other instanceof SemVer)) { |
|
other = new SemVer(other, this.options); |
|
} |
|
|
|
return this.compareMain(other) || this.comparePre(other); |
|
}; |
|
|
|
SemVer.prototype.compareMain = function (other) { |
|
if (!(other instanceof SemVer)) { |
|
other = new SemVer(other, this.options); |
|
} |
|
|
|
return compareIdentifiers(this.major, other.major) || compareIdentifiers(this.minor, other.minor) || compareIdentifiers(this.patch, other.patch); |
|
}; |
|
|
|
SemVer.prototype.comparePre = function (other) { |
|
if (!(other instanceof SemVer)) { |
|
other = new SemVer(other, this.options); |
|
} // NOT having a prerelease is > having one |
|
|
|
|
|
if (this.prerelease.length && !other.prerelease.length) { |
|
return -1; |
|
} else if (!this.prerelease.length && other.prerelease.length) { |
|
return 1; |
|
} else if (!this.prerelease.length && !other.prerelease.length) { |
|
return 0; |
|
} |
|
|
|
var i = 0; |
|
|
|
do { |
|
var a = this.prerelease[i]; |
|
var b = other.prerelease[i]; |
|
debug('prerelease compare', i, a, b); |
|
|
|
if (a === undefined && b === undefined) { |
|
return 0; |
|
} else if (b === undefined) { |
|
return 1; |
|
} else if (a === undefined) { |
|
return -1; |
|
} else if (a === b) { |
|
continue; |
|
} else { |
|
return compareIdentifiers(a, b); |
|
} |
|
} while (++i); |
|
}; |
|
|
|
SemVer.prototype.compareBuild = function (other) { |
|
if (!(other instanceof SemVer)) { |
|
other = new SemVer(other, this.options); |
|
} |
|
|
|
var i = 0; |
|
|
|
do { |
|
var a = this.build[i]; |
|
var b = other.build[i]; |
|
debug('prerelease compare', i, a, b); |
|
|
|
if (a === undefined && b === undefined) { |
|
return 0; |
|
} else if (b === undefined) { |
|
return 1; |
|
} else if (a === undefined) { |
|
return -1; |
|
} else if (a === b) { |
|
continue; |
|
} else { |
|
return compareIdentifiers(a, b); |
|
} |
|
} while (++i); |
|
}; // preminor will bump the version up to the next minor release, and immediately |
|
// down to pre-release. premajor and prepatch work the same way. |
|
|
|
|
|
SemVer.prototype.inc = function (release, identifier) { |
|
switch (release) { |
|
case 'premajor': |
|
this.prerelease.length = 0; |
|
this.patch = 0; |
|
this.minor = 0; |
|
this.major++; |
|
this.inc('pre', identifier); |
|
break; |
|
|
|
case 'preminor': |
|
this.prerelease.length = 0; |
|
this.patch = 0; |
|
this.minor++; |
|
this.inc('pre', identifier); |
|
break; |
|
|
|
case 'prepatch': |
|
// If this is already a prerelease, it will bump to the next version |
|
// drop any prereleases that might already exist, since they are not |
|
// relevant at this point. |
|
this.prerelease.length = 0; |
|
this.inc('patch', identifier); |
|
this.inc('pre', identifier); |
|
break; |
|
// If the input is a non-prerelease version, this acts the same as |
|
// prepatch. |
|
|
|
case 'prerelease': |
|
if (this.prerelease.length === 0) { |
|
this.inc('patch', identifier); |
|
} |
|
|
|
this.inc('pre', identifier); |
|
break; |
|
|
|
case 'major': |
|
// If this is a pre-major version, bump up to the same major version. |
|
// Otherwise increment major. |
|
// 1.0.0-5 bumps to 1.0.0 |
|
// 1.1.0 bumps to 2.0.0 |
|
if (this.minor !== 0 || this.patch !== 0 || this.prerelease.length === 0) { |
|
this.major++; |
|
} |
|
|
|
this.minor = 0; |
|
this.patch = 0; |
|
this.prerelease = []; |
|
break; |
|
|
|
case 'minor': |
|
// If this is a pre-minor version, bump up to the same minor version. |
|
// Otherwise increment minor. |
|
// 1.2.0-5 bumps to 1.2.0 |
|
// 1.2.1 bumps to 1.3.0 |
|
if (this.patch !== 0 || this.prerelease.length === 0) { |
|
this.minor++; |
|
} |
|
|
|
this.patch = 0; |
|
this.prerelease = []; |
|
break; |
|
|
|
case 'patch': |
|
// If this is not a pre-release version, it will increment the patch. |
|
// If it is a pre-release it will bump up to the same patch version. |
|
// 1.2.0-5 patches to 1.2.0 |
|
// 1.2.0 patches to 1.2.1 |
|
if (this.prerelease.length === 0) { |
|
this.patch++; |
|
} |
|
|
|
this.prerelease = []; |
|
break; |
|
// This probably shouldn't be used publicly. |
|
// 1.0.0 "pre" would become 1.0.0-0 which is the wrong direction. |
|
|
|
case 'pre': |
|
if (this.prerelease.length === 0) { |
|
this.prerelease = [0]; |
|
} else { |
|
var i = this.prerelease.length; |
|
|
|
while (--i >= 0) { |
|
if (typeof this.prerelease[i] === 'number') { |
|
this.prerelease[i]++; |
|
i = -2; |
|
} |
|
} |
|
|
|
if (i === -1) { |
|
// didn't increment anything |
|
this.prerelease.push(0); |
|
} |
|
} |
|
|
|
if (identifier) { |
|
// 1.2.0-beta.1 bumps to 1.2.0-beta.2, |
|
// 1.2.0-beta.fooblz or 1.2.0-beta bumps to 1.2.0-beta.0 |
|
if (this.prerelease[0] === identifier) { |
|
if (isNaN(this.prerelease[1])) { |
|
this.prerelease = [identifier, 0]; |
|
} |
|
} else { |
|
this.prerelease = [identifier, 0]; |
|
} |
|
} |
|
|
|
break; |
|
|
|
default: |
|
throw new Error('invalid increment argument: ' + release); |
|
} |
|
|
|
this.format(); |
|
this.raw = this.version; |
|
return this; |
|
}; |
|
|
|
exports.inc = inc; |
|
|
|
function inc(version, release, loose, identifier) { |
|
if (typeof loose === 'string') { |
|
identifier = loose; |
|
loose = undefined; |
|
} |
|
|
|
try { |
|
return new SemVer(version, loose).inc(release, identifier).version; |
|
} catch (er) { |
|
return null; |
|
} |
|
} |
|
|
|
exports.diff = diff; |
|
|
|
function diff(version1, version2) { |
|
if (eq(version1, version2)) { |
|
return null; |
|
} else { |
|
var v1 = parse(version1); |
|
var v2 = parse(version2); |
|
var prefix = ''; |
|
|
|
if (v1.prerelease.length || v2.prerelease.length) { |
|
prefix = 'pre'; |
|
var defaultResult = 'prerelease'; |
|
} |
|
|
|
for (var key in v1) { |
|
if (key === 'major' || key === 'minor' || key === 'patch') { |
|
if (v1[key] !== v2[key]) { |
|
return prefix + key; |
|
} |
|
} |
|
} |
|
|
|
return defaultResult; // may be undefined |
|
} |
|
} |
|
|
|
exports.compareIdentifiers = compareIdentifiers; |
|
var numeric = /^[0-9]+$/; |
|
|
|
function compareIdentifiers(a, b) { |
|
var anum = numeric.test(a); |
|
var bnum = numeric.test(b); |
|
|
|
if (anum && bnum) { |
|
a = +a; |
|
b = +b; |
|
} |
|
|
|
return a === b ? 0 : anum && !bnum ? -1 : bnum && !anum ? 1 : a < b ? -1 : 1; |
|
} |
|
|
|
exports.rcompareIdentifiers = rcompareIdentifiers; |
|
|
|
function rcompareIdentifiers(a, b) { |
|
return compareIdentifiers(b, a); |
|
} |
|
|
|
exports.major = major; |
|
|
|
function major(a, loose) { |
|
return new SemVer(a, loose).major; |
|
} |
|
|
|
exports.minor = minor; |
|
|
|
function minor(a, loose) { |
|
return new SemVer(a, loose).minor; |
|
} |
|
|
|
exports.patch = patch; |
|
|
|
function patch(a, loose) { |
|
return new SemVer(a, loose).patch; |
|
} |
|
|
|
exports.compare = compare; |
|
|
|
function compare(a, b, loose) { |
|
return new SemVer(a, loose).compare(new SemVer(b, loose)); |
|
} |
|
|
|
exports.compareLoose = compareLoose; |
|
|
|
function compareLoose(a, b) { |
|
return compare(a, b, true); |
|
} |
|
|
|
exports.compareBuild = compareBuild; |
|
|
|
function compareBuild(a, b, loose) { |
|
var versionA = new SemVer(a, loose); |
|
var versionB = new SemVer(b, loose); |
|
return versionA.compare(versionB) || versionA.compareBuild(versionB); |
|
} |
|
|
|
exports.rcompare = rcompare; |
|
|
|
function rcompare(a, b, loose) { |
|
return compare(b, a, loose); |
|
} |
|
|
|
exports.sort = sort; |
|
|
|
function sort(list, loose) { |
|
return list.sort(function (a, b) { |
|
return exports.compareBuild(a, b, loose); |
|
}); |
|
} |
|
|
|
exports.rsort = rsort; |
|
|
|
function rsort(list, loose) { |
|
return list.sort(function (a, b) { |
|
return exports.compareBuild(b, a, loose); |
|
}); |
|
} |
|
|
|
exports.gt = gt; |
|
|
|
function gt(a, b, loose) { |
|
return compare(a, b, loose) > 0; |
|
} |
|
|
|
exports.lt = lt; |
|
|
|
function lt(a, b, loose) { |
|
return compare(a, b, loose) < 0; |
|
} |
|
|
|
exports.eq = eq; |
|
|
|
function eq(a, b, loose) { |
|
return compare(a, b, loose) === 0; |
|
} |
|
|
|
exports.neq = neq; |
|
|
|
function neq(a, b, loose) { |
|
return compare(a, b, loose) !== 0; |
|
} |
|
|
|
exports.gte = gte; |
|
|
|
function gte(a, b, loose) { |
|
return compare(a, b, loose) >= 0; |
|
} |
|
|
|
exports.lte = lte; |
|
|
|
function lte(a, b, loose) { |
|
return compare(a, b, loose) <= 0; |
|
} |
|
|
|
exports.cmp = cmp; |
|
|
|
function cmp(a, op, b, loose) { |
|
switch (op) { |
|
case '===': |
|
if (_typeof(a) === 'object') a = a.version; |
|
if (_typeof(b) === 'object') b = b.version; |
|
return a === b; |
|
|
|
case '!==': |
|
if (_typeof(a) === 'object') a = a.version; |
|
if (_typeof(b) === 'object') b = b.version; |
|
return a !== b; |
|
|
|
case '': |
|
case '=': |
|
case '==': |
|
return eq(a, b, loose); |
|
|
|
case '!=': |
|
return neq(a, b, loose); |
|
|
|
case '>': |
|
return gt(a, b, loose); |
|
|
|
case '>=': |
|
return gte(a, b, loose); |
|
|
|
case '<': |
|
return lt(a, b, loose); |
|
|
|
case '<=': |
|
return lte(a, b, loose); |
|
|
|
default: |
|
throw new TypeError('Invalid operator: ' + op); |
|
} |
|
} |
|
|
|
exports.Comparator = Comparator; |
|
|
|
function Comparator(comp, options) { |
|
if (!options || _typeof(options) !== 'object') { |
|
options = { |
|
loose: !!options, |
|
includePrerelease: false |
|
}; |
|
} |
|
|
|
if (comp instanceof Comparator) { |
|
if (comp.loose === !!options.loose) { |
|
return comp; |
|
} else { |
|
comp = comp.value; |
|
} |
|
} |
|
|
|
if (!(this instanceof Comparator)) { |
|
return new Comparator(comp, options); |
|
} |
|
|
|
debug('comparator', comp, options); |
|
this.options = options; |
|
this.loose = !!options.loose; |
|
this.parse(comp); |
|
|
|
if (this.semver === ANY) { |
|
this.value = ''; |
|
} else { |
|
this.value = this.operator + this.semver.version; |
|
} |
|
|
|
debug('comp', this); |
|
} |
|
|
|
var ANY = {}; |
|
|
|
Comparator.prototype.parse = function (comp) { |
|
var r = this.options.loose ? re[t.COMPARATORLOOSE] : re[t.COMPARATOR]; |
|
var m = comp.match(r); |
|
|
|
if (!m) { |
|
throw new TypeError('Invalid comparator: ' + comp); |
|
} |
|
|
|
this.operator = m[1] !== undefined ? m[1] : ''; |
|
|
|
if (this.operator === '=') { |
|
this.operator = ''; |
|
} // if it literally is just '>' or '' then allow anything. |
|
|
|
|
|
if (!m[2]) { |
|
this.semver = ANY; |
|
} else { |
|
this.semver = new SemVer(m[2], this.options.loose); |
|
} |
|
}; |
|
|
|
Comparator.prototype.toString = function () { |
|
return this.value; |
|
}; |
|
|
|
Comparator.prototype.test = function (version) { |
|
debug('Comparator.test', version, this.options.loose); |
|
|
|
if (this.semver === ANY || version === ANY) { |
|
return true; |
|
} |
|
|
|
if (typeof version === 'string') { |
|
try { |
|
version = new SemVer(version, this.options); |
|
} catch (er) { |
|
return false; |
|
} |
|
} |
|
|
|
return cmp(version, this.operator, this.semver, this.options); |
|
}; |
|
|
|
Comparator.prototype.intersects = function (comp, options) { |
|
if (!(comp instanceof Comparator)) { |
|
throw new TypeError('a Comparator is required'); |
|
} |
|
|
|
if (!options || _typeof(options) !== 'object') { |
|
options = { |
|
loose: !!options, |
|
includePrerelease: false |
|
}; |
|
} |
|
|
|
var rangeTmp; |
|
|
|
if (this.operator === '') { |
|
if (this.value === '') { |
|
return true; |
|
} |
|
|
|
rangeTmp = new Range(comp.value, options); |
|
return satisfies(this.value, rangeTmp, options); |
|
} else if (comp.operator === '') { |
|
if (comp.value === '') { |
|
return true; |
|
} |
|
|
|
rangeTmp = new Range(this.value, options); |
|
return satisfies(comp.semver, rangeTmp, options); |
|
} |
|
|
|
var sameDirectionIncreasing = (this.operator === '>=' || this.operator === '>') && (comp.operator === '>=' || comp.operator === '>'); |
|
var sameDirectionDecreasing = (this.operator === '<=' || this.operator === '<') && (comp.operator === '<=' || comp.operator === '<'); |
|
var sameSemVer = this.semver.version === comp.semver.version; |
|
var differentDirectionsInclusive = (this.operator === '>=' || this.operator === '<=') && (comp.operator === '>=' || comp.operator === '<='); |
|
var oppositeDirectionsLessThan = cmp(this.semver, '<', comp.semver, options) && (this.operator === '>=' || this.operator === '>') && (comp.operator === '<=' || comp.operator === '<'); |
|
var oppositeDirectionsGreaterThan = cmp(this.semver, '>', comp.semver, options) && (this.operator === '<=' || this.operator === '<') && (comp.operator === '>=' || comp.operator === '>'); |
|
return sameDirectionIncreasing || sameDirectionDecreasing || sameSemVer && differentDirectionsInclusive || oppositeDirectionsLessThan || oppositeDirectionsGreaterThan; |
|
}; |
|
|
|
exports.Range = Range; |
|
|
|
function Range(range, options) { |
|
if (!options || _typeof(options) !== 'object') { |
|
options = { |
|
loose: !!options, |
|
includePrerelease: false |
|
}; |
|
} |
|
|
|
if (range instanceof Range) { |
|
if (range.loose === !!options.loose && range.includePrerelease === !!options.includePrerelease) { |
|
return range; |
|
} else { |
|
return new Range(range.raw, options); |
|
} |
|
} |
|
|
|
if (range instanceof Comparator) { |
|
return new Range(range.value, options); |
|
} |
|
|
|
if (!(this instanceof Range)) { |
|
return new Range(range, options); |
|
} |
|
|
|
this.options = options; |
|
this.loose = !!options.loose; |
|
this.includePrerelease = !!options.includePrerelease; // First, split based on boolean or || |
|
|
|
this.raw = range; |
|
this.set = range.split(/\s*\|\|\s*/).map(function (range) { |
|
return this.parseRange(range.trim()); |
|
}, this).filter(function (c) { |
|
// throw out any that are not relevant for whatever reason |
|
return c.length; |
|
}); |
|
|
|
if (!this.set.length) { |
|
throw new TypeError('Invalid SemVer Range: ' + range); |
|
} |
|
|
|
this.format(); |
|
} |
|
|
|
Range.prototype.format = function () { |
|
this.range = this.set.map(function (comps) { |
|
return comps.join(' ').trim(); |
|
}).join('||').trim(); |
|
return this.range; |
|
}; |
|
|
|
Range.prototype.toString = function () { |
|
return this.range; |
|
}; |
|
|
|
Range.prototype.parseRange = function (range) { |
|
var loose = this.options.loose; |
|
range = range.trim(); // `1.2.3 - 1.2.4` => `>=1.2.3 <=1.2.4` |
|
|
|
var hr = loose ? re[t.HYPHENRANGELOOSE] : re[t.HYPHENRANGE]; |
|
range = range.replace(hr, hyphenReplace); |
|
debug('hyphen replace', range); // `> 1.2.3 < 1.2.5` => `>1.2.3 <1.2.5` |
|
|
|
range = range.replace(re[t.COMPARATORTRIM], comparatorTrimReplace); |
|
debug('comparator trim', range, re[t.COMPARATORTRIM]); // `~ 1.2.3` => `~1.2.3` |
|
|
|
range = range.replace(re[t.TILDETRIM], tildeTrimReplace); // `^ 1.2.3` => `^1.2.3` |
|
|
|
range = range.replace(re[t.CARETTRIM], caretTrimReplace); // normalize spaces |
|
|
|
range = range.split(/\s+/).join(' '); // At this point, the range is completely trimmed and |
|
// ready to be split into comparators. |
|
|
|
var compRe = loose ? re[t.COMPARATORLOOSE] : re[t.COMPARATOR]; |
|
var set = range.split(' ').map(function (comp) { |
|
return parseComparator(comp, this.options); |
|
}, this).join(' ').split(/\s+/); |
|
|
|
if (this.options.loose) { |
|
// in loose mode, throw out any that are not valid comparators |
|
set = set.filter(function (comp) { |
|
return !!comp.match(compRe); |
|
}); |
|
} |
|
|
|
set = set.map(function (comp) { |
|
return new Comparator(comp, this.options); |
|
}, this); |
|
return set; |
|
}; |
|
|
|
Range.prototype.intersects = function (range, options) { |
|
if (!(range instanceof Range)) { |
|
throw new TypeError('a Range is required'); |
|
} |
|
|
|
return this.set.some(function (thisComparators) { |
|
return isSatisfiable(thisComparators, options) && range.set.some(function (rangeComparators) { |
|
return isSatisfiable(rangeComparators, options) && thisComparators.every(function (thisComparator) { |
|
return rangeComparators.every(function (rangeComparator) { |
|
return thisComparator.intersects(rangeComparator, options); |
|
}); |
|
}); |
|
}); |
|
}); |
|
}; // take a set of comparators and determine whether there |
|
// exists a version which can satisfy it |
|
|
|
|
|
function isSatisfiable(comparators, options) { |
|
var result = true; |
|
var remainingComparators = comparators.slice(); |
|
var testComparator = remainingComparators.pop(); |
|
|
|
while (result && remainingComparators.length) { |
|
result = remainingComparators.every(function (otherComparator) { |
|
return testComparator.intersects(otherComparator, options); |
|
}); |
|
testComparator = remainingComparators.pop(); |
|
} |
|
|
|
return result; |
|
} // Mostly just for testing and legacy API reasons |
|
|
|
|
|
exports.toComparators = toComparators; |
|
|
|
function toComparators(range, options) { |
|
return new Range(range, options).set.map(function (comp) { |
|
return comp.map(function (c) { |
|
return c.value; |
|
}).join(' ').trim().split(' '); |
|
}); |
|
} // comprised of xranges, tildes, stars, and gtlt's at this point. |
|
// already replaced the hyphen ranges |
|
// turn into a set of JUST comparators. |
|
|
|
|
|
function parseComparator(comp, options) { |
|
debug('comp', comp, options); |
|
comp = replaceCarets(comp, options); |
|
debug('caret', comp); |
|
comp = replaceTildes(comp, options); |
|
debug('tildes', comp); |
|
comp = replaceXRanges(comp, options); |
|
debug('xrange', comp); |
|
comp = replaceStars(comp, options); |
|
debug('stars', comp); |
|
return comp; |
|
} |
|
|
|
function isX(id) { |
|
return !id || id.toLowerCase() === 'x' || id === '*'; |
|
} // ~, ~> --> * (any, kinda silly) |
|
// ~2, ~2.x, ~2.x.x, ~>2, ~>2.x ~>2.x.x --> >=2.0.0 <3.0.0 |
|
// ~2.0, ~2.0.x, ~>2.0, ~>2.0.x --> >=2.0.0 <2.1.0 |
|
// ~1.2, ~1.2.x, ~>1.2, ~>1.2.x --> >=1.2.0 <1.3.0 |
|
// ~1.2.3, ~>1.2.3 --> >=1.2.3 <1.3.0 |
|
// ~1.2.0, ~>1.2.0 --> >=1.2.0 <1.3.0 |
|
|
|
|
|
function replaceTildes(comp, options) { |
|
return comp.trim().split(/\s+/).map(function (comp) { |
|
return replaceTilde(comp, options); |
|
}).join(' '); |
|
} |
|
|
|
function replaceTilde(comp, options) { |
|
var r = options.loose ? re[t.TILDELOOSE] : re[t.TILDE]; |
|
return comp.replace(r, function (_, M, m, p, pr) { |
|
debug('tilde', comp, _, M, m, p, pr); |
|
var ret; |
|
|
|
if (isX(M)) { |
|
ret = ''; |
|
} else if (isX(m)) { |
|
ret = '>=' + M + '.0.0 <' + (+M + 1) + '.0.0'; |
|
} else if (isX(p)) { |
|
// ~1.2 == >=1.2.0 <1.3.0 |
|
ret = '>=' + M + '.' + m + '.0 <' + M + '.' + (+m + 1) + '.0'; |
|
} else if (pr) { |
|
debug('replaceTilde pr', pr); |
|
ret = '>=' + M + '.' + m + '.' + p + '-' + pr + ' <' + M + '.' + (+m + 1) + '.0'; |
|
} else { |
|
// ~1.2.3 == >=1.2.3 <1.3.0 |
|
ret = '>=' + M + '.' + m + '.' + p + ' <' + M + '.' + (+m + 1) + '.0'; |
|
} |
|
|
|
debug('tilde return', ret); |
|
return ret; |
|
}); |
|
} // ^ --> * (any, kinda silly) |
|
// ^2, ^2.x, ^2.x.x --> >=2.0.0 <3.0.0 |
|
// ^2.0, ^2.0.x --> >=2.0.0 <3.0.0 |
|
// ^1.2, ^1.2.x --> >=1.2.0 <2.0.0 |
|
// ^1.2.3 --> >=1.2.3 <2.0.0 |
|
// ^1.2.0 --> >=1.2.0 <2.0.0 |
|
|
|
|
|
function replaceCarets(comp, options) { |
|
return comp.trim().split(/\s+/).map(function (comp) { |
|
return replaceCaret(comp, options); |
|
}).join(' '); |
|
} |
|
|
|
function replaceCaret(comp, options) { |
|
debug('caret', comp, options); |
|
var r = options.loose ? re[t.CARETLOOSE] : re[t.CARET]; |
|
return comp.replace(r, function (_, M, m, p, pr) { |
|
debug('caret', comp, _, M, m, p, pr); |
|
var ret; |
|
|
|
if (isX(M)) { |
|
ret = ''; |
|
} else if (isX(m)) { |
|
ret = '>=' + M + '.0.0 <' + (+M + 1) + '.0.0'; |
|
} else if (isX(p)) { |
|
if (M === '0') { |
|
ret = '>=' + M + '.' + m + '.0 <' + M + '.' + (+m + 1) + '.0'; |
|
} else { |
|
ret = '>=' + M + '.' + m + '.0 <' + (+M + 1) + '.0.0'; |
|
} |
|
} else if (pr) { |
|
debug('replaceCaret pr', pr); |
|
|
|
if (M === '0') { |
|
if (m === '0') { |
|
ret = '>=' + M + '.' + m + '.' + p + '-' + pr + ' <' + M + '.' + m + '.' + (+p + 1); |
|
} else { |
|
ret = '>=' + M + '.' + m + '.' + p + '-' + pr + ' <' + M + '.' + (+m + 1) + '.0'; |
|
} |
|
} else { |
|
ret = '>=' + M + '.' + m + '.' + p + '-' + pr + ' <' + (+M + 1) + '.0.0'; |
|
} |
|
} else { |
|
debug('no pr'); |
|
|
|
if (M === '0') { |
|
if (m === '0') { |
|
ret = '>=' + M + '.' + m + '.' + p + ' <' + M + '.' + m + '.' + (+p + 1); |
|
} else { |
|
ret = '>=' + M + '.' + m + '.' + p + ' <' + M + '.' + (+m + 1) + '.0'; |
|
} |
|
} else { |
|
ret = '>=' + M + '.' + m + '.' + p + ' <' + (+M + 1) + '.0.0'; |
|
} |
|
} |
|
|
|
debug('caret return', ret); |
|
return ret; |
|
}); |
|
} |
|
|
|
function replaceXRanges(comp, options) { |
|
debug('replaceXRanges', comp, options); |
|
return comp.split(/\s+/).map(function (comp) { |
|
return replaceXRange(comp, options); |
|
}).join(' '); |
|
} |
|
|
|
function replaceXRange(comp, options) { |
|
comp = comp.trim(); |
|
var r = options.loose ? re[t.XRANGELOOSE] : re[t.XRANGE]; |
|
return comp.replace(r, function (ret, gtlt, M, m, p, pr) { |
|
debug('xRange', comp, ret, gtlt, M, m, p, pr); |
|
var xM = isX(M); |
|
var xm = xM || isX(m); |
|
var xp = xm || isX(p); |
|
var anyX = xp; |
|
|
|
if (gtlt === '=' && anyX) { |
|
gtlt = ''; |
|
} // if we're including prereleases in the match, then we need |
|
// to fix this to -0, the lowest possible prerelease value |
|
|
|
|
|
pr = options.includePrerelease ? '-0' : ''; |
|
|
|
if (xM) { |
|
if (gtlt === '>' || gtlt === '<') { |
|
// nothing is allowed |
|
ret = '<0.0.0-0'; |
|
} else { |
|
// nothing is forbidden |
|
ret = '*'; |
|
} |
|
} else if (gtlt && anyX) { |
|
// we know patch is an x, because we have any x at all. |
|
// replace X with 0 |
|
if (xm) { |
|
m = 0; |
|
} |
|
|
|
p = 0; |
|
|
|
if (gtlt === '>') { |
|
// >1 => >=2.0.0 |
|
// >1.2 => >=1.3.0 |
|
// >1.2.3 => >= 1.2.4 |
|
gtlt = '>='; |
|
|
|
if (xm) { |
|
M = +M + 1; |
|
m = 0; |
|
p = 0; |
|
} else { |
|
m = +m + 1; |
|
p = 0; |
|
} |
|
} else if (gtlt === '<=') { |
|
// <=0.7.x is actually <0.8.0, since any 0.7.x should |
|
// pass. Similarly, <=7.x is actually <8.0.0, etc. |
|
gtlt = '<'; |
|
|
|
if (xm) { |
|
M = +M + 1; |
|
} else { |
|
m = +m + 1; |
|
} |
|
} |
|
|
|
ret = gtlt + M + '.' + m + '.' + p + pr; |
|
} else if (xm) { |
|
ret = '>=' + M + '.0.0' + pr + ' <' + (+M + 1) + '.0.0' + pr; |
|
} else if (xp) { |
|
ret = '>=' + M + '.' + m + '.0' + pr + ' <' + M + '.' + (+m + 1) + '.0' + pr; |
|
} |
|
|
|
debug('xRange return', ret); |
|
return ret; |
|
}); |
|
} // Because * is AND-ed with everything else in the comparator, |
|
// and '' means "any version", just remove the *s entirely. |
|
|
|
|
|
function replaceStars(comp, options) { |
|
debug('replaceStars', comp, options); // Looseness is ignored here. star is always as loose as it gets! |
|
|
|
return comp.trim().replace(re[t.STAR], ''); |
|
} // This function is passed to string.replace(re[t.HYPHENRANGE]) |
|
// M, m, patch, prerelease, build |
|
// 1.2 - 3.4.5 => >=1.2.0 <=3.4.5 |
|
// 1.2.3 - 3.4 => >=1.2.0 <3.5.0 Any 3.4.x will do |
|
// 1.2 - 3.4 => >=1.2.0 <3.5.0 |
|
|
|
|
|
function hyphenReplace($0, from, fM, fm, fp, fpr, fb, to, tM, tm, tp, tpr, tb) { |
|
if (isX(fM)) { |
|
from = ''; |
|
} else if (isX(fm)) { |
|
from = '>=' + fM + '.0.0'; |
|
} else if (isX(fp)) { |
|
from = '>=' + fM + '.' + fm + '.0'; |
|
} else { |
|
from = '>=' + from; |
|
} |
|
|
|
if (isX(tM)) { |
|
to = ''; |
|
} else if (isX(tm)) { |
|
to = '<' + (+tM + 1) + '.0.0'; |
|
} else if (isX(tp)) { |
|
to = '<' + tM + '.' + (+tm + 1) + '.0'; |
|
} else if (tpr) { |
|
to = '<=' + tM + '.' + tm + '.' + tp + '-' + tpr; |
|
} else { |
|
to = '<=' + to; |
|
} |
|
|
|
return (from + ' ' + to).trim(); |
|
} // if ANY of the sets match ALL of its comparators, then pass |
|
|
|
|
|
Range.prototype.test = function (version) { |
|
if (!version) { |
|
return false; |
|
} |
|
|
|
if (typeof version === 'string') { |
|
try { |
|
version = new SemVer(version, this.options); |
|
} catch (er) { |
|
return false; |
|
} |
|
} |
|
|
|
for (var i = 0; i < this.set.length; i++) { |
|
if (testSet(this.set[i], version, this.options)) { |
|
return true; |
|
} |
|
} |
|
|
|
return false; |
|
}; |
|
|
|
function testSet(set, version, options) { |
|
for (var i = 0; i < set.length; i++) { |
|
if (!set[i].test(version)) { |
|
return false; |
|
} |
|
} |
|
|
|
if (version.prerelease.length && !options.includePrerelease) { |
|
// Find the set of versions that are allowed to have prereleases |
|
// For example, ^1.2.3-pr.1 desugars to >=1.2.3-pr.1 <2.0.0 |
|
// That should allow `1.2.3-pr.2` to pass. |
|
// However, `1.2.4-alpha.notready` should NOT be allowed, |
|
// even though it's within the range set by the comparators. |
|
for (i = 0; i < set.length; i++) { |
|
debug(set[i].semver); |
|
|
|
if (set[i].semver === ANY) { |
|
continue; |
|
} |
|
|
|
if (set[i].semver.prerelease.length > 0) { |
|
var allowed = set[i].semver; |
|
|
|
if (allowed.major === version.major && allowed.minor === version.minor && allowed.patch === version.patch) { |
|
return true; |
|
} |
|
} |
|
} // Version has a -pre, but it's not one of the ones we like. |
|
|
|
|
|
return false; |
|
} |
|
|
|
return true; |
|
} |
|
|
|
exports.satisfies = satisfies; |
|
|
|
function satisfies(version, range, options) { |
|
try { |
|
range = new Range(range, options); |
|
} catch (er) { |
|
return false; |
|
} |
|
|
|
return range.test(version); |
|
} |
|
|
|
exports.maxSatisfying = maxSatisfying; |
|
|
|
function maxSatisfying(versions, range, options) { |
|
var max = null; |
|
var maxSV = null; |
|
|
|
try { |
|
var rangeObj = new Range(range, options); |
|
} catch (er) { |
|
return null; |
|
} |
|
|
|
versions.forEach(function (v) { |
|
if (rangeObj.test(v)) { |
|
// satisfies(v, range, options) |
|
if (!max || maxSV.compare(v) === -1) { |
|
// compare(max, v, true) |
|
max = v; |
|
maxSV = new SemVer(max, options); |
|
} |
|
} |
|
}); |
|
return max; |
|
} |
|
|
|
exports.minSatisfying = minSatisfying; |
|
|
|
function minSatisfying(versions, range, options) { |
|
var min = null; |
|
var minSV = null; |
|
|
|
try { |
|
var rangeObj = new Range(range, options); |
|
} catch (er) { |
|
return null; |
|
} |
|
|
|
versions.forEach(function (v) { |
|
if (rangeObj.test(v)) { |
|
// satisfies(v, range, options) |
|
if (!min || minSV.compare(v) === 1) { |
|
// compare(min, v, true) |
|
min = v; |
|
minSV = new SemVer(min, options); |
|
} |
|
} |
|
}); |
|
return min; |
|
} |
|
|
|
exports.minVersion = minVersion; |
|
|
|
function minVersion(range, loose) { |
|
range = new Range(range, loose); |
|
var minver = new SemVer('0.0.0'); |
|
|
|
if (range.test(minver)) { |
|
return minver; |
|
} |
|
|
|
minver = new SemVer('0.0.0-0'); |
|
|
|
if (range.test(minver)) { |
|
return minver; |
|
} |
|
|
|
minver = null; |
|
|
|
for (var i = 0; i < range.set.length; ++i) { |
|
var comparators = range.set[i]; |
|
comparators.forEach(function (comparator) { |
|
// Clone to avoid manipulating the comparator's semver object. |
|
var compver = new SemVer(comparator.semver.version); |
|
|
|
switch (comparator.operator) { |
|
case '>': |
|
if (compver.prerelease.length === 0) { |
|
compver.patch++; |
|
} else { |
|
compver.prerelease.push(0); |
|
} |
|
|
|
compver.raw = compver.format(); |
|
|
|
/* fallthrough */ |
|
|
|
case '': |
|
case '>=': |
|
if (!minver || gt(minver, compver)) { |
|
minver = compver; |
|
} |
|
|
|
break; |
|
|
|
case '<': |
|
case '<=': |
|
/* Ignore maximum versions */ |
|
break; |
|
|
|
/* istanbul ignore next */ |
|
|
|
default: |
|
throw new Error('Unexpected operation: ' + comparator.operator); |
|
} |
|
}); |
|
} |
|
|
|
if (minver && range.test(minver)) { |
|
return minver; |
|
} |
|
|
|
return null; |
|
} |
|
|
|
exports.validRange = validRange; |
|
|
|
function validRange(range, options) { |
|
try { |
|
// Return '*' instead of '' so that truthiness works. |
|
// This will throw if it's invalid anyway |
|
return new Range(range, options).range || '*'; |
|
} catch (er) { |
|
return null; |
|
} |
|
} // Determine if version is less than all the versions possible in the range |
|
|
|
|
|
exports.ltr = ltr; |
|
|
|
function ltr(version, range, options) { |
|
return outside(version, range, '<', options); |
|
} // Determine if version is greater than all the versions possible in the range. |
|
|
|
|
|
exports.gtr = gtr; |
|
|
|
function gtr(version, range, options) { |
|
return outside(version, range, '>', options); |
|
} |
|
|
|
exports.outside = outside; |
|
|
|
function outside(version, range, hilo, options) { |
|
version = new SemVer(version, options); |
|
range = new Range(range, options); |
|
var gtfn, ltefn, ltfn, comp, ecomp; |
|
|
|
switch (hilo) { |
|
case '>': |
|
gtfn = gt; |
|
ltefn = lte; |
|
ltfn = lt; |
|
comp = '>'; |
|
ecomp = '>='; |
|
break; |
|
|
|
case '<': |
|
gtfn = lt; |
|
ltefn = gte; |
|
ltfn = gt; |
|
comp = '<'; |
|
ecomp = '<='; |
|
break; |
|
|
|
default: |
|
throw new TypeError('Must provide a hilo val of "<" or ">"'); |
|
} // If it satisifes the range it is not outside |
|
|
|
|
|
if (satisfies(version, range, options)) { |
|
return false; |
|
} // From now on, variable terms are as if we're in "gtr" mode. |
|
// but note that everything is flipped for the "ltr" function. |
|
|
|
|
|
for (var i = 0; i < range.set.length; ++i) { |
|
var comparators = range.set[i]; |
|
var high = null; |
|
var low = null; |
|
comparators.forEach(function (comparator) { |
|
if (comparator.semver === ANY) { |
|
comparator = new Comparator('>=0.0.0'); |
|
} |
|
|
|
high = high || comparator; |
|
low = low || comparator; |
|
|
|
if (gtfn(comparator.semver, high.semver, options)) { |
|
high = comparator; |
|
} else if (ltfn(comparator.semver, low.semver, options)) { |
|
low = comparator; |
|
} |
|
}); // If the edge version comparator has a operator then our version |
|
// isn't outside it |
|
|
|
if (high.operator === comp || high.operator === ecomp) { |
|
return false; |
|
} // If the lowest version comparator has an operator and our version |
|
// is less than it then it isn't higher than the range |
|
|
|
|
|
if ((!low.operator || low.operator === comp) && ltefn(version, low.semver)) { |
|
return false; |
|
} else if (low.operator === ecomp && ltfn(version, low.semver)) { |
|
return false; |
|
} |
|
} |
|
|
|
return true; |
|
} |
|
|
|
exports.prerelease = prerelease; |
|
|
|
function prerelease(version, options) { |
|
var parsed = parse(version, options); |
|
return parsed && parsed.prerelease.length ? parsed.prerelease : null; |
|
} |
|
|
|
exports.intersects = intersects; |
|
|
|
function intersects(r1, r2, options) { |
|
r1 = new Range(r1, options); |
|
r2 = new Range(r2, options); |
|
return r1.intersects(r2); |
|
} |
|
|
|
exports.coerce = coerce; |
|
|
|
function coerce(version, options) { |
|
if (version instanceof SemVer) { |
|
return version; |
|
} |
|
|
|
if (typeof version === 'number') { |
|
version = String(version); |
|
} |
|
|
|
if (typeof version !== 'string') { |
|
return null; |
|
} |
|
|
|
options = options || {}; |
|
var match = null; |
|
|
|
if (!options.rtl) { |
|
match = version.match(re[t.COERCE]); |
|
} else { |
|
// Find the right-most coercible string that does not share |
|
// a terminus with a more left-ward coercible string. |
|
// Eg, '1.2.3.4' wants to coerce '2.3.4', not '3.4' or '4' |
|
// |
|
// Walk through the string checking with a /g regexp |
|
// Manually set the index so as to pick up overlapping matches. |
|
// Stop when we get a match that ends at the string end, since no |
|
// coercible string can be more right-ward without the same terminus. |
|
var next; |
|
|
|
while ((next = re[t.COERCERTL].exec(version)) && (!match || match.index + match[0].length !== version.length)) { |
|
if (!match || next.index + next[0].length !== match.index + match[0].length) { |
|
match = next; |
|
} |
|
|
|
re[t.COERCERTL].lastIndex = next.index + next[1].length + next[2].length; |
|
} // leave it in a clean state |
|
|
|
|
|
re[t.COERCERTL].lastIndex = -1; |
|
} |
|
|
|
if (match === null) { |
|
return null; |
|
} |
|
|
|
return parse(match[2] + '.' + (match[3] || '0') + '.' + (match[4] || '0'), options); |
|
} |
|
}); |
|
var semver_1 = semver.SEMVER_SPEC_VERSION; |
|
var semver_2 = semver.re; |
|
var semver_3 = semver.src; |
|
var semver_4 = semver.tokens; |
|
var semver_5 = semver.parse; |
|
var semver_6 = semver.valid; |
|
var semver_7 = semver.clean; |
|
var semver_8 = semver.SemVer; |
|
var semver_9 = semver.inc; |
|
var semver_10 = semver.diff; |
|
var semver_11 = semver.compareIdentifiers; |
|
var semver_12 = semver.rcompareIdentifiers; |
|
var semver_13 = semver.major; |
|
var semver_14 = semver.minor; |
|
var semver_15 = semver.patch; |
|
var semver_16 = semver.compare; |
|
var semver_17 = semver.compareLoose; |
|
var semver_18 = semver.compareBuild; |
|
var semver_19 = semver.rcompare; |
|
var semver_20 = semver.sort; |
|
var semver_21 = semver.rsort; |
|
var semver_22 = semver.gt; |
|
var semver_23 = semver.lt; |
|
var semver_24 = semver.eq; |
|
var semver_25 = semver.neq; |
|
var semver_26 = semver.gte; |
|
var semver_27 = semver.lte; |
|
var semver_28 = semver.cmp; |
|
var semver_29 = semver.Comparator; |
|
var semver_30 = semver.Range; |
|
var semver_31 = semver.toComparators; |
|
var semver_32 = semver.satisfies; |
|
var semver_33 = semver.maxSatisfying; |
|
var semver_34 = semver.minSatisfying; |
|
var semver_35 = semver.minVersion; |
|
var semver_36 = semver.validRange; |
|
var semver_37 = semver.ltr; |
|
var semver_38 = semver.gtr; |
|
var semver_39 = semver.outside; |
|
var semver_40 = semver.prerelease; |
|
var semver_41 = semver.intersects; |
|
var semver_42 = semver.coerce; |
|
|
|
var arrayify = function arrayify(object, keyName) { |
|
return Object.keys(object).reduce(function (array, key) { |
|
return array.concat(Object.assign(_defineProperty({}, keyName, key), object[key])); |
|
}, []); |
|
}; |
|
|
|
var dedent_1 = createCommonjsModule(function (module) { |
|
|
|
function dedent(strings) { |
|
var raw = void 0; |
|
|
|
if (typeof strings === "string") { |
|
// dedent can be used as a plain function |
|
raw = [strings]; |
|
} else { |
|
raw = strings.raw; |
|
} // first, perform interpolation |
|
|
|
|
|
var result = ""; |
|
|
|
for (var i = 0; i < raw.length; i++) { |
|
result += raw[i]. // join lines when there is a suppressed newline |
|
replace(/\\\n[ \t]*/g, ""). // handle escaped backticks |
|
replace(/\\`/g, "`"); |
|
|
|
if (i < (arguments.length <= 1 ? 0 : arguments.length - 1)) { |
|
result += arguments.length <= i + 1 ? undefined : arguments[i + 1]; |
|
} |
|
} // now strip indentation |
|
|
|
|
|
var lines = result.split("\n"); |
|
var mindent = null; |
|
lines.forEach(function (l) { |
|
var m = l.match(/^(\s+)\S+/); |
|
|
|
if (m) { |
|
var indent = m[1].length; |
|
|
|
if (!mindent) { |
|
// this is the first indented line |
|
mindent = indent; |
|
} else { |
|
mindent = Math.min(mindent, indent); |
|
} |
|
} |
|
}); |
|
|
|
if (mindent !== null) { |
|
result = lines.map(function (l) { |
|
return l[0] === " " ? l.slice(mindent) : l; |
|
}).join("\n"); |
|
} // dedent eats leading and trailing whitespace too |
|
|
|
|
|
result = result.trim(); // handle escaped newlines at the end to ensure they don't get stripped too |
|
|
|
return result.replace(/\\n/g, "\n"); |
|
} |
|
|
|
{ |
|
module.exports = dedent; |
|
} |
|
}); |
|
|
|
function _templateObject6() { |
|
var data = _taggedTemplateLiteral(["\n Require either '@prettier' or '@format' to be present in the file's first docblock comment\n in order for it to be formatted.\n "]); |
|
|
|
_templateObject6 = function _templateObject6() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _templateObject5() { |
|
var data = _taggedTemplateLiteral(["\n Format code starting at a given character offset.\n The range will extend backwards to the start of the first line containing the selected statement.\n This option cannot be used with --cursor-offset.\n "]); |
|
|
|
_templateObject5 = function _templateObject5() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _templateObject4() { |
|
var data = _taggedTemplateLiteral(["\n Format code ending at a given character offset (exclusive).\n The range will extend forwards to the end of the selected statement.\n This option cannot be used with --cursor-offset.\n "]); |
|
|
|
_templateObject4 = function _templateObject4() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _templateObject3() { |
|
var data = _taggedTemplateLiteral(["\n Custom directory that contains prettier plugins in node_modules subdirectory.\n Overrides default behavior when plugins are searched relatively to the location of Prettier.\n Multiple values are accepted.\n "]); |
|
|
|
_templateObject3 = function _templateObject3() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _templateObject2() { |
|
var data = _taggedTemplateLiteral(["\n Maintain existing\n (mixed values within one file are normalised by looking at what's used after the first line)\n "]); |
|
|
|
_templateObject2 = function _templateObject2() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _templateObject() { |
|
var data = _taggedTemplateLiteral(["\n Print (to stderr) where a cursor at the given position would move to after formatting.\n This option cannot be used with --range-start and --range-end.\n "]); |
|
|
|
_templateObject = function _templateObject() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
var CATEGORY_CONFIG = "Config"; |
|
var CATEGORY_EDITOR = "Editor"; |
|
var CATEGORY_FORMAT = "Format"; |
|
var CATEGORY_OTHER = "Other"; |
|
var CATEGORY_OUTPUT = "Output"; |
|
var CATEGORY_GLOBAL = "Global"; |
|
var CATEGORY_SPECIAL = "Special"; |
|
/** |
|
* @typedef {Object} OptionInfo |
|
* @property {string} [since] - available since version |
|
* @property {string} category |
|
* @property {'int' | 'boolean' | 'choice' | 'path'} type |
|
* @property {boolean} [array] - indicate it's an array of the specified type |
|
* @property {OptionValueInfo} [default] |
|
* @property {OptionRangeInfo} [range] - for type int |
|
* @property {string} description |
|
* @property {string} [deprecated] - deprecated since version |
|
* @property {OptionRedirectInfo} [redirect] - redirect deprecated option |
|
* @property {(value: any) => boolean} [exception] |
|
* @property {OptionChoiceInfo[]} [choices] - for type choice |
|
* @property {string} [cliName] |
|
* @property {string} [cliCategory] |
|
* @property {string} [cliDescription] |
|
* |
|
* @typedef {number | boolean | string} OptionValue |
|
* @typedef {OptionValue | [{ value: OptionValue[] }] | Array<{ since: string, value: OptionValue}>} OptionValueInfo |
|
* |
|
* @typedef {Object} OptionRedirectInfo |
|
* @property {string} option |
|
* @property {OptionValue} value |
|
* |
|
* @typedef {Object} OptionRangeInfo |
|
* @property {number} start - recommended range start |
|
* @property {number} end - recommended range end |
|
* @property {number} step - recommended range step |
|
* |
|
* @typedef {Object} OptionChoiceInfo |
|
* @property {boolean | string} value - boolean for the option that is originally boolean type |
|
* @property {string} description |
|
* @property {string} [since] - undefined if available since the first version of the option |
|
* @property {string} [deprecated] - deprecated since version |
|
* @property {OptionValueInfo} [redirect] - redirect deprecated value |
|
*/ |
|
|
|
/** @type {{ [name: string]: OptionInfo }} */ |
|
|
|
var options = { |
|
cursorOffset: { |
|
since: "1.4.0", |
|
category: CATEGORY_SPECIAL, |
|
type: "int", |
|
default: -1, |
|
range: { |
|
start: -1, |
|
end: Infinity, |
|
step: 1 |
|
}, |
|
description: dedent_1(_templateObject()), |
|
cliCategory: CATEGORY_EDITOR |
|
}, |
|
endOfLine: { |
|
since: "1.15.0", |
|
category: CATEGORY_GLOBAL, |
|
type: "choice", |
|
default: "auto", |
|
description: "Which end of line characters to apply.", |
|
choices: [{ |
|
value: "auto", |
|
description: dedent_1(_templateObject2()) |
|
}, { |
|
value: "lf", |
|
description: "Line Feed only (\\n), common on Linux and macOS as well as inside git repos" |
|
}, { |
|
value: "crlf", |
|
description: "Carriage Return + Line Feed characters (\\r\\n), common on Windows" |
|
}, { |
|
value: "cr", |
|
description: "Carriage Return character only (\\r), used very rarely" |
|
}] |
|
}, |
|
filepath: { |
|
since: "1.4.0", |
|
category: CATEGORY_SPECIAL, |
|
type: "path", |
|
description: "Specify the input filepath. This will be used to do parser inference.", |
|
cliName: "stdin-filepath", |
|
cliCategory: CATEGORY_OTHER, |
|
cliDescription: "Path to the file to pretend that stdin comes from." |
|
}, |
|
insertPragma: { |
|
since: "1.8.0", |
|
category: CATEGORY_SPECIAL, |
|
type: "boolean", |
|
default: false, |
|
description: "Insert @format pragma into file's first docblock comment.", |
|
cliCategory: CATEGORY_OTHER |
|
}, |
|
parser: { |
|
since: "0.0.10", |
|
category: CATEGORY_GLOBAL, |
|
type: "choice", |
|
default: [{ |
|
since: "0.0.10", |
|
value: "babylon" |
|
}, { |
|
since: "1.13.0", |
|
value: undefined |
|
}], |
|
description: "Which parser to use.", |
|
exception: function exception(value) { |
|
return typeof value === "string" || typeof value === "function"; |
|
}, |
|
choices: [{ |
|
value: "flow", |
|
description: "Flow" |
|
}, { |
|
value: "babylon", |
|
description: "JavaScript", |
|
deprecated: "1.16.0", |
|
redirect: "babel" |
|
}, { |
|
value: "babel", |
|
since: "1.16.0", |
|
description: "JavaScript" |
|
}, { |
|
value: "babel-flow", |
|
since: "1.16.0", |
|
description: "Flow" |
|
}, { |
|
value: "typescript", |
|
since: "1.4.0", |
|
description: "TypeScript" |
|
}, { |
|
value: "css", |
|
since: "1.7.1", |
|
description: "CSS" |
|
}, { |
|
value: "postcss", |
|
since: "1.4.0", |
|
description: "CSS/Less/SCSS", |
|
deprecated: "1.7.1", |
|
redirect: "css" |
|
}, { |
|
value: "less", |
|
since: "1.7.1", |
|
description: "Less" |
|
}, { |
|
value: "scss", |
|
since: "1.7.1", |
|
description: "SCSS" |
|
}, { |
|
value: "json", |
|
since: "1.5.0", |
|
description: "JSON" |
|
}, { |
|
value: "json5", |
|
since: "1.13.0", |
|
description: "JSON5" |
|
}, { |
|
value: "json-stringify", |
|
since: "1.13.0", |
|
description: "JSON.stringify" |
|
}, { |
|
value: "graphql", |
|
since: "1.5.0", |
|
description: "GraphQL" |
|
}, { |
|
value: "markdown", |
|
since: "1.8.0", |
|
description: "Markdown" |
|
}, { |
|
value: "mdx", |
|
since: "1.15.0", |
|
description: "MDX" |
|
}, { |
|
value: "vue", |
|
since: "1.10.0", |
|
description: "Vue" |
|
}, { |
|
value: "yaml", |
|
since: "1.14.0", |
|
description: "YAML" |
|
}, { |
|
value: "glimmer", |
|
since: null, |
|
description: "Handlebars" |
|
}, { |
|
value: "html", |
|
since: "1.15.0", |
|
description: "HTML" |
|
}, { |
|
value: "angular", |
|
since: "1.15.0", |
|
description: "Angular" |
|
}, { |
|
value: "lwc", |
|
since: "1.17.0", |
|
description: "Lightning Web Components" |
|
}] |
|
}, |
|
plugins: { |
|
since: "1.10.0", |
|
type: "path", |
|
array: true, |
|
default: [{ |
|
value: [] |
|
}], |
|
category: CATEGORY_GLOBAL, |
|
description: "Add a plugin. Multiple plugins can be passed as separate `--plugin`s.", |
|
exception: function exception(value) { |
|
return typeof value === "string" || _typeof(value) === "object"; |
|
}, |
|
cliName: "plugin", |
|
cliCategory: CATEGORY_CONFIG |
|
}, |
|
pluginSearchDirs: { |
|
since: "1.13.0", |
|
type: "path", |
|
array: true, |
|
default: [{ |
|
value: [] |
|
}], |
|
category: CATEGORY_GLOBAL, |
|
description: dedent_1(_templateObject3()), |
|
exception: function exception(value) { |
|
return typeof value === "string" || _typeof(value) === "object"; |
|
}, |
|
cliName: "plugin-search-dir", |
|
cliCategory: CATEGORY_CONFIG |
|
}, |
|
printWidth: { |
|
since: "0.0.0", |
|
category: CATEGORY_GLOBAL, |
|
type: "int", |
|
default: 80, |
|
description: "The line length where Prettier will try wrap.", |
|
range: { |
|
start: 0, |
|
end: Infinity, |
|
step: 1 |
|
} |
|
}, |
|
rangeEnd: { |
|
since: "1.4.0", |
|
category: CATEGORY_SPECIAL, |
|
type: "int", |
|
default: Infinity, |
|
range: { |
|
start: 0, |
|
end: Infinity, |
|
step: 1 |
|
}, |
|
description: dedent_1(_templateObject4()), |
|
cliCategory: CATEGORY_EDITOR |
|
}, |
|
rangeStart: { |
|
since: "1.4.0", |
|
category: CATEGORY_SPECIAL, |
|
type: "int", |
|
default: 0, |
|
range: { |
|
start: 0, |
|
end: Infinity, |
|
step: 1 |
|
}, |
|
description: dedent_1(_templateObject5()), |
|
cliCategory: CATEGORY_EDITOR |
|
}, |
|
requirePragma: { |
|
since: "1.7.0", |
|
category: CATEGORY_SPECIAL, |
|
type: "boolean", |
|
default: false, |
|
description: dedent_1(_templateObject6()), |
|
cliCategory: CATEGORY_OTHER |
|
}, |
|
tabWidth: { |
|
type: "int", |
|
category: CATEGORY_GLOBAL, |
|
default: 2, |
|
description: "Number of spaces per indentation level.", |
|
range: { |
|
start: 0, |
|
end: Infinity, |
|
step: 1 |
|
} |
|
}, |
|
useFlowParser: { |
|
since: "0.0.0", |
|
category: CATEGORY_GLOBAL, |
|
type: "boolean", |
|
default: [{ |
|
since: "0.0.0", |
|
value: false |
|
}, { |
|
since: "1.15.0", |
|
value: undefined |
|
}], |
|
deprecated: "0.0.10", |
|
description: "Use flow parser.", |
|
redirect: { |
|
option: "parser", |
|
value: "flow" |
|
}, |
|
cliName: "flow-parser" |
|
}, |
|
useTabs: { |
|
since: "1.0.0", |
|
category: CATEGORY_GLOBAL, |
|
type: "boolean", |
|
default: false, |
|
description: "Indent with tabs instead of spaces." |
|
} |
|
}; |
|
var coreOptions = { |
|
CATEGORY_CONFIG: CATEGORY_CONFIG, |
|
CATEGORY_EDITOR: CATEGORY_EDITOR, |
|
CATEGORY_FORMAT: CATEGORY_FORMAT, |
|
CATEGORY_OTHER: CATEGORY_OTHER, |
|
CATEGORY_OUTPUT: CATEGORY_OUTPUT, |
|
CATEGORY_GLOBAL: CATEGORY_GLOBAL, |
|
CATEGORY_SPECIAL: CATEGORY_SPECIAL, |
|
options: options |
|
}; |
|
|
|
var require$$0 = getCjsExportFromNamespace(_package$1); |
|
|
|
var currentVersion = require$$0.version; |
|
var coreOptions$1 = coreOptions.options; |
|
|
|
function getSupportInfo(version, opts) { |
|
opts = Object.assign({ |
|
plugins: [], |
|
showUnreleased: false, |
|
showDeprecated: false, |
|
showInternal: false |
|
}, opts); |
|
|
|
if (!version) { |
|
// pre-release version is smaller than the normal version in semver, |
|
// we need to treat it as the normal one so as to test new features. |
|
version = currentVersion.split("-", 1)[0]; |
|
} |
|
|
|
var plugins = opts.plugins; |
|
var options = arrayify(Object.assign(plugins.reduce(function (currentOptions, plugin) { |
|
return Object.assign(currentOptions, plugin.options); |
|
}, {}), coreOptions$1), "name").sort(function (a, b) { |
|
return a.name === b.name ? 0 : a.name < b.name ? -1 : 1; |
|
}).filter(filterSince).filter(filterDeprecated).map(mapDeprecated).map(mapInternal).map(function (option) { |
|
var newOption = Object.assign({}, option); |
|
|
|
if (Array.isArray(newOption.default)) { |
|
newOption.default = newOption.default.length === 1 ? newOption.default[0].value : newOption.default.filter(filterSince).sort(function (info1, info2) { |
|
return semver.compare(info2.since, info1.since); |
|
})[0].value; |
|
} |
|
|
|
if (Array.isArray(newOption.choices)) { |
|
newOption.choices = newOption.choices.filter(filterSince).filter(filterDeprecated).map(mapDeprecated); |
|
} |
|
|
|
return newOption; |
|
}).map(function (option) { |
|
var filteredPlugins = plugins.filter(function (plugin) { |
|
return plugin.defaultOptions && plugin.defaultOptions[option.name] !== undefined; |
|
}); |
|
var pluginDefaults = filteredPlugins.reduce(function (reduced, plugin) { |
|
reduced[plugin.name] = plugin.defaultOptions[option.name]; |
|
return reduced; |
|
}, {}); |
|
return Object.assign(option, { |
|
pluginDefaults: pluginDefaults |
|
}); |
|
}); |
|
var usePostCssParser = semver.lt(version, "1.7.1"); |
|
var useBabylonParser = semver.lt(version, "1.16.0"); |
|
var languages = plugins.reduce(function (all, plugin) { |
|
return all.concat(plugin.languages || []); |
|
}, []).filter(filterSince).map(function (language) { |
|
// Prevent breaking changes |
|
if (language.name === "Markdown") { |
|
return Object.assign({}, language, { |
|
parsers: ["markdown"] |
|
}); |
|
} |
|
|
|
if (language.name === "TypeScript") { |
|
return Object.assign({}, language, { |
|
parsers: ["typescript"] |
|
}); |
|
} // "babylon" was renamed to "babel" in 1.16.0 |
|
|
|
|
|
if (useBabylonParser && language.parsers.indexOf("babel") !== -1) { |
|
return Object.assign({}, language, { |
|
parsers: language.parsers.map(function (parser) { |
|
return parser === "babel" ? "babylon" : parser; |
|
}) |
|
}); |
|
} |
|
|
|
if (usePostCssParser && (language.name === "CSS" || language.group === "CSS")) { |
|
return Object.assign({}, language, { |
|
parsers: ["postcss"] |
|
}); |
|
} |
|
|
|
return language; |
|
}); |
|
return { |
|
languages: languages, |
|
options: options |
|
}; |
|
|
|
function filterSince(object) { |
|
return opts.showUnreleased || !("since" in object) || object.since && semver.gte(version, object.since); |
|
} |
|
|
|
function filterDeprecated(object) { |
|
return opts.showDeprecated || !("deprecated" in object) || object.deprecated && semver.lt(version, object.deprecated); |
|
} |
|
|
|
function mapDeprecated(object) { |
|
if (!object.deprecated || opts.showDeprecated) { |
|
return object; |
|
} |
|
|
|
var newObject = Object.assign({}, object); |
|
delete newObject.deprecated; |
|
delete newObject.redirect; |
|
return newObject; |
|
} |
|
|
|
function mapInternal(object) { |
|
if (opts.showInternal) { |
|
return object; |
|
} |
|
|
|
var newObject = Object.assign({}, object); |
|
delete newObject.cliName; |
|
delete newObject.cliCategory; |
|
delete newObject.cliDescription; |
|
return newObject; |
|
} |
|
} |
|
|
|
var support = { |
|
getSupportInfo: getSupportInfo |
|
}; |
|
|
|
/*! *****************************************************************************
|
|
Copyright (c) Microsoft Corporation. All rights reserved.
|
|
Licensed under the Apache License, Version 2.0 (the "License"); you may not use
|
|
this file except in compliance with the License. You may obtain a copy of the
|
|
License at http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
THIS CODE IS PROVIDED ON AN *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
|
KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED
|
|
WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE,
|
|
MERCHANTABLITY OR NON-INFRINGEMENT.
|
|
|
|
See the Apache Version 2.0 License for specific language governing permissions
|
|
and limitations under the License.
|
|
***************************************************************************** */ |
|
|
|
/* global Reflect, Promise */ |
|
var _extendStatics = function extendStatics(d, b) { |
|
_extendStatics = Object.setPrototypeOf || { |
|
__proto__: [] |
|
} instanceof Array && function (d, b) { |
|
d.__proto__ = b; |
|
} || function (d, b) { |
|
for (var p in b) { |
|
if (b.hasOwnProperty(p)) d[p] = b[p]; |
|
} |
|
}; |
|
|
|
return _extendStatics(d, b); |
|
}; |
|
|
|
function __extends(d, b) { |
|
_extendStatics(d, b); |
|
|
|
function __() { |
|
this.constructor = d; |
|
} |
|
|
|
d.prototype = b === null ? Object.create(b) : (__.prototype = b.prototype, new __()); |
|
} |
|
|
|
var _assign = function __assign() { |
|
_assign = Object.assign || function __assign(t) { |
|
for (var s, i = 1, n = arguments.length; i < n; i++) { |
|
s = arguments[i]; |
|
|
|
for (var p in s) { |
|
if (Object.prototype.hasOwnProperty.call(s, p)) t[p] = s[p]; |
|
} |
|
} |
|
|
|
return t; |
|
}; |
|
|
|
return _assign.apply(this, arguments); |
|
}; |
|
function __rest(s, e) { |
|
var t = {}; |
|
|
|
for (var p in s) { |
|
if (Object.prototype.hasOwnProperty.call(s, p) && e.indexOf(p) < 0) t[p] = s[p]; |
|
} |
|
|
|
if (s != null && typeof Object.getOwnPropertySymbols === "function") for (var i = 0, p = Object.getOwnPropertySymbols(s); i < p.length; i++) { |
|
if (e.indexOf(p[i]) < 0 && Object.prototype.propertyIsEnumerable.call(s, p[i])) t[p[i]] = s[p[i]]; |
|
} |
|
return t; |
|
} |
|
function __decorate(decorators, target, key, desc) { |
|
var c = arguments.length, |
|
r = c < 3 ? target : desc === null ? desc = Object.getOwnPropertyDescriptor(target, key) : desc, |
|
d; |
|
if ((typeof Reflect === "undefined" ? "undefined" : _typeof(Reflect)) === "object" && typeof Reflect.decorate === "function") r = Reflect.decorate(decorators, target, key, desc);else for (var i = decorators.length - 1; i >= 0; i--) { |
|
if (d = decorators[i]) r = (c < 3 ? d(r) : c > 3 ? d(target, key, r) : d(target, key)) || r; |
|
} |
|
return c > 3 && r && Object.defineProperty(target, key, r), r; |
|
} |
|
function __param(paramIndex, decorator) { |
|
return function (target, key) { |
|
decorator(target, key, paramIndex); |
|
}; |
|
} |
|
function __metadata(metadataKey, metadataValue) { |
|
if ((typeof Reflect === "undefined" ? "undefined" : _typeof(Reflect)) === "object" && typeof Reflect.metadata === "function") return Reflect.metadata(metadataKey, metadataValue); |
|
} |
|
function __awaiter(thisArg, _arguments, P, generator) { |
|
return new (P || (P = Promise))(function (resolve, reject) { |
|
function fulfilled(value) { |
|
try { |
|
step(generator.next(value)); |
|
} catch (e) { |
|
reject(e); |
|
} |
|
} |
|
|
|
function rejected(value) { |
|
try { |
|
step(generator["throw"](value)); |
|
} catch (e) { |
|
reject(e); |
|
} |
|
} |
|
|
|
function step(result) { |
|
result.done ? resolve(result.value) : new P(function (resolve) { |
|
resolve(result.value); |
|
}).then(fulfilled, rejected); |
|
} |
|
|
|
step((generator = generator.apply(thisArg, _arguments || [])).next()); |
|
}); |
|
} |
|
function __generator(thisArg, body) { |
|
var _ = { |
|
label: 0, |
|
sent: function sent() { |
|
if (t[0] & 1) throw t[1]; |
|
return t[1]; |
|
}, |
|
trys: [], |
|
ops: [] |
|
}, |
|
f, |
|
y, |
|
t, |
|
g; |
|
return g = { |
|
next: verb(0), |
|
"throw": verb(1), |
|
"return": verb(2) |
|
}, typeof Symbol === "function" && (g[Symbol.iterator] = function () { |
|
return this; |
|
}), g; |
|
|
|
function verb(n) { |
|
return function (v) { |
|
return step([n, v]); |
|
}; |
|
} |
|
|
|
function step(op) { |
|
if (f) throw new TypeError("Generator is already executing."); |
|
|
|
while (_) { |
|
try { |
|
if (f = 1, y && (t = op[0] & 2 ? y["return"] : op[0] ? y["throw"] || ((t = y["return"]) && t.call(y), 0) : y.next) && !(t = t.call(y, op[1])).done) return t; |
|
if (y = 0, t) op = [op[0] & 2, t.value]; |
|
|
|
switch (op[0]) { |
|
case 0: |
|
case 1: |
|
t = op; |
|
break; |
|
|
|
case 4: |
|
_.label++; |
|
return { |
|
value: op[1], |
|
done: false |
|
}; |
|
|
|
case 5: |
|
_.label++; |
|
y = op[1]; |
|
op = [0]; |
|
continue; |
|
|
|
case 7: |
|
op = _.ops.pop(); |
|
|
|
_.trys.pop(); |
|
|
|
continue; |
|
|
|
default: |
|
if (!(t = _.trys, t = t.length > 0 && t[t.length - 1]) && (op[0] === 6 || op[0] === 2)) { |
|
_ = 0; |
|
continue; |
|
} |
|
|
|
if (op[0] === 3 && (!t || op[1] > t[0] && op[1] < t[3])) { |
|
_.label = op[1]; |
|
break; |
|
} |
|
|
|
if (op[0] === 6 && _.label < t[1]) { |
|
_.label = t[1]; |
|
t = op; |
|
break; |
|
} |
|
|
|
if (t && _.label < t[2]) { |
|
_.label = t[2]; |
|
|
|
_.ops.push(op); |
|
|
|
break; |
|
} |
|
|
|
if (t[2]) _.ops.pop(); |
|
|
|
_.trys.pop(); |
|
|
|
continue; |
|
} |
|
|
|
op = body.call(thisArg, _); |
|
} catch (e) { |
|
op = [6, e]; |
|
y = 0; |
|
} finally { |
|
f = t = 0; |
|
} |
|
} |
|
|
|
if (op[0] & 5) throw op[1]; |
|
return { |
|
value: op[0] ? op[1] : void 0, |
|
done: true |
|
}; |
|
} |
|
} |
|
function __exportStar(m, exports) { |
|
for (var p in m) { |
|
if (!exports.hasOwnProperty(p)) exports[p] = m[p]; |
|
} |
|
} |
|
function __values(o) { |
|
var m = typeof Symbol === "function" && o[Symbol.iterator], |
|
i = 0; |
|
if (m) return m.call(o); |
|
return { |
|
next: function next() { |
|
if (o && i >= o.length) o = void 0; |
|
return { |
|
value: o && o[i++], |
|
done: !o |
|
}; |
|
} |
|
}; |
|
} |
|
function __read(o, n) { |
|
var m = typeof Symbol === "function" && o[Symbol.iterator]; |
|
if (!m) return o; |
|
var i = m.call(o), |
|
r, |
|
ar = [], |
|
e; |
|
|
|
try { |
|
while ((n === void 0 || n-- > 0) && !(r = i.next()).done) { |
|
ar.push(r.value); |
|
} |
|
} catch (error) { |
|
e = { |
|
error: error |
|
}; |
|
} finally { |
|
try { |
|
if (r && !r.done && (m = i["return"])) m.call(i); |
|
} finally { |
|
if (e) throw e.error; |
|
} |
|
} |
|
|
|
return ar; |
|
} |
|
function __spread() { |
|
for (var ar = [], i = 0; i < arguments.length; i++) { |
|
ar = ar.concat(__read(arguments[i])); |
|
} |
|
|
|
return ar; |
|
} |
|
function __spreadArrays() { |
|
for (var s = 0, i = 0, il = arguments.length; i < il; i++) { |
|
s += arguments[i].length; |
|
} |
|
|
|
for (var r = Array(s), k = 0, i = 0; i < il; i++) { |
|
for (var a = arguments[i], j = 0, jl = a.length; j < jl; j++, k++) { |
|
r[k] = a[j]; |
|
} |
|
} |
|
|
|
return r; |
|
} |
|
function __await(v) { |
|
return this instanceof __await ? (this.v = v, this) : new __await(v); |
|
} |
|
function __asyncGenerator(thisArg, _arguments, generator) { |
|
if (!Symbol.asyncIterator) throw new TypeError("Symbol.asyncIterator is not defined."); |
|
var g = generator.apply(thisArg, _arguments || []), |
|
i, |
|
q = []; |
|
return i = {}, verb("next"), verb("throw"), verb("return"), i[Symbol.asyncIterator] = function () { |
|
return this; |
|
}, i; |
|
|
|
function verb(n) { |
|
if (g[n]) i[n] = function (v) { |
|
return new Promise(function (a, b) { |
|
q.push([n, v, a, b]) > 1 || resume(n, v); |
|
}); |
|
}; |
|
} |
|
|
|
function resume(n, v) { |
|
try { |
|
step(g[n](v)); |
|
} catch (e) { |
|
settle(q[0][3], e); |
|
} |
|
} |
|
|
|
function step(r) { |
|
r.value instanceof __await ? Promise.resolve(r.value.v).then(fulfill, reject) : settle(q[0][2], r); |
|
} |
|
|
|
function fulfill(value) { |
|
resume("next", value); |
|
} |
|
|
|
function reject(value) { |
|
resume("throw", value); |
|
} |
|
|
|
function settle(f, v) { |
|
if (f(v), q.shift(), q.length) resume(q[0][0], q[0][1]); |
|
} |
|
} |
|
function __asyncDelegator(o) { |
|
var i, p; |
|
return i = {}, verb("next"), verb("throw", function (e) { |
|
throw e; |
|
}), verb("return"), i[Symbol.iterator] = function () { |
|
return this; |
|
}, i; |
|
|
|
function verb(n, f) { |
|
i[n] = o[n] ? function (v) { |
|
return (p = !p) ? { |
|
value: __await(o[n](v)), |
|
done: n === "return" |
|
} : f ? f(v) : v; |
|
} : f; |
|
} |
|
} |
|
function __asyncValues(o) { |
|
if (!Symbol.asyncIterator) throw new TypeError("Symbol.asyncIterator is not defined."); |
|
var m = o[Symbol.asyncIterator], |
|
i; |
|
return m ? m.call(o) : (o = typeof __values === "function" ? __values(o) : o[Symbol.iterator](), i = {}, verb("next"), verb("throw"), verb("return"), i[Symbol.asyncIterator] = function () { |
|
return this; |
|
}, i); |
|
|
|
function verb(n) { |
|
i[n] = o[n] && function (v) { |
|
return new Promise(function (resolve, reject) { |
|
v = o[n](v), settle(resolve, reject, v.done, v.value); |
|
}); |
|
}; |
|
} |
|
|
|
function settle(resolve, reject, d, v) { |
|
Promise.resolve(v).then(function (v) { |
|
resolve({ |
|
value: v, |
|
done: d |
|
}); |
|
}, reject); |
|
} |
|
} |
|
function __makeTemplateObject(cooked, raw) { |
|
if (Object.defineProperty) { |
|
Object.defineProperty(cooked, "raw", { |
|
value: raw |
|
}); |
|
} else { |
|
cooked.raw = raw; |
|
} |
|
|
|
return cooked; |
|
} |
|
function __importStar(mod) { |
|
if (mod && mod.__esModule) return mod; |
|
var result = {}; |
|
if (mod != null) for (var k in mod) { |
|
if (Object.hasOwnProperty.call(mod, k)) result[k] = mod[k]; |
|
} |
|
result.default = mod; |
|
return result; |
|
} |
|
function __importDefault(mod) { |
|
return mod && mod.__esModule ? mod : { |
|
default: mod |
|
}; |
|
} |
|
|
|
var tslib_es6 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
__extends: __extends, |
|
get __assign () { return _assign; }, |
|
__rest: __rest, |
|
__decorate: __decorate, |
|
__param: __param, |
|
__metadata: __metadata, |
|
__awaiter: __awaiter, |
|
__generator: __generator, |
|
__exportStar: __exportStar, |
|
__values: __values, |
|
__read: __read, |
|
__spread: __spread, |
|
__spreadArrays: __spreadArrays, |
|
__await: __await, |
|
__asyncGenerator: __asyncGenerator, |
|
__asyncDelegator: __asyncDelegator, |
|
__asyncValues: __asyncValues, |
|
__makeTemplateObject: __makeTemplateObject, |
|
__importStar: __importStar, |
|
__importDefault: __importDefault |
|
}); |
|
|
|
var api = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
exports.apiDescriptor = { |
|
key: function key(_key) { |
|
return /^[$_a-zA-Z][$_a-zA-Z0-9]*$/.test(_key) ? _key : JSON.stringify(_key); |
|
}, |
|
value: function value(_value) { |
|
if (_value === null || _typeof(_value) !== 'object') { |
|
return JSON.stringify(_value); |
|
} |
|
|
|
if (Array.isArray(_value)) { |
|
return "[".concat(_value.map(function (subValue) { |
|
return exports.apiDescriptor.value(subValue); |
|
}).join(', '), "]"); |
|
} |
|
|
|
var keys = Object.keys(_value); |
|
return keys.length === 0 ? '{}' : "{ ".concat(keys.map(function (key) { |
|
return "".concat(exports.apiDescriptor.key(key), ": ").concat(exports.apiDescriptor.value(_value[key])); |
|
}).join(', '), " }"); |
|
}, |
|
pair: function pair(_ref) { |
|
var key = _ref.key, |
|
value = _ref.value; |
|
return exports.apiDescriptor.value(_defineProperty({}, key, value)); |
|
} |
|
}; |
|
}); |
|
unwrapExports(api); |
|
var api_1 = api.apiDescriptor; |
|
|
|
var tslib_1 = getCjsExportFromNamespace(tslib_es6); |
|
|
|
var descriptors = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
tslib_1.__exportStar(api, exports); |
|
}); |
|
unwrapExports(descriptors); |
|
|
|
var matchOperatorsRe = /[|\\{}()[\]^$+*?.]/g; |
|
|
|
var escapeStringRegexp = function escapeStringRegexp(str) { |
|
if (typeof str !== 'string') { |
|
throw new TypeError('Expected a string'); |
|
} |
|
|
|
return str.replace(matchOperatorsRe, '\\$&'); |
|
}; |
|
|
|
var colorName = { |
|
"aliceblue": [240, 248, 255], |
|
"antiquewhite": [250, 235, 215], |
|
"aqua": [0, 255, 255], |
|
"aquamarine": [127, 255, 212], |
|
"azure": [240, 255, 255], |
|
"beige": [245, 245, 220], |
|
"bisque": [255, 228, 196], |
|
"black": [0, 0, 0], |
|
"blanchedalmond": [255, 235, 205], |
|
"blue": [0, 0, 255], |
|
"blueviolet": [138, 43, 226], |
|
"brown": [165, 42, 42], |
|
"burlywood": [222, 184, 135], |
|
"cadetblue": [95, 158, 160], |
|
"chartreuse": [127, 255, 0], |
|
"chocolate": [210, 105, 30], |
|
"coral": [255, 127, 80], |
|
"cornflowerblue": [100, 149, 237], |
|
"cornsilk": [255, 248, 220], |
|
"crimson": [220, 20, 60], |
|
"cyan": [0, 255, 255], |
|
"darkblue": [0, 0, 139], |
|
"darkcyan": [0, 139, 139], |
|
"darkgoldenrod": [184, 134, 11], |
|
"darkgray": [169, 169, 169], |
|
"darkgreen": [0, 100, 0], |
|
"darkgrey": [169, 169, 169], |
|
"darkkhaki": [189, 183, 107], |
|
"darkmagenta": [139, 0, 139], |
|
"darkolivegreen": [85, 107, 47], |
|
"darkorange": [255, 140, 0], |
|
"darkorchid": [153, 50, 204], |
|
"darkred": [139, 0, 0], |
|
"darksalmon": [233, 150, 122], |
|
"darkseagreen": [143, 188, 143], |
|
"darkslateblue": [72, 61, 139], |
|
"darkslategray": [47, 79, 79], |
|
"darkslategrey": [47, 79, 79], |
|
"darkturquoise": [0, 206, 209], |
|
"darkviolet": [148, 0, 211], |
|
"deeppink": [255, 20, 147], |
|
"deepskyblue": [0, 191, 255], |
|
"dimgray": [105, 105, 105], |
|
"dimgrey": [105, 105, 105], |
|
"dodgerblue": [30, 144, 255], |
|
"firebrick": [178, 34, 34], |
|
"floralwhite": [255, 250, 240], |
|
"forestgreen": [34, 139, 34], |
|
"fuchsia": [255, 0, 255], |
|
"gainsboro": [220, 220, 220], |
|
"ghostwhite": [248, 248, 255], |
|
"gold": [255, 215, 0], |
|
"goldenrod": [218, 165, 32], |
|
"gray": [128, 128, 128], |
|
"green": [0, 128, 0], |
|
"greenyellow": [173, 255, 47], |
|
"grey": [128, 128, 128], |
|
"honeydew": [240, 255, 240], |
|
"hotpink": [255, 105, 180], |
|
"indianred": [205, 92, 92], |
|
"indigo": [75, 0, 130], |
|
"ivory": [255, 255, 240], |
|
"khaki": [240, 230, 140], |
|
"lavender": [230, 230, 250], |
|
"lavenderblush": [255, 240, 245], |
|
"lawngreen": [124, 252, 0], |
|
"lemonchiffon": [255, 250, 205], |
|
"lightblue": [173, 216, 230], |
|
"lightcoral": [240, 128, 128], |
|
"lightcyan": [224, 255, 255], |
|
"lightgoldenrodyellow": [250, 250, 210], |
|
"lightgray": [211, 211, 211], |
|
"lightgreen": [144, 238, 144], |
|
"lightgrey": [211, 211, 211], |
|
"lightpink": [255, 182, 193], |
|
"lightsalmon": [255, 160, 122], |
|
"lightseagreen": [32, 178, 170], |
|
"lightskyblue": [135, 206, 250], |
|
"lightslategray": [119, 136, 153], |
|
"lightslategrey": [119, 136, 153], |
|
"lightsteelblue": [176, 196, 222], |
|
"lightyellow": [255, 255, 224], |
|
"lime": [0, 255, 0], |
|
"limegreen": [50, 205, 50], |
|
"linen": [250, 240, 230], |
|
"magenta": [255, 0, 255], |
|
"maroon": [128, 0, 0], |
|
"mediumaquamarine": [102, 205, 170], |
|
"mediumblue": [0, 0, 205], |
|
"mediumorchid": [186, 85, 211], |
|
"mediumpurple": [147, 112, 219], |
|
"mediumseagreen": [60, 179, 113], |
|
"mediumslateblue": [123, 104, 238], |
|
"mediumspringgreen": [0, 250, 154], |
|
"mediumturquoise": [72, 209, 204], |
|
"mediumvioletred": [199, 21, 133], |
|
"midnightblue": [25, 25, 112], |
|
"mintcream": [245, 255, 250], |
|
"mistyrose": [255, 228, 225], |
|
"moccasin": [255, 228, 181], |
|
"navajowhite": [255, 222, 173], |
|
"navy": [0, 0, 128], |
|
"oldlace": [253, 245, 230], |
|
"olive": [128, 128, 0], |
|
"olivedrab": [107, 142, 35], |
|
"orange": [255, 165, 0], |
|
"orangered": [255, 69, 0], |
|
"orchid": [218, 112, 214], |
|
"palegoldenrod": [238, 232, 170], |
|
"palegreen": [152, 251, 152], |
|
"paleturquoise": [175, 238, 238], |
|
"palevioletred": [219, 112, 147], |
|
"papayawhip": [255, 239, 213], |
|
"peachpuff": [255, 218, 185], |
|
"peru": [205, 133, 63], |
|
"pink": [255, 192, 203], |
|
"plum": [221, 160, 221], |
|
"powderblue": [176, 224, 230], |
|
"purple": [128, 0, 128], |
|
"rebeccapurple": [102, 51, 153], |
|
"red": [255, 0, 0], |
|
"rosybrown": [188, 143, 143], |
|
"royalblue": [65, 105, 225], |
|
"saddlebrown": [139, 69, 19], |
|
"salmon": [250, 128, 114], |
|
"sandybrown": [244, 164, 96], |
|
"seagreen": [46, 139, 87], |
|
"seashell": [255, 245, 238], |
|
"sienna": [160, 82, 45], |
|
"silver": [192, 192, 192], |
|
"skyblue": [135, 206, 235], |
|
"slateblue": [106, 90, 205], |
|
"slategray": [112, 128, 144], |
|
"slategrey": [112, 128, 144], |
|
"snow": [255, 250, 250], |
|
"springgreen": [0, 255, 127], |
|
"steelblue": [70, 130, 180], |
|
"tan": [210, 180, 140], |
|
"teal": [0, 128, 128], |
|
"thistle": [216, 191, 216], |
|
"tomato": [255, 99, 71], |
|
"turquoise": [64, 224, 208], |
|
"violet": [238, 130, 238], |
|
"wheat": [245, 222, 179], |
|
"white": [255, 255, 255], |
|
"whitesmoke": [245, 245, 245], |
|
"yellow": [255, 255, 0], |
|
"yellowgreen": [154, 205, 50] |
|
}; |
|
|
|
var conversions = createCommonjsModule(function (module) { |
|
/* MIT license */ |
|
// NOTE: conversions should only return primitive values (i.e. arrays, or |
|
// values that give correct `typeof` results). |
|
// do not use box values types (i.e. Number(), String(), etc.) |
|
var reverseKeywords = {}; |
|
|
|
for (var key in colorName) { |
|
if (colorName.hasOwnProperty(key)) { |
|
reverseKeywords[colorName[key]] = key; |
|
} |
|
} |
|
|
|
var convert = module.exports = { |
|
rgb: { |
|
channels: 3, |
|
labels: 'rgb' |
|
}, |
|
hsl: { |
|
channels: 3, |
|
labels: 'hsl' |
|
}, |
|
hsv: { |
|
channels: 3, |
|
labels: 'hsv' |
|
}, |
|
hwb: { |
|
channels: 3, |
|
labels: 'hwb' |
|
}, |
|
cmyk: { |
|
channels: 4, |
|
labels: 'cmyk' |
|
}, |
|
xyz: { |
|
channels: 3, |
|
labels: 'xyz' |
|
}, |
|
lab: { |
|
channels: 3, |
|
labels: 'lab' |
|
}, |
|
lch: { |
|
channels: 3, |
|
labels: 'lch' |
|
}, |
|
hex: { |
|
channels: 1, |
|
labels: ['hex'] |
|
}, |
|
keyword: { |
|
channels: 1, |
|
labels: ['keyword'] |
|
}, |
|
ansi16: { |
|
channels: 1, |
|
labels: ['ansi16'] |
|
}, |
|
ansi256: { |
|
channels: 1, |
|
labels: ['ansi256'] |
|
}, |
|
hcg: { |
|
channels: 3, |
|
labels: ['h', 'c', 'g'] |
|
}, |
|
apple: { |
|
channels: 3, |
|
labels: ['r16', 'g16', 'b16'] |
|
}, |
|
gray: { |
|
channels: 1, |
|
labels: ['gray'] |
|
} |
|
}; // hide .channels and .labels properties |
|
|
|
for (var model in convert) { |
|
if (convert.hasOwnProperty(model)) { |
|
if (!('channels' in convert[model])) { |
|
throw new Error('missing channels property: ' + model); |
|
} |
|
|
|
if (!('labels' in convert[model])) { |
|
throw new Error('missing channel labels property: ' + model); |
|
} |
|
|
|
if (convert[model].labels.length !== convert[model].channels) { |
|
throw new Error('channel and label counts mismatch: ' + model); |
|
} |
|
|
|
var channels = convert[model].channels; |
|
var labels = convert[model].labels; |
|
delete convert[model].channels; |
|
delete convert[model].labels; |
|
Object.defineProperty(convert[model], 'channels', { |
|
value: channels |
|
}); |
|
Object.defineProperty(convert[model], 'labels', { |
|
value: labels |
|
}); |
|
} |
|
} |
|
|
|
convert.rgb.hsl = function (rgb) { |
|
var r = rgb[0] / 255; |
|
var g = rgb[1] / 255; |
|
var b = rgb[2] / 255; |
|
var min = Math.min(r, g, b); |
|
var max = Math.max(r, g, b); |
|
var delta = max - min; |
|
var h; |
|
var s; |
|
var l; |
|
|
|
if (max === min) { |
|
h = 0; |
|
} else if (r === max) { |
|
h = (g - b) / delta; |
|
} else if (g === max) { |
|
h = 2 + (b - r) / delta; |
|
} else if (b === max) { |
|
h = 4 + (r - g) / delta; |
|
} |
|
|
|
h = Math.min(h * 60, 360); |
|
|
|
if (h < 0) { |
|
h += 360; |
|
} |
|
|
|
l = (min + max) / 2; |
|
|
|
if (max === min) { |
|
s = 0; |
|
} else if (l <= 0.5) { |
|
s = delta / (max + min); |
|
} else { |
|
s = delta / (2 - max - min); |
|
} |
|
|
|
return [h, s * 100, l * 100]; |
|
}; |
|
|
|
convert.rgb.hsv = function (rgb) { |
|
var r = rgb[0]; |
|
var g = rgb[1]; |
|
var b = rgb[2]; |
|
var min = Math.min(r, g, b); |
|
var max = Math.max(r, g, b); |
|
var delta = max - min; |
|
var h; |
|
var s; |
|
var v; |
|
|
|
if (max === 0) { |
|
s = 0; |
|
} else { |
|
s = delta / max * 1000 / 10; |
|
} |
|
|
|
if (max === min) { |
|
h = 0; |
|
} else if (r === max) { |
|
h = (g - b) / delta; |
|
} else if (g === max) { |
|
h = 2 + (b - r) / delta; |
|
} else if (b === max) { |
|
h = 4 + (r - g) / delta; |
|
} |
|
|
|
h = Math.min(h * 60, 360); |
|
|
|
if (h < 0) { |
|
h += 360; |
|
} |
|
|
|
v = max / 255 * 1000 / 10; |
|
return [h, s, v]; |
|
}; |
|
|
|
convert.rgb.hwb = function (rgb) { |
|
var r = rgb[0]; |
|
var g = rgb[1]; |
|
var b = rgb[2]; |
|
var h = convert.rgb.hsl(rgb)[0]; |
|
var w = 1 / 255 * Math.min(r, Math.min(g, b)); |
|
b = 1 - 1 / 255 * Math.max(r, Math.max(g, b)); |
|
return [h, w * 100, b * 100]; |
|
}; |
|
|
|
convert.rgb.cmyk = function (rgb) { |
|
var r = rgb[0] / 255; |
|
var g = rgb[1] / 255; |
|
var b = rgb[2] / 255; |
|
var c; |
|
var m; |
|
var y; |
|
var k; |
|
k = Math.min(1 - r, 1 - g, 1 - b); |
|
c = (1 - r - k) / (1 - k) || 0; |
|
m = (1 - g - k) / (1 - k) || 0; |
|
y = (1 - b - k) / (1 - k) || 0; |
|
return [c * 100, m * 100, y * 100, k * 100]; |
|
}; |
|
/** |
|
* See https://en.m.wikipedia.org/wiki/Euclidean_distance#Squared_Euclidean_distance |
|
* */ |
|
|
|
|
|
function comparativeDistance(x, y) { |
|
return Math.pow(x[0] - y[0], 2) + Math.pow(x[1] - y[1], 2) + Math.pow(x[2] - y[2], 2); |
|
} |
|
|
|
convert.rgb.keyword = function (rgb) { |
|
var reversed = reverseKeywords[rgb]; |
|
|
|
if (reversed) { |
|
return reversed; |
|
} |
|
|
|
var currentClosestDistance = Infinity; |
|
var currentClosestKeyword; |
|
|
|
for (var keyword in colorName) { |
|
if (colorName.hasOwnProperty(keyword)) { |
|
var value = colorName[keyword]; // Compute comparative distance |
|
|
|
var distance = comparativeDistance(rgb, value); // Check if its less, if so set as closest |
|
|
|
if (distance < currentClosestDistance) { |
|
currentClosestDistance = distance; |
|
currentClosestKeyword = keyword; |
|
} |
|
} |
|
} |
|
|
|
return currentClosestKeyword; |
|
}; |
|
|
|
convert.keyword.rgb = function (keyword) { |
|
return colorName[keyword]; |
|
}; |
|
|
|
convert.rgb.xyz = function (rgb) { |
|
var r = rgb[0] / 255; |
|
var g = rgb[1] / 255; |
|
var b = rgb[2] / 255; // assume sRGB |
|
|
|
r = r > 0.04045 ? Math.pow((r + 0.055) / 1.055, 2.4) : r / 12.92; |
|
g = g > 0.04045 ? Math.pow((g + 0.055) / 1.055, 2.4) : g / 12.92; |
|
b = b > 0.04045 ? Math.pow((b + 0.055) / 1.055, 2.4) : b / 12.92; |
|
var x = r * 0.4124 + g * 0.3576 + b * 0.1805; |
|
var y = r * 0.2126 + g * 0.7152 + b * 0.0722; |
|
var z = r * 0.0193 + g * 0.1192 + b * 0.9505; |
|
return [x * 100, y * 100, z * 100]; |
|
}; |
|
|
|
convert.rgb.lab = function (rgb) { |
|
var xyz = convert.rgb.xyz(rgb); |
|
var x = xyz[0]; |
|
var y = xyz[1]; |
|
var z = xyz[2]; |
|
var l; |
|
var a; |
|
var b; |
|
x /= 95.047; |
|
y /= 100; |
|
z /= 108.883; |
|
x = x > 0.008856 ? Math.pow(x, 1 / 3) : 7.787 * x + 16 / 116; |
|
y = y > 0.008856 ? Math.pow(y, 1 / 3) : 7.787 * y + 16 / 116; |
|
z = z > 0.008856 ? Math.pow(z, 1 / 3) : 7.787 * z + 16 / 116; |
|
l = 116 * y - 16; |
|
a = 500 * (x - y); |
|
b = 200 * (y - z); |
|
return [l, a, b]; |
|
}; |
|
|
|
convert.hsl.rgb = function (hsl) { |
|
var h = hsl[0] / 360; |
|
var s = hsl[1] / 100; |
|
var l = hsl[2] / 100; |
|
var t1; |
|
var t2; |
|
var t3; |
|
var rgb; |
|
var val; |
|
|
|
if (s === 0) { |
|
val = l * 255; |
|
return [val, val, val]; |
|
} |
|
|
|
if (l < 0.5) { |
|
t2 = l * (1 + s); |
|
} else { |
|
t2 = l + s - l * s; |
|
} |
|
|
|
t1 = 2 * l - t2; |
|
rgb = [0, 0, 0]; |
|
|
|
for (var i = 0; i < 3; i++) { |
|
t3 = h + 1 / 3 * -(i - 1); |
|
|
|
if (t3 < 0) { |
|
t3++; |
|
} |
|
|
|
if (t3 > 1) { |
|
t3--; |
|
} |
|
|
|
if (6 * t3 < 1) { |
|
val = t1 + (t2 - t1) * 6 * t3; |
|
} else if (2 * t3 < 1) { |
|
val = t2; |
|
} else if (3 * t3 < 2) { |
|
val = t1 + (t2 - t1) * (2 / 3 - t3) * 6; |
|
} else { |
|
val = t1; |
|
} |
|
|
|
rgb[i] = val * 255; |
|
} |
|
|
|
return rgb; |
|
}; |
|
|
|
convert.hsl.hsv = function (hsl) { |
|
var h = hsl[0]; |
|
var s = hsl[1] / 100; |
|
var l = hsl[2] / 100; |
|
var smin = s; |
|
var lmin = Math.max(l, 0.01); |
|
var sv; |
|
var v; |
|
l *= 2; |
|
s *= l <= 1 ? l : 2 - l; |
|
smin *= lmin <= 1 ? lmin : 2 - lmin; |
|
v = (l + s) / 2; |
|
sv = l === 0 ? 2 * smin / (lmin + smin) : 2 * s / (l + s); |
|
return [h, sv * 100, v * 100]; |
|
}; |
|
|
|
convert.hsv.rgb = function (hsv) { |
|
var h = hsv[0] / 60; |
|
var s = hsv[1] / 100; |
|
var v = hsv[2] / 100; |
|
var hi = Math.floor(h) % 6; |
|
var f = h - Math.floor(h); |
|
var p = 255 * v * (1 - s); |
|
var q = 255 * v * (1 - s * f); |
|
var t = 255 * v * (1 - s * (1 - f)); |
|
v *= 255; |
|
|
|
switch (hi) { |
|
case 0: |
|
return [v, t, p]; |
|
|
|
case 1: |
|
return [q, v, p]; |
|
|
|
case 2: |
|
return [p, v, t]; |
|
|
|
case 3: |
|
return [p, q, v]; |
|
|
|
case 4: |
|
return [t, p, v]; |
|
|
|
case 5: |
|
return [v, p, q]; |
|
} |
|
}; |
|
|
|
convert.hsv.hsl = function (hsv) { |
|
var h = hsv[0]; |
|
var s = hsv[1] / 100; |
|
var v = hsv[2] / 100; |
|
var vmin = Math.max(v, 0.01); |
|
var lmin; |
|
var sl; |
|
var l; |
|
l = (2 - s) * v; |
|
lmin = (2 - s) * vmin; |
|
sl = s * vmin; |
|
sl /= lmin <= 1 ? lmin : 2 - lmin; |
|
sl = sl || 0; |
|
l /= 2; |
|
return [h, sl * 100, l * 100]; |
|
}; // http://dev.w3.org/csswg/css-color/#hwb-to-rgb |
|
|
|
|
|
convert.hwb.rgb = function (hwb) { |
|
var h = hwb[0] / 360; |
|
var wh = hwb[1] / 100; |
|
var bl = hwb[2] / 100; |
|
var ratio = wh + bl; |
|
var i; |
|
var v; |
|
var f; |
|
var n; // wh + bl cant be > 1 |
|
|
|
if (ratio > 1) { |
|
wh /= ratio; |
|
bl /= ratio; |
|
} |
|
|
|
i = Math.floor(6 * h); |
|
v = 1 - bl; |
|
f = 6 * h - i; |
|
|
|
if ((i & 0x01) !== 0) { |
|
f = 1 - f; |
|
} |
|
|
|
n = wh + f * (v - wh); // linear interpolation |
|
|
|
var r; |
|
var g; |
|
var b; |
|
|
|
switch (i) { |
|
default: |
|
case 6: |
|
case 0: |
|
r = v; |
|
g = n; |
|
b = wh; |
|
break; |
|
|
|
case 1: |
|
r = n; |
|
g = v; |
|
b = wh; |
|
break; |
|
|
|
case 2: |
|
r = wh; |
|
g = v; |
|
b = n; |
|
break; |
|
|
|
case 3: |
|
r = wh; |
|
g = n; |
|
b = v; |
|
break; |
|
|
|
case 4: |
|
r = n; |
|
g = wh; |
|
b = v; |
|
break; |
|
|
|
case 5: |
|
r = v; |
|
g = wh; |
|
b = n; |
|
break; |
|
} |
|
|
|
return [r * 255, g * 255, b * 255]; |
|
}; |
|
|
|
convert.cmyk.rgb = function (cmyk) { |
|
var c = cmyk[0] / 100; |
|
var m = cmyk[1] / 100; |
|
var y = cmyk[2] / 100; |
|
var k = cmyk[3] / 100; |
|
var r; |
|
var g; |
|
var b; |
|
r = 1 - Math.min(1, c * (1 - k) + k); |
|
g = 1 - Math.min(1, m * (1 - k) + k); |
|
b = 1 - Math.min(1, y * (1 - k) + k); |
|
return [r * 255, g * 255, b * 255]; |
|
}; |
|
|
|
convert.xyz.rgb = function (xyz) { |
|
var x = xyz[0] / 100; |
|
var y = xyz[1] / 100; |
|
var z = xyz[2] / 100; |
|
var r; |
|
var g; |
|
var b; |
|
r = x * 3.2406 + y * -1.5372 + z * -0.4986; |
|
g = x * -0.9689 + y * 1.8758 + z * 0.0415; |
|
b = x * 0.0557 + y * -0.2040 + z * 1.0570; // assume sRGB |
|
|
|
r = r > 0.0031308 ? 1.055 * Math.pow(r, 1.0 / 2.4) - 0.055 : r * 12.92; |
|
g = g > 0.0031308 ? 1.055 * Math.pow(g, 1.0 / 2.4) - 0.055 : g * 12.92; |
|
b = b > 0.0031308 ? 1.055 * Math.pow(b, 1.0 / 2.4) - 0.055 : b * 12.92; |
|
r = Math.min(Math.max(0, r), 1); |
|
g = Math.min(Math.max(0, g), 1); |
|
b = Math.min(Math.max(0, b), 1); |
|
return [r * 255, g * 255, b * 255]; |
|
}; |
|
|
|
convert.xyz.lab = function (xyz) { |
|
var x = xyz[0]; |
|
var y = xyz[1]; |
|
var z = xyz[2]; |
|
var l; |
|
var a; |
|
var b; |
|
x /= 95.047; |
|
y /= 100; |
|
z /= 108.883; |
|
x = x > 0.008856 ? Math.pow(x, 1 / 3) : 7.787 * x + 16 / 116; |
|
y = y > 0.008856 ? Math.pow(y, 1 / 3) : 7.787 * y + 16 / 116; |
|
z = z > 0.008856 ? Math.pow(z, 1 / 3) : 7.787 * z + 16 / 116; |
|
l = 116 * y - 16; |
|
a = 500 * (x - y); |
|
b = 200 * (y - z); |
|
return [l, a, b]; |
|
}; |
|
|
|
convert.lab.xyz = function (lab) { |
|
var l = lab[0]; |
|
var a = lab[1]; |
|
var b = lab[2]; |
|
var x; |
|
var y; |
|
var z; |
|
y = (l + 16) / 116; |
|
x = a / 500 + y; |
|
z = y - b / 200; |
|
var y2 = Math.pow(y, 3); |
|
var x2 = Math.pow(x, 3); |
|
var z2 = Math.pow(z, 3); |
|
y = y2 > 0.008856 ? y2 : (y - 16 / 116) / 7.787; |
|
x = x2 > 0.008856 ? x2 : (x - 16 / 116) / 7.787; |
|
z = z2 > 0.008856 ? z2 : (z - 16 / 116) / 7.787; |
|
x *= 95.047; |
|
y *= 100; |
|
z *= 108.883; |
|
return [x, y, z]; |
|
}; |
|
|
|
convert.lab.lch = function (lab) { |
|
var l = lab[0]; |
|
var a = lab[1]; |
|
var b = lab[2]; |
|
var hr; |
|
var h; |
|
var c; |
|
hr = Math.atan2(b, a); |
|
h = hr * 360 / 2 / Math.PI; |
|
|
|
if (h < 0) { |
|
h += 360; |
|
} |
|
|
|
c = Math.sqrt(a * a + b * b); |
|
return [l, c, h]; |
|
}; |
|
|
|
convert.lch.lab = function (lch) { |
|
var l = lch[0]; |
|
var c = lch[1]; |
|
var h = lch[2]; |
|
var a; |
|
var b; |
|
var hr; |
|
hr = h / 360 * 2 * Math.PI; |
|
a = c * Math.cos(hr); |
|
b = c * Math.sin(hr); |
|
return [l, a, b]; |
|
}; |
|
|
|
convert.rgb.ansi16 = function (args) { |
|
var r = args[0]; |
|
var g = args[1]; |
|
var b = args[2]; |
|
var value = 1 in arguments ? arguments[1] : convert.rgb.hsv(args)[2]; // hsv -> ansi16 optimization |
|
|
|
value = Math.round(value / 50); |
|
|
|
if (value === 0) { |
|
return 30; |
|
} |
|
|
|
var ansi = 30 + (Math.round(b / 255) << 2 | Math.round(g / 255) << 1 | Math.round(r / 255)); |
|
|
|
if (value === 2) { |
|
ansi += 60; |
|
} |
|
|
|
return ansi; |
|
}; |
|
|
|
convert.hsv.ansi16 = function (args) { |
|
// optimization here; we already know the value and don't need to get |
|
// it converted for us. |
|
return convert.rgb.ansi16(convert.hsv.rgb(args), args[2]); |
|
}; |
|
|
|
convert.rgb.ansi256 = function (args) { |
|
var r = args[0]; |
|
var g = args[1]; |
|
var b = args[2]; // we use the extended greyscale palette here, with the exception of |
|
// black and white. normal palette only has 4 greyscale shades. |
|
|
|
if (r === g && g === b) { |
|
if (r < 8) { |
|
return 16; |
|
} |
|
|
|
if (r > 248) { |
|
return 231; |
|
} |
|
|
|
return Math.round((r - 8) / 247 * 24) + 232; |
|
} |
|
|
|
var ansi = 16 + 36 * Math.round(r / 255 * 5) + 6 * Math.round(g / 255 * 5) + Math.round(b / 255 * 5); |
|
return ansi; |
|
}; |
|
|
|
convert.ansi16.rgb = function (args) { |
|
var color = args % 10; // handle greyscale |
|
|
|
if (color === 0 || color === 7) { |
|
if (args > 50) { |
|
color += 3.5; |
|
} |
|
|
|
color = color / 10.5 * 255; |
|
return [color, color, color]; |
|
} |
|
|
|
var mult = (~~(args > 50) + 1) * 0.5; |
|
var r = (color & 1) * mult * 255; |
|
var g = (color >> 1 & 1) * mult * 255; |
|
var b = (color >> 2 & 1) * mult * 255; |
|
return [r, g, b]; |
|
}; |
|
|
|
convert.ansi256.rgb = function (args) { |
|
// handle greyscale |
|
if (args >= 232) { |
|
var c = (args - 232) * 10 + 8; |
|
return [c, c, c]; |
|
} |
|
|
|
args -= 16; |
|
var rem; |
|
var r = Math.floor(args / 36) / 5 * 255; |
|
var g = Math.floor((rem = args % 36) / 6) / 5 * 255; |
|
var b = rem % 6 / 5 * 255; |
|
return [r, g, b]; |
|
}; |
|
|
|
convert.rgb.hex = function (args) { |
|
var integer = ((Math.round(args[0]) & 0xFF) << 16) + ((Math.round(args[1]) & 0xFF) << 8) + (Math.round(args[2]) & 0xFF); |
|
var string = integer.toString(16).toUpperCase(); |
|
return '000000'.substring(string.length) + string; |
|
}; |
|
|
|
convert.hex.rgb = function (args) { |
|
var match = args.toString(16).match(/[a-f0-9]{6}|[a-f0-9]{3}/i); |
|
|
|
if (!match) { |
|
return [0, 0, 0]; |
|
} |
|
|
|
var colorString = match[0]; |
|
|
|
if (match[0].length === 3) { |
|
colorString = colorString.split('').map(function (char) { |
|
return char + char; |
|
}).join(''); |
|
} |
|
|
|
var integer = parseInt(colorString, 16); |
|
var r = integer >> 16 & 0xFF; |
|
var g = integer >> 8 & 0xFF; |
|
var b = integer & 0xFF; |
|
return [r, g, b]; |
|
}; |
|
|
|
convert.rgb.hcg = function (rgb) { |
|
var r = rgb[0] / 255; |
|
var g = rgb[1] / 255; |
|
var b = rgb[2] / 255; |
|
var max = Math.max(Math.max(r, g), b); |
|
var min = Math.min(Math.min(r, g), b); |
|
var chroma = max - min; |
|
var grayscale; |
|
var hue; |
|
|
|
if (chroma < 1) { |
|
grayscale = min / (1 - chroma); |
|
} else { |
|
grayscale = 0; |
|
} |
|
|
|
if (chroma <= 0) { |
|
hue = 0; |
|
} else if (max === r) { |
|
hue = (g - b) / chroma % 6; |
|
} else if (max === g) { |
|
hue = 2 + (b - r) / chroma; |
|
} else { |
|
hue = 4 + (r - g) / chroma + 4; |
|
} |
|
|
|
hue /= 6; |
|
hue %= 1; |
|
return [hue * 360, chroma * 100, grayscale * 100]; |
|
}; |
|
|
|
convert.hsl.hcg = function (hsl) { |
|
var s = hsl[1] / 100; |
|
var l = hsl[2] / 100; |
|
var c = 1; |
|
var f = 0; |
|
|
|
if (l < 0.5) { |
|
c = 2.0 * s * l; |
|
} else { |
|
c = 2.0 * s * (1.0 - l); |
|
} |
|
|
|
if (c < 1.0) { |
|
f = (l - 0.5 * c) / (1.0 - c); |
|
} |
|
|
|
return [hsl[0], c * 100, f * 100]; |
|
}; |
|
|
|
convert.hsv.hcg = function (hsv) { |
|
var s = hsv[1] / 100; |
|
var v = hsv[2] / 100; |
|
var c = s * v; |
|
var f = 0; |
|
|
|
if (c < 1.0) { |
|
f = (v - c) / (1 - c); |
|
} |
|
|
|
return [hsv[0], c * 100, f * 100]; |
|
}; |
|
|
|
convert.hcg.rgb = function (hcg) { |
|
var h = hcg[0] / 360; |
|
var c = hcg[1] / 100; |
|
var g = hcg[2] / 100; |
|
|
|
if (c === 0.0) { |
|
return [g * 255, g * 255, g * 255]; |
|
} |
|
|
|
var pure = [0, 0, 0]; |
|
var hi = h % 1 * 6; |
|
var v = hi % 1; |
|
var w = 1 - v; |
|
var mg = 0; |
|
|
|
switch (Math.floor(hi)) { |
|
case 0: |
|
pure[0] = 1; |
|
pure[1] = v; |
|
pure[2] = 0; |
|
break; |
|
|
|
case 1: |
|
pure[0] = w; |
|
pure[1] = 1; |
|
pure[2] = 0; |
|
break; |
|
|
|
case 2: |
|
pure[0] = 0; |
|
pure[1] = 1; |
|
pure[2] = v; |
|
break; |
|
|
|
case 3: |
|
pure[0] = 0; |
|
pure[1] = w; |
|
pure[2] = 1; |
|
break; |
|
|
|
case 4: |
|
pure[0] = v; |
|
pure[1] = 0; |
|
pure[2] = 1; |
|
break; |
|
|
|
default: |
|
pure[0] = 1; |
|
pure[1] = 0; |
|
pure[2] = w; |
|
} |
|
|
|
mg = (1.0 - c) * g; |
|
return [(c * pure[0] + mg) * 255, (c * pure[1] + mg) * 255, (c * pure[2] + mg) * 255]; |
|
}; |
|
|
|
convert.hcg.hsv = function (hcg) { |
|
var c = hcg[1] / 100; |
|
var g = hcg[2] / 100; |
|
var v = c + g * (1.0 - c); |
|
var f = 0; |
|
|
|
if (v > 0.0) { |
|
f = c / v; |
|
} |
|
|
|
return [hcg[0], f * 100, v * 100]; |
|
}; |
|
|
|
convert.hcg.hsl = function (hcg) { |
|
var c = hcg[1] / 100; |
|
var g = hcg[2] / 100; |
|
var l = g * (1.0 - c) + 0.5 * c; |
|
var s = 0; |
|
|
|
if (l > 0.0 && l < 0.5) { |
|
s = c / (2 * l); |
|
} else if (l >= 0.5 && l < 1.0) { |
|
s = c / (2 * (1 - l)); |
|
} |
|
|
|
return [hcg[0], s * 100, l * 100]; |
|
}; |
|
|
|
convert.hcg.hwb = function (hcg) { |
|
var c = hcg[1] / 100; |
|
var g = hcg[2] / 100; |
|
var v = c + g * (1.0 - c); |
|
return [hcg[0], (v - c) * 100, (1 - v) * 100]; |
|
}; |
|
|
|
convert.hwb.hcg = function (hwb) { |
|
var w = hwb[1] / 100; |
|
var b = hwb[2] / 100; |
|
var v = 1 - b; |
|
var c = v - w; |
|
var g = 0; |
|
|
|
if (c < 1) { |
|
g = (v - c) / (1 - c); |
|
} |
|
|
|
return [hwb[0], c * 100, g * 100]; |
|
}; |
|
|
|
convert.apple.rgb = function (apple) { |
|
return [apple[0] / 65535 * 255, apple[1] / 65535 * 255, apple[2] / 65535 * 255]; |
|
}; |
|
|
|
convert.rgb.apple = function (rgb) { |
|
return [rgb[0] / 255 * 65535, rgb[1] / 255 * 65535, rgb[2] / 255 * 65535]; |
|
}; |
|
|
|
convert.gray.rgb = function (args) { |
|
return [args[0] / 100 * 255, args[0] / 100 * 255, args[0] / 100 * 255]; |
|
}; |
|
|
|
convert.gray.hsl = convert.gray.hsv = function (args) { |
|
return [0, 0, args[0]]; |
|
}; |
|
|
|
convert.gray.hwb = function (gray) { |
|
return [0, 100, gray[0]]; |
|
}; |
|
|
|
convert.gray.cmyk = function (gray) { |
|
return [0, 0, 0, gray[0]]; |
|
}; |
|
|
|
convert.gray.lab = function (gray) { |
|
return [gray[0], 0, 0]; |
|
}; |
|
|
|
convert.gray.hex = function (gray) { |
|
var val = Math.round(gray[0] / 100 * 255) & 0xFF; |
|
var integer = (val << 16) + (val << 8) + val; |
|
var string = integer.toString(16).toUpperCase(); |
|
return '000000'.substring(string.length) + string; |
|
}; |
|
|
|
convert.rgb.gray = function (rgb) { |
|
var val = (rgb[0] + rgb[1] + rgb[2]) / 3; |
|
return [val / 255 * 100]; |
|
}; |
|
}); |
|
var conversions_1 = conversions.rgb; |
|
var conversions_2 = conversions.hsl; |
|
var conversions_3 = conversions.hsv; |
|
var conversions_4 = conversions.hwb; |
|
var conversions_5 = conversions.cmyk; |
|
var conversions_6 = conversions.xyz; |
|
var conversions_7 = conversions.lab; |
|
var conversions_8 = conversions.lch; |
|
var conversions_9 = conversions.hex; |
|
var conversions_10 = conversions.keyword; |
|
var conversions_11 = conversions.ansi16; |
|
var conversions_12 = conversions.ansi256; |
|
var conversions_13 = conversions.hcg; |
|
var conversions_14 = conversions.apple; |
|
var conversions_15 = conversions.gray; |
|
|
|
/* |
|
this function routes a model to all other models. |
|
|
|
all functions that are routed have a property `.conversion` attached |
|
to the returned synthetic function. This property is an array |
|
of strings, each with the steps in between the 'from' and 'to' |
|
color models (inclusive). |
|
|
|
conversions that are not possible simply are not included. |
|
*/ |
|
// https://jsperf.com/object-keys-vs-for-in-with-closure/3 |
|
|
|
var models = Object.keys(conversions); |
|
|
|
function buildGraph() { |
|
var graph = {}; |
|
|
|
for (var len = models.length, i = 0; i < len; i++) { |
|
graph[models[i]] = { |
|
// http://jsperf.com/1-vs-infinity |
|
// micro-opt, but this is simple. |
|
distance: -1, |
|
parent: null |
|
}; |
|
} |
|
|
|
return graph; |
|
} // https://en.wikipedia.org/wiki/Breadth-first_search |
|
|
|
|
|
function deriveBFS(fromModel) { |
|
var graph = buildGraph(); |
|
var queue = [fromModel]; // unshift -> queue -> pop |
|
|
|
graph[fromModel].distance = 0; |
|
|
|
while (queue.length) { |
|
var current = queue.pop(); |
|
var adjacents = Object.keys(conversions[current]); |
|
|
|
for (var len = adjacents.length, i = 0; i < len; i++) { |
|
var adjacent = adjacents[i]; |
|
var node = graph[adjacent]; |
|
|
|
if (node.distance === -1) { |
|
node.distance = graph[current].distance + 1; |
|
node.parent = current; |
|
queue.unshift(adjacent); |
|
} |
|
} |
|
} |
|
|
|
return graph; |
|
} |
|
|
|
function link(from, to) { |
|
return function (args) { |
|
return to(from(args)); |
|
}; |
|
} |
|
|
|
function wrapConversion(toModel, graph) { |
|
var path = [graph[toModel].parent, toModel]; |
|
var fn = conversions[graph[toModel].parent][toModel]; |
|
var cur = graph[toModel].parent; |
|
|
|
while (graph[cur].parent) { |
|
path.unshift(graph[cur].parent); |
|
fn = link(conversions[graph[cur].parent][cur], fn); |
|
cur = graph[cur].parent; |
|
} |
|
|
|
fn.conversion = path; |
|
return fn; |
|
} |
|
|
|
var route = function route(fromModel) { |
|
var graph = deriveBFS(fromModel); |
|
var conversion = {}; |
|
var models = Object.keys(graph); |
|
|
|
for (var len = models.length, i = 0; i < len; i++) { |
|
var toModel = models[i]; |
|
var node = graph[toModel]; |
|
|
|
if (node.parent === null) { |
|
// no possible conversion, or this node is the source model. |
|
continue; |
|
} |
|
|
|
conversion[toModel] = wrapConversion(toModel, graph); |
|
} |
|
|
|
return conversion; |
|
}; |
|
|
|
var convert = {}; |
|
var models$1 = Object.keys(conversions); |
|
|
|
function wrapRaw(fn) { |
|
var wrappedFn = function wrappedFn(args) { |
|
if (args === undefined || args === null) { |
|
return args; |
|
} |
|
|
|
if (arguments.length > 1) { |
|
args = Array.prototype.slice.call(arguments); |
|
} |
|
|
|
return fn(args); |
|
}; // preserve .conversion property if there is one |
|
|
|
|
|
if ('conversion' in fn) { |
|
wrappedFn.conversion = fn.conversion; |
|
} |
|
|
|
return wrappedFn; |
|
} |
|
|
|
function wrapRounded(fn) { |
|
var wrappedFn = function wrappedFn(args) { |
|
if (args === undefined || args === null) { |
|
return args; |
|
} |
|
|
|
if (arguments.length > 1) { |
|
args = Array.prototype.slice.call(arguments); |
|
} |
|
|
|
var result = fn(args); // we're assuming the result is an array here. |
|
// see notice in conversions.js; don't use box types |
|
// in conversion functions. |
|
|
|
if (_typeof(result) === 'object') { |
|
for (var len = result.length, i = 0; i < len; i++) { |
|
result[i] = Math.round(result[i]); |
|
} |
|
} |
|
|
|
return result; |
|
}; // preserve .conversion property if there is one |
|
|
|
|
|
if ('conversion' in fn) { |
|
wrappedFn.conversion = fn.conversion; |
|
} |
|
|
|
return wrappedFn; |
|
} |
|
|
|
models$1.forEach(function (fromModel) { |
|
convert[fromModel] = {}; |
|
Object.defineProperty(convert[fromModel], 'channels', { |
|
value: conversions[fromModel].channels |
|
}); |
|
Object.defineProperty(convert[fromModel], 'labels', { |
|
value: conversions[fromModel].labels |
|
}); |
|
var routes = route(fromModel); |
|
var routeModels = Object.keys(routes); |
|
routeModels.forEach(function (toModel) { |
|
var fn = routes[toModel]; |
|
convert[fromModel][toModel] = wrapRounded(fn); |
|
convert[fromModel][toModel].raw = wrapRaw(fn); |
|
}); |
|
}); |
|
var colorConvert = convert; |
|
|
|
var ansiStyles = createCommonjsModule(function (module) { |
|
|
|
var wrapAnsi16 = function wrapAnsi16(fn, offset) { |
|
return function () { |
|
var code = fn.apply(colorConvert, arguments); |
|
return "\x1B[".concat(code + offset, "m"); |
|
}; |
|
}; |
|
|
|
var wrapAnsi256 = function wrapAnsi256(fn, offset) { |
|
return function () { |
|
var code = fn.apply(colorConvert, arguments); |
|
return "\x1B[".concat(38 + offset, ";5;").concat(code, "m"); |
|
}; |
|
}; |
|
|
|
var wrapAnsi16m = function wrapAnsi16m(fn, offset) { |
|
return function () { |
|
var rgb = fn.apply(colorConvert, arguments); |
|
return "\x1B[".concat(38 + offset, ";2;").concat(rgb[0], ";").concat(rgb[1], ";").concat(rgb[2], "m"); |
|
}; |
|
}; |
|
|
|
function assembleStyles() { |
|
var codes = new Map(); |
|
var styles = { |
|
modifier: { |
|
reset: [0, 0], |
|
// 21 isn't widely supported and 22 does the same thing |
|
bold: [1, 22], |
|
dim: [2, 22], |
|
italic: [3, 23], |
|
underline: [4, 24], |
|
inverse: [7, 27], |
|
hidden: [8, 28], |
|
strikethrough: [9, 29] |
|
}, |
|
color: { |
|
black: [30, 39], |
|
red: [31, 39], |
|
green: [32, 39], |
|
yellow: [33, 39], |
|
blue: [34, 39], |
|
magenta: [35, 39], |
|
cyan: [36, 39], |
|
white: [37, 39], |
|
gray: [90, 39], |
|
// Bright color |
|
redBright: [91, 39], |
|
greenBright: [92, 39], |
|
yellowBright: [93, 39], |
|
blueBright: [94, 39], |
|
magentaBright: [95, 39], |
|
cyanBright: [96, 39], |
|
whiteBright: [97, 39] |
|
}, |
|
bgColor: { |
|
bgBlack: [40, 49], |
|
bgRed: [41, 49], |
|
bgGreen: [42, 49], |
|
bgYellow: [43, 49], |
|
bgBlue: [44, 49], |
|
bgMagenta: [45, 49], |
|
bgCyan: [46, 49], |
|
bgWhite: [47, 49], |
|
// Bright color |
|
bgBlackBright: [100, 49], |
|
bgRedBright: [101, 49], |
|
bgGreenBright: [102, 49], |
|
bgYellowBright: [103, 49], |
|
bgBlueBright: [104, 49], |
|
bgMagentaBright: [105, 49], |
|
bgCyanBright: [106, 49], |
|
bgWhiteBright: [107, 49] |
|
} |
|
}; // Fix humans |
|
|
|
styles.color.grey = styles.color.gray; |
|
|
|
for (var _i = 0, _Object$keys = Object.keys(styles); _i < _Object$keys.length; _i++) { |
|
var groupName = _Object$keys[_i]; |
|
var group = styles[groupName]; |
|
|
|
for (var _i3 = 0, _Object$keys3 = Object.keys(group); _i3 < _Object$keys3.length; _i3++) { |
|
var styleName = _Object$keys3[_i3]; |
|
var style = group[styleName]; |
|
styles[styleName] = { |
|
open: "\x1B[".concat(style[0], "m"), |
|
close: "\x1B[".concat(style[1], "m") |
|
}; |
|
group[styleName] = styles[styleName]; |
|
codes.set(style[0], style[1]); |
|
} |
|
|
|
Object.defineProperty(styles, groupName, { |
|
value: group, |
|
enumerable: false |
|
}); |
|
Object.defineProperty(styles, 'codes', { |
|
value: codes, |
|
enumerable: false |
|
}); |
|
} |
|
|
|
var ansi2ansi = function ansi2ansi(n) { |
|
return n; |
|
}; |
|
|
|
var rgb2rgb = function rgb2rgb(r, g, b) { |
|
return [r, g, b]; |
|
}; |
|
|
|
styles.color.close = "\x1B[39m"; |
|
styles.bgColor.close = "\x1B[49m"; |
|
styles.color.ansi = { |
|
ansi: wrapAnsi16(ansi2ansi, 0) |
|
}; |
|
styles.color.ansi256 = { |
|
ansi256: wrapAnsi256(ansi2ansi, 0) |
|
}; |
|
styles.color.ansi16m = { |
|
rgb: wrapAnsi16m(rgb2rgb, 0) |
|
}; |
|
styles.bgColor.ansi = { |
|
ansi: wrapAnsi16(ansi2ansi, 10) |
|
}; |
|
styles.bgColor.ansi256 = { |
|
ansi256: wrapAnsi256(ansi2ansi, 10) |
|
}; |
|
styles.bgColor.ansi16m = { |
|
rgb: wrapAnsi16m(rgb2rgb, 10) |
|
}; |
|
|
|
for (var _i2 = 0, _Object$keys2 = Object.keys(colorConvert); _i2 < _Object$keys2.length; _i2++) { |
|
var key = _Object$keys2[_i2]; |
|
|
|
if (_typeof(colorConvert[key]) !== 'object') { |
|
continue; |
|
} |
|
|
|
var suite = colorConvert[key]; |
|
|
|
if (key === 'ansi16') { |
|
key = 'ansi'; |
|
} |
|
|
|
if ('ansi16' in suite) { |
|
styles.color.ansi[key] = wrapAnsi16(suite.ansi16, 0); |
|
styles.bgColor.ansi[key] = wrapAnsi16(suite.ansi16, 10); |
|
} |
|
|
|
if ('ansi256' in suite) { |
|
styles.color.ansi256[key] = wrapAnsi256(suite.ansi256, 0); |
|
styles.bgColor.ansi256[key] = wrapAnsi256(suite.ansi256, 10); |
|
} |
|
|
|
if ('rgb' in suite) { |
|
styles.color.ansi16m[key] = wrapAnsi16m(suite.rgb, 0); |
|
styles.bgColor.ansi16m[key] = wrapAnsi16m(suite.rgb, 10); |
|
} |
|
} |
|
|
|
return styles; |
|
} // Make the export immutable |
|
|
|
|
|
Object.defineProperty(module, 'exports', { |
|
enumerable: true, |
|
get: assembleStyles |
|
}); |
|
}); |
|
|
|
var require$$0$1 = { |
|
EOL: "\n" |
|
}; |
|
|
|
var hasFlag = function hasFlag(flag, argv) { |
|
argv = argv || process.argv; |
|
var prefix = flag.startsWith('-') ? '' : flag.length === 1 ? '-' : '--'; |
|
var pos = argv.indexOf(prefix + flag); |
|
var terminatorPos = argv.indexOf('--'); |
|
return pos !== -1 && (terminatorPos === -1 ? true : pos < terminatorPos); |
|
}; |
|
|
|
var env$1 = process.env; |
|
var forceColor; |
|
|
|
if (hasFlag('no-color') || hasFlag('no-colors') || hasFlag('color=false')) { |
|
forceColor = false; |
|
} else if (hasFlag('color') || hasFlag('colors') || hasFlag('color=true') || hasFlag('color=always')) { |
|
forceColor = true; |
|
} |
|
|
|
if ('FORCE_COLOR' in env$1) { |
|
forceColor = env$1.FORCE_COLOR.length === 0 || parseInt(env$1.FORCE_COLOR, 10) !== 0; |
|
} |
|
|
|
function translateLevel(level) { |
|
if (level === 0) { |
|
return false; |
|
} |
|
|
|
return { |
|
level: level, |
|
hasBasic: true, |
|
has256: level >= 2, |
|
has16m: level >= 3 |
|
}; |
|
} |
|
|
|
function supportsColor(stream) { |
|
if (forceColor === false) { |
|
return 0; |
|
} |
|
|
|
if (hasFlag('color=16m') || hasFlag('color=full') || hasFlag('color=truecolor')) { |
|
return 3; |
|
} |
|
|
|
if (hasFlag('color=256')) { |
|
return 2; |
|
} |
|
|
|
if (stream && !stream.isTTY && forceColor !== true) { |
|
return 0; |
|
} |
|
|
|
var min = forceColor ? 1 : 0; |
|
|
|
if (process.platform === 'win32') { |
|
// Node.js 7.5.0 is the first version of Node.js to include a patch to |
|
// libuv that enables 256 color output on Windows. Anything earlier and it |
|
// won't work. However, here we target Node.js 8 at minimum as it is an LTS |
|
// release, and Node.js 7 is not. Windows 10 build 10586 is the first Windows |
|
// release that supports 256 colors. Windows 10 build 14931 is the first release |
|
// that supports 16m/TrueColor. |
|
var osRelease = require$$0$1.release().split('.'); |
|
|
|
if (Number(process.versions.node.split('.')[0]) >= 8 && Number(osRelease[0]) >= 10 && Number(osRelease[2]) >= 10586) { |
|
return Number(osRelease[2]) >= 14931 ? 3 : 2; |
|
} |
|
|
|
return 1; |
|
} |
|
|
|
if ('CI' in env$1) { |
|
if (['TRAVIS', 'CIRCLECI', 'APPVEYOR', 'GITLAB_CI'].some(function (sign) { |
|
return sign in env$1; |
|
}) || env$1.CI_NAME === 'codeship') { |
|
return 1; |
|
} |
|
|
|
return min; |
|
} |
|
|
|
if ('TEAMCITY_VERSION' in env$1) { |
|
return /^(9\.(0*[1-9]\d*)\.|\d{2,}\.)/.test(env$1.TEAMCITY_VERSION) ? 1 : 0; |
|
} |
|
|
|
if (env$1.COLORTERM === 'truecolor') { |
|
return 3; |
|
} |
|
|
|
if ('TERM_PROGRAM' in env$1) { |
|
var version = parseInt((env$1.TERM_PROGRAM_VERSION || '').split('.')[0], 10); |
|
|
|
switch (env$1.TERM_PROGRAM) { |
|
case 'iTerm.app': |
|
return version >= 3 ? 3 : 2; |
|
|
|
case 'Apple_Terminal': |
|
return 2; |
|
// No default |
|
} |
|
} |
|
|
|
if (/-256(color)?$/i.test(env$1.TERM)) { |
|
return 2; |
|
} |
|
|
|
if (/^screen|^xterm|^vt100|^vt220|^rxvt|color|ansi|cygwin|linux/i.test(env$1.TERM)) { |
|
return 1; |
|
} |
|
|
|
if ('COLORTERM' in env$1) { |
|
return 1; |
|
} |
|
|
|
if (env$1.TERM === 'dumb') { |
|
return min; |
|
} |
|
|
|
return min; |
|
} |
|
|
|
function getSupportLevel(stream) { |
|
var level = supportsColor(stream); |
|
return translateLevel(level); |
|
} |
|
|
|
var supportsColor_1 = { |
|
supportsColor: getSupportLevel, |
|
stdout: getSupportLevel(process.stdout), |
|
stderr: getSupportLevel(process.stderr) |
|
}; |
|
|
|
var TEMPLATE_REGEX = /(?:\\(u[a-f\d]{4}|x[a-f\d]{2}|.))|(?:\{(~)?(\w+(?:\([^)]*\))?(?:\.\w+(?:\([^)]*\))?)*)(?:[ \t]|(?=\r?\n)))|(\})|((?:.|[\r\n\f])+?)/gi; |
|
var STYLE_REGEX = /(?:^|\.)(\w+)(?:\(([^)]*)\))?/g; |
|
var STRING_REGEX = /^(['"])((?:\\.|(?!\1)[^\\])*)\1$/; |
|
var ESCAPE_REGEX = /\\(u[a-f\d]{4}|x[a-f\d]{2}|.)|([^\\])/gi; |
|
var ESCAPES = new Map([['n', '\n'], ['r', '\r'], ['t', '\t'], ['b', '\b'], ['f', '\f'], ['v', '\v'], ['0', '\0'], ['\\', '\\'], ['e', "\x1B"], ['a', "\x07"]]); |
|
|
|
function unescape(c) { |
|
if (c[0] === 'u' && c.length === 5 || c[0] === 'x' && c.length === 3) { |
|
return String.fromCharCode(parseInt(c.slice(1), 16)); |
|
} |
|
|
|
return ESCAPES.get(c) || c; |
|
} |
|
|
|
function parseArguments(name, args) { |
|
var results = []; |
|
var chunks = args.trim().split(/\s*,\s*/g); |
|
var matches; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = chunks[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var chunk = _step.value; |
|
|
|
if (!isNaN(chunk)) { |
|
results.push(Number(chunk)); |
|
} else if (matches = chunk.match(STRING_REGEX)) { |
|
results.push(matches[2].replace(ESCAPE_REGEX, function (m, escape, chr) { |
|
return escape ? unescape(escape) : chr; |
|
})); |
|
} else { |
|
throw new Error("Invalid Chalk template style argument: ".concat(chunk, " (in style '").concat(name, "')")); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return results; |
|
} |
|
|
|
function parseStyle(style) { |
|
STYLE_REGEX.lastIndex = 0; |
|
var results = []; |
|
var matches; |
|
|
|
while ((matches = STYLE_REGEX.exec(style)) !== null) { |
|
var name = matches[1]; |
|
|
|
if (matches[2]) { |
|
var args = parseArguments(name, matches[2]); |
|
results.push([name].concat(args)); |
|
} else { |
|
results.push([name]); |
|
} |
|
} |
|
|
|
return results; |
|
} |
|
|
|
function buildStyle(chalk, styles) { |
|
var enabled = {}; |
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = styles[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var layer = _step2.value; |
|
var _iteratorNormalCompletion3 = true; |
|
var _didIteratorError3 = false; |
|
var _iteratorError3 = undefined; |
|
|
|
try { |
|
for (var _iterator3 = layer.styles[Symbol.iterator](), _step3; !(_iteratorNormalCompletion3 = (_step3 = _iterator3.next()).done); _iteratorNormalCompletion3 = true) { |
|
var style = _step3.value; |
|
enabled[style[0]] = layer.inverse ? null : style.slice(1); |
|
} |
|
} catch (err) { |
|
_didIteratorError3 = true; |
|
_iteratorError3 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion3 && _iterator3.return != null) { |
|
_iterator3.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError3) { |
|
throw _iteratorError3; |
|
} |
|
} |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
|
|
var current = chalk; |
|
|
|
for (var _i = 0, _Object$keys = Object.keys(enabled); _i < _Object$keys.length; _i++) { |
|
var styleName = _Object$keys[_i]; |
|
|
|
if (Array.isArray(enabled[styleName])) { |
|
if (!(styleName in current)) { |
|
throw new Error("Unknown Chalk style: ".concat(styleName)); |
|
} |
|
|
|
if (enabled[styleName].length > 0) { |
|
current = current[styleName].apply(current, enabled[styleName]); |
|
} else { |
|
current = current[styleName]; |
|
} |
|
} |
|
} |
|
|
|
return current; |
|
} |
|
|
|
var templates = function templates(chalk, tmp) { |
|
var styles = []; |
|
var chunks = []; |
|
var chunk = []; // eslint-disable-next-line max-params |
|
|
|
tmp.replace(TEMPLATE_REGEX, function (m, escapeChar, inverse, style, close, chr) { |
|
if (escapeChar) { |
|
chunk.push(unescape(escapeChar)); |
|
} else if (style) { |
|
var str = chunk.join(''); |
|
chunk = []; |
|
chunks.push(styles.length === 0 ? str : buildStyle(chalk, styles)(str)); |
|
styles.push({ |
|
inverse: inverse, |
|
styles: parseStyle(style) |
|
}); |
|
} else if (close) { |
|
if (styles.length === 0) { |
|
throw new Error('Found extraneous } in Chalk template literal'); |
|
} |
|
|
|
chunks.push(buildStyle(chalk, styles)(chunk.join(''))); |
|
chunk = []; |
|
styles.pop(); |
|
} else { |
|
chunk.push(chr); |
|
} |
|
}); |
|
chunks.push(chunk.join('')); |
|
|
|
if (styles.length > 0) { |
|
var errMsg = "Chalk template literal is missing ".concat(styles.length, " closing bracket").concat(styles.length === 1 ? '' : 's', " (`}`)"); |
|
throw new Error(errMsg); |
|
} |
|
|
|
return chunks.join(''); |
|
}; |
|
|
|
var chalk = createCommonjsModule(function (module) { |
|
|
|
var stdoutColor = supportsColor_1.stdout; |
|
var isSimpleWindowsTerm = process.platform === 'win32' && !(process.env.TERM || '').toLowerCase().startsWith('xterm'); // `supportsColor.level` → `ansiStyles.color[name]` mapping |
|
|
|
var levelMapping = ['ansi', 'ansi', 'ansi256', 'ansi16m']; // `color-convert` models to exclude from the Chalk API due to conflicts and such |
|
|
|
var skipModels = new Set(['gray']); |
|
var styles = Object.create(null); |
|
|
|
function applyOptions(obj, options) { |
|
options = options || {}; // Detect level if not set manually |
|
|
|
var scLevel = stdoutColor ? stdoutColor.level : 0; |
|
obj.level = options.level === undefined ? scLevel : options.level; |
|
obj.enabled = 'enabled' in options ? options.enabled : obj.level > 0; |
|
} |
|
|
|
function Chalk(options) { |
|
// We check for this.template here since calling `chalk.constructor()` |
|
// by itself will have a `this` of a previously constructed chalk object |
|
if (!this || !(this instanceof Chalk) || this.template) { |
|
var _chalk = {}; |
|
applyOptions(_chalk, options); |
|
|
|
_chalk.template = function () { |
|
var args = [].slice.call(arguments); |
|
return chalkTag.apply(null, [_chalk.template].concat(args)); |
|
}; |
|
|
|
Object.setPrototypeOf(_chalk, Chalk.prototype); |
|
Object.setPrototypeOf(_chalk.template, _chalk); |
|
_chalk.template.constructor = Chalk; |
|
return _chalk.template; |
|
} |
|
|
|
applyOptions(this, options); |
|
} // Use bright blue on Windows as the normal blue color is illegible |
|
|
|
|
|
if (isSimpleWindowsTerm) { |
|
ansiStyles.blue.open = "\x1B[94m"; |
|
} |
|
|
|
var _loop = function _loop() { |
|
var key = _Object$keys[_i]; |
|
ansiStyles[key].closeRe = new RegExp(escapeStringRegexp(ansiStyles[key].close), 'g'); |
|
styles[key] = { |
|
get: function get() { |
|
var codes = ansiStyles[key]; |
|
return build.call(this, this._styles ? this._styles.concat(codes) : [codes], this._empty, key); |
|
} |
|
}; |
|
}; |
|
|
|
for (var _i = 0, _Object$keys = Object.keys(ansiStyles); _i < _Object$keys.length; _i++) { |
|
_loop(); |
|
} |
|
|
|
styles.visible = { |
|
get: function get() { |
|
return build.call(this, this._styles || [], true, 'visible'); |
|
} |
|
}; |
|
ansiStyles.color.closeRe = new RegExp(escapeStringRegexp(ansiStyles.color.close), 'g'); |
|
|
|
var _loop2 = function _loop2() { |
|
var model = _Object$keys2[_i2]; |
|
|
|
if (skipModels.has(model)) { |
|
return "continue"; |
|
} |
|
|
|
styles[model] = { |
|
get: function get() { |
|
var level = this.level; |
|
return function () { |
|
var open = ansiStyles.color[levelMapping[level]][model].apply(null, arguments); |
|
var codes = { |
|
open: open, |
|
close: ansiStyles.color.close, |
|
closeRe: ansiStyles.color.closeRe |
|
}; |
|
return build.call(this, this._styles ? this._styles.concat(codes) : [codes], this._empty, model); |
|
}; |
|
} |
|
}; |
|
}; |
|
|
|
for (var _i2 = 0, _Object$keys2 = Object.keys(ansiStyles.color.ansi); _i2 < _Object$keys2.length; _i2++) { |
|
var _ret = _loop2(); |
|
|
|
if (_ret === "continue") continue; |
|
} |
|
|
|
ansiStyles.bgColor.closeRe = new RegExp(escapeStringRegexp(ansiStyles.bgColor.close), 'g'); |
|
|
|
var _loop3 = function _loop3() { |
|
var model = _Object$keys3[_i3]; |
|
|
|
if (skipModels.has(model)) { |
|
return "continue"; |
|
} |
|
|
|
var bgModel = 'bg' + model[0].toUpperCase() + model.slice(1); |
|
styles[bgModel] = { |
|
get: function get() { |
|
var level = this.level; |
|
return function () { |
|
var open = ansiStyles.bgColor[levelMapping[level]][model].apply(null, arguments); |
|
var codes = { |
|
open: open, |
|
close: ansiStyles.bgColor.close, |
|
closeRe: ansiStyles.bgColor.closeRe |
|
}; |
|
return build.call(this, this._styles ? this._styles.concat(codes) : [codes], this._empty, model); |
|
}; |
|
} |
|
}; |
|
}; |
|
|
|
for (var _i3 = 0, _Object$keys3 = Object.keys(ansiStyles.bgColor.ansi); _i3 < _Object$keys3.length; _i3++) { |
|
var _ret2 = _loop3(); |
|
|
|
if (_ret2 === "continue") continue; |
|
} |
|
|
|
var proto = Object.defineProperties(function () {}, styles); |
|
|
|
function build(_styles, _empty, key) { |
|
var builder = function builder() { |
|
return applyStyle.apply(builder, arguments); |
|
}; |
|
|
|
builder._styles = _styles; |
|
builder._empty = _empty; |
|
var self = this; |
|
Object.defineProperty(builder, 'level', { |
|
enumerable: true, |
|
get: function get() { |
|
return self.level; |
|
}, |
|
set: function set(level) { |
|
self.level = level; |
|
} |
|
}); |
|
Object.defineProperty(builder, 'enabled', { |
|
enumerable: true, |
|
get: function get() { |
|
return self.enabled; |
|
}, |
|
set: function set(enabled) { |
|
self.enabled = enabled; |
|
} |
|
}); // See below for fix regarding invisible grey/dim combination on Windows |
|
|
|
builder.hasGrey = this.hasGrey || key === 'gray' || key === 'grey'; // `__proto__` is used because we must return a function, but there is |
|
// no way to create a function with a different prototype |
|
|
|
builder.__proto__ = proto; // eslint-disable-line no-proto |
|
|
|
return builder; |
|
} |
|
|
|
function applyStyle() { |
|
// Support varags, but simply cast to string in case there's only one arg |
|
var args = arguments; |
|
var argsLen = args.length; |
|
var str = String(arguments[0]); |
|
|
|
if (argsLen === 0) { |
|
return ''; |
|
} |
|
|
|
if (argsLen > 1) { |
|
// Don't slice `arguments`, it prevents V8 optimizations |
|
for (var a = 1; a < argsLen; a++) { |
|
str += ' ' + args[a]; |
|
} |
|
} |
|
|
|
if (!this.enabled || this.level <= 0 || !str) { |
|
return this._empty ? '' : str; |
|
} // Turns out that on Windows dimmed gray text becomes invisible in cmd.exe, |
|
// see https://github.com/chalk/chalk/issues/58 |
|
// If we're on Windows and we're dealing with a gray color, temporarily make 'dim' a noop. |
|
|
|
|
|
var originalDim = ansiStyles.dim.open; |
|
|
|
if (isSimpleWindowsTerm && this.hasGrey) { |
|
ansiStyles.dim.open = ''; |
|
} |
|
|
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = this._styles.slice().reverse()[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var code = _step.value; |
|
// Replace any instances already present with a re-opening code |
|
// otherwise only the part of the string until said closing code |
|
// will be colored, and the rest will simply be 'plain'. |
|
str = code.open + str.replace(code.closeRe, code.open) + code.close; // Close the styling before a linebreak and reopen |
|
// after next line to fix a bleed issue on macOS |
|
// https://github.com/chalk/chalk/pull/92 |
|
|
|
str = str.replace(/\r?\n/g, "".concat(code.close, "$&").concat(code.open)); |
|
} // Reset the original `dim` if we changed it to work around the Windows dimmed gray issue |
|
|
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
ansiStyles.dim.open = originalDim; |
|
return str; |
|
} |
|
|
|
function chalkTag(chalk, strings) { |
|
if (!Array.isArray(strings)) { |
|
// If chalk() was called by itself or with a string, |
|
// return the string itself as a string. |
|
return [].slice.call(arguments, 1).join(' '); |
|
} |
|
|
|
var args = [].slice.call(arguments, 2); |
|
var parts = [strings.raw[0]]; |
|
|
|
for (var i = 1; i < strings.length; i++) { |
|
parts.push(String(args[i - 1]).replace(/[{}\\]/g, '\\$&')); |
|
parts.push(String(strings.raw[i])); |
|
} |
|
|
|
return templates(chalk, parts.join('')); |
|
} |
|
|
|
Object.defineProperties(Chalk.prototype, styles); |
|
module.exports = Chalk(); // eslint-disable-line new-cap |
|
|
|
module.exports.supportsColor = stdoutColor; |
|
module.exports.default = module.exports; // For TypeScript |
|
}); |
|
var chalk_1 = chalk.supportsColor; |
|
|
|
var common = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
exports.commonDeprecatedHandler = function (keyOrPair, redirectTo, _ref) { |
|
var descriptor = _ref.descriptor; |
|
var messages = ["".concat(chalk.default.yellow(typeof keyOrPair === 'string' ? descriptor.key(keyOrPair) : descriptor.pair(keyOrPair)), " is deprecated")]; |
|
|
|
if (redirectTo) { |
|
messages.push("we now treat it as ".concat(chalk.default.blue(typeof redirectTo === 'string' ? descriptor.key(redirectTo) : descriptor.pair(redirectTo)))); |
|
} |
|
|
|
return messages.join('; ') + '.'; |
|
}; |
|
}); |
|
unwrapExports(common); |
|
var common_1 = common.commonDeprecatedHandler; |
|
|
|
var deprecated = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
tslib_1.__exportStar(common, exports); |
|
}); |
|
unwrapExports(deprecated); |
|
|
|
var common$1 = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
exports.commonInvalidHandler = function (key, value, utils) { |
|
return ["Invalid ".concat(chalk.default.red(utils.descriptor.key(key)), " value."), "Expected ".concat(chalk.default.blue(utils.schemas[key].expected(utils)), ","), "but received ".concat(chalk.default.red(utils.descriptor.value(value)), ".")].join(' '); |
|
}; |
|
}); |
|
unwrapExports(common$1); |
|
var common_1$1 = common$1.commonInvalidHandler; |
|
|
|
var invalid = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
tslib_1.__exportStar(common$1, exports); |
|
}); |
|
unwrapExports(invalid); |
|
|
|
/* eslint-disable no-nested-ternary */ |
|
|
|
var arr = []; |
|
var charCodeCache = []; |
|
|
|
var leven = function leven(a, b) { |
|
if (a === b) { |
|
return 0; |
|
} |
|
|
|
var swap = a; // Swapping the strings if `a` is longer than `b` so we know which one is the |
|
// shortest & which one is the longest |
|
|
|
if (a.length > b.length) { |
|
a = b; |
|
b = swap; |
|
} |
|
|
|
var aLen = a.length; |
|
var bLen = b.length; |
|
|
|
if (aLen === 0) { |
|
return bLen; |
|
} |
|
|
|
if (bLen === 0) { |
|
return aLen; |
|
} // Performing suffix trimming: |
|
// We can linearly drop suffix common to both strings since they |
|
// don't increase distance at all |
|
// Note: `~-` is the bitwise way to perform a `- 1` operation |
|
|
|
|
|
while (aLen > 0 && a.charCodeAt(~-aLen) === b.charCodeAt(~-bLen)) { |
|
aLen--; |
|
bLen--; |
|
} |
|
|
|
if (aLen === 0) { |
|
return bLen; |
|
} // Performing prefix trimming |
|
// We can linearly drop prefix common to both strings since they |
|
// don't increase distance at all |
|
|
|
|
|
var start = 0; |
|
|
|
while (start < aLen && a.charCodeAt(start) === b.charCodeAt(start)) { |
|
start++; |
|
} |
|
|
|
aLen -= start; |
|
bLen -= start; |
|
|
|
if (aLen === 0) { |
|
return bLen; |
|
} |
|
|
|
var bCharCode; |
|
var ret; |
|
var tmp; |
|
var tmp2; |
|
var i = 0; |
|
var j = 0; |
|
|
|
while (i < aLen) { |
|
charCodeCache[start + i] = a.charCodeAt(start + i); |
|
arr[i] = ++i; |
|
} |
|
|
|
while (j < bLen) { |
|
bCharCode = b.charCodeAt(start + j); |
|
tmp = j++; |
|
ret = j; |
|
|
|
for (i = 0; i < aLen; i++) { |
|
tmp2 = bCharCode === charCodeCache[start + i] ? tmp : tmp + 1; |
|
tmp = arr[i]; |
|
ret = arr[i] = tmp > ret ? tmp2 > ret ? ret + 1 : tmp2 : tmp2 > tmp ? tmp + 1 : tmp2; |
|
} |
|
} |
|
|
|
return ret; |
|
}; |
|
|
|
var leven_1 = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
exports.levenUnknownHandler = function (key, value, _ref) { |
|
var descriptor = _ref.descriptor, |
|
logger = _ref.logger, |
|
schemas = _ref.schemas; |
|
var messages = ["Ignored unknown option ".concat(chalk.default.yellow(descriptor.pair({ |
|
key: key, |
|
value: value |
|
})), ".")]; |
|
var suggestion = Object.keys(schemas).sort().find(function (knownKey) { |
|
return leven(key, knownKey) < 3; |
|
}); |
|
|
|
if (suggestion) { |
|
messages.push("Did you mean ".concat(chalk.default.blue(descriptor.key(suggestion)), "?")); |
|
} |
|
|
|
logger.warn(messages.join(' ')); |
|
}; |
|
}); |
|
unwrapExports(leven_1); |
|
var leven_2 = leven_1.levenUnknownHandler; |
|
|
|
var unknown = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
tslib_1.__exportStar(leven_1, exports); |
|
}); |
|
unwrapExports(unknown); |
|
|
|
var handlers = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
tslib_1.__exportStar(deprecated, exports); |
|
|
|
tslib_1.__exportStar(invalid, exports); |
|
|
|
tslib_1.__exportStar(unknown, exports); |
|
}); |
|
unwrapExports(handlers); |
|
|
|
var schema = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
var HANDLER_KEYS = ['default', 'expected', 'validate', 'deprecated', 'forward', 'redirect', 'overlap', 'preprocess', 'postprocess']; |
|
|
|
function createSchema(SchemaConstructor, parameters) { |
|
var schema = new SchemaConstructor(parameters); |
|
var subSchema = Object.create(schema); |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = HANDLER_KEYS[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var handlerKey = _step.value; |
|
|
|
if (handlerKey in parameters) { |
|
subSchema[handlerKey] = normalizeHandler(parameters[handlerKey], schema, Schema.prototype[handlerKey].length); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return subSchema; |
|
} |
|
|
|
exports.createSchema = createSchema; |
|
|
|
var Schema = |
|
/*#__PURE__*/ |
|
function () { |
|
function Schema(parameters) { |
|
_classCallCheck(this, Schema); |
|
|
|
this.name = parameters.name; |
|
} |
|
|
|
_createClass(Schema, [{ |
|
key: "default", |
|
value: function _default(_utils) { |
|
return undefined; |
|
} // istanbul ignore next: this is actually an abstract method but we need a placeholder to get `function.length` |
|
|
|
}, { |
|
key: "expected", |
|
value: function expected(_utils) { |
|
return 'nothing'; |
|
} // istanbul ignore next: this is actually an abstract method but we need a placeholder to get `function.length` |
|
|
|
}, { |
|
key: "validate", |
|
value: function validate(_value, _utils) { |
|
return false; |
|
} |
|
}, { |
|
key: "deprecated", |
|
value: function deprecated(_value, _utils) { |
|
return false; |
|
} |
|
}, { |
|
key: "forward", |
|
value: function forward(_value, _utils) { |
|
return undefined; |
|
} |
|
}, { |
|
key: "redirect", |
|
value: function redirect(_value, _utils) { |
|
return undefined; |
|
} |
|
}, { |
|
key: "overlap", |
|
value: function overlap(currentValue, _newValue, _utils) { |
|
return currentValue; |
|
} |
|
}, { |
|
key: "preprocess", |
|
value: function preprocess(value, _utils) { |
|
return value; |
|
} |
|
}, { |
|
key: "postprocess", |
|
value: function postprocess(value, _utils) { |
|
return value; |
|
} |
|
}], [{ |
|
key: "create", |
|
value: function create(parameters) { |
|
// @ts-ignore: https://github.com/Microsoft/TypeScript/issues/5863 |
|
return createSchema(this, parameters); |
|
} |
|
}]); |
|
|
|
return Schema; |
|
}(); |
|
|
|
exports.Schema = Schema; |
|
|
|
function normalizeHandler(handler, superSchema, handlerArgumentsLength) { |
|
return typeof handler === 'function' ? function () { |
|
for (var _len = arguments.length, args = new Array(_len), _key = 0; _key < _len; _key++) { |
|
args[_key] = arguments[_key]; |
|
} |
|
|
|
return handler.apply(void 0, _toConsumableArray(args.slice(0, handlerArgumentsLength - 1)).concat([superSchema], _toConsumableArray(args.slice(handlerArgumentsLength - 1)))); |
|
} : function () { |
|
return handler; |
|
}; |
|
} |
|
}); |
|
unwrapExports(schema); |
|
var schema_1 = schema.createSchema; |
|
var schema_2 = schema.Schema; |
|
|
|
var alias = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
var AliasSchema = |
|
/*#__PURE__*/ |
|
function (_schema_1$Schema) { |
|
_inherits(AliasSchema, _schema_1$Schema); |
|
|
|
function AliasSchema(parameters) { |
|
var _this; |
|
|
|
_classCallCheck(this, AliasSchema); |
|
|
|
_this = _possibleConstructorReturn(this, _getPrototypeOf(AliasSchema).call(this, parameters)); |
|
_this._sourceName = parameters.sourceName; |
|
return _this; |
|
} |
|
|
|
_createClass(AliasSchema, [{ |
|
key: "expected", |
|
value: function expected(utils) { |
|
return utils.schemas[this._sourceName].expected(utils); |
|
} |
|
}, { |
|
key: "validate", |
|
value: function validate(value, utils) { |
|
return utils.schemas[this._sourceName].validate(value, utils); |
|
} |
|
}, { |
|
key: "redirect", |
|
value: function redirect(_value, _utils) { |
|
return this._sourceName; |
|
} |
|
}]); |
|
|
|
return AliasSchema; |
|
}(schema.Schema); |
|
|
|
exports.AliasSchema = AliasSchema; |
|
}); |
|
unwrapExports(alias); |
|
var alias_1 = alias.AliasSchema; |
|
|
|
var any = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
var AnySchema = |
|
/*#__PURE__*/ |
|
function (_schema_1$Schema) { |
|
_inherits(AnySchema, _schema_1$Schema); |
|
|
|
function AnySchema() { |
|
_classCallCheck(this, AnySchema); |
|
|
|
return _possibleConstructorReturn(this, _getPrototypeOf(AnySchema).apply(this, arguments)); |
|
} |
|
|
|
_createClass(AnySchema, [{ |
|
key: "expected", |
|
value: function expected() { |
|
return 'anything'; |
|
} |
|
}, { |
|
key: "validate", |
|
value: function validate() { |
|
return true; |
|
} |
|
}]); |
|
|
|
return AnySchema; |
|
}(schema.Schema); |
|
|
|
exports.AnySchema = AnySchema; |
|
}); |
|
unwrapExports(any); |
|
var any_1 = any.AnySchema; |
|
|
|
var array = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
var ArraySchema = |
|
/*#__PURE__*/ |
|
function (_schema_1$Schema) { |
|
_inherits(ArraySchema, _schema_1$Schema); |
|
|
|
function ArraySchema(_a) { |
|
var _this; |
|
|
|
_classCallCheck(this, ArraySchema); |
|
|
|
var valueSchema = _a.valueSchema, |
|
_a$name = _a.name, |
|
name = _a$name === void 0 ? valueSchema.name : _a$name, |
|
handlers = tslib_1.__rest(_a, ["valueSchema", "name"]); |
|
|
|
_this = _possibleConstructorReturn(this, _getPrototypeOf(ArraySchema).call(this, Object.assign({}, handlers, { |
|
name: name |
|
}))); |
|
_this._valueSchema = valueSchema; |
|
return _this; |
|
} |
|
|
|
_createClass(ArraySchema, [{ |
|
key: "expected", |
|
value: function expected(utils) { |
|
return "an array of ".concat(this._valueSchema.expected(utils)); |
|
} |
|
}, { |
|
key: "validate", |
|
value: function validate(value, utils) { |
|
if (!Array.isArray(value)) { |
|
return false; |
|
} |
|
|
|
var invalidValues = []; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = value[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var subValue = _step.value; |
|
var subValidateResult = utils.normalizeValidateResult(this._valueSchema.validate(subValue, utils), subValue); |
|
|
|
if (subValidateResult !== true) { |
|
invalidValues.push(subValidateResult.value); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return invalidValues.length === 0 ? true : { |
|
value: invalidValues |
|
}; |
|
} |
|
}, { |
|
key: "deprecated", |
|
value: function deprecated(value, utils) { |
|
var deprecatedResult = []; |
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = value[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var subValue = _step2.value; |
|
var subDeprecatedResult = utils.normalizeDeprecatedResult(this._valueSchema.deprecated(subValue, utils), subValue); |
|
|
|
if (subDeprecatedResult !== false) { |
|
deprecatedResult.push.apply(deprecatedResult, _toConsumableArray(subDeprecatedResult.map(function (_ref) { |
|
var deprecatedValue = _ref.value; |
|
return { |
|
value: [deprecatedValue] |
|
}; |
|
}))); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
|
|
return deprecatedResult; |
|
} |
|
}, { |
|
key: "forward", |
|
value: function forward(value, utils) { |
|
var forwardResult = []; |
|
var _iteratorNormalCompletion3 = true; |
|
var _didIteratorError3 = false; |
|
var _iteratorError3 = undefined; |
|
|
|
try { |
|
for (var _iterator3 = value[Symbol.iterator](), _step3; !(_iteratorNormalCompletion3 = (_step3 = _iterator3.next()).done); _iteratorNormalCompletion3 = true) { |
|
var subValue = _step3.value; |
|
var subForwardResult = utils.normalizeForwardResult(this._valueSchema.forward(subValue, utils), subValue); |
|
forwardResult.push.apply(forwardResult, _toConsumableArray(subForwardResult.map(wrapTransferResult))); |
|
} |
|
} catch (err) { |
|
_didIteratorError3 = true; |
|
_iteratorError3 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion3 && _iterator3.return != null) { |
|
_iterator3.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError3) { |
|
throw _iteratorError3; |
|
} |
|
} |
|
} |
|
|
|
return forwardResult; |
|
} |
|
}, { |
|
key: "redirect", |
|
value: function redirect(value, utils) { |
|
var remain = []; |
|
var redirect = []; |
|
var _iteratorNormalCompletion4 = true; |
|
var _didIteratorError4 = false; |
|
var _iteratorError4 = undefined; |
|
|
|
try { |
|
for (var _iterator4 = value[Symbol.iterator](), _step4; !(_iteratorNormalCompletion4 = (_step4 = _iterator4.next()).done); _iteratorNormalCompletion4 = true) { |
|
var subValue = _step4.value; |
|
var subRedirectResult = utils.normalizeRedirectResult(this._valueSchema.redirect(subValue, utils), subValue); |
|
|
|
if ('remain' in subRedirectResult) { |
|
remain.push(subRedirectResult.remain); |
|
} |
|
|
|
redirect.push.apply(redirect, _toConsumableArray(subRedirectResult.redirect.map(wrapTransferResult))); |
|
} |
|
} catch (err) { |
|
_didIteratorError4 = true; |
|
_iteratorError4 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion4 && _iterator4.return != null) { |
|
_iterator4.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError4) { |
|
throw _iteratorError4; |
|
} |
|
} |
|
} |
|
|
|
return remain.length === 0 ? { |
|
redirect: redirect |
|
} : { |
|
redirect: redirect, |
|
remain: remain |
|
}; |
|
} |
|
}, { |
|
key: "overlap", |
|
value: function overlap(currentValue, newValue) { |
|
return currentValue.concat(newValue); |
|
} |
|
}]); |
|
|
|
return ArraySchema; |
|
}(schema.Schema); |
|
|
|
exports.ArraySchema = ArraySchema; |
|
|
|
function wrapTransferResult(_ref2) { |
|
var from = _ref2.from, |
|
to = _ref2.to; |
|
return { |
|
from: [from], |
|
to: to |
|
}; |
|
} |
|
}); |
|
unwrapExports(array); |
|
var array_1 = array.ArraySchema; |
|
|
|
var boolean_1 = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
var BooleanSchema = |
|
/*#__PURE__*/ |
|
function (_schema_1$Schema) { |
|
_inherits(BooleanSchema, _schema_1$Schema); |
|
|
|
function BooleanSchema() { |
|
_classCallCheck(this, BooleanSchema); |
|
|
|
return _possibleConstructorReturn(this, _getPrototypeOf(BooleanSchema).apply(this, arguments)); |
|
} |
|
|
|
_createClass(BooleanSchema, [{ |
|
key: "expected", |
|
value: function expected() { |
|
return 'true or false'; |
|
} |
|
}, { |
|
key: "validate", |
|
value: function validate(value) { |
|
return typeof value === 'boolean'; |
|
} |
|
}]); |
|
|
|
return BooleanSchema; |
|
}(schema.Schema); |
|
|
|
exports.BooleanSchema = BooleanSchema; |
|
}); |
|
unwrapExports(boolean_1); |
|
var boolean_2 = boolean_1.BooleanSchema; |
|
|
|
var utils = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
function recordFromArray(array, mainKey) { |
|
var record = Object.create(null); |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = array[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var value = _step.value; |
|
var key = value[mainKey]; // istanbul ignore next |
|
|
|
if (record[key]) { |
|
throw new Error("Duplicate ".concat(mainKey, " ").concat(JSON.stringify(key))); |
|
} // @ts-ignore |
|
|
|
|
|
record[key] = value; |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return record; |
|
} |
|
|
|
exports.recordFromArray = recordFromArray; |
|
|
|
function mapFromArray(array, mainKey) { |
|
var map = new Map(); |
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = array[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var value = _step2.value; |
|
var key = value[mainKey]; // istanbul ignore next |
|
|
|
if (map.has(key)) { |
|
throw new Error("Duplicate ".concat(mainKey, " ").concat(JSON.stringify(key))); |
|
} |
|
|
|
map.set(key, value); |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
|
|
return map; |
|
} |
|
|
|
exports.mapFromArray = mapFromArray; |
|
|
|
function createAutoChecklist() { |
|
var map = Object.create(null); |
|
return function (id) { |
|
var idString = JSON.stringify(id); |
|
|
|
if (map[idString]) { |
|
return true; |
|
} |
|
|
|
map[idString] = true; |
|
return false; |
|
}; |
|
} |
|
|
|
exports.createAutoChecklist = createAutoChecklist; |
|
|
|
function partition(array, predicate) { |
|
var trueArray = []; |
|
var falseArray = []; |
|
var _iteratorNormalCompletion3 = true; |
|
var _didIteratorError3 = false; |
|
var _iteratorError3 = undefined; |
|
|
|
try { |
|
for (var _iterator3 = array[Symbol.iterator](), _step3; !(_iteratorNormalCompletion3 = (_step3 = _iterator3.next()).done); _iteratorNormalCompletion3 = true) { |
|
var value = _step3.value; |
|
|
|
if (predicate(value)) { |
|
trueArray.push(value); |
|
} else { |
|
falseArray.push(value); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError3 = true; |
|
_iteratorError3 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion3 && _iterator3.return != null) { |
|
_iterator3.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError3) { |
|
throw _iteratorError3; |
|
} |
|
} |
|
} |
|
|
|
return [trueArray, falseArray]; |
|
} |
|
|
|
exports.partition = partition; |
|
|
|
function isInt(value) { |
|
return value === Math.floor(value); |
|
} |
|
|
|
exports.isInt = isInt; |
|
|
|
function comparePrimitive(a, b) { |
|
if (a === b) { |
|
return 0; |
|
} |
|
|
|
var typeofA = _typeof(a); |
|
|
|
var typeofB = _typeof(b); |
|
|
|
var orders = ['undefined', 'object', 'boolean', 'number', 'string']; |
|
|
|
if (typeofA !== typeofB) { |
|
return orders.indexOf(typeofA) - orders.indexOf(typeofB); |
|
} |
|
|
|
if (typeofA !== 'string') { |
|
return Number(a) - Number(b); |
|
} |
|
|
|
return a.localeCompare(b); |
|
} |
|
|
|
exports.comparePrimitive = comparePrimitive; |
|
|
|
function normalizeDefaultResult(result) { |
|
return result === undefined ? {} : result; |
|
} |
|
|
|
exports.normalizeDefaultResult = normalizeDefaultResult; |
|
|
|
function normalizeValidateResult(result, value) { |
|
return result === true ? true : result === false ? { |
|
value: value |
|
} : result; |
|
} |
|
|
|
exports.normalizeValidateResult = normalizeValidateResult; |
|
|
|
function normalizeDeprecatedResult(result, value) { |
|
var doNotNormalizeTrue = arguments.length > 2 && arguments[2] !== undefined ? arguments[2] : false; |
|
return result === false ? false : result === true ? doNotNormalizeTrue ? true : [{ |
|
value: value |
|
}] : 'value' in result ? [result] : result.length === 0 ? false : result; |
|
} |
|
|
|
exports.normalizeDeprecatedResult = normalizeDeprecatedResult; |
|
|
|
function normalizeTransferResult(result, value) { |
|
return typeof result === 'string' || 'key' in result ? { |
|
from: value, |
|
to: result |
|
} : 'from' in result ? { |
|
from: result.from, |
|
to: result.to |
|
} : { |
|
from: value, |
|
to: result.to |
|
}; |
|
} |
|
|
|
exports.normalizeTransferResult = normalizeTransferResult; |
|
|
|
function normalizeForwardResult(result, value) { |
|
return result === undefined ? [] : Array.isArray(result) ? result.map(function (transferResult) { |
|
return normalizeTransferResult(transferResult, value); |
|
}) : [normalizeTransferResult(result, value)]; |
|
} |
|
|
|
exports.normalizeForwardResult = normalizeForwardResult; |
|
|
|
function normalizeRedirectResult(result, value) { |
|
var redirect = normalizeForwardResult(_typeof(result) === 'object' && 'redirect' in result ? result.redirect : result, value); |
|
return redirect.length === 0 ? { |
|
remain: value, |
|
redirect: redirect |
|
} : _typeof(result) === 'object' && 'remain' in result ? { |
|
remain: result.remain, |
|
redirect: redirect |
|
} : { |
|
redirect: redirect |
|
}; |
|
} |
|
|
|
exports.normalizeRedirectResult = normalizeRedirectResult; |
|
}); |
|
unwrapExports(utils); |
|
var utils_1 = utils.recordFromArray; |
|
var utils_2 = utils.mapFromArray; |
|
var utils_3 = utils.createAutoChecklist; |
|
var utils_4 = utils.partition; |
|
var utils_5 = utils.isInt; |
|
var utils_6 = utils.comparePrimitive; |
|
var utils_7 = utils.normalizeDefaultResult; |
|
var utils_8 = utils.normalizeValidateResult; |
|
var utils_9 = utils.normalizeDeprecatedResult; |
|
var utils_10 = utils.normalizeTransferResult; |
|
var utils_11 = utils.normalizeForwardResult; |
|
var utils_12 = utils.normalizeRedirectResult; |
|
|
|
var choice = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
var ChoiceSchema = |
|
/*#__PURE__*/ |
|
function (_schema_1$Schema) { |
|
_inherits(ChoiceSchema, _schema_1$Schema); |
|
|
|
function ChoiceSchema(parameters) { |
|
var _this; |
|
|
|
_classCallCheck(this, ChoiceSchema); |
|
|
|
_this = _possibleConstructorReturn(this, _getPrototypeOf(ChoiceSchema).call(this, parameters)); |
|
_this._choices = utils.mapFromArray(parameters.choices.map(function (choice) { |
|
return choice && _typeof(choice) === 'object' ? choice : { |
|
value: choice |
|
}; |
|
}), 'value'); |
|
return _this; |
|
} |
|
|
|
_createClass(ChoiceSchema, [{ |
|
key: "expected", |
|
value: function expected(_ref) { |
|
var _this2 = this; |
|
|
|
var descriptor = _ref.descriptor; |
|
var choiceValues = Array.from(this._choices.keys()).map(function (value) { |
|
return _this2._choices.get(value); |
|
}).filter(function (choiceInfo) { |
|
return !choiceInfo.deprecated; |
|
}).map(function (choiceInfo) { |
|
return choiceInfo.value; |
|
}).sort(utils.comparePrimitive).map(descriptor.value); |
|
var head = choiceValues.slice(0, -2); |
|
var tail = choiceValues.slice(-2); |
|
return head.concat(tail.join(' or ')).join(', '); |
|
} |
|
}, { |
|
key: "validate", |
|
value: function validate(value) { |
|
return this._choices.has(value); |
|
} |
|
}, { |
|
key: "deprecated", |
|
value: function deprecated(value) { |
|
var choiceInfo = this._choices.get(value); |
|
|
|
return choiceInfo && choiceInfo.deprecated ? { |
|
value: value |
|
} : false; |
|
} |
|
}, { |
|
key: "forward", |
|
value: function forward(value) { |
|
var choiceInfo = this._choices.get(value); |
|
|
|
return choiceInfo ? choiceInfo.forward : undefined; |
|
} |
|
}, { |
|
key: "redirect", |
|
value: function redirect(value) { |
|
var choiceInfo = this._choices.get(value); |
|
|
|
return choiceInfo ? choiceInfo.redirect : undefined; |
|
} |
|
}]); |
|
|
|
return ChoiceSchema; |
|
}(schema.Schema); |
|
|
|
exports.ChoiceSchema = ChoiceSchema; |
|
}); |
|
unwrapExports(choice); |
|
var choice_1 = choice.ChoiceSchema; |
|
|
|
var number = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
var NumberSchema = |
|
/*#__PURE__*/ |
|
function (_schema_1$Schema) { |
|
_inherits(NumberSchema, _schema_1$Schema); |
|
|
|
function NumberSchema() { |
|
_classCallCheck(this, NumberSchema); |
|
|
|
return _possibleConstructorReturn(this, _getPrototypeOf(NumberSchema).apply(this, arguments)); |
|
} |
|
|
|
_createClass(NumberSchema, [{ |
|
key: "expected", |
|
value: function expected() { |
|
return 'a number'; |
|
} |
|
}, { |
|
key: "validate", |
|
value: function validate(value, _utils) { |
|
return typeof value === 'number'; |
|
} |
|
}]); |
|
|
|
return NumberSchema; |
|
}(schema.Schema); |
|
|
|
exports.NumberSchema = NumberSchema; |
|
}); |
|
unwrapExports(number); |
|
var number_1 = number.NumberSchema; |
|
|
|
var integer = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
var IntegerSchema = |
|
/*#__PURE__*/ |
|
function (_number_1$NumberSchem) { |
|
_inherits(IntegerSchema, _number_1$NumberSchem); |
|
|
|
function IntegerSchema() { |
|
_classCallCheck(this, IntegerSchema); |
|
|
|
return _possibleConstructorReturn(this, _getPrototypeOf(IntegerSchema).apply(this, arguments)); |
|
} |
|
|
|
_createClass(IntegerSchema, [{ |
|
key: "expected", |
|
value: function expected() { |
|
return 'an integer'; |
|
} |
|
}, { |
|
key: "validate", |
|
value: function validate(value, utils$1) { |
|
return utils$1.normalizeValidateResult(_get(_getPrototypeOf(IntegerSchema.prototype), "validate", this).call(this, value, utils$1), value) === true && utils.isInt(value); |
|
} |
|
}]); |
|
|
|
return IntegerSchema; |
|
}(number.NumberSchema); |
|
|
|
exports.IntegerSchema = IntegerSchema; |
|
}); |
|
unwrapExports(integer); |
|
var integer_1 = integer.IntegerSchema; |
|
|
|
var string = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
var StringSchema = |
|
/*#__PURE__*/ |
|
function (_schema_1$Schema) { |
|
_inherits(StringSchema, _schema_1$Schema); |
|
|
|
function StringSchema() { |
|
_classCallCheck(this, StringSchema); |
|
|
|
return _possibleConstructorReturn(this, _getPrototypeOf(StringSchema).apply(this, arguments)); |
|
} |
|
|
|
_createClass(StringSchema, [{ |
|
key: "expected", |
|
value: function expected() { |
|
return 'a string'; |
|
} |
|
}, { |
|
key: "validate", |
|
value: function validate(value) { |
|
return typeof value === 'string'; |
|
} |
|
}]); |
|
|
|
return StringSchema; |
|
}(schema.Schema); |
|
|
|
exports.StringSchema = StringSchema; |
|
}); |
|
unwrapExports(string); |
|
var string_1 = string.StringSchema; |
|
|
|
var schemas = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
tslib_1.__exportStar(alias, exports); |
|
|
|
tslib_1.__exportStar(any, exports); |
|
|
|
tslib_1.__exportStar(array, exports); |
|
|
|
tslib_1.__exportStar(boolean_1, exports); |
|
|
|
tslib_1.__exportStar(choice, exports); |
|
|
|
tslib_1.__exportStar(integer, exports); |
|
|
|
tslib_1.__exportStar(number, exports); |
|
|
|
tslib_1.__exportStar(string, exports); |
|
}); |
|
unwrapExports(schemas); |
|
|
|
var defaults = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
exports.defaultDescriptor = api.apiDescriptor; |
|
exports.defaultUnknownHandler = leven_1.levenUnknownHandler; |
|
exports.defaultInvalidHandler = invalid.commonInvalidHandler; |
|
exports.defaultDeprecatedHandler = common.commonDeprecatedHandler; |
|
}); |
|
unwrapExports(defaults); |
|
var defaults_1 = defaults.defaultDescriptor; |
|
var defaults_2 = defaults.defaultUnknownHandler; |
|
var defaults_3 = defaults.defaultInvalidHandler; |
|
var defaults_4 = defaults.defaultDeprecatedHandler; |
|
|
|
var normalize = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
exports.normalize = function (options, schemas, opts) { |
|
return new Normalizer(schemas, opts).normalize(options); |
|
}; |
|
|
|
var Normalizer = |
|
/*#__PURE__*/ |
|
function () { |
|
function Normalizer(schemas, opts) { |
|
_classCallCheck(this, Normalizer); |
|
|
|
// istanbul ignore next |
|
var _ref = opts || {}, |
|
_ref$logger = _ref.logger, |
|
logger = _ref$logger === void 0 ? console : _ref$logger, |
|
_ref$descriptor = _ref.descriptor, |
|
descriptor = _ref$descriptor === void 0 ? defaults.defaultDescriptor : _ref$descriptor, |
|
_ref$unknown = _ref.unknown, |
|
unknown = _ref$unknown === void 0 ? defaults.defaultUnknownHandler : _ref$unknown, |
|
_ref$invalid = _ref.invalid, |
|
invalid = _ref$invalid === void 0 ? defaults.defaultInvalidHandler : _ref$invalid, |
|
_ref$deprecated = _ref.deprecated, |
|
deprecated = _ref$deprecated === void 0 ? defaults.defaultDeprecatedHandler : _ref$deprecated; |
|
|
|
this._utils = { |
|
descriptor: descriptor, |
|
logger: |
|
/* istanbul ignore next */ |
|
logger || { |
|
warn: function warn() {} |
|
}, |
|
schemas: utils.recordFromArray(schemas, 'name'), |
|
normalizeDefaultResult: utils.normalizeDefaultResult, |
|
normalizeDeprecatedResult: utils.normalizeDeprecatedResult, |
|
normalizeForwardResult: utils.normalizeForwardResult, |
|
normalizeRedirectResult: utils.normalizeRedirectResult, |
|
normalizeValidateResult: utils.normalizeValidateResult |
|
}; |
|
this._unknownHandler = unknown; |
|
this._invalidHandler = invalid; |
|
this._deprecatedHandler = deprecated; |
|
this.cleanHistory(); |
|
} |
|
|
|
_createClass(Normalizer, [{ |
|
key: "cleanHistory", |
|
value: function cleanHistory() { |
|
this._hasDeprecationWarned = utils.createAutoChecklist(); |
|
} |
|
}, { |
|
key: "normalize", |
|
value: function normalize(options) { |
|
var _this = this; |
|
|
|
var normalized = {}; |
|
var restOptionsArray = [options]; |
|
|
|
var applyNormalization = function applyNormalization() { |
|
while (restOptionsArray.length !== 0) { |
|
var currentOptions = restOptionsArray.shift(); |
|
|
|
var transferredOptionsArray = _this._applyNormalization(currentOptions, normalized); |
|
|
|
restOptionsArray.push.apply(restOptionsArray, _toConsumableArray(transferredOptionsArray)); |
|
} |
|
}; |
|
|
|
applyNormalization(); |
|
|
|
for (var _i = 0, _Object$keys = Object.keys(this._utils.schemas); _i < _Object$keys.length; _i++) { |
|
var key = _Object$keys[_i]; |
|
var schema = this._utils.schemas[key]; |
|
|
|
if (!(key in normalized)) { |
|
var defaultResult = utils.normalizeDefaultResult(schema.default(this._utils)); |
|
|
|
if ('value' in defaultResult) { |
|
restOptionsArray.push(_defineProperty({}, key, defaultResult.value)); |
|
} |
|
} |
|
} |
|
|
|
applyNormalization(); |
|
|
|
for (var _i2 = 0, _Object$keys2 = Object.keys(this._utils.schemas); _i2 < _Object$keys2.length; _i2++) { |
|
var _key = _Object$keys2[_i2]; |
|
var _schema = this._utils.schemas[_key]; |
|
|
|
if (_key in normalized) { |
|
normalized[_key] = _schema.postprocess(normalized[_key], this._utils); |
|
} |
|
} |
|
|
|
return normalized; |
|
} |
|
}, { |
|
key: "_applyNormalization", |
|
value: function _applyNormalization(options, normalized) { |
|
var _this2 = this; |
|
|
|
var transferredOptionsArray = []; |
|
|
|
var _utils_1$partition = utils.partition(Object.keys(options), function (key) { |
|
return key in _this2._utils.schemas; |
|
}), |
|
_utils_1$partition2 = _slicedToArray(_utils_1$partition, 2), |
|
knownOptionNames = _utils_1$partition2[0], |
|
unknownOptionNames = _utils_1$partition2[1]; |
|
|
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
var _loop = function _loop() { |
|
var key = _step.value; |
|
var schema = _this2._utils.schemas[key]; |
|
var value = schema.preprocess(options[key], _this2._utils); |
|
var validateResult = utils.normalizeValidateResult(schema.validate(value, _this2._utils), value); |
|
|
|
if (validateResult !== true) { |
|
var invalidValue = validateResult.value; |
|
|
|
var errorMessageOrError = _this2._invalidHandler(key, invalidValue, _this2._utils); |
|
|
|
throw typeof errorMessageOrError === 'string' ? new Error(errorMessageOrError) : |
|
/* istanbul ignore next*/ |
|
errorMessageOrError; |
|
} |
|
|
|
var appendTransferredOptions = function appendTransferredOptions(_ref2) { |
|
var from = _ref2.from, |
|
to = _ref2.to; |
|
transferredOptionsArray.push(typeof to === 'string' ? _defineProperty({}, to, from) : _defineProperty({}, to.key, to.value)); |
|
}; |
|
|
|
var warnDeprecated = function warnDeprecated(_ref5) { |
|
var currentValue = _ref5.value, |
|
redirectTo = _ref5.redirectTo; |
|
var deprecatedResult = utils.normalizeDeprecatedResult(schema.deprecated(currentValue, _this2._utils), value, |
|
/* doNotNormalizeTrue */ |
|
true); |
|
|
|
if (deprecatedResult === false) { |
|
return; |
|
} |
|
|
|
if (deprecatedResult === true) { |
|
if (!_this2._hasDeprecationWarned(key)) { |
|
_this2._utils.logger.warn(_this2._deprecatedHandler(key, redirectTo, _this2._utils)); |
|
} |
|
} else { |
|
var _iteratorNormalCompletion3 = true; |
|
var _didIteratorError3 = false; |
|
var _iteratorError3 = undefined; |
|
|
|
try { |
|
for (var _iterator3 = deprecatedResult[Symbol.iterator](), _step3; !(_iteratorNormalCompletion3 = (_step3 = _iterator3.next()).done); _iteratorNormalCompletion3 = true) { |
|
var deprecatedValue = _step3.value.value; |
|
var pair = { |
|
key: key, |
|
value: deprecatedValue |
|
}; |
|
|
|
if (!_this2._hasDeprecationWarned(pair)) { |
|
var redirectToPair = typeof redirectTo === 'string' ? { |
|
key: redirectTo, |
|
value: deprecatedValue |
|
} : redirectTo; |
|
|
|
_this2._utils.logger.warn(_this2._deprecatedHandler(pair, redirectToPair, _this2._utils)); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError3 = true; |
|
_iteratorError3 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion3 && _iterator3.return != null) { |
|
_iterator3.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError3) { |
|
throw _iteratorError3; |
|
} |
|
} |
|
} |
|
} |
|
}; |
|
|
|
var forwardResult = utils.normalizeForwardResult(schema.forward(value, _this2._utils), value); |
|
forwardResult.forEach(appendTransferredOptions); |
|
var redirectResult = utils.normalizeRedirectResult(schema.redirect(value, _this2._utils), value); |
|
redirectResult.redirect.forEach(appendTransferredOptions); |
|
|
|
if ('remain' in redirectResult) { |
|
var remainingValue = redirectResult.remain; |
|
normalized[key] = key in normalized ? schema.overlap(normalized[key], remainingValue, _this2._utils) : remainingValue; |
|
warnDeprecated({ |
|
value: remainingValue |
|
}); |
|
} |
|
|
|
var _iteratorNormalCompletion4 = true; |
|
var _didIteratorError4 = false; |
|
var _iteratorError4 = undefined; |
|
|
|
try { |
|
for (var _iterator4 = redirectResult.redirect[Symbol.iterator](), _step4; !(_iteratorNormalCompletion4 = (_step4 = _iterator4.next()).done); _iteratorNormalCompletion4 = true) { |
|
var _step4$value = _step4.value, |
|
from = _step4$value.from, |
|
to = _step4$value.to; |
|
warnDeprecated({ |
|
value: from, |
|
redirectTo: to |
|
}); |
|
} |
|
} catch (err) { |
|
_didIteratorError4 = true; |
|
_iteratorError4 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion4 && _iterator4.return != null) { |
|
_iterator4.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError4) { |
|
throw _iteratorError4; |
|
} |
|
} |
|
} |
|
}; |
|
|
|
for (var _iterator = knownOptionNames[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
_loop(); |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = unknownOptionNames[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var key = _step2.value; |
|
var value = options[key]; |
|
|
|
var unknownResult = this._unknownHandler(key, value, this._utils); |
|
|
|
if (unknownResult) { |
|
for (var _i3 = 0, _Object$keys3 = Object.keys(unknownResult); _i3 < _Object$keys3.length; _i3++) { |
|
var unknownKey = _Object$keys3[_i3]; |
|
|
|
var unknownOption = _defineProperty({}, unknownKey, unknownResult[unknownKey]); |
|
|
|
if (unknownKey in this._utils.schemas) { |
|
transferredOptionsArray.push(unknownOption); |
|
} else { |
|
Object.assign(normalized, unknownOption); |
|
} |
|
} |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
|
|
return transferredOptionsArray; |
|
} |
|
}]); |
|
|
|
return Normalizer; |
|
}(); |
|
|
|
exports.Normalizer = Normalizer; |
|
}); |
|
unwrapExports(normalize); |
|
var normalize_1 = normalize.normalize; |
|
var normalize_2 = normalize.Normalizer; |
|
|
|
var lib = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
|
|
tslib_1.__exportStar(descriptors, exports); |
|
|
|
tslib_1.__exportStar(handlers, exports); |
|
|
|
tslib_1.__exportStar(schemas, exports); |
|
|
|
tslib_1.__exportStar(normalize, exports); |
|
|
|
tslib_1.__exportStar(schema, exports); |
|
}); |
|
unwrapExports(lib); |
|
|
|
var array$1 = []; |
|
var charCodeCache$1 = []; |
|
|
|
var leven$1 = function leven(left, right) { |
|
if (left === right) { |
|
return 0; |
|
} |
|
|
|
var swap = left; // Swapping the strings if `a` is longer than `b` so we know which one is the |
|
// shortest & which one is the longest |
|
|
|
if (left.length > right.length) { |
|
left = right; |
|
right = swap; |
|
} |
|
|
|
var leftLength = left.length; |
|
var rightLength = right.length; // Performing suffix trimming: |
|
// We can linearly drop suffix common to both strings since they |
|
// don't increase distance at all |
|
// Note: `~-` is the bitwise way to perform a `- 1` operation |
|
|
|
while (leftLength > 0 && left.charCodeAt(~-leftLength) === right.charCodeAt(~-rightLength)) { |
|
leftLength--; |
|
rightLength--; |
|
} // Performing prefix trimming |
|
// We can linearly drop prefix common to both strings since they |
|
// don't increase distance at all |
|
|
|
|
|
var start = 0; |
|
|
|
while (start < leftLength && left.charCodeAt(start) === right.charCodeAt(start)) { |
|
start++; |
|
} |
|
|
|
leftLength -= start; |
|
rightLength -= start; |
|
|
|
if (leftLength === 0) { |
|
return rightLength; |
|
} |
|
|
|
var bCharCode; |
|
var result; |
|
var temp; |
|
var temp2; |
|
var i = 0; |
|
var j = 0; |
|
|
|
while (i < leftLength) { |
|
charCodeCache$1[i] = left.charCodeAt(start + i); |
|
array$1[i] = ++i; |
|
} |
|
|
|
while (j < rightLength) { |
|
bCharCode = right.charCodeAt(start + j); |
|
temp = j++; |
|
result = j; |
|
|
|
for (i = 0; i < leftLength; i++) { |
|
temp2 = bCharCode === charCodeCache$1[i] ? temp : temp + 1; |
|
temp = array$1[i]; // eslint-disable-next-line no-multi-assign |
|
|
|
result = array$1[i] = temp > result ? temp2 > result ? result + 1 : temp2 : temp2 > temp ? temp + 1 : temp2; |
|
} |
|
} |
|
|
|
return result; |
|
}; |
|
|
|
var leven_1$1 = leven$1; // TODO: Remove this for the next major release |
|
|
|
var default_1 = leven$1; |
|
leven_1$1.default = default_1; |
|
|
|
var cliDescriptor = { |
|
key: function key(_key) { |
|
return _key.length === 1 ? "-".concat(_key) : "--".concat(_key); |
|
}, |
|
value: function value(_value) { |
|
return lib.apiDescriptor.value(_value); |
|
}, |
|
pair: function pair(_ref) { |
|
var key = _ref.key, |
|
value = _ref.value; |
|
return value === false ? "--no-".concat(key) : value === true ? cliDescriptor.key(key) : value === "" ? "".concat(cliDescriptor.key(key), " without an argument") : "".concat(cliDescriptor.key(key), "=").concat(value); |
|
} |
|
}; |
|
|
|
var FlagSchema = |
|
/*#__PURE__*/ |
|
function (_vnopts$ChoiceSchema) { |
|
_inherits(FlagSchema, _vnopts$ChoiceSchema); |
|
|
|
function FlagSchema(_ref2) { |
|
var _this; |
|
|
|
var name = _ref2.name, |
|
flags = _ref2.flags; |
|
|
|
_classCallCheck(this, FlagSchema); |
|
|
|
_this = _possibleConstructorReturn(this, _getPrototypeOf(FlagSchema).call(this, { |
|
name: name, |
|
choices: flags |
|
})); |
|
_this._flags = flags.slice().sort(); |
|
return _this; |
|
} |
|
|
|
_createClass(FlagSchema, [{ |
|
key: "preprocess", |
|
value: function preprocess(value, utils) { |
|
if (typeof value === "string" && value.length !== 0 && this._flags.indexOf(value) === -1) { |
|
var suggestion = this._flags.find(function (flag) { |
|
return leven_1$1(flag, value) < 3; |
|
}); |
|
|
|
if (suggestion) { |
|
utils.logger.warn(["Unknown flag ".concat(chalk.yellow(utils.descriptor.value(value)), ","), "did you mean ".concat(chalk.blue(utils.descriptor.value(suggestion)), "?")].join(" ")); |
|
return suggestion; |
|
} |
|
} |
|
|
|
return value; |
|
} |
|
}, { |
|
key: "expected", |
|
value: function expected() { |
|
return "a flag"; |
|
} |
|
}]); |
|
|
|
return FlagSchema; |
|
}(lib.ChoiceSchema); |
|
|
|
var hasDeprecationWarned; |
|
|
|
function normalizeOptions(options, optionInfos) { |
|
var _ref3 = arguments.length > 2 && arguments[2] !== undefined ? arguments[2] : {}, |
|
logger = _ref3.logger, |
|
_ref3$isCLI = _ref3.isCLI, |
|
isCLI = _ref3$isCLI === void 0 ? false : _ref3$isCLI, |
|
_ref3$passThrough = _ref3.passThrough, |
|
passThrough = _ref3$passThrough === void 0 ? false : _ref3$passThrough; |
|
|
|
var unknown = !passThrough ? lib.levenUnknownHandler : Array.isArray(passThrough) ? function (key, value) { |
|
return passThrough.indexOf(key) === -1 ? undefined : _defineProperty({}, key, value); |
|
} : function (key, value) { |
|
return _defineProperty({}, key, value); |
|
}; |
|
var descriptor = isCLI ? cliDescriptor : lib.apiDescriptor; |
|
var schemas = optionInfosToSchemas(optionInfos, { |
|
isCLI: isCLI |
|
}); |
|
var normalizer = new lib.Normalizer(schemas, { |
|
logger: logger, |
|
unknown: unknown, |
|
descriptor: descriptor |
|
}); |
|
var shouldSuppressDuplicateDeprecationWarnings = logger !== false; |
|
|
|
if (shouldSuppressDuplicateDeprecationWarnings && hasDeprecationWarned) { |
|
normalizer._hasDeprecationWarned = hasDeprecationWarned; |
|
} |
|
|
|
var normalized = normalizer.normalize(options); |
|
|
|
if (shouldSuppressDuplicateDeprecationWarnings) { |
|
hasDeprecationWarned = normalizer._hasDeprecationWarned; |
|
} |
|
|
|
return normalized; |
|
} |
|
|
|
function optionInfosToSchemas(optionInfos, _ref6) { |
|
var isCLI = _ref6.isCLI; |
|
var schemas = []; |
|
|
|
if (isCLI) { |
|
schemas.push(lib.AnySchema.create({ |
|
name: "_" |
|
})); |
|
} |
|
|
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = optionInfos[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var optionInfo = _step.value; |
|
schemas.push(optionInfoToSchema(optionInfo, { |
|
isCLI: isCLI, |
|
optionInfos: optionInfos |
|
})); |
|
|
|
if (optionInfo.alias && isCLI) { |
|
schemas.push(lib.AliasSchema.create({ |
|
name: optionInfo.alias, |
|
sourceName: optionInfo.name |
|
})); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return schemas; |
|
} |
|
|
|
function optionInfoToSchema(optionInfo, _ref7) { |
|
var isCLI = _ref7.isCLI, |
|
optionInfos = _ref7.optionInfos; |
|
var SchemaConstructor; |
|
var parameters = { |
|
name: optionInfo.name |
|
}; |
|
var handlers = {}; |
|
|
|
switch (optionInfo.type) { |
|
case "int": |
|
SchemaConstructor = lib.IntegerSchema; |
|
|
|
if (isCLI) { |
|
parameters.preprocess = function (value) { |
|
return Number(value); |
|
}; |
|
} |
|
|
|
break; |
|
|
|
case "string": |
|
SchemaConstructor = lib.StringSchema; |
|
break; |
|
|
|
case "choice": |
|
SchemaConstructor = lib.ChoiceSchema; |
|
parameters.choices = optionInfo.choices.map(function (choiceInfo) { |
|
return _typeof(choiceInfo) === "object" && choiceInfo.redirect ? Object.assign({}, choiceInfo, { |
|
redirect: { |
|
to: { |
|
key: optionInfo.name, |
|
value: choiceInfo.redirect |
|
} |
|
} |
|
}) : choiceInfo; |
|
}); |
|
break; |
|
|
|
case "boolean": |
|
SchemaConstructor = lib.BooleanSchema; |
|
break; |
|
|
|
case "flag": |
|
SchemaConstructor = FlagSchema; |
|
parameters.flags = optionInfos.map(function (optionInfo) { |
|
return [].concat(optionInfo.alias || [], optionInfo.description ? optionInfo.name : [], optionInfo.oppositeDescription ? "no-".concat(optionInfo.name) : []); |
|
}).reduce(function (a, b) { |
|
return a.concat(b); |
|
}, []); |
|
break; |
|
|
|
case "path": |
|
SchemaConstructor = lib.StringSchema; |
|
break; |
|
|
|
default: |
|
throw new Error("Unexpected type ".concat(optionInfo.type)); |
|
} |
|
|
|
if (optionInfo.exception) { |
|
parameters.validate = function (value, schema, utils) { |
|
return optionInfo.exception(value) || schema.validate(value, utils); |
|
}; |
|
} else { |
|
parameters.validate = function (value, schema, utils) { |
|
return value === undefined || schema.validate(value, utils); |
|
}; |
|
} |
|
|
|
if (optionInfo.redirect) { |
|
handlers.redirect = function (value) { |
|
return !value ? undefined : { |
|
to: { |
|
key: optionInfo.redirect.option, |
|
value: optionInfo.redirect.value |
|
} |
|
}; |
|
}; |
|
} |
|
|
|
if (optionInfo.deprecated) { |
|
handlers.deprecated = true; |
|
} // allow CLI overriding, e.g., prettier package.json --tab-width 1 --tab-width 2 |
|
|
|
|
|
if (isCLI && !optionInfo.array) { |
|
var originalPreprocess = parameters.preprocess || function (x) { |
|
return x; |
|
}; |
|
|
|
parameters.preprocess = function (value, schema, utils) { |
|
return schema.preprocess(originalPreprocess(Array.isArray(value) ? value[value.length - 1] : value), utils); |
|
}; |
|
} |
|
|
|
return optionInfo.array ? lib.ArraySchema.create(Object.assign(isCLI ? { |
|
preprocess: function preprocess(v) { |
|
return [].concat(v); |
|
} |
|
} : {}, handlers, { |
|
valueSchema: SchemaConstructor.create(parameters) |
|
})) : SchemaConstructor.create(Object.assign({}, parameters, handlers)); |
|
} |
|
|
|
function normalizeApiOptions(options, optionInfos, opts) { |
|
return normalizeOptions(options, optionInfos, opts); |
|
} |
|
|
|
function normalizeCliOptions(options, optionInfos, opts) { |
|
return normalizeOptions(options, optionInfos, Object.assign({ |
|
isCLI: true |
|
}, opts)); |
|
} |
|
|
|
var optionsNormalizer = { |
|
normalizeApiOptions: normalizeApiOptions, |
|
normalizeCliOptions: normalizeCliOptions |
|
}; |
|
|
|
var getLast = function getLast(arr) { |
|
return arr.length > 0 ? arr[arr.length - 1] : null; |
|
}; |
|
|
|
function locStart(node, opts) { |
|
opts = opts || {}; // Handle nodes with decorators. They should start at the first decorator |
|
|
|
if (!opts.ignoreDecorators && node.declaration && node.declaration.decorators && node.declaration.decorators.length > 0) { |
|
return locStart(node.declaration.decorators[0]); |
|
} |
|
|
|
if (!opts.ignoreDecorators && node.decorators && node.decorators.length > 0) { |
|
return locStart(node.decorators[0]); |
|
} |
|
|
|
if (node.__location) { |
|
return node.__location.startOffset; |
|
} |
|
|
|
if (node.range) { |
|
return node.range[0]; |
|
} |
|
|
|
if (typeof node.start === "number") { |
|
return node.start; |
|
} |
|
|
|
if (node.loc) { |
|
return node.loc.start; |
|
} |
|
|
|
return null; |
|
} |
|
|
|
function locEnd(node) { |
|
var endNode = node.nodes && getLast(node.nodes); |
|
|
|
if (endNode && node.source && !node.source.end) { |
|
node = endNode; |
|
} |
|
|
|
if (node.__location) { |
|
return node.__location.endOffset; |
|
} |
|
|
|
var loc = node.range ? node.range[1] : typeof node.end === "number" ? node.end : null; |
|
|
|
if (node.typeAnnotation) { |
|
return Math.max(loc, locEnd(node.typeAnnotation)); |
|
} |
|
|
|
if (node.loc && !loc) { |
|
return node.loc.end; |
|
} |
|
|
|
return loc; |
|
} |
|
|
|
var loc = { |
|
locStart: locStart, |
|
locEnd: locEnd |
|
}; |
|
|
|
var jsTokens = createCommonjsModule(function (module, exports) { |
|
// Copyright 2014, 2015, 2016, 2017, 2018 Simon Lydell |
|
// License: MIT. (See LICENSE.) |
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); // This regex comes from regex.coffee, and is inserted here by generate-index.js |
|
// (run `npm run build`). |
|
|
|
exports.default = /((['"])(?:(?!\2|\\).|\\(?:\r\n|[\s\S]))*(\2)?|`(?:[^`\\$]|\\[\s\S]|\$(?!\{)|\$\{(?:[^{}]|\{[^}]*\}?)*\}?)*(`)?)|(\/\/.*)|(\/\*(?:[^*]|\*(?!\/))*(\*\/)?)|(\/(?!\*)(?:\[(?:(?![\]\\]).|\\.)*\]|(?![\/\]\\]).|\\.)+\/(?:(?!\s*(?:\b|[\u0080-\uFFFF$\\'"~({]|[+\-!](?!=)|\.?\d))|[gmiyus]{1,6}\b(?![\u0080-\uFFFF$\\]|\s*(?:[+\-*%&|^<>!=?({]|\/(?![\/*])))))|(0[xX][\da-fA-F]+|0[oO][0-7]+|0[bB][01]+|(?:\d*\.\d+|\d+\.?)(?:[eE][+-]?\d+)?)|((?!\d)(?:(?!\s)[$\w\u0080-\uFFFF]|\\u[\da-fA-F]{4}|\\u\{[\da-fA-F]+\})+)|(--|\+\+|&&|\|\||=>|\.{3}|(?:[+\-\/%&|^]|\*{1,2}|<{1,2}|>{1,3}|!=?|={1,2})=?|[?~.,:;[\](){}])|(\s+)|(^$|[\s\S])/g; |
|
|
|
exports.matchToToken = function (match) { |
|
var token = { |
|
type: "invalid", |
|
value: match[0], |
|
closed: undefined |
|
}; |
|
if (match[1]) token.type = "string", token.closed = !!(match[3] || match[4]);else if (match[5]) token.type = "comment";else if (match[6]) token.type = "comment", token.closed = !!match[7];else if (match[8]) token.type = "regex";else if (match[9]) token.type = "number";else if (match[10]) token.type = "name";else if (match[11]) token.type = "punctuator";else if (match[12]) token.type = "whitespace"; |
|
return token; |
|
}; |
|
}); |
|
unwrapExports(jsTokens); |
|
var jsTokens_1 = jsTokens.matchToToken; |
|
|
|
var ast = createCommonjsModule(function (module) { |
|
/* |
|
Copyright (C) 2013 Yusuke Suzuki <utatane.tea@gmail.com> |
|
|
|
Redistribution and use in source and binary forms, with or without |
|
modification, are permitted provided that the following conditions are met: |
|
|
|
* Redistributions of source code must retain the above copyright |
|
notice, this list of conditions and the following disclaimer. |
|
* Redistributions in binary form must reproduce the above copyright |
|
notice, this list of conditions and the following disclaimer in the |
|
documentation and/or other materials provided with the distribution. |
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 'AS IS' |
|
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
|
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
|
ARE DISCLAIMED. IN NO EVENT SHALL <COPYRIGHT HOLDER> BE LIABLE FOR ANY |
|
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES |
|
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
|
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND |
|
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
|
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
|
THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
|
*/ |
|
(function () { |
|
|
|
function isExpression(node) { |
|
if (node == null) { |
|
return false; |
|
} |
|
|
|
switch (node.type) { |
|
case 'ArrayExpression': |
|
case 'AssignmentExpression': |
|
case 'BinaryExpression': |
|
case 'CallExpression': |
|
case 'ConditionalExpression': |
|
case 'FunctionExpression': |
|
case 'Identifier': |
|
case 'Literal': |
|
case 'LogicalExpression': |
|
case 'MemberExpression': |
|
case 'NewExpression': |
|
case 'ObjectExpression': |
|
case 'SequenceExpression': |
|
case 'ThisExpression': |
|
case 'UnaryExpression': |
|
case 'UpdateExpression': |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isIterationStatement(node) { |
|
if (node == null) { |
|
return false; |
|
} |
|
|
|
switch (node.type) { |
|
case 'DoWhileStatement': |
|
case 'ForInStatement': |
|
case 'ForStatement': |
|
case 'WhileStatement': |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isStatement(node) { |
|
if (node == null) { |
|
return false; |
|
} |
|
|
|
switch (node.type) { |
|
case 'BlockStatement': |
|
case 'BreakStatement': |
|
case 'ContinueStatement': |
|
case 'DebuggerStatement': |
|
case 'DoWhileStatement': |
|
case 'EmptyStatement': |
|
case 'ExpressionStatement': |
|
case 'ForInStatement': |
|
case 'ForStatement': |
|
case 'IfStatement': |
|
case 'LabeledStatement': |
|
case 'ReturnStatement': |
|
case 'SwitchStatement': |
|
case 'ThrowStatement': |
|
case 'TryStatement': |
|
case 'VariableDeclaration': |
|
case 'WhileStatement': |
|
case 'WithStatement': |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isSourceElement(node) { |
|
return isStatement(node) || node != null && node.type === 'FunctionDeclaration'; |
|
} |
|
|
|
function trailingStatement(node) { |
|
switch (node.type) { |
|
case 'IfStatement': |
|
if (node.alternate != null) { |
|
return node.alternate; |
|
} |
|
|
|
return node.consequent; |
|
|
|
case 'LabeledStatement': |
|
case 'ForStatement': |
|
case 'ForInStatement': |
|
case 'WhileStatement': |
|
case 'WithStatement': |
|
return node.body; |
|
} |
|
|
|
return null; |
|
} |
|
|
|
function isProblematicIfStatement(node) { |
|
var current; |
|
|
|
if (node.type !== 'IfStatement') { |
|
return false; |
|
} |
|
|
|
if (node.alternate == null) { |
|
return false; |
|
} |
|
|
|
current = node.consequent; |
|
|
|
do { |
|
if (current.type === 'IfStatement') { |
|
if (current.alternate == null) { |
|
return true; |
|
} |
|
} |
|
|
|
current = trailingStatement(current); |
|
} while (current); |
|
|
|
return false; |
|
} |
|
|
|
module.exports = { |
|
isExpression: isExpression, |
|
isStatement: isStatement, |
|
isIterationStatement: isIterationStatement, |
|
isSourceElement: isSourceElement, |
|
isProblematicIfStatement: isProblematicIfStatement, |
|
trailingStatement: trailingStatement |
|
}; |
|
})(); |
|
/* vim: set sw=4 ts=4 et tw=80 : */ |
|
|
|
}); |
|
var ast_1 = ast.isExpression; |
|
var ast_2 = ast.isStatement; |
|
var ast_3 = ast.isIterationStatement; |
|
var ast_4 = ast.isSourceElement; |
|
var ast_5 = ast.isProblematicIfStatement; |
|
var ast_6 = ast.trailingStatement; |
|
|
|
var code = createCommonjsModule(function (module) { |
|
/* |
|
Copyright (C) 2013-2014 Yusuke Suzuki <utatane.tea@gmail.com> |
|
Copyright (C) 2014 Ivan Nikulin <ifaaan@gmail.com> |
|
|
|
Redistribution and use in source and binary forms, with or without |
|
modification, are permitted provided that the following conditions are met: |
|
|
|
* Redistributions of source code must retain the above copyright |
|
notice, this list of conditions and the following disclaimer. |
|
* Redistributions in binary form must reproduce the above copyright |
|
notice, this list of conditions and the following disclaimer in the |
|
documentation and/or other materials provided with the distribution. |
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
|
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
|
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
|
ARE DISCLAIMED. IN NO EVENT SHALL <COPYRIGHT HOLDER> BE LIABLE FOR ANY |
|
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES |
|
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
|
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND |
|
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
|
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
|
THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
|
*/ |
|
(function () { |
|
|
|
var ES6Regex, ES5Regex, NON_ASCII_WHITESPACES, IDENTIFIER_START, IDENTIFIER_PART, ch; // See `tools/generate-identifier-regex.js`. |
|
|
|
ES5Regex = { |
|
// ECMAScript 5.1/Unicode v9.0.0 NonAsciiIdentifierStart: |
|
NonAsciiIdentifierStart: /[\xAA\xB5\xBA\xC0-\xD6\xD8-\xF6\xF8-\u02C1\u02C6-\u02D1\u02E0-\u02E4\u02EC\u02EE\u0370-\u0374\u0376\u0377\u037A-\u037D\u037F\u0386\u0388-\u038A\u038C\u038E-\u03A1\u03A3-\u03F5\u03F7-\u0481\u048A-\u052F\u0531-\u0556\u0559\u0561-\u0587\u05D0-\u05EA\u05F0-\u05F2\u0620-\u064A\u066E\u066F\u0671-\u06D3\u06D5\u06E5\u06E6\u06EE\u06EF\u06FA-\u06FC\u06FF\u0710\u0712-\u072F\u074D-\u07A5\u07B1\u07CA-\u07EA\u07F4\u07F5\u07FA\u0800-\u0815\u081A\u0824\u0828\u0840-\u0858\u08A0-\u08B4\u08B6-\u08BD\u0904-\u0939\u093D\u0950\u0958-\u0961\u0971-\u0980\u0985-\u098C\u098F\u0990\u0993-\u09A8\u09AA-\u09B0\u09B2\u09B6-\u09B9\u09BD\u09CE\u09DC\u09DD\u09DF-\u09E1\u09F0\u09F1\u0A05-\u0A0A\u0A0F\u0A10\u0A13-\u0A28\u0A2A-\u0A30\u0A32\u0A33\u0A35\u0A36\u0A38\u0A39\u0A59-\u0A5C\u0A5E\u0A72-\u0A74\u0A85-\u0A8D\u0A8F-\u0A91\u0A93-\u0AA8\u0AAA-\u0AB0\u0AB2\u0AB3\u0AB5-\u0AB9\u0ABD\u0AD0\u0AE0\u0AE1\u0AF9\u0B05-\u0B0C\u0B0F\u0B10\u0B13-\u0B28\u0B2A-\u0B30\u0B32\u0B33\u0B35-\u0B39\u0B3D\u0B5C\u0B5D\u0B5F-\u0B61\u0B71\u0B83\u0B85-\u0B8A\u0B8E-\u0B90\u0B92-\u0B95\u0B99\u0B9A\u0B9C\u0B9E\u0B9F\u0BA3\u0BA4\u0BA8-\u0BAA\u0BAE-\u0BB9\u0BD0\u0C05-\u0C0C\u0C0E-\u0C10\u0C12-\u0C28\u0C2A-\u0C39\u0C3D\u0C58-\u0C5A\u0C60\u0C61\u0C80\u0C85-\u0C8C\u0C8E-\u0C90\u0C92-\u0CA8\u0CAA-\u0CB3\u0CB5-\u0CB9\u0CBD\u0CDE\u0CE0\u0CE1\u0CF1\u0CF2\u0D05-\u0D0C\u0D0E-\u0D10\u0D12-\u0D3A\u0D3D\u0D4E\u0D54-\u0D56\u0D5F-\u0D61\u0D7A-\u0D7F\u0D85-\u0D96\u0D9A-\u0DB1\u0DB3-\u0DBB\u0DBD\u0DC0-\u0DC6\u0E01-\u0E30\u0E32\u0E33\u0E40-\u0E46\u0E81\u0E82\u0E84\u0E87\u0E88\u0E8A\u0E8D\u0E94-\u0E97\u0E99-\u0E9F\u0EA1-\u0EA3\u0EA5\u0EA7\u0EAA\u0EAB\u0EAD-\u0EB0\u0EB2\u0EB3\u0EBD\u0EC0-\u0EC4\u0EC6\u0EDC-\u0EDF\u0F00\u0F40-\u0F47\u0F49-\u0F6C\u0F88-\u0F8C\u1000-\u102A\u103F\u1050-\u1055\u105A-\u105D\u1061\u1065\u1066\u106E-\u1070\u1075-\u1081\u108E\u10A0-\u10C5\u10C7\u10CD\u10D0-\u10FA\u10FC-\u1248\u124A-\u124D\u1250-\u1256\u1258\u125A-\u125D\u1260-\u1288\u128A-\u128D\u1290-\u12B0\u12B2-\u12B5\u12B8-\u12BE\u12C0\u12C2-\u12C5\u12C8-\u12D6\u12D8-\u1310\u1312-\u1315\u1318-\u135A\u1380-\u138F\u13A0-\u13F5\u13F8-\u13FD\u1401-\u166C\u166F-\u167F\u1681-\u169A\u16A0-\u16EA\u16EE-\u16F8\u1700-\u170C\u170E-\u1711\u1720-\u1731\u1740-\u1751\u1760-\u176C\u176E-\u1770\u1780-\u17B3\u17D7\u17DC\u1820-\u1877\u1880-\u1884\u1887-\u18A8\u18AA\u18B0-\u18F5\u1900-\u191E\u1950-\u196D\u1970-\u1974\u1980-\u19AB\u19B0-\u19C9\u1A00-\u1A16\u1A20-\u1A54\u1AA7\u1B05-\u1B33\u1B45-\u1B4B\u1B83-\u1BA0\u1BAE\u1BAF\u1BBA-\u1BE5\u1C00-\u1C23\u1C4D-\u1C4F\u1C5A-\u1C7D\u1C80-\u1C88\u1CE9-\u1CEC\u1CEE-\u1CF1\u1CF5\u1CF6\u1D00-\u1DBF\u1E00-\u1F15\u1F18-\u1F1D\u1F20-\u1F45\u1F48-\u1F4D\u1F50-\u1F57\u1F59\u1F5B\u1F5D\u1F5F-\u1F7D\u1F80-\u1FB4\u1FB6-\u1FBC\u1FBE\u1FC2-\u1FC4\u1FC6-\u1FCC\u1FD0-\u1FD3\u1FD6-\u1FDB\u1FE0-\u1FEC\u1FF2-\u1FF4\u1FF6-\u1FFC\u2071\u207F\u2090-\u209C\u2102\u2107\u210A-\u2113\u2115\u2119-\u211D\u2124\u2126\u2128\u212A-\u212D\u212F-\u2139\u213C-\u213F\u2145-\u2149\u214E\u2160-\u2188\u2C00-\u2C2E\u2C30-\u2C5E\u2C60-\u2CE4\u2CEB-\u2CEE\u2CF2\u2CF3\u2D00-\u2D25\u2D27\u2D2D\u2D30-\u2D67\u2D6F\u2D80-\u2D96\u2DA0-\u2DA6\u2DA8-\u2DAE\u2DB0-\u2DB6\u2DB8-\u2DBE\u2DC0-\u2DC6\u2DC8-\u2DCE\u2DD0-\u2DD6\u2DD8-\u2DDE\u2E2F\u3005-\u3007\u3021-\u3029\u3031-\u3035\u3038-\u303C\u3041-\u3096\u309D-\u309F\u30A1-\u30FA\u30FC-\u30FF\u3105-\u312D\u3131-\u318E\u31A0-\u31BA\u31F0-\u31FF\u3400-\u4DB5\u4E00-\u9FD5\uA000-\uA48C\uA4D0-\uA4FD\uA500-\uA60C\uA610-\uA61F\uA62A\uA62B\uA640-\uA66E\uA67F-\uA69D\uA6A0-\uA6EF\uA717-\uA71F\uA722-\uA788\uA78B-\uA7AE\uA7B0-\uA7B7\uA7F7-\uA801\uA803-\uA805\uA807-\uA80A\uA80C-\uA822\uA840-\uA873\uA882-\uA8B3\uA8F2-\uA8F7\uA8FB\uA8FD\uA90A-\uA925\uA930-\uA946\uA960-\uA97C\uA984-\uA9B2\uA9CF\uA9E0-\uA9E4\uA9E6-\uA9EF\uA9FA-\uA9FE\uAA00-\uAA28\uAA40-\uAA42\uAA44-\uAA4B\uAA60-\uAA76\uAA7A\uAA7E-\uAAAF\uAAB1\uAAB5\uAAB6\uAAB9-\uAABD\uAAC0\uAAC2\uAADB-\uAADD\uAAE0-\uAAEA\uAAF2-\uAAF4\uAB01-\uAB06\uAB09-\uAB0E\uAB11-\uAB16\uAB20-\uAB26\uAB28-\uAB2E\uAB30-\uAB5A\uAB5C-\uAB65\uAB70-\uABE2\uAC00-\uD7A3\uD7B0-\uD7C6\uD7CB-\uD7FB\uF900-\uFA6D\uFA70-\uFAD9\uFB00-\uFB06\uFB13-\uFB17\uFB1D\uFB1F-\uFB28\uFB2A-\uFB36\uFB38-\uFB3C\uFB3E\uFB40\uFB41\uFB43\uFB44\uFB46-\uFBB1\uFBD3-\uFD3D\uFD50-\uFD8F\uFD92-\uFDC7\uFDF0-\uFDFB\uFE70-\uFE74\uFE76-\uFEFC\uFF21-\uFF3A\uFF41-\uFF5A\uFF66-\uFFBE\uFFC2-\uFFC7\uFFCA-\uFFCF\uFFD2-\uFFD7\uFFDA-\uFFDC]/, |
|
// ECMAScript 5.1/Unicode v9.0.0 NonAsciiIdentifierPart: |
|
NonAsciiIdentifierPart: /[\xAA\xB5\xBA\xC0-\xD6\xD8-\xF6\xF8-\u02C1\u02C6-\u02D1\u02E0-\u02E4\u02EC\u02EE\u0300-\u0374\u0376\u0377\u037A-\u037D\u037F\u0386\u0388-\u038A\u038C\u038E-\u03A1\u03A3-\u03F5\u03F7-\u0481\u0483-\u0487\u048A-\u052F\u0531-\u0556\u0559\u0561-\u0587\u0591-\u05BD\u05BF\u05C1\u05C2\u05C4\u05C5\u05C7\u05D0-\u05EA\u05F0-\u05F2\u0610-\u061A\u0620-\u0669\u066E-\u06D3\u06D5-\u06DC\u06DF-\u06E8\u06EA-\u06FC\u06FF\u0710-\u074A\u074D-\u07B1\u07C0-\u07F5\u07FA\u0800-\u082D\u0840-\u085B\u08A0-\u08B4\u08B6-\u08BD\u08D4-\u08E1\u08E3-\u0963\u0966-\u096F\u0971-\u0983\u0985-\u098C\u098F\u0990\u0993-\u09A8\u09AA-\u09B0\u09B2\u09B6-\u09B9\u09BC-\u09C4\u09C7\u09C8\u09CB-\u09CE\u09D7\u09DC\u09DD\u09DF-\u09E3\u09E6-\u09F1\u0A01-\u0A03\u0A05-\u0A0A\u0A0F\u0A10\u0A13-\u0A28\u0A2A-\u0A30\u0A32\u0A33\u0A35\u0A36\u0A38\u0A39\u0A3C\u0A3E-\u0A42\u0A47\u0A48\u0A4B-\u0A4D\u0A51\u0A59-\u0A5C\u0A5E\u0A66-\u0A75\u0A81-\u0A83\u0A85-\u0A8D\u0A8F-\u0A91\u0A93-\u0AA8\u0AAA-\u0AB0\u0AB2\u0AB3\u0AB5-\u0AB9\u0ABC-\u0AC5\u0AC7-\u0AC9\u0ACB-\u0ACD\u0AD0\u0AE0-\u0AE3\u0AE6-\u0AEF\u0AF9\u0B01-\u0B03\u0B05-\u0B0C\u0B0F\u0B10\u0B13-\u0B28\u0B2A-\u0B30\u0B32\u0B33\u0B35-\u0B39\u0B3C-\u0B44\u0B47\u0B48\u0B4B-\u0B4D\u0B56\u0B57\u0B5C\u0B5D\u0B5F-\u0B63\u0B66-\u0B6F\u0B71\u0B82\u0B83\u0B85-\u0B8A\u0B8E-\u0B90\u0B92-\u0B95\u0B99\u0B9A\u0B9C\u0B9E\u0B9F\u0BA3\u0BA4\u0BA8-\u0BAA\u0BAE-\u0BB9\u0BBE-\u0BC2\u0BC6-\u0BC8\u0BCA-\u0BCD\u0BD0\u0BD7\u0BE6-\u0BEF\u0C00-\u0C03\u0C05-\u0C0C\u0C0E-\u0C10\u0C12-\u0C28\u0C2A-\u0C39\u0C3D-\u0C44\u0C46-\u0C48\u0C4A-\u0C4D\u0C55\u0C56\u0C58-\u0C5A\u0C60-\u0C63\u0C66-\u0C6F\u0C80-\u0C83\u0C85-\u0C8C\u0C8E-\u0C90\u0C92-\u0CA8\u0CAA-\u0CB3\u0CB5-\u0CB9\u0CBC-\u0CC4\u0CC6-\u0CC8\u0CCA-\u0CCD\u0CD5\u0CD6\u0CDE\u0CE0-\u0CE3\u0CE6-\u0CEF\u0CF1\u0CF2\u0D01-\u0D03\u0D05-\u0D0C\u0D0E-\u0D10\u0D12-\u0D3A\u0D3D-\u0D44\u0D46-\u0D48\u0D4A-\u0D4E\u0D54-\u0D57\u0D5F-\u0D63\u0D66-\u0D6F\u0D7A-\u0D7F\u0D82\u0D83\u0D85-\u0D96\u0D9A-\u0DB1\u0DB3-\u0DBB\u0DBD\u0DC0-\u0DC6\u0DCA\u0DCF-\u0DD4\u0DD6\u0DD8-\u0DDF\u0DE6-\u0DEF\u0DF2\u0DF3\u0E01-\u0E3A\u0E40-\u0E4E\u0E50-\u0E59\u0E81\u0E82\u0E84\u0E87\u0E88\u0E8A\u0E8D\u0E94-\u0E97\u0E99-\u0E9F\u0EA1-\u0EA3\u0EA5\u0EA7\u0EAA\u0EAB\u0EAD-\u0EB9\u0EBB-\u0EBD\u0EC0-\u0EC4\u0EC6\u0EC8-\u0ECD\u0ED0-\u0ED9\u0EDC-\u0EDF\u0F00\u0F18\u0F19\u0F20-\u0F29\u0F35\u0F37\u0F39\u0F3E-\u0F47\u0F49-\u0F6C\u0F71-\u0F84\u0F86-\u0F97\u0F99-\u0FBC\u0FC6\u1000-\u1049\u1050-\u109D\u10A0-\u10C5\u10C7\u10CD\u10D0-\u10FA\u10FC-\u1248\u124A-\u124D\u1250-\u1256\u1258\u125A-\u125D\u1260-\u1288\u128A-\u128D\u1290-\u12B0\u12B2-\u12B5\u12B8-\u12BE\u12C0\u12C2-\u12C5\u12C8-\u12D6\u12D8-\u1310\u1312-\u1315\u1318-\u135A\u135D-\u135F\u1380-\u138F\u13A0-\u13F5\u13F8-\u13FD\u1401-\u166C\u166F-\u167F\u1681-\u169A\u16A0-\u16EA\u16EE-\u16F8\u1700-\u170C\u170E-\u1714\u1720-\u1734\u1740-\u1753\u1760-\u176C\u176E-\u1770\u1772\u1773\u1780-\u17D3\u17D7\u17DC\u17DD\u17E0-\u17E9\u180B-\u180D\u1810-\u1819\u1820-\u1877\u1880-\u18AA\u18B0-\u18F5\u1900-\u191E\u1920-\u192B\u1930-\u193B\u1946-\u196D\u1970-\u1974\u1980-\u19AB\u19B0-\u19C9\u19D0-\u19D9\u1A00-\u1A1B\u1A20-\u1A5E\u1A60-\u1A7C\u1A7F-\u1A89\u1A90-\u1A99\u1AA7\u1AB0-\u1ABD\u1B00-\u1B4B\u1B50-\u1B59\u1B6B-\u1B73\u1B80-\u1BF3\u1C00-\u1C37\u1C40-\u1C49\u1C4D-\u1C7D\u1C80-\u1C88\u1CD0-\u1CD2\u1CD4-\u1CF6\u1CF8\u1CF9\u1D00-\u1DF5\u1DFB-\u1F15\u1F18-\u1F1D\u1F20-\u1F45\u1F48-\u1F4D\u1F50-\u1F57\u1F59\u1F5B\u1F5D\u1F5F-\u1F7D\u1F80-\u1FB4\u1FB6-\u1FBC\u1FBE\u1FC2-\u1FC4\u1FC6-\u1FCC\u1FD0-\u1FD3\u1FD6-\u1FDB\u1FE0-\u1FEC\u1FF2-\u1FF4\u1FF6-\u1FFC\u200C\u200D\u203F\u2040\u2054\u2071\u207F\u2090-\u209C\u20D0-\u20DC\u20E1\u20E5-\u20F0\u2102\u2107\u210A-\u2113\u2115\u2119-\u211D\u2124\u2126\u2128\u212A-\u212D\u212F-\u2139\u213C-\u213F\u2145-\u2149\u214E\u2160-\u2188\u2C00-\u2C2E\u2C30-\u2C5E\u2C60-\u2CE4\u2CEB-\u2CF3\u2D00-\u2D25\u2D27\u2D2D\u2D30-\u2D67\u2D6F\u2D7F-\u2D96\u2DA0-\u2DA6\u2DA8-\u2DAE\u2DB0-\u2DB6\u2DB8-\u2DBE\u2DC0-\u2DC6\u2DC8-\u2DCE\u2DD0-\u2DD6\u2DD8-\u2DDE\u2DE0-\u2DFF\u2E2F\u3005-\u3007\u3021-\u302F\u3031-\u3035\u3038-\u303C\u3041-\u3096\u3099\u309A\u309D-\u309F\u30A1-\u30FA\u30FC-\u30FF\u3105-\u312D\u3131-\u318E\u31A0-\u31BA\u31F0-\u31FF\u3400-\u4DB5\u4E00-\u9FD5\uA000-\uA48C\uA4D0-\uA4FD\uA500-\uA60C\uA610-\uA62B\uA640-\uA66F\uA674-\uA67D\uA67F-\uA6F1\uA717-\uA71F\uA722-\uA788\uA78B-\uA7AE\uA7B0-\uA7B7\uA7F7-\uA827\uA840-\uA873\uA880-\uA8C5\uA8D0-\uA8D9\uA8E0-\uA8F7\uA8FB\uA8FD\uA900-\uA92D\uA930-\uA953\uA960-\uA97C\uA980-\uA9C0\uA9CF-\uA9D9\uA9E0-\uA9FE\uAA00-\uAA36\uAA40-\uAA4D\uAA50-\uAA59\uAA60-\uAA76\uAA7A-\uAAC2\uAADB-\uAADD\uAAE0-\uAAEF\uAAF2-\uAAF6\uAB01-\uAB06\uAB09-\uAB0E\uAB11-\uAB16\uAB20-\uAB26\uAB28-\uAB2E\uAB30-\uAB5A\uAB5C-\uAB65\uAB70-\uABEA\uABEC\uABED\uABF0-\uABF9\uAC00-\uD7A3\uD7B0-\uD7C6\uD7CB-\uD7FB\uF900-\uFA6D\uFA70-\uFAD9\uFB00-\uFB06\uFB13-\uFB17\uFB1D-\uFB28\uFB2A-\uFB36\uFB38-\uFB3C\uFB3E\uFB40\uFB41\uFB43\uFB44\uFB46-\uFBB1\uFBD3-\uFD3D\uFD50-\uFD8F\uFD92-\uFDC7\uFDF0-\uFDFB\uFE00-\uFE0F\uFE20-\uFE2F\uFE33\uFE34\uFE4D-\uFE4F\uFE70-\uFE74\uFE76-\uFEFC\uFF10-\uFF19\uFF21-\uFF3A\uFF3F\uFF41-\uFF5A\uFF66-\uFFBE\uFFC2-\uFFC7\uFFCA-\uFFCF\uFFD2-\uFFD7\uFFDA-\uFFDC]/ |
|
}; |
|
ES6Regex = { |
|
// ECMAScript 6/Unicode v9.0.0 NonAsciiIdentifierStart: |
|
NonAsciiIdentifierStart: /[\xAA\xB5\xBA\xC0-\xD6\xD8-\xF6\xF8-\u02C1\u02C6-\u02D1\u02E0-\u02E4\u02EC\u02EE\u0370-\u0374\u0376\u0377\u037A-\u037D\u037F\u0386\u0388-\u038A\u038C\u038E-\u03A1\u03A3-\u03F5\u03F7-\u0481\u048A-\u052F\u0531-\u0556\u0559\u0561-\u0587\u05D0-\u05EA\u05F0-\u05F2\u0620-\u064A\u066E\u066F\u0671-\u06D3\u06D5\u06E5\u06E6\u06EE\u06EF\u06FA-\u06FC\u06FF\u0710\u0712-\u072F\u074D-\u07A5\u07B1\u07CA-\u07EA\u07F4\u07F5\u07FA\u0800-\u0815\u081A\u0824\u0828\u0840-\u0858\u08A0-\u08B4\u08B6-\u08BD\u0904-\u0939\u093D\u0950\u0958-\u0961\u0971-\u0980\u0985-\u098C\u098F\u0990\u0993-\u09A8\u09AA-\u09B0\u09B2\u09B6-\u09B9\u09BD\u09CE\u09DC\u09DD\u09DF-\u09E1\u09F0\u09F1\u0A05-\u0A0A\u0A0F\u0A10\u0A13-\u0A28\u0A2A-\u0A30\u0A32\u0A33\u0A35\u0A36\u0A38\u0A39\u0A59-\u0A5C\u0A5E\u0A72-\u0A74\u0A85-\u0A8D\u0A8F-\u0A91\u0A93-\u0AA8\u0AAA-\u0AB0\u0AB2\u0AB3\u0AB5-\u0AB9\u0ABD\u0AD0\u0AE0\u0AE1\u0AF9\u0B05-\u0B0C\u0B0F\u0B10\u0B13-\u0B28\u0B2A-\u0B30\u0B32\u0B33\u0B35-\u0B39\u0B3D\u0B5C\u0B5D\u0B5F-\u0B61\u0B71\u0B83\u0B85-\u0B8A\u0B8E-\u0B90\u0B92-\u0B95\u0B99\u0B9A\u0B9C\u0B9E\u0B9F\u0BA3\u0BA4\u0BA8-\u0BAA\u0BAE-\u0BB9\u0BD0\u0C05-\u0C0C\u0C0E-\u0C10\u0C12-\u0C28\u0C2A-\u0C39\u0C3D\u0C58-\u0C5A\u0C60\u0C61\u0C80\u0C85-\u0C8C\u0C8E-\u0C90\u0C92-\u0CA8\u0CAA-\u0CB3\u0CB5-\u0CB9\u0CBD\u0CDE\u0CE0\u0CE1\u0CF1\u0CF2\u0D05-\u0D0C\u0D0E-\u0D10\u0D12-\u0D3A\u0D3D\u0D4E\u0D54-\u0D56\u0D5F-\u0D61\u0D7A-\u0D7F\u0D85-\u0D96\u0D9A-\u0DB1\u0DB3-\u0DBB\u0DBD\u0DC0-\u0DC6\u0E01-\u0E30\u0E32\u0E33\u0E40-\u0E46\u0E81\u0E82\u0E84\u0E87\u0E88\u0E8A\u0E8D\u0E94-\u0E97\u0E99-\u0E9F\u0EA1-\u0EA3\u0EA5\u0EA7\u0EAA\u0EAB\u0EAD-\u0EB0\u0EB2\u0EB3\u0EBD\u0EC0-\u0EC4\u0EC6\u0EDC-\u0EDF\u0F00\u0F40-\u0F47\u0F49-\u0F6C\u0F88-\u0F8C\u1000-\u102A\u103F\u1050-\u1055\u105A-\u105D\u1061\u1065\u1066\u106E-\u1070\u1075-\u1081\u108E\u10A0-\u10C5\u10C7\u10CD\u10D0-\u10FA\u10FC-\u1248\u124A-\u124D\u1250-\u1256\u1258\u125A-\u125D\u1260-\u1288\u128A-\u128D\u1290-\u12B0\u12B2-\u12B5\u12B8-\u12BE\u12C0\u12C2-\u12C5\u12C8-\u12D6\u12D8-\u1310\u1312-\u1315\u1318-\u135A\u1380-\u138F\u13A0-\u13F5\u13F8-\u13FD\u1401-\u166C\u166F-\u167F\u1681-\u169A\u16A0-\u16EA\u16EE-\u16F8\u1700-\u170C\u170E-\u1711\u1720-\u1731\u1740-\u1751\u1760-\u176C\u176E-\u1770\u1780-\u17B3\u17D7\u17DC\u1820-\u1877\u1880-\u18A8\u18AA\u18B0-\u18F5\u1900-\u191E\u1950-\u196D\u1970-\u1974\u1980-\u19AB\u19B0-\u19C9\u1A00-\u1A16\u1A20-\u1A54\u1AA7\u1B05-\u1B33\u1B45-\u1B4B\u1B83-\u1BA0\u1BAE\u1BAF\u1BBA-\u1BE5\u1C00-\u1C23\u1C4D-\u1C4F\u1C5A-\u1C7D\u1C80-\u1C88\u1CE9-\u1CEC\u1CEE-\u1CF1\u1CF5\u1CF6\u1D00-\u1DBF\u1E00-\u1F15\u1F18-\u1F1D\u1F20-\u1F45\u1F48-\u1F4D\u1F50-\u1F57\u1F59\u1F5B\u1F5D\u1F5F-\u1F7D\u1F80-\u1FB4\u1FB6-\u1FBC\u1FBE\u1FC2-\u1FC4\u1FC6-\u1FCC\u1FD0-\u1FD3\u1FD6-\u1FDB\u1FE0-\u1FEC\u1FF2-\u1FF4\u1FF6-\u1FFC\u2071\u207F\u2090-\u209C\u2102\u2107\u210A-\u2113\u2115\u2118-\u211D\u2124\u2126\u2128\u212A-\u2139\u213C-\u213F\u2145-\u2149\u214E\u2160-\u2188\u2C00-\u2C2E\u2C30-\u2C5E\u2C60-\u2CE4\u2CEB-\u2CEE\u2CF2\u2CF3\u2D00-\u2D25\u2D27\u2D2D\u2D30-\u2D67\u2D6F\u2D80-\u2D96\u2DA0-\u2DA6\u2DA8-\u2DAE\u2DB0-\u2DB6\u2DB8-\u2DBE\u2DC0-\u2DC6\u2DC8-\u2DCE\u2DD0-\u2DD6\u2DD8-\u2DDE\u3005-\u3007\u3021-\u3029\u3031-\u3035\u3038-\u303C\u3041-\u3096\u309B-\u309F\u30A1-\u30FA\u30FC-\u30FF\u3105-\u312D\u3131-\u318E\u31A0-\u31BA\u31F0-\u31FF\u3400-\u4DB5\u4E00-\u9FD5\uA000-\uA48C\uA4D0-\uA4FD\uA500-\uA60C\uA610-\uA61F\uA62A\uA62B\uA640-\uA66E\uA67F-\uA69D\uA6A0-\uA6EF\uA717-\uA71F\uA722-\uA788\uA78B-\uA7AE\uA7B0-\uA7B7\uA7F7-\uA801\uA803-\uA805\uA807-\uA80A\uA80C-\uA822\uA840-\uA873\uA882-\uA8B3\uA8F2-\uA8F7\uA8FB\uA8FD\uA90A-\uA925\uA930-\uA946\uA960-\uA97C\uA984-\uA9B2\uA9CF\uA9E0-\uA9E4\uA9E6-\uA9EF\uA9FA-\uA9FE\uAA00-\uAA28\uAA40-\uAA42\uAA44-\uAA4B\uAA60-\uAA76\uAA7A\uAA7E-\uAAAF\uAAB1\uAAB5\uAAB6\uAAB9-\uAABD\uAAC0\uAAC2\uAADB-\uAADD\uAAE0-\uAAEA\uAAF2-\uAAF4\uAB01-\uAB06\uAB09-\uAB0E\uAB11-\uAB16\uAB20-\uAB26\uAB28-\uAB2E\uAB30-\uAB5A\uAB5C-\uAB65\uAB70-\uABE2\uAC00-\uD7A3\uD7B0-\uD7C6\uD7CB-\uD7FB\uF900-\uFA6D\uFA70-\uFAD9\uFB00-\uFB06\uFB13-\uFB17\uFB1D\uFB1F-\uFB28\uFB2A-\uFB36\uFB38-\uFB3C\uFB3E\uFB40\uFB41\uFB43\uFB44\uFB46-\uFBB1\uFBD3-\uFD3D\uFD50-\uFD8F\uFD92-\uFDC7\uFDF0-\uFDFB\uFE70-\uFE74\uFE76-\uFEFC\uFF21-\uFF3A\uFF41-\uFF5A\uFF66-\uFFBE\uFFC2-\uFFC7\uFFCA-\uFFCF\uFFD2-\uFFD7\uFFDA-\uFFDC]|\uD800[\uDC00-\uDC0B\uDC0D-\uDC26\uDC28-\uDC3A\uDC3C\uDC3D\uDC3F-\uDC4D\uDC50-\uDC5D\uDC80-\uDCFA\uDD40-\uDD74\uDE80-\uDE9C\uDEA0-\uDED0\uDF00-\uDF1F\uDF30-\uDF4A\uDF50-\uDF75\uDF80-\uDF9D\uDFA0-\uDFC3\uDFC8-\uDFCF\uDFD1-\uDFD5]|\uD801[\uDC00-\uDC9D\uDCB0-\uDCD3\uDCD8-\uDCFB\uDD00-\uDD27\uDD30-\uDD63\uDE00-\uDF36\uDF40-\uDF55\uDF60-\uDF67]|\uD802[\uDC00-\uDC05\uDC08\uDC0A-\uDC35\uDC37\uDC38\uDC3C\uDC3F-\uDC55\uDC60-\uDC76\uDC80-\uDC9E\uDCE0-\uDCF2\uDCF4\uDCF5\uDD00-\uDD15\uDD20-\uDD39\uDD80-\uDDB7\uDDBE\uDDBF\uDE00\uDE10-\uDE13\uDE15-\uDE17\uDE19-\uDE33\uDE60-\uDE7C\uDE80-\uDE9C\uDEC0-\uDEC7\uDEC9-\uDEE4\uDF00-\uDF35\uDF40-\uDF55\uDF60-\uDF72\uDF80-\uDF91]|\uD803[\uDC00-\uDC48\uDC80-\uDCB2\uDCC0-\uDCF2]|\uD804[\uDC03-\uDC37\uDC83-\uDCAF\uDCD0-\uDCE8\uDD03-\uDD26\uDD50-\uDD72\uDD76\uDD83-\uDDB2\uDDC1-\uDDC4\uDDDA\uDDDC\uDE00-\uDE11\uDE13-\uDE2B\uDE80-\uDE86\uDE88\uDE8A-\uDE8D\uDE8F-\uDE9D\uDE9F-\uDEA8\uDEB0-\uDEDE\uDF05-\uDF0C\uDF0F\uDF10\uDF13-\uDF28\uDF2A-\uDF30\uDF32\uDF33\uDF35-\uDF39\uDF3D\uDF50\uDF5D-\uDF61]|\uD805[\uDC00-\uDC34\uDC47-\uDC4A\uDC80-\uDCAF\uDCC4\uDCC5\uDCC7\uDD80-\uDDAE\uDDD8-\uDDDB\uDE00-\uDE2F\uDE44\uDE80-\uDEAA\uDF00-\uDF19]|\uD806[\uDCA0-\uDCDF\uDCFF\uDEC0-\uDEF8]|\uD807[\uDC00-\uDC08\uDC0A-\uDC2E\uDC40\uDC72-\uDC8F]|\uD808[\uDC00-\uDF99]|\uD809[\uDC00-\uDC6E\uDC80-\uDD43]|[\uD80C\uD81C-\uD820\uD840-\uD868\uD86A-\uD86C\uD86F-\uD872][\uDC00-\uDFFF]|\uD80D[\uDC00-\uDC2E]|\uD811[\uDC00-\uDE46]|\uD81A[\uDC00-\uDE38\uDE40-\uDE5E\uDED0-\uDEED\uDF00-\uDF2F\uDF40-\uDF43\uDF63-\uDF77\uDF7D-\uDF8F]|\uD81B[\uDF00-\uDF44\uDF50\uDF93-\uDF9F\uDFE0]|\uD821[\uDC00-\uDFEC]|\uD822[\uDC00-\uDEF2]|\uD82C[\uDC00\uDC01]|\uD82F[\uDC00-\uDC6A\uDC70-\uDC7C\uDC80-\uDC88\uDC90-\uDC99]|\uD835[\uDC00-\uDC54\uDC56-\uDC9C\uDC9E\uDC9F\uDCA2\uDCA5\uDCA6\uDCA9-\uDCAC\uDCAE-\uDCB9\uDCBB\uDCBD-\uDCC3\uDCC5-\uDD05\uDD07-\uDD0A\uDD0D-\uDD14\uDD16-\uDD1C\uDD1E-\uDD39\uDD3B-\uDD3E\uDD40-\uDD44\uDD46\uDD4A-\uDD50\uDD52-\uDEA5\uDEA8-\uDEC0\uDEC2-\uDEDA\uDEDC-\uDEFA\uDEFC-\uDF14\uDF16-\uDF34\uDF36-\uDF4E\uDF50-\uDF6E\uDF70-\uDF88\uDF8A-\uDFA8\uDFAA-\uDFC2\uDFC4-\uDFCB]|\uD83A[\uDC00-\uDCC4\uDD00-\uDD43]|\uD83B[\uDE00-\uDE03\uDE05-\uDE1F\uDE21\uDE22\uDE24\uDE27\uDE29-\uDE32\uDE34-\uDE37\uDE39\uDE3B\uDE42\uDE47\uDE49\uDE4B\uDE4D-\uDE4F\uDE51\uDE52\uDE54\uDE57\uDE59\uDE5B\uDE5D\uDE5F\uDE61\uDE62\uDE64\uDE67-\uDE6A\uDE6C-\uDE72\uDE74-\uDE77\uDE79-\uDE7C\uDE7E\uDE80-\uDE89\uDE8B-\uDE9B\uDEA1-\uDEA3\uDEA5-\uDEA9\uDEAB-\uDEBB]|\uD869[\uDC00-\uDED6\uDF00-\uDFFF]|\uD86D[\uDC00-\uDF34\uDF40-\uDFFF]|\uD86E[\uDC00-\uDC1D\uDC20-\uDFFF]|\uD873[\uDC00-\uDEA1]|\uD87E[\uDC00-\uDE1D]/, |
|
// ECMAScript 6/Unicode v9.0.0 NonAsciiIdentifierPart: |
|
NonAsciiIdentifierPart: /[\xAA\xB5\xB7\xBA\xC0-\xD6\xD8-\xF6\xF8-\u02C1\u02C6-\u02D1\u02E0-\u02E4\u02EC\u02EE\u0300-\u0374\u0376\u0377\u037A-\u037D\u037F\u0386-\u038A\u038C\u038E-\u03A1\u03A3-\u03F5\u03F7-\u0481\u0483-\u0487\u048A-\u052F\u0531-\u0556\u0559\u0561-\u0587\u0591-\u05BD\u05BF\u05C1\u05C2\u05C4\u05C5\u05C7\u05D0-\u05EA\u05F0-\u05F2\u0610-\u061A\u0620-\u0669\u066E-\u06D3\u06D5-\u06DC\u06DF-\u06E8\u06EA-\u06FC\u06FF\u0710-\u074A\u074D-\u07B1\u07C0-\u07F5\u07FA\u0800-\u082D\u0840-\u085B\u08A0-\u08B4\u08B6-\u08BD\u08D4-\u08E1\u08E3-\u0963\u0966-\u096F\u0971-\u0983\u0985-\u098C\u098F\u0990\u0993-\u09A8\u09AA-\u09B0\u09B2\u09B6-\u09B9\u09BC-\u09C4\u09C7\u09C8\u09CB-\u09CE\u09D7\u09DC\u09DD\u09DF-\u09E3\u09E6-\u09F1\u0A01-\u0A03\u0A05-\u0A0A\u0A0F\u0A10\u0A13-\u0A28\u0A2A-\u0A30\u0A32\u0A33\u0A35\u0A36\u0A38\u0A39\u0A3C\u0A3E-\u0A42\u0A47\u0A48\u0A4B-\u0A4D\u0A51\u0A59-\u0A5C\u0A5E\u0A66-\u0A75\u0A81-\u0A83\u0A85-\u0A8D\u0A8F-\u0A91\u0A93-\u0AA8\u0AAA-\u0AB0\u0AB2\u0AB3\u0AB5-\u0AB9\u0ABC-\u0AC5\u0AC7-\u0AC9\u0ACB-\u0ACD\u0AD0\u0AE0-\u0AE3\u0AE6-\u0AEF\u0AF9\u0B01-\u0B03\u0B05-\u0B0C\u0B0F\u0B10\u0B13-\u0B28\u0B2A-\u0B30\u0B32\u0B33\u0B35-\u0B39\u0B3C-\u0B44\u0B47\u0B48\u0B4B-\u0B4D\u0B56\u0B57\u0B5C\u0B5D\u0B5F-\u0B63\u0B66-\u0B6F\u0B71\u0B82\u0B83\u0B85-\u0B8A\u0B8E-\u0B90\u0B92-\u0B95\u0B99\u0B9A\u0B9C\u0B9E\u0B9F\u0BA3\u0BA4\u0BA8-\u0BAA\u0BAE-\u0BB9\u0BBE-\u0BC2\u0BC6-\u0BC8\u0BCA-\u0BCD\u0BD0\u0BD7\u0BE6-\u0BEF\u0C00-\u0C03\u0C05-\u0C0C\u0C0E-\u0C10\u0C12-\u0C28\u0C2A-\u0C39\u0C3D-\u0C44\u0C46-\u0C48\u0C4A-\u0C4D\u0C55\u0C56\u0C58-\u0C5A\u0C60-\u0C63\u0C66-\u0C6F\u0C80-\u0C83\u0C85-\u0C8C\u0C8E-\u0C90\u0C92-\u0CA8\u0CAA-\u0CB3\u0CB5-\u0CB9\u0CBC-\u0CC4\u0CC6-\u0CC8\u0CCA-\u0CCD\u0CD5\u0CD6\u0CDE\u0CE0-\u0CE3\u0CE6-\u0CEF\u0CF1\u0CF2\u0D01-\u0D03\u0D05-\u0D0C\u0D0E-\u0D10\u0D12-\u0D3A\u0D3D-\u0D44\u0D46-\u0D48\u0D4A-\u0D4E\u0D54-\u0D57\u0D5F-\u0D63\u0D66-\u0D6F\u0D7A-\u0D7F\u0D82\u0D83\u0D85-\u0D96\u0D9A-\u0DB1\u0DB3-\u0DBB\u0DBD\u0DC0-\u0DC6\u0DCA\u0DCF-\u0DD4\u0DD6\u0DD8-\u0DDF\u0DE6-\u0DEF\u0DF2\u0DF3\u0E01-\u0E3A\u0E40-\u0E4E\u0E50-\u0E59\u0E81\u0E82\u0E84\u0E87\u0E88\u0E8A\u0E8D\u0E94-\u0E97\u0E99-\u0E9F\u0EA1-\u0EA3\u0EA5\u0EA7\u0EAA\u0EAB\u0EAD-\u0EB9\u0EBB-\u0EBD\u0EC0-\u0EC4\u0EC6\u0EC8-\u0ECD\u0ED0-\u0ED9\u0EDC-\u0EDF\u0F00\u0F18\u0F19\u0F20-\u0F29\u0F35\u0F37\u0F39\u0F3E-\u0F47\u0F49-\u0F6C\u0F71-\u0F84\u0F86-\u0F97\u0F99-\u0FBC\u0FC6\u1000-\u1049\u1050-\u109D\u10A0-\u10C5\u10C7\u10CD\u10D0-\u10FA\u10FC-\u1248\u124A-\u124D\u1250-\u1256\u1258\u125A-\u125D\u1260-\u1288\u128A-\u128D\u1290-\u12B0\u12B2-\u12B5\u12B8-\u12BE\u12C0\u12C2-\u12C5\u12C8-\u12D6\u12D8-\u1310\u1312-\u1315\u1318-\u135A\u135D-\u135F\u1369-\u1371\u1380-\u138F\u13A0-\u13F5\u13F8-\u13FD\u1401-\u166C\u166F-\u167F\u1681-\u169A\u16A0-\u16EA\u16EE-\u16F8\u1700-\u170C\u170E-\u1714\u1720-\u1734\u1740-\u1753\u1760-\u176C\u176E-\u1770\u1772\u1773\u1780-\u17D3\u17D7\u17DC\u17DD\u17E0-\u17E9\u180B-\u180D\u1810-\u1819\u1820-\u1877\u1880-\u18AA\u18B0-\u18F5\u1900-\u191E\u1920-\u192B\u1930-\u193B\u1946-\u196D\u1970-\u1974\u1980-\u19AB\u19B0-\u19C9\u19D0-\u19DA\u1A00-\u1A1B\u1A20-\u1A5E\u1A60-\u1A7C\u1A7F-\u1A89\u1A90-\u1A99\u1AA7\u1AB0-\u1ABD\u1B00-\u1B4B\u1B50-\u1B59\u1B6B-\u1B73\u1B80-\u1BF3\u1C00-\u1C37\u1C40-\u1C49\u1C4D-\u1C7D\u1C80-\u1C88\u1CD0-\u1CD2\u1CD4-\u1CF6\u1CF8\u1CF9\u1D00-\u1DF5\u1DFB-\u1F15\u1F18-\u1F1D\u1F20-\u1F45\u1F48-\u1F4D\u1F50-\u1F57\u1F59\u1F5B\u1F5D\u1F5F-\u1F7D\u1F80-\u1FB4\u1FB6-\u1FBC\u1FBE\u1FC2-\u1FC4\u1FC6-\u1FCC\u1FD0-\u1FD3\u1FD6-\u1FDB\u1FE0-\u1FEC\u1FF2-\u1FF4\u1FF6-\u1FFC\u200C\u200D\u203F\u2040\u2054\u2071\u207F\u2090-\u209C\u20D0-\u20DC\u20E1\u20E5-\u20F0\u2102\u2107\u210A-\u2113\u2115\u2118-\u211D\u2124\u2126\u2128\u212A-\u2139\u213C-\u213F\u2145-\u2149\u214E\u2160-\u2188\u2C00-\u2C2E\u2C30-\u2C5E\u2C60-\u2CE4\u2CEB-\u2CF3\u2D00-\u2D25\u2D27\u2D2D\u2D30-\u2D67\u2D6F\u2D7F-\u2D96\u2DA0-\u2DA6\u2DA8-\u2DAE\u2DB0-\u2DB6\u2DB8-\u2DBE\u2DC0-\u2DC6\u2DC8-\u2DCE\u2DD0-\u2DD6\u2DD8-\u2DDE\u2DE0-\u2DFF\u3005-\u3007\u3021-\u302F\u3031-\u3035\u3038-\u303C\u3041-\u3096\u3099-\u309F\u30A1-\u30FA\u30FC-\u30FF\u3105-\u312D\u3131-\u318E\u31A0-\u31BA\u31F0-\u31FF\u3400-\u4DB5\u4E00-\u9FD5\uA000-\uA48C\uA4D0-\uA4FD\uA500-\uA60C\uA610-\uA62B\uA640-\uA66F\uA674-\uA67D\uA67F-\uA6F1\uA717-\uA71F\uA722-\uA788\uA78B-\uA7AE\uA7B0-\uA7B7\uA7F7-\uA827\uA840-\uA873\uA880-\uA8C5\uA8D0-\uA8D9\uA8E0-\uA8F7\uA8FB\uA8FD\uA900-\uA92D\uA930-\uA953\uA960-\uA97C\uA980-\uA9C0\uA9CF-\uA9D9\uA9E0-\uA9FE\uAA00-\uAA36\uAA40-\uAA4D\uAA50-\uAA59\uAA60-\uAA76\uAA7A-\uAAC2\uAADB-\uAADD\uAAE0-\uAAEF\uAAF2-\uAAF6\uAB01-\uAB06\uAB09-\uAB0E\uAB11-\uAB16\uAB20-\uAB26\uAB28-\uAB2E\uAB30-\uAB5A\uAB5C-\uAB65\uAB70-\uABEA\uABEC\uABED\uABF0-\uABF9\uAC00-\uD7A3\uD7B0-\uD7C6\uD7CB-\uD7FB\uF900-\uFA6D\uFA70-\uFAD9\uFB00-\uFB06\uFB13-\uFB17\uFB1D-\uFB28\uFB2A-\uFB36\uFB38-\uFB3C\uFB3E\uFB40\uFB41\uFB43\uFB44\uFB46-\uFBB1\uFBD3-\uFD3D\uFD50-\uFD8F\uFD92-\uFDC7\uFDF0-\uFDFB\uFE00-\uFE0F\uFE20-\uFE2F\uFE33\uFE34\uFE4D-\uFE4F\uFE70-\uFE74\uFE76-\uFEFC\uFF10-\uFF19\uFF21-\uFF3A\uFF3F\uFF41-\uFF5A\uFF66-\uFFBE\uFFC2-\uFFC7\uFFCA-\uFFCF\uFFD2-\uFFD7\uFFDA-\uFFDC]|\uD800[\uDC00-\uDC0B\uDC0D-\uDC26\uDC28-\uDC3A\uDC3C\uDC3D\uDC3F-\uDC4D\uDC50-\uDC5D\uDC80-\uDCFA\uDD40-\uDD74\uDDFD\uDE80-\uDE9C\uDEA0-\uDED0\uDEE0\uDF00-\uDF1F\uDF30-\uDF4A\uDF50-\uDF7A\uDF80-\uDF9D\uDFA0-\uDFC3\uDFC8-\uDFCF\uDFD1-\uDFD5]|\uD801[\uDC00-\uDC9D\uDCA0-\uDCA9\uDCB0-\uDCD3\uDCD8-\uDCFB\uDD00-\uDD27\uDD30-\uDD63\uDE00-\uDF36\uDF40-\uDF55\uDF60-\uDF67]|\uD802[\uDC00-\uDC05\uDC08\uDC0A-\uDC35\uDC37\uDC38\uDC3C\uDC3F-\uDC55\uDC60-\uDC76\uDC80-\uDC9E\uDCE0-\uDCF2\uDCF4\uDCF5\uDD00-\uDD15\uDD20-\uDD39\uDD80-\uDDB7\uDDBE\uDDBF\uDE00-\uDE03\uDE05\uDE06\uDE0C-\uDE13\uDE15-\uDE17\uDE19-\uDE33\uDE38-\uDE3A\uDE3F\uDE60-\uDE7C\uDE80-\uDE9C\uDEC0-\uDEC7\uDEC9-\uDEE6\uDF00-\uDF35\uDF40-\uDF55\uDF60-\uDF72\uDF80-\uDF91]|\uD803[\uDC00-\uDC48\uDC80-\uDCB2\uDCC0-\uDCF2]|\uD804[\uDC00-\uDC46\uDC66-\uDC6F\uDC7F-\uDCBA\uDCD0-\uDCE8\uDCF0-\uDCF9\uDD00-\uDD34\uDD36-\uDD3F\uDD50-\uDD73\uDD76\uDD80-\uDDC4\uDDCA-\uDDCC\uDDD0-\uDDDA\uDDDC\uDE00-\uDE11\uDE13-\uDE37\uDE3E\uDE80-\uDE86\uDE88\uDE8A-\uDE8D\uDE8F-\uDE9D\uDE9F-\uDEA8\uDEB0-\uDEEA\uDEF0-\uDEF9\uDF00-\uDF03\uDF05-\uDF0C\uDF0F\uDF10\uDF13-\uDF28\uDF2A-\uDF30\uDF32\uDF33\uDF35-\uDF39\uDF3C-\uDF44\uDF47\uDF48\uDF4B-\uDF4D\uDF50\uDF57\uDF5D-\uDF63\uDF66-\uDF6C\uDF70-\uDF74]|\uD805[\uDC00-\uDC4A\uDC50-\uDC59\uDC80-\uDCC5\uDCC7\uDCD0-\uDCD9\uDD80-\uDDB5\uDDB8-\uDDC0\uDDD8-\uDDDD\uDE00-\uDE40\uDE44\uDE50-\uDE59\uDE80-\uDEB7\uDEC0-\uDEC9\uDF00-\uDF19\uDF1D-\uDF2B\uDF30-\uDF39]|\uD806[\uDCA0-\uDCE9\uDCFF\uDEC0-\uDEF8]|\uD807[\uDC00-\uDC08\uDC0A-\uDC36\uDC38-\uDC40\uDC50-\uDC59\uDC72-\uDC8F\uDC92-\uDCA7\uDCA9-\uDCB6]|\uD808[\uDC00-\uDF99]|\uD809[\uDC00-\uDC6E\uDC80-\uDD43]|[\uD80C\uD81C-\uD820\uD840-\uD868\uD86A-\uD86C\uD86F-\uD872][\uDC00-\uDFFF]|\uD80D[\uDC00-\uDC2E]|\uD811[\uDC00-\uDE46]|\uD81A[\uDC00-\uDE38\uDE40-\uDE5E\uDE60-\uDE69\uDED0-\uDEED\uDEF0-\uDEF4\uDF00-\uDF36\uDF40-\uDF43\uDF50-\uDF59\uDF63-\uDF77\uDF7D-\uDF8F]|\uD81B[\uDF00-\uDF44\uDF50-\uDF7E\uDF8F-\uDF9F\uDFE0]|\uD821[\uDC00-\uDFEC]|\uD822[\uDC00-\uDEF2]|\uD82C[\uDC00\uDC01]|\uD82F[\uDC00-\uDC6A\uDC70-\uDC7C\uDC80-\uDC88\uDC90-\uDC99\uDC9D\uDC9E]|\uD834[\uDD65-\uDD69\uDD6D-\uDD72\uDD7B-\uDD82\uDD85-\uDD8B\uDDAA-\uDDAD\uDE42-\uDE44]|\uD835[\uDC00-\uDC54\uDC56-\uDC9C\uDC9E\uDC9F\uDCA2\uDCA5\uDCA6\uDCA9-\uDCAC\uDCAE-\uDCB9\uDCBB\uDCBD-\uDCC3\uDCC5-\uDD05\uDD07-\uDD0A\uDD0D-\uDD14\uDD16-\uDD1C\uDD1E-\uDD39\uDD3B-\uDD3E\uDD40-\uDD44\uDD46\uDD4A-\uDD50\uDD52-\uDEA5\uDEA8-\uDEC0\uDEC2-\uDEDA\uDEDC-\uDEFA\uDEFC-\uDF14\uDF16-\uDF34\uDF36-\uDF4E\uDF50-\uDF6E\uDF70-\uDF88\uDF8A-\uDFA8\uDFAA-\uDFC2\uDFC4-\uDFCB\uDFCE-\uDFFF]|\uD836[\uDE00-\uDE36\uDE3B-\uDE6C\uDE75\uDE84\uDE9B-\uDE9F\uDEA1-\uDEAF]|\uD838[\uDC00-\uDC06\uDC08-\uDC18\uDC1B-\uDC21\uDC23\uDC24\uDC26-\uDC2A]|\uD83A[\uDC00-\uDCC4\uDCD0-\uDCD6\uDD00-\uDD4A\uDD50-\uDD59]|\uD83B[\uDE00-\uDE03\uDE05-\uDE1F\uDE21\uDE22\uDE24\uDE27\uDE29-\uDE32\uDE34-\uDE37\uDE39\uDE3B\uDE42\uDE47\uDE49\uDE4B\uDE4D-\uDE4F\uDE51\uDE52\uDE54\uDE57\uDE59\uDE5B\uDE5D\uDE5F\uDE61\uDE62\uDE64\uDE67-\uDE6A\uDE6C-\uDE72\uDE74-\uDE77\uDE79-\uDE7C\uDE7E\uDE80-\uDE89\uDE8B-\uDE9B\uDEA1-\uDEA3\uDEA5-\uDEA9\uDEAB-\uDEBB]|\uD869[\uDC00-\uDED6\uDF00-\uDFFF]|\uD86D[\uDC00-\uDF34\uDF40-\uDFFF]|\uD86E[\uDC00-\uDC1D\uDC20-\uDFFF]|\uD873[\uDC00-\uDEA1]|\uD87E[\uDC00-\uDE1D]|\uDB40[\uDD00-\uDDEF]/ |
|
}; |
|
|
|
function isDecimalDigit(ch) { |
|
return 0x30 <= ch && ch <= 0x39; // 0..9 |
|
} |
|
|
|
function isHexDigit(ch) { |
|
return 0x30 <= ch && ch <= 0x39 || // 0..9 |
|
0x61 <= ch && ch <= 0x66 || // a..f |
|
0x41 <= ch && ch <= 0x46; // A..F |
|
} |
|
|
|
function isOctalDigit(ch) { |
|
return ch >= 0x30 && ch <= 0x37; // 0..7 |
|
} // 7.2 White Space |
|
|
|
|
|
NON_ASCII_WHITESPACES = [0x1680, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, 0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202F, 0x205F, 0x3000, 0xFEFF]; |
|
|
|
function isWhiteSpace(ch) { |
|
return ch === 0x20 || ch === 0x09 || ch === 0x0B || ch === 0x0C || ch === 0xA0 || ch >= 0x1680 && NON_ASCII_WHITESPACES.indexOf(ch) >= 0; |
|
} // 7.3 Line Terminators |
|
|
|
|
|
function isLineTerminator(ch) { |
|
return ch === 0x0A || ch === 0x0D || ch === 0x2028 || ch === 0x2029; |
|
} // 7.6 Identifier Names and Identifiers |
|
|
|
|
|
function fromCodePoint(cp) { |
|
if (cp <= 0xFFFF) { |
|
return String.fromCharCode(cp); |
|
} |
|
|
|
var cu1 = String.fromCharCode(Math.floor((cp - 0x10000) / 0x400) + 0xD800); |
|
var cu2 = String.fromCharCode((cp - 0x10000) % 0x400 + 0xDC00); |
|
return cu1 + cu2; |
|
} |
|
|
|
IDENTIFIER_START = new Array(0x80); |
|
|
|
for (ch = 0; ch < 0x80; ++ch) { |
|
IDENTIFIER_START[ch] = ch >= 0x61 && ch <= 0x7A || // a..z |
|
ch >= 0x41 && ch <= 0x5A || // A..Z |
|
ch === 0x24 || ch === 0x5F; // $ (dollar) and _ (underscore) |
|
} |
|
|
|
IDENTIFIER_PART = new Array(0x80); |
|
|
|
for (ch = 0; ch < 0x80; ++ch) { |
|
IDENTIFIER_PART[ch] = ch >= 0x61 && ch <= 0x7A || // a..z |
|
ch >= 0x41 && ch <= 0x5A || // A..Z |
|
ch >= 0x30 && ch <= 0x39 || // 0..9 |
|
ch === 0x24 || ch === 0x5F; // $ (dollar) and _ (underscore) |
|
} |
|
|
|
function isIdentifierStartES5(ch) { |
|
return ch < 0x80 ? IDENTIFIER_START[ch] : ES5Regex.NonAsciiIdentifierStart.test(fromCodePoint(ch)); |
|
} |
|
|
|
function isIdentifierPartES5(ch) { |
|
return ch < 0x80 ? IDENTIFIER_PART[ch] : ES5Regex.NonAsciiIdentifierPart.test(fromCodePoint(ch)); |
|
} |
|
|
|
function isIdentifierStartES6(ch) { |
|
return ch < 0x80 ? IDENTIFIER_START[ch] : ES6Regex.NonAsciiIdentifierStart.test(fromCodePoint(ch)); |
|
} |
|
|
|
function isIdentifierPartES6(ch) { |
|
return ch < 0x80 ? IDENTIFIER_PART[ch] : ES6Regex.NonAsciiIdentifierPart.test(fromCodePoint(ch)); |
|
} |
|
|
|
module.exports = { |
|
isDecimalDigit: isDecimalDigit, |
|
isHexDigit: isHexDigit, |
|
isOctalDigit: isOctalDigit, |
|
isWhiteSpace: isWhiteSpace, |
|
isLineTerminator: isLineTerminator, |
|
isIdentifierStartES5: isIdentifierStartES5, |
|
isIdentifierPartES5: isIdentifierPartES5, |
|
isIdentifierStartES6: isIdentifierStartES6, |
|
isIdentifierPartES6: isIdentifierPartES6 |
|
}; |
|
})(); |
|
/* vim: set sw=4 ts=4 et tw=80 : */ |
|
|
|
}); |
|
var code_1 = code.isDecimalDigit; |
|
var code_2 = code.isHexDigit; |
|
var code_3 = code.isOctalDigit; |
|
var code_4 = code.isWhiteSpace; |
|
var code_5 = code.isLineTerminator; |
|
var code_6 = code.isIdentifierStartES5; |
|
var code_7 = code.isIdentifierPartES5; |
|
var code_8 = code.isIdentifierStartES6; |
|
var code_9 = code.isIdentifierPartES6; |
|
|
|
var keyword = createCommonjsModule(function (module) { |
|
/* |
|
Copyright (C) 2013 Yusuke Suzuki <utatane.tea@gmail.com> |
|
|
|
Redistribution and use in source and binary forms, with or without |
|
modification, are permitted provided that the following conditions are met: |
|
|
|
* Redistributions of source code must retain the above copyright |
|
notice, this list of conditions and the following disclaimer. |
|
* Redistributions in binary form must reproduce the above copyright |
|
notice, this list of conditions and the following disclaimer in the |
|
documentation and/or other materials provided with the distribution. |
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
|
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
|
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
|
ARE DISCLAIMED. IN NO EVENT SHALL <COPYRIGHT HOLDER> BE LIABLE FOR ANY |
|
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES |
|
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
|
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND |
|
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
|
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
|
THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
|
*/ |
|
(function () { |
|
|
|
var code$1 = code; |
|
|
|
function isStrictModeReservedWordES6(id) { |
|
switch (id) { |
|
case 'implements': |
|
case 'interface': |
|
case 'package': |
|
case 'private': |
|
case 'protected': |
|
case 'public': |
|
case 'static': |
|
case 'let': |
|
return true; |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function isKeywordES5(id, strict) { |
|
// yield should not be treated as keyword under non-strict mode. |
|
if (!strict && id === 'yield') { |
|
return false; |
|
} |
|
|
|
return isKeywordES6(id, strict); |
|
} |
|
|
|
function isKeywordES6(id, strict) { |
|
if (strict && isStrictModeReservedWordES6(id)) { |
|
return true; |
|
} |
|
|
|
switch (id.length) { |
|
case 2: |
|
return id === 'if' || id === 'in' || id === 'do'; |
|
|
|
case 3: |
|
return id === 'var' || id === 'for' || id === 'new' || id === 'try'; |
|
|
|
case 4: |
|
return id === 'this' || id === 'else' || id === 'case' || id === 'void' || id === 'with' || id === 'enum'; |
|
|
|
case 5: |
|
return id === 'while' || id === 'break' || id === 'catch' || id === 'throw' || id === 'const' || id === 'yield' || id === 'class' || id === 'super'; |
|
|
|
case 6: |
|
return id === 'return' || id === 'typeof' || id === 'delete' || id === 'switch' || id === 'export' || id === 'import'; |
|
|
|
case 7: |
|
return id === 'default' || id === 'finally' || id === 'extends'; |
|
|
|
case 8: |
|
return id === 'function' || id === 'continue' || id === 'debugger'; |
|
|
|
case 10: |
|
return id === 'instanceof'; |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function isReservedWordES5(id, strict) { |
|
return id === 'null' || id === 'true' || id === 'false' || isKeywordES5(id, strict); |
|
} |
|
|
|
function isReservedWordES6(id, strict) { |
|
return id === 'null' || id === 'true' || id === 'false' || isKeywordES6(id, strict); |
|
} |
|
|
|
function isRestrictedWord(id) { |
|
return id === 'eval' || id === 'arguments'; |
|
} |
|
|
|
function isIdentifierNameES5(id) { |
|
var i, iz, ch; |
|
|
|
if (id.length === 0) { |
|
return false; |
|
} |
|
|
|
ch = id.charCodeAt(0); |
|
|
|
if (!code$1.isIdentifierStartES5(ch)) { |
|
return false; |
|
} |
|
|
|
for (i = 1, iz = id.length; i < iz; ++i) { |
|
ch = id.charCodeAt(i); |
|
|
|
if (!code$1.isIdentifierPartES5(ch)) { |
|
return false; |
|
} |
|
} |
|
|
|
return true; |
|
} |
|
|
|
function decodeUtf16(lead, trail) { |
|
return (lead - 0xD800) * 0x400 + (trail - 0xDC00) + 0x10000; |
|
} |
|
|
|
function isIdentifierNameES6(id) { |
|
var i, iz, ch, lowCh, check; |
|
|
|
if (id.length === 0) { |
|
return false; |
|
} |
|
|
|
check = code$1.isIdentifierStartES6; |
|
|
|
for (i = 0, iz = id.length; i < iz; ++i) { |
|
ch = id.charCodeAt(i); |
|
|
|
if (0xD800 <= ch && ch <= 0xDBFF) { |
|
++i; |
|
|
|
if (i >= iz) { |
|
return false; |
|
} |
|
|
|
lowCh = id.charCodeAt(i); |
|
|
|
if (!(0xDC00 <= lowCh && lowCh <= 0xDFFF)) { |
|
return false; |
|
} |
|
|
|
ch = decodeUtf16(ch, lowCh); |
|
} |
|
|
|
if (!check(ch)) { |
|
return false; |
|
} |
|
|
|
check = code$1.isIdentifierPartES6; |
|
} |
|
|
|
return true; |
|
} |
|
|
|
function isIdentifierES5(id, strict) { |
|
return isIdentifierNameES5(id) && !isReservedWordES5(id, strict); |
|
} |
|
|
|
function isIdentifierES6(id, strict) { |
|
return isIdentifierNameES6(id) && !isReservedWordES6(id, strict); |
|
} |
|
|
|
module.exports = { |
|
isKeywordES5: isKeywordES5, |
|
isKeywordES6: isKeywordES6, |
|
isReservedWordES5: isReservedWordES5, |
|
isReservedWordES6: isReservedWordES6, |
|
isRestrictedWord: isRestrictedWord, |
|
isIdentifierNameES5: isIdentifierNameES5, |
|
isIdentifierNameES6: isIdentifierNameES6, |
|
isIdentifierES5: isIdentifierES5, |
|
isIdentifierES6: isIdentifierES6 |
|
}; |
|
})(); |
|
/* vim: set sw=4 ts=4 et tw=80 : */ |
|
|
|
}); |
|
var keyword_1 = keyword.isKeywordES5; |
|
var keyword_2 = keyword.isKeywordES6; |
|
var keyword_3 = keyword.isReservedWordES5; |
|
var keyword_4 = keyword.isReservedWordES6; |
|
var keyword_5 = keyword.isRestrictedWord; |
|
var keyword_6 = keyword.isIdentifierNameES5; |
|
var keyword_7 = keyword.isIdentifierNameES6; |
|
var keyword_8 = keyword.isIdentifierES5; |
|
var keyword_9 = keyword.isIdentifierES6; |
|
|
|
var utils$1 = createCommonjsModule(function (module, exports) { |
|
/* |
|
Copyright (C) 2013 Yusuke Suzuki <utatane.tea@gmail.com> |
|
|
|
Redistribution and use in source and binary forms, with or without |
|
modification, are permitted provided that the following conditions are met: |
|
|
|
* Redistributions of source code must retain the above copyright |
|
notice, this list of conditions and the following disclaimer. |
|
* Redistributions in binary form must reproduce the above copyright |
|
notice, this list of conditions and the following disclaimer in the |
|
documentation and/or other materials provided with the distribution. |
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
|
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
|
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
|
ARE DISCLAIMED. IN NO EVENT SHALL <COPYRIGHT HOLDER> BE LIABLE FOR ANY |
|
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES |
|
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
|
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND |
|
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
|
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
|
THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
|
*/ |
|
(function () { |
|
|
|
exports.ast = ast; |
|
exports.code = code; |
|
exports.keyword = keyword; |
|
})(); |
|
/* vim: set sw=4 ts=4 et tw=80 : */ |
|
|
|
}); |
|
var utils_1$1 = utils$1.ast; |
|
var utils_2$1 = utils$1.code; |
|
var utils_3$1 = utils$1.keyword; |
|
|
|
var lib$1 = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
exports.shouldHighlight = shouldHighlight; |
|
exports.getChalk = getChalk; |
|
exports.default = highlight; |
|
|
|
function _jsTokens() { |
|
var data = _interopRequireWildcard(jsTokens); |
|
|
|
_jsTokens = function _jsTokens() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _esutils() { |
|
var data = _interopRequireDefault(utils$1); |
|
|
|
_esutils = function _esutils() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _chalk() { |
|
var data = _interopRequireDefault(chalk); |
|
|
|
_chalk = function _chalk() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _interopRequireDefault(obj) { |
|
return obj && obj.__esModule ? obj : { |
|
default: obj |
|
}; |
|
} |
|
|
|
function _interopRequireWildcard(obj) { |
|
if (obj && obj.__esModule) { |
|
return obj; |
|
} else { |
|
var newObj = {}; |
|
|
|
if (obj != null) { |
|
for (var key in obj) { |
|
if (Object.prototype.hasOwnProperty.call(obj, key)) { |
|
var desc = Object.defineProperty && Object.getOwnPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : {}; |
|
|
|
if (desc.get || desc.set) { |
|
Object.defineProperty(newObj, key, desc); |
|
} else { |
|
newObj[key] = obj[key]; |
|
} |
|
} |
|
} |
|
} |
|
|
|
newObj.default = obj; |
|
return newObj; |
|
} |
|
} |
|
|
|
function getDefs(chalk) { |
|
return { |
|
keyword: chalk.cyan, |
|
capitalized: chalk.yellow, |
|
jsx_tag: chalk.yellow, |
|
punctuator: chalk.yellow, |
|
number: chalk.magenta, |
|
string: chalk.green, |
|
regex: chalk.magenta, |
|
comment: chalk.grey, |
|
invalid: chalk.white.bgRed.bold |
|
}; |
|
} |
|
|
|
var NEWLINE = /\r\n|[\n\r\u2028\u2029]/; |
|
var JSX_TAG = /^[a-z][\w-]*$/i; |
|
var BRACKET = /^[()[\]{}]$/; |
|
|
|
function getTokenType(match) { |
|
var _match$slice = match.slice(-2), |
|
_match$slice2 = _slicedToArray(_match$slice, 2), |
|
offset = _match$slice2[0], |
|
text = _match$slice2[1]; |
|
|
|
var token = (0, _jsTokens().matchToToken)(match); |
|
|
|
if (token.type === "name") { |
|
if (_esutils().default.keyword.isReservedWordES6(token.value)) { |
|
return "keyword"; |
|
} |
|
|
|
if (JSX_TAG.test(token.value) && (text[offset - 1] === "<" || text.substr(offset - 2, 2) == "</")) { |
|
return "jsx_tag"; |
|
} |
|
|
|
if (token.value[0] !== token.value[0].toLowerCase()) { |
|
return "capitalized"; |
|
} |
|
} |
|
|
|
if (token.type === "punctuator" && BRACKET.test(token.value)) { |
|
return "bracket"; |
|
} |
|
|
|
if (token.type === "invalid" && (token.value === "@" || token.value === "#")) { |
|
return "punctuator"; |
|
} |
|
|
|
return token.type; |
|
} |
|
|
|
function highlightTokens(defs, text) { |
|
return text.replace(_jsTokens().default, function () { |
|
for (var _len = arguments.length, args = new Array(_len), _key = 0; _key < _len; _key++) { |
|
args[_key] = arguments[_key]; |
|
} |
|
|
|
var type = getTokenType(args); |
|
var colorize = defs[type]; |
|
|
|
if (colorize) { |
|
return args[0].split(NEWLINE).map(function (str) { |
|
return colorize(str); |
|
}).join("\n"); |
|
} else { |
|
return args[0]; |
|
} |
|
}); |
|
} |
|
|
|
function shouldHighlight(options) { |
|
return _chalk().default.supportsColor || options.forceColor; |
|
} |
|
|
|
function getChalk(options) { |
|
var chalk = _chalk().default; |
|
|
|
if (options.forceColor) { |
|
chalk = new (_chalk().default.constructor)({ |
|
enabled: true, |
|
level: 1 |
|
}); |
|
} |
|
|
|
return chalk; |
|
} |
|
|
|
function highlight(code) { |
|
var options = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {}; |
|
|
|
if (shouldHighlight(options)) { |
|
var chalk = getChalk(options); |
|
var defs = getDefs(chalk); |
|
return highlightTokens(defs, code); |
|
} else { |
|
return code; |
|
} |
|
} |
|
}); |
|
unwrapExports(lib$1); |
|
var lib_1 = lib$1.shouldHighlight; |
|
var lib_2 = lib$1.getChalk; |
|
|
|
var lib$2 = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, "__esModule", { |
|
value: true |
|
}); |
|
exports.codeFrameColumns = codeFrameColumns; |
|
exports.default = _default; |
|
|
|
function _highlight() { |
|
var data = _interopRequireWildcard(lib$1); |
|
|
|
_highlight = function _highlight() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _interopRequireWildcard(obj) { |
|
if (obj && obj.__esModule) { |
|
return obj; |
|
} else { |
|
var newObj = {}; |
|
|
|
if (obj != null) { |
|
for (var key in obj) { |
|
if (Object.prototype.hasOwnProperty.call(obj, key)) { |
|
var desc = Object.defineProperty && Object.getOwnPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : {}; |
|
|
|
if (desc.get || desc.set) { |
|
Object.defineProperty(newObj, key, desc); |
|
} else { |
|
newObj[key] = obj[key]; |
|
} |
|
} |
|
} |
|
} |
|
|
|
newObj.default = obj; |
|
return newObj; |
|
} |
|
} |
|
|
|
var deprecationWarningShown = false; |
|
|
|
function getDefs(chalk) { |
|
return { |
|
gutter: chalk.grey, |
|
marker: chalk.red.bold, |
|
message: chalk.red.bold |
|
}; |
|
} |
|
|
|
var NEWLINE = /\r\n|[\n\r\u2028\u2029]/; |
|
|
|
function getMarkerLines(loc, source, opts) { |
|
var startLoc = Object.assign({ |
|
column: 0, |
|
line: -1 |
|
}, loc.start); |
|
var endLoc = Object.assign({}, startLoc, loc.end); |
|
|
|
var _ref = opts || {}, |
|
_ref$linesAbove = _ref.linesAbove, |
|
linesAbove = _ref$linesAbove === void 0 ? 2 : _ref$linesAbove, |
|
_ref$linesBelow = _ref.linesBelow, |
|
linesBelow = _ref$linesBelow === void 0 ? 3 : _ref$linesBelow; |
|
|
|
var startLine = startLoc.line; |
|
var startColumn = startLoc.column; |
|
var endLine = endLoc.line; |
|
var endColumn = endLoc.column; |
|
var start = Math.max(startLine - (linesAbove + 1), 0); |
|
var end = Math.min(source.length, endLine + linesBelow); |
|
|
|
if (startLine === -1) { |
|
start = 0; |
|
} |
|
|
|
if (endLine === -1) { |
|
end = source.length; |
|
} |
|
|
|
var lineDiff = endLine - startLine; |
|
var markerLines = {}; |
|
|
|
if (lineDiff) { |
|
for (var i = 0; i <= lineDiff; i++) { |
|
var lineNumber = i + startLine; |
|
|
|
if (!startColumn) { |
|
markerLines[lineNumber] = true; |
|
} else if (i === 0) { |
|
var sourceLength = source[lineNumber - 1].length; |
|
markerLines[lineNumber] = [startColumn, sourceLength - startColumn + 1]; |
|
} else if (i === lineDiff) { |
|
markerLines[lineNumber] = [0, endColumn]; |
|
} else { |
|
var _sourceLength = source[lineNumber - i].length; |
|
markerLines[lineNumber] = [0, _sourceLength]; |
|
} |
|
} |
|
} else { |
|
if (startColumn === endColumn) { |
|
if (startColumn) { |
|
markerLines[startLine] = [startColumn, 0]; |
|
} else { |
|
markerLines[startLine] = true; |
|
} |
|
} else { |
|
markerLines[startLine] = [startColumn, endColumn - startColumn]; |
|
} |
|
} |
|
|
|
return { |
|
start: start, |
|
end: end, |
|
markerLines: markerLines |
|
}; |
|
} |
|
|
|
function codeFrameColumns(rawLines, loc) { |
|
var opts = arguments.length > 2 && arguments[2] !== undefined ? arguments[2] : {}; |
|
var highlighted = (opts.highlightCode || opts.forceColor) && (0, _highlight().shouldHighlight)(opts); |
|
var chalk = (0, _highlight().getChalk)(opts); |
|
var defs = getDefs(chalk); |
|
|
|
var maybeHighlight = function maybeHighlight(chalkFn, string) { |
|
return highlighted ? chalkFn(string) : string; |
|
}; |
|
|
|
var lines = rawLines.split(NEWLINE); |
|
|
|
var _getMarkerLines = getMarkerLines(loc, lines, opts), |
|
start = _getMarkerLines.start, |
|
end = _getMarkerLines.end, |
|
markerLines = _getMarkerLines.markerLines; |
|
|
|
var hasColumns = loc.start && typeof loc.start.column === "number"; |
|
var numberMaxWidth = String(end).length; |
|
var highlightedLines = highlighted ? (0, _highlight().default)(rawLines, opts) : rawLines; |
|
var frame = highlightedLines.split(NEWLINE).slice(start, end).map(function (line, index) { |
|
var number = start + 1 + index; |
|
var paddedNumber = " ".concat(number).slice(-numberMaxWidth); |
|
var gutter = " ".concat(paddedNumber, " | "); |
|
var hasMarker = markerLines[number]; |
|
var lastMarkerLine = !markerLines[number + 1]; |
|
|
|
if (hasMarker) { |
|
var markerLine = ""; |
|
|
|
if (Array.isArray(hasMarker)) { |
|
var markerSpacing = line.slice(0, Math.max(hasMarker[0] - 1, 0)).replace(/[^\t]/g, " "); |
|
var numberOfMarkers = hasMarker[1] || 1; |
|
markerLine = ["\n ", maybeHighlight(defs.gutter, gutter.replace(/\d/g, " ")), markerSpacing, maybeHighlight(defs.marker, "^").repeat(numberOfMarkers)].join(""); |
|
|
|
if (lastMarkerLine && opts.message) { |
|
markerLine += " " + maybeHighlight(defs.message, opts.message); |
|
} |
|
} |
|
|
|
return [maybeHighlight(defs.marker, ">"), maybeHighlight(defs.gutter, gutter), line, markerLine].join(""); |
|
} else { |
|
return " ".concat(maybeHighlight(defs.gutter, gutter)).concat(line); |
|
} |
|
}).join("\n"); |
|
|
|
if (opts.message && !hasColumns) { |
|
frame = "".concat(" ".repeat(numberMaxWidth + 1)).concat(opts.message, "\n").concat(frame); |
|
} |
|
|
|
if (highlighted) { |
|
return chalk.reset(frame); |
|
} else { |
|
return frame; |
|
} |
|
} |
|
|
|
function _default(rawLines, lineNumber, colNumber) { |
|
var opts = arguments.length > 3 && arguments[3] !== undefined ? arguments[3] : {}; |
|
|
|
if (!deprecationWarningShown) { |
|
deprecationWarningShown = true; |
|
var message = "Passing lineNumber and colNumber is deprecated to @babel/code-frame. Please use `codeFrameColumns`."; |
|
|
|
if (process.emitWarning) { |
|
process.emitWarning(message, "DeprecationWarning"); |
|
} else { |
|
var deprecationError = new Error(message); |
|
deprecationError.name = "DeprecationWarning"; |
|
console.warn(new Error(message)); |
|
} |
|
} |
|
|
|
colNumber = Math.max(colNumber, 0); |
|
var location = { |
|
start: { |
|
column: colNumber, |
|
line: lineNumber |
|
} |
|
}; |
|
return codeFrameColumns(rawLines, location, opts); |
|
} |
|
}); |
|
unwrapExports(lib$2); |
|
var lib_1$1 = lib$2.codeFrameColumns; |
|
|
|
var ConfigError$1 = errors.ConfigError; |
|
var locStart$1 = loc.locStart, |
|
locEnd$1 = loc.locEnd; // Use defineProperties()/getOwnPropertyDescriptor() to prevent |
|
// triggering the parsers getters. |
|
|
|
var ownNames = Object.getOwnPropertyNames; |
|
var ownDescriptor = Object.getOwnPropertyDescriptor; |
|
|
|
function getParsers(options) { |
|
var parsers = {}; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = options.plugins[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var plugin = _step.value; |
|
|
|
if (!plugin.parsers) { |
|
continue; |
|
} |
|
|
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = ownNames(plugin.parsers)[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var name = _step2.value; |
|
Object.defineProperty(parsers, name, ownDescriptor(plugin.parsers, name)); |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return parsers; |
|
} |
|
|
|
function resolveParser(opts, parsers) { |
|
parsers = parsers || getParsers(opts); |
|
|
|
if (typeof opts.parser === "function") { |
|
// Custom parser API always works with JavaScript. |
|
return { |
|
parse: opts.parser, |
|
astFormat: "estree", |
|
locStart: locStart$1, |
|
locEnd: locEnd$1 |
|
}; |
|
} |
|
|
|
if (typeof opts.parser === "string") { |
|
if (Object.prototype.hasOwnProperty.call(parsers, opts.parser)) { |
|
return parsers[opts.parser]; |
|
} |
|
/* istanbul ignore next */ |
|
|
|
|
|
{ |
|
throw new ConfigError$1("Couldn't resolve parser \"".concat(opts.parser, "\". Parsers must be explicitly added to the standalone bundle.")); |
|
} |
|
} |
|
} |
|
|
|
function parse(text, opts) { |
|
var parsers = getParsers(opts); // Create a new object {parserName: parseFn}. Uses defineProperty() to only call |
|
// the parsers getters when actually calling the parser `parse` function. |
|
|
|
var parsersForCustomParserApi = Object.keys(parsers).reduce(function (object, parserName) { |
|
return Object.defineProperty(object, parserName, { |
|
enumerable: true, |
|
get: function get() { |
|
return parsers[parserName].parse; |
|
} |
|
}); |
|
}, {}); |
|
var parser = resolveParser(opts, parsers); |
|
|
|
try { |
|
if (parser.preprocess) { |
|
text = parser.preprocess(text, opts); |
|
} |
|
|
|
return { |
|
text: text, |
|
ast: parser.parse(text, parsersForCustomParserApi, opts) |
|
}; |
|
} catch (error) { |
|
var loc = error.loc; |
|
|
|
if (loc) { |
|
var codeFrame = lib$2; |
|
error.codeFrame = codeFrame.codeFrameColumns(text, loc, { |
|
highlightCode: true |
|
}); |
|
error.message += "\n" + error.codeFrame; |
|
throw error; |
|
} |
|
/* istanbul ignore next */ |
|
|
|
|
|
throw error.stack; |
|
} |
|
} |
|
|
|
var parser = { |
|
parse: parse, |
|
resolveParser: resolveParser |
|
}; |
|
|
|
var UndefinedParserError$1 = errors.UndefinedParserError; |
|
var getSupportInfo$1 = support.getSupportInfo; |
|
var resolveParser$1 = parser.resolveParser; |
|
var hiddenDefaults = { |
|
astFormat: "estree", |
|
printer: {}, |
|
originalText: undefined, |
|
locStart: null, |
|
locEnd: null |
|
}; // Copy options and fill in default values. |
|
|
|
function normalize$1(options, opts) { |
|
opts = opts || {}; |
|
var rawOptions = Object.assign({}, options); |
|
var supportOptions = getSupportInfo$1(null, { |
|
plugins: options.plugins, |
|
showUnreleased: true, |
|
showDeprecated: true |
|
}).options; |
|
var defaults = supportOptions.reduce(function (reduced, optionInfo) { |
|
return optionInfo.default !== undefined ? Object.assign(reduced, _defineProperty({}, optionInfo.name, optionInfo.default)) : reduced; |
|
}, Object.assign({}, hiddenDefaults)); |
|
|
|
if (!rawOptions.parser) { |
|
if (!rawOptions.filepath) { |
|
var logger = opts.logger || console; |
|
logger.warn("No parser and no filepath given, using 'babel' the parser now " + "but this will throw an error in the future. " + "Please specify a parser or a filepath so one can be inferred."); |
|
rawOptions.parser = "babel"; |
|
} else { |
|
rawOptions.parser = inferParser(rawOptions.filepath, rawOptions.plugins); |
|
|
|
if (!rawOptions.parser) { |
|
throw new UndefinedParserError$1("No parser could be inferred for file: ".concat(rawOptions.filepath)); |
|
} |
|
} |
|
} |
|
|
|
var parser = resolveParser$1(optionsNormalizer.normalizeApiOptions(rawOptions, [supportOptions.find(function (x) { |
|
return x.name === "parser"; |
|
})], { |
|
passThrough: true, |
|
logger: false |
|
})); |
|
rawOptions.astFormat = parser.astFormat; |
|
rawOptions.locEnd = parser.locEnd; |
|
rawOptions.locStart = parser.locStart; |
|
var plugin = getPlugin(rawOptions); |
|
rawOptions.printer = plugin.printers[rawOptions.astFormat]; |
|
var pluginDefaults = supportOptions.filter(function (optionInfo) { |
|
return optionInfo.pluginDefaults && optionInfo.pluginDefaults[plugin.name] !== undefined; |
|
}).reduce(function (reduced, optionInfo) { |
|
return Object.assign(reduced, _defineProperty({}, optionInfo.name, optionInfo.pluginDefaults[plugin.name])); |
|
}, {}); |
|
var mixedDefaults = Object.assign({}, defaults, pluginDefaults); |
|
Object.keys(mixedDefaults).forEach(function (k) { |
|
if (rawOptions[k] == null) { |
|
rawOptions[k] = mixedDefaults[k]; |
|
} |
|
}); |
|
|
|
if (rawOptions.parser === "json") { |
|
rawOptions.trailingComma = "none"; |
|
} |
|
|
|
return optionsNormalizer.normalizeApiOptions(rawOptions, supportOptions, Object.assign({ |
|
passThrough: Object.keys(hiddenDefaults) |
|
}, opts)); |
|
} |
|
|
|
function getPlugin(options) { |
|
var astFormat = options.astFormat; |
|
|
|
if (!astFormat) { |
|
throw new Error("getPlugin() requires astFormat to be set"); |
|
} |
|
|
|
var printerPlugin = options.plugins.find(function (plugin) { |
|
return plugin.printers && plugin.printers[astFormat]; |
|
}); |
|
|
|
if (!printerPlugin) { |
|
throw new Error("Couldn't find plugin for AST format \"".concat(astFormat, "\"")); |
|
} |
|
|
|
return printerPlugin; |
|
} |
|
|
|
function getInterpreter(filepath) { |
|
if (typeof filepath !== "string") { |
|
return ""; |
|
} |
|
|
|
var fd; |
|
|
|
try { |
|
fd = fs.openSync(filepath, "r"); |
|
} catch (err) { |
|
return ""; |
|
} |
|
|
|
try { |
|
var liner = new readlines(fd); |
|
var firstLine = liner.next().toString("utf8"); // #!/bin/env node, #!/usr/bin/env node |
|
|
|
var m1 = firstLine.match(/^#!\/(?:usr\/)?bin\/env\s+(\S+)/); |
|
|
|
if (m1) { |
|
return m1[1]; |
|
} // #!/bin/node, #!/usr/bin/node, #!/usr/local/bin/node |
|
|
|
|
|
var m2 = firstLine.match(/^#!\/(?:usr\/(?:local\/)?)?bin\/(\S+)/); |
|
|
|
if (m2) { |
|
return m2[1]; |
|
} |
|
|
|
return ""; |
|
} catch (err) { |
|
// There are some weird cases where paths are missing, causing Jest |
|
// failures. It's unclear what these correspond to in the real world. |
|
return ""; |
|
} finally { |
|
try { |
|
// There are some weird cases where paths are missing, causing Jest |
|
// failures. It's unclear what these correspond to in the real world. |
|
fs.closeSync(fd); |
|
} catch (err) {// nop |
|
} |
|
} |
|
} |
|
|
|
function inferParser(filepath, plugins) { |
|
var filepathParts = normalizePath(filepath).split("/"); |
|
var filename = filepathParts[filepathParts.length - 1].toLowerCase(); // If the file has no extension, we can try to infer the language from the |
|
// interpreter in the shebang line, if any; but since this requires FS access, |
|
// do it last. |
|
|
|
var language = getSupportInfo$1(null, { |
|
plugins: plugins |
|
}).languages.find(function (language) { |
|
return language.since !== null && (language.extensions && language.extensions.some(function (extension) { |
|
return filename.endsWith(extension); |
|
}) || language.filenames && language.filenames.find(function (name) { |
|
return name.toLowerCase() === filename; |
|
}) || filename.indexOf(".") === -1 && language.interpreters && language.interpreters.indexOf(getInterpreter(filepath)) !== -1); |
|
}); |
|
return language && language.parsers[0]; |
|
} |
|
|
|
var options$1 = { |
|
normalize: normalize$1, |
|
hiddenDefaults: hiddenDefaults, |
|
inferParser: inferParser |
|
}; |
|
|
|
function massageAST(ast, options, parent) { |
|
if (Array.isArray(ast)) { |
|
return ast.map(function (e) { |
|
return massageAST(e, options, parent); |
|
}).filter(function (e) { |
|
return e; |
|
}); |
|
} |
|
|
|
if (!ast || _typeof(ast) !== "object") { |
|
return ast; |
|
} |
|
|
|
var newObj = {}; |
|
|
|
for (var _i = 0, _Object$keys = Object.keys(ast); _i < _Object$keys.length; _i++) { |
|
var key = _Object$keys[_i]; |
|
|
|
if (typeof ast[key] !== "function") { |
|
newObj[key] = massageAST(ast[key], options, ast); |
|
} |
|
} |
|
|
|
if (options.printer.massageAstNode) { |
|
var result = options.printer.massageAstNode(ast, newObj, parent); |
|
|
|
if (result === null) { |
|
return undefined; |
|
} |
|
|
|
if (result) { |
|
return result; |
|
} |
|
} |
|
|
|
return newObj; |
|
} |
|
|
|
var massageAst = massageAST; |
|
|
|
function assert() {} |
|
|
|
assert.ok = function () {}; |
|
|
|
assert.strictEqual = function () {}; |
|
|
|
/** |
|
* @param {Doc[]} parts |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function concat(parts) { |
|
// access the internals of a document directly. |
|
// if(parts.length === 1) { |
|
// // If it's a single document, no need to concat it. |
|
// return parts[0]; |
|
// } |
|
|
|
|
|
return { |
|
type: "concat", |
|
parts: parts |
|
}; |
|
} |
|
/** |
|
* @param {Doc} contents |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function indent(contents) { |
|
|
|
return { |
|
type: "indent", |
|
contents: contents |
|
}; |
|
} |
|
/** |
|
* @param {number} n |
|
* @param {Doc} contents |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function align(n, contents) { |
|
|
|
return { |
|
type: "align", |
|
contents: contents, |
|
n: n |
|
}; |
|
} |
|
/** |
|
* @param {Doc} contents |
|
* @param {object} [opts] - TBD ??? |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function group(contents, opts) { |
|
opts = opts || {}; |
|
|
|
return { |
|
type: "group", |
|
id: opts.id, |
|
contents: contents, |
|
break: !!opts.shouldBreak, |
|
expandedStates: opts.expandedStates |
|
}; |
|
} |
|
/** |
|
* @param {Doc} contents |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function dedentToRoot(contents) { |
|
return align(-Infinity, contents); |
|
} |
|
/** |
|
* @param {Doc} contents |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function markAsRoot(contents) { |
|
// @ts-ignore - TBD ???: |
|
return align({ |
|
type: "root" |
|
}, contents); |
|
} |
|
/** |
|
* @param {Doc} contents |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function dedent(contents) { |
|
return align(-1, contents); |
|
} |
|
/** |
|
* @param {Doc[]} states |
|
* @param {object} [opts] - TBD ??? |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function conditionalGroup(states, opts) { |
|
return group(states[0], Object.assign(opts || {}, { |
|
expandedStates: states |
|
})); |
|
} |
|
/** |
|
* @param {Doc[]} parts |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function fill(parts) { |
|
|
|
return { |
|
type: "fill", |
|
parts: parts |
|
}; |
|
} |
|
/** |
|
* @param {Doc} [breakContents] |
|
* @param {Doc} [flatContents] |
|
* @param {object} [opts] - TBD ??? |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function ifBreak(breakContents, flatContents, opts) { |
|
opts = opts || {}; |
|
|
|
return { |
|
type: "if-break", |
|
breakContents: breakContents, |
|
flatContents: flatContents, |
|
groupId: opts.groupId |
|
}; |
|
} |
|
/** |
|
* @param {Doc} contents |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function lineSuffix(contents) { |
|
|
|
return { |
|
type: "line-suffix", |
|
contents: contents |
|
}; |
|
} |
|
|
|
var lineSuffixBoundary = { |
|
type: "line-suffix-boundary" |
|
}; |
|
var breakParent = { |
|
type: "break-parent" |
|
}; |
|
var trim = { |
|
type: "trim" |
|
}; |
|
var line = { |
|
type: "line" |
|
}; |
|
var softline = { |
|
type: "line", |
|
soft: true |
|
}; |
|
var hardline = concat([{ |
|
type: "line", |
|
hard: true |
|
}, breakParent]); |
|
var literalline = concat([{ |
|
type: "line", |
|
hard: true, |
|
literal: true |
|
}, breakParent]); |
|
var cursor = { |
|
type: "cursor", |
|
placeholder: Symbol("cursor") |
|
}; |
|
/** |
|
* @param {Doc} sep |
|
* @param {Doc[]} arr |
|
* @returns Doc |
|
*/ |
|
|
|
function join(sep, arr) { |
|
var res = []; |
|
|
|
for (var i = 0; i < arr.length; i++) { |
|
if (i !== 0) { |
|
res.push(sep); |
|
} |
|
|
|
res.push(arr[i]); |
|
} |
|
|
|
return concat(res); |
|
} |
|
/** |
|
* @param {Doc} doc |
|
* @param {number} size |
|
* @param {number} tabWidth |
|
*/ |
|
|
|
|
|
function addAlignmentToDoc(doc, size, tabWidth) { |
|
var aligned = doc; |
|
|
|
if (size > 0) { |
|
// Use indent to add tabs for all the levels of tabs we need |
|
for (var i = 0; i < Math.floor(size / tabWidth); ++i) { |
|
aligned = indent(aligned); |
|
} // Use align for all the spaces that are needed |
|
|
|
|
|
aligned = align(size % tabWidth, aligned); // size is absolute from 0 and not relative to the current |
|
// indentation, so we use -Infinity to reset the indentation to 0 |
|
|
|
aligned = align(-Infinity, aligned); |
|
} |
|
|
|
return aligned; |
|
} |
|
|
|
var docBuilders = { |
|
concat: concat, |
|
join: join, |
|
line: line, |
|
softline: softline, |
|
hardline: hardline, |
|
literalline: literalline, |
|
group: group, |
|
conditionalGroup: conditionalGroup, |
|
fill: fill, |
|
lineSuffix: lineSuffix, |
|
lineSuffixBoundary: lineSuffixBoundary, |
|
cursor: cursor, |
|
breakParent: breakParent, |
|
ifBreak: ifBreak, |
|
trim: trim, |
|
indent: indent, |
|
align: align, |
|
addAlignmentToDoc: addAlignmentToDoc, |
|
markAsRoot: markAsRoot, |
|
dedentToRoot: dedentToRoot, |
|
dedent: dedent |
|
}; |
|
|
|
var ansiRegex = function ansiRegex(options) { |
|
options = Object.assign({ |
|
onlyFirst: false |
|
}, options); |
|
var pattern = ["[\\u001B\\u009B][[\\]()#;?]*(?:(?:(?:[a-zA-Z\\d]*(?:;[-a-zA-Z\\d\\/#&.:=?%@~_]*)*)?\\u0007)", '(?:(?:\\d{1,4}(?:;\\d{0,4})*)?[\\dA-PR-TZcf-ntqry=><~]))'].join('|'); |
|
return new RegExp(pattern, options.onlyFirst ? undefined : 'g'); |
|
}; |
|
|
|
var stripAnsi = function stripAnsi(string) { |
|
return typeof string === 'string' ? string.replace(ansiRegex(), '') : string; |
|
}; |
|
|
|
var stripAnsi_1 = stripAnsi; |
|
var default_1$1 = stripAnsi; |
|
stripAnsi_1.default = default_1$1; |
|
|
|
/* eslint-disable yoda */ |
|
|
|
var isFullwidthCodePoint = function isFullwidthCodePoint(codePoint) { |
|
if (Number.isNaN(codePoint)) { |
|
return false; |
|
} // Code points are derived from: |
|
// http://www.unix.org/Public/UNIDATA/EastAsianWidth.txt |
|
|
|
|
|
if (codePoint >= 0x1100 && (codePoint <= 0x115F || // Hangul Jamo |
|
codePoint === 0x2329 || // LEFT-POINTING ANGLE BRACKET |
|
codePoint === 0x232A || // RIGHT-POINTING ANGLE BRACKET |
|
// CJK Radicals Supplement .. Enclosed CJK Letters and Months |
|
0x2E80 <= codePoint && codePoint <= 0x3247 && codePoint !== 0x303F || // Enclosed CJK Letters and Months .. CJK Unified Ideographs Extension A |
|
0x3250 <= codePoint && codePoint <= 0x4DBF || // CJK Unified Ideographs .. Yi Radicals |
|
0x4E00 <= codePoint && codePoint <= 0xA4C6 || // Hangul Jamo Extended-A |
|
0xA960 <= codePoint && codePoint <= 0xA97C || // Hangul Syllables |
|
0xAC00 <= codePoint && codePoint <= 0xD7A3 || // CJK Compatibility Ideographs |
|
0xF900 <= codePoint && codePoint <= 0xFAFF || // Vertical Forms |
|
0xFE10 <= codePoint && codePoint <= 0xFE19 || // CJK Compatibility Forms .. Small Form Variants |
|
0xFE30 <= codePoint && codePoint <= 0xFE6B || // Halfwidth and Fullwidth Forms |
|
0xFF01 <= codePoint && codePoint <= 0xFF60 || 0xFFE0 <= codePoint && codePoint <= 0xFFE6 || // Kana Supplement |
|
0x1B000 <= codePoint && codePoint <= 0x1B001 || // Enclosed Ideographic Supplement |
|
0x1F200 <= codePoint && codePoint <= 0x1F251 || // CJK Unified Ideographs Extension B .. Tertiary Ideographic Plane |
|
0x20000 <= codePoint && codePoint <= 0x3FFFD)) { |
|
return true; |
|
} |
|
|
|
return false; |
|
}; |
|
|
|
var isFullwidthCodePoint_1 = isFullwidthCodePoint; |
|
var default_1$2 = isFullwidthCodePoint; |
|
isFullwidthCodePoint_1.default = default_1$2; |
|
|
|
var emojiRegex = function emojiRegex() { |
|
// https://mths.be/emoji |
|
return /\uD83C\uDFF4\uDB40\uDC67\uDB40\uDC62(?:\uDB40\uDC65\uDB40\uDC6E\uDB40\uDC67|\uDB40\uDC73\uDB40\uDC63\uDB40\uDC74|\uDB40\uDC77\uDB40\uDC6C\uDB40\uDC73)\uDB40\uDC7F|\uD83D\uDC68(?:\uD83C\uDFFC\u200D(?:\uD83E\uDD1D\u200D\uD83D\uDC68\uD83C\uDFFB|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\uD83C\uDFFF\u200D(?:\uD83E\uDD1D\u200D\uD83D\uDC68(?:\uD83C[\uDFFB-\uDFFE])|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\uD83C\uDFFE\u200D(?:\uD83E\uDD1D\u200D\uD83D\uDC68(?:\uD83C[\uDFFB-\uDFFD])|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\uD83C\uDFFD\u200D(?:\uD83E\uDD1D\u200D\uD83D\uDC68(?:\uD83C[\uDFFB\uDFFC])|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\u200D(?:\u2764\uFE0F\u200D(?:\uD83D\uDC8B\u200D)?\uD83D\uDC68|(?:\uD83D[\uDC68\uDC69])\u200D(?:\uD83D\uDC66\u200D\uD83D\uDC66|\uD83D\uDC67\u200D(?:\uD83D[\uDC66\uDC67]))|\uD83D\uDC66\u200D\uD83D\uDC66|\uD83D\uDC67\u200D(?:\uD83D[\uDC66\uDC67])|(?:\uD83D[\uDC68\uDC69])\u200D(?:\uD83D[\uDC66\uDC67])|[\u2695\u2696\u2708]\uFE0F|\uD83D[\uDC66\uDC67]|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|(?:\uD83C\uDFFB\u200D[\u2695\u2696\u2708]|\uD83C\uDFFF\u200D[\u2695\u2696\u2708]|\uD83C\uDFFE\u200D[\u2695\u2696\u2708]|\uD83C\uDFFD\u200D[\u2695\u2696\u2708]|\uD83C\uDFFC\u200D[\u2695\u2696\u2708])\uFE0F|\uD83C\uDFFB\u200D(?:\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\uD83C[\uDFFB-\uDFFF])|(?:\uD83E\uDDD1\uD83C\uDFFB\u200D\uD83E\uDD1D\u200D\uD83E\uDDD1|\uD83D\uDC69\uD83C\uDFFC\u200D\uD83E\uDD1D\u200D\uD83D\uDC69)\uD83C\uDFFB|\uD83E\uDDD1(?:\uD83C\uDFFF\u200D\uD83E\uDD1D\u200D\uD83E\uDDD1(?:\uD83C[\uDFFB-\uDFFF])|\u200D\uD83E\uDD1D\u200D\uD83E\uDDD1)|(?:\uD83E\uDDD1\uD83C\uDFFE\u200D\uD83E\uDD1D\u200D\uD83E\uDDD1|\uD83D\uDC69\uD83C\uDFFF\u200D\uD83E\uDD1D\u200D(?:\uD83D[\uDC68\uDC69]))(?:\uD83C[\uDFFB-\uDFFE])|(?:\uD83E\uDDD1\uD83C\uDFFC\u200D\uD83E\uDD1D\u200D\uD83E\uDDD1|\uD83D\uDC69\uD83C\uDFFD\u200D\uD83E\uDD1D\u200D\uD83D\uDC69)(?:\uD83C[\uDFFB\uDFFC])|\uD83D\uDC69(?:\uD83C\uDFFE\u200D(?:\uD83E\uDD1D\u200D\uD83D\uDC68(?:\uD83C[\uDFFB-\uDFFD\uDFFF])|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\uD83C\uDFFC\u200D(?:\uD83E\uDD1D\u200D\uD83D\uDC68(?:\uD83C[\uDFFB\uDFFD-\uDFFF])|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\uD83C\uDFFB\u200D(?:\uD83E\uDD1D\u200D\uD83D\uDC68(?:\uD83C[\uDFFC-\uDFFF])|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\uD83C\uDFFD\u200D(?:\uD83E\uDD1D\u200D\uD83D\uDC68(?:\uD83C[\uDFFB\uDFFC\uDFFE\uDFFF])|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\u200D(?:\u2764\uFE0F\u200D(?:\uD83D\uDC8B\u200D(?:\uD83D[\uDC68\uDC69])|\uD83D[\uDC68\uDC69])|\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD])|\uD83C\uDFFF\u200D(?:\uD83C[\uDF3E\uDF73\uDF93\uDFA4\uDFA8\uDFEB\uDFED]|\uD83D[\uDCBB\uDCBC\uDD27\uDD2C\uDE80\uDE92]|\uD83E[\uDDAF-\uDDB3\uDDBC\uDDBD]))|\uD83D\uDC69\u200D\uD83D\uDC69\u200D(?:\uD83D\uDC66\u200D\uD83D\uDC66|\uD83D\uDC67\u200D(?:\uD83D[\uDC66\uDC67]))|(?:\uD83E\uDDD1\uD83C\uDFFD\u200D\uD83E\uDD1D\u200D\uD83E\uDDD1|\uD83D\uDC69\uD83C\uDFFE\u200D\uD83E\uDD1D\u200D\uD83D\uDC69)(?:\uD83C[\uDFFB-\uDFFD])|\uD83D\uDC69\u200D\uD83D\uDC66\u200D\uD83D\uDC66|\uD83D\uDC69\u200D\uD83D\uDC69\u200D(?:\uD83D[\uDC66\uDC67])|(?:\uD83D\uDC41\uFE0F\u200D\uD83D\uDDE8|\uD83D\uDC69(?:\uD83C\uDFFF\u200D[\u2695\u2696\u2708]|\uD83C\uDFFE\u200D[\u2695\u2696\u2708]|\uD83C\uDFFC\u200D[\u2695\u2696\u2708]|\uD83C\uDFFB\u200D[\u2695\u2696\u2708]|\uD83C\uDFFD\u200D[\u2695\u2696\u2708]|\u200D[\u2695\u2696\u2708])|(?:(?:\u26F9|\uD83C[\uDFCB\uDFCC]|\uD83D\uDD75)\uFE0F|\uD83D\uDC6F|\uD83E[\uDD3C\uDDDE\uDDDF])\u200D[\u2640\u2642]|(?:\u26F9|\uD83C[\uDFCB\uDFCC]|\uD83D\uDD75)(?:\uD83C[\uDFFB-\uDFFF])\u200D[\u2640\u2642]|(?:\uD83C[\uDFC3\uDFC4\uDFCA]|\uD83D[\uDC6E\uDC71\uDC73\uDC77\uDC81\uDC82\uDC86\uDC87\uDE45-\uDE47\uDE4B\uDE4D\uDE4E\uDEA3\uDEB4-\uDEB6]|\uD83E[\uDD26\uDD37-\uDD39\uDD3D\uDD3E\uDDB8\uDDB9\uDDCD-\uDDCF\uDDD6-\uDDDD])(?:(?:\uD83C[\uDFFB-\uDFFF])\u200D[\u2640\u2642]|\u200D[\u2640\u2642])|\uD83C\uDFF4\u200D\u2620)\uFE0F|\uD83D\uDC69\u200D\uD83D\uDC67\u200D(?:\uD83D[\uDC66\uDC67])|\uD83C\uDFF3\uFE0F\u200D\uD83C\uDF08|\uD83D\uDC15\u200D\uD83E\uDDBA|\uD83D\uDC69\u200D\uD83D\uDC66|\uD83D\uDC69\u200D\uD83D\uDC67|\uD83C\uDDFD\uD83C\uDDF0|\uD83C\uDDF4\uD83C\uDDF2|\uD83C\uDDF6\uD83C\uDDE6|[#\*0-9]\uFE0F\u20E3|\uD83C\uDDE7(?:\uD83C[\uDDE6\uDDE7\uDDE9-\uDDEF\uDDF1-\uDDF4\uDDF6-\uDDF9\uDDFB\uDDFC\uDDFE\uDDFF])|\uD83C\uDDF9(?:\uD83C[\uDDE6\uDDE8\uDDE9\uDDEB-\uDDED\uDDEF-\uDDF4\uDDF7\uDDF9\uDDFB\uDDFC\uDDFF])|\uD83C\uDDEA(?:\uD83C[\uDDE6\uDDE8\uDDEA\uDDEC\uDDED\uDDF7-\uDDFA])|\uD83E\uDDD1(?:\uD83C[\uDFFB-\uDFFF])|\uD83C\uDDF7(?:\uD83C[\uDDEA\uDDF4\uDDF8\uDDFA\uDDFC])|\uD83D\uDC69(?:\uD83C[\uDFFB-\uDFFF])|\uD83C\uDDF2(?:\uD83C[\uDDE6\uDDE8-\uDDED\uDDF0-\uDDFF])|\uD83C\uDDE6(?:\uD83C[\uDDE8-\uDDEC\uDDEE\uDDF1\uDDF2\uDDF4\uDDF6-\uDDFA\uDDFC\uDDFD\uDDFF])|\uD83C\uDDF0(?:\uD83C[\uDDEA\uDDEC-\uDDEE\uDDF2\uDDF3\uDDF5\uDDF7\uDDFC\uDDFE\uDDFF])|\uD83C\uDDED(?:\uD83C[\uDDF0\uDDF2\uDDF3\uDDF7\uDDF9\uDDFA])|\uD83C\uDDE9(?:\uD83C[\uDDEA\uDDEC\uDDEF\uDDF0\uDDF2\uDDF4\uDDFF])|\uD83C\uDDFE(?:\uD83C[\uDDEA\uDDF9])|\uD83C\uDDEC(?:\uD83C[\uDDE6\uDDE7\uDDE9-\uDDEE\uDDF1-\uDDF3\uDDF5-\uDDFA\uDDFC\uDDFE])|\uD83C\uDDF8(?:\uD83C[\uDDE6-\uDDEA\uDDEC-\uDDF4\uDDF7-\uDDF9\uDDFB\uDDFD-\uDDFF])|\uD83C\uDDEB(?:\uD83C[\uDDEE-\uDDF0\uDDF2\uDDF4\uDDF7])|\uD83C\uDDF5(?:\uD83C[\uDDE6\uDDEA-\uDDED\uDDF0-\uDDF3\uDDF7-\uDDF9\uDDFC\uDDFE])|\uD83C\uDDFB(?:\uD83C[\uDDE6\uDDE8\uDDEA\uDDEC\uDDEE\uDDF3\uDDFA])|\uD83C\uDDF3(?:\uD83C[\uDDE6\uDDE8\uDDEA-\uDDEC\uDDEE\uDDF1\uDDF4\uDDF5\uDDF7\uDDFA\uDDFF])|\uD83C\uDDE8(?:\uD83C[\uDDE6\uDDE8\uDDE9\uDDEB-\uDDEE\uDDF0-\uDDF5\uDDF7\uDDFA-\uDDFF])|\uD83C\uDDF1(?:\uD83C[\uDDE6-\uDDE8\uDDEE\uDDF0\uDDF7-\uDDFB\uDDFE])|\uD83C\uDDFF(?:\uD83C[\uDDE6\uDDF2\uDDFC])|\uD83C\uDDFC(?:\uD83C[\uDDEB\uDDF8])|\uD83C\uDDFA(?:\uD83C[\uDDE6\uDDEC\uDDF2\uDDF3\uDDF8\uDDFE\uDDFF])|\uD83C\uDDEE(?:\uD83C[\uDDE8-\uDDEA\uDDF1-\uDDF4\uDDF6-\uDDF9])|\uD83C\uDDEF(?:\uD83C[\uDDEA\uDDF2\uDDF4\uDDF5])|(?:\uD83C[\uDFC3\uDFC4\uDFCA]|\uD83D[\uDC6E\uDC71\uDC73\uDC77\uDC81\uDC82\uDC86\uDC87\uDE45-\uDE47\uDE4B\uDE4D\uDE4E\uDEA3\uDEB4-\uDEB6]|\uD83E[\uDD26\uDD37-\uDD39\uDD3D\uDD3E\uDDB8\uDDB9\uDDCD-\uDDCF\uDDD6-\uDDDD])(?:\uD83C[\uDFFB-\uDFFF])|(?:\u26F9|\uD83C[\uDFCB\uDFCC]|\uD83D\uDD75)(?:\uD83C[\uDFFB-\uDFFF])|(?:[\u261D\u270A-\u270D]|\uD83C[\uDF85\uDFC2\uDFC7]|\uD83D[\uDC42\uDC43\uDC46-\uDC50\uDC66\uDC67\uDC6B-\uDC6D\uDC70\uDC72\uDC74-\uDC76\uDC78\uDC7C\uDC83\uDC85\uDCAA\uDD74\uDD7A\uDD90\uDD95\uDD96\uDE4C\uDE4F\uDEC0\uDECC]|\uD83E[\uDD0F\uDD18-\uDD1C\uDD1E\uDD1F\uDD30-\uDD36\uDDB5\uDDB6\uDDBB\uDDD2-\uDDD5])(?:\uD83C[\uDFFB-\uDFFF])|(?:[\u231A\u231B\u23E9-\u23EC\u23F0\u23F3\u25FD\u25FE\u2614\u2615\u2648-\u2653\u267F\u2693\u26A1\u26AA\u26AB\u26BD\u26BE\u26C4\u26C5\u26CE\u26D4\u26EA\u26F2\u26F3\u26F5\u26FA\u26FD\u2705\u270A\u270B\u2728\u274C\u274E\u2753-\u2755\u2757\u2795-\u2797\u27B0\u27BF\u2B1B\u2B1C\u2B50\u2B55]|\uD83C[\uDC04\uDCCF\uDD8E\uDD91-\uDD9A\uDDE6-\uDDFF\uDE01\uDE1A\uDE2F\uDE32-\uDE36\uDE38-\uDE3A\uDE50\uDE51\uDF00-\uDF20\uDF2D-\uDF35\uDF37-\uDF7C\uDF7E-\uDF93\uDFA0-\uDFCA\uDFCF-\uDFD3\uDFE0-\uDFF0\uDFF4\uDFF8-\uDFFF]|\uD83D[\uDC00-\uDC3E\uDC40\uDC42-\uDCFC\uDCFF-\uDD3D\uDD4B-\uDD4E\uDD50-\uDD67\uDD7A\uDD95\uDD96\uDDA4\uDDFB-\uDE4F\uDE80-\uDEC5\uDECC\uDED0-\uDED2\uDED5\uDEEB\uDEEC\uDEF4-\uDEFA\uDFE0-\uDFEB]|\uD83E[\uDD0D-\uDD3A\uDD3C-\uDD45\uDD47-\uDD71\uDD73-\uDD76\uDD7A-\uDDA2\uDDA5-\uDDAA\uDDAE-\uDDCA\uDDCD-\uDDFF\uDE70-\uDE73\uDE78-\uDE7A\uDE80-\uDE82\uDE90-\uDE95])|(?:[#\*0-9\xA9\xAE\u203C\u2049\u2122\u2139\u2194-\u2199\u21A9\u21AA\u231A\u231B\u2328\u23CF\u23E9-\u23F3\u23F8-\u23FA\u24C2\u25AA\u25AB\u25B6\u25C0\u25FB-\u25FE\u2600-\u2604\u260E\u2611\u2614\u2615\u2618\u261D\u2620\u2622\u2623\u2626\u262A\u262E\u262F\u2638-\u263A\u2640\u2642\u2648-\u2653\u265F\u2660\u2663\u2665\u2666\u2668\u267B\u267E\u267F\u2692-\u2697\u2699\u269B\u269C\u26A0\u26A1\u26AA\u26AB\u26B0\u26B1\u26BD\u26BE\u26C4\u26C5\u26C8\u26CE\u26CF\u26D1\u26D3\u26D4\u26E9\u26EA\u26F0-\u26F5\u26F7-\u26FA\u26FD\u2702\u2705\u2708-\u270D\u270F\u2712\u2714\u2716\u271D\u2721\u2728\u2733\u2734\u2744\u2747\u274C\u274E\u2753-\u2755\u2757\u2763\u2764\u2795-\u2797\u27A1\u27B0\u27BF\u2934\u2935\u2B05-\u2B07\u2B1B\u2B1C\u2B50\u2B55\u3030\u303D\u3297\u3299]|\uD83C[\uDC04\uDCCF\uDD70\uDD71\uDD7E\uDD7F\uDD8E\uDD91-\uDD9A\uDDE6-\uDDFF\uDE01\uDE02\uDE1A\uDE2F\uDE32-\uDE3A\uDE50\uDE51\uDF00-\uDF21\uDF24-\uDF93\uDF96\uDF97\uDF99-\uDF9B\uDF9E-\uDFF0\uDFF3-\uDFF5\uDFF7-\uDFFF]|\uD83D[\uDC00-\uDCFD\uDCFF-\uDD3D\uDD49-\uDD4E\uDD50-\uDD67\uDD6F\uDD70\uDD73-\uDD7A\uDD87\uDD8A-\uDD8D\uDD90\uDD95\uDD96\uDDA4\uDDA5\uDDA8\uDDB1\uDDB2\uDDBC\uDDC2-\uDDC4\uDDD1-\uDDD3\uDDDC-\uDDDE\uDDE1\uDDE3\uDDE8\uDDEF\uDDF3\uDDFA-\uDE4F\uDE80-\uDEC5\uDECB-\uDED2\uDED5\uDEE0-\uDEE5\uDEE9\uDEEB\uDEEC\uDEF0\uDEF3-\uDEFA\uDFE0-\uDFEB]|\uD83E[\uDD0D-\uDD3A\uDD3C-\uDD45\uDD47-\uDD71\uDD73-\uDD76\uDD7A-\uDDA2\uDDA5-\uDDAA\uDDAE-\uDDCA\uDDCD-\uDDFF\uDE70-\uDE73\uDE78-\uDE7A\uDE80-\uDE82\uDE90-\uDE95])\uFE0F|(?:[\u261D\u26F9\u270A-\u270D]|\uD83C[\uDF85\uDFC2-\uDFC4\uDFC7\uDFCA-\uDFCC]|\uD83D[\uDC42\uDC43\uDC46-\uDC50\uDC66-\uDC78\uDC7C\uDC81-\uDC83\uDC85-\uDC87\uDC8F\uDC91\uDCAA\uDD74\uDD75\uDD7A\uDD90\uDD95\uDD96\uDE45-\uDE47\uDE4B-\uDE4F\uDEA3\uDEB4-\uDEB6\uDEC0\uDECC]|\uD83E[\uDD0F\uDD18-\uDD1F\uDD26\uDD30-\uDD39\uDD3C-\uDD3E\uDDB5\uDDB6\uDDB8\uDDB9\uDDBB\uDDCD-\uDDCF\uDDD1-\uDDDD])/g; |
|
}; |
|
|
|
var stringWidth = function stringWidth(string) { |
|
string = string.replace(emojiRegex(), ' '); |
|
|
|
if (typeof string !== 'string' || string.length === 0) { |
|
return 0; |
|
} |
|
|
|
string = stripAnsi_1(string); |
|
var width = 0; |
|
|
|
for (var i = 0; i < string.length; i++) { |
|
var code = string.codePointAt(i); // Ignore control characters |
|
|
|
if (code <= 0x1F || code >= 0x7F && code <= 0x9F) { |
|
continue; |
|
} // Ignore combining characters |
|
|
|
|
|
if (code >= 0x300 && code <= 0x36F) { |
|
continue; |
|
} // Surrogates |
|
|
|
|
|
if (code > 0xFFFF) { |
|
i++; |
|
} |
|
|
|
width += isFullwidthCodePoint_1(code) ? 2 : 1; |
|
} |
|
|
|
return width; |
|
}; |
|
|
|
var stringWidth_1 = stringWidth; // TODO: remove this in the next major version |
|
|
|
var default_1$3 = stringWidth; |
|
stringWidth_1.default = default_1$3; |
|
|
|
var notAsciiRegex = /[^\x20-\x7F]/; |
|
|
|
function isExportDeclaration(node) { |
|
if (node) { |
|
switch (node.type) { |
|
case "ExportDefaultDeclaration": |
|
case "ExportDefaultSpecifier": |
|
case "DeclareExportDeclaration": |
|
case "ExportNamedDeclaration": |
|
case "ExportAllDeclaration": |
|
return true; |
|
} |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function getParentExportDeclaration(path) { |
|
var parentNode = path.getParentNode(); |
|
|
|
if (path.getName() === "declaration" && isExportDeclaration(parentNode)) { |
|
return parentNode; |
|
} |
|
|
|
return null; |
|
} |
|
|
|
function getPenultimate(arr) { |
|
if (arr.length > 1) { |
|
return arr[arr.length - 2]; |
|
} |
|
|
|
return null; |
|
} |
|
/** |
|
* @typedef {{backwards?: boolean}} SkipOptions |
|
*/ |
|
|
|
/** |
|
* @param {string | RegExp} chars |
|
* @returns {(text: string, index: number | false, opts?: SkipOptions) => number | false} |
|
*/ |
|
|
|
|
|
function skip(chars) { |
|
return function (text, index, opts) { |
|
var backwards = opts && opts.backwards; // Allow `skip` functions to be threaded together without having |
|
// to check for failures (did someone say monads?). |
|
|
|
if (index === false) { |
|
return false; |
|
} |
|
|
|
var length = text.length; |
|
var cursor = index; |
|
|
|
while (cursor >= 0 && cursor < length) { |
|
var c = text.charAt(cursor); |
|
|
|
if (chars instanceof RegExp) { |
|
if (!chars.test(c)) { |
|
return cursor; |
|
} |
|
} else if (chars.indexOf(c) === -1) { |
|
return cursor; |
|
} |
|
|
|
backwards ? cursor-- : cursor++; |
|
} |
|
|
|
if (cursor === -1 || cursor === length) { |
|
// If we reached the beginning or end of the file, return the |
|
// out-of-bounds cursor. It's up to the caller to handle this |
|
// correctly. We don't want to indicate `false` though if it |
|
// actually skipped valid characters. |
|
return cursor; |
|
} |
|
|
|
return false; |
|
}; |
|
} |
|
/** |
|
* @type {(text: string, index: number | false, opts?: SkipOptions) => number | false} |
|
*/ |
|
|
|
|
|
var skipWhitespace = skip(/\s/); |
|
/** |
|
* @type {(text: string, index: number | false, opts?: SkipOptions) => number | false} |
|
*/ |
|
|
|
var skipSpaces = skip(" \t"); |
|
/** |
|
* @type {(text: string, index: number | false, opts?: SkipOptions) => number | false} |
|
*/ |
|
|
|
var skipToLineEnd = skip(",; \t"); |
|
/** |
|
* @type {(text: string, index: number | false, opts?: SkipOptions) => number | false} |
|
*/ |
|
|
|
var skipEverythingButNewLine = skip(/[^\r\n]/); |
|
/** |
|
* @param {string} text |
|
* @param {number | false} index |
|
* @returns {number | false} |
|
*/ |
|
|
|
function skipInlineComment(text, index) { |
|
if (index === false) { |
|
return false; |
|
} |
|
|
|
if (text.charAt(index) === "/" && text.charAt(index + 1) === "*") { |
|
for (var i = index + 2; i < text.length; ++i) { |
|
if (text.charAt(i) === "*" && text.charAt(i + 1) === "/") { |
|
return i + 2; |
|
} |
|
} |
|
} |
|
|
|
return index; |
|
} |
|
/** |
|
* @param {string} text |
|
* @param {number | false} index |
|
* @returns {number | false} |
|
*/ |
|
|
|
|
|
function skipTrailingComment(text, index) { |
|
if (index === false) { |
|
return false; |
|
} |
|
|
|
if (text.charAt(index) === "/" && text.charAt(index + 1) === "/") { |
|
return skipEverythingButNewLine(text, index); |
|
} |
|
|
|
return index; |
|
} // This one doesn't use the above helper function because it wants to |
|
// test \r\n in order and `skip` doesn't support ordering and we only |
|
// want to skip one newline. It's simple to implement. |
|
|
|
/** |
|
* @param {string} text |
|
* @param {number | false} index |
|
* @param {SkipOptions=} opts |
|
* @returns {number | false} |
|
*/ |
|
|
|
|
|
function skipNewline(text, index, opts) { |
|
var backwards = opts && opts.backwards; |
|
|
|
if (index === false) { |
|
return false; |
|
} |
|
|
|
var atIndex = text.charAt(index); |
|
|
|
if (backwards) { |
|
if (text.charAt(index - 1) === "\r" && atIndex === "\n") { |
|
return index - 2; |
|
} |
|
|
|
if (atIndex === "\n" || atIndex === "\r" || atIndex === "\u2028" || atIndex === "\u2029") { |
|
return index - 1; |
|
} |
|
} else { |
|
if (atIndex === "\r" && text.charAt(index + 1) === "\n") { |
|
return index + 2; |
|
} |
|
|
|
if (atIndex === "\n" || atIndex === "\r" || atIndex === "\u2028" || atIndex === "\u2029") { |
|
return index + 1; |
|
} |
|
} |
|
|
|
return index; |
|
} |
|
/** |
|
* @param {string} text |
|
* @param {number} index |
|
* @param {SkipOptions=} opts |
|
* @returns {boolean} |
|
*/ |
|
|
|
|
|
function hasNewline(text, index, opts) { |
|
opts = opts || {}; |
|
var idx = skipSpaces(text, opts.backwards ? index - 1 : index, opts); |
|
var idx2 = skipNewline(text, idx, opts); |
|
return idx !== idx2; |
|
} |
|
/** |
|
* @param {string} text |
|
* @param {number} start |
|
* @param {number} end |
|
* @returns {boolean} |
|
*/ |
|
|
|
|
|
function hasNewlineInRange(text, start, end) { |
|
for (var i = start; i < end; ++i) { |
|
if (text.charAt(i) === "\n") { |
|
return true; |
|
} |
|
} |
|
|
|
return false; |
|
} // Note: this function doesn't ignore leading comments unlike isNextLineEmpty |
|
|
|
/** |
|
* @template N |
|
* @param {string} text |
|
* @param {N} node |
|
* @param {(node: N) => number} locStart |
|
*/ |
|
|
|
|
|
function isPreviousLineEmpty(text, node, locStart) { |
|
/** @type {number | false} */ |
|
var idx = locStart(node) - 1; |
|
idx = skipSpaces(text, idx, { |
|
backwards: true |
|
}); |
|
idx = skipNewline(text, idx, { |
|
backwards: true |
|
}); |
|
idx = skipSpaces(text, idx, { |
|
backwards: true |
|
}); |
|
var idx2 = skipNewline(text, idx, { |
|
backwards: true |
|
}); |
|
return idx !== idx2; |
|
} |
|
/** |
|
* @param {string} text |
|
* @param {number} index |
|
* @returns {boolean} |
|
*/ |
|
|
|
|
|
function isNextLineEmptyAfterIndex(text, index) { |
|
/** @type {number | false} */ |
|
var oldIdx = null; |
|
/** @type {number | false} */ |
|
|
|
var idx = index; |
|
|
|
while (idx !== oldIdx) { |
|
// We need to skip all the potential trailing inline comments |
|
oldIdx = idx; |
|
idx = skipToLineEnd(text, idx); |
|
idx = skipInlineComment(text, idx); |
|
idx = skipSpaces(text, idx); |
|
} |
|
|
|
idx = skipTrailingComment(text, idx); |
|
idx = skipNewline(text, idx); |
|
return idx !== false && hasNewline(text, idx); |
|
} |
|
/** |
|
* @template N |
|
* @param {string} text |
|
* @param {N} node |
|
* @param {(node: N) => number} locEnd |
|
* @returns {boolean} |
|
*/ |
|
|
|
|
|
function isNextLineEmpty(text, node, locEnd) { |
|
return isNextLineEmptyAfterIndex(text, locEnd(node)); |
|
} |
|
/** |
|
* @param {string} text |
|
* @param {number} idx |
|
* @returns {number | false} |
|
*/ |
|
|
|
|
|
function getNextNonSpaceNonCommentCharacterIndexWithStartIndex(text, idx) { |
|
/** @type {number | false} */ |
|
var oldIdx = null; |
|
/** @type {number | false} */ |
|
|
|
var nextIdx = idx; |
|
|
|
while (nextIdx !== oldIdx) { |
|
oldIdx = nextIdx; |
|
nextIdx = skipSpaces(text, nextIdx); |
|
nextIdx = skipInlineComment(text, nextIdx); |
|
nextIdx = skipTrailingComment(text, nextIdx); |
|
nextIdx = skipNewline(text, nextIdx); |
|
} |
|
|
|
return nextIdx; |
|
} |
|
/** |
|
* @template N |
|
* @param {string} text |
|
* @param {N} node |
|
* @param {(node: N) => number} locEnd |
|
* @returns {number | false} |
|
*/ |
|
|
|
|
|
function getNextNonSpaceNonCommentCharacterIndex(text, node, locEnd) { |
|
return getNextNonSpaceNonCommentCharacterIndexWithStartIndex(text, locEnd(node)); |
|
} |
|
/** |
|
* @template N |
|
* @param {string} text |
|
* @param {N} node |
|
* @param {(node: N) => number} locEnd |
|
* @returns {string} |
|
*/ |
|
|
|
|
|
function getNextNonSpaceNonCommentCharacter(text, node, locEnd) { |
|
return text.charAt( // @ts-ignore => TBD: can return false, should we define a fallback? |
|
getNextNonSpaceNonCommentCharacterIndex(text, node, locEnd)); |
|
} |
|
/** |
|
* @param {string} text |
|
* @param {number} index |
|
* @param {SkipOptions=} opts |
|
* @returns {boolean} |
|
*/ |
|
|
|
|
|
function hasSpaces(text, index, opts) { |
|
opts = opts || {}; |
|
var idx = skipSpaces(text, opts.backwards ? index - 1 : index, opts); |
|
return idx !== index; |
|
} |
|
/** |
|
* @param {{range?: [number, number], start?: number}} node |
|
* @param {number} index |
|
*/ |
|
|
|
|
|
function setLocStart(node, index) { |
|
if (node.range) { |
|
node.range[0] = index; |
|
} else { |
|
node.start = index; |
|
} |
|
} |
|
/** |
|
* @param {{range?: [number, number], end?: number}} node |
|
* @param {number} index |
|
*/ |
|
|
|
|
|
function setLocEnd(node, index) { |
|
if (node.range) { |
|
node.range[1] = index; |
|
} else { |
|
node.end = index; |
|
} |
|
} |
|
|
|
var PRECEDENCE = {}; |
|
[["|>"], ["??"], ["||"], ["&&"], ["|"], ["^"], ["&"], ["==", "===", "!=", "!=="], ["<", ">", "<=", ">=", "in", "instanceof"], [">>", "<<", ">>>"], ["+", "-"], ["*", "/", "%"], ["**"]].forEach(function (tier, i) { |
|
tier.forEach(function (op) { |
|
PRECEDENCE[op] = i; |
|
}); |
|
}); |
|
|
|
function getPrecedence(op) { |
|
return PRECEDENCE[op]; |
|
} |
|
|
|
var equalityOperators = { |
|
"==": true, |
|
"!=": true, |
|
"===": true, |
|
"!==": true |
|
}; |
|
var multiplicativeOperators = { |
|
"*": true, |
|
"/": true, |
|
"%": true |
|
}; |
|
var bitshiftOperators = { |
|
">>": true, |
|
">>>": true, |
|
"<<": true |
|
}; |
|
|
|
function shouldFlatten(parentOp, nodeOp) { |
|
if (getPrecedence(nodeOp) !== getPrecedence(parentOp)) { |
|
return false; |
|
} // ** is right-associative |
|
// x ** y ** z --> x ** (y ** z) |
|
|
|
|
|
if (parentOp === "**") { |
|
return false; |
|
} // x == y == z --> (x == y) == z |
|
|
|
|
|
if (equalityOperators[parentOp] && equalityOperators[nodeOp]) { |
|
return false; |
|
} // x * y % z --> (x * y) % z |
|
|
|
|
|
if (nodeOp === "%" && multiplicativeOperators[parentOp] || parentOp === "%" && multiplicativeOperators[nodeOp]) { |
|
return false; |
|
} // x * y / z --> (x * y) / z |
|
// x / y * z --> (x / y) * z |
|
|
|
|
|
if (nodeOp !== parentOp && multiplicativeOperators[nodeOp] && multiplicativeOperators[parentOp]) { |
|
return false; |
|
} // x << y << z --> (x << y) << z |
|
|
|
|
|
if (bitshiftOperators[parentOp] && bitshiftOperators[nodeOp]) { |
|
return false; |
|
} |
|
|
|
return true; |
|
} |
|
|
|
function isBitwiseOperator(operator) { |
|
return !!bitshiftOperators[operator] || operator === "|" || operator === "^" || operator === "&"; |
|
} // Tests if an expression starts with `{`, or (if forbidFunctionClassAndDoExpr |
|
// holds) `function`, `class`, or `do {}`. Will be overzealous if there's |
|
// already necessary grouping parentheses. |
|
|
|
|
|
function startsWithNoLookaheadToken(node, forbidFunctionClassAndDoExpr) { |
|
node = getLeftMost(node); |
|
|
|
switch (node.type) { |
|
case "FunctionExpression": |
|
case "ClassExpression": |
|
case "DoExpression": |
|
return forbidFunctionClassAndDoExpr; |
|
|
|
case "ObjectExpression": |
|
return true; |
|
|
|
case "MemberExpression": |
|
case "OptionalMemberExpression": |
|
return startsWithNoLookaheadToken(node.object, forbidFunctionClassAndDoExpr); |
|
|
|
case "TaggedTemplateExpression": |
|
if (node.tag.type === "FunctionExpression") { |
|
// IIFEs are always already parenthesized |
|
return false; |
|
} |
|
|
|
return startsWithNoLookaheadToken(node.tag, forbidFunctionClassAndDoExpr); |
|
|
|
case "CallExpression": |
|
case "OptionalCallExpression": |
|
if (node.callee.type === "FunctionExpression") { |
|
// IIFEs are always already parenthesized |
|
return false; |
|
} |
|
|
|
return startsWithNoLookaheadToken(node.callee, forbidFunctionClassAndDoExpr); |
|
|
|
case "ConditionalExpression": |
|
return startsWithNoLookaheadToken(node.test, forbidFunctionClassAndDoExpr); |
|
|
|
case "UpdateExpression": |
|
return !node.prefix && startsWithNoLookaheadToken(node.argument, forbidFunctionClassAndDoExpr); |
|
|
|
case "BindExpression": |
|
return node.object && startsWithNoLookaheadToken(node.object, forbidFunctionClassAndDoExpr); |
|
|
|
case "SequenceExpression": |
|
return startsWithNoLookaheadToken(node.expressions[0], forbidFunctionClassAndDoExpr); |
|
|
|
case "TSAsExpression": |
|
return startsWithNoLookaheadToken(node.expression, forbidFunctionClassAndDoExpr); |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function getLeftMost(node) { |
|
if (node.left) { |
|
return getLeftMost(node.left); |
|
} |
|
|
|
return node; |
|
} |
|
/** |
|
* @param {string} value |
|
* @param {number} tabWidth |
|
* @param {number=} startIndex |
|
* @returns {number} |
|
*/ |
|
|
|
|
|
function getAlignmentSize(value, tabWidth, startIndex) { |
|
startIndex = startIndex || 0; |
|
var size = 0; |
|
|
|
for (var i = startIndex; i < value.length; ++i) { |
|
if (value[i] === "\t") { |
|
// Tabs behave in a way that they are aligned to the nearest |
|
// multiple of tabWidth: |
|
// 0 -> 4, 1 -> 4, 2 -> 4, 3 -> 4 |
|
// 4 -> 8, 5 -> 8, 6 -> 8, 7 -> 8 ... |
|
size = size + tabWidth - size % tabWidth; |
|
} else { |
|
size++; |
|
} |
|
} |
|
|
|
return size; |
|
} |
|
/** |
|
* @param {string} value |
|
* @param {number} tabWidth |
|
* @returns {number} |
|
*/ |
|
|
|
|
|
function getIndentSize(value, tabWidth) { |
|
var lastNewlineIndex = value.lastIndexOf("\n"); |
|
|
|
if (lastNewlineIndex === -1) { |
|
return 0; |
|
} |
|
|
|
return getAlignmentSize( // All the leading whitespaces |
|
value.slice(lastNewlineIndex + 1).match(/^[ \t]*/)[0], tabWidth); |
|
} |
|
/** |
|
* @typedef {'"' | "'"} Quote |
|
*/ |
|
|
|
/** |
|
* |
|
* @param {string} raw |
|
* @param {Quote} preferredQuote |
|
* @returns {Quote} |
|
*/ |
|
|
|
|
|
function getPreferredQuote(raw, preferredQuote) { |
|
// `rawContent` is the string exactly like it appeared in the input source |
|
// code, without its enclosing quotes. |
|
var rawContent = raw.slice(1, -1); |
|
/** @type {{ quote: '"', regex: RegExp }} */ |
|
|
|
var double = { |
|
quote: '"', |
|
regex: /"/g |
|
}; |
|
/** @type {{ quote: "'", regex: RegExp }} */ |
|
|
|
var single = { |
|
quote: "'", |
|
regex: /'/g |
|
}; |
|
var preferred = preferredQuote === "'" ? single : double; |
|
var alternate = preferred === single ? double : single; |
|
var result = preferred.quote; // If `rawContent` contains at least one of the quote preferred for enclosing |
|
// the string, we might want to enclose with the alternate quote instead, to |
|
// minimize the number of escaped quotes. |
|
|
|
if (rawContent.includes(preferred.quote) || rawContent.includes(alternate.quote)) { |
|
var numPreferredQuotes = (rawContent.match(preferred.regex) || []).length; |
|
var numAlternateQuotes = (rawContent.match(alternate.regex) || []).length; |
|
result = numPreferredQuotes > numAlternateQuotes ? alternate.quote : preferred.quote; |
|
} |
|
|
|
return result; |
|
} |
|
|
|
function printString(raw, options, isDirectiveLiteral) { |
|
// `rawContent` is the string exactly like it appeared in the input source |
|
// code, without its enclosing quotes. |
|
var rawContent = raw.slice(1, -1); // Check for the alternate quote, to determine if we're allowed to swap |
|
// the quotes on a DirectiveLiteral. |
|
|
|
var canChangeDirectiveQuotes = !rawContent.includes('"') && !rawContent.includes("'"); |
|
/** @type {Quote} */ |
|
|
|
var enclosingQuote = options.parser === "json" ? '"' : options.__isInHtmlAttribute ? "'" : getPreferredQuote(raw, options.singleQuote ? "'" : '"'); // Directives are exact code unit sequences, which means that you can't |
|
// change the escape sequences they use. |
|
// See https://github.com/prettier/prettier/issues/1555 |
|
// and https://tc39.github.io/ecma262/#directive-prologue |
|
|
|
if (isDirectiveLiteral) { |
|
if (canChangeDirectiveQuotes) { |
|
return enclosingQuote + rawContent + enclosingQuote; |
|
} |
|
|
|
return raw; |
|
} // It might sound unnecessary to use `makeString` even if the string already |
|
// is enclosed with `enclosingQuote`, but it isn't. The string could contain |
|
// unnecessary escapes (such as in `"\'"`). Always using `makeString` makes |
|
// sure that we consistently output the minimum amount of escaped quotes. |
|
|
|
|
|
return makeString(rawContent, enclosingQuote, !(options.parser === "css" || options.parser === "less" || options.parser === "scss" || options.embeddedInHtml)); |
|
} |
|
/** |
|
* @param {string} rawContent |
|
* @param {Quote} enclosingQuote |
|
* @param {boolean=} unescapeUnnecessaryEscapes |
|
* @returns {string} |
|
*/ |
|
|
|
|
|
function makeString(rawContent, enclosingQuote, unescapeUnnecessaryEscapes) { |
|
var otherQuote = enclosingQuote === '"' ? "'" : '"'; // Matches _any_ escape and unescaped quotes (both single and double). |
|
|
|
var regex = /\\([\s\S])|(['"])/g; // Escape and unescape single and double quotes as needed to be able to |
|
// enclose `rawContent` with `enclosingQuote`. |
|
|
|
var newContent = rawContent.replace(regex, function (match, escaped, quote) { |
|
// If we matched an escape, and the escaped character is a quote of the |
|
// other type than we intend to enclose the string with, there's no need for |
|
// it to be escaped, so return it _without_ the backslash. |
|
if (escaped === otherQuote) { |
|
return escaped; |
|
} // If we matched an unescaped quote and it is of the _same_ type as we |
|
// intend to enclose the string with, it must be escaped, so return it with |
|
// a backslash. |
|
|
|
|
|
if (quote === enclosingQuote) { |
|
return "\\" + quote; |
|
} |
|
|
|
if (quote) { |
|
return quote; |
|
} // Unescape any unnecessarily escaped character. |
|
// Adapted from https://github.com/eslint/eslint/blob/de0b4ad7bd820ade41b1f606008bea68683dc11a/lib/rules/no-useless-escape.js#L27 |
|
|
|
|
|
return unescapeUnnecessaryEscapes && /^[^\\nrvtbfux\r\n\u2028\u2029"'0-7]$/.test(escaped) ? escaped : "\\" + escaped; |
|
}); |
|
return enclosingQuote + newContent + enclosingQuote; |
|
} |
|
|
|
function printNumber(rawNumber) { |
|
return rawNumber.toLowerCase() // Remove unnecessary plus and zeroes from scientific notation. |
|
.replace(/^([+-]?[\d.]+e)(?:\+|(-))?0*(\d)/, "$1$2$3") // Remove unnecessary scientific notation (1e0). |
|
.replace(/^([+-]?[\d.]+)e[+-]?0+$/, "$1") // Make sure numbers always start with a digit. |
|
.replace(/^([+-])?\./, "$10.") // Remove extraneous trailing decimal zeroes. |
|
.replace(/(\.\d+?)0+(?=e|$)/, "$1") // Remove trailing dot. |
|
.replace(/\.(?=e|$)/, ""); |
|
} |
|
/** |
|
* @param {string} str |
|
* @param {string} target |
|
* @returns {number} |
|
*/ |
|
|
|
|
|
function getMaxContinuousCount(str, target) { |
|
var results = str.match(new RegExp("(".concat(escapeStringRegexp(target), ")+"), "g")); |
|
|
|
if (results === null) { |
|
return 0; |
|
} |
|
|
|
return results.reduce(function (maxCount, result) { |
|
return Math.max(maxCount, result.length / target.length); |
|
}, 0); |
|
} |
|
|
|
function getMinNotPresentContinuousCount(str, target) { |
|
var matches = str.match(new RegExp("(".concat(escapeStringRegexp(target), ")+"), "g")); |
|
|
|
if (matches === null) { |
|
return 0; |
|
} |
|
|
|
var countPresent = new Map(); |
|
var max = 0; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = matches[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var match = _step.value; |
|
var count = match.length / target.length; |
|
countPresent.set(count, true); |
|
|
|
if (count > max) { |
|
max = count; |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
for (var i = 1; i < max; i++) { |
|
if (!countPresent.get(i)) { |
|
return i; |
|
} |
|
} |
|
|
|
return max + 1; |
|
} |
|
/** |
|
* @param {string} text |
|
* @returns {number} |
|
*/ |
|
|
|
|
|
function getStringWidth(text) { |
|
if (!text) { |
|
return 0; |
|
} // shortcut to avoid needless string `RegExp`s, replacements, and allocations within `string-width` |
|
|
|
|
|
if (!notAsciiRegex.test(text)) { |
|
return text.length; |
|
} |
|
|
|
return stringWidth_1(text); |
|
} |
|
|
|
function hasIgnoreComment(path) { |
|
var node = path.getValue(); |
|
return hasNodeIgnoreComment(node); |
|
} |
|
|
|
function hasNodeIgnoreComment(node) { |
|
return node && node.comments && node.comments.length > 0 && node.comments.some(function (comment) { |
|
return comment.value.trim() === "prettier-ignore"; |
|
}); |
|
} |
|
|
|
function matchAncestorTypes(path, types, index) { |
|
index = index || 0; |
|
types = types.slice(); |
|
|
|
while (types.length) { |
|
var parent = path.getParentNode(index); |
|
var type = types.shift(); |
|
|
|
if (!parent || parent.type !== type) { |
|
return false; |
|
} |
|
|
|
index++; |
|
} |
|
|
|
return true; |
|
} |
|
|
|
function addCommentHelper(node, comment) { |
|
var comments = node.comments || (node.comments = []); |
|
comments.push(comment); |
|
comment.printed = false; // For some reason, TypeScript parses `// x` inside of JSXText as a comment |
|
// We already "print" it via the raw text, we don't need to re-print it as a |
|
// comment |
|
|
|
if (node.type === "JSXText") { |
|
comment.printed = true; |
|
} |
|
} |
|
|
|
function addLeadingComment(node, comment) { |
|
comment.leading = true; |
|
comment.trailing = false; |
|
addCommentHelper(node, comment); |
|
} |
|
|
|
function addDanglingComment(node, comment) { |
|
comment.leading = false; |
|
comment.trailing = false; |
|
addCommentHelper(node, comment); |
|
} |
|
|
|
function addTrailingComment(node, comment) { |
|
comment.leading = false; |
|
comment.trailing = true; |
|
addCommentHelper(node, comment); |
|
} |
|
|
|
function isWithinParentArrayProperty(path, propertyName) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
|
|
if (parent == null) { |
|
return false; |
|
} |
|
|
|
if (!Array.isArray(parent[propertyName])) { |
|
return false; |
|
} |
|
|
|
var key = path.getName(); |
|
return parent[propertyName][key] === node; |
|
} |
|
|
|
function replaceEndOfLineWith(text, replacement) { |
|
var parts = []; |
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = text.split("\n")[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var part = _step2.value; |
|
|
|
if (parts.length !== 0) { |
|
parts.push(replacement); |
|
} |
|
|
|
parts.push(part); |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
|
|
return parts; |
|
} |
|
|
|
var util = { |
|
replaceEndOfLineWith: replaceEndOfLineWith, |
|
getStringWidth: getStringWidth, |
|
getMaxContinuousCount: getMaxContinuousCount, |
|
getMinNotPresentContinuousCount: getMinNotPresentContinuousCount, |
|
getPrecedence: getPrecedence, |
|
shouldFlatten: shouldFlatten, |
|
isBitwiseOperator: isBitwiseOperator, |
|
isExportDeclaration: isExportDeclaration, |
|
getParentExportDeclaration: getParentExportDeclaration, |
|
getPenultimate: getPenultimate, |
|
getLast: getLast, |
|
getNextNonSpaceNonCommentCharacterIndexWithStartIndex: getNextNonSpaceNonCommentCharacterIndexWithStartIndex, |
|
getNextNonSpaceNonCommentCharacterIndex: getNextNonSpaceNonCommentCharacterIndex, |
|
getNextNonSpaceNonCommentCharacter: getNextNonSpaceNonCommentCharacter, |
|
skip: skip, |
|
skipWhitespace: skipWhitespace, |
|
skipSpaces: skipSpaces, |
|
skipToLineEnd: skipToLineEnd, |
|
skipEverythingButNewLine: skipEverythingButNewLine, |
|
skipInlineComment: skipInlineComment, |
|
skipTrailingComment: skipTrailingComment, |
|
skipNewline: skipNewline, |
|
isNextLineEmptyAfterIndex: isNextLineEmptyAfterIndex, |
|
isNextLineEmpty: isNextLineEmpty, |
|
isPreviousLineEmpty: isPreviousLineEmpty, |
|
hasNewline: hasNewline, |
|
hasNewlineInRange: hasNewlineInRange, |
|
hasSpaces: hasSpaces, |
|
setLocStart: setLocStart, |
|
setLocEnd: setLocEnd, |
|
startsWithNoLookaheadToken: startsWithNoLookaheadToken, |
|
getAlignmentSize: getAlignmentSize, |
|
getIndentSize: getIndentSize, |
|
getPreferredQuote: getPreferredQuote, |
|
printString: printString, |
|
printNumber: printNumber, |
|
hasIgnoreComment: hasIgnoreComment, |
|
hasNodeIgnoreComment: hasNodeIgnoreComment, |
|
makeString: makeString, |
|
matchAncestorTypes: matchAncestorTypes, |
|
addLeadingComment: addLeadingComment, |
|
addDanglingComment: addDanglingComment, |
|
addTrailingComment: addTrailingComment, |
|
isWithinParentArrayProperty: isWithinParentArrayProperty |
|
}; |
|
|
|
function guessEndOfLine(text) { |
|
var index = text.indexOf("\r"); |
|
|
|
if (index >= 0) { |
|
return text.charAt(index + 1) === "\n" ? "crlf" : "cr"; |
|
} |
|
|
|
return "lf"; |
|
} |
|
|
|
function convertEndOfLineToChars(value) { |
|
switch (value) { |
|
case "cr": |
|
return "\r"; |
|
|
|
case "crlf": |
|
return "\r\n"; |
|
|
|
default: |
|
return "\n"; |
|
} |
|
} |
|
|
|
var endOfLine = { |
|
guessEndOfLine: guessEndOfLine, |
|
convertEndOfLineToChars: convertEndOfLineToChars |
|
}; |
|
|
|
var getStringWidth$1 = util.getStringWidth; |
|
var convertEndOfLineToChars$1 = endOfLine.convertEndOfLineToChars; |
|
var concat$1 = docBuilders.concat, |
|
fill$1 = docBuilders.fill, |
|
cursor$1 = docBuilders.cursor; |
|
/** @type {Record<symbol, typeof MODE_BREAK | typeof MODE_FLAT>} */ |
|
|
|
var groupModeMap; |
|
var MODE_BREAK = 1; |
|
var MODE_FLAT = 2; |
|
|
|
function rootIndent() { |
|
return { |
|
value: "", |
|
length: 0, |
|
queue: [] |
|
}; |
|
} |
|
|
|
function makeIndent(ind, options) { |
|
return generateInd(ind, { |
|
type: "indent" |
|
}, options); |
|
} |
|
|
|
function makeAlign(ind, n, options) { |
|
return n === -Infinity ? ind.root || rootIndent() : n < 0 ? generateInd(ind, { |
|
type: "dedent" |
|
}, options) : !n ? ind : n.type === "root" ? Object.assign({}, ind, { |
|
root: ind |
|
}) : typeof n === "string" ? generateInd(ind, { |
|
type: "stringAlign", |
|
n: n |
|
}, options) : generateInd(ind, { |
|
type: "numberAlign", |
|
n: n |
|
}, options); |
|
} |
|
|
|
function generateInd(ind, newPart, options) { |
|
var queue = newPart.type === "dedent" ? ind.queue.slice(0, -1) : ind.queue.concat(newPart); |
|
var value = ""; |
|
var length = 0; |
|
var lastTabs = 0; |
|
var lastSpaces = 0; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = queue[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var part = _step.value; |
|
|
|
switch (part.type) { |
|
case "indent": |
|
flush(); |
|
|
|
if (options.useTabs) { |
|
addTabs(1); |
|
} else { |
|
addSpaces(options.tabWidth); |
|
} |
|
|
|
break; |
|
|
|
case "stringAlign": |
|
flush(); |
|
value += part.n; |
|
length += part.n.length; |
|
break; |
|
|
|
case "numberAlign": |
|
lastTabs += 1; |
|
lastSpaces += part.n; |
|
break; |
|
|
|
/* istanbul ignore next */ |
|
|
|
default: |
|
throw new Error("Unexpected type '".concat(part.type, "'")); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
flushSpaces(); |
|
return Object.assign({}, ind, { |
|
value: value, |
|
length: length, |
|
queue: queue |
|
}); |
|
|
|
function addTabs(count) { |
|
value += "\t".repeat(count); |
|
length += options.tabWidth * count; |
|
} |
|
|
|
function addSpaces(count) { |
|
value += " ".repeat(count); |
|
length += count; |
|
} |
|
|
|
function flush() { |
|
if (options.useTabs) { |
|
flushTabs(); |
|
} else { |
|
flushSpaces(); |
|
} |
|
} |
|
|
|
function flushTabs() { |
|
if (lastTabs > 0) { |
|
addTabs(lastTabs); |
|
} |
|
|
|
resetLast(); |
|
} |
|
|
|
function flushSpaces() { |
|
if (lastSpaces > 0) { |
|
addSpaces(lastSpaces); |
|
} |
|
|
|
resetLast(); |
|
} |
|
|
|
function resetLast() { |
|
lastTabs = 0; |
|
lastSpaces = 0; |
|
} |
|
} |
|
|
|
function trim$1(out) { |
|
if (out.length === 0) { |
|
return 0; |
|
} |
|
|
|
var trimCount = 0; // Trim whitespace at the end of line |
|
|
|
while (out.length > 0 && typeof out[out.length - 1] === "string" && out[out.length - 1].match(/^[ \t]*$/)) { |
|
trimCount += out.pop().length; |
|
} |
|
|
|
if (out.length && typeof out[out.length - 1] === "string") { |
|
var trimmed = out[out.length - 1].replace(/[ \t]*$/, ""); |
|
trimCount += out[out.length - 1].length - trimmed.length; |
|
out[out.length - 1] = trimmed; |
|
} |
|
|
|
return trimCount; |
|
} |
|
|
|
function fits(next, restCommands, width, options, mustBeFlat) { |
|
var restIdx = restCommands.length; |
|
var cmds = [next]; // `out` is only used for width counting because `trim` requires to look |
|
// backwards for space characters. |
|
|
|
var out = []; |
|
|
|
while (width >= 0) { |
|
if (cmds.length === 0) { |
|
if (restIdx === 0) { |
|
return true; |
|
} |
|
|
|
cmds.push(restCommands[restIdx - 1]); |
|
restIdx--; |
|
continue; |
|
} |
|
|
|
var x = cmds.pop(); |
|
var ind = x[0]; |
|
var mode = x[1]; |
|
var doc = x[2]; |
|
|
|
if (typeof doc === "string") { |
|
out.push(doc); |
|
width -= getStringWidth$1(doc); |
|
} else { |
|
switch (doc.type) { |
|
case "concat": |
|
for (var i = doc.parts.length - 1; i >= 0; i--) { |
|
cmds.push([ind, mode, doc.parts[i]]); |
|
} |
|
|
|
break; |
|
|
|
case "indent": |
|
cmds.push([makeIndent(ind, options), mode, doc.contents]); |
|
break; |
|
|
|
case "align": |
|
cmds.push([makeAlign(ind, doc.n, options), mode, doc.contents]); |
|
break; |
|
|
|
case "trim": |
|
width += trim$1(out); |
|
break; |
|
|
|
case "group": |
|
if (mustBeFlat && doc.break) { |
|
return false; |
|
} |
|
|
|
cmds.push([ind, doc.break ? MODE_BREAK : mode, doc.contents]); |
|
|
|
if (doc.id) { |
|
groupModeMap[doc.id] = cmds[cmds.length - 1][1]; |
|
} |
|
|
|
break; |
|
|
|
case "fill": |
|
for (var _i = doc.parts.length - 1; _i >= 0; _i--) { |
|
cmds.push([ind, mode, doc.parts[_i]]); |
|
} |
|
|
|
break; |
|
|
|
case "if-break": |
|
{ |
|
var groupMode = doc.groupId ? groupModeMap[doc.groupId] : mode; |
|
|
|
if (groupMode === MODE_BREAK) { |
|
if (doc.breakContents) { |
|
cmds.push([ind, mode, doc.breakContents]); |
|
} |
|
} |
|
|
|
if (groupMode === MODE_FLAT) { |
|
if (doc.flatContents) { |
|
cmds.push([ind, mode, doc.flatContents]); |
|
} |
|
} |
|
|
|
break; |
|
} |
|
|
|
case "line": |
|
switch (mode) { |
|
// fallthrough |
|
case MODE_FLAT: |
|
if (!doc.hard) { |
|
if (!doc.soft) { |
|
out.push(" "); |
|
width -= 1; |
|
} |
|
|
|
break; |
|
} |
|
|
|
return true; |
|
|
|
case MODE_BREAK: |
|
return true; |
|
} |
|
|
|
break; |
|
} |
|
} |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function printDocToString(doc, options) { |
|
groupModeMap = {}; |
|
var width = options.printWidth; |
|
var newLine = convertEndOfLineToChars$1(options.endOfLine); |
|
var pos = 0; // cmds is basically a stack. We've turned a recursive call into a |
|
// while loop which is much faster. The while loop below adds new |
|
// cmds to the array instead of recursively calling `print`. |
|
|
|
var cmds = [[rootIndent(), MODE_BREAK, doc]]; |
|
var out = []; |
|
var shouldRemeasure = false; |
|
var lineSuffix = []; |
|
|
|
while (cmds.length !== 0) { |
|
var x = cmds.pop(); |
|
var ind = x[0]; |
|
var mode = x[1]; |
|
var _doc = x[2]; |
|
|
|
if (typeof _doc === "string") { |
|
out.push(_doc); |
|
pos += getStringWidth$1(_doc); |
|
} else { |
|
switch (_doc.type) { |
|
case "cursor": |
|
out.push(cursor$1.placeholder); |
|
break; |
|
|
|
case "concat": |
|
for (var i = _doc.parts.length - 1; i >= 0; i--) { |
|
cmds.push([ind, mode, _doc.parts[i]]); |
|
} |
|
|
|
break; |
|
|
|
case "indent": |
|
cmds.push([makeIndent(ind, options), mode, _doc.contents]); |
|
break; |
|
|
|
case "align": |
|
cmds.push([makeAlign(ind, _doc.n, options), mode, _doc.contents]); |
|
break; |
|
|
|
case "trim": |
|
pos -= trim$1(out); |
|
break; |
|
|
|
case "group": |
|
switch (mode) { |
|
case MODE_FLAT: |
|
if (!shouldRemeasure) { |
|
cmds.push([ind, _doc.break ? MODE_BREAK : MODE_FLAT, _doc.contents]); |
|
break; |
|
} |
|
|
|
// fallthrough |
|
|
|
case MODE_BREAK: |
|
{ |
|
shouldRemeasure = false; |
|
var next = [ind, MODE_FLAT, _doc.contents]; |
|
var rem = width - pos; |
|
|
|
if (!_doc.break && fits(next, cmds, rem, options)) { |
|
cmds.push(next); |
|
} else { |
|
// Expanded states are a rare case where a document |
|
// can manually provide multiple representations of |
|
// itself. It provides an array of documents |
|
// going from the least expanded (most flattened) |
|
// representation first to the most expanded. If a |
|
// group has these, we need to manually go through |
|
// these states and find the first one that fits. |
|
if (_doc.expandedStates) { |
|
var mostExpanded = _doc.expandedStates[_doc.expandedStates.length - 1]; |
|
|
|
if (_doc.break) { |
|
cmds.push([ind, MODE_BREAK, mostExpanded]); |
|
break; |
|
} else { |
|
for (var _i2 = 1; _i2 < _doc.expandedStates.length + 1; _i2++) { |
|
if (_i2 >= _doc.expandedStates.length) { |
|
cmds.push([ind, MODE_BREAK, mostExpanded]); |
|
break; |
|
} else { |
|
var state = _doc.expandedStates[_i2]; |
|
var cmd = [ind, MODE_FLAT, state]; |
|
|
|
if (fits(cmd, cmds, rem, options)) { |
|
cmds.push(cmd); |
|
break; |
|
} |
|
} |
|
} |
|
} |
|
} else { |
|
cmds.push([ind, MODE_BREAK, _doc.contents]); |
|
} |
|
} |
|
|
|
break; |
|
} |
|
} |
|
|
|
if (_doc.id) { |
|
groupModeMap[_doc.id] = cmds[cmds.length - 1][1]; |
|
} |
|
|
|
break; |
|
// Fills each line with as much code as possible before moving to a new |
|
// line with the same indentation. |
|
// |
|
// Expects doc.parts to be an array of alternating content and |
|
// whitespace. The whitespace contains the linebreaks. |
|
// |
|
// For example: |
|
// ["I", line, "love", line, "monkeys"] |
|
// or |
|
// [{ type: group, ... }, softline, { type: group, ... }] |
|
// |
|
// It uses this parts structure to handle three main layout cases: |
|
// * The first two content items fit on the same line without |
|
// breaking |
|
// -> output the first content item and the whitespace "flat". |
|
// * Only the first content item fits on the line without breaking |
|
// -> output the first content item "flat" and the whitespace with |
|
// "break". |
|
// * Neither content item fits on the line without breaking |
|
// -> output the first content item and the whitespace with "break". |
|
|
|
case "fill": |
|
{ |
|
var _rem = width - pos; |
|
|
|
var parts = _doc.parts; |
|
|
|
if (parts.length === 0) { |
|
break; |
|
} |
|
|
|
var content = parts[0]; |
|
var contentFlatCmd = [ind, MODE_FLAT, content]; |
|
var contentBreakCmd = [ind, MODE_BREAK, content]; |
|
var contentFits = fits(contentFlatCmd, [], _rem, options, true); |
|
|
|
if (parts.length === 1) { |
|
if (contentFits) { |
|
cmds.push(contentFlatCmd); |
|
} else { |
|
cmds.push(contentBreakCmd); |
|
} |
|
|
|
break; |
|
} |
|
|
|
var whitespace = parts[1]; |
|
var whitespaceFlatCmd = [ind, MODE_FLAT, whitespace]; |
|
var whitespaceBreakCmd = [ind, MODE_BREAK, whitespace]; |
|
|
|
if (parts.length === 2) { |
|
if (contentFits) { |
|
cmds.push(whitespaceFlatCmd); |
|
cmds.push(contentFlatCmd); |
|
} else { |
|
cmds.push(whitespaceBreakCmd); |
|
cmds.push(contentBreakCmd); |
|
} |
|
|
|
break; |
|
} // At this point we've handled the first pair (context, separator) |
|
// and will create a new fill doc for the rest of the content. |
|
// Ideally we wouldn't mutate the array here but coping all the |
|
// elements to a new array would make this algorithm quadratic, |
|
// which is unusable for large arrays (e.g. large texts in JSX). |
|
|
|
|
|
parts.splice(0, 2); |
|
var remainingCmd = [ind, mode, fill$1(parts)]; |
|
var secondContent = parts[0]; |
|
var firstAndSecondContentFlatCmd = [ind, MODE_FLAT, concat$1([content, whitespace, secondContent])]; |
|
var firstAndSecondContentFits = fits(firstAndSecondContentFlatCmd, [], _rem, options, true); |
|
|
|
if (firstAndSecondContentFits) { |
|
cmds.push(remainingCmd); |
|
cmds.push(whitespaceFlatCmd); |
|
cmds.push(contentFlatCmd); |
|
} else if (contentFits) { |
|
cmds.push(remainingCmd); |
|
cmds.push(whitespaceBreakCmd); |
|
cmds.push(contentFlatCmd); |
|
} else { |
|
cmds.push(remainingCmd); |
|
cmds.push(whitespaceBreakCmd); |
|
cmds.push(contentBreakCmd); |
|
} |
|
|
|
break; |
|
} |
|
|
|
case "if-break": |
|
{ |
|
var groupMode = _doc.groupId ? groupModeMap[_doc.groupId] : mode; |
|
|
|
if (groupMode === MODE_BREAK) { |
|
if (_doc.breakContents) { |
|
cmds.push([ind, mode, _doc.breakContents]); |
|
} |
|
} |
|
|
|
if (groupMode === MODE_FLAT) { |
|
if (_doc.flatContents) { |
|
cmds.push([ind, mode, _doc.flatContents]); |
|
} |
|
} |
|
|
|
break; |
|
} |
|
|
|
case "line-suffix": |
|
lineSuffix.push([ind, mode, _doc.contents]); |
|
break; |
|
|
|
case "line-suffix-boundary": |
|
if (lineSuffix.length > 0) { |
|
cmds.push([ind, mode, { |
|
type: "line", |
|
hard: true |
|
}]); |
|
} |
|
|
|
break; |
|
|
|
case "line": |
|
switch (mode) { |
|
case MODE_FLAT: |
|
if (!_doc.hard) { |
|
if (!_doc.soft) { |
|
out.push(" "); |
|
pos += 1; |
|
} |
|
|
|
break; |
|
} else { |
|
// This line was forced into the output even if we |
|
// were in flattened mode, so we need to tell the next |
|
// group that no matter what, it needs to remeasure |
|
// because the previous measurement didn't accurately |
|
// capture the entire expression (this is necessary |
|
// for nested groups) |
|
shouldRemeasure = true; |
|
} |
|
|
|
// fallthrough |
|
|
|
case MODE_BREAK: |
|
if (lineSuffix.length) { |
|
cmds.push([ind, mode, _doc]); |
|
[].push.apply(cmds, lineSuffix.reverse()); |
|
lineSuffix = []; |
|
break; |
|
} |
|
|
|
if (_doc.literal) { |
|
if (ind.root) { |
|
out.push(newLine, ind.root.value); |
|
pos = ind.root.length; |
|
} else { |
|
out.push(newLine); |
|
pos = 0; |
|
} |
|
} else { |
|
pos -= trim$1(out); |
|
out.push(newLine + ind.value); |
|
pos = ind.length; |
|
} |
|
|
|
break; |
|
} |
|
|
|
break; |
|
} |
|
} |
|
} |
|
|
|
var cursorPlaceholderIndex = out.indexOf(cursor$1.placeholder); |
|
|
|
if (cursorPlaceholderIndex !== -1) { |
|
var otherCursorPlaceholderIndex = out.indexOf(cursor$1.placeholder, cursorPlaceholderIndex + 1); |
|
var beforeCursor = out.slice(0, cursorPlaceholderIndex).join(""); |
|
var aroundCursor = out.slice(cursorPlaceholderIndex + 1, otherCursorPlaceholderIndex).join(""); |
|
var afterCursor = out.slice(otherCursorPlaceholderIndex + 1).join(""); |
|
return { |
|
formatted: beforeCursor + aroundCursor + afterCursor, |
|
cursorNodeStart: beforeCursor.length, |
|
cursorNodeText: aroundCursor |
|
}; |
|
} |
|
|
|
return { |
|
formatted: out.join("") |
|
}; |
|
} |
|
|
|
var docPrinter = { |
|
printDocToString: printDocToString |
|
}; |
|
|
|
var traverseDocOnExitStackMarker = {}; |
|
|
|
function traverseDoc(doc, onEnter, onExit, shouldTraverseConditionalGroups) { |
|
var docsStack = [doc]; |
|
|
|
while (docsStack.length !== 0) { |
|
var _doc = docsStack.pop(); |
|
|
|
if (_doc === traverseDocOnExitStackMarker) { |
|
onExit(docsStack.pop()); |
|
continue; |
|
} |
|
|
|
var shouldRecurse = true; |
|
|
|
if (onEnter) { |
|
if (onEnter(_doc) === false) { |
|
shouldRecurse = false; |
|
} |
|
} |
|
|
|
if (onExit) { |
|
docsStack.push(_doc); |
|
docsStack.push(traverseDocOnExitStackMarker); |
|
} |
|
|
|
if (shouldRecurse) { |
|
// When there are multiple parts to process, |
|
// the parts need to be pushed onto the stack in reverse order, |
|
// so that they are processed in the original order |
|
// when the stack is popped. |
|
if (_doc.type === "concat" || _doc.type === "fill") { |
|
for (var ic = _doc.parts.length, i = ic - 1; i >= 0; --i) { |
|
docsStack.push(_doc.parts[i]); |
|
} |
|
} else if (_doc.type === "if-break") { |
|
if (_doc.flatContents) { |
|
docsStack.push(_doc.flatContents); |
|
} |
|
|
|
if (_doc.breakContents) { |
|
docsStack.push(_doc.breakContents); |
|
} |
|
} else if (_doc.type === "group" && _doc.expandedStates) { |
|
if (shouldTraverseConditionalGroups) { |
|
for (var _ic = _doc.expandedStates.length, _i = _ic - 1; _i >= 0; --_i) { |
|
docsStack.push(_doc.expandedStates[_i]); |
|
} |
|
} else { |
|
docsStack.push(_doc.contents); |
|
} |
|
} else if (_doc.contents) { |
|
docsStack.push(_doc.contents); |
|
} |
|
} |
|
} |
|
} |
|
|
|
function mapDoc(doc, cb) { |
|
if (doc.type === "concat" || doc.type === "fill") { |
|
var parts = doc.parts.map(function (part) { |
|
return mapDoc(part, cb); |
|
}); |
|
return cb(Object.assign({}, doc, { |
|
parts: parts |
|
})); |
|
} else if (doc.type === "if-break") { |
|
var breakContents = doc.breakContents && mapDoc(doc.breakContents, cb); |
|
var flatContents = doc.flatContents && mapDoc(doc.flatContents, cb); |
|
return cb(Object.assign({}, doc, { |
|
breakContents: breakContents, |
|
flatContents: flatContents |
|
})); |
|
} else if (doc.contents) { |
|
var contents = mapDoc(doc.contents, cb); |
|
return cb(Object.assign({}, doc, { |
|
contents: contents |
|
})); |
|
} |
|
|
|
return cb(doc); |
|
} |
|
|
|
function findInDoc(doc, fn, defaultValue) { |
|
var result = defaultValue; |
|
var hasStopped = false; |
|
|
|
function findInDocOnEnterFn(doc) { |
|
var maybeResult = fn(doc); |
|
|
|
if (maybeResult !== undefined) { |
|
hasStopped = true; |
|
result = maybeResult; |
|
} |
|
|
|
if (hasStopped) { |
|
return false; |
|
} |
|
} |
|
|
|
traverseDoc(doc, findInDocOnEnterFn); |
|
return result; |
|
} |
|
|
|
function isEmpty(n) { |
|
return typeof n === "string" && n.length === 0; |
|
} |
|
|
|
function isLineNextFn(doc) { |
|
if (typeof doc === "string") { |
|
return false; |
|
} |
|
|
|
if (doc.type === "line") { |
|
return true; |
|
} |
|
} |
|
|
|
function isLineNext(doc) { |
|
return findInDoc(doc, isLineNextFn, false); |
|
} |
|
|
|
function willBreakFn(doc) { |
|
if (doc.type === "group" && doc.break) { |
|
return true; |
|
} |
|
|
|
if (doc.type === "line" && doc.hard) { |
|
return true; |
|
} |
|
|
|
if (doc.type === "break-parent") { |
|
return true; |
|
} |
|
} |
|
|
|
function willBreak(doc) { |
|
return findInDoc(doc, willBreakFn, false); |
|
} |
|
|
|
function breakParentGroup(groupStack) { |
|
if (groupStack.length > 0) { |
|
var parentGroup = groupStack[groupStack.length - 1]; // Breaks are not propagated through conditional groups because |
|
// the user is expected to manually handle what breaks. |
|
|
|
if (!parentGroup.expandedStates) { |
|
parentGroup.break = true; |
|
} |
|
} |
|
|
|
return null; |
|
} |
|
|
|
function propagateBreaks(doc) { |
|
var alreadyVisitedSet = new Set(); |
|
var groupStack = []; |
|
|
|
function propagateBreaksOnEnterFn(doc) { |
|
if (doc.type === "break-parent") { |
|
breakParentGroup(groupStack); |
|
} |
|
|
|
if (doc.type === "group") { |
|
groupStack.push(doc); |
|
|
|
if (alreadyVisitedSet.has(doc)) { |
|
return false; |
|
} |
|
|
|
alreadyVisitedSet.add(doc); |
|
} |
|
} |
|
|
|
function propagateBreaksOnExitFn(doc) { |
|
if (doc.type === "group") { |
|
var group = groupStack.pop(); |
|
|
|
if (group.break) { |
|
breakParentGroup(groupStack); |
|
} |
|
} |
|
} |
|
|
|
traverseDoc(doc, propagateBreaksOnEnterFn, propagateBreaksOnExitFn, |
|
/* shouldTraverseConditionalGroups */ |
|
true); |
|
} |
|
|
|
function removeLinesFn(doc) { |
|
// Force this doc into flat mode by statically converting all |
|
// lines into spaces (or soft lines into nothing). Hard lines |
|
// should still output because there's too great of a chance |
|
// of breaking existing assumptions otherwise. |
|
if (doc.type === "line" && !doc.hard) { |
|
return doc.soft ? "" : " "; |
|
} else if (doc.type === "if-break") { |
|
return doc.flatContents || ""; |
|
} |
|
|
|
return doc; |
|
} |
|
|
|
function removeLines(doc) { |
|
return mapDoc(doc, removeLinesFn); |
|
} |
|
|
|
function stripTrailingHardline(doc) { |
|
// HACK remove ending hardline, original PR: #1984 |
|
if (doc.type === "concat" && doc.parts.length !== 0) { |
|
var lastPart = doc.parts[doc.parts.length - 1]; |
|
|
|
if (lastPart.type === "concat") { |
|
if (lastPart.parts.length === 2 && lastPart.parts[0].hard && lastPart.parts[1].type === "break-parent") { |
|
return { |
|
type: "concat", |
|
parts: doc.parts.slice(0, -1) |
|
}; |
|
} |
|
|
|
return { |
|
type: "concat", |
|
parts: doc.parts.slice(0, -1).concat(stripTrailingHardline(lastPart)) |
|
}; |
|
} |
|
} |
|
|
|
return doc; |
|
} |
|
|
|
var docUtils = { |
|
isEmpty: isEmpty, |
|
willBreak: willBreak, |
|
isLineNext: isLineNext, |
|
traverseDoc: traverseDoc, |
|
findInDoc: findInDoc, |
|
mapDoc: mapDoc, |
|
propagateBreaks: propagateBreaks, |
|
removeLines: removeLines, |
|
stripTrailingHardline: stripTrailingHardline |
|
}; |
|
|
|
function flattenDoc(doc) { |
|
if (doc.type === "concat") { |
|
var res = []; |
|
|
|
for (var i = 0; i < doc.parts.length; ++i) { |
|
var doc2 = doc.parts[i]; |
|
|
|
if (typeof doc2 !== "string" && doc2.type === "concat") { |
|
[].push.apply(res, flattenDoc(doc2).parts); |
|
} else { |
|
var flattened = flattenDoc(doc2); |
|
|
|
if (flattened !== "") { |
|
res.push(flattened); |
|
} |
|
} |
|
} |
|
|
|
return Object.assign({}, doc, { |
|
parts: res |
|
}); |
|
} else if (doc.type === "if-break") { |
|
return Object.assign({}, doc, { |
|
breakContents: doc.breakContents != null ? flattenDoc(doc.breakContents) : null, |
|
flatContents: doc.flatContents != null ? flattenDoc(doc.flatContents) : null |
|
}); |
|
} else if (doc.type === "group") { |
|
return Object.assign({}, doc, { |
|
contents: flattenDoc(doc.contents), |
|
expandedStates: doc.expandedStates ? doc.expandedStates.map(flattenDoc) : doc.expandedStates |
|
}); |
|
} else if (doc.contents) { |
|
return Object.assign({}, doc, { |
|
contents: flattenDoc(doc.contents) |
|
}); |
|
} |
|
|
|
return doc; |
|
} |
|
|
|
function printDoc(doc) { |
|
if (typeof doc === "string") { |
|
return JSON.stringify(doc); |
|
} |
|
|
|
if (doc.type === "line") { |
|
if (doc.literal) { |
|
return "literalline"; |
|
} |
|
|
|
if (doc.hard) { |
|
return "hardline"; |
|
} |
|
|
|
if (doc.soft) { |
|
return "softline"; |
|
} |
|
|
|
return "line"; |
|
} |
|
|
|
if (doc.type === "break-parent") { |
|
return "breakParent"; |
|
} |
|
|
|
if (doc.type === "trim") { |
|
return "trim"; |
|
} |
|
|
|
if (doc.type === "concat") { |
|
return "[" + doc.parts.map(printDoc).join(", ") + "]"; |
|
} |
|
|
|
if (doc.type === "indent") { |
|
return "indent(" + printDoc(doc.contents) + ")"; |
|
} |
|
|
|
if (doc.type === "align") { |
|
return doc.n === -Infinity ? "dedentToRoot(" + printDoc(doc.contents) + ")" : doc.n < 0 ? "dedent(" + printDoc(doc.contents) + ")" : doc.n.type === "root" ? "markAsRoot(" + printDoc(doc.contents) + ")" : "align(" + JSON.stringify(doc.n) + ", " + printDoc(doc.contents) + ")"; |
|
} |
|
|
|
if (doc.type === "if-break") { |
|
return "ifBreak(" + printDoc(doc.breakContents) + (doc.flatContents ? ", " + printDoc(doc.flatContents) : "") + ")"; |
|
} |
|
|
|
if (doc.type === "group") { |
|
if (doc.expandedStates) { |
|
return "conditionalGroup(" + "[" + doc.expandedStates.map(printDoc).join(",") + "])"; |
|
} |
|
|
|
return (doc.break ? "wrappedGroup" : "group") + "(" + printDoc(doc.contents) + ")"; |
|
} |
|
|
|
if (doc.type === "fill") { |
|
return "fill" + "(" + doc.parts.map(printDoc).join(", ") + ")"; |
|
} |
|
|
|
if (doc.type === "line-suffix") { |
|
return "lineSuffix(" + printDoc(doc.contents) + ")"; |
|
} |
|
|
|
if (doc.type === "line-suffix-boundary") { |
|
return "lineSuffixBoundary"; |
|
} |
|
|
|
throw new Error("Unknown doc type " + doc.type); |
|
} |
|
|
|
var docDebug = { |
|
printDocToDebug: function printDocToDebug(doc) { |
|
return printDoc(flattenDoc(doc)); |
|
} |
|
}; |
|
|
|
var doc = { |
|
builders: docBuilders, |
|
printer: docPrinter, |
|
utils: docUtils, |
|
debug: docDebug |
|
}; |
|
|
|
var mapDoc$1 = doc.utils.mapDoc; |
|
|
|
function isNextLineEmpty$1(text, node, options) { |
|
return util.isNextLineEmpty(text, node, options.locEnd); |
|
} |
|
|
|
function isPreviousLineEmpty$1(text, node, options) { |
|
return util.isPreviousLineEmpty(text, node, options.locStart); |
|
} |
|
|
|
function getNextNonSpaceNonCommentCharacterIndex$1(text, node, options) { |
|
return util.getNextNonSpaceNonCommentCharacterIndex(text, node, options.locEnd); |
|
} |
|
|
|
var utilShared = { |
|
getMaxContinuousCount: util.getMaxContinuousCount, |
|
getStringWidth: util.getStringWidth, |
|
getAlignmentSize: util.getAlignmentSize, |
|
getIndentSize: util.getIndentSize, |
|
skip: util.skip, |
|
skipWhitespace: util.skipWhitespace, |
|
skipSpaces: util.skipSpaces, |
|
skipNewline: util.skipNewline, |
|
skipToLineEnd: util.skipToLineEnd, |
|
skipEverythingButNewLine: util.skipEverythingButNewLine, |
|
skipInlineComment: util.skipInlineComment, |
|
skipTrailingComment: util.skipTrailingComment, |
|
hasNewline: util.hasNewline, |
|
hasNewlineInRange: util.hasNewlineInRange, |
|
hasSpaces: util.hasSpaces, |
|
isNextLineEmpty: isNextLineEmpty$1, |
|
isNextLineEmptyAfterIndex: util.isNextLineEmptyAfterIndex, |
|
isPreviousLineEmpty: isPreviousLineEmpty$1, |
|
getNextNonSpaceNonCommentCharacterIndex: getNextNonSpaceNonCommentCharacterIndex$1, |
|
mapDoc: mapDoc$1, |
|
// TODO: remove in 2.0, we already exposed it in docUtils |
|
makeString: util.makeString, |
|
addLeadingComment: util.addLeadingComment, |
|
addDanglingComment: util.addDanglingComment, |
|
addTrailingComment: util.addTrailingComment |
|
}; |
|
|
|
var _require$$0$builders = doc.builders, |
|
concat$2 = _require$$0$builders.concat, |
|
hardline$1 = _require$$0$builders.hardline, |
|
breakParent$1 = _require$$0$builders.breakParent, |
|
indent$1 = _require$$0$builders.indent, |
|
lineSuffix$1 = _require$$0$builders.lineSuffix, |
|
join$1 = _require$$0$builders.join, |
|
cursor$2 = _require$$0$builders.cursor; |
|
var hasNewline$1 = util.hasNewline, |
|
skipNewline$1 = util.skipNewline, |
|
isPreviousLineEmpty$2 = util.isPreviousLineEmpty; |
|
var addLeadingComment$1 = utilShared.addLeadingComment, |
|
addDanglingComment$1 = utilShared.addDanglingComment, |
|
addTrailingComment$1 = utilShared.addTrailingComment; |
|
var childNodesCacheKey = Symbol("child-nodes"); |
|
|
|
function getSortedChildNodes(node, options, resultArray) { |
|
if (!node) { |
|
return; |
|
} |
|
|
|
var printer = options.printer, |
|
locStart = options.locStart, |
|
locEnd = options.locEnd; |
|
|
|
if (resultArray) { |
|
if (node && printer.canAttachComment && printer.canAttachComment(node)) { |
|
// This reverse insertion sort almost always takes constant |
|
// time because we almost always (maybe always?) append the |
|
// nodes in order anyway. |
|
var i; |
|
|
|
for (i = resultArray.length - 1; i >= 0; --i) { |
|
if (locStart(resultArray[i]) <= locStart(node) && locEnd(resultArray[i]) <= locEnd(node)) { |
|
break; |
|
} |
|
} |
|
|
|
resultArray.splice(i + 1, 0, node); |
|
return; |
|
} |
|
} else if (node[childNodesCacheKey]) { |
|
return node[childNodesCacheKey]; |
|
} |
|
|
|
var childNodes; |
|
|
|
if (printer.getCommentChildNodes) { |
|
childNodes = printer.getCommentChildNodes(node); |
|
} else if (node && _typeof(node) === "object") { |
|
childNodes = Object.keys(node).filter(function (n) { |
|
return n !== "enclosingNode" && n !== "precedingNode" && n !== "followingNode"; |
|
}).map(function (n) { |
|
return node[n]; |
|
}); |
|
} |
|
|
|
if (!childNodes) { |
|
return; |
|
} |
|
|
|
if (!resultArray) { |
|
Object.defineProperty(node, childNodesCacheKey, { |
|
value: resultArray = [], |
|
enumerable: false |
|
}); |
|
} |
|
|
|
childNodes.forEach(function (childNode) { |
|
getSortedChildNodes(childNode, options, resultArray); |
|
}); |
|
return resultArray; |
|
} // As efficiently as possible, decorate the comment object with |
|
// .precedingNode, .enclosingNode, and/or .followingNode properties, at |
|
// least one of which is guaranteed to be defined. |
|
|
|
|
|
function decorateComment(node, comment, options) { |
|
var locStart = options.locStart, |
|
locEnd = options.locEnd; |
|
var childNodes = getSortedChildNodes(node, options); |
|
var precedingNode; |
|
var followingNode; // Time to dust off the old binary search robes and wizard hat. |
|
|
|
var left = 0; |
|
var right = childNodes.length; |
|
|
|
while (left < right) { |
|
var middle = left + right >> 1; |
|
var child = childNodes[middle]; |
|
|
|
if (locStart(child) - locStart(comment) <= 0 && locEnd(comment) - locEnd(child) <= 0) { |
|
// The comment is completely contained by this child node. |
|
comment.enclosingNode = child; |
|
decorateComment(child, comment, options); |
|
return; // Abandon the binary search at this level. |
|
} |
|
|
|
if (locEnd(child) - locStart(comment) <= 0) { |
|
// This child node falls completely before the comment. |
|
// Because we will never consider this node or any nodes |
|
// before it again, this node must be the closest preceding |
|
// node we have encountered so far. |
|
precedingNode = child; |
|
left = middle + 1; |
|
continue; |
|
} |
|
|
|
if (locEnd(comment) - locStart(child) <= 0) { |
|
// This child node falls completely after the comment. |
|
// Because we will never consider this node or any nodes after |
|
// it again, this node must be the closest following node we |
|
// have encountered so far. |
|
followingNode = child; |
|
right = middle; |
|
continue; |
|
} |
|
/* istanbul ignore next */ |
|
|
|
|
|
throw new Error("Comment location overlaps with node location"); |
|
} // We don't want comments inside of different expressions inside of the same |
|
// template literal to move to another expression. |
|
|
|
|
|
if (comment.enclosingNode && comment.enclosingNode.type === "TemplateLiteral") { |
|
var quasis = comment.enclosingNode.quasis; |
|
var commentIndex = findExpressionIndexForComment(quasis, comment, options); |
|
|
|
if (precedingNode && findExpressionIndexForComment(quasis, precedingNode, options) !== commentIndex) { |
|
precedingNode = null; |
|
} |
|
|
|
if (followingNode && findExpressionIndexForComment(quasis, followingNode, options) !== commentIndex) { |
|
followingNode = null; |
|
} |
|
} |
|
|
|
if (precedingNode) { |
|
comment.precedingNode = precedingNode; |
|
} |
|
|
|
if (followingNode) { |
|
comment.followingNode = followingNode; |
|
} |
|
} |
|
|
|
function attach(comments, ast, text, options) { |
|
if (!Array.isArray(comments)) { |
|
return; |
|
} |
|
|
|
var tiesToBreak = []; |
|
var locStart = options.locStart, |
|
locEnd = options.locEnd; |
|
comments.forEach(function (comment, i) { |
|
if (options.parser === "json" || options.parser === "json5" || options.parser === "__js_expression" || options.parser === "__vue_expression") { |
|
if (locStart(comment) - locStart(ast) <= 0) { |
|
addLeadingComment$1(ast, comment); |
|
return; |
|
} |
|
|
|
if (locEnd(comment) - locEnd(ast) >= 0) { |
|
addTrailingComment$1(ast, comment); |
|
return; |
|
} |
|
} |
|
|
|
decorateComment(ast, comment, options); |
|
var precedingNode = comment.precedingNode, |
|
enclosingNode = comment.enclosingNode, |
|
followingNode = comment.followingNode; |
|
var pluginHandleOwnLineComment = options.printer.handleComments && options.printer.handleComments.ownLine ? options.printer.handleComments.ownLine : function () { |
|
return false; |
|
}; |
|
var pluginHandleEndOfLineComment = options.printer.handleComments && options.printer.handleComments.endOfLine ? options.printer.handleComments.endOfLine : function () { |
|
return false; |
|
}; |
|
var pluginHandleRemainingComment = options.printer.handleComments && options.printer.handleComments.remaining ? options.printer.handleComments.remaining : function () { |
|
return false; |
|
}; |
|
var isLastComment = comments.length - 1 === i; |
|
|
|
if (hasNewline$1(text, locStart(comment), { |
|
backwards: true |
|
})) { |
|
// If a comment exists on its own line, prefer a leading comment. |
|
// We also need to check if it's the first line of the file. |
|
if (pluginHandleOwnLineComment(comment, text, options, ast, isLastComment)) ; else if (followingNode) { |
|
// Always a leading comment. |
|
addLeadingComment$1(followingNode, comment); |
|
} else if (precedingNode) { |
|
addTrailingComment$1(precedingNode, comment); |
|
} else if (enclosingNode) { |
|
addDanglingComment$1(enclosingNode, comment); |
|
} else { |
|
// There are no nodes, let's attach it to the root of the ast |
|
|
|
/* istanbul ignore next */ |
|
addDanglingComment$1(ast, comment); |
|
} |
|
} else if (hasNewline$1(text, locEnd(comment))) { |
|
if (pluginHandleEndOfLineComment(comment, text, options, ast, isLastComment)) ; else if (precedingNode) { |
|
// There is content before this comment on the same line, but |
|
// none after it, so prefer a trailing comment of the previous node. |
|
addTrailingComment$1(precedingNode, comment); |
|
} else if (followingNode) { |
|
addLeadingComment$1(followingNode, comment); |
|
} else if (enclosingNode) { |
|
addDanglingComment$1(enclosingNode, comment); |
|
} else { |
|
// There are no nodes, let's attach it to the root of the ast |
|
|
|
/* istanbul ignore next */ |
|
addDanglingComment$1(ast, comment); |
|
} |
|
} else { |
|
if (pluginHandleRemainingComment(comment, text, options, ast, isLastComment)) ; else if (precedingNode && followingNode) { |
|
// Otherwise, text exists both before and after the comment on |
|
// the same line. If there is both a preceding and following |
|
// node, use a tie-breaking algorithm to determine if it should |
|
// be attached to the next or previous node. In the last case, |
|
// simply attach the right node; |
|
var tieCount = tiesToBreak.length; |
|
|
|
if (tieCount > 0) { |
|
var lastTie = tiesToBreak[tieCount - 1]; |
|
|
|
if (lastTie.followingNode !== comment.followingNode) { |
|
breakTies(tiesToBreak, text, options); |
|
} |
|
} |
|
|
|
tiesToBreak.push(comment); |
|
} else if (precedingNode) { |
|
addTrailingComment$1(precedingNode, comment); |
|
} else if (followingNode) { |
|
addLeadingComment$1(followingNode, comment); |
|
} else if (enclosingNode) { |
|
addDanglingComment$1(enclosingNode, comment); |
|
} else { |
|
// There are no nodes, let's attach it to the root of the ast |
|
|
|
/* istanbul ignore next */ |
|
addDanglingComment$1(ast, comment); |
|
} |
|
} |
|
}); |
|
breakTies(tiesToBreak, text, options); |
|
comments.forEach(function (comment) { |
|
// These node references were useful for breaking ties, but we |
|
// don't need them anymore, and they create cycles in the AST that |
|
// may lead to infinite recursion if we don't delete them here. |
|
delete comment.precedingNode; |
|
delete comment.enclosingNode; |
|
delete comment.followingNode; |
|
}); |
|
} |
|
|
|
function breakTies(tiesToBreak, text, options) { |
|
var tieCount = tiesToBreak.length; |
|
|
|
if (tieCount === 0) { |
|
return; |
|
} |
|
|
|
var _tiesToBreak$ = tiesToBreak[0], |
|
precedingNode = _tiesToBreak$.precedingNode, |
|
followingNode = _tiesToBreak$.followingNode; |
|
var gapEndPos = options.locStart(followingNode); // Iterate backwards through tiesToBreak, examining the gaps |
|
// between the tied comments. In order to qualify as leading, a |
|
// comment must be separated from followingNode by an unbroken series of |
|
// gaps (or other comments). Gaps should only contain whitespace or open |
|
// parentheses. |
|
|
|
var indexOfFirstLeadingComment; |
|
|
|
for (indexOfFirstLeadingComment = tieCount; indexOfFirstLeadingComment > 0; --indexOfFirstLeadingComment) { |
|
var comment = tiesToBreak[indexOfFirstLeadingComment - 1]; |
|
assert.strictEqual(comment.precedingNode, precedingNode); |
|
assert.strictEqual(comment.followingNode, followingNode); |
|
var gap = text.slice(options.locEnd(comment), gapEndPos); |
|
|
|
if (/^[\s(]*$/.test(gap)) { |
|
gapEndPos = options.locStart(comment); |
|
} else { |
|
// The gap string contained something other than whitespace or open |
|
// parentheses. |
|
break; |
|
} |
|
} |
|
|
|
tiesToBreak.forEach(function (comment, i) { |
|
if (i < indexOfFirstLeadingComment) { |
|
addTrailingComment$1(precedingNode, comment); |
|
} else { |
|
addLeadingComment$1(followingNode, comment); |
|
} |
|
}); |
|
tiesToBreak.length = 0; |
|
} |
|
|
|
function printComment(commentPath, options) { |
|
var comment = commentPath.getValue(); |
|
comment.printed = true; |
|
return options.printer.printComment(commentPath, options); |
|
} |
|
|
|
function findExpressionIndexForComment(quasis, comment, options) { |
|
var startPos = options.locStart(comment) - 1; |
|
|
|
for (var i = 1; i < quasis.length; ++i) { |
|
if (startPos < getQuasiRange(quasis[i]).start) { |
|
return i - 1; |
|
} |
|
} // We haven't found it, it probably means that some of the locations are off. |
|
// Let's just return the first one. |
|
|
|
/* istanbul ignore next */ |
|
|
|
|
|
return 0; |
|
} |
|
|
|
function getQuasiRange(expr) { |
|
if (expr.start !== undefined) { |
|
// Babel |
|
return { |
|
start: expr.start, |
|
end: expr.end |
|
}; |
|
} // Flow |
|
|
|
|
|
return { |
|
start: expr.range[0], |
|
end: expr.range[1] |
|
}; |
|
} |
|
|
|
function printLeadingComment(commentPath, print, options) { |
|
var comment = commentPath.getValue(); |
|
var contents = printComment(commentPath, options); |
|
|
|
if (!contents) { |
|
return ""; |
|
} |
|
|
|
var isBlock = options.printer.isBlockComment && options.printer.isBlockComment(comment); // Leading block comments should see if they need to stay on the |
|
// same line or not. |
|
|
|
if (isBlock) { |
|
return concat$2([contents, hasNewline$1(options.originalText, options.locEnd(comment)) ? hardline$1 : " "]); |
|
} |
|
|
|
return concat$2([contents, hardline$1]); |
|
} |
|
|
|
function printTrailingComment(commentPath, print, options) { |
|
var comment = commentPath.getValue(); |
|
var contents = printComment(commentPath, options); |
|
|
|
if (!contents) { |
|
return ""; |
|
} |
|
|
|
var isBlock = options.printer.isBlockComment && options.printer.isBlockComment(comment); // We don't want the line to break |
|
// when the parentParentNode is a ClassDeclaration/-Expression |
|
// And the parentNode is in the superClass property |
|
|
|
var parentNode = commentPath.getNode(1); |
|
var parentParentNode = commentPath.getNode(2); |
|
var isParentSuperClass = parentParentNode && (parentParentNode.type === "ClassDeclaration" || parentParentNode.type === "ClassExpression") && parentParentNode.superClass === parentNode; |
|
|
|
if (hasNewline$1(options.originalText, options.locStart(comment), { |
|
backwards: true |
|
})) { |
|
// This allows comments at the end of nested structures: |
|
// { |
|
// x: 1, |
|
// y: 2 |
|
// // A comment |
|
// } |
|
// Those kinds of comments are almost always leading comments, but |
|
// here it doesn't go "outside" the block and turns it into a |
|
// trailing comment for `2`. We can simulate the above by checking |
|
// if this a comment on its own line; normal trailing comments are |
|
// always at the end of another expression. |
|
var isLineBeforeEmpty = isPreviousLineEmpty$2(options.originalText, comment, options.locStart); |
|
return lineSuffix$1(concat$2([hardline$1, isLineBeforeEmpty ? hardline$1 : "", contents])); |
|
} else if (isBlock || isParentSuperClass) { |
|
// Trailing block comments never need a newline |
|
return concat$2([" ", contents]); |
|
} |
|
|
|
return concat$2([lineSuffix$1(concat$2([" ", contents])), !isBlock ? breakParent$1 : ""]); |
|
} |
|
|
|
function printDanglingComments(path, options, sameIndent, filter) { |
|
var parts = []; |
|
var node = path.getValue(); |
|
|
|
if (!node || !node.comments) { |
|
return ""; |
|
} |
|
|
|
path.each(function (commentPath) { |
|
var comment = commentPath.getValue(); |
|
|
|
if (comment && !comment.leading && !comment.trailing && (!filter || filter(comment))) { |
|
parts.push(printComment(commentPath, options)); |
|
} |
|
}, "comments"); |
|
|
|
if (parts.length === 0) { |
|
return ""; |
|
} |
|
|
|
if (sameIndent) { |
|
return join$1(hardline$1, parts); |
|
} |
|
|
|
return indent$1(concat$2([hardline$1, join$1(hardline$1, parts)])); |
|
} |
|
|
|
function prependCursorPlaceholder(path, options, printed) { |
|
if (path.getNode() === options.cursorNode && path.getValue()) { |
|
return concat$2([cursor$2, printed, cursor$2]); |
|
} |
|
|
|
return printed; |
|
} |
|
|
|
function printComments(path, print, options, needsSemi) { |
|
var value = path.getValue(); |
|
var printed = print(path); |
|
var comments = value && value.comments; |
|
|
|
if (!comments || comments.length === 0) { |
|
return prependCursorPlaceholder(path, options, printed); |
|
} |
|
|
|
var leadingParts = []; |
|
var trailingParts = [needsSemi ? ";" : "", printed]; |
|
path.each(function (commentPath) { |
|
var comment = commentPath.getValue(); |
|
var leading = comment.leading, |
|
trailing = comment.trailing; |
|
|
|
if (leading) { |
|
var contents = printLeadingComment(commentPath, print, options); |
|
|
|
if (!contents) { |
|
return; |
|
} |
|
|
|
leadingParts.push(contents); |
|
var text = options.originalText; |
|
var index = skipNewline$1(text, options.locEnd(comment)); |
|
|
|
if (index !== false && hasNewline$1(text, index)) { |
|
leadingParts.push(hardline$1); |
|
} |
|
} else if (trailing) { |
|
trailingParts.push(printTrailingComment(commentPath, print, options)); |
|
} |
|
}, "comments"); |
|
return prependCursorPlaceholder(path, options, concat$2(leadingParts.concat(trailingParts))); |
|
} |
|
|
|
var comments = { |
|
attach: attach, |
|
printComments: printComments, |
|
printDanglingComments: printDanglingComments, |
|
getSortedChildNodes: getSortedChildNodes |
|
}; |
|
|
|
function FastPath(value) { |
|
assert.ok(this instanceof FastPath); |
|
this.stack = [value]; |
|
} // The name of the current property is always the penultimate element of |
|
// this.stack, and always a String. |
|
|
|
|
|
FastPath.prototype.getName = function getName() { |
|
var s = this.stack; |
|
var len = s.length; |
|
|
|
if (len > 1) { |
|
return s[len - 2]; |
|
} // Since the name is always a string, null is a safe sentinel value to |
|
// return if we do not know the name of the (root) value. |
|
|
|
/* istanbul ignore next */ |
|
|
|
|
|
return null; |
|
}; // The value of the current property is always the final element of |
|
// this.stack. |
|
|
|
|
|
FastPath.prototype.getValue = function getValue() { |
|
var s = this.stack; |
|
return s[s.length - 1]; |
|
}; |
|
|
|
function getNodeHelper(path, count) { |
|
var stackIndex = getNodeStackIndexHelper(path.stack, count); |
|
return stackIndex === -1 ? null : path.stack[stackIndex]; |
|
} |
|
|
|
function getNodeStackIndexHelper(stack, count) { |
|
for (var i = stack.length - 1; i >= 0; i -= 2) { |
|
var value = stack[i]; |
|
|
|
if (value && !Array.isArray(value) && --count < 0) { |
|
return i; |
|
} |
|
} |
|
|
|
return -1; |
|
} |
|
|
|
FastPath.prototype.getNode = function getNode(count) { |
|
return getNodeHelper(this, ~~count); |
|
}; |
|
|
|
FastPath.prototype.getParentNode = function getParentNode(count) { |
|
return getNodeHelper(this, ~~count + 1); |
|
}; // Temporarily push properties named by string arguments given after the |
|
// callback function onto this.stack, then call the callback with a |
|
// reference to this (modified) FastPath object. Note that the stack will |
|
// be restored to its original state after the callback is finished, so it |
|
// is probably a mistake to retain a reference to the path. |
|
|
|
|
|
FastPath.prototype.call = function call(callback |
|
/*, name1, name2, ... */ |
|
) { |
|
var s = this.stack; |
|
var origLen = s.length; |
|
var value = s[origLen - 1]; |
|
var argc = arguments.length; |
|
|
|
for (var i = 1; i < argc; ++i) { |
|
var name = arguments[i]; |
|
value = value[name]; |
|
s.push(name, value); |
|
} |
|
|
|
var result = callback(this); |
|
s.length = origLen; |
|
return result; |
|
}; |
|
|
|
FastPath.prototype.callParent = function callParent(callback, count) { |
|
var stackIndex = getNodeStackIndexHelper(this.stack, ~~count + 1); |
|
var parentValues = this.stack.splice(stackIndex + 1); |
|
var result = callback(this); |
|
Array.prototype.push.apply(this.stack, parentValues); |
|
return result; |
|
}; // Similar to FastPath.prototype.call, except that the value obtained by |
|
// accessing this.getValue()[name1][name2]... should be array-like. The |
|
// callback will be called with a reference to this path object for each |
|
// element of the array. |
|
|
|
|
|
FastPath.prototype.each = function each(callback |
|
/*, name1, name2, ... */ |
|
) { |
|
var s = this.stack; |
|
var origLen = s.length; |
|
var value = s[origLen - 1]; |
|
var argc = arguments.length; |
|
|
|
for (var i = 1; i < argc; ++i) { |
|
var name = arguments[i]; |
|
value = value[name]; |
|
s.push(name, value); |
|
} |
|
|
|
for (var _i = 0; _i < value.length; ++_i) { |
|
if (_i in value) { |
|
s.push(_i, value[_i]); // If the callback needs to know the value of i, call |
|
// path.getName(), assuming path is the parameter name. |
|
|
|
callback(this); |
|
s.length -= 2; |
|
} |
|
} |
|
|
|
s.length = origLen; |
|
}; // Similar to FastPath.prototype.each, except that the results of the |
|
// callback function invocations are stored in an array and returned at |
|
// the end of the iteration. |
|
|
|
|
|
FastPath.prototype.map = function map(callback |
|
/*, name1, name2, ... */ |
|
) { |
|
var s = this.stack; |
|
var origLen = s.length; |
|
var value = s[origLen - 1]; |
|
var argc = arguments.length; |
|
|
|
for (var i = 1; i < argc; ++i) { |
|
var name = arguments[i]; |
|
value = value[name]; |
|
s.push(name, value); |
|
} |
|
|
|
var result = new Array(value.length); |
|
|
|
for (var _i2 = 0; _i2 < value.length; ++_i2) { |
|
if (_i2 in value) { |
|
s.push(_i2, value[_i2]); |
|
result[_i2] = callback(this, _i2); |
|
s.length -= 2; |
|
} |
|
} |
|
|
|
s.length = origLen; |
|
return result; |
|
}; |
|
|
|
var fastPath = FastPath; |
|
|
|
var normalize$2 = options$1.normalize; |
|
|
|
function printSubtree(path, print, options, printAstToDoc) { |
|
if (options.printer.embed) { |
|
return options.printer.embed(path, print, function (text, partialNextOptions) { |
|
return textToDoc(text, partialNextOptions, options, printAstToDoc); |
|
}, options); |
|
} |
|
} |
|
|
|
function textToDoc(text, partialNextOptions, parentOptions, printAstToDoc) { |
|
var nextOptions = normalize$2(Object.assign({}, parentOptions, partialNextOptions, { |
|
parentParser: parentOptions.parser, |
|
embeddedInHtml: !!(parentOptions.embeddedInHtml || parentOptions.parser === "html" || parentOptions.parser === "vue" || parentOptions.parser === "angular" || parentOptions.parser === "lwc"), |
|
originalText: text |
|
}), { |
|
passThrough: true |
|
}); |
|
var result = parser.parse(text, nextOptions); |
|
var ast = result.ast; |
|
text = result.text; |
|
var astComments = ast.comments; |
|
delete ast.comments; |
|
comments.attach(astComments, ast, text, nextOptions); |
|
return printAstToDoc(ast, nextOptions); |
|
} |
|
|
|
var multiparser = { |
|
printSubtree: printSubtree |
|
}; |
|
|
|
var doc$1 = doc; |
|
var docBuilders$1 = doc$1.builders; |
|
var concat$3 = docBuilders$1.concat; |
|
var hardline$2 = docBuilders$1.hardline; |
|
var addAlignmentToDoc$1 = docBuilders$1.addAlignmentToDoc; |
|
var docUtils$1 = doc$1.utils; |
|
/** |
|
* Takes an abstract syntax tree (AST) and recursively converts it to a |
|
* document (series of printing primitives). |
|
* |
|
* This is done by descending down the AST recursively. The recursion |
|
* involves two functions that call each other: |
|
* |
|
* 1. printGenerically(), which is defined as an inner function here. |
|
* It basically takes care of node caching. |
|
* 2. callPluginPrintFunction(), which checks for some options, and |
|
* ultimately calls the print() function provided by the plugin. |
|
* |
|
* The plugin function will call printGenerically() again for child nodes |
|
* of the current node, which will do its housekeeping, then call the |
|
* plugin function again, and so on. |
|
* |
|
* All the while, these functions pass a "path" variable around, which |
|
* is a stack-like data structure (FastPath) that maintains the current |
|
* state of the recursion. It is called "path", because it represents |
|
* the path to the current node through the Abstract Syntax Tree. |
|
*/ |
|
|
|
function printAstToDoc(ast, options) { |
|
var alignmentSize = arguments.length > 2 && arguments[2] !== undefined ? arguments[2] : 0; |
|
var printer = options.printer; |
|
|
|
if (printer.preprocess) { |
|
ast = printer.preprocess(ast, options); |
|
} |
|
|
|
var cache = new Map(); |
|
|
|
function printGenerically(path, args) { |
|
var node = path.getValue(); |
|
var shouldCache = node && _typeof(node) === "object" && args === undefined; |
|
|
|
if (shouldCache && cache.has(node)) { |
|
return cache.get(node); |
|
} // We let JSXElement print its comments itself because it adds () around |
|
// UnionTypeAnnotation has to align the child without the comments |
|
|
|
|
|
var res; |
|
|
|
if (printer.willPrintOwnComments && printer.willPrintOwnComments(path, options)) { |
|
res = callPluginPrintFunction(path, options, printGenerically, args); |
|
} else { |
|
// printComments will call the plugin print function and check for |
|
// comments to print |
|
res = comments.printComments(path, function (p) { |
|
return callPluginPrintFunction(p, options, printGenerically, args); |
|
}, options, args && args.needsSemi); |
|
} |
|
|
|
if (shouldCache) { |
|
cache.set(node, res); |
|
} |
|
|
|
return res; |
|
} |
|
|
|
var doc = printGenerically(new fastPath(ast)); |
|
|
|
if (alignmentSize > 0) { |
|
// Add a hardline to make the indents take effect |
|
// It should be removed in index.js format() |
|
doc = addAlignmentToDoc$1(concat$3([hardline$2, doc]), alignmentSize, options.tabWidth); |
|
} |
|
|
|
docUtils$1.propagateBreaks(doc); |
|
return doc; |
|
} |
|
|
|
function callPluginPrintFunction(path, options, printPath, args) { |
|
assert.ok(path instanceof fastPath); |
|
var node = path.getValue(); |
|
var printer = options.printer; // Escape hatch |
|
|
|
if (printer.hasPrettierIgnore && printer.hasPrettierIgnore(path)) { |
|
return options.originalText.slice(options.locStart(node), options.locEnd(node)); |
|
} |
|
|
|
if (node) { |
|
try { |
|
// Potentially switch to a different parser |
|
var sub = multiparser.printSubtree(path, printPath, options, printAstToDoc); |
|
|
|
if (sub) { |
|
return sub; |
|
} |
|
} catch (error) { |
|
/* istanbul ignore if */ |
|
if (commonjsGlobal.PRETTIER_DEBUG) { |
|
throw error; |
|
} // Continue with current parser |
|
|
|
} |
|
} |
|
|
|
return printer.print(path, options, printPath, args); |
|
} |
|
|
|
var astToDoc = printAstToDoc; |
|
|
|
function findSiblingAncestors(startNodeAndParents, endNodeAndParents, opts) { |
|
var resultStartNode = startNodeAndParents.node; |
|
var resultEndNode = endNodeAndParents.node; |
|
|
|
if (resultStartNode === resultEndNode) { |
|
return { |
|
startNode: resultStartNode, |
|
endNode: resultEndNode |
|
}; |
|
} |
|
|
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = endNodeAndParents.parentNodes[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var endParent = _step.value; |
|
|
|
if (endParent.type !== "Program" && endParent.type !== "File" && opts.locStart(endParent) >= opts.locStart(startNodeAndParents.node)) { |
|
resultEndNode = endParent; |
|
} else { |
|
break; |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = startNodeAndParents.parentNodes[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var startParent = _step2.value; |
|
|
|
if (startParent.type !== "Program" && startParent.type !== "File" && opts.locEnd(startParent) <= opts.locEnd(endNodeAndParents.node)) { |
|
resultStartNode = startParent; |
|
} else { |
|
break; |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
|
|
return { |
|
startNode: resultStartNode, |
|
endNode: resultEndNode |
|
}; |
|
} |
|
|
|
function findNodeAtOffset(node, offset, options, predicate, parentNodes) { |
|
predicate = predicate || function () { |
|
return true; |
|
}; |
|
|
|
parentNodes = parentNodes || []; |
|
var start = options.locStart(node, options.locStart); |
|
var end = options.locEnd(node, options.locEnd); |
|
|
|
if (start <= offset && offset <= end) { |
|
var _iteratorNormalCompletion3 = true; |
|
var _didIteratorError3 = false; |
|
var _iteratorError3 = undefined; |
|
|
|
try { |
|
for (var _iterator3 = comments.getSortedChildNodes(node, options)[Symbol.iterator](), _step3; !(_iteratorNormalCompletion3 = (_step3 = _iterator3.next()).done); _iteratorNormalCompletion3 = true) { |
|
var childNode = _step3.value; |
|
var childResult = findNodeAtOffset(childNode, offset, options, predicate, [node].concat(parentNodes)); |
|
|
|
if (childResult) { |
|
return childResult; |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError3 = true; |
|
_iteratorError3 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion3 && _iterator3.return != null) { |
|
_iterator3.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError3) { |
|
throw _iteratorError3; |
|
} |
|
} |
|
} |
|
|
|
if (predicate(node)) { |
|
return { |
|
node: node, |
|
parentNodes: parentNodes |
|
}; |
|
} |
|
} |
|
} // See https://www.ecma-international.org/ecma-262/5.1/#sec-A.5 |
|
|
|
|
|
function isSourceElement(opts, node) { |
|
if (node == null) { |
|
return false; |
|
} // JS and JS like to avoid repetitions |
|
|
|
|
|
var jsSourceElements = ["FunctionDeclaration", "BlockStatement", "BreakStatement", "ContinueStatement", "DebuggerStatement", "DoWhileStatement", "EmptyStatement", "ExpressionStatement", "ForInStatement", "ForStatement", "IfStatement", "LabeledStatement", "ReturnStatement", "SwitchStatement", "ThrowStatement", "TryStatement", "VariableDeclaration", "WhileStatement", "WithStatement", "ClassDeclaration", // ES 2015 |
|
"ImportDeclaration", // Module |
|
"ExportDefaultDeclaration", // Module |
|
"ExportNamedDeclaration", // Module |
|
"ExportAllDeclaration", // Module |
|
"TypeAlias", // Flow |
|
"InterfaceDeclaration", // Flow, TypeScript |
|
"TypeAliasDeclaration", // TypeScript |
|
"ExportAssignment", // TypeScript |
|
"ExportDeclaration" // TypeScript |
|
]; |
|
var jsonSourceElements = ["ObjectExpression", "ArrayExpression", "StringLiteral", "NumericLiteral", "BooleanLiteral", "NullLiteral"]; |
|
var graphqlSourceElements = ["OperationDefinition", "FragmentDefinition", "VariableDefinition", "TypeExtensionDefinition", "ObjectTypeDefinition", "FieldDefinition", "DirectiveDefinition", "EnumTypeDefinition", "EnumValueDefinition", "InputValueDefinition", "InputObjectTypeDefinition", "SchemaDefinition", "OperationTypeDefinition", "InterfaceTypeDefinition", "UnionTypeDefinition", "ScalarTypeDefinition"]; |
|
|
|
switch (opts.parser) { |
|
case "flow": |
|
case "babel": |
|
case "typescript": |
|
return jsSourceElements.indexOf(node.type) > -1; |
|
|
|
case "json": |
|
return jsonSourceElements.indexOf(node.type) > -1; |
|
|
|
case "graphql": |
|
return graphqlSourceElements.indexOf(node.kind) > -1; |
|
|
|
case "vue": |
|
return node.tag !== "root"; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function calculateRange(text, opts, ast) { |
|
// Contract the range so that it has non-whitespace characters at its endpoints. |
|
// This ensures we can format a range that doesn't end on a node. |
|
var rangeStringOrig = text.slice(opts.rangeStart, opts.rangeEnd); |
|
var startNonWhitespace = Math.max(opts.rangeStart + rangeStringOrig.search(/\S/), opts.rangeStart); |
|
var endNonWhitespace; |
|
|
|
for (endNonWhitespace = opts.rangeEnd; endNonWhitespace > opts.rangeStart; --endNonWhitespace) { |
|
if (text[endNonWhitespace - 1].match(/\S/)) { |
|
break; |
|
} |
|
} |
|
|
|
var startNodeAndParents = findNodeAtOffset(ast, startNonWhitespace, opts, function (node) { |
|
return isSourceElement(opts, node); |
|
}); |
|
var endNodeAndParents = findNodeAtOffset(ast, endNonWhitespace, opts, function (node) { |
|
return isSourceElement(opts, node); |
|
}); |
|
|
|
if (!startNodeAndParents || !endNodeAndParents) { |
|
return { |
|
rangeStart: 0, |
|
rangeEnd: 0 |
|
}; |
|
} |
|
|
|
var siblingAncestors = findSiblingAncestors(startNodeAndParents, endNodeAndParents, opts); |
|
var startNode = siblingAncestors.startNode, |
|
endNode = siblingAncestors.endNode; |
|
var rangeStart = Math.min(opts.locStart(startNode, opts.locStart), opts.locStart(endNode, opts.locStart)); |
|
var rangeEnd = Math.max(opts.locEnd(startNode, opts.locEnd), opts.locEnd(endNode, opts.locEnd)); |
|
return { |
|
rangeStart: rangeStart, |
|
rangeEnd: rangeEnd |
|
}; |
|
} |
|
|
|
var rangeUtil = { |
|
calculateRange: calculateRange, |
|
findNodeAtOffset: findNodeAtOffset |
|
}; |
|
|
|
var diff = getCjsExportFromNamespace(index_es6); |
|
|
|
var normalizeOptions$1 = options$1.normalize; |
|
var guessEndOfLine$1 = endOfLine.guessEndOfLine, |
|
convertEndOfLineToChars$2 = endOfLine.convertEndOfLineToChars; |
|
var mapDoc$2 = doc.utils.mapDoc, |
|
_printDocToString = doc.printer.printDocToString, |
|
printDocToDebug = doc.debug.printDocToDebug; |
|
var UTF8BOM = 0xfeff; |
|
var CURSOR = Symbol("cursor"); |
|
var PLACEHOLDERS = { |
|
cursorOffset: "<<<PRETTIER_CURSOR>>>", |
|
rangeStart: "<<<PRETTIER_RANGE_START>>>", |
|
rangeEnd: "<<<PRETTIER_RANGE_END>>>" |
|
}; |
|
|
|
function ensureAllCommentsPrinted(astComments) { |
|
if (!astComments) { |
|
return; |
|
} |
|
|
|
for (var i = 0; i < astComments.length; ++i) { |
|
if (astComments[i].value.trim() === "prettier-ignore") { |
|
// If there's a prettier-ignore, we're not printing that sub-tree so we |
|
// don't know if the comments was printed or not. |
|
return; |
|
} |
|
} |
|
|
|
astComments.forEach(function (comment) { |
|
if (!comment.printed) { |
|
throw new Error('Comment "' + comment.value.trim() + '" was not printed. Please report this error!'); |
|
} |
|
|
|
delete comment.printed; |
|
}); |
|
} |
|
|
|
function attachComments(text, ast, opts) { |
|
var astComments = ast.comments; |
|
|
|
if (astComments) { |
|
delete ast.comments; |
|
comments.attach(astComments, ast, text, opts); |
|
} |
|
|
|
ast.tokens = []; |
|
opts.originalText = opts.parser === "yaml" ? text : text.trimRight(); |
|
return astComments; |
|
} |
|
|
|
function coreFormat(text, opts, addAlignmentSize) { |
|
if (!text || !text.trim().length) { |
|
return { |
|
formatted: "", |
|
cursorOffset: 0 |
|
}; |
|
} |
|
|
|
addAlignmentSize = addAlignmentSize || 0; |
|
var parsed = parser.parse(text, opts); |
|
var ast = parsed.ast; |
|
text = parsed.text; |
|
|
|
if (opts.cursorOffset >= 0) { |
|
var nodeResult = rangeUtil.findNodeAtOffset(ast, opts.cursorOffset, opts); |
|
|
|
if (nodeResult && nodeResult.node) { |
|
opts.cursorNode = nodeResult.node; |
|
} |
|
} |
|
|
|
var astComments = attachComments(text, ast, opts); |
|
var doc = astToDoc(ast, opts, addAlignmentSize); |
|
var eol = convertEndOfLineToChars$2(opts.endOfLine); |
|
|
|
var result = _printDocToString(opts.endOfLine === "lf" ? doc : mapDoc$2(doc, function (currentDoc) { |
|
return typeof currentDoc === "string" && currentDoc.indexOf("\n") !== -1 ? currentDoc.replace(/\n/g, eol) : currentDoc; |
|
}), opts); |
|
|
|
ensureAllCommentsPrinted(astComments); // Remove extra leading indentation as well as the added indentation after last newline |
|
|
|
if (addAlignmentSize > 0) { |
|
var trimmed = result.formatted.trim(); |
|
|
|
if (result.cursorNodeStart !== undefined) { |
|
result.cursorNodeStart -= result.formatted.indexOf(trimmed); |
|
} |
|
|
|
result.formatted = trimmed + convertEndOfLineToChars$2(opts.endOfLine); |
|
} |
|
|
|
if (opts.cursorOffset >= 0) { |
|
var oldCursorNodeStart; |
|
var oldCursorNodeText; |
|
var cursorOffsetRelativeToOldCursorNode; |
|
var newCursorNodeStart; |
|
var newCursorNodeText; |
|
|
|
if (opts.cursorNode && result.cursorNodeText) { |
|
oldCursorNodeStart = opts.locStart(opts.cursorNode); |
|
oldCursorNodeText = text.slice(oldCursorNodeStart, opts.locEnd(opts.cursorNode)); |
|
cursorOffsetRelativeToOldCursorNode = opts.cursorOffset - oldCursorNodeStart; |
|
newCursorNodeStart = result.cursorNodeStart; |
|
newCursorNodeText = result.cursorNodeText; |
|
} else { |
|
oldCursorNodeStart = 0; |
|
oldCursorNodeText = text; |
|
cursorOffsetRelativeToOldCursorNode = opts.cursorOffset; |
|
newCursorNodeStart = 0; |
|
newCursorNodeText = result.formatted; |
|
} |
|
|
|
if (oldCursorNodeText === newCursorNodeText) { |
|
return { |
|
formatted: result.formatted, |
|
cursorOffset: newCursorNodeStart + cursorOffsetRelativeToOldCursorNode |
|
}; |
|
} // diff old and new cursor node texts, with a special cursor |
|
// symbol inserted to find out where it moves to |
|
|
|
|
|
var oldCursorNodeCharArray = oldCursorNodeText.split(""); |
|
oldCursorNodeCharArray.splice(cursorOffsetRelativeToOldCursorNode, 0, CURSOR); |
|
var newCursorNodeCharArray = newCursorNodeText.split(""); |
|
var cursorNodeDiff = diff.diffArrays(oldCursorNodeCharArray, newCursorNodeCharArray); |
|
var cursorOffset = newCursorNodeStart; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = cursorNodeDiff[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var entry = _step.value; |
|
|
|
if (entry.removed) { |
|
if (entry.value.indexOf(CURSOR) > -1) { |
|
break; |
|
} |
|
} else { |
|
cursorOffset += entry.count; |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return { |
|
formatted: result.formatted, |
|
cursorOffset: cursorOffset |
|
}; |
|
} |
|
|
|
return { |
|
formatted: result.formatted |
|
}; |
|
} |
|
|
|
function formatRange(text, opts) { |
|
var parsed = parser.parse(text, opts); |
|
var ast = parsed.ast; |
|
text = parsed.text; |
|
var range = rangeUtil.calculateRange(text, opts, ast); |
|
var rangeStart = range.rangeStart; |
|
var rangeEnd = range.rangeEnd; |
|
var rangeString = text.slice(rangeStart, rangeEnd); // Try to extend the range backwards to the beginning of the line. |
|
// This is so we can detect indentation correctly and restore it. |
|
// Use `Math.min` since `lastIndexOf` returns 0 when `rangeStart` is 0 |
|
|
|
var rangeStart2 = Math.min(rangeStart, text.lastIndexOf("\n", rangeStart) + 1); |
|
var indentString = text.slice(rangeStart2, rangeStart); |
|
var alignmentSize = util.getAlignmentSize(indentString, opts.tabWidth); |
|
var rangeResult = coreFormat(rangeString, Object.assign({}, opts, { |
|
rangeStart: 0, |
|
rangeEnd: Infinity, |
|
// track the cursor offset only if it's within our range |
|
cursorOffset: opts.cursorOffset >= rangeStart && opts.cursorOffset < rangeEnd ? opts.cursorOffset - rangeStart : -1 |
|
}), alignmentSize); // Since the range contracts to avoid trailing whitespace, |
|
// we need to remove the newline that was inserted by the `format` call. |
|
|
|
var rangeTrimmed = rangeResult.formatted.trimRight(); |
|
var rangeLeft = text.slice(0, rangeStart); |
|
var rangeRight = text.slice(rangeEnd); |
|
var cursorOffset = opts.cursorOffset; |
|
|
|
if (opts.cursorOffset >= rangeEnd) { |
|
// handle the case where the cursor was past the end of the range |
|
cursorOffset = opts.cursorOffset - rangeEnd + (rangeStart + rangeTrimmed.length); |
|
} else if (rangeResult.cursorOffset !== undefined) { |
|
// handle the case where the cursor was in the range |
|
cursorOffset = rangeResult.cursorOffset + rangeStart; |
|
} // keep the cursor as it was if it was before the start of the range |
|
|
|
|
|
var formatted; |
|
|
|
if (opts.endOfLine === "lf") { |
|
formatted = rangeLeft + rangeTrimmed + rangeRight; |
|
} else { |
|
var eol = convertEndOfLineToChars$2(opts.endOfLine); |
|
|
|
if (cursorOffset >= 0) { |
|
var parts = [rangeLeft, rangeTrimmed, rangeRight]; |
|
var partIndex = 0; |
|
var partOffset = cursorOffset; |
|
|
|
while (partIndex < parts.length) { |
|
var part = parts[partIndex]; |
|
|
|
if (partOffset < part.length) { |
|
parts[partIndex] = parts[partIndex].slice(0, partOffset) + PLACEHOLDERS.cursorOffset + parts[partIndex].slice(partOffset); |
|
break; |
|
} |
|
|
|
partIndex++; |
|
partOffset -= part.length; |
|
} |
|
|
|
var newRangeLeft = parts[0], |
|
newRangeTrimmed = parts[1], |
|
newRangeRight = parts[2]; |
|
formatted = (newRangeLeft.replace(/\n/g, eol) + newRangeTrimmed + newRangeRight.replace(/\n/g, eol)).replace(PLACEHOLDERS.cursorOffset, function (_, index) { |
|
cursorOffset = index; |
|
return ""; |
|
}); |
|
} else { |
|
formatted = rangeLeft.replace(/\n/g, eol) + rangeTrimmed + rangeRight.replace(/\n/g, eol); |
|
} |
|
} |
|
|
|
return { |
|
formatted: formatted, |
|
cursorOffset: cursorOffset |
|
}; |
|
} |
|
|
|
function format(text, opts) { |
|
var selectedParser = parser.resolveParser(opts); |
|
var hasPragma = !selectedParser.hasPragma || selectedParser.hasPragma(text); |
|
|
|
if (opts.requirePragma && !hasPragma) { |
|
return { |
|
formatted: text |
|
}; |
|
} |
|
|
|
if (opts.endOfLine === "auto") { |
|
opts.endOfLine = guessEndOfLine$1(text); |
|
} |
|
|
|
var hasCursor = opts.cursorOffset >= 0; |
|
var hasRangeStart = opts.rangeStart > 0; |
|
var hasRangeEnd = opts.rangeEnd < text.length; // get rid of CR/CRLF parsing |
|
|
|
if (text.indexOf("\r") !== -1) { |
|
var offsetKeys = [hasCursor && "cursorOffset", hasRangeStart && "rangeStart", hasRangeEnd && "rangeEnd"].filter(Boolean).sort(function (aKey, bKey) { |
|
return opts[aKey] - opts[bKey]; |
|
}); |
|
|
|
for (var i = offsetKeys.length - 1; i >= 0; i--) { |
|
var key = offsetKeys[i]; |
|
text = text.slice(0, opts[key]) + PLACEHOLDERS[key] + text.slice(opts[key]); |
|
} |
|
|
|
text = text.replace(/\r\n?/g, "\n"); |
|
|
|
var _loop = function _loop(_i) { |
|
var key = offsetKeys[_i]; |
|
text = text.replace(PLACEHOLDERS[key], function (_, index) { |
|
opts[key] = index; |
|
return ""; |
|
}); |
|
}; |
|
|
|
for (var _i = 0; _i < offsetKeys.length; _i++) { |
|
_loop(_i); |
|
} |
|
} |
|
|
|
var hasUnicodeBOM = text.charCodeAt(0) === UTF8BOM; |
|
|
|
if (hasUnicodeBOM) { |
|
text = text.substring(1); |
|
|
|
if (hasCursor) { |
|
opts.cursorOffset++; |
|
} |
|
|
|
if (hasRangeStart) { |
|
opts.rangeStart++; |
|
} |
|
|
|
if (hasRangeEnd) { |
|
opts.rangeEnd++; |
|
} |
|
} |
|
|
|
if (!hasCursor) { |
|
opts.cursorOffset = -1; |
|
} |
|
|
|
if (opts.rangeStart < 0) { |
|
opts.rangeStart = 0; |
|
} |
|
|
|
if (opts.rangeEnd > text.length) { |
|
opts.rangeEnd = text.length; |
|
} |
|
|
|
var result = hasRangeStart || hasRangeEnd ? formatRange(text, opts) : coreFormat(opts.insertPragma && opts.printer.insertPragma && !hasPragma ? opts.printer.insertPragma(text) : text, opts); |
|
|
|
if (hasUnicodeBOM) { |
|
result.formatted = String.fromCharCode(UTF8BOM) + result.formatted; |
|
|
|
if (hasCursor) { |
|
result.cursorOffset++; |
|
} |
|
} |
|
|
|
return result; |
|
} |
|
|
|
var core = { |
|
formatWithCursor: function formatWithCursor(text, opts) { |
|
opts = normalizeOptions$1(opts); |
|
return format(text, opts); |
|
}, |
|
parse: function parse(text, opts, massage) { |
|
opts = normalizeOptions$1(opts); |
|
|
|
if (text.indexOf("\r") !== -1) { |
|
text = text.replace(/\r\n?/g, "\n"); |
|
} |
|
|
|
var parsed = parser.parse(text, opts); |
|
|
|
if (massage) { |
|
parsed.ast = massageAst(parsed.ast, opts); |
|
} |
|
|
|
return parsed; |
|
}, |
|
formatAST: function formatAST(ast, opts) { |
|
opts = normalizeOptions$1(opts); |
|
var doc = astToDoc(ast, opts); |
|
return _printDocToString(doc, opts); |
|
}, |
|
// Doesn't handle shebang for now |
|
formatDoc: function formatDoc(doc, opts) { |
|
var debug = printDocToDebug(doc); |
|
opts = normalizeOptions$1(Object.assign({}, opts, { |
|
parser: "babel" |
|
})); |
|
return format(debug, opts).formatted; |
|
}, |
|
printToDoc: function printToDoc(text, opts) { |
|
opts = normalizeOptions$1(opts); |
|
var parsed = parser.parse(text, opts); |
|
var ast = parsed.ast; |
|
text = parsed.text; |
|
attachComments(text, ast, opts); |
|
return astToDoc(ast, opts); |
|
}, |
|
printDocToString: function printDocToString(doc, opts) { |
|
return _printDocToString(doc, normalizeOptions$1(opts)); |
|
} |
|
}; |
|
|
|
var index = [ |
|
"a", |
|
"abbr", |
|
"acronym", |
|
"address", |
|
"applet", |
|
"area", |
|
"article", |
|
"aside", |
|
"audio", |
|
"b", |
|
"base", |
|
"basefont", |
|
"bdi", |
|
"bdo", |
|
"bgsound", |
|
"big", |
|
"blink", |
|
"blockquote", |
|
"body", |
|
"br", |
|
"button", |
|
"canvas", |
|
"caption", |
|
"center", |
|
"cite", |
|
"code", |
|
"col", |
|
"colgroup", |
|
"command", |
|
"content", |
|
"data", |
|
"datalist", |
|
"dd", |
|
"del", |
|
"details", |
|
"dfn", |
|
"dialog", |
|
"dir", |
|
"div", |
|
"dl", |
|
"dt", |
|
"element", |
|
"em", |
|
"embed", |
|
"fieldset", |
|
"figcaption", |
|
"figure", |
|
"font", |
|
"footer", |
|
"form", |
|
"frame", |
|
"frameset", |
|
"h1", |
|
"h2", |
|
"h3", |
|
"h4", |
|
"h5", |
|
"h6", |
|
"head", |
|
"header", |
|
"hgroup", |
|
"hr", |
|
"html", |
|
"i", |
|
"iframe", |
|
"image", |
|
"img", |
|
"input", |
|
"ins", |
|
"isindex", |
|
"kbd", |
|
"keygen", |
|
"label", |
|
"legend", |
|
"li", |
|
"link", |
|
"listing", |
|
"main", |
|
"map", |
|
"mark", |
|
"marquee", |
|
"math", |
|
"menu", |
|
"menuitem", |
|
"meta", |
|
"meter", |
|
"multicol", |
|
"nav", |
|
"nextid", |
|
"nobr", |
|
"noembed", |
|
"noframes", |
|
"noscript", |
|
"object", |
|
"ol", |
|
"optgroup", |
|
"option", |
|
"output", |
|
"p", |
|
"param", |
|
"picture", |
|
"plaintext", |
|
"pre", |
|
"progress", |
|
"q", |
|
"rb", |
|
"rbc", |
|
"rp", |
|
"rt", |
|
"rtc", |
|
"ruby", |
|
"s", |
|
"samp", |
|
"script", |
|
"section", |
|
"select", |
|
"shadow", |
|
"slot", |
|
"small", |
|
"source", |
|
"spacer", |
|
"span", |
|
"strike", |
|
"strong", |
|
"style", |
|
"sub", |
|
"summary", |
|
"sup", |
|
"svg", |
|
"table", |
|
"tbody", |
|
"td", |
|
"template", |
|
"textarea", |
|
"tfoot", |
|
"th", |
|
"thead", |
|
"time", |
|
"title", |
|
"tr", |
|
"track", |
|
"tt", |
|
"u", |
|
"ul", |
|
"var", |
|
"video", |
|
"wbr", |
|
"xmp" |
|
]; |
|
|
|
var htmlTagNames = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
'default': index |
|
}); |
|
|
|
var htmlTagNames$1 = getCjsExportFromNamespace(htmlTagNames); |
|
|
|
function clean(ast, newObj, parent) { |
|
["raw", // front-matter |
|
"raws", "sourceIndex", "source", "before", "after", "trailingComma"].forEach(function (name) { |
|
delete newObj[name]; |
|
}); |
|
|
|
if (ast.type === "yaml") { |
|
delete newObj.value; |
|
} // --insert-pragma |
|
|
|
|
|
if (ast.type === "css-comment" && parent.type === "css-root" && parent.nodes.length !== 0 && ( // first non-front-matter comment |
|
parent.nodes[0] === ast || (parent.nodes[0].type === "yaml" || parent.nodes[0].type === "toml") && parent.nodes[1] === ast)) { |
|
/** |
|
* something |
|
* |
|
* @format |
|
*/ |
|
delete newObj.text; // standalone pragma |
|
|
|
if (/^\*\s*@(format|prettier)\s*$/.test(ast.text)) { |
|
return null; |
|
} |
|
} |
|
|
|
if (ast.type === "media-query" || ast.type === "media-query-list" || ast.type === "media-feature-expression") { |
|
delete newObj.value; |
|
} |
|
|
|
if (ast.type === "css-rule") { |
|
delete newObj.params; |
|
} |
|
|
|
if (ast.type === "selector-combinator") { |
|
newObj.value = newObj.value.replace(/\s+/g, " "); |
|
} |
|
|
|
if (ast.type === "media-feature") { |
|
newObj.value = newObj.value.replace(/ /g, ""); |
|
} |
|
|
|
if (ast.type === "value-word" && (ast.isColor && ast.isHex || ["initial", "inherit", "unset", "revert"].indexOf(newObj.value.replace().toLowerCase()) !== -1) || ast.type === "media-feature" || ast.type === "selector-root-invalid" || ast.type === "selector-pseudo") { |
|
newObj.value = newObj.value.toLowerCase(); |
|
} |
|
|
|
if (ast.type === "css-decl") { |
|
newObj.prop = newObj.prop.toLowerCase(); |
|
} |
|
|
|
if (ast.type === "css-atrule" || ast.type === "css-import") { |
|
newObj.name = newObj.name.toLowerCase(); |
|
} |
|
|
|
if (ast.type === "value-number") { |
|
newObj.unit = newObj.unit.toLowerCase(); |
|
} |
|
|
|
if ((ast.type === "media-feature" || ast.type === "media-keyword" || ast.type === "media-type" || ast.type === "media-unknown" || ast.type === "media-url" || ast.type === "media-value" || ast.type === "selector-attribute" || ast.type === "selector-string" || ast.type === "selector-class" || ast.type === "selector-combinator" || ast.type === "value-string") && newObj.value) { |
|
newObj.value = cleanCSSStrings(newObj.value); |
|
} |
|
|
|
if (ast.type === "selector-attribute") { |
|
newObj.attribute = newObj.attribute.trim(); |
|
|
|
if (newObj.namespace) { |
|
if (typeof newObj.namespace === "string") { |
|
newObj.namespace = newObj.namespace.trim(); |
|
|
|
if (newObj.namespace.length === 0) { |
|
newObj.namespace = true; |
|
} |
|
} |
|
} |
|
|
|
if (newObj.value) { |
|
newObj.value = newObj.value.trim().replace(/^['"]|['"]$/g, ""); |
|
delete newObj.quoted; |
|
} |
|
} |
|
|
|
if ((ast.type === "media-value" || ast.type === "media-type" || ast.type === "value-number" || ast.type === "selector-root-invalid" || ast.type === "selector-class" || ast.type === "selector-combinator" || ast.type === "selector-tag") && newObj.value) { |
|
newObj.value = newObj.value.replace(/([\d.eE+-]+)([a-zA-Z]*)/g, function (match, numStr, unit) { |
|
var num = Number(numStr); |
|
return isNaN(num) ? match : num + unit.toLowerCase(); |
|
}); |
|
} |
|
|
|
if (ast.type === "selector-tag") { |
|
var lowercasedValue = ast.value.toLowerCase(); |
|
|
|
if (htmlTagNames$1.indexOf(lowercasedValue) !== -1) { |
|
newObj.value = lowercasedValue; |
|
} |
|
|
|
if (["from", "to"].indexOf(lowercasedValue) !== -1) { |
|
newObj.value = lowercasedValue; |
|
} |
|
} // Workaround when `postcss-values-parser` parse `not`, `and` or `or` keywords as `value-func` |
|
|
|
|
|
if (ast.type === "css-atrule" && ast.name.toLowerCase() === "supports") { |
|
delete newObj.value; |
|
} // Workaround for SCSS nested properties |
|
|
|
|
|
if (ast.type === "selector-unknown") { |
|
delete newObj.value; |
|
} |
|
} |
|
|
|
function cleanCSSStrings(value) { |
|
return value.replace(/'/g, '"').replace(/\\([^a-fA-F\d])/g, "$1"); |
|
} |
|
|
|
var clean_1 = clean; |
|
|
|
var _require$$0$builders$1 = doc.builders, |
|
hardline$3 = _require$$0$builders$1.hardline, |
|
literalline$1 = _require$$0$builders$1.literalline, |
|
concat$4 = _require$$0$builders$1.concat, |
|
markAsRoot$1 = _require$$0$builders$1.markAsRoot, |
|
mapDoc$3 = doc.utils.mapDoc; |
|
|
|
function embed(path, print, textToDoc |
|
/*, options */ |
|
) { |
|
var node = path.getValue(); |
|
|
|
if (node.type === "yaml") { |
|
return markAsRoot$1(concat$4(["---", hardline$3, node.value.trim() ? replaceNewlinesWithLiterallines(textToDoc(node.value, { |
|
parser: "yaml" |
|
})) : "", "---", hardline$3])); |
|
} |
|
|
|
return null; |
|
|
|
function replaceNewlinesWithLiterallines(doc) { |
|
return mapDoc$3(doc, function (currentDoc) { |
|
return typeof currentDoc === "string" && currentDoc.includes("\n") ? concat$4(currentDoc.split(/(\n)/g).map(function (v, i) { |
|
return i % 2 === 0 ? v : literalline$1; |
|
})) : currentDoc; |
|
}); |
|
} |
|
} |
|
|
|
var embed_1 = embed; |
|
|
|
var detectNewline = createCommonjsModule(function (module) { |
|
|
|
module.exports = function (str) { |
|
if (typeof str !== 'string') { |
|
throw new TypeError('Expected a string'); |
|
} |
|
|
|
var newlines = str.match(/(?:\r?\n)/g) || []; |
|
|
|
if (newlines.length === 0) { |
|
return null; |
|
} |
|
|
|
var crlf = newlines.filter(function (el) { |
|
return el === '\r\n'; |
|
}).length; |
|
var lf = newlines.length - crlf; |
|
return crlf > lf ? '\r\n' : '\n'; |
|
}; |
|
|
|
module.exports.graceful = function (str) { |
|
return module.exports(str) || '\n'; |
|
}; |
|
}); |
|
var detectNewline_1 = detectNewline.graceful; |
|
|
|
var build = createCommonjsModule(function (module, exports) { |
|
|
|
Object.defineProperty(exports, '__esModule', { |
|
value: true |
|
}); |
|
exports.extract = extract; |
|
exports.strip = strip; |
|
exports.parse = parse; |
|
exports.parseWithComments = parseWithComments; |
|
exports.print = print; |
|
|
|
function _os() { |
|
var data = require$$0$1; |
|
|
|
_os = function _os() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _detectNewline() { |
|
var data = _interopRequireDefault(detectNewline); |
|
|
|
_detectNewline = function _detectNewline() { |
|
return data; |
|
}; |
|
|
|
return data; |
|
} |
|
|
|
function _interopRequireDefault(obj) { |
|
return obj && obj.__esModule ? obj : { |
|
default: obj |
|
}; |
|
} |
|
/** |
|
* Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved. |
|
* |
|
* This source code is licensed under the MIT license found in the |
|
* LICENSE file in the root directory of this source tree. |
|
*/ |
|
|
|
|
|
var commentEndRe = /\*\/$/; |
|
var commentStartRe = /^\/\*\*/; |
|
var docblockRe = /^\s*(\/\*\*?(.|\r?\n)*?\*\/)/; |
|
var lineCommentRe = /(^|\s+)\/\/([^\r\n]*)/g; |
|
var ltrimNewlineRe = /^(\r?\n)+/; |
|
var multilineRe = /(?:^|\r?\n) *(@[^\r\n]*?) *\r?\n *(?![^@\r\n]*\/\/[^]*)([^@\r\n\s][^@\r\n]+?) *\r?\n/g; |
|
var propertyRe = /(?:^|\r?\n) *@(\S+) *([^\r\n]*)/g; |
|
var stringStartRe = /(\r?\n|^) *\* ?/g; |
|
|
|
function extract(contents) { |
|
var match = contents.match(docblockRe); |
|
return match ? match[0].trimLeft() : ''; |
|
} |
|
|
|
function strip(contents) { |
|
var match = contents.match(docblockRe); |
|
return match && match[0] ? contents.substring(match[0].length) : contents; |
|
} |
|
|
|
function parse(docblock) { |
|
return parseWithComments(docblock).pragmas; |
|
} |
|
|
|
function parseWithComments(docblock) { |
|
var line = (0, _detectNewline().default)(docblock) || _os().EOL; |
|
|
|
docblock = docblock.replace(commentStartRe, '').replace(commentEndRe, '').replace(stringStartRe, '$1'); // Normalize multi-line directives |
|
|
|
var prev = ''; |
|
|
|
while (prev !== docblock) { |
|
prev = docblock; |
|
docblock = docblock.replace(multilineRe, "".concat(line, "$1 $2").concat(line)); |
|
} |
|
|
|
docblock = docblock.replace(ltrimNewlineRe, '').trimRight(); |
|
var result = Object.create(null); |
|
var comments = docblock.replace(propertyRe, '').replace(ltrimNewlineRe, '').trimRight(); |
|
var match; |
|
|
|
while (match = propertyRe.exec(docblock)) { |
|
// strip linecomments from pragmas |
|
var nextPragma = match[2].replace(lineCommentRe, ''); |
|
|
|
if (typeof result[match[1]] === 'string' || Array.isArray(result[match[1]])) { |
|
result[match[1]] = [].concat(result[match[1]], nextPragma); |
|
} else { |
|
result[match[1]] = nextPragma; |
|
} |
|
} |
|
|
|
return { |
|
comments: comments, |
|
pragmas: result |
|
}; |
|
} |
|
|
|
function print(_ref) { |
|
var _ref$comments = _ref.comments, |
|
comments = _ref$comments === void 0 ? '' : _ref$comments, |
|
_ref$pragmas = _ref.pragmas, |
|
pragmas = _ref$pragmas === void 0 ? {} : _ref$pragmas; |
|
|
|
var line = (0, _detectNewline().default)(comments) || _os().EOL; |
|
|
|
var head = '/**'; |
|
var start = ' *'; |
|
var tail = ' */'; |
|
var keys = Object.keys(pragmas); |
|
var printedObject = keys.map(function (key) { |
|
return printKeyValues(key, pragmas[key]); |
|
}).reduce(function (arr, next) { |
|
return arr.concat(next); |
|
}, []).map(function (keyValue) { |
|
return start + ' ' + keyValue + line; |
|
}).join(''); |
|
|
|
if (!comments) { |
|
if (keys.length === 0) { |
|
return ''; |
|
} |
|
|
|
if (keys.length === 1 && !Array.isArray(pragmas[keys[0]])) { |
|
var value = pragmas[keys[0]]; |
|
return "".concat(head, " ").concat(printKeyValues(keys[0], value)[0]).concat(tail); |
|
} |
|
} |
|
|
|
var printedComments = comments.split(line).map(function (textLine) { |
|
return "".concat(start, " ").concat(textLine); |
|
}).join(line) + line; |
|
return head + line + (comments ? printedComments : '') + (comments && keys.length ? start + line : '') + printedObject + tail; |
|
} |
|
|
|
function printKeyValues(key, valueOrArray) { |
|
return [].concat(valueOrArray).map(function (value) { |
|
return "@".concat(key, " ").concat(value).trim(); |
|
}); |
|
} |
|
}); |
|
unwrapExports(build); |
|
var build_1 = build.extract; |
|
var build_2 = build.strip; |
|
var build_3 = build.parse; |
|
var build_4 = build.parseWithComments; |
|
var build_5 = build.print; |
|
|
|
function hasPragma(text) { |
|
var pragmas = Object.keys(build.parse(build.extract(text))); |
|
return pragmas.indexOf("prettier") !== -1 || pragmas.indexOf("format") !== -1; |
|
} |
|
|
|
function insertPragma(text) { |
|
var parsedDocblock = build.parseWithComments(build.extract(text)); |
|
var pragmas = Object.assign({ |
|
format: "" |
|
}, parsedDocblock.pragmas); |
|
var newDocblock = build.print({ |
|
pragmas: pragmas, |
|
comments: parsedDocblock.comments.replace(/^(\s+?\r?\n)+/, "") // remove leading newlines |
|
|
|
}).replace(/(\r\n|\r)/g, "\n"); // normalise newlines (mitigate use of os.EOL by jest-docblock) |
|
|
|
var strippedText = build.strip(text); |
|
var separatingNewlines = strippedText.startsWith("\n") ? "\n" : "\n\n"; |
|
return newDocblock + separatingNewlines + strippedText; |
|
} |
|
|
|
var pragma = { |
|
hasPragma: hasPragma, |
|
insertPragma: insertPragma |
|
}; |
|
|
|
var DELIMITER_MAP = { |
|
"---": "yaml", |
|
"+++": "toml" |
|
}; |
|
|
|
function parse$1(text) { |
|
var delimiterRegex = Object.keys(DELIMITER_MAP).map(escapeStringRegexp).join("|"); |
|
var match = text.match( // trailing spaces after delimiters are allowed |
|
new RegExp("^(".concat(delimiterRegex, ")[^\\n\\S]*\\n(?:([\\s\\S]*?)\\n)?\\1[^\\n\\S]*(\\n|$)"))); |
|
|
|
if (match === null) { |
|
return { |
|
frontMatter: null, |
|
content: text |
|
}; |
|
} |
|
|
|
var raw = match[0].replace(/\n$/, ""); |
|
var delimiter = match[1]; |
|
var value = match[2]; |
|
return { |
|
frontMatter: { |
|
type: DELIMITER_MAP[delimiter], |
|
value: value, |
|
raw: raw |
|
}, |
|
content: match[0].replace(/[^\n]/g, " ") + text.slice(match[0].length) |
|
}; |
|
} |
|
|
|
var frontMatter = parse$1; |
|
|
|
function hasPragma$1(text) { |
|
return pragma.hasPragma(frontMatter(text).content); |
|
} |
|
|
|
function insertPragma$1(text) { |
|
var _parseFrontMatter = frontMatter(text), |
|
frontMatter$1 = _parseFrontMatter.frontMatter, |
|
content = _parseFrontMatter.content; |
|
|
|
return (frontMatter$1 ? frontMatter$1.raw + "\n\n" : "") + pragma.insertPragma(content); |
|
} |
|
|
|
var pragma$1 = { |
|
hasPragma: hasPragma$1, |
|
insertPragma: insertPragma$1 |
|
}; |
|
|
|
var colorAdjusterFunctions = ["red", "green", "blue", "alpha", "a", "rgb", "hue", "h", "saturation", "s", "lightness", "l", "whiteness", "w", "blackness", "b", "tint", "shade", "blend", "blenda", "contrast", "hsl", "hsla", "hwb", "hwba"]; |
|
|
|
function getAncestorCounter(path, typeOrTypes) { |
|
var types = [].concat(typeOrTypes); |
|
var counter = -1; |
|
var ancestorNode; |
|
|
|
while (ancestorNode = path.getParentNode(++counter)) { |
|
if (types.indexOf(ancestorNode.type) !== -1) { |
|
return counter; |
|
} |
|
} |
|
|
|
return -1; |
|
} |
|
|
|
function getAncestorNode(path, typeOrTypes) { |
|
var counter = getAncestorCounter(path, typeOrTypes); |
|
return counter === -1 ? null : path.getParentNode(counter); |
|
} |
|
|
|
function getPropOfDeclNode(path) { |
|
var declAncestorNode = getAncestorNode(path, "css-decl"); |
|
return declAncestorNode && declAncestorNode.prop && declAncestorNode.prop.toLowerCase(); |
|
} |
|
|
|
function isSCSS(parser, text) { |
|
var hasExplicitParserChoice = parser === "less" || parser === "scss"; |
|
var IS_POSSIBLY_SCSS = /(\w\s*: [^}:]+|#){|@import[^\n]+(url|,)/; |
|
return hasExplicitParserChoice ? parser === "scss" : IS_POSSIBLY_SCSS.test(text); |
|
} |
|
|
|
function isWideKeywords(value) { |
|
return ["initial", "inherit", "unset", "revert"].indexOf(value.toLowerCase()) !== -1; |
|
} |
|
|
|
function isKeyframeAtRuleKeywords(path, value) { |
|
var atRuleAncestorNode = getAncestorNode(path, "css-atrule"); |
|
return atRuleAncestorNode && atRuleAncestorNode.name && atRuleAncestorNode.name.toLowerCase().endsWith("keyframes") && ["from", "to"].indexOf(value.toLowerCase()) !== -1; |
|
} |
|
|
|
function maybeToLowerCase(value) { |
|
return value.includes("$") || value.includes("@") || value.includes("#") || value.startsWith("%") || value.startsWith("--") || value.startsWith(":--") || value.includes("(") && value.includes(")") ? value : value.toLowerCase(); |
|
} |
|
|
|
function insideValueFunctionNode(path, functionName) { |
|
var funcAncestorNode = getAncestorNode(path, "value-func"); |
|
return funcAncestorNode && funcAncestorNode.value && funcAncestorNode.value.toLowerCase() === functionName; |
|
} |
|
|
|
function insideICSSRuleNode(path) { |
|
var ruleAncestorNode = getAncestorNode(path, "css-rule"); |
|
return ruleAncestorNode && ruleAncestorNode.raws && ruleAncestorNode.raws.selector && (ruleAncestorNode.raws.selector.startsWith(":import") || ruleAncestorNode.raws.selector.startsWith(":export")); |
|
} |
|
|
|
function insideAtRuleNode(path, atRuleNameOrAtRuleNames) { |
|
var atRuleNames = [].concat(atRuleNameOrAtRuleNames); |
|
var atRuleAncestorNode = getAncestorNode(path, "css-atrule"); |
|
return atRuleAncestorNode && atRuleNames.indexOf(atRuleAncestorNode.name.toLowerCase()) !== -1; |
|
} |
|
|
|
function insideURLFunctionInImportAtRuleNode(path) { |
|
var node = path.getValue(); |
|
var atRuleAncestorNode = getAncestorNode(path, "css-atrule"); |
|
return atRuleAncestorNode && atRuleAncestorNode.name === "import" && node.groups[0].value === "url" && node.groups.length === 2; |
|
} |
|
|
|
function isURLFunctionNode(node) { |
|
return node.type === "value-func" && node.value.toLowerCase() === "url"; |
|
} |
|
|
|
function isLastNode(path, node) { |
|
var parentNode = path.getParentNode(); |
|
|
|
if (!parentNode) { |
|
return false; |
|
} |
|
|
|
var nodes = parentNode.nodes; |
|
return nodes && nodes.indexOf(node) === nodes.length - 1; |
|
} |
|
|
|
function isHTMLTag(value) { |
|
return htmlTagNames$1.indexOf(value.toLowerCase()) !== -1; |
|
} |
|
|
|
function isDetachedRulesetDeclarationNode(node) { |
|
// If a Less file ends up being parsed with the SCSS parser, Less |
|
// variable declarations will be parsed as atrules with names ending |
|
// with a colon, so keep the original case then. |
|
if (!node.selector) { |
|
return false; |
|
} |
|
|
|
return typeof node.selector === "string" && /^@.+:.*$/.test(node.selector) || node.selector.value && /^@.+:.*$/.test(node.selector.value); |
|
} |
|
|
|
function isForKeywordNode(node) { |
|
return node.type === "value-word" && ["from", "through", "end"].indexOf(node.value) !== -1; |
|
} |
|
|
|
function isIfElseKeywordNode(node) { |
|
return node.type === "value-word" && ["and", "or", "not"].indexOf(node.value) !== -1; |
|
} |
|
|
|
function isEachKeywordNode(node) { |
|
return node.type === "value-word" && node.value === "in"; |
|
} |
|
|
|
function isMultiplicationNode(node) { |
|
return node.type === "value-operator" && node.value === "*"; |
|
} |
|
|
|
function isDivisionNode(node) { |
|
return node.type === "value-operator" && node.value === "/"; |
|
} |
|
|
|
function isAdditionNode(node) { |
|
return node.type === "value-operator" && node.value === "+"; |
|
} |
|
|
|
function isSubtractionNode(node) { |
|
return node.type === "value-operator" && node.value === "-"; |
|
} |
|
|
|
function isModuloNode(node) { |
|
return node.type === "value-operator" && node.value === "%"; |
|
} |
|
|
|
function isMathOperatorNode(node) { |
|
return isMultiplicationNode(node) || isDivisionNode(node) || isAdditionNode(node) || isSubtractionNode(node) || isModuloNode(node); |
|
} |
|
|
|
function isEqualityOperatorNode(node) { |
|
return node.type === "value-word" && ["==", "!="].indexOf(node.value) !== -1; |
|
} |
|
|
|
function isRelationalOperatorNode(node) { |
|
return node.type === "value-word" && ["<", ">", "<=", ">="].indexOf(node.value) !== -1; |
|
} |
|
|
|
function isSCSSControlDirectiveNode(node) { |
|
return node.type === "css-atrule" && ["if", "else", "for", "each", "while"].indexOf(node.name) !== -1; |
|
} |
|
|
|
function isSCSSNestedPropertyNode(node) { |
|
if (!node.selector) { |
|
return false; |
|
} |
|
|
|
return node.selector.replace(/\/\*.*?\*\//, "").replace(/\/\/.*?\n/, "").trim().endsWith(":"); |
|
} |
|
|
|
function isDetachedRulesetCallNode(node) { |
|
return node.raws && node.raws.params && /^\(\s*\)$/.test(node.raws.params); |
|
} |
|
|
|
function isTemplatePlaceholderNode(node) { |
|
return node.name.startsWith("prettier-placeholder"); |
|
} |
|
|
|
function isTemplatePropNode(node) { |
|
return node.prop.startsWith("@prettier-placeholder"); |
|
} |
|
|
|
function isPostcssSimpleVarNode(currentNode, nextNode) { |
|
return currentNode.value === "$$" && currentNode.type === "value-func" && nextNode && nextNode.type === "value-word" && !nextNode.raws.before; |
|
} |
|
|
|
function hasComposesNode(node) { |
|
return node.value && node.value.type === "value-root" && node.value.group && node.value.group.type === "value-value" && node.prop.toLowerCase() === "composes"; |
|
} |
|
|
|
function hasParensAroundNode(node) { |
|
return node.value && node.value.group && node.value.group.group && node.value.group.group.type === "value-paren_group" && node.value.group.group.open !== null && node.value.group.group.close !== null; |
|
} |
|
|
|
function hasEmptyRawBefore(node) { |
|
return node.raws && node.raws.before === ""; |
|
} |
|
|
|
function isKeyValuePairNode(node) { |
|
return node.type === "value-comma_group" && node.groups && node.groups[1] && node.groups[1].type === "value-colon"; |
|
} |
|
|
|
function isKeyValuePairInParenGroupNode(node) { |
|
return node.type === "value-paren_group" && node.groups && node.groups[0] && isKeyValuePairNode(node.groups[0]); |
|
} |
|
|
|
function isSCSSMapItemNode(path) { |
|
var node = path.getValue(); // Ignore empty item (i.e. `$key: ()`) |
|
|
|
if (node.groups.length === 0) { |
|
return false; |
|
} |
|
|
|
var parentParentNode = path.getParentNode(1); // Check open parens contain key/value pair (i.e. `(key: value)` and `(key: (value, other-value)`) |
|
|
|
if (!isKeyValuePairInParenGroupNode(node) && !(parentParentNode && isKeyValuePairInParenGroupNode(parentParentNode))) { |
|
return false; |
|
} |
|
|
|
var declNode = getAncestorNode(path, "css-decl"); // SCSS map declaration (i.e. `$map: (key: value, other-key: other-value)`) |
|
|
|
if (declNode && declNode.prop && declNode.prop.startsWith("$")) { |
|
return true; |
|
} // List as value of key inside SCSS map (i.e. `$map: (key: (value other-value other-other-value))`) |
|
|
|
|
|
if (isKeyValuePairInParenGroupNode(parentParentNode)) { |
|
return true; |
|
} // SCSS Map is argument of function (i.e. `func((key: value, other-key: other-value))`) |
|
|
|
|
|
if (parentParentNode.type === "value-func") { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isInlineValueCommentNode(node) { |
|
return node.type === "value-comment" && node.inline; |
|
} |
|
|
|
function isHashNode(node) { |
|
return node.type === "value-word" && node.value === "#"; |
|
} |
|
|
|
function isLeftCurlyBraceNode(node) { |
|
return node.type === "value-word" && node.value === "{"; |
|
} |
|
|
|
function isRightCurlyBraceNode(node) { |
|
return node.type === "value-word" && node.value === "}"; |
|
} |
|
|
|
function isWordNode(node) { |
|
return ["value-word", "value-atword"].indexOf(node.type) !== -1; |
|
} |
|
|
|
function isColonNode(node) { |
|
return node.type === "value-colon"; |
|
} |
|
|
|
function isMediaAndSupportsKeywords(node) { |
|
return node.value && ["not", "and", "or"].indexOf(node.value.toLowerCase()) !== -1; |
|
} |
|
|
|
function isColorAdjusterFuncNode(node) { |
|
if (node.type !== "value-func") { |
|
return false; |
|
} |
|
|
|
return colorAdjusterFunctions.indexOf(node.value.toLowerCase()) !== -1; |
|
} |
|
|
|
var utils$2 = { |
|
getAncestorCounter: getAncestorCounter, |
|
getAncestorNode: getAncestorNode, |
|
getPropOfDeclNode: getPropOfDeclNode, |
|
maybeToLowerCase: maybeToLowerCase, |
|
insideValueFunctionNode: insideValueFunctionNode, |
|
insideICSSRuleNode: insideICSSRuleNode, |
|
insideAtRuleNode: insideAtRuleNode, |
|
insideURLFunctionInImportAtRuleNode: insideURLFunctionInImportAtRuleNode, |
|
isKeyframeAtRuleKeywords: isKeyframeAtRuleKeywords, |
|
isHTMLTag: isHTMLTag, |
|
isWideKeywords: isWideKeywords, |
|
isSCSS: isSCSS, |
|
isLastNode: isLastNode, |
|
isSCSSControlDirectiveNode: isSCSSControlDirectiveNode, |
|
isDetachedRulesetDeclarationNode: isDetachedRulesetDeclarationNode, |
|
isRelationalOperatorNode: isRelationalOperatorNode, |
|
isEqualityOperatorNode: isEqualityOperatorNode, |
|
isMultiplicationNode: isMultiplicationNode, |
|
isDivisionNode: isDivisionNode, |
|
isAdditionNode: isAdditionNode, |
|
isSubtractionNode: isSubtractionNode, |
|
isModuloNode: isModuloNode, |
|
isMathOperatorNode: isMathOperatorNode, |
|
isEachKeywordNode: isEachKeywordNode, |
|
isForKeywordNode: isForKeywordNode, |
|
isURLFunctionNode: isURLFunctionNode, |
|
isIfElseKeywordNode: isIfElseKeywordNode, |
|
hasComposesNode: hasComposesNode, |
|
hasParensAroundNode: hasParensAroundNode, |
|
hasEmptyRawBefore: hasEmptyRawBefore, |
|
isSCSSNestedPropertyNode: isSCSSNestedPropertyNode, |
|
isDetachedRulesetCallNode: isDetachedRulesetCallNode, |
|
isTemplatePlaceholderNode: isTemplatePlaceholderNode, |
|
isTemplatePropNode: isTemplatePropNode, |
|
isPostcssSimpleVarNode: isPostcssSimpleVarNode, |
|
isKeyValuePairNode: isKeyValuePairNode, |
|
isKeyValuePairInParenGroupNode: isKeyValuePairInParenGroupNode, |
|
isSCSSMapItemNode: isSCSSMapItemNode, |
|
isInlineValueCommentNode: isInlineValueCommentNode, |
|
isHashNode: isHashNode, |
|
isLeftCurlyBraceNode: isLeftCurlyBraceNode, |
|
isRightCurlyBraceNode: isRightCurlyBraceNode, |
|
isWordNode: isWordNode, |
|
isColonNode: isColonNode, |
|
isMediaAndSupportsKeywords: isMediaAndSupportsKeywords, |
|
isColorAdjusterFuncNode: isColorAdjusterFuncNode |
|
}; |
|
|
|
var insertPragma$2 = pragma$1.insertPragma; |
|
var printNumber$1 = util.printNumber, |
|
printString$1 = util.printString, |
|
hasIgnoreComment$1 = util.hasIgnoreComment, |
|
hasNewline$2 = util.hasNewline; |
|
var isNextLineEmpty$2 = utilShared.isNextLineEmpty; |
|
var _require$$3$builders = doc.builders, |
|
concat$5 = _require$$3$builders.concat, |
|
join$2 = _require$$3$builders.join, |
|
line$1 = _require$$3$builders.line, |
|
hardline$4 = _require$$3$builders.hardline, |
|
softline$1 = _require$$3$builders.softline, |
|
group$1 = _require$$3$builders.group, |
|
fill$2 = _require$$3$builders.fill, |
|
indent$2 = _require$$3$builders.indent, |
|
dedent$1 = _require$$3$builders.dedent, |
|
ifBreak$1 = _require$$3$builders.ifBreak, |
|
removeLines$1 = doc.utils.removeLines; |
|
var getAncestorNode$1 = utils$2.getAncestorNode, |
|
getPropOfDeclNode$1 = utils$2.getPropOfDeclNode, |
|
maybeToLowerCase$1 = utils$2.maybeToLowerCase, |
|
insideValueFunctionNode$1 = utils$2.insideValueFunctionNode, |
|
insideICSSRuleNode$1 = utils$2.insideICSSRuleNode, |
|
insideAtRuleNode$1 = utils$2.insideAtRuleNode, |
|
insideURLFunctionInImportAtRuleNode$1 = utils$2.insideURLFunctionInImportAtRuleNode, |
|
isKeyframeAtRuleKeywords$1 = utils$2.isKeyframeAtRuleKeywords, |
|
isHTMLTag$1 = utils$2.isHTMLTag, |
|
isWideKeywords$1 = utils$2.isWideKeywords, |
|
isSCSS$1 = utils$2.isSCSS, |
|
isLastNode$1 = utils$2.isLastNode, |
|
isSCSSControlDirectiveNode$1 = utils$2.isSCSSControlDirectiveNode, |
|
isDetachedRulesetDeclarationNode$1 = utils$2.isDetachedRulesetDeclarationNode, |
|
isRelationalOperatorNode$1 = utils$2.isRelationalOperatorNode, |
|
isEqualityOperatorNode$1 = utils$2.isEqualityOperatorNode, |
|
isMultiplicationNode$1 = utils$2.isMultiplicationNode, |
|
isDivisionNode$1 = utils$2.isDivisionNode, |
|
isAdditionNode$1 = utils$2.isAdditionNode, |
|
isSubtractionNode$1 = utils$2.isSubtractionNode, |
|
isMathOperatorNode$1 = utils$2.isMathOperatorNode, |
|
isEachKeywordNode$1 = utils$2.isEachKeywordNode, |
|
isForKeywordNode$1 = utils$2.isForKeywordNode, |
|
isURLFunctionNode$1 = utils$2.isURLFunctionNode, |
|
isIfElseKeywordNode$1 = utils$2.isIfElseKeywordNode, |
|
hasComposesNode$1 = utils$2.hasComposesNode, |
|
hasParensAroundNode$1 = utils$2.hasParensAroundNode, |
|
hasEmptyRawBefore$1 = utils$2.hasEmptyRawBefore, |
|
isKeyValuePairNode$1 = utils$2.isKeyValuePairNode, |
|
isDetachedRulesetCallNode$1 = utils$2.isDetachedRulesetCallNode, |
|
isTemplatePlaceholderNode$1 = utils$2.isTemplatePlaceholderNode, |
|
isTemplatePropNode$1 = utils$2.isTemplatePropNode, |
|
isPostcssSimpleVarNode$1 = utils$2.isPostcssSimpleVarNode, |
|
isSCSSMapItemNode$1 = utils$2.isSCSSMapItemNode, |
|
isInlineValueCommentNode$1 = utils$2.isInlineValueCommentNode, |
|
isHashNode$1 = utils$2.isHashNode, |
|
isLeftCurlyBraceNode$1 = utils$2.isLeftCurlyBraceNode, |
|
isRightCurlyBraceNode$1 = utils$2.isRightCurlyBraceNode, |
|
isWordNode$1 = utils$2.isWordNode, |
|
isColonNode$1 = utils$2.isColonNode, |
|
isMediaAndSupportsKeywords$1 = utils$2.isMediaAndSupportsKeywords, |
|
isColorAdjusterFuncNode$1 = utils$2.isColorAdjusterFuncNode; |
|
|
|
function shouldPrintComma(options) { |
|
switch (options.trailingComma) { |
|
case "all": |
|
case "es5": |
|
return true; |
|
|
|
case "none": |
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function genericPrint(path, options, print) { |
|
var node = path.getValue(); |
|
/* istanbul ignore if */ |
|
|
|
if (!node) { |
|
return ""; |
|
} |
|
|
|
if (typeof node === "string") { |
|
return node; |
|
} |
|
|
|
switch (node.type) { |
|
case "yaml": |
|
case "toml": |
|
return concat$5([node.raw, hardline$4]); |
|
|
|
case "css-root": |
|
{ |
|
var nodes = printNodeSequence(path, options, print); |
|
|
|
if (nodes.parts.length) { |
|
return concat$5([nodes, hardline$4]); |
|
} |
|
|
|
return nodes; |
|
} |
|
|
|
case "css-comment": |
|
{ |
|
if (node.raws.content) { |
|
return node.raws.content; |
|
} |
|
|
|
var text = options.originalText.slice(options.locStart(node), options.locEnd(node)); |
|
var rawText = node.raws.text || node.text; // Workaround a bug where the location is off. |
|
// https://github.com/postcss/postcss-scss/issues/63 |
|
|
|
if (text.indexOf(rawText) === -1) { |
|
if (node.raws.inline) { |
|
return concat$5(["// ", rawText]); |
|
} |
|
|
|
return concat$5(["/* ", rawText, " */"]); |
|
} |
|
|
|
return text; |
|
} |
|
|
|
case "css-rule": |
|
{ |
|
return concat$5([path.call(print, "selector"), node.important ? " !important" : "", node.nodes ? concat$5([" {", node.nodes.length > 0 ? indent$2(concat$5([hardline$4, printNodeSequence(path, options, print)])) : "", hardline$4, "}", isDetachedRulesetDeclarationNode$1(node) ? ";" : ""]) : ";"]); |
|
} |
|
|
|
case "css-decl": |
|
{ |
|
var parentNode = path.getParentNode(); |
|
return concat$5([node.raws.before.replace(/[\s;]/g, ""), insideICSSRuleNode$1(path) ? node.prop : maybeToLowerCase$1(node.prop), node.raws.between.trim() === ":" ? ":" : node.raws.between.trim(), node.extend ? "" : " ", hasComposesNode$1(node) ? removeLines$1(path.call(print, "value")) : path.call(print, "value"), node.raws.important ? node.raws.important.replace(/\s*!\s*important/i, " !important") : node.important ? " !important" : "", node.raws.scssDefault ? node.raws.scssDefault.replace(/\s*!default/i, " !default") : node.scssDefault ? " !default" : "", node.raws.scssGlobal ? node.raws.scssGlobal.replace(/\s*!global/i, " !global") : node.scssGlobal ? " !global" : "", node.nodes ? concat$5([" {", indent$2(concat$5([softline$1, printNodeSequence(path, options, print)])), softline$1, "}"]) : isTemplatePropNode$1(node) && !parentNode.raws.semicolon && options.originalText[options.locEnd(node) - 1] !== ";" ? "" : ";"]); |
|
} |
|
|
|
case "css-atrule": |
|
{ |
|
var _parentNode = path.getParentNode(); |
|
|
|
return concat$5(["@", // If a Less file ends up being parsed with the SCSS parser, Less |
|
// variable declarations will be parsed as at-rules with names ending |
|
// with a colon, so keep the original case then. |
|
isDetachedRulesetCallNode$1(node) || node.name.endsWith(":") ? node.name : maybeToLowerCase$1(node.name), node.params ? concat$5([isDetachedRulesetCallNode$1(node) ? "" : isTemplatePlaceholderNode$1(node) && /^\s*\n/.test(node.raws.afterName) ? /^\s*\n\s*\n/.test(node.raws.afterName) ? concat$5([hardline$4, hardline$4]) : hardline$4 : " ", path.call(print, "params")]) : "", node.selector ? indent$2(concat$5([" ", path.call(print, "selector")])) : "", node.value ? group$1(concat$5([" ", path.call(print, "value"), isSCSSControlDirectiveNode$1(node) ? hasParensAroundNode$1(node) ? " " : line$1 : ""])) : node.name === "else" ? " " : "", node.nodes ? concat$5([isSCSSControlDirectiveNode$1(node) ? "" : " ", "{", indent$2(concat$5([node.nodes.length > 0 ? softline$1 : "", printNodeSequence(path, options, print)])), softline$1, "}"]) : isTemplatePlaceholderNode$1(node) && !_parentNode.raws.semicolon && options.originalText[options.locEnd(node) - 1] !== ";" ? "" : ";"]); |
|
} |
|
// postcss-media-query-parser |
|
|
|
case "media-query-list": |
|
{ |
|
var parts = []; |
|
path.each(function (childPath) { |
|
var node = childPath.getValue(); |
|
|
|
if (node.type === "media-query" && node.value === "") { |
|
return; |
|
} |
|
|
|
parts.push(childPath.call(print)); |
|
}, "nodes"); |
|
return group$1(indent$2(join$2(line$1, parts))); |
|
} |
|
|
|
case "media-query": |
|
{ |
|
return concat$5([join$2(" ", path.map(print, "nodes")), isLastNode$1(path, node) ? "" : ","]); |
|
} |
|
|
|
case "media-type": |
|
{ |
|
return adjustNumbers(adjustStrings(node.value, options)); |
|
} |
|
|
|
case "media-feature-expression": |
|
{ |
|
if (!node.nodes) { |
|
return node.value; |
|
} |
|
|
|
return concat$5(["(", concat$5(path.map(print, "nodes")), ")"]); |
|
} |
|
|
|
case "media-feature": |
|
{ |
|
return maybeToLowerCase$1(adjustStrings(node.value.replace(/ +/g, " "), options)); |
|
} |
|
|
|
case "media-colon": |
|
{ |
|
return concat$5([node.value, " "]); |
|
} |
|
|
|
case "media-value": |
|
{ |
|
return adjustNumbers(adjustStrings(node.value, options)); |
|
} |
|
|
|
case "media-keyword": |
|
{ |
|
return adjustStrings(node.value, options); |
|
} |
|
|
|
case "media-url": |
|
{ |
|
return adjustStrings(node.value.replace(/^url\(\s+/gi, "url(").replace(/\s+\)$/gi, ")"), options); |
|
} |
|
|
|
case "media-unknown": |
|
{ |
|
return node.value; |
|
} |
|
// postcss-selector-parser |
|
|
|
case "selector-root": |
|
{ |
|
return group$1(concat$5([insideAtRuleNode$1(path, "custom-selector") ? concat$5([getAncestorNode$1(path, "css-atrule").customSelector, line$1]) : "", join$2(concat$5([",", insideAtRuleNode$1(path, ["extend", "custom-selector", "nest"]) ? line$1 : hardline$4]), path.map(print, "nodes"))])); |
|
} |
|
|
|
case "selector-selector": |
|
{ |
|
return group$1(indent$2(concat$5(path.map(print, "nodes")))); |
|
} |
|
|
|
case "selector-comment": |
|
{ |
|
return node.value; |
|
} |
|
|
|
case "selector-string": |
|
{ |
|
return adjustStrings(node.value, options); |
|
} |
|
|
|
case "selector-tag": |
|
{ |
|
var _parentNode2 = path.getParentNode(); |
|
|
|
var index = _parentNode2 && _parentNode2.nodes.indexOf(node); |
|
|
|
var prevNode = index && _parentNode2.nodes[index - 1]; |
|
return concat$5([node.namespace ? concat$5([node.namespace === true ? "" : node.namespace.trim(), "|"]) : "", prevNode.type === "selector-nesting" ? node.value : adjustNumbers(isHTMLTag$1(node.value) || isKeyframeAtRuleKeywords$1(path, node.value) ? node.value.toLowerCase() : node.value)]); |
|
} |
|
|
|
case "selector-id": |
|
{ |
|
return concat$5(["#", node.value]); |
|
} |
|
|
|
case "selector-class": |
|
{ |
|
return concat$5([".", adjustNumbers(adjustStrings(node.value, options))]); |
|
} |
|
|
|
case "selector-attribute": |
|
{ |
|
return concat$5(["[", node.namespace ? concat$5([node.namespace === true ? "" : node.namespace.trim(), "|"]) : "", node.attribute.trim(), node.operator ? node.operator : "", node.value ? quoteAttributeValue(adjustStrings(node.value.trim(), options), options) : "", node.insensitive ? " i" : "", "]"]); |
|
} |
|
|
|
case "selector-combinator": |
|
{ |
|
if (node.value === "+" || node.value === ">" || node.value === "~" || node.value === ">>>") { |
|
var _parentNode3 = path.getParentNode(); |
|
|
|
var _leading = _parentNode3.type === "selector-selector" && _parentNode3.nodes[0] === node ? "" : line$1; |
|
|
|
return concat$5([_leading, node.value, isLastNode$1(path, node) ? "" : " "]); |
|
} |
|
|
|
var leading = node.value.trim().startsWith("(") ? line$1 : ""; |
|
var value = adjustNumbers(adjustStrings(node.value.trim(), options)) || line$1; |
|
return concat$5([leading, value]); |
|
} |
|
|
|
case "selector-universal": |
|
{ |
|
return concat$5([node.namespace ? concat$5([node.namespace === true ? "" : node.namespace.trim(), "|"]) : "", node.value]); |
|
} |
|
|
|
case "selector-pseudo": |
|
{ |
|
return concat$5([maybeToLowerCase$1(node.value), node.nodes && node.nodes.length > 0 ? concat$5(["(", join$2(", ", path.map(print, "nodes")), ")"]) : ""]); |
|
} |
|
|
|
case "selector-nesting": |
|
{ |
|
return node.value; |
|
} |
|
|
|
case "selector-unknown": |
|
{ |
|
var ruleAncestorNode = getAncestorNode$1(path, "css-rule"); // Nested SCSS property |
|
|
|
if (ruleAncestorNode && ruleAncestorNode.isSCSSNesterProperty) { |
|
return adjustNumbers(adjustStrings(maybeToLowerCase$1(node.value), options)); |
|
} |
|
|
|
return node.value; |
|
} |
|
// postcss-values-parser |
|
|
|
case "value-value": |
|
case "value-root": |
|
{ |
|
return path.call(print, "group"); |
|
} |
|
|
|
case "value-comment": |
|
{ |
|
return concat$5([node.inline ? "//" : "/*", node.value, node.inline ? "" : "*/"]); |
|
} |
|
|
|
case "value-comma_group": |
|
{ |
|
var _parentNode4 = path.getParentNode(); |
|
|
|
var parentParentNode = path.getParentNode(1); |
|
var declAncestorProp = getPropOfDeclNode$1(path); |
|
var isGridValue = declAncestorProp && _parentNode4.type === "value-value" && (declAncestorProp === "grid" || declAncestorProp.startsWith("grid-template")); |
|
var atRuleAncestorNode = getAncestorNode$1(path, "css-atrule"); |
|
var isControlDirective = atRuleAncestorNode && isSCSSControlDirectiveNode$1(atRuleAncestorNode); |
|
var printed = path.map(print, "groups"); |
|
var _parts = []; |
|
var insideURLFunction = insideValueFunctionNode$1(path, "url"); |
|
var insideSCSSInterpolationInString = false; |
|
var didBreak = false; |
|
|
|
for (var i = 0; i < node.groups.length; ++i) { |
|
_parts.push(printed[i]); // Ignore value inside `url()` |
|
|
|
|
|
if (insideURLFunction) { |
|
continue; |
|
} |
|
|
|
var iPrevNode = node.groups[i - 1]; |
|
var iNode = node.groups[i]; |
|
var iNextNode = node.groups[i + 1]; |
|
var iNextNextNode = node.groups[i + 2]; // Ignore after latest node (i.e. before semicolon) |
|
|
|
if (!iNextNode) { |
|
continue; |
|
} // Ignore spaces before/after string interpolation (i.e. `"#{my-fn("_")}"`) |
|
|
|
|
|
var isStartSCSSInterpolationInString = iNode.type === "value-string" && iNode.value.startsWith("#{"); |
|
var isEndingSCSSInterpolationInString = insideSCSSInterpolationInString && iNextNode.type === "value-string" && iNextNode.value.endsWith("}"); |
|
|
|
if (isStartSCSSInterpolationInString || isEndingSCSSInterpolationInString) { |
|
insideSCSSInterpolationInString = !insideSCSSInterpolationInString; |
|
continue; |
|
} |
|
|
|
if (insideSCSSInterpolationInString) { |
|
continue; |
|
} // Ignore colon (i.e. `:`) |
|
|
|
|
|
if (isColonNode$1(iNode) || isColonNode$1(iNextNode)) { |
|
continue; |
|
} // Ignore `@` in Less (i.e. `@@var;`) |
|
|
|
|
|
if (iNode.type === "value-atword" && iNode.value === "") { |
|
continue; |
|
} // Ignore `~` in Less (i.e. `content: ~"^//* some horrible but needed css hack";`) |
|
|
|
|
|
if (iNode.value === "~") { |
|
continue; |
|
} // Ignore escape `\` |
|
|
|
|
|
if (iNode.value && iNode.value.indexOf("\\") !== -1 && iNextNode && iNextNode.type !== "value-comment") { |
|
continue; |
|
} // Ignore escaped `/` |
|
|
|
|
|
if (iPrevNode && iPrevNode.value && iPrevNode.value.indexOf("\\") === iPrevNode.value.length - 1 && iNode.type === "value-operator" && iNode.value === "/") { |
|
continue; |
|
} // Ignore `\` (i.e. `$variable: \@small;`) |
|
|
|
|
|
if (iNode.value === "\\") { |
|
continue; |
|
} // Ignore `$$` (i.e. `background-color: $$(style)Color;`) |
|
|
|
|
|
if (isPostcssSimpleVarNode$1(iNode, iNextNode)) { |
|
continue; |
|
} // Ignore spaces after `#` and after `{` and before `}` in SCSS interpolation (i.e. `#{variable}`) |
|
|
|
|
|
if (isHashNode$1(iNode) || isLeftCurlyBraceNode$1(iNode) || isRightCurlyBraceNode$1(iNextNode) || isLeftCurlyBraceNode$1(iNextNode) && hasEmptyRawBefore$1(iNextNode) || isRightCurlyBraceNode$1(iNode) && hasEmptyRawBefore$1(iNextNode)) { |
|
continue; |
|
} // Ignore css variables and interpolation in SCSS (i.e. `--#{$var}`) |
|
|
|
|
|
if (iNode.value === "--" && isHashNode$1(iNextNode)) { |
|
continue; |
|
} // Formatting math operations |
|
|
|
|
|
var isMathOperator = isMathOperatorNode$1(iNode); |
|
var isNextMathOperator = isMathOperatorNode$1(iNextNode); // Print spaces before and after math operators beside SCSS interpolation as is |
|
// (i.e. `#{$var}+5`, `#{$var} +5`, `#{$var}+ 5`, `#{$var} + 5`) |
|
// (i.e. `5+#{$var}`, `5 +#{$var}`, `5+ #{$var}`, `5 + #{$var}`) |
|
|
|
if ((isMathOperator && isHashNode$1(iNextNode) || isNextMathOperator && isRightCurlyBraceNode$1(iNode)) && hasEmptyRawBefore$1(iNextNode)) { |
|
continue; |
|
} // Print spaces before and after addition and subtraction math operators as is in `calc` function |
|
// due to the fact that it is not valid syntax |
|
// (i.e. `calc(1px+1px)`, `calc(1px+ 1px)`, `calc(1px +1px)`, `calc(1px + 1px)`) |
|
|
|
|
|
if (insideValueFunctionNode$1(path, "calc") && (isAdditionNode$1(iNode) || isAdditionNode$1(iNextNode) || isSubtractionNode$1(iNode) || isSubtractionNode$1(iNextNode)) && hasEmptyRawBefore$1(iNextNode)) { |
|
continue; |
|
} // Print spaces after `+` and `-` in color adjuster functions as is (e.g. `color(red l(+ 20%))`) |
|
// Adjusters with signed numbers (e.g. `color(red l(+20%))`) output as-is. |
|
|
|
|
|
var isColorAdjusterNode = (isAdditionNode$1(iNode) || isSubtractionNode$1(iNode)) && i === 0 && (iNextNode.type === "value-number" || iNextNode.isHex) && parentParentNode && isColorAdjusterFuncNode$1(parentParentNode) && !hasEmptyRawBefore$1(iNextNode); |
|
var requireSpaceBeforeOperator = iNextNextNode && iNextNextNode.type === "value-func" || iNextNextNode && isWordNode$1(iNextNextNode) || iNode.type === "value-func" || isWordNode$1(iNode); |
|
var requireSpaceAfterOperator = iNextNode.type === "value-func" || isWordNode$1(iNextNode) || iPrevNode && iPrevNode.type === "value-func" || iPrevNode && isWordNode$1(iPrevNode); // Formatting `/`, `+`, `-` sign |
|
|
|
if (!(isMultiplicationNode$1(iNextNode) || isMultiplicationNode$1(iNode)) && !insideValueFunctionNode$1(path, "calc") && !isColorAdjusterNode && (isDivisionNode$1(iNextNode) && !requireSpaceBeforeOperator || isDivisionNode$1(iNode) && !requireSpaceAfterOperator || isAdditionNode$1(iNextNode) && !requireSpaceBeforeOperator || isAdditionNode$1(iNode) && !requireSpaceAfterOperator || isSubtractionNode$1(iNextNode) || isSubtractionNode$1(iNode)) && (hasEmptyRawBefore$1(iNextNode) || isMathOperator && (!iPrevNode || iPrevNode && isMathOperatorNode$1(iPrevNode)))) { |
|
continue; |
|
} // Add `hardline` after inline comment (i.e. `// comment\n foo: bar;`) |
|
|
|
|
|
if (isInlineValueCommentNode$1(iNode)) { |
|
_parts.push(hardline$4); |
|
|
|
continue; |
|
} // Handle keywords in SCSS control directive |
|
|
|
|
|
if (isControlDirective && (isEqualityOperatorNode$1(iNextNode) || isRelationalOperatorNode$1(iNextNode) || isIfElseKeywordNode$1(iNextNode) || isEachKeywordNode$1(iNode) || isForKeywordNode$1(iNode))) { |
|
_parts.push(" "); |
|
|
|
continue; |
|
} // At-rule `namespace` should be in one line |
|
|
|
|
|
if (atRuleAncestorNode && atRuleAncestorNode.name.toLowerCase() === "namespace") { |
|
_parts.push(" "); |
|
|
|
continue; |
|
} // Formatting `grid` property |
|
|
|
|
|
if (isGridValue) { |
|
if (iNode.source && iNextNode.source && iNode.source.start.line !== iNextNode.source.start.line) { |
|
_parts.push(hardline$4); |
|
|
|
didBreak = true; |
|
} else { |
|
_parts.push(" "); |
|
} |
|
|
|
continue; |
|
} // Add `space` before next math operation |
|
// Note: `grip` property have `/` delimiter and it is not math operation, so |
|
// `grid` property handles above |
|
|
|
|
|
if (isNextMathOperator) { |
|
_parts.push(" "); |
|
|
|
continue; |
|
} // Be default all values go through `line` |
|
|
|
|
|
_parts.push(line$1); |
|
} |
|
|
|
if (didBreak) { |
|
_parts.unshift(hardline$4); |
|
} |
|
|
|
if (isControlDirective) { |
|
return group$1(indent$2(concat$5(_parts))); |
|
} // Indent is not needed for import url when url is very long |
|
// and node has two groups |
|
// when type is value-comma_group |
|
// example @import url("verylongurl") projection,tv |
|
|
|
|
|
if (insideURLFunctionInImportAtRuleNode$1(path)) { |
|
return group$1(fill$2(_parts)); |
|
} |
|
|
|
return group$1(indent$2(fill$2(_parts))); |
|
} |
|
|
|
case "value-paren_group": |
|
{ |
|
var _parentNode5 = path.getParentNode(); |
|
|
|
if (_parentNode5 && isURLFunctionNode$1(_parentNode5) && (node.groups.length === 1 || node.groups.length > 0 && node.groups[0].type === "value-comma_group" && node.groups[0].groups.length > 0 && node.groups[0].groups[0].type === "value-word" && node.groups[0].groups[0].value.startsWith("data:"))) { |
|
return concat$5([node.open ? path.call(print, "open") : "", join$2(",", path.map(print, "groups")), node.close ? path.call(print, "close") : ""]); |
|
} |
|
|
|
if (!node.open) { |
|
var _printed = path.map(print, "groups"); |
|
|
|
var res = []; |
|
|
|
for (var _i = 0; _i < _printed.length; _i++) { |
|
if (_i !== 0) { |
|
res.push(concat$5([",", line$1])); |
|
} |
|
|
|
res.push(_printed[_i]); |
|
} |
|
|
|
return group$1(indent$2(fill$2(res))); |
|
} |
|
|
|
var isSCSSMapItem = isSCSSMapItemNode$1(path); |
|
return group$1(concat$5([node.open ? path.call(print, "open") : "", indent$2(concat$5([softline$1, join$2(concat$5([",", line$1]), path.map(function (childPath) { |
|
var node = childPath.getValue(); |
|
var printed = print(childPath); // Key/Value pair in open paren already indented |
|
|
|
if (isKeyValuePairNode$1(node) && node.type === "value-comma_group" && node.groups && node.groups[2] && node.groups[2].type === "value-paren_group") { |
|
printed.contents.contents.parts[1] = group$1(printed.contents.contents.parts[1]); |
|
return group$1(dedent$1(printed)); |
|
} |
|
|
|
return printed; |
|
}, "groups"))])), ifBreak$1(isSCSS$1(options.parser, options.originalText) && isSCSSMapItem && shouldPrintComma(options) ? "," : ""), softline$1, node.close ? path.call(print, "close") : ""]), { |
|
shouldBreak: isSCSSMapItem |
|
}); |
|
} |
|
|
|
case "value-func": |
|
{ |
|
return concat$5([node.value, insideAtRuleNode$1(path, "supports") && isMediaAndSupportsKeywords$1(node) ? " " : "", path.call(print, "group")]); |
|
} |
|
|
|
case "value-paren": |
|
{ |
|
return node.value; |
|
} |
|
|
|
case "value-number": |
|
{ |
|
return concat$5([printCssNumber(node.value), maybeToLowerCase$1(node.unit)]); |
|
} |
|
|
|
case "value-operator": |
|
{ |
|
return node.value; |
|
} |
|
|
|
case "value-word": |
|
{ |
|
if (node.isColor && node.isHex || isWideKeywords$1(node.value)) { |
|
return node.value.toLowerCase(); |
|
} |
|
|
|
return node.value; |
|
} |
|
|
|
case "value-colon": |
|
{ |
|
return concat$5([node.value, // Don't add spaces on `:` in `url` function (i.e. `url(fbglyph: cross-outline, fig-white)`) |
|
insideValueFunctionNode$1(path, "url") ? "" : line$1]); |
|
} |
|
|
|
case "value-comma": |
|
{ |
|
return concat$5([node.value, " "]); |
|
} |
|
|
|
case "value-string": |
|
{ |
|
return printString$1(node.raws.quote + node.value + node.raws.quote, options); |
|
} |
|
|
|
case "value-atword": |
|
{ |
|
return concat$5(["@", node.value]); |
|
} |
|
|
|
case "value-unicode-range": |
|
{ |
|
return node.value; |
|
} |
|
|
|
case "value-unknown": |
|
{ |
|
return node.value; |
|
} |
|
|
|
default: |
|
/* istanbul ignore next */ |
|
throw new Error("Unknown postcss type ".concat(JSON.stringify(node.type))); |
|
} |
|
} |
|
|
|
function printNodeSequence(path, options, print) { |
|
var node = path.getValue(); |
|
var parts = []; |
|
var i = 0; |
|
path.map(function (pathChild) { |
|
var prevNode = node.nodes[i - 1]; |
|
|
|
if (prevNode && prevNode.type === "css-comment" && prevNode.text.trim() === "prettier-ignore") { |
|
var childNode = pathChild.getValue(); |
|
parts.push(options.originalText.slice(options.locStart(childNode), options.locEnd(childNode))); |
|
} else { |
|
parts.push(pathChild.call(print)); |
|
} |
|
|
|
if (i !== node.nodes.length - 1) { |
|
if (node.nodes[i + 1].type === "css-comment" && !hasNewline$2(options.originalText, options.locStart(node.nodes[i + 1]), { |
|
backwards: true |
|
}) && node.nodes[i].type !== "yaml" && node.nodes[i].type !== "toml" || node.nodes[i + 1].type === "css-atrule" && node.nodes[i + 1].name === "else" && node.nodes[i].type !== "css-comment") { |
|
parts.push(" "); |
|
} else { |
|
parts.push(hardline$4); |
|
|
|
if (isNextLineEmpty$2(options.originalText, pathChild.getValue(), options) && node.nodes[i].type !== "yaml" && node.nodes[i].type !== "toml") { |
|
parts.push(hardline$4); |
|
} |
|
} |
|
} |
|
|
|
i++; |
|
}, "nodes"); |
|
return concat$5(parts); |
|
} |
|
|
|
var STRING_REGEX$1 = /(['"])(?:(?!\1)[^\\]|\\[\s\S])*\1/g; |
|
var NUMBER_REGEX = /(?:\d*\.\d+|\d+\.?)(?:[eE][+-]?\d+)?/g; |
|
var STANDARD_UNIT_REGEX = /[a-zA-Z]+/g; |
|
var WORD_PART_REGEX = /[$@]?[a-zA-Z_\u0080-\uFFFF][\w\-\u0080-\uFFFF]*/g; |
|
var ADJUST_NUMBERS_REGEX = RegExp(STRING_REGEX$1.source + "|" + "(".concat(WORD_PART_REGEX.source, ")?") + "(".concat(NUMBER_REGEX.source, ")") + "(".concat(STANDARD_UNIT_REGEX.source, ")?"), "g"); |
|
|
|
function adjustStrings(value, options) { |
|
return value.replace(STRING_REGEX$1, function (match) { |
|
return printString$1(match, options); |
|
}); |
|
} |
|
|
|
function quoteAttributeValue(value, options) { |
|
var quote = options.singleQuote ? "'" : '"'; |
|
return value.includes('"') || value.includes("'") ? value : quote + value + quote; |
|
} |
|
|
|
function adjustNumbers(value) { |
|
return value.replace(ADJUST_NUMBERS_REGEX, function (match, quote, wordPart, number, unit) { |
|
return !wordPart && number ? (wordPart || "") + printCssNumber(number) + maybeToLowerCase$1(unit || "") : match; |
|
}); |
|
} |
|
|
|
function printCssNumber(rawNumber) { |
|
return printNumber$1(rawNumber) // Remove trailing `.0`. |
|
.replace(/\.0(?=$|e)/, ""); |
|
} |
|
|
|
var printerPostcss = { |
|
print: genericPrint, |
|
embed: embed_1, |
|
insertPragma: insertPragma$2, |
|
hasPrettierIgnore: hasIgnoreComment$1, |
|
massageAstNode: clean_1 |
|
}; |
|
|
|
var CATEGORY_COMMON = "Common"; // format based on https://github.com/prettier/prettier/blob/master/src/main/core-options.js |
|
|
|
var commonOptions = { |
|
bracketSpacing: { |
|
since: "0.0.0", |
|
category: CATEGORY_COMMON, |
|
type: "boolean", |
|
default: true, |
|
description: "Print spaces between brackets.", |
|
oppositeDescription: "Do not print spaces between brackets." |
|
}, |
|
singleQuote: { |
|
since: "0.0.0", |
|
category: CATEGORY_COMMON, |
|
type: "boolean", |
|
default: false, |
|
description: "Use single quotes instead of double quotes." |
|
}, |
|
proseWrap: { |
|
since: "1.8.2", |
|
category: CATEGORY_COMMON, |
|
type: "choice", |
|
default: [{ |
|
since: "1.8.2", |
|
value: true |
|
}, { |
|
since: "1.9.0", |
|
value: "preserve" |
|
}], |
|
description: "How to wrap prose.", |
|
choices: [{ |
|
since: "1.9.0", |
|
value: "always", |
|
description: "Wrap prose if it exceeds the print width." |
|
}, { |
|
since: "1.9.0", |
|
value: "never", |
|
description: "Do not wrap prose." |
|
}, { |
|
since: "1.9.0", |
|
value: "preserve", |
|
description: "Wrap prose as-is." |
|
}, { |
|
value: false, |
|
deprecated: "1.9.0", |
|
redirect: "never" |
|
}, { |
|
value: true, |
|
deprecated: "1.9.0", |
|
redirect: "always" |
|
}] |
|
} |
|
}; |
|
|
|
var options$2 = { |
|
singleQuote: commonOptions.singleQuote |
|
}; |
|
|
|
var createLanguage = function createLanguage(linguistData, transform) { |
|
var language = {}; |
|
|
|
for (var key in linguistData) { |
|
var newKey = key === "languageId" ? "linguistLanguageId" : key; |
|
language[newKey] = linguistData[key]; |
|
} |
|
|
|
return transform(language); |
|
}; |
|
|
|
var name$1 = "CSS"; |
|
var type = "markup"; |
|
var tmScope = "source.css"; |
|
var aceMode = "css"; |
|
var codemirrorMode = "css"; |
|
var codemirrorMimeType = "text/css"; |
|
var color = "#563d7c"; |
|
var extensions = [ |
|
".css" |
|
]; |
|
var languageId = 50; |
|
var CSS = { |
|
name: name$1, |
|
type: type, |
|
tmScope: tmScope, |
|
aceMode: aceMode, |
|
codemirrorMode: codemirrorMode, |
|
codemirrorMimeType: codemirrorMimeType, |
|
color: color, |
|
extensions: extensions, |
|
languageId: languageId |
|
}; |
|
|
|
var CSS$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$1, |
|
type: type, |
|
tmScope: tmScope, |
|
aceMode: aceMode, |
|
codemirrorMode: codemirrorMode, |
|
codemirrorMimeType: codemirrorMimeType, |
|
color: color, |
|
extensions: extensions, |
|
languageId: languageId, |
|
'default': CSS |
|
}); |
|
|
|
var name$2 = "PostCSS"; |
|
var type$1 = "markup"; |
|
var tmScope$1 = "source.postcss"; |
|
var group$2 = "CSS"; |
|
var extensions$1 = [ |
|
".pcss" |
|
]; |
|
var aceMode$1 = "text"; |
|
var languageId$1 = 262764437; |
|
var PostCSS = { |
|
name: name$2, |
|
type: type$1, |
|
tmScope: tmScope$1, |
|
group: group$2, |
|
extensions: extensions$1, |
|
aceMode: aceMode$1, |
|
languageId: languageId$1 |
|
}; |
|
|
|
var PostCSS$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$2, |
|
type: type$1, |
|
tmScope: tmScope$1, |
|
group: group$2, |
|
extensions: extensions$1, |
|
aceMode: aceMode$1, |
|
languageId: languageId$1, |
|
'default': PostCSS |
|
}); |
|
|
|
var name$3 = "Less"; |
|
var type$2 = "markup"; |
|
var group$3 = "CSS"; |
|
var extensions$2 = [ |
|
".less" |
|
]; |
|
var tmScope$2 = "source.css.less"; |
|
var aceMode$2 = "less"; |
|
var codemirrorMode$1 = "css"; |
|
var codemirrorMimeType$1 = "text/css"; |
|
var languageId$2 = 198; |
|
var Less = { |
|
name: name$3, |
|
type: type$2, |
|
group: group$3, |
|
extensions: extensions$2, |
|
tmScope: tmScope$2, |
|
aceMode: aceMode$2, |
|
codemirrorMode: codemirrorMode$1, |
|
codemirrorMimeType: codemirrorMimeType$1, |
|
languageId: languageId$2 |
|
}; |
|
|
|
var Less$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$3, |
|
type: type$2, |
|
group: group$3, |
|
extensions: extensions$2, |
|
tmScope: tmScope$2, |
|
aceMode: aceMode$2, |
|
codemirrorMode: codemirrorMode$1, |
|
codemirrorMimeType: codemirrorMimeType$1, |
|
languageId: languageId$2, |
|
'default': Less |
|
}); |
|
|
|
var name$4 = "SCSS"; |
|
var type$3 = "markup"; |
|
var tmScope$3 = "source.css.scss"; |
|
var group$4 = "CSS"; |
|
var aceMode$3 = "scss"; |
|
var codemirrorMode$2 = "css"; |
|
var codemirrorMimeType$2 = "text/x-scss"; |
|
var extensions$3 = [ |
|
".scss" |
|
]; |
|
var languageId$3 = 329; |
|
var SCSS = { |
|
name: name$4, |
|
type: type$3, |
|
tmScope: tmScope$3, |
|
group: group$4, |
|
aceMode: aceMode$3, |
|
codemirrorMode: codemirrorMode$2, |
|
codemirrorMimeType: codemirrorMimeType$2, |
|
extensions: extensions$3, |
|
languageId: languageId$3 |
|
}; |
|
|
|
var SCSS$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$4, |
|
type: type$3, |
|
tmScope: tmScope$3, |
|
group: group$4, |
|
aceMode: aceMode$3, |
|
codemirrorMode: codemirrorMode$2, |
|
codemirrorMimeType: codemirrorMimeType$2, |
|
extensions: extensions$3, |
|
languageId: languageId$3, |
|
'default': SCSS |
|
}); |
|
|
|
var require$$0$2 = getCjsExportFromNamespace(CSS$1); |
|
|
|
var require$$1 = getCjsExportFromNamespace(PostCSS$1); |
|
|
|
var require$$2 = getCjsExportFromNamespace(Less$1); |
|
|
|
var require$$3 = getCjsExportFromNamespace(SCSS$1); |
|
|
|
var languages = [createLanguage(require$$0$2, function (data) { |
|
return Object.assign(data, { |
|
since: "1.4.0", |
|
parsers: ["css"], |
|
vscodeLanguageIds: ["css"] |
|
}); |
|
}), createLanguage(require$$1, function (data) { |
|
return Object.assign(data, { |
|
since: "1.4.0", |
|
parsers: ["css"], |
|
vscodeLanguageIds: ["postcss"], |
|
extensions: data.extensions.concat(".postcss") |
|
}); |
|
}), createLanguage(require$$2, function (data) { |
|
return Object.assign(data, { |
|
since: "1.4.0", |
|
parsers: ["less"], |
|
vscodeLanguageIds: ["less"] |
|
}); |
|
}), createLanguage(require$$3, function (data) { |
|
return Object.assign(data, { |
|
since: "1.4.0", |
|
parsers: ["scss"], |
|
vscodeLanguageIds: ["scss"] |
|
}); |
|
})]; |
|
var printers = { |
|
postcss: printerPostcss |
|
}; |
|
var languageCss = { |
|
languages: languages, |
|
options: options$2, |
|
printers: printers |
|
}; |
|
|
|
function hasPragma$2(text) { |
|
return /^\s*#[^\n\S]*@(format|prettier)\s*(\n|$)/.test(text); |
|
} |
|
|
|
function insertPragma$3(text) { |
|
return "# @format\n\n" + text; |
|
} |
|
|
|
var pragma$2 = { |
|
hasPragma: hasPragma$2, |
|
insertPragma: insertPragma$3 |
|
}; |
|
|
|
var _require$$0$builders$2 = doc.builders, |
|
concat$6 = _require$$0$builders$2.concat, |
|
join$3 = _require$$0$builders$2.join, |
|
hardline$5 = _require$$0$builders$2.hardline, |
|
line$2 = _require$$0$builders$2.line, |
|
softline$2 = _require$$0$builders$2.softline, |
|
group$5 = _require$$0$builders$2.group, |
|
indent$3 = _require$$0$builders$2.indent, |
|
ifBreak$2 = _require$$0$builders$2.ifBreak; |
|
var hasIgnoreComment$2 = util.hasIgnoreComment; |
|
var isNextLineEmpty$3 = utilShared.isNextLineEmpty; |
|
var insertPragma$4 = pragma$2.insertPragma; |
|
|
|
function genericPrint$1(path, options, print) { |
|
var n = path.getValue(); |
|
|
|
if (!n) { |
|
return ""; |
|
} |
|
|
|
if (typeof n === "string") { |
|
return n; |
|
} |
|
|
|
switch (n.kind) { |
|
case "Document": |
|
{ |
|
var parts = []; |
|
path.map(function (pathChild, index) { |
|
parts.push(concat$6([pathChild.call(print)])); |
|
|
|
if (index !== n.definitions.length - 1) { |
|
parts.push(hardline$5); |
|
|
|
if (isNextLineEmpty$3(options.originalText, pathChild.getValue(), options)) { |
|
parts.push(hardline$5); |
|
} |
|
} |
|
}, "definitions"); |
|
return concat$6([concat$6(parts), hardline$5]); |
|
} |
|
|
|
case "OperationDefinition": |
|
{ |
|
var hasOperation = options.originalText[options.locStart(n)] !== "{"; |
|
var hasName = !!n.name; |
|
return concat$6([hasOperation ? n.operation : "", hasOperation && hasName ? concat$6([" ", path.call(print, "name")]) : "", n.variableDefinitions && n.variableDefinitions.length ? group$5(concat$6(["(", indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", ", "), softline$2]), path.map(print, "variableDefinitions"))])), softline$2, ")"])) : "", printDirectives(path, print, n), n.selectionSet ? !hasOperation && !hasName ? "" : " " : "", path.call(print, "selectionSet")]); |
|
} |
|
|
|
case "FragmentDefinition": |
|
{ |
|
return concat$6(["fragment ", path.call(print, "name"), n.variableDefinitions && n.variableDefinitions.length ? group$5(concat$6(["(", indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", ", "), softline$2]), path.map(print, "variableDefinitions"))])), softline$2, ")"])) : "", " on ", path.call(print, "typeCondition"), printDirectives(path, print, n), " ", path.call(print, "selectionSet")]); |
|
} |
|
|
|
case "SelectionSet": |
|
{ |
|
return concat$6(["{", indent$3(concat$6([hardline$5, join$3(hardline$5, path.call(function (selectionsPath) { |
|
return printSequence(selectionsPath, options, print); |
|
}, "selections"))])), hardline$5, "}"]); |
|
} |
|
|
|
case "Field": |
|
{ |
|
return group$5(concat$6([n.alias ? concat$6([path.call(print, "alias"), ": "]) : "", path.call(print, "name"), n.arguments.length > 0 ? group$5(concat$6(["(", indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", ", "), softline$2]), path.call(function (argsPath) { |
|
return printSequence(argsPath, options, print); |
|
}, "arguments"))])), softline$2, ")"])) : "", printDirectives(path, print, n), n.selectionSet ? " " : "", path.call(print, "selectionSet")])); |
|
} |
|
|
|
case "Name": |
|
{ |
|
return n.value; |
|
} |
|
|
|
case "StringValue": |
|
{ |
|
if (n.block) { |
|
return concat$6(['"""', hardline$5, join$3(hardline$5, n.value.replace(/"""/g, "\\$&").split("\n")), hardline$5, '"""']); |
|
} |
|
|
|
return concat$6(['"', n.value.replace(/["\\]/g, "\\$&").replace(/\n/g, "\\n"), '"']); |
|
} |
|
|
|
case "IntValue": |
|
case "FloatValue": |
|
case "EnumValue": |
|
{ |
|
return n.value; |
|
} |
|
|
|
case "BooleanValue": |
|
{ |
|
return n.value ? "true" : "false"; |
|
} |
|
|
|
case "NullValue": |
|
{ |
|
return "null"; |
|
} |
|
|
|
case "Variable": |
|
{ |
|
return concat$6(["$", path.call(print, "name")]); |
|
} |
|
|
|
case "ListValue": |
|
{ |
|
return group$5(concat$6(["[", indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", ", "), softline$2]), path.map(print, "values"))])), softline$2, "]"])); |
|
} |
|
|
|
case "ObjectValue": |
|
{ |
|
return group$5(concat$6(["{", options.bracketSpacing && n.fields.length > 0 ? " " : "", indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", ", "), softline$2]), path.map(print, "fields"))])), softline$2, ifBreak$2("", options.bracketSpacing && n.fields.length > 0 ? " " : ""), "}"])); |
|
} |
|
|
|
case "ObjectField": |
|
case "Argument": |
|
{ |
|
return concat$6([path.call(print, "name"), ": ", path.call(print, "value")]); |
|
} |
|
|
|
case "Directive": |
|
{ |
|
return concat$6(["@", path.call(print, "name"), n.arguments.length > 0 ? group$5(concat$6(["(", indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", ", "), softline$2]), path.call(function (argsPath) { |
|
return printSequence(argsPath, options, print); |
|
}, "arguments"))])), softline$2, ")"])) : ""]); |
|
} |
|
|
|
case "NamedType": |
|
{ |
|
return path.call(print, "name"); |
|
} |
|
|
|
case "VariableDefinition": |
|
{ |
|
return concat$6([path.call(print, "variable"), ": ", path.call(print, "type"), n.defaultValue ? concat$6([" = ", path.call(print, "defaultValue")]) : "", printDirectives(path, print, n)]); |
|
} |
|
|
|
case "TypeExtensionDefinition": |
|
{ |
|
return concat$6(["extend ", path.call(print, "definition")]); |
|
} |
|
|
|
case "ObjectTypeExtension": |
|
case "ObjectTypeDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? hardline$5 : "", n.kind === "ObjectTypeExtension" ? "extend " : "", "type ", path.call(print, "name"), n.interfaces.length > 0 ? concat$6([" implements ", join$3(determineInterfaceSeparator(options.originalText.substr(options.locStart(n), options.locEnd(n))), path.map(print, "interfaces"))]) : "", printDirectives(path, print, n), n.fields.length > 0 ? concat$6([" {", indent$3(concat$6([hardline$5, join$3(hardline$5, path.call(function (fieldsPath) { |
|
return printSequence(fieldsPath, options, print); |
|
}, "fields"))])), hardline$5, "}"]) : ""]); |
|
} |
|
|
|
case "FieldDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? hardline$5 : "", path.call(print, "name"), n.arguments.length > 0 ? group$5(concat$6(["(", indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", ", "), softline$2]), path.call(function (argsPath) { |
|
return printSequence(argsPath, options, print); |
|
}, "arguments"))])), softline$2, ")"])) : "", ": ", path.call(print, "type"), printDirectives(path, print, n)]); |
|
} |
|
|
|
case "DirectiveDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? hardline$5 : "", "directive ", "@", path.call(print, "name"), n.arguments.length > 0 ? group$5(concat$6(["(", indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", ", "), softline$2]), path.call(function (argsPath) { |
|
return printSequence(argsPath, options, print); |
|
}, "arguments"))])), softline$2, ")"])) : "", concat$6([" on ", join$3(" | ", path.map(print, "locations"))])]); |
|
} |
|
|
|
case "EnumTypeExtension": |
|
case "EnumTypeDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? hardline$5 : "", n.kind === "EnumTypeExtension" ? "extend " : "", "enum ", path.call(print, "name"), printDirectives(path, print, n), n.values.length > 0 ? concat$6([" {", indent$3(concat$6([hardline$5, join$3(hardline$5, path.call(function (valuesPath) { |
|
return printSequence(valuesPath, options, print); |
|
}, "values"))])), hardline$5, "}"]) : ""]); |
|
} |
|
|
|
case "EnumValueDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? hardline$5 : "", path.call(print, "name"), printDirectives(path, print, n)]); |
|
} |
|
|
|
case "InputValueDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? n.description.block ? hardline$5 : line$2 : "", path.call(print, "name"), ": ", path.call(print, "type"), n.defaultValue ? concat$6([" = ", path.call(print, "defaultValue")]) : "", printDirectives(path, print, n)]); |
|
} |
|
|
|
case "InputObjectTypeExtension": |
|
case "InputObjectTypeDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? hardline$5 : "", n.kind === "InputObjectTypeExtension" ? "extend " : "", "input ", path.call(print, "name"), printDirectives(path, print, n), n.fields.length > 0 ? concat$6([" {", indent$3(concat$6([hardline$5, join$3(hardline$5, path.call(function (fieldsPath) { |
|
return printSequence(fieldsPath, options, print); |
|
}, "fields"))])), hardline$5, "}"]) : ""]); |
|
} |
|
|
|
case "SchemaDefinition": |
|
{ |
|
return concat$6(["schema", printDirectives(path, print, n), " {", n.operationTypes.length > 0 ? indent$3(concat$6([hardline$5, join$3(hardline$5, path.call(function (opsPath) { |
|
return printSequence(opsPath, options, print); |
|
}, "operationTypes"))])) : "", hardline$5, "}"]); |
|
} |
|
|
|
case "OperationTypeDefinition": |
|
{ |
|
return concat$6([path.call(print, "operation"), ": ", path.call(print, "type")]); |
|
} |
|
|
|
case "InterfaceTypeExtension": |
|
case "InterfaceTypeDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? hardline$5 : "", n.kind === "InterfaceTypeExtension" ? "extend " : "", "interface ", path.call(print, "name"), printDirectives(path, print, n), n.fields.length > 0 ? concat$6([" {", indent$3(concat$6([hardline$5, join$3(hardline$5, path.call(function (fieldsPath) { |
|
return printSequence(fieldsPath, options, print); |
|
}, "fields"))])), hardline$5, "}"]) : ""]); |
|
} |
|
|
|
case "FragmentSpread": |
|
{ |
|
return concat$6(["...", path.call(print, "name"), printDirectives(path, print, n)]); |
|
} |
|
|
|
case "InlineFragment": |
|
{ |
|
return concat$6(["...", n.typeCondition ? concat$6([" on ", path.call(print, "typeCondition")]) : "", printDirectives(path, print, n), " ", path.call(print, "selectionSet")]); |
|
} |
|
|
|
case "UnionTypeExtension": |
|
case "UnionTypeDefinition": |
|
{ |
|
return group$5(concat$6([path.call(print, "description"), n.description ? hardline$5 : "", group$5(concat$6([n.kind === "UnionTypeExtension" ? "extend " : "", "union ", path.call(print, "name"), printDirectives(path, print, n), n.types.length > 0 ? concat$6([" =", ifBreak$2("", " "), indent$3(concat$6([ifBreak$2(concat$6([line$2, " "])), join$3(concat$6([line$2, "| "]), path.map(print, "types"))]))]) : ""]))])); |
|
} |
|
|
|
case "ScalarTypeExtension": |
|
case "ScalarTypeDefinition": |
|
{ |
|
return concat$6([path.call(print, "description"), n.description ? hardline$5 : "", n.kind === "ScalarTypeExtension" ? "extend " : "", "scalar ", path.call(print, "name"), printDirectives(path, print, n)]); |
|
} |
|
|
|
case "NonNullType": |
|
{ |
|
return concat$6([path.call(print, "type"), "!"]); |
|
} |
|
|
|
case "ListType": |
|
{ |
|
return concat$6(["[", path.call(print, "type"), "]"]); |
|
} |
|
|
|
default: |
|
/* istanbul ignore next */ |
|
throw new Error("unknown graphql type: " + JSON.stringify(n.kind)); |
|
} |
|
} |
|
|
|
function printDirectives(path, print, n) { |
|
if (n.directives.length === 0) { |
|
return ""; |
|
} |
|
|
|
return concat$6([" ", group$5(indent$3(concat$6([softline$2, join$3(concat$6([ifBreak$2("", " "), softline$2]), path.map(print, "directives"))])))]); |
|
} |
|
|
|
function printSequence(sequencePath, options, print) { |
|
var count = sequencePath.getValue().length; |
|
return sequencePath.map(function (path, i) { |
|
var printed = print(path); |
|
|
|
if (isNextLineEmpty$3(options.originalText, path.getValue(), options) && i < count - 1) { |
|
return concat$6([printed, hardline$5]); |
|
} |
|
|
|
return printed; |
|
}); |
|
} |
|
|
|
function canAttachComment(node) { |
|
return node.kind && node.kind !== "Comment"; |
|
} |
|
|
|
function printComment$1(commentPath) { |
|
var comment = commentPath.getValue(); |
|
|
|
if (comment.kind === "Comment") { |
|
return "#" + comment.value.trimRight(); |
|
} |
|
|
|
throw new Error("Not a comment: " + JSON.stringify(comment)); |
|
} |
|
|
|
function determineInterfaceSeparator(originalSource) { |
|
var start = originalSource.indexOf("implements"); |
|
|
|
if (start === -1) { |
|
throw new Error("Must implement interfaces: " + originalSource); |
|
} |
|
|
|
var end = originalSource.indexOf("{"); |
|
|
|
if (end === -1) { |
|
end = originalSource.length; |
|
} |
|
|
|
return originalSource.substr(start, end).includes("&") ? " & " : ", "; |
|
} |
|
|
|
function clean$1(node, newNode |
|
/*, parent*/ |
|
) { |
|
delete newNode.loc; |
|
delete newNode.comments; |
|
} |
|
|
|
var printerGraphql = { |
|
print: genericPrint$1, |
|
massageAstNode: clean$1, |
|
hasPrettierIgnore: hasIgnoreComment$2, |
|
insertPragma: insertPragma$4, |
|
printComment: printComment$1, |
|
canAttachComment: canAttachComment |
|
}; |
|
|
|
var options$3 = { |
|
bracketSpacing: commonOptions.bracketSpacing |
|
}; |
|
|
|
var name$5 = "GraphQL"; |
|
var type$4 = "data"; |
|
var extensions$4 = [ |
|
".graphql", |
|
".gql", |
|
".graphqls" |
|
]; |
|
var tmScope$4 = "source.graphql"; |
|
var aceMode$4 = "text"; |
|
var languageId$4 = 139; |
|
var GraphQL = { |
|
name: name$5, |
|
type: type$4, |
|
extensions: extensions$4, |
|
tmScope: tmScope$4, |
|
aceMode: aceMode$4, |
|
languageId: languageId$4 |
|
}; |
|
|
|
var GraphQL$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$5, |
|
type: type$4, |
|
extensions: extensions$4, |
|
tmScope: tmScope$4, |
|
aceMode: aceMode$4, |
|
languageId: languageId$4, |
|
'default': GraphQL |
|
}); |
|
|
|
var require$$0$3 = getCjsExportFromNamespace(GraphQL$1); |
|
|
|
var languages$1 = [createLanguage(require$$0$3, function (data) { |
|
return Object.assign(data, { |
|
since: "1.5.0", |
|
parsers: ["graphql"], |
|
vscodeLanguageIds: ["graphql"] |
|
}); |
|
})]; |
|
var printers$1 = { |
|
graphql: printerGraphql |
|
}; |
|
var languageGraphql = { |
|
languages: languages$1, |
|
options: options$3, |
|
printers: printers$1 |
|
}; |
|
|
|
var _require$$0$builders$3 = doc.builders, |
|
concat$7 = _require$$0$builders$3.concat, |
|
join$4 = _require$$0$builders$3.join, |
|
softline$3 = _require$$0$builders$3.softline, |
|
hardline$6 = _require$$0$builders$3.hardline, |
|
line$3 = _require$$0$builders$3.line, |
|
group$6 = _require$$0$builders$3.group, |
|
indent$4 = _require$$0$builders$3.indent, |
|
ifBreak$3 = _require$$0$builders$3.ifBreak; // http://w3c.github.io/html/single-page.html#void-elements |
|
|
|
var voidTags = ["area", "base", "br", "col", "embed", "hr", "img", "input", "link", "meta", "param", "source", "track", "wbr"]; // Formatter based on @glimmerjs/syntax's built-in test formatter: |
|
// https://github.com/glimmerjs/glimmer-vm/blob/master/packages/%40glimmer/syntax/lib/generation/print.ts |
|
|
|
function printChildren(path, options, print) { |
|
return concat$7(path.map(function (childPath, childIndex) { |
|
var childNode = path.getValue(); |
|
var isFirstNode = childIndex === 0; |
|
var isLastNode = childIndex == path.getParentNode(0).children.length - 1; |
|
var isLastNodeInMultiNodeList = isLastNode && !isFirstNode; |
|
var isWhitespace = isWhitespaceNode(childNode); |
|
|
|
if (isWhitespace && isLastNodeInMultiNodeList) { |
|
return print(childPath, options, print); |
|
} else if (isFirstNode) { |
|
return concat$7([softline$3, print(childPath, options, print)]); |
|
} |
|
|
|
return print(childPath, options, print); |
|
}, "children")); |
|
} |
|
|
|
function print(path, options, print) { |
|
var n = path.getValue(); |
|
/* istanbul ignore if*/ |
|
|
|
if (!n) { |
|
return ""; |
|
} |
|
|
|
switch (n.type) { |
|
case "Block": |
|
case "Program": |
|
case "Template": |
|
{ |
|
return group$6(concat$7(path.map(print, "body").filter(function (text) { |
|
return text !== ""; |
|
}))); |
|
} |
|
|
|
case "ElementNode": |
|
{ |
|
var tagFirstChar = n.tag[0]; |
|
var isLocal = n.tag.indexOf(".") !== -1; |
|
var isGlimmerComponent = tagFirstChar.toUpperCase() === tagFirstChar || isLocal; |
|
var hasChildren = n.children.length > 0; |
|
var hasNonWhitespaceChildren = n.children.some(function (n) { |
|
return !isWhitespaceNode(n); |
|
}); |
|
var isVoid = isGlimmerComponent && (!hasChildren || !hasNonWhitespaceChildren) || voidTags.indexOf(n.tag) !== -1; |
|
var closeTagForNoBreak = isVoid ? concat$7([" />", softline$3]) : ">"; |
|
var closeTagForBreak = isVoid ? "/>" : ">"; |
|
|
|
var _getParams = function _getParams(path, print) { |
|
return indent$4(concat$7([n.attributes.length ? line$3 : "", join$4(line$3, path.map(print, "attributes")), n.modifiers.length ? line$3 : "", join$4(line$3, path.map(print, "modifiers")), n.comments.length ? line$3 : "", join$4(line$3, path.map(print, "comments"))])); |
|
}; |
|
|
|
var nextNode = getNextNode(path); |
|
return concat$7([group$6(concat$7(["<", n.tag, _getParams(path, print), n.blockParams.length ? " as |".concat(n.blockParams.join(" "), "|") : "", ifBreak$3(softline$3, ""), ifBreak$3(closeTagForBreak, closeTagForNoBreak)])), !isVoid ? group$6(concat$7([hasNonWhitespaceChildren ? indent$4(printChildren(path, options, print)) : "", ifBreak$3(hasChildren ? hardline$6 : "", ""), concat$7(["</", n.tag, ">"])])) : "", nextNode && nextNode.type === "ElementNode" ? hardline$6 : ""]); |
|
} |
|
|
|
case "BlockStatement": |
|
{ |
|
var pp = path.getParentNode(1); |
|
var isElseIf = pp && pp.inverse && pp.inverse.body.length === 1 && pp.inverse.body[0] === n && pp.inverse.body[0].path.parts[0] === "if"; |
|
var hasElseIf = n.inverse && n.inverse.body.length === 1 && n.inverse.body[0].type === "BlockStatement" && n.inverse.body[0].path.parts[0] === "if"; |
|
var indentElse = hasElseIf ? function (a) { |
|
return a; |
|
} : indent$4; |
|
|
|
if (n.inverse) { |
|
return concat$7([isElseIf ? concat$7(["{{else ", printPathParams(path, print), "}}"]) : printOpenBlock(path, print), indent$4(concat$7([hardline$6, path.call(print, "program")])), n.inverse && !hasElseIf ? concat$7([hardline$6, "{{else}}"]) : "", n.inverse ? indentElse(concat$7([hardline$6, path.call(print, "inverse")])) : "", isElseIf ? "" : concat$7([hardline$6, printCloseBlock(path, print)])]); |
|
} else if (isElseIf) { |
|
return concat$7([concat$7(["{{else ", printPathParams(path, print), "}}"]), indent$4(concat$7([hardline$6, path.call(print, "program")]))]); |
|
} |
|
|
|
var _hasNonWhitespaceChildren = n.program.body.some(function (n) { |
|
return !isWhitespaceNode(n); |
|
}); |
|
|
|
return concat$7([printOpenBlock(path, print), group$6(concat$7([indent$4(concat$7([softline$3, path.call(print, "program")])), _hasNonWhitespaceChildren ? hardline$6 : softline$3, printCloseBlock(path, print)]))]); |
|
} |
|
|
|
case "ElementModifierStatement": |
|
case "MustacheStatement": |
|
{ |
|
var _pp = path.getParentNode(1); |
|
|
|
var isConcat = _pp && _pp.type === "ConcatStatement"; |
|
return group$6(concat$7([n.escaped === false ? "{{{" : "{{", printPathParams(path, print, { |
|
group: false |
|
}), isConcat ? "" : softline$3, n.escaped === false ? "}}}" : "}}"])); |
|
} |
|
|
|
case "SubExpression": |
|
{ |
|
var params = getParams(path, print); |
|
var printedParams = params.length > 0 ? indent$4(concat$7([line$3, group$6(join$4(line$3, params))])) : ""; |
|
return group$6(concat$7(["(", printPath(path, print), printedParams, softline$3, ")"])); |
|
} |
|
|
|
case "AttrNode": |
|
{ |
|
var isText = n.value.type === "TextNode"; |
|
|
|
if (isText && n.value.loc.start.column === n.value.loc.end.column) { |
|
return concat$7([n.name]); |
|
} |
|
|
|
var value = path.call(print, "value"); |
|
var quotedValue = isText ? printStringLiteral(value.parts.join(), options) : value; |
|
return concat$7([n.name, "=", quotedValue]); |
|
} |
|
|
|
case "ConcatStatement": |
|
{ |
|
return concat$7(['"', group$6(indent$4(join$4(softline$3, path.map(function (partPath) { |
|
return print(partPath); |
|
}, "parts").filter(function (a) { |
|
return a !== ""; |
|
})))), '"']); |
|
} |
|
|
|
case "Hash": |
|
{ |
|
return concat$7([join$4(line$3, path.map(print, "pairs"))]); |
|
} |
|
|
|
case "HashPair": |
|
{ |
|
return concat$7([n.key, "=", path.call(print, "value")]); |
|
} |
|
|
|
case "TextNode": |
|
{ |
|
var maxLineBreaksToPreserve = 2; |
|
var isFirstElement = !getPreviousNode(path); |
|
var isLastElement = !getNextNode(path); |
|
var isWhitespaceOnly = !/\S/.test(n.chars); |
|
var lineBreaksCount = countNewLines(n.chars); |
|
var hasBlockParent = path.getParentNode(0).type === "Block"; |
|
var hasElementParent = path.getParentNode(0).type === "ElementNode"; |
|
var hasTemplateParent = path.getParentNode(0).type === "Template"; |
|
var leadingLineBreaksCount = countLeadingNewLines(n.chars); |
|
var trailingLineBreaksCount = countTrailingNewLines(n.chars); |
|
|
|
if ((isFirstElement || isLastElement) && isWhitespaceOnly && (hasBlockParent || hasElementParent || hasTemplateParent)) { |
|
return ""; |
|
} |
|
|
|
if (isWhitespaceOnly && lineBreaksCount) { |
|
leadingLineBreaksCount = Math.min(lineBreaksCount, maxLineBreaksToPreserve); |
|
trailingLineBreaksCount = 0; |
|
} else { |
|
if (isNextNodeOfType(path, "ElementNode") || isNextNodeOfType(path, "BlockStatement")) { |
|
trailingLineBreaksCount = Math.max(trailingLineBreaksCount, 1); |
|
} |
|
|
|
if (isPreviousNodeOfSomeType(path, ["ElementNode"]) || isPreviousNodeOfSomeType(path, ["BlockStatement"])) { |
|
leadingLineBreaksCount = Math.max(leadingLineBreaksCount, 1); |
|
} |
|
} |
|
|
|
var leadingSpace = ""; |
|
var trailingSpace = ""; // preserve a space inside of an attribute node where whitespace present, |
|
// when next to mustache statement. |
|
|
|
var inAttrNode = path.stack.indexOf("attributes") >= 0; |
|
|
|
if (inAttrNode) { |
|
var parentNode = path.getParentNode(0); |
|
|
|
var _isConcat = parentNode.type === "ConcatStatement"; |
|
|
|
if (_isConcat) { |
|
var parts = parentNode.parts; |
|
var partIndex = parts.indexOf(n); |
|
|
|
if (partIndex > 0) { |
|
var partType = parts[partIndex - 1].type; |
|
var isMustache = partType === "MustacheStatement"; |
|
|
|
if (isMustache) { |
|
leadingSpace = " "; |
|
} |
|
} |
|
|
|
if (partIndex < parts.length - 1) { |
|
var _partType = parts[partIndex + 1].type; |
|
|
|
var _isMustache = _partType === "MustacheStatement"; |
|
|
|
if (_isMustache) { |
|
trailingSpace = " "; |
|
} |
|
} |
|
} |
|
} else { |
|
if (trailingLineBreaksCount === 0 && isNextNodeOfType(path, "MustacheStatement")) { |
|
trailingSpace = " "; |
|
} |
|
|
|
if (leadingLineBreaksCount === 0 && isPreviousNodeOfSomeType(path, ["MustacheStatement"])) { |
|
leadingSpace = " "; |
|
} |
|
|
|
if (isFirstElement) { |
|
leadingLineBreaksCount = 0; |
|
leadingSpace = ""; |
|
} |
|
|
|
if (isLastElement) { |
|
trailingLineBreaksCount = 0; |
|
trailingSpace = ""; |
|
} |
|
} |
|
|
|
return concat$7([].concat(_toConsumableArray(generateHardlines(leadingLineBreaksCount, maxLineBreaksToPreserve)), [n.chars.replace(/^[\s ]+/g, leadingSpace).replace(/[\s ]+$/, trailingSpace)], _toConsumableArray(generateHardlines(trailingLineBreaksCount, maxLineBreaksToPreserve))).filter(Boolean)); |
|
} |
|
|
|
case "MustacheCommentStatement": |
|
{ |
|
var dashes = n.value.indexOf("}}") > -1 ? "--" : ""; |
|
return concat$7(["{{!", dashes, n.value, dashes, "}}"]); |
|
} |
|
|
|
case "PathExpression": |
|
{ |
|
return n.original; |
|
} |
|
|
|
case "BooleanLiteral": |
|
{ |
|
return String(n.value); |
|
} |
|
|
|
case "CommentStatement": |
|
{ |
|
return concat$7(["<!--", n.value, "-->"]); |
|
} |
|
|
|
case "StringLiteral": |
|
{ |
|
return printStringLiteral(n.value, options); |
|
} |
|
|
|
case "NumberLiteral": |
|
{ |
|
return String(n.value); |
|
} |
|
|
|
case "UndefinedLiteral": |
|
{ |
|
return "undefined"; |
|
} |
|
|
|
case "NullLiteral": |
|
{ |
|
return "null"; |
|
} |
|
|
|
/* istanbul ignore next */ |
|
|
|
default: |
|
throw new Error("unknown glimmer type: " + JSON.stringify(n.type)); |
|
} |
|
} |
|
/** |
|
* Prints a string literal with the correct surrounding quotes based on |
|
* `options.singleQuote` and the number of escaped quotes contained in |
|
* the string literal. This function is the glimmer equivalent of `printString` |
|
* in `common/util`, but has differences because of the way escaped characters |
|
* are treated in hbs string literals. |
|
* @param {string} stringLiteral - the string literal value |
|
* @param {object} options - the prettier options object |
|
*/ |
|
|
|
|
|
function printStringLiteral(stringLiteral, options) { |
|
var double = { |
|
quote: '"', |
|
regex: /"/g |
|
}; |
|
var single = { |
|
quote: "'", |
|
regex: /'/g |
|
}; |
|
var preferred = options.singleQuote ? single : double; |
|
var alternate = preferred === single ? double : single; |
|
var shouldUseAlternateQuote = false; // If `stringLiteral` contains at least one of the quote preferred for |
|
// enclosing the string, we might want to enclose with the alternate quote |
|
// instead, to minimize the number of escaped quotes. |
|
|
|
if (stringLiteral.includes(preferred.quote) || stringLiteral.includes(alternate.quote)) { |
|
var numPreferredQuotes = (stringLiteral.match(preferred.regex) || []).length; |
|
var numAlternateQuotes = (stringLiteral.match(alternate.regex) || []).length; |
|
shouldUseAlternateQuote = numPreferredQuotes > numAlternateQuotes; |
|
} |
|
|
|
var enclosingQuote = shouldUseAlternateQuote ? alternate : preferred; |
|
var escapedStringLiteral = stringLiteral.replace(enclosingQuote.regex, "\\".concat(enclosingQuote.quote)); |
|
return "".concat(enclosingQuote.quote).concat(escapedStringLiteral).concat(enclosingQuote.quote); |
|
} |
|
|
|
function printPath(path, print) { |
|
return path.call(print, "path"); |
|
} |
|
|
|
function getParams(path, print) { |
|
var node = path.getValue(); |
|
var parts = []; |
|
|
|
if (node.params.length > 0) { |
|
parts = parts.concat(path.map(print, "params")); |
|
} |
|
|
|
if (node.hash && node.hash.pairs.length > 0) { |
|
parts.push(path.call(print, "hash")); |
|
} |
|
|
|
return parts; |
|
} |
|
|
|
function printPathParams(path, print, options) { |
|
var parts = []; |
|
options = Object.assign({ |
|
group: true |
|
}, options || {}); |
|
parts.push(printPath(path, print)); |
|
parts = parts.concat(getParams(path, print)); |
|
|
|
if (!options.group) { |
|
return indent$4(join$4(line$3, parts)); |
|
} |
|
|
|
return indent$4(group$6(join$4(line$3, parts))); |
|
} |
|
|
|
function printBlockParams(path) { |
|
var block = path.getValue(); |
|
|
|
if (!block.program || !block.program.blockParams.length) { |
|
return ""; |
|
} |
|
|
|
return concat$7([" as |", block.program.blockParams.join(" "), "|"]); |
|
} |
|
|
|
function printOpenBlock(path, print) { |
|
return group$6(concat$7(["{{#", printPathParams(path, print), printBlockParams(path), softline$3, "}}"])); |
|
} |
|
|
|
function printCloseBlock(path, print) { |
|
return concat$7(["{{/", path.call(print, "path"), "}}"]); |
|
} |
|
|
|
function isWhitespaceNode(node) { |
|
return node.type === "TextNode" && !/\S/.test(node.chars); |
|
} |
|
|
|
function getPreviousNode(path) { |
|
var node = path.getValue(); |
|
var parentNode = path.getParentNode(0); |
|
var children = parentNode.children || parentNode.body; |
|
|
|
if (children) { |
|
var nodeIndex = children.indexOf(node); |
|
|
|
if (nodeIndex > 0) { |
|
var previousNode = children[nodeIndex - 1]; |
|
return previousNode; |
|
} |
|
} |
|
} |
|
|
|
function getNextNode(path) { |
|
var node = path.getValue(); |
|
var parentNode = path.getParentNode(0); |
|
var children = parentNode.children || parentNode.body; |
|
|
|
if (children) { |
|
var nodeIndex = children.indexOf(node); |
|
|
|
if (nodeIndex < children.length) { |
|
var nextNode = children[nodeIndex + 1]; |
|
return nextNode; |
|
} |
|
} |
|
} |
|
|
|
function isPreviousNodeOfSomeType(path, types) { |
|
var previousNode = getPreviousNode(path); |
|
|
|
if (previousNode) { |
|
return types.some(function (type) { |
|
return previousNode.type === type; |
|
}); |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isNextNodeOfType(path, type) { |
|
var nextNode = getNextNode(path); |
|
return nextNode && nextNode.type === type; |
|
} |
|
|
|
function clean$2(ast, newObj) { |
|
delete newObj.loc; |
|
delete newObj.selfClosing; // (Glimmer/HTML) ignore TextNode whitespace |
|
|
|
if (ast.type === "TextNode") { |
|
if (ast.chars.replace(/\s+/, "") === "") { |
|
return null; |
|
} |
|
|
|
newObj.chars = ast.chars.replace(/^\s+/, "").replace(/\s+$/, ""); |
|
} |
|
} |
|
|
|
function countNewLines(string) { |
|
/* istanbul ignore next */ |
|
string = typeof string === "string" ? string : ""; |
|
return string.split("\n").length - 1; |
|
} |
|
|
|
function countLeadingNewLines(string) { |
|
/* istanbul ignore next */ |
|
string = typeof string === "string" ? string : ""; |
|
var newLines = (string.match(/^([^\S\r\n]*[\r\n])+/g) || [])[0] || ""; |
|
return countNewLines(newLines); |
|
} |
|
|
|
function countTrailingNewLines(string) { |
|
/* istanbul ignore next */ |
|
string = typeof string === "string" ? string : ""; |
|
var newLines = (string.match(/([\r\n][^\S\r\n]*)+$/g) || [])[0] || ""; |
|
return countNewLines(newLines); |
|
} |
|
|
|
function generateHardlines() { |
|
var number = arguments.length > 0 && arguments[0] !== undefined ? arguments[0] : 0; |
|
var max = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : 0; |
|
return new Array(Math.min(number, max)).fill(hardline$6); |
|
} |
|
|
|
var printerGlimmer = { |
|
print: print, |
|
massageAstNode: clean$2 |
|
}; |
|
|
|
var name$6 = "Handlebars"; |
|
var type$5 = "markup"; |
|
var group$7 = "HTML"; |
|
var aliases = [ |
|
"hbs", |
|
"htmlbars" |
|
]; |
|
var extensions$5 = [ |
|
".handlebars", |
|
".hbs" |
|
]; |
|
var tmScope$5 = "text.html.handlebars"; |
|
var aceMode$5 = "handlebars"; |
|
var languageId$5 = 155; |
|
var Handlebars = { |
|
name: name$6, |
|
type: type$5, |
|
group: group$7, |
|
aliases: aliases, |
|
extensions: extensions$5, |
|
tmScope: tmScope$5, |
|
aceMode: aceMode$5, |
|
languageId: languageId$5 |
|
}; |
|
|
|
var Handlebars$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$6, |
|
type: type$5, |
|
group: group$7, |
|
aliases: aliases, |
|
extensions: extensions$5, |
|
tmScope: tmScope$5, |
|
aceMode: aceMode$5, |
|
languageId: languageId$5, |
|
'default': Handlebars |
|
}); |
|
|
|
var require$$0$4 = getCjsExportFromNamespace(Handlebars$1); |
|
|
|
var languages$2 = [createLanguage(require$$0$4, function (data) { |
|
return Object.assign(data, { |
|
since: null, |
|
// unreleased |
|
parsers: ["glimmer"], |
|
vscodeLanguageIds: ["handlebars"] |
|
}); |
|
})]; |
|
var printers$2 = { |
|
glimmer: printerGlimmer |
|
}; |
|
var languageHandlebars = { |
|
languages: languages$2, |
|
printers: printers$2 |
|
}; |
|
|
|
var clean$3 = function clean(ast, newNode) { |
|
delete newNode.sourceSpan; |
|
delete newNode.startSourceSpan; |
|
delete newNode.endSourceSpan; |
|
delete newNode.nameSpan; |
|
delete newNode.valueSpan; |
|
|
|
if (ast.type === "text" || ast.type === "comment") { |
|
return null; |
|
} // may be formatted by multiparser |
|
|
|
|
|
if (ast.type === "yaml" || ast.type === "toml") { |
|
return null; |
|
} |
|
|
|
if (ast.type === "attribute") { |
|
delete newNode.value; |
|
} |
|
|
|
if (ast.type === "docType") { |
|
delete newNode.value; |
|
} |
|
}; |
|
|
|
var json = { |
|
"CSS_DISPLAY_TAGS": { |
|
"area": "none", |
|
"base": "none", |
|
"basefont": "none", |
|
"datalist": "none", |
|
"head": "none", |
|
"link": "none", |
|
"meta": "none", |
|
"noembed": "none", |
|
"noframes": "none", |
|
"param": "none", |
|
"rp": "none", |
|
"script": "block", |
|
"source": "block", |
|
"style": "none", |
|
"template": "inline", |
|
"track": "block", |
|
"title": "none", |
|
"html": "block", |
|
"body": "block", |
|
"address": "block", |
|
"blockquote": "block", |
|
"center": "block", |
|
"div": "block", |
|
"figure": "block", |
|
"figcaption": "block", |
|
"footer": "block", |
|
"form": "block", |
|
"header": "block", |
|
"hr": "block", |
|
"legend": "block", |
|
"listing": "block", |
|
"main": "block", |
|
"p": "block", |
|
"plaintext": "block", |
|
"pre": "block", |
|
"xmp": "block", |
|
"slot": "contents", |
|
"ruby": "ruby", |
|
"rt": "ruby-text", |
|
"article": "block", |
|
"aside": "block", |
|
"h1": "block", |
|
"h2": "block", |
|
"h3": "block", |
|
"h4": "block", |
|
"h5": "block", |
|
"h6": "block", |
|
"hgroup": "block", |
|
"nav": "block", |
|
"section": "block", |
|
"dir": "block", |
|
"dd": "block", |
|
"dl": "block", |
|
"dt": "block", |
|
"ol": "block", |
|
"ul": "block", |
|
"li": "list-item", |
|
"table": "table", |
|
"caption": "table-caption", |
|
"colgroup": "table-column-group", |
|
"col": "table-column", |
|
"thead": "table-header-group", |
|
"tbody": "table-row-group", |
|
"tfoot": "table-footer-group", |
|
"tr": "table-row", |
|
"td": "table-cell", |
|
"th": "table-cell", |
|
"fieldset": "block", |
|
"button": "inline-block", |
|
"video": "inline-block", |
|
"audio": "inline-block" |
|
}, |
|
"CSS_DISPLAY_DEFAULT": "inline", |
|
"CSS_WHITE_SPACE_TAGS": { |
|
"listing": "pre", |
|
"plaintext": "pre", |
|
"pre": "pre", |
|
"xmp": "pre", |
|
"nobr": "nowrap", |
|
"table": "initial", |
|
"textarea": "pre-wrap" |
|
}, |
|
"CSS_WHITE_SPACE_DEFAULT": "normal" |
|
}; |
|
|
|
var a = [ |
|
"accesskey", |
|
"charset", |
|
"coords", |
|
"download", |
|
"href", |
|
"hreflang", |
|
"name", |
|
"ping", |
|
"referrerpolicy", |
|
"rel", |
|
"rev", |
|
"shape", |
|
"tabindex", |
|
"target", |
|
"type" |
|
]; |
|
var abbr = [ |
|
"title" |
|
]; |
|
var applet = [ |
|
"align", |
|
"alt", |
|
"archive", |
|
"code", |
|
"codebase", |
|
"height", |
|
"hspace", |
|
"name", |
|
"object", |
|
"vspace", |
|
"width" |
|
]; |
|
var area = [ |
|
"accesskey", |
|
"alt", |
|
"coords", |
|
"download", |
|
"href", |
|
"hreflang", |
|
"nohref", |
|
"ping", |
|
"referrerpolicy", |
|
"rel", |
|
"shape", |
|
"tabindex", |
|
"target", |
|
"type" |
|
]; |
|
var audio = [ |
|
"autoplay", |
|
"controls", |
|
"crossorigin", |
|
"loop", |
|
"muted", |
|
"preload", |
|
"src" |
|
]; |
|
var base = [ |
|
"href", |
|
"target" |
|
]; |
|
var basefont = [ |
|
"color", |
|
"face", |
|
"size" |
|
]; |
|
var bdo = [ |
|
"dir" |
|
]; |
|
var blockquote = [ |
|
"cite" |
|
]; |
|
var body = [ |
|
"alink", |
|
"background", |
|
"bgcolor", |
|
"link", |
|
"text", |
|
"vlink" |
|
]; |
|
var br = [ |
|
"clear" |
|
]; |
|
var button = [ |
|
"accesskey", |
|
"autofocus", |
|
"disabled", |
|
"form", |
|
"formaction", |
|
"formenctype", |
|
"formmethod", |
|
"formnovalidate", |
|
"formtarget", |
|
"name", |
|
"tabindex", |
|
"type", |
|
"value" |
|
]; |
|
var canvas = [ |
|
"height", |
|
"width" |
|
]; |
|
var caption = [ |
|
"align" |
|
]; |
|
var col = [ |
|
"align", |
|
"char", |
|
"charoff", |
|
"span", |
|
"valign", |
|
"width" |
|
]; |
|
var colgroup = [ |
|
"align", |
|
"char", |
|
"charoff", |
|
"span", |
|
"valign", |
|
"width" |
|
]; |
|
var data = [ |
|
"value" |
|
]; |
|
var del = [ |
|
"cite", |
|
"datetime" |
|
]; |
|
var details = [ |
|
"open" |
|
]; |
|
var dfn = [ |
|
"title" |
|
]; |
|
var dialog = [ |
|
"open" |
|
]; |
|
var dir = [ |
|
"compact" |
|
]; |
|
var div = [ |
|
"align" |
|
]; |
|
var dl = [ |
|
"compact" |
|
]; |
|
var embed$1 = [ |
|
"height", |
|
"src", |
|
"type", |
|
"width" |
|
]; |
|
var fieldset = [ |
|
"disabled", |
|
"form", |
|
"name" |
|
]; |
|
var font = [ |
|
"color", |
|
"face", |
|
"size" |
|
]; |
|
var form = [ |
|
"accept", |
|
"accept-charset", |
|
"action", |
|
"autocomplete", |
|
"enctype", |
|
"method", |
|
"name", |
|
"novalidate", |
|
"target" |
|
]; |
|
var frame = [ |
|
"frameborder", |
|
"longdesc", |
|
"marginheight", |
|
"marginwidth", |
|
"name", |
|
"noresize", |
|
"scrolling", |
|
"src" |
|
]; |
|
var frameset = [ |
|
"cols", |
|
"rows" |
|
]; |
|
var h1 = [ |
|
"align" |
|
]; |
|
var h2 = [ |
|
"align" |
|
]; |
|
var h3 = [ |
|
"align" |
|
]; |
|
var h4 = [ |
|
"align" |
|
]; |
|
var h5 = [ |
|
"align" |
|
]; |
|
var h6 = [ |
|
"align" |
|
]; |
|
var head = [ |
|
"profile" |
|
]; |
|
var hr = [ |
|
"align", |
|
"noshade", |
|
"size", |
|
"width" |
|
]; |
|
var html = [ |
|
"manifest", |
|
"version" |
|
]; |
|
var iframe = [ |
|
"align", |
|
"allow", |
|
"allowfullscreen", |
|
"allowpaymentrequest", |
|
"allowusermedia", |
|
"frameborder", |
|
"height", |
|
"longdesc", |
|
"marginheight", |
|
"marginwidth", |
|
"name", |
|
"referrerpolicy", |
|
"sandbox", |
|
"scrolling", |
|
"src", |
|
"srcdoc", |
|
"width" |
|
]; |
|
var img = [ |
|
"align", |
|
"alt", |
|
"border", |
|
"crossorigin", |
|
"decoding", |
|
"height", |
|
"hspace", |
|
"ismap", |
|
"longdesc", |
|
"name", |
|
"referrerpolicy", |
|
"sizes", |
|
"src", |
|
"srcset", |
|
"usemap", |
|
"vspace", |
|
"width" |
|
]; |
|
var input = [ |
|
"accept", |
|
"accesskey", |
|
"align", |
|
"alt", |
|
"autocomplete", |
|
"autofocus", |
|
"checked", |
|
"dirname", |
|
"disabled", |
|
"form", |
|
"formaction", |
|
"formenctype", |
|
"formmethod", |
|
"formnovalidate", |
|
"formtarget", |
|
"height", |
|
"ismap", |
|
"list", |
|
"max", |
|
"maxlength", |
|
"min", |
|
"minlength", |
|
"multiple", |
|
"name", |
|
"pattern", |
|
"placeholder", |
|
"readonly", |
|
"required", |
|
"size", |
|
"src", |
|
"step", |
|
"tabindex", |
|
"title", |
|
"type", |
|
"usemap", |
|
"value", |
|
"width" |
|
]; |
|
var ins = [ |
|
"cite", |
|
"datetime" |
|
]; |
|
var isindex = [ |
|
"prompt" |
|
]; |
|
var label = [ |
|
"accesskey", |
|
"for", |
|
"form" |
|
]; |
|
var legend = [ |
|
"accesskey", |
|
"align" |
|
]; |
|
var li = [ |
|
"type", |
|
"value" |
|
]; |
|
var link$1 = [ |
|
"as", |
|
"charset", |
|
"color", |
|
"crossorigin", |
|
"href", |
|
"hreflang", |
|
"imagesizes", |
|
"imagesrcset", |
|
"integrity", |
|
"media", |
|
"nonce", |
|
"referrerpolicy", |
|
"rel", |
|
"rev", |
|
"sizes", |
|
"target", |
|
"title", |
|
"type" |
|
]; |
|
var map = [ |
|
"name" |
|
]; |
|
var menu = [ |
|
"compact" |
|
]; |
|
var meta = [ |
|
"charset", |
|
"content", |
|
"http-equiv", |
|
"name", |
|
"scheme" |
|
]; |
|
var meter = [ |
|
"high", |
|
"low", |
|
"max", |
|
"min", |
|
"optimum", |
|
"value" |
|
]; |
|
var object = [ |
|
"align", |
|
"archive", |
|
"border", |
|
"classid", |
|
"codebase", |
|
"codetype", |
|
"data", |
|
"declare", |
|
"form", |
|
"height", |
|
"hspace", |
|
"name", |
|
"standby", |
|
"tabindex", |
|
"type", |
|
"typemustmatch", |
|
"usemap", |
|
"vspace", |
|
"width" |
|
]; |
|
var ol = [ |
|
"compact", |
|
"reversed", |
|
"start", |
|
"type" |
|
]; |
|
var optgroup = [ |
|
"disabled", |
|
"label" |
|
]; |
|
var option = [ |
|
"disabled", |
|
"label", |
|
"selected", |
|
"value" |
|
]; |
|
var output = [ |
|
"for", |
|
"form", |
|
"name" |
|
]; |
|
var p = [ |
|
"align" |
|
]; |
|
var param = [ |
|
"name", |
|
"type", |
|
"value", |
|
"valuetype" |
|
]; |
|
var pre = [ |
|
"width" |
|
]; |
|
var progress = [ |
|
"max", |
|
"value" |
|
]; |
|
var q = [ |
|
"cite" |
|
]; |
|
var script = [ |
|
"async", |
|
"charset", |
|
"crossorigin", |
|
"defer", |
|
"integrity", |
|
"language", |
|
"nomodule", |
|
"nonce", |
|
"referrerpolicy", |
|
"src", |
|
"type" |
|
]; |
|
var select = [ |
|
"autocomplete", |
|
"autofocus", |
|
"disabled", |
|
"form", |
|
"multiple", |
|
"name", |
|
"required", |
|
"size", |
|
"tabindex" |
|
]; |
|
var slot = [ |
|
"name" |
|
]; |
|
var source = [ |
|
"media", |
|
"sizes", |
|
"src", |
|
"srcset", |
|
"type" |
|
]; |
|
var style = [ |
|
"media", |
|
"nonce", |
|
"title", |
|
"type" |
|
]; |
|
var table = [ |
|
"align", |
|
"bgcolor", |
|
"border", |
|
"cellpadding", |
|
"cellspacing", |
|
"frame", |
|
"rules", |
|
"summary", |
|
"width" |
|
]; |
|
var tbody = [ |
|
"align", |
|
"char", |
|
"charoff", |
|
"valign" |
|
]; |
|
var td = [ |
|
"abbr", |
|
"align", |
|
"axis", |
|
"bgcolor", |
|
"char", |
|
"charoff", |
|
"colspan", |
|
"headers", |
|
"height", |
|
"nowrap", |
|
"rowspan", |
|
"scope", |
|
"valign", |
|
"width" |
|
]; |
|
var textarea = [ |
|
"accesskey", |
|
"autocomplete", |
|
"autofocus", |
|
"cols", |
|
"dirname", |
|
"disabled", |
|
"form", |
|
"maxlength", |
|
"minlength", |
|
"name", |
|
"placeholder", |
|
"readonly", |
|
"required", |
|
"rows", |
|
"tabindex", |
|
"wrap" |
|
]; |
|
var tfoot = [ |
|
"align", |
|
"char", |
|
"charoff", |
|
"valign" |
|
]; |
|
var th = [ |
|
"abbr", |
|
"align", |
|
"axis", |
|
"bgcolor", |
|
"char", |
|
"charoff", |
|
"colspan", |
|
"headers", |
|
"height", |
|
"nowrap", |
|
"rowspan", |
|
"scope", |
|
"valign", |
|
"width" |
|
]; |
|
var thead = [ |
|
"align", |
|
"char", |
|
"charoff", |
|
"valign" |
|
]; |
|
var time = [ |
|
"datetime" |
|
]; |
|
var tr = [ |
|
"align", |
|
"bgcolor", |
|
"char", |
|
"charoff", |
|
"valign" |
|
]; |
|
var track = [ |
|
"default", |
|
"kind", |
|
"label", |
|
"src", |
|
"srclang" |
|
]; |
|
var ul = [ |
|
"compact", |
|
"type" |
|
]; |
|
var video = [ |
|
"autoplay", |
|
"controls", |
|
"crossorigin", |
|
"height", |
|
"loop", |
|
"muted", |
|
"playsinline", |
|
"poster", |
|
"preload", |
|
"src", |
|
"width" |
|
]; |
|
var index$1 = { |
|
"*": [ |
|
"accesskey", |
|
"autocapitalize", |
|
"autofocus", |
|
"class", |
|
"contenteditable", |
|
"dir", |
|
"draggable", |
|
"enterkeyhint", |
|
"hidden", |
|
"id", |
|
"inputmode", |
|
"is", |
|
"itemid", |
|
"itemprop", |
|
"itemref", |
|
"itemscope", |
|
"itemtype", |
|
"lang", |
|
"nonce", |
|
"slot", |
|
"spellcheck", |
|
"style", |
|
"tabindex", |
|
"title", |
|
"translate" |
|
], |
|
a: a, |
|
abbr: abbr, |
|
applet: applet, |
|
area: area, |
|
audio: audio, |
|
base: base, |
|
basefont: basefont, |
|
bdo: bdo, |
|
blockquote: blockquote, |
|
body: body, |
|
br: br, |
|
button: button, |
|
canvas: canvas, |
|
caption: caption, |
|
col: col, |
|
colgroup: colgroup, |
|
data: data, |
|
del: del, |
|
details: details, |
|
dfn: dfn, |
|
dialog: dialog, |
|
dir: dir, |
|
div: div, |
|
dl: dl, |
|
embed: embed$1, |
|
fieldset: fieldset, |
|
font: font, |
|
form: form, |
|
frame: frame, |
|
frameset: frameset, |
|
h1: h1, |
|
h2: h2, |
|
h3: h3, |
|
h4: h4, |
|
h5: h5, |
|
h6: h6, |
|
head: head, |
|
hr: hr, |
|
html: html, |
|
iframe: iframe, |
|
img: img, |
|
input: input, |
|
ins: ins, |
|
isindex: isindex, |
|
label: label, |
|
legend: legend, |
|
li: li, |
|
link: link$1, |
|
map: map, |
|
menu: menu, |
|
meta: meta, |
|
meter: meter, |
|
object: object, |
|
ol: ol, |
|
optgroup: optgroup, |
|
option: option, |
|
output: output, |
|
p: p, |
|
param: param, |
|
pre: pre, |
|
progress: progress, |
|
q: q, |
|
script: script, |
|
select: select, |
|
slot: slot, |
|
source: source, |
|
style: style, |
|
table: table, |
|
tbody: tbody, |
|
td: td, |
|
textarea: textarea, |
|
tfoot: tfoot, |
|
th: th, |
|
thead: thead, |
|
time: time, |
|
tr: tr, |
|
track: track, |
|
ul: ul, |
|
video: video |
|
}; |
|
|
|
var htmlElementAttributes = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
a: a, |
|
abbr: abbr, |
|
applet: applet, |
|
area: area, |
|
audio: audio, |
|
base: base, |
|
basefont: basefont, |
|
bdo: bdo, |
|
blockquote: blockquote, |
|
body: body, |
|
br: br, |
|
button: button, |
|
canvas: canvas, |
|
caption: caption, |
|
col: col, |
|
colgroup: colgroup, |
|
data: data, |
|
del: del, |
|
details: details, |
|
dfn: dfn, |
|
dialog: dialog, |
|
dir: dir, |
|
div: div, |
|
dl: dl, |
|
embed: embed$1, |
|
fieldset: fieldset, |
|
font: font, |
|
form: form, |
|
frame: frame, |
|
frameset: frameset, |
|
h1: h1, |
|
h2: h2, |
|
h3: h3, |
|
h4: h4, |
|
h5: h5, |
|
h6: h6, |
|
head: head, |
|
hr: hr, |
|
html: html, |
|
iframe: iframe, |
|
img: img, |
|
input: input, |
|
ins: ins, |
|
isindex: isindex, |
|
label: label, |
|
legend: legend, |
|
li: li, |
|
link: link$1, |
|
map: map, |
|
menu: menu, |
|
meta: meta, |
|
meter: meter, |
|
object: object, |
|
ol: ol, |
|
optgroup: optgroup, |
|
option: option, |
|
output: output, |
|
p: p, |
|
param: param, |
|
pre: pre, |
|
progress: progress, |
|
q: q, |
|
script: script, |
|
select: select, |
|
slot: slot, |
|
source: source, |
|
style: style, |
|
table: table, |
|
tbody: tbody, |
|
td: td, |
|
textarea: textarea, |
|
tfoot: tfoot, |
|
th: th, |
|
thead: thead, |
|
time: time, |
|
tr: tr, |
|
track: track, |
|
ul: ul, |
|
video: video, |
|
'default': index$1 |
|
}); |
|
|
|
var htmlElementAttributes$1 = getCjsExportFromNamespace(htmlElementAttributes); |
|
|
|
var CSS_DISPLAY_TAGS = json.CSS_DISPLAY_TAGS, |
|
CSS_DISPLAY_DEFAULT = json.CSS_DISPLAY_DEFAULT, |
|
CSS_WHITE_SPACE_TAGS = json.CSS_WHITE_SPACE_TAGS, |
|
CSS_WHITE_SPACE_DEFAULT = json.CSS_WHITE_SPACE_DEFAULT; |
|
var HTML_TAGS = arrayToMap(htmlTagNames$1); |
|
var HTML_ELEMENT_ATTRIBUTES = mapObject(htmlElementAttributes$1, arrayToMap); |
|
|
|
function arrayToMap(array) { |
|
var map = Object.create(null); |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = array[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var value = _step.value; |
|
map[value] = true; |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return map; |
|
} |
|
|
|
function mapObject(object, fn) { |
|
var newObject = Object.create(null); |
|
|
|
for (var _i = 0, _Object$keys = Object.keys(object); _i < _Object$keys.length; _i++) { |
|
var key = _Object$keys[_i]; |
|
newObject[key] = fn(object[key], key); |
|
} |
|
|
|
return newObject; |
|
} |
|
|
|
function shouldPreserveContent(node, options) { |
|
if (node.type === "element" && node.fullName === "template" && node.attrMap.lang && node.attrMap.lang !== "html") { |
|
return true; |
|
} // unterminated node in ie conditional comment |
|
// e.g. <!--[if lt IE 9]><html><![endif]--> |
|
|
|
|
|
if (node.type === "ieConditionalComment" && node.lastChild && !node.lastChild.isSelfClosing && !node.lastChild.endSourceSpan) { |
|
return true; |
|
} // incomplete html in ie conditional comment |
|
// e.g. <!--[if lt IE 9]></div><![endif]--> |
|
|
|
|
|
if (node.type === "ieConditionalComment" && !node.complete) { |
|
return true; |
|
} // top-level elements (excluding <template>, <style> and <script>) in Vue SFC are considered custom block |
|
// custom blocks can be written in other languages so we should preserve them to not break the code |
|
|
|
|
|
if (options.parser === "vue" && node.type === "element" && node.parent.type === "root" && ["template", "style", "script", // vue parser can be used for vue dom template as well, so we should still format top-level <html> |
|
"html"].indexOf(node.fullName) === -1) { |
|
return true; |
|
} // TODO: handle non-text children in <pre> |
|
|
|
|
|
if (isPreLikeNode(node) && node.children.some(function (child) { |
|
return child.type !== "text" && child.type !== "interpolation"; |
|
})) { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function hasPrettierIgnore(node) { |
|
if (node.type === "attribute" || isTextLikeNode(node)) { |
|
return false; |
|
} |
|
|
|
if (!node.parent) { |
|
return false; |
|
} |
|
|
|
if (typeof node.index !== "number" || node.index === 0) { |
|
return false; |
|
} |
|
|
|
var prevNode = node.parent.children[node.index - 1]; |
|
return isPrettierIgnore(prevNode); |
|
} |
|
|
|
function isPrettierIgnore(node) { |
|
return node.type === "comment" && node.value.trim() === "prettier-ignore"; |
|
} |
|
|
|
function getPrettierIgnoreAttributeCommentData(value) { |
|
var match = value.trim().match(/^prettier-ignore-attribute(?:\s+([^]+))?$/); |
|
|
|
if (!match) { |
|
return false; |
|
} |
|
|
|
if (!match[1]) { |
|
return true; |
|
} |
|
|
|
return match[1].split(/\s+/); |
|
} |
|
/** there's no opening/closing tag or it's considered not breakable */ |
|
|
|
|
|
function isTextLikeNode(node) { |
|
return node.type === "text" || node.type === "comment"; |
|
} |
|
|
|
function isScriptLikeTag(node) { |
|
return node.type === "element" && (node.fullName === "script" || node.fullName === "style" || node.fullName === "svg:style"); |
|
} |
|
|
|
function isFrontMatterNode(node) { |
|
return node.type === "yaml" || node.type === "toml"; |
|
} |
|
|
|
function canHaveInterpolation(node) { |
|
return node.children && !isScriptLikeTag(node); |
|
} |
|
|
|
function isWhitespaceSensitiveNode(node) { |
|
return isScriptLikeTag(node) || node.type === "interpolation" || isIndentationSensitiveNode(node); |
|
} |
|
|
|
function isIndentationSensitiveNode(node) { |
|
return getNodeCssStyleWhiteSpace(node).startsWith("pre"); |
|
} |
|
|
|
function isLeadingSpaceSensitiveNode(node) { |
|
var isLeadingSpaceSensitive = _isLeadingSpaceSensitiveNode(); |
|
|
|
if (isLeadingSpaceSensitive && !node.prev && node.parent && node.parent.tagDefinition && node.parent.tagDefinition.ignoreFirstLf) { |
|
return node.type === "interpolation"; |
|
} |
|
|
|
return isLeadingSpaceSensitive; |
|
|
|
function _isLeadingSpaceSensitiveNode() { |
|
if (isFrontMatterNode(node)) { |
|
return false; |
|
} |
|
|
|
if ((node.type === "text" || node.type === "interpolation") && node.prev && (node.prev.type === "text" || node.prev.type === "interpolation")) { |
|
return true; |
|
} |
|
|
|
if (!node.parent || node.parent.cssDisplay === "none") { |
|
return false; |
|
} |
|
|
|
if (isPreLikeNode(node.parent)) { |
|
return true; |
|
} |
|
|
|
if (!node.prev && (node.parent.type === "root" || isScriptLikeTag(node.parent) || !isFirstChildLeadingSpaceSensitiveCssDisplay(node.parent.cssDisplay))) { |
|
return false; |
|
} |
|
|
|
if (node.prev && !isNextLeadingSpaceSensitiveCssDisplay(node.prev.cssDisplay)) { |
|
return false; |
|
} |
|
|
|
return true; |
|
} |
|
} |
|
|
|
function isTrailingSpaceSensitiveNode(node) { |
|
if (isFrontMatterNode(node)) { |
|
return false; |
|
} |
|
|
|
if ((node.type === "text" || node.type === "interpolation") && node.next && (node.next.type === "text" || node.next.type === "interpolation")) { |
|
return true; |
|
} |
|
|
|
if (!node.parent || node.parent.cssDisplay === "none") { |
|
return false; |
|
} |
|
|
|
if (isPreLikeNode(node.parent)) { |
|
return true; |
|
} |
|
|
|
if (!node.next && (node.parent.type === "root" || isScriptLikeTag(node.parent) || !isLastChildTrailingSpaceSensitiveCssDisplay(node.parent.cssDisplay))) { |
|
return false; |
|
} |
|
|
|
if (node.next && !isPrevTrailingSpaceSensitiveCssDisplay(node.next.cssDisplay)) { |
|
return false; |
|
} |
|
|
|
return true; |
|
} |
|
|
|
function isDanglingSpaceSensitiveNode(node) { |
|
return isDanglingSpaceSensitiveCssDisplay(node.cssDisplay) && !isScriptLikeTag(node); |
|
} |
|
|
|
function forceNextEmptyLine(node) { |
|
return isFrontMatterNode(node) || node.next && node.sourceSpan.end.line + 1 < node.next.sourceSpan.start.line; |
|
} |
|
/** firstChild leadingSpaces and lastChild trailingSpaces */ |
|
|
|
|
|
function forceBreakContent(node) { |
|
return forceBreakChildren(node) || node.type === "element" && node.children.length !== 0 && (["body", "script", "style"].indexOf(node.name) !== -1 || node.children.some(function (child) { |
|
return hasNonTextChild(child); |
|
})) || node.firstChild && node.firstChild === node.lastChild && hasLeadingLineBreak(node.firstChild) && (!node.lastChild.isTrailingSpaceSensitive || hasTrailingLineBreak(node.lastChild)); |
|
} |
|
/** spaces between children */ |
|
|
|
|
|
function forceBreakChildren(node) { |
|
return node.type === "element" && node.children.length !== 0 && (["html", "head", "ul", "ol", "select"].indexOf(node.name) !== -1 || node.cssDisplay.startsWith("table") && node.cssDisplay !== "table-cell"); |
|
} |
|
|
|
function preferHardlineAsLeadingSpaces(node) { |
|
return preferHardlineAsSurroundingSpaces(node) || node.prev && preferHardlineAsTrailingSpaces(node.prev) || hasSurroundingLineBreak(node); |
|
} |
|
|
|
function preferHardlineAsTrailingSpaces(node) { |
|
return preferHardlineAsSurroundingSpaces(node) || node.type === "element" && node.fullName === "br" || hasSurroundingLineBreak(node); |
|
} |
|
|
|
function hasSurroundingLineBreak(node) { |
|
return hasLeadingLineBreak(node) && hasTrailingLineBreak(node); |
|
} |
|
|
|
function hasLeadingLineBreak(node) { |
|
return node.hasLeadingSpaces && (node.prev ? node.prev.sourceSpan.end.line < node.sourceSpan.start.line : node.parent.type === "root" || node.parent.startSourceSpan.end.line < node.sourceSpan.start.line); |
|
} |
|
|
|
function hasTrailingLineBreak(node) { |
|
return node.hasTrailingSpaces && (node.next ? node.next.sourceSpan.start.line > node.sourceSpan.end.line : node.parent.type === "root" || node.parent.endSourceSpan.start.line > node.sourceSpan.end.line); |
|
} |
|
|
|
function preferHardlineAsSurroundingSpaces(node) { |
|
switch (node.type) { |
|
case "ieConditionalComment": |
|
case "comment": |
|
case "directive": |
|
return true; |
|
|
|
case "element": |
|
return ["script", "select"].indexOf(node.name) !== -1; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function getLastDescendant(node) { |
|
return node.lastChild ? getLastDescendant(node.lastChild) : node; |
|
} |
|
|
|
function hasNonTextChild(node) { |
|
return node.children && node.children.some(function (child) { |
|
return child.type !== "text"; |
|
}); |
|
} |
|
|
|
function inferScriptParser(node) { |
|
if (node.name === "script" && !node.attrMap.src) { |
|
if (!node.attrMap.lang && !node.attrMap.type || node.attrMap.type === "module" || node.attrMap.type === "text/javascript" || node.attrMap.type === "text/babel" || node.attrMap.type === "application/javascript") { |
|
return "babel"; |
|
} |
|
|
|
if (node.attrMap.type === "application/x-typescript" || node.attrMap.lang === "ts" || node.attrMap.lang === "tsx") { |
|
return "typescript"; |
|
} |
|
|
|
if (node.attrMap.type === "text/markdown") { |
|
return "markdown"; |
|
} |
|
|
|
if (node.attrMap.type.endsWith("json") || node.attrMap.type.endsWith("importmap")) { |
|
return "json"; |
|
} |
|
} |
|
|
|
if (node.name === "style") { |
|
if (!node.attrMap.lang || node.attrMap.lang === "postcss" || node.attrMap.lang === "css") { |
|
return "css"; |
|
} |
|
|
|
if (node.attrMap.lang === "scss") { |
|
return "scss"; |
|
} |
|
|
|
if (node.attrMap.lang === "less") { |
|
return "less"; |
|
} |
|
} |
|
|
|
return null; |
|
} |
|
|
|
function isBlockLikeCssDisplay(cssDisplay) { |
|
return cssDisplay === "block" || cssDisplay === "list-item" || cssDisplay.startsWith("table"); |
|
} |
|
|
|
function isFirstChildLeadingSpaceSensitiveCssDisplay(cssDisplay) { |
|
return !isBlockLikeCssDisplay(cssDisplay) && cssDisplay !== "inline-block"; |
|
} |
|
|
|
function isLastChildTrailingSpaceSensitiveCssDisplay(cssDisplay) { |
|
return !isBlockLikeCssDisplay(cssDisplay) && cssDisplay !== "inline-block"; |
|
} |
|
|
|
function isPrevTrailingSpaceSensitiveCssDisplay(cssDisplay) { |
|
return !isBlockLikeCssDisplay(cssDisplay); |
|
} |
|
|
|
function isNextLeadingSpaceSensitiveCssDisplay(cssDisplay) { |
|
return !isBlockLikeCssDisplay(cssDisplay); |
|
} |
|
|
|
function isDanglingSpaceSensitiveCssDisplay(cssDisplay) { |
|
return !isBlockLikeCssDisplay(cssDisplay) && cssDisplay !== "inline-block"; |
|
} |
|
|
|
function isPreLikeNode(node) { |
|
return getNodeCssStyleWhiteSpace(node).startsWith("pre"); |
|
} |
|
|
|
function countParents(path) { |
|
var predicate = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : function () { |
|
return true; |
|
}; |
|
var counter = 0; |
|
|
|
for (var i = path.stack.length - 1; i >= 0; i--) { |
|
var value = path.stack[i]; |
|
|
|
if (value && _typeof(value) === "object" && !Array.isArray(value) && predicate(value)) { |
|
counter++; |
|
} |
|
} |
|
|
|
return counter; |
|
} |
|
|
|
function hasParent(node, fn) { |
|
var current = node; |
|
|
|
while (current) { |
|
if (fn(current)) { |
|
return true; |
|
} |
|
|
|
current = current.parent; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function getNodeCssStyleDisplay(node, options) { |
|
if (node.prev && node.prev.type === "comment") { |
|
// <!-- display: block --> |
|
var match = node.prev.value.match(/^\s*display:\s*([a-z]+)\s*$/); |
|
|
|
if (match) { |
|
return match[1]; |
|
} |
|
} |
|
|
|
var isInSvgForeignObject = false; |
|
|
|
if (node.type === "element" && node.namespace === "svg") { |
|
if (hasParent(node, function (parent) { |
|
return parent.fullName === "svg:foreignObject"; |
|
})) { |
|
isInSvgForeignObject = true; |
|
} else { |
|
return node.name === "svg" ? "inline-block" : "block"; |
|
} |
|
} |
|
|
|
switch (options.htmlWhitespaceSensitivity) { |
|
case "strict": |
|
return "inline"; |
|
|
|
case "ignore": |
|
return "block"; |
|
|
|
default: |
|
return node.type === "element" && (!node.namespace || isInSvgForeignObject) && CSS_DISPLAY_TAGS[node.name] || CSS_DISPLAY_DEFAULT; |
|
} |
|
} |
|
|
|
function getNodeCssStyleWhiteSpace(node) { |
|
return node.type === "element" && !node.namespace && CSS_WHITE_SPACE_TAGS[node.name] || CSS_WHITE_SPACE_DEFAULT; |
|
} |
|
|
|
function getMinIndentation(text) { |
|
var minIndentation = Infinity; |
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = text.split("\n")[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var lineText = _step2.value; |
|
|
|
if (lineText.length === 0) { |
|
continue; |
|
} |
|
|
|
if (/\S/.test(lineText[0])) { |
|
return 0; |
|
} |
|
|
|
var indentation = lineText.match(/^\s*/)[0].length; |
|
|
|
if (lineText.length === indentation) { |
|
continue; |
|
} |
|
|
|
if (indentation < minIndentation) { |
|
minIndentation = indentation; |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
|
|
return minIndentation === Infinity ? 0 : minIndentation; |
|
} |
|
|
|
function dedentString(text) { |
|
var minIndent = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : getMinIndentation(text); |
|
return minIndent === 0 ? text : text.split("\n").map(function (lineText) { |
|
return lineText.slice(minIndent); |
|
}).join("\n"); |
|
} |
|
|
|
function normalizeParts(parts) { |
|
var newParts = []; |
|
var restParts = parts.slice(); |
|
|
|
while (restParts.length !== 0) { |
|
var part = restParts.shift(); |
|
|
|
if (!part) { |
|
continue; |
|
} |
|
|
|
if (part.type === "concat") { |
|
Array.prototype.unshift.apply(restParts, part.parts); |
|
continue; |
|
} |
|
|
|
if (newParts.length !== 0 && typeof newParts[newParts.length - 1] === "string" && typeof part === "string") { |
|
newParts.push(newParts.pop() + part); |
|
continue; |
|
} |
|
|
|
newParts.push(part); |
|
} |
|
|
|
return newParts; |
|
} |
|
|
|
function identity(x) { |
|
return x; |
|
} |
|
|
|
function shouldNotPrintClosingTag(node, options) { |
|
return !node.isSelfClosing && !node.endSourceSpan && (hasPrettierIgnore(node) || shouldPreserveContent(node.parent, options)); |
|
} |
|
|
|
function countChars(text, char) { |
|
var counter = 0; |
|
|
|
for (var i = 0; i < text.length; i++) { |
|
if (text[i] === char) { |
|
counter++; |
|
} |
|
} |
|
|
|
return counter; |
|
} |
|
|
|
function unescapeQuoteEntities(text) { |
|
return text.replace(/'/g, "'").replace(/"/g, '"'); |
|
} |
|
|
|
var utils$3 = { |
|
HTML_ELEMENT_ATTRIBUTES: HTML_ELEMENT_ATTRIBUTES, |
|
HTML_TAGS: HTML_TAGS, |
|
canHaveInterpolation: canHaveInterpolation, |
|
countChars: countChars, |
|
countParents: countParents, |
|
dedentString: dedentString, |
|
forceBreakChildren: forceBreakChildren, |
|
forceBreakContent: forceBreakContent, |
|
forceNextEmptyLine: forceNextEmptyLine, |
|
getLastDescendant: getLastDescendant, |
|
getNodeCssStyleDisplay: getNodeCssStyleDisplay, |
|
getNodeCssStyleWhiteSpace: getNodeCssStyleWhiteSpace, |
|
getPrettierIgnoreAttributeCommentData: getPrettierIgnoreAttributeCommentData, |
|
hasPrettierIgnore: hasPrettierIgnore, |
|
identity: identity, |
|
inferScriptParser: inferScriptParser, |
|
isDanglingSpaceSensitiveNode: isDanglingSpaceSensitiveNode, |
|
isFrontMatterNode: isFrontMatterNode, |
|
isIndentationSensitiveNode: isIndentationSensitiveNode, |
|
isLeadingSpaceSensitiveNode: isLeadingSpaceSensitiveNode, |
|
isPreLikeNode: isPreLikeNode, |
|
isScriptLikeTag: isScriptLikeTag, |
|
isTextLikeNode: isTextLikeNode, |
|
isTrailingSpaceSensitiveNode: isTrailingSpaceSensitiveNode, |
|
isWhitespaceSensitiveNode: isWhitespaceSensitiveNode, |
|
normalizeParts: normalizeParts, |
|
preferHardlineAsLeadingSpaces: preferHardlineAsLeadingSpaces, |
|
preferHardlineAsTrailingSpaces: preferHardlineAsTrailingSpaces, |
|
shouldNotPrintClosingTag: shouldNotPrintClosingTag, |
|
shouldPreserveContent: shouldPreserveContent, |
|
unescapeQuoteEntities: unescapeQuoteEntities |
|
}; |
|
|
|
var canHaveInterpolation$1 = utils$3.canHaveInterpolation, |
|
getNodeCssStyleDisplay$1 = utils$3.getNodeCssStyleDisplay, |
|
isDanglingSpaceSensitiveNode$1 = utils$3.isDanglingSpaceSensitiveNode, |
|
isIndentationSensitiveNode$1 = utils$3.isIndentationSensitiveNode, |
|
isLeadingSpaceSensitiveNode$1 = utils$3.isLeadingSpaceSensitiveNode, |
|
isTrailingSpaceSensitiveNode$1 = utils$3.isTrailingSpaceSensitiveNode, |
|
isWhitespaceSensitiveNode$1 = utils$3.isWhitespaceSensitiveNode; |
|
var PREPROCESS_PIPELINE = [removeIgnorableFirstLf, mergeIeConditonalStartEndCommentIntoElementOpeningTag, mergeCdataIntoText, extractInterpolation, extractWhitespaces, addCssDisplay, addIsSelfClosing, addHasHtmComponentClosingTag, addIsSpaceSensitive, mergeSimpleElementIntoText]; |
|
|
|
function preprocess(ast, options) { |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = PREPROCESS_PIPELINE[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var fn = _step.value; |
|
ast = fn(ast, options); |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return ast; |
|
} |
|
|
|
function removeIgnorableFirstLf(ast |
|
/*, options */ |
|
) { |
|
return ast.map(function (node) { |
|
if (node.type === "element" && node.tagDefinition.ignoreFirstLf && node.children.length !== 0 && node.children[0].type === "text" && node.children[0].value[0] === "\n") { |
|
var text = node.children[0]; |
|
return node.clone({ |
|
children: text.value.length === 1 ? node.children.slice(1) : [].concat(text.clone({ |
|
value: text.value.slice(1) |
|
}), node.children.slice(1)) |
|
}); |
|
} |
|
|
|
return node; |
|
}); |
|
} |
|
|
|
function mergeIeConditonalStartEndCommentIntoElementOpeningTag(ast |
|
/*, options */ |
|
) { |
|
/** |
|
* <!--[if ...]><!--><target><!--<![endif]--> |
|
*/ |
|
var isTarget = function isTarget(node) { |
|
return node.type === "element" && node.prev && node.prev.type === "ieConditionalStartComment" && node.prev.sourceSpan.end.offset === node.startSourceSpan.start.offset && node.firstChild && node.firstChild.type === "ieConditionalEndComment" && node.firstChild.sourceSpan.start.offset === node.startSourceSpan.end.offset; |
|
}; |
|
|
|
return ast.map(function (node) { |
|
if (node.children) { |
|
var isTargetResults = node.children.map(isTarget); |
|
|
|
if (isTargetResults.some(Boolean)) { |
|
var newChildren = []; |
|
|
|
for (var i = 0; i < node.children.length; i++) { |
|
var child = node.children[i]; |
|
|
|
if (isTargetResults[i + 1]) { |
|
// ieConditionalStartComment |
|
continue; |
|
} |
|
|
|
if (isTargetResults[i]) { |
|
var ieConditionalStartComment = child.prev; |
|
var ieConditionalEndComment = child.firstChild; |
|
var ParseSourceSpan = child.sourceSpan.constructor; |
|
var startSourceSpan = new ParseSourceSpan(ieConditionalStartComment.sourceSpan.start, ieConditionalEndComment.sourceSpan.end); |
|
var sourceSpan = new ParseSourceSpan(startSourceSpan.start, child.sourceSpan.end); |
|
newChildren.push(child.clone({ |
|
condition: ieConditionalStartComment.condition, |
|
sourceSpan: sourceSpan, |
|
startSourceSpan: startSourceSpan, |
|
children: child.children.slice(1) |
|
})); |
|
continue; |
|
} |
|
|
|
newChildren.push(child); |
|
} |
|
|
|
return node.clone({ |
|
children: newChildren |
|
}); |
|
} |
|
} |
|
|
|
return node; |
|
}); |
|
} |
|
|
|
function mergeNodeIntoText(ast, shouldMerge, getValue) { |
|
return ast.map(function (node) { |
|
if (node.children) { |
|
var shouldMergeResults = node.children.map(shouldMerge); |
|
|
|
if (shouldMergeResults.some(Boolean)) { |
|
var newChildren = []; |
|
|
|
for (var i = 0; i < node.children.length; i++) { |
|
var child = node.children[i]; |
|
|
|
if (child.type !== "text" && !shouldMergeResults[i]) { |
|
newChildren.push(child); |
|
continue; |
|
} |
|
|
|
var newChild = child.type === "text" ? child : child.clone({ |
|
type: "text", |
|
value: getValue(child) |
|
}); |
|
|
|
if (newChildren.length === 0 || newChildren[newChildren.length - 1].type !== "text") { |
|
newChildren.push(newChild); |
|
continue; |
|
} |
|
|
|
var lastChild = newChildren.pop(); |
|
var ParseSourceSpan = lastChild.sourceSpan.constructor; |
|
newChildren.push(lastChild.clone({ |
|
value: lastChild.value + newChild.value, |
|
sourceSpan: new ParseSourceSpan(lastChild.sourceSpan.start, newChild.sourceSpan.end) |
|
})); |
|
} |
|
|
|
return node.clone({ |
|
children: newChildren |
|
}); |
|
} |
|
} |
|
|
|
return node; |
|
}); |
|
} |
|
|
|
function mergeCdataIntoText(ast |
|
/*, options */ |
|
) { |
|
return mergeNodeIntoText(ast, function (node) { |
|
return node.type === "cdata"; |
|
}, function (node) { |
|
return "<![CDATA[".concat(node.value, "]]>"); |
|
}); |
|
} |
|
|
|
function mergeSimpleElementIntoText(ast |
|
/*, options */ |
|
) { |
|
var isSimpleElement = function isSimpleElement(node) { |
|
return node.type === "element" && node.attrs.length === 0 && node.children.length === 1 && node.firstChild.type === "text" && // \xA0: non-breaking whitespace |
|
!/[^\S\xA0]/.test(node.children[0].value) && !node.firstChild.hasLeadingSpaces && !node.firstChild.hasTrailingSpaces && node.isLeadingSpaceSensitive && !node.hasLeadingSpaces && node.isTrailingSpaceSensitive && !node.hasTrailingSpaces && node.prev && node.prev.type === "text" && node.next && node.next.type === "text"; |
|
}; |
|
|
|
return ast.map(function (node) { |
|
if (node.children) { |
|
var isSimpleElementResults = node.children.map(isSimpleElement); |
|
|
|
if (isSimpleElementResults.some(Boolean)) { |
|
var newChildren = []; |
|
|
|
for (var i = 0; i < node.children.length; i++) { |
|
var child = node.children[i]; |
|
|
|
if (isSimpleElementResults[i]) { |
|
var lastChild = newChildren.pop(); |
|
var nextChild = node.children[++i]; |
|
var ParseSourceSpan = node.sourceSpan.constructor; |
|
var isTrailingSpaceSensitive = nextChild.isTrailingSpaceSensitive, |
|
hasTrailingSpaces = nextChild.hasTrailingSpaces; |
|
newChildren.push(lastChild.clone({ |
|
value: lastChild.value + "<".concat(child.rawName, ">") + child.firstChild.value + "</".concat(child.rawName, ">") + nextChild.value, |
|
sourceSpan: new ParseSourceSpan(lastChild.sourceSpan.start, nextChild.sourceSpan.end), |
|
isTrailingSpaceSensitive: isTrailingSpaceSensitive, |
|
hasTrailingSpaces: hasTrailingSpaces |
|
})); |
|
} else { |
|
newChildren.push(child); |
|
} |
|
} |
|
|
|
return node.clone({ |
|
children: newChildren |
|
}); |
|
} |
|
} |
|
|
|
return node; |
|
}); |
|
} |
|
|
|
function extractInterpolation(ast, options) { |
|
if (options.parser === "html") { |
|
return ast; |
|
} |
|
|
|
var interpolationRegex = /\{\{([\s\S]+?)\}\}/g; |
|
return ast.map(function (node) { |
|
if (!canHaveInterpolation$1(node)) { |
|
return node; |
|
} |
|
|
|
var newChildren = []; |
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = node.children[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var child = _step2.value; |
|
|
|
if (child.type !== "text") { |
|
newChildren.push(child); |
|
continue; |
|
} |
|
|
|
var ParseSourceSpan = child.sourceSpan.constructor; |
|
var startSourceSpan = child.sourceSpan.start; |
|
var endSourceSpan = null; |
|
var components = child.value.split(interpolationRegex); |
|
|
|
for (var i = 0; i < components.length; i++, startSourceSpan = endSourceSpan) { |
|
var value = components[i]; |
|
|
|
if (i % 2 === 0) { |
|
endSourceSpan = startSourceSpan.moveBy(value.length); |
|
|
|
if (value.length !== 0) { |
|
newChildren.push({ |
|
type: "text", |
|
value: value, |
|
sourceSpan: new ParseSourceSpan(startSourceSpan, endSourceSpan) |
|
}); |
|
} |
|
|
|
continue; |
|
} |
|
|
|
endSourceSpan = startSourceSpan.moveBy(value.length + 4); // `{{` + `}}` |
|
|
|
newChildren.push({ |
|
type: "interpolation", |
|
sourceSpan: new ParseSourceSpan(startSourceSpan, endSourceSpan), |
|
children: value.length === 0 ? [] : [{ |
|
type: "text", |
|
value: value, |
|
sourceSpan: new ParseSourceSpan(startSourceSpan.moveBy(2), endSourceSpan.moveBy(-2)) |
|
}] |
|
}); |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
|
|
return node.clone({ |
|
children: newChildren |
|
}); |
|
}); |
|
} |
|
/** |
|
* - add `hasLeadingSpaces` field |
|
* - add `hasTrailingSpaces` field |
|
* - add `hasDanglingSpaces` field for parent nodes |
|
* - add `isWhitespaceSensitive`, `isIndentationSensitive` field for text nodes |
|
* - remove insensitive whitespaces |
|
*/ |
|
|
|
|
|
function extractWhitespaces(ast |
|
/*, options*/ |
|
) { |
|
var TYPE_WHITESPACE = "whitespace"; |
|
return ast.map(function (node) { |
|
if (!node.children) { |
|
return node; |
|
} |
|
|
|
if (node.children.length === 0 || node.children.length === 1 && node.children[0].type === "text" && node.children[0].value.trim().length === 0) { |
|
return node.clone({ |
|
children: [], |
|
hasDanglingSpaces: node.children.length !== 0 |
|
}); |
|
} |
|
|
|
var isWhitespaceSensitive = isWhitespaceSensitiveNode$1(node); |
|
var isIndentationSensitive = isIndentationSensitiveNode$1(node); |
|
return node.clone({ |
|
isWhitespaceSensitive: isWhitespaceSensitive, |
|
isIndentationSensitive: isIndentationSensitive, |
|
children: node.children // extract whitespace nodes |
|
.reduce(function (newChildren, child) { |
|
if (child.type !== "text" || isWhitespaceSensitive) { |
|
return newChildren.concat(child); |
|
} |
|
|
|
var localChildren = []; |
|
|
|
var _child$value$match = child.value.match(/^(\s*)([\s\S]*?)(\s*)$/), |
|
_child$value$match2 = _slicedToArray(_child$value$match, 4), |
|
leadingSpaces = _child$value$match2[1], |
|
text = _child$value$match2[2], |
|
trailingSpaces = _child$value$match2[3]; |
|
|
|
if (leadingSpaces) { |
|
localChildren.push({ |
|
type: TYPE_WHITESPACE |
|
}); |
|
} |
|
|
|
var ParseSourceSpan = child.sourceSpan.constructor; |
|
|
|
if (text) { |
|
localChildren.push({ |
|
type: "text", |
|
value: text, |
|
sourceSpan: new ParseSourceSpan(child.sourceSpan.start.moveBy(leadingSpaces.length), child.sourceSpan.end.moveBy(-trailingSpaces.length)) |
|
}); |
|
} |
|
|
|
if (trailingSpaces) { |
|
localChildren.push({ |
|
type: TYPE_WHITESPACE |
|
}); |
|
} |
|
|
|
return newChildren.concat(localChildren); |
|
}, []) // set hasLeadingSpaces/hasTrailingSpaces and filter whitespace nodes |
|
.reduce(function (newChildren, child, i, children) { |
|
if (child.type === TYPE_WHITESPACE) { |
|
return newChildren; |
|
} |
|
|
|
var hasLeadingSpaces = i !== 0 && children[i - 1].type === TYPE_WHITESPACE; |
|
var hasTrailingSpaces = i !== children.length - 1 && children[i + 1].type === TYPE_WHITESPACE; |
|
return newChildren.concat(Object.assign({}, child, { |
|
hasLeadingSpaces: hasLeadingSpaces, |
|
hasTrailingSpaces: hasTrailingSpaces |
|
})); |
|
}, []) |
|
}); |
|
}); |
|
} |
|
|
|
function addIsSelfClosing(ast |
|
/*, options */ |
|
) { |
|
return ast.map(function (node) { |
|
return Object.assign(node, { |
|
isSelfClosing: !node.children || node.type === "element" && (node.tagDefinition.isVoid || // self-closing |
|
node.startSourceSpan === node.endSourceSpan) |
|
}); |
|
}); |
|
} |
|
|
|
function addHasHtmComponentClosingTag(ast, options) { |
|
return ast.map(function (node) { |
|
return node.type !== "element" ? node : Object.assign(node, { |
|
hasHtmComponentClosingTag: node.endSourceSpan && /^<\s*\/\s*\/\s*>$/.test(options.originalText.slice(node.endSourceSpan.start.offset, node.endSourceSpan.end.offset)) |
|
}); |
|
}); |
|
} |
|
|
|
function addCssDisplay(ast, options) { |
|
return ast.map(function (node) { |
|
return Object.assign(node, { |
|
cssDisplay: getNodeCssStyleDisplay$1(node, options) |
|
}); |
|
}); |
|
} |
|
/** |
|
* - add `isLeadingSpaceSensitive` field |
|
* - add `isTrailingSpaceSensitive` field |
|
* - add `isDanglingSpaceSensitive` field for parent nodes |
|
*/ |
|
|
|
|
|
function addIsSpaceSensitive(ast |
|
/*, options */ |
|
) { |
|
return ast.map(function (node) { |
|
if (!node.children) { |
|
return node; |
|
} |
|
|
|
if (node.children.length === 0) { |
|
return node.clone({ |
|
isDanglingSpaceSensitive: isDanglingSpaceSensitiveNode$1(node) |
|
}); |
|
} |
|
|
|
return node.clone({ |
|
children: node.children.map(function (child) { |
|
return Object.assign({}, child, { |
|
isLeadingSpaceSensitive: isLeadingSpaceSensitiveNode$1(child), |
|
isTrailingSpaceSensitive: isTrailingSpaceSensitiveNode$1(child) |
|
}); |
|
}).map(function (child, index, children) { |
|
return Object.assign({}, child, { |
|
isLeadingSpaceSensitive: index === 0 ? child.isLeadingSpaceSensitive : children[index - 1].isTrailingSpaceSensitive && child.isLeadingSpaceSensitive, |
|
isTrailingSpaceSensitive: index === children.length - 1 ? child.isTrailingSpaceSensitive : children[index + 1].isLeadingSpaceSensitive && child.isTrailingSpaceSensitive |
|
}); |
|
}) |
|
}); |
|
}); |
|
} |
|
|
|
var preprocess_1 = preprocess; |
|
|
|
function hasPragma$3(text) { |
|
return /^\s*<!--\s*@(format|prettier)\s*-->/.test(text); |
|
} |
|
|
|
function insertPragma$5(text) { |
|
return "<!-- @format -->\n\n" + text.replace(/^\s*\n/, ""); |
|
} |
|
|
|
var pragma$3 = { |
|
hasPragma: hasPragma$3, |
|
insertPragma: insertPragma$5 |
|
}; |
|
|
|
var _require$$0$builders$4 = doc.builders, |
|
concat$8 = _require$$0$builders$4.concat, |
|
group$8 = _require$$0$builders$4.group; |
|
/** |
|
* v-for="... in ..." |
|
* v-for="... of ..." |
|
* v-for="(..., ...) in ..." |
|
* v-for="(..., ...) of ..." |
|
*/ |
|
|
|
function printVueFor(value, textToDoc) { |
|
var _parseVueFor = parseVueFor(value), |
|
left = _parseVueFor.left, |
|
operator = _parseVueFor.operator, |
|
right = _parseVueFor.right; |
|
|
|
return concat$8([group$8(textToDoc("function _(".concat(left, ") {}"), { |
|
parser: "babel", |
|
__isVueForBindingLeft: true |
|
})), " ", operator, " ", textToDoc(right, { |
|
parser: "__js_expression" |
|
})]); |
|
} // modified from https://github.com/vuejs/vue/blob/v2.5.17/src/compiler/parser/index.js#L370-L387 |
|
|
|
|
|
function parseVueFor(value) { |
|
var forAliasRE = /([^]*?)\s+(in|of)\s+([^]*)/; |
|
var forIteratorRE = /,([^,}\]]*)(?:,([^,}\]]*))?$/; |
|
var stripParensRE = /^\(|\)$/g; |
|
var inMatch = value.match(forAliasRE); |
|
|
|
if (!inMatch) { |
|
return; |
|
} |
|
|
|
var res = {}; |
|
res.for = inMatch[3].trim(); |
|
var alias = inMatch[1].trim().replace(stripParensRE, ""); |
|
var iteratorMatch = alias.match(forIteratorRE); |
|
|
|
if (iteratorMatch) { |
|
res.alias = alias.replace(forIteratorRE, ""); |
|
res.iterator1 = iteratorMatch[1].trim(); |
|
|
|
if (iteratorMatch[2]) { |
|
res.iterator2 = iteratorMatch[2].trim(); |
|
} |
|
} else { |
|
res.alias = alias; |
|
} |
|
|
|
return { |
|
left: "".concat([res.alias, res.iterator1, res.iterator2].filter(Boolean).join(",")), |
|
operator: inMatch[2], |
|
right: res.for |
|
}; |
|
} |
|
|
|
function printVueSlotScope(value, textToDoc) { |
|
return textToDoc("function _(".concat(value, ") {}"), { |
|
parser: "babel", |
|
__isVueSlotScope: true |
|
}); |
|
} |
|
|
|
function isVueEventBindingExpression(eventBindingValue) { |
|
// https://github.com/vuejs/vue/blob/v2.5.17/src/compiler/codegen/events.js#L3-L4 |
|
// arrow function or anonymous function |
|
var fnExpRE = /^([\w$_]+|\([^)]*?\))\s*=>|^function\s*\(/; // simple member expression chain (a, a.b, a['b'], a["b"], a[0], a[b]) |
|
|
|
var simplePathRE = /^[A-Za-z_$][\w$]*(?:\.[A-Za-z_$][\w$]*|\['[^']*?']|\["[^"]*?"]|\[\d+]|\[[A-Za-z_$][\w$]*])*$/; // https://github.com/vuejs/vue/blob/v2.5.17/src/compiler/helpers.js#L104 |
|
|
|
var value = eventBindingValue.trim(); |
|
return fnExpRE.test(value) || simplePathRE.test(value); |
|
} |
|
|
|
var syntaxVue = { |
|
isVueEventBindingExpression: isVueEventBindingExpression, |
|
printVueFor: printVueFor, |
|
printVueSlotScope: printVueSlotScope |
|
}; |
|
|
|
var parseSrcset = createCommonjsModule(function (module) { |
|
/** |
|
* Srcset Parser |
|
* |
|
* By Alex Bell | MIT License |
|
* |
|
* JS Parser for the string value that appears in markup <img srcset="here"> |
|
* |
|
* @returns Array [{url: _, d: _, w: _, h:_}, ...] |
|
* |
|
* Based super duper closely on the reference algorithm at: |
|
* https://html.spec.whatwg.org/multipage/embedded-content.html#parse-a-srcset-attribute |
|
* |
|
* Most comments are copied in directly from the spec |
|
* (except for comments in parens). |
|
*/ |
|
(function (root, factory) { |
|
if ( module.exports) { |
|
// Node. Does not work with strict CommonJS, but |
|
// only CommonJS-like environments that support module.exports, |
|
// like Node. |
|
module.exports = factory(); |
|
} else { |
|
// Browser globals (root is window) |
|
root.parseSrcset = factory(); |
|
} |
|
})(commonjsGlobal, function () { |
|
// 1. Let input be the value passed to this algorithm. |
|
return function (input, options) { |
|
var logger = options && options.logger || console; // UTILITY FUNCTIONS |
|
// Manual is faster than RegEx |
|
// http://bjorn.tipling.com/state-and-regular-expressions-in-javascript |
|
// http://jsperf.com/whitespace-character/5 |
|
|
|
function isSpace(c) { |
|
return c === " " || // space |
|
c === "\t" || // horizontal tab |
|
c === "\n" || // new line |
|
c === "\f" || // form feed |
|
c === "\r"; // carriage return |
|
} |
|
|
|
function collectCharacters(regEx) { |
|
var chars, |
|
match = regEx.exec(input.substring(pos)); |
|
|
|
if (match) { |
|
chars = match[0]; |
|
pos += chars.length; |
|
return chars; |
|
} |
|
} |
|
|
|
var inputLength = input.length, |
|
// (Don't use \s, to avoid matching non-breaking space) |
|
regexLeadingSpaces = /^[ \t\n\r\u000c]+/, |
|
regexLeadingCommasOrSpaces = /^[, \t\n\r\u000c]+/, |
|
regexLeadingNotSpaces = /^[^ \t\n\r\u000c]+/, |
|
regexTrailingCommas = /[,]+$/, |
|
regexNonNegativeInteger = /^\d+$/, |
|
// ( Positive or negative or unsigned integers or decimals, without or without exponents. |
|
// Must include at least one digit. |
|
// According to spec tests any decimal point must be followed by a digit. |
|
// No leading plus sign is allowed.) |
|
// https://html.spec.whatwg.org/multipage/infrastructure.html#valid-floating-point-number |
|
regexFloatingPoint = /^-?(?:[0-9]+|[0-9]*\.[0-9]+)(?:[eE][+-]?[0-9]+)?$/, |
|
url, |
|
descriptors, |
|
currentDescriptor, |
|
state, |
|
c, |
|
// 2. Let position be a pointer into input, initially pointing at the start |
|
// of the string. |
|
pos = 0, |
|
// 3. Let candidates be an initially empty source set. |
|
candidates = []; // 4. Splitting loop: Collect a sequence of characters that are space |
|
// characters or U+002C COMMA characters. If any U+002C COMMA characters |
|
// were collected, that is a parse error. |
|
|
|
while (true) { |
|
collectCharacters(regexLeadingCommasOrSpaces); // 5. If position is past the end of input, return candidates and abort these steps. |
|
|
|
if (pos >= inputLength) { |
|
return candidates; // (we're done, this is the sole return path) |
|
} // 6. Collect a sequence of characters that are not space characters, |
|
// and let that be url. |
|
|
|
|
|
url = collectCharacters(regexLeadingNotSpaces); // 7. Let descriptors be a new empty list. |
|
|
|
descriptors = []; // 8. If url ends with a U+002C COMMA character (,), follow these substeps: |
|
// (1). Remove all trailing U+002C COMMA characters from url. If this removed |
|
// more than one character, that is a parse error. |
|
|
|
if (url.slice(-1) === ",") { |
|
url = url.replace(regexTrailingCommas, ""); // (Jump ahead to step 9 to skip tokenization and just push the candidate). |
|
|
|
parseDescriptors(); // Otherwise, follow these substeps: |
|
} else { |
|
tokenize(); |
|
} // (close else of step 8) |
|
// 16. Return to the step labeled splitting loop. |
|
|
|
} // (Close of big while loop.) |
|
|
|
/** |
|
* Tokenizes descriptor properties prior to parsing |
|
* Returns undefined. |
|
*/ |
|
|
|
|
|
function tokenize() { |
|
// 8.1. Descriptor tokeniser: Skip whitespace |
|
collectCharacters(regexLeadingSpaces); // 8.2. Let current descriptor be the empty string. |
|
|
|
currentDescriptor = ""; // 8.3. Let state be in descriptor. |
|
|
|
state = "in descriptor"; |
|
|
|
while (true) { |
|
// 8.4. Let c be the character at position. |
|
c = input.charAt(pos); // Do the following depending on the value of state. |
|
// For the purpose of this step, "EOF" is a special character representing |
|
// that position is past the end of input. |
|
// In descriptor |
|
|
|
if (state === "in descriptor") { |
|
// Do the following, depending on the value of c: |
|
// Space character |
|
// If current descriptor is not empty, append current descriptor to |
|
// descriptors and let current descriptor be the empty string. |
|
// Set state to after descriptor. |
|
if (isSpace(c)) { |
|
if (currentDescriptor) { |
|
descriptors.push(currentDescriptor); |
|
currentDescriptor = ""; |
|
state = "after descriptor"; |
|
} // U+002C COMMA (,) |
|
// Advance position to the next character in input. If current descriptor |
|
// is not empty, append current descriptor to descriptors. Jump to the step |
|
// labeled descriptor parser. |
|
|
|
} else if (c === ",") { |
|
pos += 1; |
|
|
|
if (currentDescriptor) { |
|
descriptors.push(currentDescriptor); |
|
} |
|
|
|
parseDescriptors(); |
|
return; // U+0028 LEFT PARENTHESIS (() |
|
// Append c to current descriptor. Set state to in parens. |
|
} else if (c === "(") { |
|
currentDescriptor = currentDescriptor + c; |
|
state = "in parens"; // EOF |
|
// If current descriptor is not empty, append current descriptor to |
|
// descriptors. Jump to the step labeled descriptor parser. |
|
} else if (c === "") { |
|
if (currentDescriptor) { |
|
descriptors.push(currentDescriptor); |
|
} |
|
|
|
parseDescriptors(); |
|
return; // Anything else |
|
// Append c to current descriptor. |
|
} else { |
|
currentDescriptor = currentDescriptor + c; |
|
} // (end "in descriptor" |
|
// In parens |
|
|
|
} else if (state === "in parens") { |
|
// U+0029 RIGHT PARENTHESIS ()) |
|
// Append c to current descriptor. Set state to in descriptor. |
|
if (c === ")") { |
|
currentDescriptor = currentDescriptor + c; |
|
state = "in descriptor"; // EOF |
|
// Append current descriptor to descriptors. Jump to the step labeled |
|
// descriptor parser. |
|
} else if (c === "") { |
|
descriptors.push(currentDescriptor); |
|
parseDescriptors(); |
|
return; // Anything else |
|
// Append c to current descriptor. |
|
} else { |
|
currentDescriptor = currentDescriptor + c; |
|
} // After descriptor |
|
|
|
} else if (state === "after descriptor") { |
|
// Do the following, depending on the value of c: |
|
// Space character: Stay in this state. |
|
if (isSpace(c)) ; else if (c === "") { |
|
parseDescriptors(); |
|
return; // Anything else |
|
// Set state to in descriptor. Set position to the previous character in input. |
|
} else { |
|
state = "in descriptor"; |
|
pos -= 1; |
|
} |
|
} // Advance position to the next character in input. |
|
|
|
|
|
pos += 1; // Repeat this step. |
|
} // (close while true loop) |
|
|
|
} |
|
/** |
|
* Adds descriptor properties to a candidate, pushes to the candidates array |
|
* @return undefined |
|
*/ |
|
// Declared outside of the while loop so that it's only created once. |
|
|
|
|
|
function parseDescriptors() { |
|
// 9. Descriptor parser: Let error be no. |
|
var pError = false, |
|
// 10. Let width be absent. |
|
// 11. Let density be absent. |
|
// 12. Let future-compat-h be absent. (We're implementing it now as h) |
|
w, |
|
d, |
|
h, |
|
i, |
|
candidate = {}, |
|
desc, |
|
lastChar, |
|
value, |
|
intVal, |
|
floatVal; // 13. For each descriptor in descriptors, run the appropriate set of steps |
|
// from the following list: |
|
|
|
for (i = 0; i < descriptors.length; i++) { |
|
desc = descriptors[i]; |
|
lastChar = desc[desc.length - 1]; |
|
value = desc.substring(0, desc.length - 1); |
|
intVal = parseInt(value, 10); |
|
floatVal = parseFloat(value); // If the descriptor consists of a valid non-negative integer followed by |
|
// a U+0077 LATIN SMALL LETTER W character |
|
|
|
if (regexNonNegativeInteger.test(value) && lastChar === "w") { |
|
// If width and density are not both absent, then let error be yes. |
|
if (w || d) { |
|
pError = true; |
|
} // Apply the rules for parsing non-negative integers to the descriptor. |
|
// If the result is zero, let error be yes. |
|
// Otherwise, let width be the result. |
|
|
|
|
|
if (intVal === 0) { |
|
pError = true; |
|
} else { |
|
w = intVal; |
|
} // If the descriptor consists of a valid floating-point number followed by |
|
// a U+0078 LATIN SMALL LETTER X character |
|
|
|
} else if (regexFloatingPoint.test(value) && lastChar === "x") { |
|
// If width, density and future-compat-h are not all absent, then let error |
|
// be yes. |
|
if (w || d || h) { |
|
pError = true; |
|
} // Apply the rules for parsing floating-point number values to the descriptor. |
|
// If the result is less than zero, let error be yes. Otherwise, let density |
|
// be the result. |
|
|
|
|
|
if (floatVal < 0) { |
|
pError = true; |
|
} else { |
|
d = floatVal; |
|
} // If the descriptor consists of a valid non-negative integer followed by |
|
// a U+0068 LATIN SMALL LETTER H character |
|
|
|
} else if (regexNonNegativeInteger.test(value) && lastChar === "h") { |
|
// If height and density are not both absent, then let error be yes. |
|
if (h || d) { |
|
pError = true; |
|
} // Apply the rules for parsing non-negative integers to the descriptor. |
|
// If the result is zero, let error be yes. Otherwise, let future-compat-h |
|
// be the result. |
|
|
|
|
|
if (intVal === 0) { |
|
pError = true; |
|
} else { |
|
h = intVal; |
|
} // Anything else, Let error be yes. |
|
|
|
} else { |
|
pError = true; |
|
} |
|
} // (close step 13 for loop) |
|
// 15. If error is still no, then append a new image source to candidates whose |
|
// URL is url, associated with a width width if not absent and a pixel |
|
// density density if not absent. Otherwise, there is a parse error. |
|
|
|
|
|
if (!pError) { |
|
candidate.url = url; |
|
|
|
if (w) { |
|
candidate.w = w; |
|
} |
|
|
|
if (d) { |
|
candidate.d = d; |
|
} |
|
|
|
if (h) { |
|
candidate.h = h; |
|
} |
|
|
|
candidates.push(candidate); |
|
} else if (logger && logger.error) { |
|
logger.error("Invalid srcset descriptor found in '" + input + "' at '" + desc + "'."); |
|
} |
|
} // (close parseDescriptors fn) |
|
|
|
}; |
|
}); |
|
}); |
|
|
|
var _require$$0$builders$5 = doc.builders, |
|
concat$9 = _require$$0$builders$5.concat, |
|
ifBreak$4 = _require$$0$builders$5.ifBreak, |
|
join$5 = _require$$0$builders$5.join, |
|
line$4 = _require$$0$builders$5.line; |
|
|
|
function printImgSrcset(value) { |
|
var srcset = parseSrcset(value, { |
|
logger: { |
|
error: function error(message) { |
|
throw new Error(message); |
|
} |
|
} |
|
}); |
|
var hasW = srcset.some(function (src) { |
|
return src.w; |
|
}); |
|
var hasH = srcset.some(function (src) { |
|
return src.h; |
|
}); |
|
var hasX = srcset.some(function (src) { |
|
return src.d; |
|
}); |
|
|
|
if (hasW + hasH + hasX !== 1) { |
|
throw new Error("Mixed descriptor in srcset is not supported"); |
|
} |
|
|
|
var key = hasW ? "w" : hasH ? "h" : "d"; |
|
var unit = hasW ? "w" : hasH ? "h" : "x"; |
|
|
|
var getMax = function getMax(values) { |
|
return Math.max.apply(Math, values); |
|
}; |
|
|
|
var urls = srcset.map(function (src) { |
|
return src.url; |
|
}); |
|
var maxUrlLength = getMax(urls.map(function (url) { |
|
return url.length; |
|
})); |
|
var descriptors = srcset.map(function (src) { |
|
return src[key]; |
|
}).map(function (descriptor) { |
|
return descriptor ? descriptor.toString() : ""; |
|
}); |
|
var descriptorLeftLengths = descriptors.map(function (descriptor) { |
|
var index = descriptor.indexOf("."); |
|
return index === -1 ? descriptor.length : index; |
|
}); |
|
var maxDescriptorLeftLength = getMax(descriptorLeftLengths); |
|
return join$5(concat$9([",", line$4]), urls.map(function (url, index) { |
|
var parts = [url]; |
|
var descriptor = descriptors[index]; |
|
|
|
if (descriptor) { |
|
var urlPadding = maxUrlLength - url.length + 1; |
|
var descriptorPadding = maxDescriptorLeftLength - descriptorLeftLengths[index]; |
|
var alignment = " ".repeat(urlPadding + descriptorPadding); |
|
parts.push(ifBreak$4(alignment, " "), descriptor + unit); |
|
} |
|
|
|
return concat$9(parts); |
|
})); |
|
} |
|
|
|
var syntaxAttribute = { |
|
printImgSrcset: printImgSrcset |
|
}; |
|
|
|
var builders = doc.builders, |
|
_require$$0$utils = doc.utils, |
|
stripTrailingHardline$1 = _require$$0$utils.stripTrailingHardline, |
|
mapDoc$4 = _require$$0$utils.mapDoc; |
|
var breakParent$2 = builders.breakParent, |
|
dedentToRoot$1 = builders.dedentToRoot, |
|
fill$3 = builders.fill, |
|
group$9 = builders.group, |
|
hardline$7 = builders.hardline, |
|
ifBreak$5 = builders.ifBreak, |
|
indent$5 = builders.indent, |
|
join$6 = builders.join, |
|
line$5 = builders.line, |
|
literalline$2 = builders.literalline, |
|
markAsRoot$2 = builders.markAsRoot, |
|
softline$4 = builders.softline; |
|
var countChars$1 = utils$3.countChars, |
|
countParents$1 = utils$3.countParents, |
|
dedentString$1 = utils$3.dedentString, |
|
forceBreakChildren$1 = utils$3.forceBreakChildren, |
|
forceBreakContent$1 = utils$3.forceBreakContent, |
|
forceNextEmptyLine$1 = utils$3.forceNextEmptyLine, |
|
getLastDescendant$1 = utils$3.getLastDescendant, |
|
getPrettierIgnoreAttributeCommentData$1 = utils$3.getPrettierIgnoreAttributeCommentData, |
|
hasPrettierIgnore$1 = utils$3.hasPrettierIgnore, |
|
inferScriptParser$1 = utils$3.inferScriptParser, |
|
isScriptLikeTag$1 = utils$3.isScriptLikeTag, |
|
isTextLikeNode$1 = utils$3.isTextLikeNode, |
|
normalizeParts$1 = utils$3.normalizeParts, |
|
preferHardlineAsLeadingSpaces$1 = utils$3.preferHardlineAsLeadingSpaces, |
|
shouldNotPrintClosingTag$1 = utils$3.shouldNotPrintClosingTag, |
|
shouldPreserveContent$1 = utils$3.shouldPreserveContent, |
|
unescapeQuoteEntities$1 = utils$3.unescapeQuoteEntities; |
|
var replaceEndOfLineWith$1 = util.replaceEndOfLineWith; |
|
var insertPragma$6 = pragma$3.insertPragma; |
|
var printVueFor$1 = syntaxVue.printVueFor, |
|
printVueSlotScope$1 = syntaxVue.printVueSlotScope, |
|
isVueEventBindingExpression$1 = syntaxVue.isVueEventBindingExpression; |
|
var printImgSrcset$1 = syntaxAttribute.printImgSrcset; |
|
|
|
function concat$a(parts) { |
|
var newParts = normalizeParts$1(parts); |
|
return newParts.length === 0 ? "" : newParts.length === 1 ? newParts[0] : builders.concat(newParts); |
|
} |
|
|
|
function embed$2(path, print, textToDoc, options) { |
|
var node = path.getValue(); |
|
|
|
switch (node.type) { |
|
case "text": |
|
{ |
|
if (isScriptLikeTag$1(node.parent)) { |
|
var parser = inferScriptParser$1(node.parent); |
|
|
|
if (parser) { |
|
var value = parser === "markdown" ? dedentString$1(node.value.replace(/^[^\S\n]*?\n/, "")) : node.value; |
|
return builders.concat([concat$a([breakParent$2, printOpeningTagPrefix(node, options), stripTrailingHardline$1(textToDoc(value, { |
|
parser: parser |
|
})), printClosingTagSuffix(node, options)])]); |
|
} |
|
} else if (node.parent.type === "interpolation") { |
|
return concat$a([indent$5(concat$a([line$5, textToDoc(node.value, Object.assign({ |
|
__isInHtmlInterpolation: true // to avoid unexpected `}}` |
|
|
|
}, options.parser === "angular" ? { |
|
parser: "__ng_interpolation", |
|
trailingComma: "none" |
|
} : options.parser === "vue" ? { |
|
parser: "__vue_expression" |
|
} : { |
|
parser: "__js_expression" |
|
}))])), node.parent.next && needsToBorrowPrevClosingTagEndMarker(node.parent.next) ? " " : line$5]); |
|
} |
|
|
|
break; |
|
} |
|
|
|
case "attribute": |
|
{ |
|
if (!node.value) { |
|
break; |
|
} // lit-html: html`<my-element obj=${obj}></my-element>` |
|
|
|
|
|
if (/^PRETTIER_HTML_PLACEHOLDER_\d+_\d+_IN_JS$/.test(options.originalText.slice(node.valueSpan.start.offset, node.valueSpan.end.offset))) { |
|
return concat$a([node.rawName, "=", node.value]); |
|
} // lwc: html`<my-element data-for={value}></my-element>` |
|
|
|
|
|
if (options.parser === "lwc") { |
|
var interpolationRegex = /^\{[\s\S]*\}$/; |
|
|
|
if (interpolationRegex.test(options.originalText.slice(node.valueSpan.start.offset, node.valueSpan.end.offset))) { |
|
return concat$a([node.rawName, "=", node.value]); |
|
} |
|
} |
|
|
|
var embeddedAttributeValueDoc = printEmbeddedAttributeValue(node, function (code, opts) { |
|
return (// strictly prefer single quote to avoid unnecessary html entity escape |
|
textToDoc(code, Object.assign({ |
|
__isInHtmlAttribute: true |
|
}, opts)) |
|
); |
|
}, options); |
|
|
|
if (embeddedAttributeValueDoc) { |
|
return concat$a([node.rawName, '="', group$9(mapDoc$4(embeddedAttributeValueDoc, function (doc) { |
|
return typeof doc === "string" ? doc.replace(/"/g, """) : doc; |
|
})), '"']); |
|
} |
|
|
|
break; |
|
} |
|
|
|
case "yaml": |
|
return markAsRoot$2(concat$a(["---", hardline$7, node.value.trim().length === 0 ? "" : textToDoc(node.value, { |
|
parser: "yaml" |
|
}), "---"])); |
|
} |
|
} |
|
|
|
function genericPrint$2(path, options, print) { |
|
var node = path.getValue(); |
|
|
|
switch (node.type) { |
|
case "root": |
|
// use original concat to not break stripTrailingHardline |
|
return builders.concat([group$9(printChildren$1(path, options, print)), hardline$7]); |
|
|
|
case "element": |
|
case "ieConditionalComment": |
|
{ |
|
/** |
|
* do not break: |
|
* |
|
* <div>{{ |
|
* ~ |
|
* interpolation |
|
* }}</div> |
|
* ~ |
|
* |
|
* exception: break if the opening tag breaks |
|
* |
|
* <div |
|
* long |
|
* ~ |
|
* >{{ |
|
* interpolation |
|
* }}</div |
|
* ~ |
|
* > |
|
*/ |
|
var shouldHugContent = node.children.length === 1 && node.firstChild.type === "interpolation" && node.firstChild.isLeadingSpaceSensitive && !node.firstChild.hasLeadingSpaces && node.lastChild.isTrailingSpaceSensitive && !node.lastChild.hasTrailingSpaces; |
|
var attrGroupId = Symbol("element-attr-group-id"); |
|
return concat$a([group$9(concat$a([group$9(printOpeningTag(path, options, print), { |
|
id: attrGroupId |
|
}), node.children.length === 0 ? node.hasDanglingSpaces && node.isDanglingSpaceSensitive ? line$5 : "" : concat$a([forceBreakContent$1(node) ? breakParent$2 : "", function (childrenDoc) { |
|
return shouldHugContent ? ifBreak$5(indent$5(childrenDoc), childrenDoc, { |
|
groupId: attrGroupId |
|
}) : isScriptLikeTag$1(node) && node.parent.type === "root" && options.parser === "vue" && !options.vueIndentScriptAndStyle ? childrenDoc : indent$5(childrenDoc); |
|
}(concat$a([shouldHugContent ? ifBreak$5(softline$4, "", { |
|
groupId: attrGroupId |
|
}) : node.firstChild.hasLeadingSpaces && node.firstChild.isLeadingSpaceSensitive ? line$5 : node.firstChild.type === "text" && node.isWhitespaceSensitive && node.isIndentationSensitive ? dedentToRoot$1(softline$4) : softline$4, printChildren$1(path, options, print)])), (node.next ? needsToBorrowPrevClosingTagEndMarker(node.next) : needsToBorrowLastChildClosingTagEndMarker(node.parent)) ? node.lastChild.hasTrailingSpaces && node.lastChild.isTrailingSpaceSensitive ? " " : "" : shouldHugContent ? ifBreak$5(softline$4, "", { |
|
groupId: attrGroupId |
|
}) : node.lastChild.hasTrailingSpaces && node.lastChild.isTrailingSpaceSensitive ? line$5 : (node.lastChild.type === "comment" || node.lastChild.type === "text" && node.isWhitespaceSensitive && node.isIndentationSensitive) && new RegExp("\\n\\s{".concat(options.tabWidth * countParents$1(path, function (n) { |
|
return n.parent && n.parent.type !== "root"; |
|
}), "}$")).test(node.lastChild.value) ? |
|
/** |
|
* <div> |
|
* <pre> |
|
* something |
|
* </pre> |
|
* ~ |
|
* </div> |
|
*/ |
|
"" : softline$4])])), printClosingTag(node, options)]); |
|
} |
|
|
|
case "ieConditionalStartComment": |
|
case "ieConditionalEndComment": |
|
return concat$a([printOpeningTagStart(node), printClosingTagEnd(node)]); |
|
|
|
case "interpolation": |
|
return concat$a([printOpeningTagStart(node, options), concat$a(path.map(print, "children")), printClosingTagEnd(node, options)]); |
|
|
|
case "text": |
|
{ |
|
if (node.parent.type === "interpolation") { |
|
// replace the trailing literalline with hardline for better readability |
|
var trailingNewlineRegex = /\n[^\S\n]*?$/; |
|
var hasTrailingNewline = trailingNewlineRegex.test(node.value); |
|
var value = hasTrailingNewline ? node.value.replace(trailingNewlineRegex, "") : node.value; |
|
return concat$a([concat$a(replaceEndOfLineWith$1(value, literalline$2)), hasTrailingNewline ? hardline$7 : ""]); |
|
} |
|
|
|
return fill$3(normalizeParts$1([].concat(printOpeningTagPrefix(node, options), getTextValueParts(node), printClosingTagSuffix(node, options)))); |
|
} |
|
|
|
case "docType": |
|
return concat$a([group$9(concat$a([printOpeningTagStart(node, options), " ", node.value.replace(/^html\b/i, "html").replace(/\s+/g, " ")])), printClosingTagEnd(node, options)]); |
|
|
|
case "comment": |
|
{ |
|
return concat$a([printOpeningTagPrefix(node, options), concat$a(replaceEndOfLineWith$1(options.originalText.slice(options.locStart(node), options.locEnd(node)), literalline$2)), printClosingTagSuffix(node, options)]); |
|
} |
|
|
|
case "attribute": |
|
{ |
|
if (node.value === null) { |
|
return node.rawName; |
|
} |
|
|
|
var _value = unescapeQuoteEntities$1(node.value); |
|
|
|
var singleQuoteCount = countChars$1(_value, "'"); |
|
var doubleQuoteCount = countChars$1(_value, '"'); |
|
var quote = singleQuoteCount < doubleQuoteCount ? "'" : '"'; |
|
return concat$a([node.rawName, concat$a(["=", quote, concat$a(replaceEndOfLineWith$1(quote === '"' ? _value.replace(/"/g, """) : _value.replace(/'/g, "'"), literalline$2)), quote])]); |
|
} |
|
|
|
case "yaml": |
|
case "toml": |
|
return concat$a(replaceEndOfLineWith$1(node.raw, literalline$2)); |
|
|
|
default: |
|
throw new Error("Unexpected node type ".concat(node.type)); |
|
} |
|
} |
|
|
|
function printChildren$1(path, options, print) { |
|
var node = path.getValue(); |
|
|
|
if (forceBreakChildren$1(node)) { |
|
return concat$a([breakParent$2, concat$a(path.map(function (childPath) { |
|
var childNode = childPath.getValue(); |
|
var prevBetweenLine = !childNode.prev ? "" : printBetweenLine(childNode.prev, childNode); |
|
return concat$a([!prevBetweenLine ? "" : concat$a([prevBetweenLine, forceNextEmptyLine$1(childNode.prev) ? hardline$7 : ""]), printChild(childPath)]); |
|
}, "children"))]); |
|
} |
|
|
|
var groupIds = node.children.map(function () { |
|
return Symbol(""); |
|
}); |
|
return concat$a(path.map(function (childPath, childIndex) { |
|
var childNode = childPath.getValue(); |
|
|
|
if (isTextLikeNode$1(childNode)) { |
|
if (childNode.prev && isTextLikeNode$1(childNode.prev)) { |
|
var _prevBetweenLine = printBetweenLine(childNode.prev, childNode); |
|
|
|
if (_prevBetweenLine) { |
|
if (forceNextEmptyLine$1(childNode.prev)) { |
|
return concat$a([hardline$7, hardline$7, printChild(childPath)]); |
|
} |
|
|
|
return concat$a([_prevBetweenLine, printChild(childPath)]); |
|
} |
|
} |
|
|
|
return printChild(childPath); |
|
} |
|
|
|
var prevParts = []; |
|
var leadingParts = []; |
|
var trailingParts = []; |
|
var nextParts = []; |
|
var prevBetweenLine = childNode.prev ? printBetweenLine(childNode.prev, childNode) : ""; |
|
var nextBetweenLine = childNode.next ? printBetweenLine(childNode, childNode.next) : ""; |
|
|
|
if (prevBetweenLine) { |
|
if (forceNextEmptyLine$1(childNode.prev)) { |
|
prevParts.push(hardline$7, hardline$7); |
|
} else if (prevBetweenLine === hardline$7) { |
|
prevParts.push(hardline$7); |
|
} else { |
|
if (isTextLikeNode$1(childNode.prev)) { |
|
leadingParts.push(prevBetweenLine); |
|
} else { |
|
leadingParts.push(ifBreak$5("", softline$4, { |
|
groupId: groupIds[childIndex - 1] |
|
})); |
|
} |
|
} |
|
} |
|
|
|
if (nextBetweenLine) { |
|
if (forceNextEmptyLine$1(childNode)) { |
|
if (isTextLikeNode$1(childNode.next)) { |
|
nextParts.push(hardline$7, hardline$7); |
|
} |
|
} else if (nextBetweenLine === hardline$7) { |
|
if (isTextLikeNode$1(childNode.next)) { |
|
nextParts.push(hardline$7); |
|
} |
|
} else { |
|
trailingParts.push(nextBetweenLine); |
|
} |
|
} |
|
|
|
return concat$a([].concat(prevParts, group$9(concat$a([concat$a(leadingParts), group$9(concat$a([printChild(childPath), concat$a(trailingParts)]), { |
|
id: groupIds[childIndex] |
|
})])), nextParts)); |
|
}, "children")); |
|
|
|
function printChild(childPath) { |
|
var child = childPath.getValue(); |
|
|
|
if (hasPrettierIgnore$1(child)) { |
|
return concat$a([].concat(printOpeningTagPrefix(child, options), replaceEndOfLineWith$1(options.originalText.slice(options.locStart(child) + (child.prev && needsToBorrowNextOpeningTagStartMarker(child.prev) ? printOpeningTagStartMarker(child).length : 0), options.locEnd(child) - (child.next && needsToBorrowPrevClosingTagEndMarker(child.next) ? printClosingTagEndMarker(child, options).length : 0)), literalline$2), printClosingTagSuffix(child, options))); |
|
} |
|
|
|
if (shouldPreserveContent$1(child, options)) { |
|
return concat$a([].concat(printOpeningTagPrefix(child, options), group$9(printOpeningTag(childPath, options, print)), replaceEndOfLineWith$1(options.originalText.slice(child.startSourceSpan.end.offset + (child.firstChild && needsToBorrowParentOpeningTagEndMarker(child.firstChild) ? -printOpeningTagEndMarker(child).length : 0), child.endSourceSpan.start.offset + (child.lastChild && needsToBorrowParentClosingTagStartMarker(child.lastChild) ? printClosingTagStartMarker(child, options).length : needsToBorrowLastChildClosingTagEndMarker(child) ? -printClosingTagEndMarker(child.lastChild, options).length : 0)), literalline$2), printClosingTag(child, options), printClosingTagSuffix(child, options))); |
|
} |
|
|
|
return print(childPath); |
|
} |
|
|
|
function printBetweenLine(prevNode, nextNode) { |
|
return isTextLikeNode$1(prevNode) && isTextLikeNode$1(nextNode) ? prevNode.isTrailingSpaceSensitive ? prevNode.hasTrailingSpaces ? preferHardlineAsLeadingSpaces$1(nextNode) ? hardline$7 : line$5 : "" : preferHardlineAsLeadingSpaces$1(nextNode) ? hardline$7 : softline$4 : needsToBorrowNextOpeningTagStartMarker(prevNode) && ( |
|
/** |
|
* 123<a |
|
* ~ |
|
* ><b> |
|
*/ |
|
nextNode.firstChild || |
|
/** |
|
* 123<!-- |
|
* ~ |
|
* --> |
|
*/ |
|
nextNode.isSelfClosing || |
|
/** |
|
* 123<span |
|
* ~ |
|
* attr |
|
*/ |
|
nextNode.type === "element" && nextNode.attrs.length !== 0) || |
|
/** |
|
* <img |
|
* src="long" |
|
* ~ |
|
* />123 |
|
*/ |
|
prevNode.type === "element" && prevNode.isSelfClosing && needsToBorrowPrevClosingTagEndMarker(nextNode) ? "" : !nextNode.isLeadingSpaceSensitive || preferHardlineAsLeadingSpaces$1(nextNode) || |
|
/** |
|
* Want to write us a letter? Use our<a |
|
* ><b><a>mailing address</a></b></a |
|
* ~ |
|
* >. |
|
*/ |
|
needsToBorrowPrevClosingTagEndMarker(nextNode) && prevNode.lastChild && needsToBorrowParentClosingTagStartMarker(prevNode.lastChild) && prevNode.lastChild.lastChild && needsToBorrowParentClosingTagStartMarker(prevNode.lastChild.lastChild) ? hardline$7 : nextNode.hasLeadingSpaces ? line$5 : softline$4; |
|
} |
|
} |
|
|
|
function printOpeningTag(path, options, print) { |
|
var node = path.getValue(); |
|
var forceNotToBreakAttrContent = node.type === "element" && node.fullName === "script" && node.attrs.length === 1 && node.attrs[0].fullName === "src" && node.children.length === 0; |
|
return concat$a([printOpeningTagStart(node, options), !node.attrs || node.attrs.length === 0 ? node.isSelfClosing ? |
|
/** |
|
* <br /> |
|
* ^ |
|
*/ |
|
" " : "" : concat$a([indent$5(concat$a([forceNotToBreakAttrContent ? " " : line$5, join$6(line$5, function (ignoreAttributeData) { |
|
var hasPrettierIgnoreAttribute = typeof ignoreAttributeData === "boolean" ? function () { |
|
return ignoreAttributeData; |
|
} : Array.isArray(ignoreAttributeData) ? function (attr) { |
|
return ignoreAttributeData.indexOf(attr.rawName) !== -1; |
|
} : function () { |
|
return false; |
|
}; |
|
return path.map(function (attrPath) { |
|
var attr = attrPath.getValue(); |
|
return hasPrettierIgnoreAttribute(attr) ? concat$a(replaceEndOfLineWith$1(options.originalText.slice(options.locStart(attr), options.locEnd(attr)), literalline$2)) : print(attrPath); |
|
}, "attrs"); |
|
}(node.prev && node.prev.type === "comment" && getPrettierIgnoreAttributeCommentData$1(node.prev.value)))])), |
|
/** |
|
* 123<a |
|
* attr |
|
* ~ |
|
* >456 |
|
*/ |
|
node.firstChild && needsToBorrowParentOpeningTagEndMarker(node.firstChild) || |
|
/** |
|
* <span |
|
* >123<meta |
|
* ~ |
|
* /></span> |
|
*/ |
|
node.isSelfClosing && needsToBorrowLastChildClosingTagEndMarker(node.parent) ? "" : node.isSelfClosing ? forceNotToBreakAttrContent ? " " : line$5 : forceNotToBreakAttrContent ? "" : softline$4]), node.isSelfClosing ? "" : printOpeningTagEnd(node)]); |
|
} |
|
|
|
function printOpeningTagStart(node, options) { |
|
return node.prev && needsToBorrowNextOpeningTagStartMarker(node.prev) ? "" : concat$a([printOpeningTagPrefix(node, options), printOpeningTagStartMarker(node)]); |
|
} |
|
|
|
function printOpeningTagEnd(node) { |
|
return node.firstChild && needsToBorrowParentOpeningTagEndMarker(node.firstChild) ? "" : printOpeningTagEndMarker(node); |
|
} |
|
|
|
function printClosingTag(node, options) { |
|
return concat$a([node.isSelfClosing ? "" : printClosingTagStart(node, options), printClosingTagEnd(node, options)]); |
|
} |
|
|
|
function printClosingTagStart(node, options) { |
|
return node.lastChild && needsToBorrowParentClosingTagStartMarker(node.lastChild) ? "" : concat$a([printClosingTagPrefix(node, options), printClosingTagStartMarker(node, options)]); |
|
} |
|
|
|
function printClosingTagEnd(node, options) { |
|
return (node.next ? needsToBorrowPrevClosingTagEndMarker(node.next) : needsToBorrowLastChildClosingTagEndMarker(node.parent)) ? "" : concat$a([printClosingTagEndMarker(node, options), printClosingTagSuffix(node, options)]); |
|
} |
|
|
|
function needsToBorrowNextOpeningTagStartMarker(node) { |
|
/** |
|
* 123<p |
|
* ^^ |
|
* > |
|
*/ |
|
return node.next && !isTextLikeNode$1(node.next) && isTextLikeNode$1(node) && node.isTrailingSpaceSensitive && !node.hasTrailingSpaces; |
|
} |
|
|
|
function needsToBorrowParentOpeningTagEndMarker(node) { |
|
/** |
|
* <p |
|
* >123 |
|
* ^ |
|
* |
|
* <p |
|
* ><a |
|
* ^ |
|
*/ |
|
return !node.prev && node.isLeadingSpaceSensitive && !node.hasLeadingSpaces; |
|
} |
|
|
|
function needsToBorrowPrevClosingTagEndMarker(node) { |
|
/** |
|
* <p></p |
|
* >123 |
|
* ^ |
|
* |
|
* <p></p |
|
* ><a |
|
* ^ |
|
*/ |
|
return node.prev && !isTextLikeNode$1(node.prev) && node.isLeadingSpaceSensitive && !node.hasLeadingSpaces; |
|
} |
|
|
|
function needsToBorrowLastChildClosingTagEndMarker(node) { |
|
/** |
|
* <p |
|
* ><a></a |
|
* ></p |
|
* ^ |
|
* > |
|
*/ |
|
return node.lastChild && node.lastChild.isTrailingSpaceSensitive && !node.lastChild.hasTrailingSpaces && !isTextLikeNode$1(getLastDescendant$1(node.lastChild)); |
|
} |
|
|
|
function needsToBorrowParentClosingTagStartMarker(node) { |
|
/** |
|
* <p> |
|
* 123</p |
|
* ^^^ |
|
* > |
|
* |
|
* 123</b |
|
* ></a |
|
* ^^^ |
|
* > |
|
*/ |
|
return !node.next && !node.hasTrailingSpaces && node.isTrailingSpaceSensitive && isTextLikeNode$1(getLastDescendant$1(node)); |
|
} |
|
|
|
function printOpeningTagPrefix(node, options) { |
|
return needsToBorrowParentOpeningTagEndMarker(node) ? printOpeningTagEndMarker(node.parent) : needsToBorrowPrevClosingTagEndMarker(node) ? printClosingTagEndMarker(node.prev, options) : ""; |
|
} |
|
|
|
function printClosingTagPrefix(node, options) { |
|
return needsToBorrowLastChildClosingTagEndMarker(node) ? printClosingTagEndMarker(node.lastChild, options) : ""; |
|
} |
|
|
|
function printClosingTagSuffix(node, options) { |
|
return needsToBorrowParentClosingTagStartMarker(node) ? printClosingTagStartMarker(node.parent, options) : needsToBorrowNextOpeningTagStartMarker(node) ? printOpeningTagStartMarker(node.next) : ""; |
|
} |
|
|
|
function printOpeningTagStartMarker(node) { |
|
switch (node.type) { |
|
case "ieConditionalComment": |
|
case "ieConditionalStartComment": |
|
return "<!--[if ".concat(node.condition); |
|
|
|
case "ieConditionalEndComment": |
|
return "<!--<!"; |
|
|
|
case "interpolation": |
|
return "{{"; |
|
|
|
case "docType": |
|
return "<!DOCTYPE"; |
|
|
|
case "element": |
|
if (node.condition) { |
|
return "<!--[if ".concat(node.condition, "]><!--><").concat(node.rawName); |
|
} |
|
|
|
// fall through |
|
|
|
default: |
|
return "<".concat(node.rawName); |
|
} |
|
} |
|
|
|
function printOpeningTagEndMarker(node) { |
|
assert(!node.isSelfClosing); |
|
|
|
switch (node.type) { |
|
case "ieConditionalComment": |
|
return "]>"; |
|
|
|
case "element": |
|
if (node.condition) { |
|
return "><!--<![endif]-->"; |
|
} |
|
|
|
// fall through |
|
|
|
default: |
|
return ">"; |
|
} |
|
} |
|
|
|
function printClosingTagStartMarker(node, options) { |
|
assert(!node.isSelfClosing); |
|
|
|
if (shouldNotPrintClosingTag$1(node, options)) { |
|
return ""; |
|
} |
|
|
|
switch (node.type) { |
|
case "ieConditionalComment": |
|
return "<!"; |
|
|
|
case "element": |
|
if (node.hasHtmComponentClosingTag) { |
|
return "<//"; |
|
} |
|
|
|
// fall through |
|
|
|
default: |
|
return "</".concat(node.rawName); |
|
} |
|
} |
|
|
|
function printClosingTagEndMarker(node, options) { |
|
if (shouldNotPrintClosingTag$1(node, options)) { |
|
return ""; |
|
} |
|
|
|
switch (node.type) { |
|
case "ieConditionalComment": |
|
case "ieConditionalEndComment": |
|
return "[endif]-->"; |
|
|
|
case "ieConditionalStartComment": |
|
return "]><!-->"; |
|
|
|
case "interpolation": |
|
return "}}"; |
|
|
|
case "element": |
|
if (node.isSelfClosing) { |
|
return "/>"; |
|
} |
|
|
|
// fall through |
|
|
|
default: |
|
return ">"; |
|
} |
|
} |
|
|
|
function getTextValueParts(node) { |
|
var value = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : node.value; |
|
return node.parent.isWhitespaceSensitive ? node.parent.isIndentationSensitive ? replaceEndOfLineWith$1(value, literalline$2) : replaceEndOfLineWith$1(dedentString$1(value.replace(/^\s*?\n|\n\s*?$/g, "")), hardline$7) : // https://infra.spec.whatwg.org/#ascii-whitespace |
|
join$6(line$5, value.split(/[\t\n\f\r ]+/)).parts; |
|
} |
|
|
|
function printEmbeddedAttributeValue(node, originalTextToDoc, options) { |
|
var isKeyMatched = function isKeyMatched(patterns) { |
|
return new RegExp(patterns.join("|")).test(node.fullName); |
|
}; |
|
|
|
var getValue = function getValue() { |
|
return unescapeQuoteEntities$1(node.value); |
|
}; |
|
|
|
var shouldHug = false; |
|
|
|
var __onHtmlBindingRoot = function __onHtmlBindingRoot(root) { |
|
var rootNode = root.type === "NGRoot" ? root.node.type === "NGMicrosyntax" && root.node.body.length === 1 && root.node.body[0].type === "NGMicrosyntaxExpression" ? root.node.body[0].expression : root.node : root.type === "JsExpressionRoot" ? root.node : root; |
|
|
|
if (rootNode && (rootNode.type === "ObjectExpression" || rootNode.type === "ArrayExpression")) { |
|
shouldHug = true; |
|
} |
|
}; |
|
|
|
var printHug = function printHug(doc) { |
|
return group$9(doc); |
|
}; |
|
|
|
var printExpand = function printExpand(doc) { |
|
return group$9(concat$a([indent$5(concat$a([softline$4, doc])), softline$4])); |
|
}; |
|
|
|
var printMaybeHug = function printMaybeHug(doc) { |
|
return shouldHug ? printHug(doc) : printExpand(doc); |
|
}; |
|
|
|
var textToDoc = function textToDoc(code, opts) { |
|
return originalTextToDoc(code, Object.assign({ |
|
__onHtmlBindingRoot: __onHtmlBindingRoot |
|
}, opts)); |
|
}; |
|
|
|
if (node.fullName === "srcset" && (node.parent.fullName === "img" || node.parent.fullName === "source")) { |
|
return printExpand(printImgSrcset$1(getValue())); |
|
} |
|
|
|
if (options.parser === "vue") { |
|
if (node.fullName === "v-for") { |
|
return printVueFor$1(getValue(), textToDoc); |
|
} |
|
|
|
if (node.fullName === "slot-scope") { |
|
return printVueSlotScope$1(getValue(), textToDoc); |
|
} |
|
/** |
|
* @click="jsStatement" |
|
* @click="jsExpression" |
|
* v-on:click="jsStatement" |
|
* v-on:click="jsExpression" |
|
*/ |
|
|
|
|
|
var vueEventBindingPatterns = ["^@", "^v-on:"]; |
|
/** |
|
* :class="vueExpression" |
|
* v-bind:id="vueExpression" |
|
*/ |
|
|
|
var vueExpressionBindingPatterns = ["^:", "^v-bind:"]; |
|
/** |
|
* v-if="jsExpression" |
|
*/ |
|
|
|
var jsExpressionBindingPatterns = ["^v-"]; |
|
|
|
if (isKeyMatched(vueEventBindingPatterns)) { |
|
var value = getValue(); |
|
return printMaybeHug(isVueEventBindingExpression$1(value) ? textToDoc(value, { |
|
parser: "__js_expression" |
|
}) : stripTrailingHardline$1(textToDoc(value, { |
|
parser: "__vue_event_binding" |
|
}))); |
|
} |
|
|
|
if (isKeyMatched(vueExpressionBindingPatterns)) { |
|
return printMaybeHug(textToDoc(getValue(), { |
|
parser: "__vue_expression" |
|
})); |
|
} |
|
|
|
if (isKeyMatched(jsExpressionBindingPatterns)) { |
|
return printMaybeHug(textToDoc(getValue(), { |
|
parser: "__js_expression" |
|
})); |
|
} |
|
} |
|
|
|
if (options.parser === "angular") { |
|
var ngTextToDoc = function ngTextToDoc(code, opts) { |
|
return (// angular does not allow trailing comma |
|
textToDoc(code, Object.assign({ |
|
trailingComma: "none" |
|
}, opts)) |
|
); |
|
}; |
|
/** |
|
* *directive="angularDirective" |
|
*/ |
|
|
|
|
|
var ngDirectiveBindingPatterns = ["^\\*"]; |
|
/** |
|
* (click)="angularStatement" |
|
* on-click="angularStatement" |
|
*/ |
|
|
|
var ngStatementBindingPatterns = ["^\\(.+\\)$", "^on-"]; |
|
/** |
|
* [target]="angularExpression" |
|
* bind-target="angularExpression" |
|
* [(target)]="angularExpression" |
|
* bindon-target="angularExpression" |
|
*/ |
|
|
|
var ngExpressionBindingPatterns = ["^\\[.+\\]$", "^bind(on)?-"]; |
|
/** |
|
* i18n="longDescription" |
|
* i18n-attr="longDescription" |
|
*/ |
|
|
|
var ngI18nPatterns = ["^i18n(-.+)?$"]; |
|
|
|
if (isKeyMatched(ngStatementBindingPatterns)) { |
|
return printMaybeHug(ngTextToDoc(getValue(), { |
|
parser: "__ng_action" |
|
})); |
|
} |
|
|
|
if (isKeyMatched(ngExpressionBindingPatterns)) { |
|
return printMaybeHug(ngTextToDoc(getValue(), { |
|
parser: "__ng_binding" |
|
})); |
|
} |
|
|
|
if (isKeyMatched(ngI18nPatterns)) { |
|
return printExpand(fill$3(getTextValueParts(node, getValue()))); |
|
} |
|
|
|
if (isKeyMatched(ngDirectiveBindingPatterns)) { |
|
return printMaybeHug(ngTextToDoc(getValue(), { |
|
parser: "__ng_directive" |
|
})); |
|
} |
|
|
|
var interpolationRegex = /\{\{([\s\S]+?)\}\}/g; |
|
|
|
var _value2 = getValue(); |
|
|
|
if (interpolationRegex.test(_value2)) { |
|
var parts = []; |
|
|
|
_value2.split(interpolationRegex).forEach(function (part, index) { |
|
if (index % 2 === 0) { |
|
parts.push(concat$a(replaceEndOfLineWith$1(part, literalline$2))); |
|
} else { |
|
try { |
|
parts.push(group$9(concat$a(["{{", indent$5(concat$a([line$5, ngTextToDoc(part, { |
|
parser: "__ng_interpolation", |
|
__isInHtmlInterpolation: true // to avoid unexpected `}}` |
|
|
|
})])), line$5, "}}"]))); |
|
} catch (e) { |
|
parts.push("{{", concat$a(replaceEndOfLineWith$1(part, literalline$2)), "}}"); |
|
} |
|
} |
|
}); |
|
|
|
return group$9(concat$a(parts)); |
|
} |
|
} |
|
|
|
return null; |
|
} |
|
|
|
var printerHtml = { |
|
preprocess: preprocess_1, |
|
print: genericPrint$2, |
|
insertPragma: insertPragma$6, |
|
massageAstNode: clean$3, |
|
embed: embed$2 |
|
}; |
|
|
|
var CATEGORY_HTML = "HTML"; // format based on https://github.com/prettier/prettier/blob/master/src/main/core-options.js |
|
|
|
var options$4 = { |
|
htmlWhitespaceSensitivity: { |
|
since: "1.15.0", |
|
category: CATEGORY_HTML, |
|
type: "choice", |
|
default: "css", |
|
description: "How to handle whitespaces in HTML.", |
|
choices: [{ |
|
value: "css", |
|
description: "Respect the default value of CSS display property." |
|
}, { |
|
value: "strict", |
|
description: "Whitespaces are considered sensitive." |
|
}, { |
|
value: "ignore", |
|
description: "Whitespaces are considered insensitive." |
|
}] |
|
}, |
|
vueIndentScriptAndStyle: { |
|
since: "1.19.0", |
|
category: CATEGORY_HTML, |
|
type: "boolean", |
|
default: false, |
|
description: "Indent script and style tags in Vue files." |
|
} |
|
}; |
|
|
|
var name$7 = "HTML"; |
|
var type$6 = "markup"; |
|
var tmScope$6 = "text.html.basic"; |
|
var aceMode$6 = "html"; |
|
var codemirrorMode$3 = "htmlmixed"; |
|
var codemirrorMimeType$3 = "text/html"; |
|
var color$1 = "#e34c26"; |
|
var aliases$1 = [ |
|
"xhtml" |
|
]; |
|
var extensions$6 = [ |
|
".html", |
|
".htm", |
|
".html.hl", |
|
".inc", |
|
".st", |
|
".xht", |
|
".xhtml" |
|
]; |
|
var languageId$6 = 146; |
|
var HTML = { |
|
name: name$7, |
|
type: type$6, |
|
tmScope: tmScope$6, |
|
aceMode: aceMode$6, |
|
codemirrorMode: codemirrorMode$3, |
|
codemirrorMimeType: codemirrorMimeType$3, |
|
color: color$1, |
|
aliases: aliases$1, |
|
extensions: extensions$6, |
|
languageId: languageId$6 |
|
}; |
|
|
|
var HTML$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$7, |
|
type: type$6, |
|
tmScope: tmScope$6, |
|
aceMode: aceMode$6, |
|
codemirrorMode: codemirrorMode$3, |
|
codemirrorMimeType: codemirrorMimeType$3, |
|
color: color$1, |
|
aliases: aliases$1, |
|
extensions: extensions$6, |
|
languageId: languageId$6, |
|
'default': HTML |
|
}); |
|
|
|
var name$8 = "Vue"; |
|
var type$7 = "markup"; |
|
var color$2 = "#2c3e50"; |
|
var extensions$7 = [ |
|
".vue" |
|
]; |
|
var tmScope$7 = "text.html.vue"; |
|
var aceMode$7 = "html"; |
|
var languageId$7 = 391; |
|
var Vue = { |
|
name: name$8, |
|
type: type$7, |
|
color: color$2, |
|
extensions: extensions$7, |
|
tmScope: tmScope$7, |
|
aceMode: aceMode$7, |
|
languageId: languageId$7 |
|
}; |
|
|
|
var Vue$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$8, |
|
type: type$7, |
|
color: color$2, |
|
extensions: extensions$7, |
|
tmScope: tmScope$7, |
|
aceMode: aceMode$7, |
|
languageId: languageId$7, |
|
'default': Vue |
|
}); |
|
|
|
var require$$0$5 = getCjsExportFromNamespace(HTML$1); |
|
|
|
var require$$1$1 = getCjsExportFromNamespace(Vue$1); |
|
|
|
var languages$3 = [createLanguage(require$$0$5, function (data) { |
|
return Object.assign(data, { |
|
name: "Angular", |
|
since: "1.15.0", |
|
parsers: ["angular"], |
|
vscodeLanguageIds: ["html"], |
|
extensions: [".component.html"], |
|
filenames: [] |
|
}); |
|
}), createLanguage(require$$0$5, function (data) { |
|
return Object.assign(data, { |
|
since: "1.15.0", |
|
parsers: ["html"], |
|
vscodeLanguageIds: ["html"], |
|
extensions: data.extensions.concat([".mjml" // MJML is considered XML in Linguist but it should be formatted as HTML |
|
]) |
|
}); |
|
}), createLanguage(require$$0$5, function (data) { |
|
return Object.assign(data, { |
|
name: "Lightning Web Components", |
|
since: "1.17.0", |
|
parsers: ["lwc"], |
|
vscodeLanguageIds: ["html"], |
|
extensions: [], |
|
filenames: [] |
|
}); |
|
}), createLanguage(require$$1$1, function (data) { |
|
return Object.assign(data, { |
|
since: "1.10.0", |
|
parsers: ["vue"], |
|
vscodeLanguageIds: ["vue"] |
|
}); |
|
})]; |
|
var printers$3 = { |
|
html: printerHtml |
|
}; |
|
var languageHtml = { |
|
languages: languages$3, |
|
printers: printers$3, |
|
options: options$4 |
|
}; |
|
|
|
var addLeadingComment$2 = utilShared.addLeadingComment, |
|
addTrailingComment$2 = utilShared.addTrailingComment, |
|
addDanglingComment$2 = utilShared.addDanglingComment; |
|
|
|
function handleOwnLineComment(comment, text, options, ast, isLastComment) { |
|
var precedingNode = comment.precedingNode, |
|
enclosingNode = comment.enclosingNode, |
|
followingNode = comment.followingNode; |
|
|
|
if (handleLastFunctionArgComments(text, precedingNode, enclosingNode, followingNode, comment, options) || handleMemberExpressionComments(enclosingNode, followingNode, comment) || handleIfStatementComments(text, precedingNode, enclosingNode, followingNode, comment, options) || handleWhileComments(text, precedingNode, enclosingNode, followingNode, comment, options) || handleTryStatementComments(enclosingNode, precedingNode, followingNode, comment) || handleClassComments(enclosingNode, precedingNode, followingNode, comment) || handleImportSpecifierComments(enclosingNode, comment) || handleForComments(enclosingNode, precedingNode, comment) || handleUnionTypeComments(precedingNode, enclosingNode, followingNode, comment) || handleOnlyComments(enclosingNode, ast, comment, isLastComment) || handleImportDeclarationComments(text, enclosingNode, precedingNode, comment, options) || handleAssignmentPatternComments(enclosingNode, comment) || handleMethodNameComments(text, enclosingNode, precedingNode, comment, options)) { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleEndOfLineComment(comment, text, options, ast, isLastComment) { |
|
var precedingNode = comment.precedingNode, |
|
enclosingNode = comment.enclosingNode, |
|
followingNode = comment.followingNode; |
|
|
|
if (handleLastFunctionArgComments(text, precedingNode, enclosingNode, followingNode, comment, options) || handleConditionalExpressionComments(enclosingNode, precedingNode, followingNode, comment, text, options) || handleImportSpecifierComments(enclosingNode, comment) || handleIfStatementComments(text, precedingNode, enclosingNode, followingNode, comment, options) || handleWhileComments(text, precedingNode, enclosingNode, followingNode, comment, options) || handleTryStatementComments(enclosingNode, precedingNode, followingNode, comment) || handleClassComments(enclosingNode, precedingNode, followingNode, comment) || handleLabeledStatementComments(enclosingNode, comment) || handleCallExpressionComments(precedingNode, enclosingNode, comment) || handlePropertyComments(enclosingNode, comment) || handleOnlyComments(enclosingNode, ast, comment, isLastComment) || handleTypeAliasComments(enclosingNode, followingNode, comment) || handleVariableDeclaratorComments(enclosingNode, followingNode, comment)) { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleRemainingComment(comment, text, options, ast, isLastComment) { |
|
var precedingNode = comment.precedingNode, |
|
enclosingNode = comment.enclosingNode, |
|
followingNode = comment.followingNode; |
|
|
|
if (handleIfStatementComments(text, precedingNode, enclosingNode, followingNode, comment, options) || handleWhileComments(text, precedingNode, enclosingNode, followingNode, comment, options) || handleObjectPropertyAssignment(enclosingNode, precedingNode, comment) || handleCommentInEmptyParens(text, enclosingNode, comment, options) || handleMethodNameComments(text, enclosingNode, precedingNode, comment, options) || handleOnlyComments(enclosingNode, ast, comment, isLastComment) || handleCommentAfterArrowParams(text, enclosingNode, comment, options) || handleFunctionNameComments(text, enclosingNode, precedingNode, comment, options) || handleTSMappedTypeComments(text, enclosingNode, precedingNode, followingNode, comment) || handleBreakAndContinueStatementComments(enclosingNode, comment)) { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function addBlockStatementFirstComment(node, comment) { |
|
var body = node.body.filter(function (n) { |
|
return n.type !== "EmptyStatement"; |
|
}); |
|
|
|
if (body.length === 0) { |
|
addDanglingComment$2(node, comment); |
|
} else { |
|
addLeadingComment$2(body[0], comment); |
|
} |
|
} |
|
|
|
function addBlockOrNotComment(node, comment) { |
|
if (node.type === "BlockStatement") { |
|
addBlockStatementFirstComment(node, comment); |
|
} else { |
|
addLeadingComment$2(node, comment); |
|
} |
|
} // There are often comments before the else clause of if statements like |
|
// |
|
// if (1) { ... } |
|
// // comment |
|
// else { ... } |
|
// |
|
// They are being attached as leading comments of the BlockExpression which |
|
// is not well printed. What we want is to instead move the comment inside |
|
// of the block and make it leadingComment of the first element of the block |
|
// or dangling comment of the block if there is nothing inside |
|
// |
|
// if (1) { ... } |
|
// else { |
|
// // comment |
|
// ... |
|
// } |
|
|
|
|
|
function handleIfStatementComments(text, precedingNode, enclosingNode, followingNode, comment, options) { |
|
if (!enclosingNode || enclosingNode.type !== "IfStatement" || !followingNode) { |
|
return false; |
|
} // We unfortunately have no way using the AST or location of nodes to know |
|
// if the comment is positioned before the condition parenthesis: |
|
// if (a /* comment */) {} |
|
// The only workaround I found is to look at the next character to see if |
|
// it is a ). |
|
|
|
|
|
var nextCharacter = util.getNextNonSpaceNonCommentCharacter(text, comment, options.locEnd); |
|
|
|
if (nextCharacter === ")") { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} // Comments before `else`: |
|
// - treat as trailing comments of the consequent, if it's a BlockStatement |
|
// - treat as a dangling comment otherwise |
|
|
|
|
|
if (precedingNode === enclosingNode.consequent && followingNode === enclosingNode.alternate) { |
|
if (precedingNode.type === "BlockStatement") { |
|
addTrailingComment$2(precedingNode, comment); |
|
} else { |
|
addDanglingComment$2(enclosingNode, comment); |
|
} |
|
|
|
return true; |
|
} |
|
|
|
if (followingNode.type === "BlockStatement") { |
|
addBlockStatementFirstComment(followingNode, comment); |
|
return true; |
|
} |
|
|
|
if (followingNode.type === "IfStatement") { |
|
addBlockOrNotComment(followingNode.consequent, comment); |
|
return true; |
|
} // For comments positioned after the condition parenthesis in an if statement |
|
// before the consequent without brackets on, such as |
|
// if (a) /* comment */ true, |
|
// we look at the next character to see if the following node |
|
// is the consequent for the if statement |
|
|
|
|
|
if (enclosingNode.consequent === followingNode) { |
|
addLeadingComment$2(followingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleWhileComments(text, precedingNode, enclosingNode, followingNode, comment, options) { |
|
if (!enclosingNode || enclosingNode.type !== "WhileStatement" || !followingNode) { |
|
return false; |
|
} // We unfortunately have no way using the AST or location of nodes to know |
|
// if the comment is positioned before the condition parenthesis: |
|
// while (a /* comment */) {} |
|
// The only workaround I found is to look at the next character to see if |
|
// it is a ). |
|
|
|
|
|
var nextCharacter = util.getNextNonSpaceNonCommentCharacter(text, comment, options.locEnd); |
|
|
|
if (nextCharacter === ")") { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} |
|
|
|
if (followingNode.type === "BlockStatement") { |
|
addBlockStatementFirstComment(followingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} // Same as IfStatement but for TryStatement |
|
|
|
|
|
function handleTryStatementComments(enclosingNode, precedingNode, followingNode, comment) { |
|
if (!enclosingNode || enclosingNode.type !== "TryStatement" && enclosingNode.type !== "CatchClause" || !followingNode) { |
|
return false; |
|
} |
|
|
|
if (enclosingNode.type === "CatchClause" && precedingNode) { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} |
|
|
|
if (followingNode.type === "BlockStatement") { |
|
addBlockStatementFirstComment(followingNode, comment); |
|
return true; |
|
} |
|
|
|
if (followingNode.type === "TryStatement") { |
|
addBlockOrNotComment(followingNode.finalizer, comment); |
|
return true; |
|
} |
|
|
|
if (followingNode.type === "CatchClause") { |
|
addBlockOrNotComment(followingNode.body, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleMemberExpressionComments(enclosingNode, followingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "MemberExpression" || enclosingNode.type === "OptionalMemberExpression") && followingNode && followingNode.type === "Identifier") { |
|
addLeadingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleConditionalExpressionComments(enclosingNode, precedingNode, followingNode, comment, text, options) { |
|
var isSameLineAsPrecedingNode = precedingNode && !util.hasNewlineInRange(text, options.locEnd(precedingNode), options.locStart(comment)); |
|
|
|
if ((!precedingNode || !isSameLineAsPrecedingNode) && enclosingNode && enclosingNode.type === "ConditionalExpression" && followingNode) { |
|
addLeadingComment$2(followingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleObjectPropertyAssignment(enclosingNode, precedingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "ObjectProperty" || enclosingNode.type === "Property") && enclosingNode.shorthand && enclosingNode.key === precedingNode && enclosingNode.value.type === "AssignmentPattern") { |
|
addTrailingComment$2(enclosingNode.value.left, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleClassComments(enclosingNode, precedingNode, followingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "ClassDeclaration" || enclosingNode.type === "ClassExpression") && enclosingNode.decorators && enclosingNode.decorators.length > 0 && !(followingNode && followingNode.type === "Decorator")) { |
|
if (!enclosingNode.decorators || enclosingNode.decorators.length === 0) { |
|
addLeadingComment$2(enclosingNode, comment); |
|
} else { |
|
addTrailingComment$2(enclosingNode.decorators[enclosingNode.decorators.length - 1], comment); |
|
} |
|
|
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleMethodNameComments(text, enclosingNode, precedingNode, comment, options) { |
|
// This is only needed for estree parsers (flow, typescript) to attach |
|
// after a method name: |
|
// obj = { fn /*comment*/() {} }; |
|
if (enclosingNode && precedingNode && (enclosingNode.type === "Property" || enclosingNode.type === "MethodDefinition") && precedingNode.type === "Identifier" && enclosingNode.key === precedingNode && // special Property case: { key: /*comment*/(value) }; |
|
// comment should be attached to value instead of key |
|
util.getNextNonSpaceNonCommentCharacter(text, precedingNode, options.locEnd) !== ":") { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} // Print comments between decorators and class methods as a trailing comment |
|
// on the decorator node instead of the method node |
|
|
|
|
|
if (precedingNode && enclosingNode && precedingNode.type === "Decorator" && (enclosingNode.type === "ClassMethod" || enclosingNode.type === "ClassProperty" || enclosingNode.type === "TSAbstractClassProperty" || enclosingNode.type === "TSAbstractMethodDefinition" || enclosingNode.type === "MethodDefinition")) { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleFunctionNameComments(text, enclosingNode, precedingNode, comment, options) { |
|
if (util.getNextNonSpaceNonCommentCharacter(text, comment, options.locEnd) !== "(") { |
|
return false; |
|
} |
|
|
|
if (precedingNode && enclosingNode && (enclosingNode.type === "FunctionDeclaration" || enclosingNode.type === "FunctionExpression" || enclosingNode.type === "ClassMethod" || enclosingNode.type === "MethodDefinition" || enclosingNode.type === "ObjectMethod")) { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleCommentAfterArrowParams(text, enclosingNode, comment, options) { |
|
if (!(enclosingNode && enclosingNode.type === "ArrowFunctionExpression")) { |
|
return false; |
|
} |
|
|
|
var index = utilShared.getNextNonSpaceNonCommentCharacterIndex(text, comment, options); |
|
|
|
if (text.substr(index, 2) === "=>") { |
|
addDanglingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleCommentInEmptyParens(text, enclosingNode, comment, options) { |
|
if (util.getNextNonSpaceNonCommentCharacter(text, comment, options.locEnd) !== ")") { |
|
return false; |
|
} // Only add dangling comments to fix the case when no params are present, |
|
// i.e. a function without any argument. |
|
|
|
|
|
if (enclosingNode && ((enclosingNode.type === "FunctionDeclaration" || enclosingNode.type === "FunctionExpression" || enclosingNode.type === "ArrowFunctionExpression" || enclosingNode.type === "ClassMethod" || enclosingNode.type === "ObjectMethod") && enclosingNode.params.length === 0 || (enclosingNode.type === "CallExpression" || enclosingNode.type === "OptionalCallExpression" || enclosingNode.type === "NewExpression") && enclosingNode.arguments.length === 0)) { |
|
addDanglingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
if (enclosingNode && enclosingNode.type === "MethodDefinition" && enclosingNode.value.params.length === 0) { |
|
addDanglingComment$2(enclosingNode.value, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleLastFunctionArgComments(text, precedingNode, enclosingNode, followingNode, comment, options) { |
|
// Type definitions functions |
|
if (precedingNode && precedingNode.type === "FunctionTypeParam" && enclosingNode && enclosingNode.type === "FunctionTypeAnnotation" && followingNode && followingNode.type !== "FunctionTypeParam") { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} // Real functions |
|
|
|
|
|
if (precedingNode && (precedingNode.type === "Identifier" || precedingNode.type === "AssignmentPattern") && enclosingNode && (enclosingNode.type === "ArrowFunctionExpression" || enclosingNode.type === "FunctionExpression" || enclosingNode.type === "FunctionDeclaration" || enclosingNode.type === "ObjectMethod" || enclosingNode.type === "ClassMethod") && util.getNextNonSpaceNonCommentCharacter(text, comment, options.locEnd) === ")") { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} |
|
|
|
if (enclosingNode && enclosingNode.type === "FunctionDeclaration" && followingNode && followingNode.type === "BlockStatement") { |
|
var functionParamRightParenIndex = function () { |
|
if (enclosingNode.params.length !== 0) { |
|
return util.getNextNonSpaceNonCommentCharacterIndexWithStartIndex(text, options.locEnd(util.getLast(enclosingNode.params))); |
|
} |
|
|
|
var functionParamLeftParenIndex = util.getNextNonSpaceNonCommentCharacterIndexWithStartIndex(text, options.locEnd(enclosingNode.id)); |
|
return util.getNextNonSpaceNonCommentCharacterIndexWithStartIndex(text, functionParamLeftParenIndex + 1); |
|
}(); |
|
|
|
if (options.locStart(comment) > functionParamRightParenIndex) { |
|
addBlockStatementFirstComment(followingNode, comment); |
|
return true; |
|
} |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleImportSpecifierComments(enclosingNode, comment) { |
|
if (enclosingNode && enclosingNode.type === "ImportSpecifier") { |
|
addLeadingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleLabeledStatementComments(enclosingNode, comment) { |
|
if (enclosingNode && enclosingNode.type === "LabeledStatement") { |
|
addLeadingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleBreakAndContinueStatementComments(enclosingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "ContinueStatement" || enclosingNode.type === "BreakStatement") && !enclosingNode.label) { |
|
addTrailingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleCallExpressionComments(precedingNode, enclosingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "CallExpression" || enclosingNode.type === "OptionalCallExpression") && precedingNode && enclosingNode.callee === precedingNode && enclosingNode.arguments.length > 0) { |
|
addLeadingComment$2(enclosingNode.arguments[0], comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleUnionTypeComments(precedingNode, enclosingNode, followingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "UnionTypeAnnotation" || enclosingNode.type === "TSUnionType")) { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handlePropertyComments(enclosingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "Property" || enclosingNode.type === "ObjectProperty")) { |
|
addLeadingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleOnlyComments(enclosingNode, ast, comment, isLastComment) { |
|
// With Flow the enclosingNode is undefined so use the AST instead. |
|
if (ast && ast.body && ast.body.length === 0) { |
|
if (isLastComment) { |
|
addDanglingComment$2(ast, comment); |
|
} else { |
|
addLeadingComment$2(ast, comment); |
|
} |
|
|
|
return true; |
|
} else if (enclosingNode && enclosingNode.type === "Program" && enclosingNode.body.length === 0 && enclosingNode.directives && enclosingNode.directives.length === 0) { |
|
if (isLastComment) { |
|
addDanglingComment$2(enclosingNode, comment); |
|
} else { |
|
addLeadingComment$2(enclosingNode, comment); |
|
} |
|
|
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleForComments(enclosingNode, precedingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "ForInStatement" || enclosingNode.type === "ForOfStatement")) { |
|
addLeadingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleImportDeclarationComments(text, enclosingNode, precedingNode, comment, options) { |
|
if (precedingNode && precedingNode.type === "ImportSpecifier" && enclosingNode && enclosingNode.type === "ImportDeclaration" && util.hasNewline(text, options.locEnd(comment))) { |
|
addTrailingComment$2(precedingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleAssignmentPatternComments(enclosingNode, comment) { |
|
if (enclosingNode && enclosingNode.type === "AssignmentPattern") { |
|
addLeadingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleTypeAliasComments(enclosingNode, followingNode, comment) { |
|
if (enclosingNode && enclosingNode.type === "TypeAlias") { |
|
addLeadingComment$2(enclosingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleVariableDeclaratorComments(enclosingNode, followingNode, comment) { |
|
if (enclosingNode && (enclosingNode.type === "VariableDeclarator" || enclosingNode.type === "AssignmentExpression") && followingNode && (followingNode.type === "ObjectExpression" || followingNode.type === "ArrayExpression" || followingNode.type === "TemplateLiteral" || followingNode.type === "TaggedTemplateExpression")) { |
|
addLeadingComment$2(followingNode, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function handleTSMappedTypeComments(text, enclosingNode, precedingNode, followingNode, comment) { |
|
if (!enclosingNode || enclosingNode.type !== "TSMappedType") { |
|
return false; |
|
} |
|
|
|
if (followingNode && followingNode.type === "TSTypeParameter" && followingNode.name) { |
|
addLeadingComment$2(followingNode.name, comment); |
|
return true; |
|
} |
|
|
|
if (precedingNode && precedingNode.type === "TSTypeParameter" && precedingNode.constraint) { |
|
addTrailingComment$2(precedingNode.constraint, comment); |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isBlockComment(comment) { |
|
return comment.type === "Block" || comment.type === "CommentBlock"; |
|
} |
|
|
|
function hasLeadingComment(node) { |
|
var fn = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : function () { |
|
return true; |
|
}; |
|
|
|
if (node.leadingComments) { |
|
return node.leadingComments.some(fn); |
|
} |
|
|
|
if (node.comments) { |
|
return node.comments.some(function (comment) { |
|
return comment.leading && fn(comment); |
|
}); |
|
} |
|
|
|
return false; |
|
} |
|
|
|
var comments$1 = { |
|
handleOwnLineComment: handleOwnLineComment, |
|
handleEndOfLineComment: handleEndOfLineComment, |
|
handleRemainingComment: handleRemainingComment, |
|
hasLeadingComment: hasLeadingComment, |
|
isBlockComment: isBlockComment |
|
}; |
|
|
|
var isBlockComment$1 = comments$1.isBlockComment, |
|
hasLeadingComment$1 = comments$1.hasLeadingComment; |
|
var _require$$1$builders = doc.builders, |
|
indent$6 = _require$$1$builders.indent, |
|
join$7 = _require$$1$builders.join, |
|
hardline$8 = _require$$1$builders.hardline, |
|
softline$5 = _require$$1$builders.softline, |
|
literalline$3 = _require$$1$builders.literalline, |
|
concat$b = _require$$1$builders.concat, |
|
group$a = _require$$1$builders.group, |
|
dedentToRoot$2 = _require$$1$builders.dedentToRoot, |
|
_require$$1$utils = doc.utils, |
|
mapDoc$5 = _require$$1$utils.mapDoc, |
|
stripTrailingHardline$2 = _require$$1$utils.stripTrailingHardline; |
|
|
|
function embed$3(path, print, textToDoc, options) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
var parentParent = path.getParentNode(1); |
|
|
|
switch (node.type) { |
|
case "TemplateLiteral": |
|
{ |
|
var isCss = [isStyledJsx, isStyledComponents, isCssProp, isAngularComponentStyles].some(function (isIt) { |
|
return isIt(path); |
|
}); |
|
|
|
if (isCss) { |
|
// Get full template literal with expressions replaced by placeholders |
|
var rawQuasis = node.quasis.map(function (q) { |
|
return q.value.raw; |
|
}); |
|
var placeholderID = 0; |
|
var text = rawQuasis.reduce(function (prevVal, currVal, idx) { |
|
return idx == 0 ? currVal : prevVal + "@prettier-placeholder-" + placeholderID++ + "-id" + currVal; |
|
}, ""); |
|
var doc = textToDoc(text, { |
|
parser: "css" |
|
}); |
|
return transformCssDoc(doc, path, print); |
|
} |
|
/* |
|
* react-relay and graphql-tag |
|
* graphql`...` |
|
* graphql.experimental`...` |
|
* gql`...` |
|
* |
|
* This intentionally excludes Relay Classic tags, as Prettier does not |
|
* support Relay Classic formatting. |
|
*/ |
|
|
|
|
|
if (isGraphQL(path)) { |
|
var expressionDocs = node.expressions ? path.map(print, "expressions") : []; |
|
var numQuasis = node.quasis.length; |
|
|
|
if (numQuasis === 1 && node.quasis[0].value.raw.trim() === "") { |
|
return "``"; |
|
} |
|
|
|
var parts = []; |
|
|
|
for (var i = 0; i < numQuasis; i++) { |
|
var templateElement = node.quasis[i]; |
|
var isFirst = i === 0; |
|
var isLast = i === numQuasis - 1; |
|
var _text = templateElement.value.cooked; // Bail out if any of the quasis have an invalid escape sequence |
|
// (which would make the `cooked` value be `null` or `undefined`) |
|
|
|
if (typeof _text !== "string") { |
|
return null; |
|
} |
|
|
|
var lines = _text.split("\n"); |
|
|
|
var numLines = lines.length; |
|
var expressionDoc = expressionDocs[i]; |
|
var startsWithBlankLine = numLines > 2 && lines[0].trim() === "" && lines[1].trim() === ""; |
|
var endsWithBlankLine = numLines > 2 && lines[numLines - 1].trim() === "" && lines[numLines - 2].trim() === ""; |
|
var commentsAndWhitespaceOnly = lines.every(function (line) { |
|
return /^\s*(?:#[^\r\n]*)?$/.test(line); |
|
}); // Bail out if an interpolation occurs within a comment. |
|
|
|
if (!isLast && /#[^\r\n]*$/.test(lines[numLines - 1])) { |
|
return null; |
|
} |
|
|
|
var _doc = null; |
|
|
|
if (commentsAndWhitespaceOnly) { |
|
_doc = printGraphqlComments(lines); |
|
} else { |
|
_doc = stripTrailingHardline$2(textToDoc(_text, { |
|
parser: "graphql" |
|
})); |
|
} |
|
|
|
if (_doc) { |
|
_doc = escapeTemplateCharacters(_doc, false); |
|
|
|
if (!isFirst && startsWithBlankLine) { |
|
parts.push(""); |
|
} |
|
|
|
parts.push(_doc); |
|
|
|
if (!isLast && endsWithBlankLine) { |
|
parts.push(""); |
|
} |
|
} else if (!isFirst && !isLast && startsWithBlankLine) { |
|
parts.push(""); |
|
} |
|
|
|
if (expressionDoc) { |
|
parts.push(concat$b(["${", expressionDoc, "}"])); |
|
} |
|
} |
|
|
|
return concat$b(["`", indent$6(concat$b([hardline$8, join$7(hardline$8, parts)])), hardline$8, "`"]); |
|
} |
|
|
|
var htmlParser = isHtml(path) ? "html" : isAngularComponentTemplate(path) ? "angular" : undefined; |
|
|
|
if (htmlParser) { |
|
return printHtmlTemplateLiteral(path, print, textToDoc, htmlParser, options.embeddedInHtml); |
|
} |
|
|
|
break; |
|
} |
|
|
|
case "TemplateElement": |
|
{ |
|
/** |
|
* md`...` |
|
* markdown`...` |
|
*/ |
|
if (parentParent && parentParent.type === "TaggedTemplateExpression" && parent.quasis.length === 1 && parentParent.tag.type === "Identifier" && (parentParent.tag.name === "md" || parentParent.tag.name === "markdown")) { |
|
var _text2 = parent.quasis[0].value.raw.replace(/((?:\\\\)*)\\`/g, function (_, backslashes) { |
|
return "\\".repeat(backslashes.length / 2) + "`"; |
|
}); |
|
|
|
var indentation = getIndentation(_text2); |
|
var hasIndent = indentation !== ""; |
|
return concat$b([hasIndent ? indent$6(concat$b([softline$5, printMarkdown(_text2.replace(new RegExp("^".concat(indentation), "gm"), ""))])) : concat$b([literalline$3, dedentToRoot$2(printMarkdown(_text2))]), softline$5]); |
|
} |
|
|
|
break; |
|
} |
|
} |
|
|
|
function printMarkdown(text) { |
|
var doc = textToDoc(text, { |
|
parser: "markdown", |
|
__inJsTemplate: true |
|
}); |
|
return stripTrailingHardline$2(escapeTemplateCharacters(doc, true)); |
|
} |
|
} |
|
|
|
function getIndentation(str) { |
|
var firstMatchedIndent = str.match(/^([^\S\n]*)\S/m); |
|
return firstMatchedIndent === null ? "" : firstMatchedIndent[1]; |
|
} |
|
|
|
function uncook(cookedValue) { |
|
return cookedValue.replace(/([\\`]|\$\{)/g, "\\$1"); |
|
} |
|
|
|
function escapeTemplateCharacters(doc, raw) { |
|
return mapDoc$5(doc, function (currentDoc) { |
|
if (!currentDoc.parts) { |
|
return currentDoc; |
|
} |
|
|
|
var parts = []; |
|
currentDoc.parts.forEach(function (part) { |
|
if (typeof part === "string") { |
|
parts.push(raw ? part.replace(/(\\*)`/g, "$1$1\\`") : uncook(part)); |
|
} else { |
|
parts.push(part); |
|
} |
|
}); |
|
return Object.assign({}, currentDoc, { |
|
parts: parts |
|
}); |
|
}); |
|
} |
|
|
|
function transformCssDoc(quasisDoc, path, print) { |
|
var parentNode = path.getValue(); |
|
var isEmpty = parentNode.quasis.length === 1 && !parentNode.quasis[0].value.raw.trim(); |
|
|
|
if (isEmpty) { |
|
return "``"; |
|
} |
|
|
|
var expressionDocs = parentNode.expressions ? path.map(print, "expressions") : []; |
|
var newDoc = replacePlaceholders(quasisDoc, expressionDocs); |
|
/* istanbul ignore if */ |
|
|
|
if (!newDoc) { |
|
throw new Error("Couldn't insert all the expressions"); |
|
} |
|
|
|
return concat$b(["`", indent$6(concat$b([hardline$8, stripTrailingHardline$2(newDoc)])), softline$5, "`"]); |
|
} // Search all the placeholders in the quasisDoc tree |
|
// and replace them with the expression docs one by one |
|
// returns a new doc with all the placeholders replaced, |
|
// or null if it couldn't replace any expression |
|
|
|
|
|
function replacePlaceholders(quasisDoc, expressionDocs) { |
|
if (!expressionDocs || !expressionDocs.length) { |
|
return quasisDoc; |
|
} |
|
|
|
var expressions = expressionDocs.slice(); |
|
var replaceCounter = 0; |
|
var newDoc = mapDoc$5(quasisDoc, function (doc) { |
|
if (!doc || !doc.parts || !doc.parts.length) { |
|
return doc; |
|
} |
|
|
|
var parts = doc.parts; |
|
var atIndex = parts.indexOf("@"); |
|
var placeholderIndex = atIndex + 1; |
|
|
|
if (atIndex > -1 && typeof parts[placeholderIndex] === "string" && parts[placeholderIndex].startsWith("prettier-placeholder")) { |
|
// If placeholder is split, join it |
|
var at = parts[atIndex]; |
|
var placeholder = parts[placeholderIndex]; |
|
var rest = parts.slice(placeholderIndex + 1); |
|
parts = parts.slice(0, atIndex).concat([at + placeholder]).concat(rest); |
|
} |
|
|
|
var atPlaceholderIndex = parts.findIndex(function (part) { |
|
return typeof part === "string" && part.startsWith("@prettier-placeholder"); |
|
}); |
|
|
|
if (atPlaceholderIndex > -1) { |
|
var _placeholder = parts[atPlaceholderIndex]; |
|
|
|
var _rest = parts.slice(atPlaceholderIndex + 1); |
|
|
|
var placeholderMatch = _placeholder.match(/@prettier-placeholder-(.+)-id([\s\S]*)/); |
|
|
|
var placeholderID = placeholderMatch[1]; // When the expression has a suffix appended, like: |
|
// animation: linear ${time}s ease-out; |
|
|
|
var suffix = placeholderMatch[2]; |
|
var expression = expressions[placeholderID]; |
|
replaceCounter++; |
|
parts = parts.slice(0, atPlaceholderIndex).concat(["${", expression, "}" + suffix]).concat(_rest); |
|
} |
|
|
|
return Object.assign({}, doc, { |
|
parts: parts |
|
}); |
|
}); |
|
return expressions.length === replaceCounter ? newDoc : null; |
|
} |
|
|
|
function printGraphqlComments(lines) { |
|
var parts = []; |
|
var seenComment = false; |
|
lines.map(function (textLine) { |
|
return textLine.trim(); |
|
}).forEach(function (textLine, i, array) { |
|
// Lines are either whitespace only, or a comment (with potential whitespace |
|
// around it). Drop whitespace-only lines. |
|
if (textLine === "") { |
|
return; |
|
} |
|
|
|
if (array[i - 1] === "" && seenComment) { |
|
// If a non-first comment is preceded by a blank (whitespace only) line, |
|
// add in a blank line. |
|
parts.push(concat$b([hardline$8, textLine])); |
|
} else { |
|
parts.push(textLine); |
|
} |
|
|
|
seenComment = true; |
|
}); // If `lines` was whitespace only, return `null`. |
|
|
|
return parts.length === 0 ? null : join$7(hardline$8, parts); |
|
} |
|
/** |
|
* Template literal in these contexts: |
|
* <style jsx>{`div{color:red}`}</style> |
|
* css`` |
|
* css.global`` |
|
* css.resolve`` |
|
*/ |
|
|
|
|
|
function isStyledJsx(path) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
var parentParent = path.getParentNode(1); |
|
return parentParent && node.quasis && parent.type === "JSXExpressionContainer" && parentParent.type === "JSXElement" && parentParent.openingElement.name.name === "style" && parentParent.openingElement.attributes.some(function (attribute) { |
|
return attribute.name.name === "jsx"; |
|
}) || parent && parent.type === "TaggedTemplateExpression" && parent.tag.type === "Identifier" && parent.tag.name === "css" || parent && parent.type === "TaggedTemplateExpression" && parent.tag.type === "MemberExpression" && parent.tag.object.name === "css" && (parent.tag.property.name === "global" || parent.tag.property.name === "resolve"); |
|
} |
|
/** |
|
* Angular Components can have: |
|
* - Inline HTML template |
|
* - Inline CSS styles |
|
* |
|
* ...which are both within template literals somewhere |
|
* inside of the Component decorator factory. |
|
* |
|
* E.g. |
|
* @Component({ |
|
* template: `<div>...</div>`, |
|
* styles: [`h1 { color: blue; }`] |
|
* }) |
|
*/ |
|
|
|
|
|
function isAngularComponentStyles(path) { |
|
return isPathMatch(path, [function (node) { |
|
return node.type === "TemplateLiteral"; |
|
}, function (node, name) { |
|
return node.type === "ArrayExpression" && name === "elements"; |
|
}, function (node, name) { |
|
return node.type === "Property" && node.key.type === "Identifier" && node.key.name === "styles" && name === "value"; |
|
}].concat(getAngularComponentObjectExpressionPredicates())); |
|
} |
|
|
|
function isAngularComponentTemplate(path) { |
|
return isPathMatch(path, [function (node) { |
|
return node.type === "TemplateLiteral"; |
|
}, function (node, name) { |
|
return node.type === "Property" && node.key.type === "Identifier" && node.key.name === "template" && name === "value"; |
|
}].concat(getAngularComponentObjectExpressionPredicates())); |
|
} |
|
|
|
function getAngularComponentObjectExpressionPredicates() { |
|
return [function (node, name) { |
|
return node.type === "ObjectExpression" && name === "properties"; |
|
}, function (node, name) { |
|
return node.type === "CallExpression" && node.callee.type === "Identifier" && node.callee.name === "Component" && name === "arguments"; |
|
}, function (node, name) { |
|
return node.type === "Decorator" && name === "expression"; |
|
}]; |
|
} |
|
/** |
|
* styled-components template literals |
|
*/ |
|
|
|
|
|
function isStyledComponents(path) { |
|
var parent = path.getParentNode(); |
|
|
|
if (!parent || parent.type !== "TaggedTemplateExpression") { |
|
return false; |
|
} |
|
|
|
var tag = parent.tag; |
|
|
|
switch (tag.type) { |
|
case "MemberExpression": |
|
return (// styled.foo`` |
|
isStyledIdentifier(tag.object) || // Component.extend`` |
|
isStyledExtend(tag) |
|
); |
|
|
|
case "CallExpression": |
|
return (// styled(Component)`` |
|
isStyledIdentifier(tag.callee) || tag.callee.type === "MemberExpression" && (tag.callee.object.type === "MemberExpression" && ( // styled.foo.attrs({})`` |
|
isStyledIdentifier(tag.callee.object.object) || // Component.extend.attrs({})`` |
|
isStyledExtend(tag.callee.object)) || // styled(Component).attrs({})`` |
|
tag.callee.object.type === "CallExpression" && isStyledIdentifier(tag.callee.object.callee)) |
|
); |
|
|
|
case "Identifier": |
|
// css`` |
|
return tag.name === "css"; |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
/** |
|
* JSX element with CSS prop |
|
*/ |
|
|
|
|
|
function isCssProp(path) { |
|
var parent = path.getParentNode(); |
|
var parentParent = path.getParentNode(1); |
|
return parentParent && parent.type === "JSXExpressionContainer" && parentParent.type === "JSXAttribute" && parentParent.name.type === "JSXIdentifier" && parentParent.name.name === "css"; |
|
} |
|
|
|
function isStyledIdentifier(node) { |
|
return node.type === "Identifier" && node.name === "styled"; |
|
} |
|
|
|
function isStyledExtend(node) { |
|
return /^[A-Z]/.test(node.object.name) && node.property.name === "extend"; |
|
} |
|
/* |
|
* react-relay and graphql-tag |
|
* graphql`...` |
|
* graphql.experimental`...` |
|
* gql`...` |
|
* GraphQL comment block |
|
* |
|
* This intentionally excludes Relay Classic tags, as Prettier does not |
|
* support Relay Classic formatting. |
|
*/ |
|
|
|
|
|
function isGraphQL(path) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
return hasLanguageComment(node, "GraphQL") || parent && (parent.type === "TaggedTemplateExpression" && (parent.tag.type === "MemberExpression" && parent.tag.object.name === "graphql" && parent.tag.property.name === "experimental" || parent.tag.type === "Identifier" && (parent.tag.name === "gql" || parent.tag.name === "graphql")) || parent.type === "CallExpression" && parent.callee.type === "Identifier" && parent.callee.name === "graphql"); |
|
} |
|
|
|
function hasLanguageComment(node, languageName) { |
|
// This checks for a leading comment that is exactly `/* GraphQL */` |
|
// In order to be in line with other implementations of this comment tag |
|
// we will not trim the comment value and we will expect exactly one space on |
|
// either side of the GraphQL string |
|
// Also see ./clean.js |
|
return hasLeadingComment$1(node, function (comment) { |
|
return isBlockComment$1(comment) && comment.value === " ".concat(languageName, " "); |
|
}); |
|
} |
|
|
|
function isPathMatch(path, predicateStack) { |
|
var stack = path.stack.slice(); |
|
var name = null; |
|
var node = stack.pop(); |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = predicateStack[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var predicate = _step.value; |
|
|
|
if (node === undefined) { |
|
return false; |
|
} // skip index/array |
|
|
|
|
|
if (typeof name === "number") { |
|
name = stack.pop(); |
|
node = stack.pop(); |
|
} |
|
|
|
if (!predicate(node, name)) { |
|
return false; |
|
} |
|
|
|
name = stack.pop(); |
|
node = stack.pop(); |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return true; |
|
} |
|
/** |
|
* - html`...` |
|
* - HTML comment block |
|
*/ |
|
|
|
|
|
function isHtml(path) { |
|
var node = path.getValue(); |
|
return hasLanguageComment(node, "HTML") || isPathMatch(path, [function (node) { |
|
return node.type === "TemplateLiteral"; |
|
}, function (node, name) { |
|
return node.type === "TaggedTemplateExpression" && node.tag.type === "Identifier" && node.tag.name === "html" && name === "quasi"; |
|
}]); |
|
} // The counter is needed to distinguish nested embeds. |
|
|
|
|
|
var htmlTemplateLiteralCounter = 0; |
|
|
|
function printHtmlTemplateLiteral(path, print, textToDoc, parser, escapeClosingScriptTag) { |
|
var node = path.getValue(); |
|
var counter = htmlTemplateLiteralCounter; |
|
htmlTemplateLiteralCounter = htmlTemplateLiteralCounter + 1 >>> 0; |
|
|
|
var composePlaceholder = function composePlaceholder(index) { |
|
return "PRETTIER_HTML_PLACEHOLDER_".concat(index, "_").concat(counter, "_IN_JS"); |
|
}; |
|
|
|
var text = node.quasis.map(function (quasi, index, quasis) { |
|
return index === quasis.length - 1 ? quasi.value.cooked : quasi.value.cooked + composePlaceholder(index); |
|
}).join(""); |
|
var expressionDocs = path.map(print, "expressions"); |
|
|
|
if (expressionDocs.length === 0 && text.trim().length === 0) { |
|
return "``"; |
|
} |
|
|
|
var placeholderRegex = RegExp(composePlaceholder("(\\d+)"), "g"); |
|
var contentDoc = mapDoc$5(stripTrailingHardline$2(textToDoc(text, { |
|
parser: parser |
|
})), function (doc) { |
|
if (typeof doc !== "string") { |
|
return doc; |
|
} |
|
|
|
var parts = []; |
|
var components = doc.split(placeholderRegex); |
|
|
|
for (var i = 0; i < components.length; i++) { |
|
var component = components[i]; |
|
|
|
if (i % 2 === 0) { |
|
if (component) { |
|
component = uncook(component); |
|
|
|
if (escapeClosingScriptTag) { |
|
component = component.replace(/<\/(script)\b/gi, "<\\/$1"); |
|
} |
|
|
|
parts.push(component); |
|
} |
|
|
|
continue; |
|
} |
|
|
|
var placeholderIndex = +component; |
|
parts.push(concat$b(["${", group$a(expressionDocs[placeholderIndex]), "}"])); |
|
} |
|
|
|
return concat$b(parts); |
|
}); |
|
return group$a(concat$b(["`", indent$6(concat$b([hardline$8, group$a(contentDoc)])), softline$5, "`"])); |
|
} |
|
|
|
var embed_1$1 = embed$3; |
|
|
|
function clean$4(ast, newObj, parent) { |
|
["range", "raw", "comments", "leadingComments", "trailingComments", "extra", "start", "end", "flags", "errors"].forEach(function (name) { |
|
delete newObj[name]; |
|
}); |
|
|
|
if (ast.type === "BigIntLiteral") { |
|
newObj.value = newObj.value.toLowerCase(); |
|
} // We remove extra `;` and add them when needed |
|
|
|
|
|
if (ast.type === "EmptyStatement") { |
|
return null; |
|
} // We move text around, including whitespaces and add {" "} |
|
|
|
|
|
if (ast.type === "JSXText") { |
|
return null; |
|
} |
|
|
|
if (ast.type === "JSXExpressionContainer" && ast.expression.type === "Literal" && ast.expression.value === " ") { |
|
return null; |
|
} // We remove unneeded parens around same-operator LogicalExpressions |
|
|
|
|
|
if (isUnbalancedLogicalTree(newObj)) { |
|
return rebalanceLogicalTree(newObj); |
|
} // (TypeScript) Ignore `static` in `constructor(static p) {}` |
|
// and `export` in `constructor(export p) {}` |
|
|
|
|
|
if (ast.type === "TSParameterProperty" && ast.accessibility === null && !ast.readonly) { |
|
return { |
|
type: "Identifier", |
|
name: ast.parameter.name, |
|
typeAnnotation: newObj.parameter.typeAnnotation, |
|
decorators: newObj.decorators |
|
}; |
|
} // (TypeScript) ignore empty `specifiers` array |
|
|
|
|
|
if (ast.type === "TSNamespaceExportDeclaration" && ast.specifiers && ast.specifiers.length === 0) { |
|
delete newObj.specifiers; |
|
} // We convert <div></div> to <div /> |
|
|
|
|
|
if (ast.type === "JSXOpeningElement") { |
|
delete newObj.selfClosing; |
|
} |
|
|
|
if (ast.type === "JSXElement") { |
|
delete newObj.closingElement; |
|
} // We change {'key': value} into {key: value} |
|
|
|
|
|
if ((ast.type === "Property" || ast.type === "ObjectProperty" || ast.type === "MethodDefinition" || ast.type === "ClassProperty" || ast.type === "TSPropertySignature" || ast.type === "ObjectTypeProperty") && _typeof(ast.key) === "object" && ast.key && (ast.key.type === "Literal" || ast.key.type === "StringLiteral" || ast.key.type === "Identifier")) { |
|
delete newObj.key; |
|
} |
|
|
|
if (ast.type === "OptionalMemberExpression" && ast.optional === false) { |
|
newObj.type = "MemberExpression"; |
|
delete newObj.optional; |
|
} // Remove raw and cooked values from TemplateElement when it's CSS |
|
// styled-jsx |
|
|
|
|
|
if (ast.type === "JSXElement" && ast.openingElement.name.name === "style" && ast.openingElement.attributes.some(function (attr) { |
|
return attr.name.name === "jsx"; |
|
})) { |
|
var templateLiterals = newObj.children.filter(function (child) { |
|
return child.type === "JSXExpressionContainer" && child.expression.type === "TemplateLiteral"; |
|
}).map(function (container) { |
|
return container.expression; |
|
}); |
|
var quasis = templateLiterals.reduce(function (quasis, templateLiteral) { |
|
return quasis.concat(templateLiteral.quasis); |
|
}, []); |
|
quasis.forEach(function (q) { |
|
return delete q.value; |
|
}); |
|
} // CSS template literals in css prop |
|
|
|
|
|
if (ast.type === "JSXAttribute" && ast.name.name === "css" && ast.value.type === "JSXExpressionContainer" && ast.value.expression.type === "TemplateLiteral") { |
|
newObj.value.expression.quasis.forEach(function (q) { |
|
return delete q.value; |
|
}); |
|
} // Angular Components: Inline HTML template and Inline CSS styles |
|
|
|
|
|
var expression = ast.expression || ast.callee; |
|
|
|
if (ast.type === "Decorator" && expression.type === "CallExpression" && expression.callee.name === "Component" && expression.arguments.length === 1) { |
|
var astProps = ast.expression.arguments[0].properties; |
|
newObj.expression.arguments[0].properties.forEach(function (prop, index) { |
|
var templateLiteral = null; |
|
|
|
switch (astProps[index].key.name) { |
|
case "styles": |
|
if (prop.value.type === "ArrayExpression") { |
|
templateLiteral = prop.value.elements[0]; |
|
} |
|
|
|
break; |
|
|
|
case "template": |
|
if (prop.value.type === "TemplateLiteral") { |
|
templateLiteral = prop.value; |
|
} |
|
|
|
break; |
|
} |
|
|
|
if (templateLiteral) { |
|
templateLiteral.quasis.forEach(function (q) { |
|
return delete q.value; |
|
}); |
|
} |
|
}); |
|
} // styled-components, graphql, markdown |
|
|
|
|
|
if (ast.type === "TaggedTemplateExpression" && (ast.tag.type === "MemberExpression" || ast.tag.type === "Identifier" && (ast.tag.name === "gql" || ast.tag.name === "graphql" || ast.tag.name === "css" || ast.tag.name === "md" || ast.tag.name === "markdown" || ast.tag.name === "html") || ast.tag.type === "CallExpression")) { |
|
newObj.quasi.quasis.forEach(function (quasi) { |
|
return delete quasi.value; |
|
}); |
|
} |
|
|
|
if (ast.type === "TemplateLiteral") { |
|
// This checks for a leading comment that is exactly `/* GraphQL */` |
|
// In order to be in line with other implementations of this comment tag |
|
// we will not trim the comment value and we will expect exactly one space on |
|
// either side of the GraphQL string |
|
// Also see ./embed.js |
|
var hasLanguageComment = ast.leadingComments && ast.leadingComments.some(function (comment) { |
|
return comment.type === "CommentBlock" && ["GraphQL", "HTML"].some(function (languageName) { |
|
return comment.value === " ".concat(languageName, " "); |
|
}); |
|
}); |
|
|
|
if (hasLanguageComment || parent.type === "CallExpression" && parent.callee.name === "graphql") { |
|
newObj.quasis.forEach(function (quasi) { |
|
return delete quasi.value; |
|
}); |
|
} |
|
} |
|
} |
|
|
|
function isUnbalancedLogicalTree(newObj) { |
|
return newObj.type === "LogicalExpression" && newObj.right.type === "LogicalExpression" && newObj.operator === newObj.right.operator; |
|
} |
|
|
|
function rebalanceLogicalTree(newObj) { |
|
if (isUnbalancedLogicalTree(newObj)) { |
|
return rebalanceLogicalTree({ |
|
type: "LogicalExpression", |
|
operator: newObj.operator, |
|
left: rebalanceLogicalTree({ |
|
type: "LogicalExpression", |
|
operator: newObj.operator, |
|
left: newObj.left, |
|
right: newObj.right.left, |
|
loc: {} |
|
}), |
|
right: newObj.right.right, |
|
loc: {} |
|
}); |
|
} |
|
|
|
return newObj; |
|
} |
|
|
|
var clean_1$1 = clean$4; |
|
|
|
var getLast$1 = util.getLast, |
|
hasNewline$3 = util.hasNewline, |
|
hasNewlineInRange$1 = util.hasNewlineInRange, |
|
hasIgnoreComment$3 = util.hasIgnoreComment, |
|
hasNodeIgnoreComment$1 = util.hasNodeIgnoreComment, |
|
skipWhitespace$1 = util.skipWhitespace; |
|
var isIdentifierName = utils$1.keyword.isIdentifierNameES5; // We match any whitespace except line terminators because |
|
// Flow annotation comments cannot be split across lines. For example: |
|
// |
|
// (this /* |
|
// : any */).foo = 5; |
|
// |
|
// is not picked up by Flow (see https://github.com/facebook/flow/issues/7050), so |
|
// removing the newline would create a type annotation that the user did not intend |
|
// to create. |
|
|
|
var NON_LINE_TERMINATING_WHITE_SPACE = "(?:(?=.)\\s)"; |
|
var FLOW_SHORTHAND_ANNOTATION = new RegExp("^".concat(NON_LINE_TERMINATING_WHITE_SPACE, "*:")); |
|
var FLOW_ANNOTATION = new RegExp("^".concat(NON_LINE_TERMINATING_WHITE_SPACE, "*::")); |
|
|
|
function hasFlowShorthandAnnotationComment(node) { |
|
// https://flow.org/en/docs/types/comments/ |
|
// Syntax example: const r = new (window.Request /*: Class<Request> */)(""); |
|
return node.extra && node.extra.parenthesized && node.trailingComments && node.trailingComments[0].value.match(FLOW_SHORTHAND_ANNOTATION); |
|
} |
|
|
|
function hasFlowAnnotationComment(comments) { |
|
return comments && comments[0].value.match(FLOW_ANNOTATION); |
|
} |
|
|
|
function hasNode(node, fn) { |
|
if (!node || _typeof(node) !== "object") { |
|
return false; |
|
} |
|
|
|
if (Array.isArray(node)) { |
|
return node.some(function (value) { |
|
return hasNode(value, fn); |
|
}); |
|
} |
|
|
|
var result = fn(node); |
|
return typeof result === "boolean" ? result : Object.keys(node).some(function (key) { |
|
return hasNode(node[key], fn); |
|
}); |
|
} |
|
|
|
function hasNakedLeftSide(node) { |
|
return node.type === "AssignmentExpression" || node.type === "BinaryExpression" || node.type === "LogicalExpression" || node.type === "NGPipeExpression" || node.type === "ConditionalExpression" || node.type === "CallExpression" || node.type === "OptionalCallExpression" || node.type === "MemberExpression" || node.type === "OptionalMemberExpression" || node.type === "SequenceExpression" || node.type === "TaggedTemplateExpression" || node.type === "BindExpression" || node.type === "UpdateExpression" && !node.prefix || node.type === "TSAsExpression" || node.type === "TSNonNullExpression"; |
|
} |
|
|
|
function getLeftSide(node) { |
|
if (node.expressions) { |
|
return node.expressions[0]; |
|
} |
|
|
|
return node.left || node.test || node.callee || node.object || node.tag || node.argument || node.expression; |
|
} |
|
|
|
function getLeftSidePathName(path, node) { |
|
if (node.expressions) { |
|
return ["expressions", 0]; |
|
} |
|
|
|
if (node.left) { |
|
return ["left"]; |
|
} |
|
|
|
if (node.test) { |
|
return ["test"]; |
|
} |
|
|
|
if (node.object) { |
|
return ["object"]; |
|
} |
|
|
|
if (node.callee) { |
|
return ["callee"]; |
|
} |
|
|
|
if (node.tag) { |
|
return ["tag"]; |
|
} |
|
|
|
if (node.argument) { |
|
return ["argument"]; |
|
} |
|
|
|
if (node.expression) { |
|
return ["expression"]; |
|
} |
|
|
|
throw new Error("Unexpected node has no left side", node); |
|
} |
|
|
|
function isLiteral(node) { |
|
return node.type === "BooleanLiteral" || node.type === "DirectiveLiteral" || node.type === "Literal" || node.type === "NullLiteral" || node.type === "NumericLiteral" || node.type === "RegExpLiteral" || node.type === "StringLiteral" || node.type === "TemplateLiteral" || node.type === "TSTypeLiteral" || node.type === "JSXText"; |
|
} |
|
|
|
function isNumericLiteral(node) { |
|
return node.type === "NumericLiteral" || node.type === "Literal" && typeof node.value === "number"; |
|
} |
|
|
|
function isStringLiteral(node) { |
|
return node.type === "StringLiteral" || node.type === "Literal" && typeof node.value === "string"; |
|
} |
|
|
|
function isObjectType(n) { |
|
return n.type === "ObjectTypeAnnotation" || n.type === "TSTypeLiteral"; |
|
} |
|
|
|
function isFunctionOrArrowExpression(node) { |
|
return node.type === "FunctionExpression" || node.type === "ArrowFunctionExpression"; |
|
} |
|
|
|
function isFunctionOrArrowExpressionWithBody(node) { |
|
return node.type === "FunctionExpression" || node.type === "ArrowFunctionExpression" && node.body.type === "BlockStatement"; |
|
} |
|
|
|
function isTemplateLiteral(node) { |
|
return node.type === "TemplateLiteral"; |
|
} // `inject` is used in AngularJS 1.x, `async` in Angular 2+ |
|
// example: https://docs.angularjs.org/guide/unit-testing#using-beforeall- |
|
|
|
|
|
function isAngularTestWrapper(node) { |
|
return (node.type === "CallExpression" || node.type === "OptionalCallExpression") && node.callee.type === "Identifier" && (node.callee.name === "async" || node.callee.name === "inject" || node.callee.name === "fakeAsync"); |
|
} |
|
|
|
function isJSXNode(node) { |
|
return node.type === "JSXElement" || node.type === "JSXFragment"; |
|
} |
|
|
|
function isTheOnlyJSXElementInMarkdown(options, path) { |
|
if (options.parentParser !== "markdown" && options.parentParser !== "mdx") { |
|
return false; |
|
} |
|
|
|
var node = path.getNode(); |
|
|
|
if (!node.expression || !isJSXNode(node.expression)) { |
|
return false; |
|
} |
|
|
|
var parent = path.getParentNode(); |
|
return parent.type === "Program" && parent.body.length == 1; |
|
} // Detect an expression node representing `{" "}` |
|
|
|
|
|
function isJSXWhitespaceExpression(node) { |
|
return node.type === "JSXExpressionContainer" && isLiteral(node.expression) && node.expression.value === " " && !node.expression.comments; |
|
} |
|
|
|
function isMemberExpressionChain(node) { |
|
if (node.type !== "MemberExpression" && node.type !== "OptionalMemberExpression") { |
|
return false; |
|
} |
|
|
|
if (node.object.type === "Identifier") { |
|
return true; |
|
} |
|
|
|
return isMemberExpressionChain(node.object); |
|
} |
|
|
|
function isGetterOrSetter(node) { |
|
return node.kind === "get" || node.kind === "set"; |
|
} |
|
|
|
function sameLocStart(nodeA, nodeB, options) { |
|
return options.locStart(nodeA) === options.locStart(nodeB); |
|
} // TODO: This is a bad hack and we need a better way to distinguish between |
|
// arrow functions and otherwise |
|
|
|
|
|
function isFunctionNotation(node, options) { |
|
return isGetterOrSetter(node) || sameLocStart(node, node.value, options); |
|
} // Hack to differentiate between the following two which have the same ast |
|
// type T = { method: () => void }; |
|
// type T = { method(): void }; |
|
|
|
|
|
function isObjectTypePropertyAFunction(node, options) { |
|
return (node.type === "ObjectTypeProperty" || node.type === "ObjectTypeInternalSlot") && node.value.type === "FunctionTypeAnnotation" && !node.static && !isFunctionNotation(node, options); |
|
} // Hack to differentiate between the following two which have the same ast |
|
// declare function f(a): void; |
|
// var f: (a) => void; |
|
|
|
|
|
function isTypeAnnotationAFunction(node, options) { |
|
return (node.type === "TypeAnnotation" || node.type === "TSTypeAnnotation") && node.typeAnnotation.type === "FunctionTypeAnnotation" && !node.static && !sameLocStart(node, node.typeAnnotation, options); |
|
} |
|
|
|
function isBinaryish(node) { |
|
return node.type === "BinaryExpression" || node.type === "LogicalExpression" || node.type === "NGPipeExpression"; |
|
} |
|
|
|
function isMemberish(node) { |
|
return node.type === "MemberExpression" || node.type === "OptionalMemberExpression" || node.type === "BindExpression" && node.object; |
|
} |
|
|
|
function isSimpleFlowType(node) { |
|
var flowTypeAnnotations = ["AnyTypeAnnotation", "NullLiteralTypeAnnotation", "GenericTypeAnnotation", "ThisTypeAnnotation", "NumberTypeAnnotation", "VoidTypeAnnotation", "EmptyTypeAnnotation", "MixedTypeAnnotation", "BooleanTypeAnnotation", "BooleanLiteralTypeAnnotation", "StringTypeAnnotation"]; |
|
return node && flowTypeAnnotations.indexOf(node.type) !== -1 && !(node.type === "GenericTypeAnnotation" && node.typeParameters); |
|
} |
|
|
|
var unitTestRe = /^(skip|[fx]?(it|describe|test))$/; |
|
|
|
function isSkipOrOnlyBlock(node) { |
|
return (node.callee.type === "MemberExpression" || node.callee.type === "OptionalMemberExpression") && node.callee.object.type === "Identifier" && node.callee.property.type === "Identifier" && unitTestRe.test(node.callee.object.name) && (node.callee.property.name === "only" || node.callee.property.name === "skip"); |
|
} |
|
|
|
function isUnitTestSetUp(n) { |
|
var unitTestSetUpRe = /^(before|after)(Each|All)$/; |
|
return n.callee.type === "Identifier" && unitTestSetUpRe.test(n.callee.name) && n.arguments.length === 1; |
|
} // eg; `describe("some string", (done) => {})` |
|
|
|
|
|
function isTestCall(n, parent) { |
|
if (n.type !== "CallExpression") { |
|
return false; |
|
} |
|
|
|
if (n.arguments.length === 1) { |
|
if (isAngularTestWrapper(n) && parent && isTestCall(parent)) { |
|
return isFunctionOrArrowExpression(n.arguments[0]); |
|
} |
|
|
|
if (isUnitTestSetUp(n)) { |
|
return isAngularTestWrapper(n.arguments[0]); |
|
} |
|
} else if (n.arguments.length === 2 || n.arguments.length === 3) { |
|
if ((n.callee.type === "Identifier" && unitTestRe.test(n.callee.name) || isSkipOrOnlyBlock(n)) && (isTemplateLiteral(n.arguments[0]) || isStringLiteral(n.arguments[0]))) { |
|
// it("name", () => { ... }, 2500) |
|
if (n.arguments[2] && !isNumericLiteral(n.arguments[2])) { |
|
return false; |
|
} |
|
|
|
return (n.arguments.length === 2 ? isFunctionOrArrowExpression(n.arguments[1]) : isFunctionOrArrowExpressionWithBody(n.arguments[1]) && n.arguments[1].params.length <= 1) || isAngularTestWrapper(n.arguments[1]); |
|
} |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function hasLeadingComment$2(node) { |
|
return node.comments && node.comments.some(function (comment) { |
|
return comment.leading; |
|
}); |
|
} |
|
|
|
function hasTrailingComment(node) { |
|
return node.comments && node.comments.some(function (comment) { |
|
return comment.trailing; |
|
}); |
|
} |
|
|
|
function isCallOrOptionalCallExpression(node) { |
|
return node.type === "CallExpression" || node.type === "OptionalCallExpression"; |
|
} |
|
|
|
function hasDanglingComments(node) { |
|
return node.comments && node.comments.some(function (comment) { |
|
return !comment.leading && !comment.trailing; |
|
}); |
|
} |
|
/** identify if an angular expression seems to have side effects */ |
|
|
|
|
|
function hasNgSideEffect(path) { |
|
return hasNode(path.getValue(), function (node) { |
|
switch (node.type) { |
|
case undefined: |
|
return false; |
|
|
|
case "CallExpression": |
|
case "OptionalCallExpression": |
|
case "AssignmentExpression": |
|
return true; |
|
} |
|
}); |
|
} |
|
|
|
function isNgForOf(node, index, parentNode) { |
|
return node.type === "NGMicrosyntaxKeyedExpression" && node.key.name === "of" && index === 1 && parentNode.body[0].type === "NGMicrosyntaxLet" && parentNode.body[0].value === null; |
|
} |
|
/** @param node {import("estree").TemplateLiteral} */ |
|
|
|
|
|
function isSimpleTemplateLiteral(node) { |
|
if (node.expressions.length === 0) { |
|
return false; |
|
} |
|
|
|
return node.expressions.every(function (expr) { |
|
// Disallow comments since printDocToString can't print them here |
|
if (expr.comments) { |
|
return false; |
|
} // Allow `x` and `this` |
|
|
|
|
|
if (expr.type === "Identifier" || expr.type === "ThisExpression") { |
|
return true; |
|
} // Allow `a.b.c`, `a.b[c]`, and `this.x.y` |
|
|
|
|
|
if (expr.type === "MemberExpression" || expr.type === "OptionalMemberExpression") { |
|
var head = expr; |
|
|
|
while (head.type === "MemberExpression" || head.type === "OptionalMemberExpression") { |
|
if (head.property.type !== "Identifier" && head.property.type !== "Literal" && head.property.type !== "StringLiteral" && head.property.type !== "NumericLiteral") { |
|
return false; |
|
} |
|
|
|
head = head.object; |
|
|
|
if (head.comments) { |
|
return false; |
|
} |
|
} |
|
|
|
if (head.type === "Identifier" || head.type === "ThisExpression") { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
return false; |
|
}); |
|
} |
|
|
|
function getFlowVariance(path) { |
|
if (!path.variance) { |
|
return null; |
|
} // Babel 7.0 currently uses variance node type, and flow should |
|
// follow suit soon: |
|
// https://github.com/babel/babel/issues/4722 |
|
|
|
|
|
var variance = path.variance.kind || path.variance; |
|
|
|
switch (variance) { |
|
case "plus": |
|
return "+"; |
|
|
|
case "minus": |
|
return "-"; |
|
|
|
default: |
|
/* istanbul ignore next */ |
|
return variance; |
|
} |
|
} |
|
|
|
function classPropMayCauseASIProblems(path) { |
|
var node = path.getNode(); |
|
|
|
if (node.type !== "ClassProperty") { |
|
return false; |
|
} |
|
|
|
var name = node.key && node.key.name; // this isn't actually possible yet with most parsers available today |
|
// so isn't properly tested yet. |
|
|
|
if ((name === "static" || name === "get" || name === "set") && !node.value && !node.typeAnnotation) { |
|
return true; |
|
} |
|
} |
|
|
|
function classChildNeedsASIProtection(node) { |
|
if (!node) { |
|
return; |
|
} |
|
|
|
if (node.static || node.accessibility // TypeScript |
|
) { |
|
return false; |
|
} |
|
|
|
if (!node.computed) { |
|
var name = node.key && node.key.name; |
|
|
|
if (name === "in" || name === "instanceof") { |
|
return true; |
|
} |
|
} |
|
|
|
switch (node.type) { |
|
case "ClassProperty": |
|
case "TSAbstractClassProperty": |
|
return node.computed; |
|
|
|
case "MethodDefinition": // Flow |
|
|
|
case "TSAbstractMethodDefinition": // TypeScript |
|
|
|
case "ClassMethod": |
|
case "ClassPrivateMethod": |
|
{ |
|
// Babel |
|
var isAsync = node.value ? node.value.async : node.async; |
|
var isGenerator = node.value ? node.value.generator : node.generator; |
|
|
|
if (isAsync || node.kind === "get" || node.kind === "set") { |
|
return false; |
|
} |
|
|
|
if (node.computed || isGenerator) { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
case "TSIndexSignature": |
|
return true; |
|
|
|
default: |
|
/* istanbul ignore next */ |
|
return false; |
|
} |
|
} |
|
|
|
function getTypeScriptMappedTypeModifier(tokenNode, keyword) { |
|
if (tokenNode === "+") { |
|
return "+" + keyword; |
|
} else if (tokenNode === "-") { |
|
return "-" + keyword; |
|
} |
|
|
|
return keyword; |
|
} |
|
|
|
function hasNewlineBetweenOrAfterDecorators(node, options) { |
|
return hasNewlineInRange$1(options.originalText, options.locStart(node.decorators[0]), options.locEnd(getLast$1(node.decorators))) || hasNewline$3(options.originalText, options.locEnd(getLast$1(node.decorators))); |
|
} // Only space, newline, carriage return, and tab are treated as whitespace |
|
// inside JSX. |
|
|
|
|
|
var jsxWhitespaceChars = " \n\r\t"; |
|
var matchJsxWhitespaceRegex = new RegExp("([" + jsxWhitespaceChars + "]+)"); |
|
var containsNonJsxWhitespaceRegex = new RegExp("[^" + jsxWhitespaceChars + "]"); // Meaningful if it contains non-whitespace characters, |
|
// or it contains whitespace without a new line. |
|
|
|
function isMeaningfulJSXText(node) { |
|
return isLiteral(node) && (containsNonJsxWhitespaceRegex.test(rawText(node)) || !/\n/.test(rawText(node))); |
|
} |
|
|
|
function hasJsxIgnoreComment(path) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
|
|
if (!parent || !node || !isJSXNode(node) || !isJSXNode(parent)) { |
|
return false; |
|
} // Lookup the previous sibling, ignoring any empty JSXText elements |
|
|
|
|
|
var index = parent.children.indexOf(node); |
|
var prevSibling = null; |
|
|
|
for (var i = index; i > 0; i--) { |
|
var candidate = parent.children[i - 1]; |
|
|
|
if (candidate.type === "JSXText" && !isMeaningfulJSXText(candidate)) { |
|
continue; |
|
} |
|
|
|
prevSibling = candidate; |
|
break; |
|
} |
|
|
|
return prevSibling && prevSibling.type === "JSXExpressionContainer" && prevSibling.expression.type === "JSXEmptyExpression" && prevSibling.expression.comments && prevSibling.expression.comments.find(function (comment) { |
|
return comment.value.trim() === "prettier-ignore"; |
|
}); |
|
} |
|
|
|
function isEmptyJSXElement(node) { |
|
if (node.children.length === 0) { |
|
return true; |
|
} |
|
|
|
if (node.children.length > 1) { |
|
return false; |
|
} // if there is one text child and does not contain any meaningful text |
|
// we can treat the element as empty. |
|
|
|
|
|
var child = node.children[0]; |
|
return isLiteral(child) && !isMeaningfulJSXText(child); |
|
} |
|
|
|
function hasPrettierIgnore$2(path) { |
|
return hasIgnoreComment$3(path) || hasJsxIgnoreComment(path); |
|
} |
|
|
|
function isLastStatement(path) { |
|
var parent = path.getParentNode(); |
|
|
|
if (!parent) { |
|
return true; |
|
} |
|
|
|
var node = path.getValue(); |
|
var body = (parent.body || parent.consequent).filter(function (stmt) { |
|
return stmt.type !== "EmptyStatement"; |
|
}); |
|
return body && body[body.length - 1] === node; |
|
} |
|
|
|
function isFlowAnnotationComment(text, typeAnnotation, options) { |
|
var start = options.locStart(typeAnnotation); |
|
var end = skipWhitespace$1(text, options.locEnd(typeAnnotation)); |
|
return text.substr(start, 2) === "/*" && text.substr(end, 2) === "*/"; |
|
} |
|
|
|
function hasLeadingOwnLineComment(text, node, options) { |
|
if (isJSXNode(node)) { |
|
return hasNodeIgnoreComment$1(node); |
|
} |
|
|
|
var res = node.comments && node.comments.some(function (comment) { |
|
return comment.leading && hasNewline$3(text, options.locEnd(comment)); |
|
}); |
|
return res; |
|
} // This recurses the return argument, looking for the first token |
|
// (the leftmost leaf node) and, if it (or its parents) has any |
|
// leadingComments, returns true (so it can be wrapped in parens). |
|
|
|
|
|
function returnArgumentHasLeadingComment(options, argument) { |
|
if (hasLeadingOwnLineComment(options.originalText, argument, options)) { |
|
return true; |
|
} |
|
|
|
if (hasNakedLeftSide(argument)) { |
|
var leftMost = argument; |
|
var newLeftMost; |
|
|
|
while (newLeftMost = getLeftSide(leftMost)) { |
|
leftMost = newLeftMost; |
|
|
|
if (hasLeadingOwnLineComment(options.originalText, leftMost, options)) { |
|
return true; |
|
} |
|
} |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isStringPropSafeToCoerceToIdentifier(node, options) { |
|
return isStringLiteral(node.key) && isIdentifierName(node.key.value) && options.parser !== "json" && !(options.parser === "typescript" && node.type === "ClassProperty"); |
|
} |
|
|
|
function isJestEachTemplateLiteral(node, parentNode) { |
|
/** |
|
* describe.each`table`(name, fn) |
|
* describe.only.each`table`(name, fn) |
|
* describe.skip.each`table`(name, fn) |
|
* test.each`table`(name, fn) |
|
* test.only.each`table`(name, fn) |
|
* test.skip.each`table`(name, fn) |
|
* |
|
* Ref: https://github.com/facebook/jest/pull/6102 |
|
*/ |
|
var jestEachTriggerRegex = /^[xf]?(describe|it|test)$/; |
|
return parentNode.type === "TaggedTemplateExpression" && parentNode.quasi === node && parentNode.tag.type === "MemberExpression" && parentNode.tag.property.type === "Identifier" && parentNode.tag.property.name === "each" && (parentNode.tag.object.type === "Identifier" && jestEachTriggerRegex.test(parentNode.tag.object.name) || parentNode.tag.object.type === "MemberExpression" && parentNode.tag.object.property.type === "Identifier" && (parentNode.tag.object.property.name === "only" || parentNode.tag.object.property.name === "skip") && parentNode.tag.object.object.type === "Identifier" && jestEachTriggerRegex.test(parentNode.tag.object.object.name)); |
|
} |
|
|
|
function templateLiteralHasNewLines(template) { |
|
return template.quasis.some(function (quasi) { |
|
return quasi.value.raw.includes("\n"); |
|
}); |
|
} |
|
|
|
function isTemplateOnItsOwnLine(n, text, options) { |
|
return (n.type === "TemplateLiteral" && templateLiteralHasNewLines(n) || n.type === "TaggedTemplateExpression" && templateLiteralHasNewLines(n.quasi)) && !hasNewline$3(text, options.locStart(n), { |
|
backwards: true |
|
}); |
|
} |
|
|
|
function needsHardlineAfterDanglingComment(node) { |
|
if (!node.comments) { |
|
return false; |
|
} |
|
|
|
var lastDanglingComment = getLast$1(node.comments.filter(function (comment) { |
|
return !comment.leading && !comment.trailing; |
|
})); |
|
return lastDanglingComment && !comments$1.isBlockComment(lastDanglingComment); |
|
} // If we have nested conditional expressions, we want to print them in JSX mode |
|
// if there's at least one JSXElement somewhere in the tree. |
|
// |
|
// A conditional expression chain like this should be printed in normal mode, |
|
// because there aren't JSXElements anywhere in it: |
|
// |
|
// isA ? "A" : isB ? "B" : isC ? "C" : "Unknown"; |
|
// |
|
// But a conditional expression chain like this should be printed in JSX mode, |
|
// because there is a JSXElement in the last ConditionalExpression: |
|
// |
|
// isA ? "A" : isB ? "B" : isC ? "C" : <span className="warning">Unknown</span>; |
|
// |
|
// This type of ConditionalExpression chain is structured like this in the AST: |
|
// |
|
// ConditionalExpression { |
|
// test: ..., |
|
// consequent: ..., |
|
// alternate: ConditionalExpression { |
|
// test: ..., |
|
// consequent: ..., |
|
// alternate: ConditionalExpression { |
|
// test: ..., |
|
// consequent: ..., |
|
// alternate: ..., |
|
// } |
|
// } |
|
// } |
|
// |
|
// We want to traverse over that shape and convert it into a flat structure so |
|
// that we can find if there's a JSXElement somewhere inside. |
|
|
|
|
|
function getConditionalChainContents(node) { |
|
// Given this code: |
|
// |
|
// // Using a ConditionalExpression as the consequent is uncommon, but should |
|
// // be handled. |
|
// A ? B : C ? D : E ? F ? G : H : I |
|
// |
|
// which has this AST: |
|
// |
|
// ConditionalExpression { |
|
// test: Identifier(A), |
|
// consequent: Identifier(B), |
|
// alternate: ConditionalExpression { |
|
// test: Identifier(C), |
|
// consequent: Identifier(D), |
|
// alternate: ConditionalExpression { |
|
// test: Identifier(E), |
|
// consequent: ConditionalExpression { |
|
// test: Identifier(F), |
|
// consequent: Identifier(G), |
|
// alternate: Identifier(H), |
|
// }, |
|
// alternate: Identifier(I), |
|
// } |
|
// } |
|
// } |
|
// |
|
// we should return this Array: |
|
// |
|
// [ |
|
// Identifier(A), |
|
// Identifier(B), |
|
// Identifier(C), |
|
// Identifier(D), |
|
// Identifier(E), |
|
// Identifier(F), |
|
// Identifier(G), |
|
// Identifier(H), |
|
// Identifier(I) |
|
// ]; |
|
// |
|
// This loses the information about whether each node was the test, |
|
// consequent, or alternate, but we don't care about that here- we are only |
|
// flattening this structure to find if there's any JSXElements inside. |
|
var nonConditionalExpressions = []; |
|
|
|
function recurse(node) { |
|
if (node.type === "ConditionalExpression") { |
|
recurse(node.test); |
|
recurse(node.consequent); |
|
recurse(node.alternate); |
|
} else { |
|
nonConditionalExpressions.push(node); |
|
} |
|
} |
|
|
|
recurse(node); |
|
return nonConditionalExpressions; |
|
} |
|
|
|
function conditionalExpressionChainContainsJSX(node) { |
|
return Boolean(getConditionalChainContents(node).find(isJSXNode)); |
|
} // Logic to check for args with multiple anonymous functions. For instance, |
|
// the following call should be split on multiple lines for readability: |
|
// source.pipe(map((x) => x + x), filter((x) => x % 2 === 0)) |
|
|
|
|
|
function isFunctionCompositionArgs(args) { |
|
if (args.length <= 1) { |
|
return false; |
|
} |
|
|
|
var count = 0; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = args[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var arg = _step.value; |
|
|
|
if (isFunctionOrArrowExpression(arg)) { |
|
count += 1; |
|
|
|
if (count > 1) { |
|
return true; |
|
} |
|
} else if (isCallOrOptionalCallExpression(arg)) { |
|
var _iteratorNormalCompletion2 = true; |
|
var _didIteratorError2 = false; |
|
var _iteratorError2 = undefined; |
|
|
|
try { |
|
for (var _iterator2 = arg.arguments[Symbol.iterator](), _step2; !(_iteratorNormalCompletion2 = (_step2 = _iterator2.next()).done); _iteratorNormalCompletion2 = true) { |
|
var childArg = _step2.value; |
|
|
|
if (isFunctionOrArrowExpression(childArg)) { |
|
return true; |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError2 = true; |
|
_iteratorError2 = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion2 && _iterator2.return != null) { |
|
_iterator2.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError2) { |
|
throw _iteratorError2; |
|
} |
|
} |
|
} |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return false; |
|
} // Logic to determine if a call is a “long curried function call”. |
|
// See https://github.com/prettier/prettier/issues/1420. |
|
// |
|
// `connect(a, b, c)(d)` |
|
// In the above call expression, the second call is the parent node and the |
|
// first call is the current node. |
|
|
|
|
|
function isLongCurriedCallExpression(path) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
return isCallOrOptionalCallExpression(node) && isCallOrOptionalCallExpression(parent) && parent.callee === node && node.arguments.length > parent.arguments.length && parent.arguments.length > 0; |
|
} |
|
|
|
function rawText(node) { |
|
return node.extra ? node.extra.raw : node.raw; |
|
} |
|
|
|
function identity$1(x) { |
|
return x; |
|
} |
|
|
|
function isTSXFile(options) { |
|
return options.filepath && /\.tsx$/i.test(options.filepath); |
|
} |
|
|
|
var utils$4 = { |
|
classChildNeedsASIProtection: classChildNeedsASIProtection, |
|
classPropMayCauseASIProblems: classPropMayCauseASIProblems, |
|
conditionalExpressionChainContainsJSX: conditionalExpressionChainContainsJSX, |
|
getFlowVariance: getFlowVariance, |
|
getLeftSidePathName: getLeftSidePathName, |
|
getTypeScriptMappedTypeModifier: getTypeScriptMappedTypeModifier, |
|
hasDanglingComments: hasDanglingComments, |
|
hasFlowAnnotationComment: hasFlowAnnotationComment, |
|
hasFlowShorthandAnnotationComment: hasFlowShorthandAnnotationComment, |
|
hasLeadingComment: hasLeadingComment$2, |
|
hasLeadingOwnLineComment: hasLeadingOwnLineComment, |
|
hasNakedLeftSide: hasNakedLeftSide, |
|
hasNewlineBetweenOrAfterDecorators: hasNewlineBetweenOrAfterDecorators, |
|
hasNgSideEffect: hasNgSideEffect, |
|
hasNode: hasNode, |
|
hasPrettierIgnore: hasPrettierIgnore$2, |
|
hasTrailingComment: hasTrailingComment, |
|
identity: identity$1, |
|
isBinaryish: isBinaryish, |
|
isCallOrOptionalCallExpression: isCallOrOptionalCallExpression, |
|
isEmptyJSXElement: isEmptyJSXElement, |
|
isFlowAnnotationComment: isFlowAnnotationComment, |
|
isFunctionCompositionArgs: isFunctionCompositionArgs, |
|
isFunctionNotation: isFunctionNotation, |
|
isFunctionOrArrowExpression: isFunctionOrArrowExpression, |
|
isGetterOrSetter: isGetterOrSetter, |
|
isJestEachTemplateLiteral: isJestEachTemplateLiteral, |
|
isJSXNode: isJSXNode, |
|
isJSXWhitespaceExpression: isJSXWhitespaceExpression, |
|
isLastStatement: isLastStatement, |
|
isLiteral: isLiteral, |
|
isLongCurriedCallExpression: isLongCurriedCallExpression, |
|
isMeaningfulJSXText: isMeaningfulJSXText, |
|
isMemberExpressionChain: isMemberExpressionChain, |
|
isMemberish: isMemberish, |
|
isNgForOf: isNgForOf, |
|
isNumericLiteral: isNumericLiteral, |
|
isObjectType: isObjectType, |
|
isObjectTypePropertyAFunction: isObjectTypePropertyAFunction, |
|
isSimpleFlowType: isSimpleFlowType, |
|
isSimpleTemplateLiteral: isSimpleTemplateLiteral, |
|
isStringLiteral: isStringLiteral, |
|
isStringPropSafeToCoerceToIdentifier: isStringPropSafeToCoerceToIdentifier, |
|
isTemplateOnItsOwnLine: isTemplateOnItsOwnLine, |
|
isTestCall: isTestCall, |
|
isTheOnlyJSXElementInMarkdown: isTheOnlyJSXElementInMarkdown, |
|
isTSXFile: isTSXFile, |
|
isTypeAnnotationAFunction: isTypeAnnotationAFunction, |
|
matchJsxWhitespaceRegex: matchJsxWhitespaceRegex, |
|
needsHardlineAfterDanglingComment: needsHardlineAfterDanglingComment, |
|
rawText: rawText, |
|
returnArgumentHasLeadingComment: returnArgumentHasLeadingComment |
|
}; |
|
|
|
var getLeftSidePathName$1 = utils$4.getLeftSidePathName, |
|
hasFlowShorthandAnnotationComment$1 = utils$4.hasFlowShorthandAnnotationComment, |
|
hasNakedLeftSide$1 = utils$4.hasNakedLeftSide, |
|
hasNode$1 = utils$4.hasNode; |
|
|
|
function hasClosureCompilerTypeCastComment(text, path) { |
|
// https://github.com/google/closure-compiler/wiki/Annotating-Types#type-casts |
|
// Syntax example: var x = /** @type {string} */ (fruit); |
|
var n = path.getValue(); |
|
return isParenthesized(n) && (hasTypeCastComment(n) || hasAncestorTypeCastComment(0)); // for sub-item: /** @type {array} */ (numberOrString).map(x => x); |
|
|
|
function hasAncestorTypeCastComment(index) { |
|
var ancestor = path.getParentNode(index); |
|
return ancestor && !isParenthesized(ancestor) ? hasTypeCastComment(ancestor) || hasAncestorTypeCastComment(index + 1) : false; |
|
} |
|
|
|
function hasTypeCastComment(node) { |
|
return node.comments && node.comments.some(function (comment) { |
|
return comment.leading && comments$1.isBlockComment(comment) && isTypeCastComment(comment.value); |
|
}); |
|
} |
|
|
|
function isParenthesized(node) { |
|
// Closure typecast comments only really make sense when _not_ using |
|
// typescript or flow parsers, so we take advantage of the babel parser's |
|
// parenthesized expressions. |
|
return node.extra && node.extra.parenthesized; |
|
} |
|
|
|
function isTypeCastComment(comment) { |
|
var cleaned = comment.trim().split("\n").map(function (line) { |
|
return line.replace(/^[\s*]+/, ""); |
|
}).join(" ").trim(); |
|
|
|
if (!/^@type\s*\{[^]+\}$/.test(cleaned)) { |
|
return false; |
|
} |
|
|
|
var isCompletelyClosed = false; |
|
var unpairedBracketCount = 0; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = cleaned[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var char = _step.value; |
|
|
|
if (char === "{") { |
|
if (isCompletelyClosed) { |
|
return false; |
|
} |
|
|
|
unpairedBracketCount++; |
|
} else if (char === "}") { |
|
if (unpairedBracketCount === 0) { |
|
return false; |
|
} |
|
|
|
unpairedBracketCount--; |
|
|
|
if (unpairedBracketCount === 0) { |
|
isCompletelyClosed = true; |
|
} |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
return unpairedBracketCount === 0; |
|
} |
|
} |
|
|
|
function needsParens(path, options) { |
|
var parent = path.getParentNode(); |
|
|
|
if (!parent) { |
|
return false; |
|
} |
|
|
|
var name = path.getName(); |
|
var node = path.getNode(); // If the value of this path is some child of a Node and not a Node |
|
// itself, then it doesn't need parentheses. Only Node objects (in |
|
// fact, only Expression nodes) need parentheses. |
|
|
|
if (path.getValue() !== node) { |
|
return false; |
|
} // to avoid unexpected `}}` in HTML interpolations |
|
|
|
|
|
if (options.__isInHtmlInterpolation && !options.bracketSpacing && endsWithRightBracket(node) && isFollowedByRightBracket(path)) { |
|
return true; |
|
} // Only statements don't need parentheses. |
|
|
|
|
|
if (isStatement(node)) { |
|
return false; |
|
} // Closure compiler requires that type casted expressions to be surrounded by |
|
// parentheses. |
|
|
|
|
|
if (hasClosureCompilerTypeCastComment(options.originalText, path)) { |
|
return true; |
|
} |
|
|
|
if ( // Preserve parens if we have a Flow annotation comment, unless we're using the Flow |
|
// parser. The Flow parser turns Flow comments into type annotation nodes in its |
|
// AST, which we handle separately. |
|
options.parser !== "flow" && hasFlowShorthandAnnotationComment$1(path.getValue())) { |
|
return true; |
|
} // Identifiers never need parentheses. |
|
|
|
|
|
if (node.type === "Identifier") { |
|
// ...unless those identifiers are embed placeholders. They might be substituted by complex |
|
// expressions, so the parens around them should not be dropped. Example (JS-in-HTML-in-JS): |
|
// let tpl = html`<script> f((${expr}) / 2); </script>`; |
|
// If the inner JS formatter removes the parens, the expression might change its meaning: |
|
// f((a + b) / 2) vs f(a + b / 2) |
|
if (node.extra && node.extra.parenthesized && /^PRETTIER_HTML_PLACEHOLDER_\d+_\d+_IN_JS$/.test(node.name)) { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
if (parent.type === "ParenthesizedExpression") { |
|
return false; |
|
} // Add parens around the extends clause of a class. It is needed for almost |
|
// all expressions. |
|
|
|
|
|
if ((parent.type === "ClassDeclaration" || parent.type === "ClassExpression") && parent.superClass === node && (node.type === "ArrowFunctionExpression" || node.type === "AssignmentExpression" || node.type === "AwaitExpression" || node.type === "BinaryExpression" || node.type === "ConditionalExpression" || node.type === "LogicalExpression" || node.type === "NewExpression" || node.type === "ObjectExpression" || node.type === "ParenthesizedExpression" || node.type === "SequenceExpression" || node.type === "TaggedTemplateExpression" || node.type === "UnaryExpression" || node.type === "UpdateExpression" || node.type === "YieldExpression")) { |
|
return true; |
|
} // `export default function` or `export default class` can't be followed by |
|
// anything after. So an expression like `export default (function(){}).toString()` |
|
// needs to be followed by a parentheses |
|
|
|
|
|
if (parent.type === "ExportDefaultDeclaration") { |
|
return shouldWrapFunctionForExportDefault(path, options); |
|
} |
|
|
|
if (parent.type === "Decorator" && parent.expression === node) { |
|
var hasCallExpression = false; |
|
var hasMemberExpression = false; |
|
var current = node; |
|
|
|
while (current) { |
|
switch (current.type) { |
|
case "MemberExpression": |
|
hasMemberExpression = true; |
|
current = current.object; |
|
break; |
|
|
|
case "CallExpression": |
|
if ( |
|
/** @(x().y) */ |
|
hasMemberExpression || |
|
/** @(x().y()) */ |
|
hasCallExpression) { |
|
return true; |
|
} |
|
|
|
hasCallExpression = true; |
|
current = current.callee; |
|
break; |
|
|
|
case "Identifier": |
|
return false; |
|
|
|
default: |
|
return true; |
|
} |
|
} |
|
|
|
return true; |
|
} |
|
|
|
if (parent.type === "ArrowFunctionExpression" && parent.body === node && node.type !== "SequenceExpression" && // these have parens added anyway |
|
util.startsWithNoLookaheadToken(node, |
|
/* forbidFunctionClassAndDoExpr */ |
|
false) || parent.type === "ExpressionStatement" && util.startsWithNoLookaheadToken(node, |
|
/* forbidFunctionClassAndDoExpr */ |
|
true)) { |
|
return true; |
|
} |
|
|
|
switch (node.type) { |
|
case "SpreadElement": |
|
case "SpreadProperty": |
|
return parent.type === "MemberExpression" && name === "object" && parent.object === node; |
|
|
|
case "UpdateExpression": |
|
if (parent.type === "UnaryExpression") { |
|
return node.prefix && (node.operator === "++" && parent.operator === "+" || node.operator === "--" && parent.operator === "-"); |
|
} |
|
|
|
// else fallthrough |
|
|
|
case "UnaryExpression": |
|
switch (parent.type) { |
|
case "UnaryExpression": |
|
return node.operator === parent.operator && (node.operator === "+" || node.operator === "-"); |
|
|
|
case "BindExpression": |
|
return true; |
|
|
|
case "MemberExpression": |
|
case "OptionalMemberExpression": |
|
return name === "object"; |
|
|
|
case "TaggedTemplateExpression": |
|
return true; |
|
|
|
case "NewExpression": |
|
case "CallExpression": |
|
case "OptionalCallExpression": |
|
return name === "callee"; |
|
|
|
case "BinaryExpression": |
|
return parent.operator === "**" && name === "left"; |
|
|
|
case "TSNonNullExpression": |
|
return true; |
|
|
|
default: |
|
return false; |
|
} |
|
|
|
case "BinaryExpression": |
|
{ |
|
if (parent.type === "UpdateExpression") { |
|
return true; |
|
} |
|
|
|
var isLeftOfAForStatement = function isLeftOfAForStatement(node) { |
|
var i = 0; |
|
|
|
while (node) { |
|
var _parent = path.getParentNode(i++); |
|
|
|
if (!_parent) { |
|
return false; |
|
} |
|
|
|
if (_parent.type === "ForStatement" && _parent.init === node) { |
|
return true; |
|
} |
|
|
|
node = _parent; |
|
} |
|
|
|
return false; |
|
}; |
|
|
|
if (node.operator === "in" && isLeftOfAForStatement(node)) { |
|
return true; |
|
} |
|
} |
|
// fallthrough |
|
|
|
case "TSTypeAssertion": |
|
case "TSAsExpression": |
|
case "LogicalExpression": |
|
switch (parent.type) { |
|
case "ConditionalExpression": |
|
return node.type === "TSAsExpression"; |
|
|
|
case "CallExpression": |
|
case "NewExpression": |
|
case "OptionalCallExpression": |
|
return name === "callee"; |
|
|
|
case "ClassExpression": |
|
case "ClassDeclaration": |
|
return name === "superClass" && parent.superClass === node; |
|
|
|
case "TSTypeAssertion": |
|
case "TaggedTemplateExpression": |
|
case "UnaryExpression": |
|
case "JSXSpreadAttribute": |
|
case "SpreadElement": |
|
case "SpreadProperty": |
|
case "BindExpression": |
|
case "AwaitExpression": |
|
case "TSAsExpression": |
|
case "TSNonNullExpression": |
|
case "UpdateExpression": |
|
return true; |
|
|
|
case "MemberExpression": |
|
case "OptionalMemberExpression": |
|
return name === "object"; |
|
|
|
case "AssignmentExpression": |
|
return parent.left === node && (node.type === "TSTypeAssertion" || node.type === "TSAsExpression"); |
|
|
|
case "LogicalExpression": |
|
if (node.type === "LogicalExpression") { |
|
return parent.operator !== node.operator; |
|
} |
|
|
|
// else fallthrough |
|
|
|
case "BinaryExpression": |
|
{ |
|
if (!node.operator && node.type !== "TSTypeAssertion") { |
|
return true; |
|
} |
|
|
|
var po = parent.operator; |
|
var pp = util.getPrecedence(po); |
|
var no = node.operator; |
|
var np = util.getPrecedence(no); |
|
|
|
if (pp > np) { |
|
return true; |
|
} |
|
|
|
if (pp === np && name === "right") { |
|
assert.strictEqual(parent.right, node); |
|
return true; |
|
} |
|
|
|
if (pp === np && !util.shouldFlatten(po, no)) { |
|
return true; |
|
} |
|
|
|
if (pp < np && no === "%") { |
|
return po === "+" || po === "-"; |
|
} // Add parenthesis when working with bitwise operators |
|
// It's not strictly needed but helps with code understanding |
|
|
|
|
|
if (util.isBitwiseOperator(po)) { |
|
return true; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
default: |
|
return false; |
|
} |
|
|
|
case "SequenceExpression": |
|
switch (parent.type) { |
|
case "ReturnStatement": |
|
return false; |
|
|
|
case "ForStatement": |
|
// Although parentheses wouldn't hurt around sequence |
|
// expressions in the head of for loops, traditional style |
|
// dictates that e.g. i++, j++ should not be wrapped with |
|
// parentheses. |
|
return false; |
|
|
|
case "ExpressionStatement": |
|
return name !== "expression"; |
|
|
|
case "ArrowFunctionExpression": |
|
// We do need parentheses, but SequenceExpressions are handled |
|
// specially when printing bodies of arrow functions. |
|
return name !== "body"; |
|
|
|
default: |
|
// Otherwise err on the side of overparenthesization, adding |
|
// explicit exceptions above if this proves overzealous. |
|
return true; |
|
} |
|
|
|
case "YieldExpression": |
|
if (parent.type === "UnaryExpression" || parent.type === "AwaitExpression" || parent.type === "TSAsExpression" || parent.type === "TSNonNullExpression") { |
|
return true; |
|
} |
|
|
|
// else fallthrough |
|
|
|
case "AwaitExpression": |
|
switch (parent.type) { |
|
case "TaggedTemplateExpression": |
|
case "UnaryExpression": |
|
case "BinaryExpression": |
|
case "LogicalExpression": |
|
case "SpreadElement": |
|
case "SpreadProperty": |
|
case "TSAsExpression": |
|
case "TSNonNullExpression": |
|
case "BindExpression": |
|
return true; |
|
|
|
case "MemberExpression": |
|
case "OptionalMemberExpression": |
|
return name === "object"; |
|
|
|
case "NewExpression": |
|
case "CallExpression": |
|
case "OptionalCallExpression": |
|
return name === "callee"; |
|
|
|
case "ConditionalExpression": |
|
return parent.test === node; |
|
|
|
default: |
|
return false; |
|
} |
|
|
|
case "TSConditionalType": |
|
if (parent.type === "TSConditionalType" && node === parent.extendsType) { |
|
return true; |
|
} |
|
|
|
// fallthrough |
|
|
|
case "TSFunctionType": |
|
case "TSConstructorType": |
|
if (parent.type === "TSConditionalType" && node === parent.checkType) { |
|
return true; |
|
} |
|
|
|
// fallthrough |
|
|
|
case "TSUnionType": |
|
case "TSIntersectionType": |
|
if (parent.type === "TSUnionType" || parent.type === "TSIntersectionType") { |
|
return true; |
|
} |
|
|
|
// fallthrough |
|
|
|
case "TSTypeOperator": |
|
case "TSInferType": |
|
return parent.type === "TSArrayType" || parent.type === "TSOptionalType" || parent.type === "TSRestType" || parent.type === "TSIndexedAccessType" && node === parent.objectType || parent.type === "TSTypeOperator"; |
|
|
|
case "ArrayTypeAnnotation": |
|
return parent.type === "NullableTypeAnnotation"; |
|
|
|
case "IntersectionTypeAnnotation": |
|
case "UnionTypeAnnotation": |
|
return parent.type === "ArrayTypeAnnotation" || parent.type === "NullableTypeAnnotation" || parent.type === "IntersectionTypeAnnotation" || parent.type === "UnionTypeAnnotation"; |
|
|
|
case "NullableTypeAnnotation": |
|
return parent.type === "ArrayTypeAnnotation"; |
|
|
|
case "FunctionTypeAnnotation": |
|
{ |
|
var ancestor = parent.type === "NullableTypeAnnotation" ? path.getParentNode(1) : parent; |
|
return ancestor.type === "UnionTypeAnnotation" || ancestor.type === "IntersectionTypeAnnotation" || ancestor.type === "ArrayTypeAnnotation" || // We should check ancestor's parent to know whether the parentheses |
|
// are really needed, but since ??T doesn't make sense this check |
|
// will almost never be true. |
|
ancestor.type === "NullableTypeAnnotation"; |
|
} |
|
|
|
case "StringLiteral": |
|
case "NumericLiteral": |
|
case "Literal": |
|
if (typeof node.value === "string" && parent.type === "ExpressionStatement" && ( // TypeScript workaround for https://github.com/JamesHenry/typescript-estree/issues/2 |
|
// See corresponding workaround in printer.js case: "Literal" |
|
options.parser !== "typescript" && !parent.directive || options.parser === "typescript" && options.originalText.substr(options.locStart(node) - 1, 1) === "(")) { |
|
// To avoid becoming a directive |
|
var grandParent = path.getParentNode(1); |
|
return grandParent.type === "Program" || grandParent.type === "BlockStatement"; |
|
} |
|
|
|
return parent.type === "MemberExpression" && typeof node.value === "number" && name === "object" && parent.object === node; |
|
|
|
case "AssignmentExpression": |
|
{ |
|
var _grandParent = path.getParentNode(1); |
|
|
|
if (parent.type === "ArrowFunctionExpression" && parent.body === node) { |
|
return true; |
|
} else if (parent.type === "ClassProperty" && parent.key === node && parent.computed) { |
|
return false; |
|
} else if (parent.type === "TSPropertySignature" && parent.name === node) { |
|
return false; |
|
} else if (parent.type === "ForStatement" && (parent.init === node || parent.update === node)) { |
|
return false; |
|
} else if (parent.type === "ExpressionStatement") { |
|
return node.left.type === "ObjectPattern"; |
|
} else if (parent.type === "TSPropertySignature" && parent.key === node) { |
|
return false; |
|
} else if (parent.type === "AssignmentExpression") { |
|
return false; |
|
} else if (parent.type === "SequenceExpression" && _grandParent && _grandParent.type === "ForStatement" && (_grandParent.init === parent || _grandParent.update === parent)) { |
|
return false; |
|
} else if (parent.type === "Property" && parent.value === node) { |
|
return false; |
|
} else if (parent.type === "NGChainedExpression") { |
|
return false; |
|
} |
|
|
|
return true; |
|
} |
|
|
|
case "ConditionalExpression": |
|
switch (parent.type) { |
|
case "TaggedTemplateExpression": |
|
case "UnaryExpression": |
|
case "SpreadElement": |
|
case "SpreadProperty": |
|
case "BinaryExpression": |
|
case "LogicalExpression": |
|
case "NGPipeExpression": |
|
case "ExportDefaultDeclaration": |
|
case "AwaitExpression": |
|
case "JSXSpreadAttribute": |
|
case "TSTypeAssertion": |
|
case "TypeCastExpression": |
|
case "TSAsExpression": |
|
case "TSNonNullExpression": |
|
return true; |
|
|
|
case "NewExpression": |
|
case "CallExpression": |
|
case "OptionalCallExpression": |
|
return name === "callee"; |
|
|
|
case "ConditionalExpression": |
|
return name === "test" && parent.test === node; |
|
|
|
case "MemberExpression": |
|
case "OptionalMemberExpression": |
|
return name === "object"; |
|
|
|
default: |
|
return false; |
|
} |
|
|
|
case "FunctionExpression": |
|
switch (parent.type) { |
|
case "NewExpression": |
|
case "CallExpression": |
|
case "OptionalCallExpression": |
|
// Not always necessary, but it's clearer to the reader if IIFEs are wrapped in parentheses. |
|
// Is necessary if it is `expression` of `ExpressionStatement`. |
|
return name === "callee"; |
|
|
|
case "TaggedTemplateExpression": |
|
return true; |
|
// This is basically a kind of IIFE. |
|
|
|
default: |
|
return false; |
|
} |
|
|
|
case "ArrowFunctionExpression": |
|
switch (parent.type) { |
|
case "NewExpression": |
|
case "CallExpression": |
|
case "OptionalCallExpression": |
|
return name === "callee"; |
|
|
|
case "MemberExpression": |
|
case "OptionalMemberExpression": |
|
return name === "object"; |
|
|
|
case "TSAsExpression": |
|
case "BindExpression": |
|
case "TaggedTemplateExpression": |
|
case "UnaryExpression": |
|
case "LogicalExpression": |
|
case "BinaryExpression": |
|
case "AwaitExpression": |
|
case "TSTypeAssertion": |
|
return true; |
|
|
|
case "ConditionalExpression": |
|
return name === "test"; |
|
|
|
default: |
|
return false; |
|
} |
|
|
|
case "ClassExpression": |
|
switch (parent.type) { |
|
case "NewExpression": |
|
return name === "callee" && parent.callee === node; |
|
|
|
default: |
|
return false; |
|
} |
|
|
|
case "OptionalMemberExpression": |
|
case "OptionalCallExpression": |
|
if ((parent.type === "MemberExpression" && name === "object" || parent.type === "CallExpression" && name === "callee") && // workaround for https://github.com/facebook/flow/issues/8159 |
|
!(options.parser === "flow" && parent.range[0] === node.range[0])) { |
|
return true; |
|
} |
|
|
|
// fallthrough |
|
|
|
case "CallExpression": |
|
case "MemberExpression": |
|
case "TaggedTemplateExpression": |
|
case "TSNonNullExpression": |
|
if ((parent.type === "BindExpression" || parent.type === "NewExpression") && name === "callee") { |
|
var object = node; |
|
|
|
while (object) { |
|
switch (object.type) { |
|
case "CallExpression": |
|
case "OptionalCallExpression": |
|
return true; |
|
|
|
case "MemberExpression": |
|
case "OptionalMemberExpression": |
|
case "BindExpression": |
|
object = object.object; |
|
break; |
|
// tagged templates are basically member expressions from a grammar perspective |
|
// see https://tc39.github.io/ecma262/#prod-MemberExpression |
|
|
|
case "TaggedTemplateExpression": |
|
object = object.tag; |
|
break; |
|
|
|
case "TSNonNullExpression": |
|
object = object.expression; |
|
break; |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
} |
|
|
|
return false; |
|
|
|
case "BindExpression": |
|
return (parent.type === "BindExpression" || parent.type === "NewExpression") && name === "callee" || (parent.type === "MemberExpression" || parent.type === "OptionalMemberExpression") && name === "object"; |
|
|
|
case "NGPipeExpression": |
|
if (parent.type === "NGRoot" || parent.type === "NGMicrosyntaxExpression" || parent.type === "ObjectProperty" || parent.type === "ArrayExpression" || (parent.type === "CallExpression" || parent.type === "OptionalCallExpression") && parent.arguments[name] === node || parent.type === "NGPipeExpression" && name === "right" || parent.type === "MemberExpression" && name === "property" || parent.type === "AssignmentExpression") { |
|
return false; |
|
} |
|
|
|
return true; |
|
|
|
case "JSXFragment": |
|
case "JSXElement": |
|
return name === "callee" || parent.type !== "ArrayExpression" && parent.type !== "ArrowFunctionExpression" && parent.type !== "AssignmentExpression" && parent.type !== "AssignmentPattern" && parent.type !== "BinaryExpression" && parent.type !== "CallExpression" && parent.type !== "ConditionalExpression" && parent.type !== "ExpressionStatement" && parent.type !== "JsExpressionRoot" && parent.type !== "JSXAttribute" && parent.type !== "JSXElement" && parent.type !== "JSXExpressionContainer" && parent.type !== "JSXFragment" && parent.type !== "LogicalExpression" && parent.type !== "ObjectProperty" && parent.type !== "OptionalCallExpression" && parent.type !== "Property" && parent.type !== "ReturnStatement" && parent.type !== "TypeCastExpression" && parent.type !== "VariableDeclarator"; |
|
|
|
case "TypeAnnotation": |
|
return name === "returnType" && parent.type === "ArrowFunctionExpression" && includesFunctionTypeInObjectType(node); |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isStatement(node) { |
|
return node.type === "BlockStatement" || node.type === "BreakStatement" || node.type === "ClassBody" || node.type === "ClassDeclaration" || node.type === "ClassMethod" || node.type === "ClassProperty" || node.type === "ClassPrivateProperty" || node.type === "ContinueStatement" || node.type === "DebuggerStatement" || node.type === "DeclareClass" || node.type === "DeclareExportAllDeclaration" || node.type === "DeclareExportDeclaration" || node.type === "DeclareFunction" || node.type === "DeclareInterface" || node.type === "DeclareModule" || node.type === "DeclareModuleExports" || node.type === "DeclareVariable" || node.type === "DoWhileStatement" || node.type === "EnumDeclaration" || node.type === "ExportAllDeclaration" || node.type === "ExportDefaultDeclaration" || node.type === "ExportNamedDeclaration" || node.type === "ExpressionStatement" || node.type === "ForAwaitStatement" || node.type === "ForInStatement" || node.type === "ForOfStatement" || node.type === "ForStatement" || node.type === "FunctionDeclaration" || node.type === "IfStatement" || node.type === "ImportDeclaration" || node.type === "InterfaceDeclaration" || node.type === "LabeledStatement" || node.type === "MethodDefinition" || node.type === "ReturnStatement" || node.type === "SwitchStatement" || node.type === "ThrowStatement" || node.type === "TryStatement" || node.type === "TSDeclareFunction" || node.type === "TSEnumDeclaration" || node.type === "TSImportEqualsDeclaration" || node.type === "TSInterfaceDeclaration" || node.type === "TSModuleDeclaration" || node.type === "TSNamespaceExportDeclaration" || node.type === "TypeAlias" || node.type === "VariableDeclaration" || node.type === "WhileStatement" || node.type === "WithStatement"; |
|
} |
|
|
|
function includesFunctionTypeInObjectType(node) { |
|
return hasNode$1(node, function (n1) { |
|
return n1.type === "ObjectTypeAnnotation" && hasNode$1(n1, function (n2) { |
|
return n2.type === "FunctionTypeAnnotation" || undefined; |
|
}) || undefined; |
|
}); |
|
} |
|
|
|
function endsWithRightBracket(node) { |
|
switch (node.type) { |
|
case "ObjectExpression": |
|
return true; |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function isFollowedByRightBracket(path) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
var name = path.getName(); |
|
|
|
switch (parent.type) { |
|
case "NGPipeExpression": |
|
if (typeof name === "number" && parent.arguments[name] === node && parent.arguments.length - 1 === name) { |
|
return path.callParent(isFollowedByRightBracket); |
|
} |
|
|
|
break; |
|
|
|
case "ObjectProperty": |
|
if (name === "value") { |
|
var parentParent = path.getParentNode(1); |
|
return parentParent.properties[parentParent.properties.length - 1] === parent; |
|
} |
|
|
|
break; |
|
|
|
case "BinaryExpression": |
|
case "LogicalExpression": |
|
if (name === "right") { |
|
return path.callParent(isFollowedByRightBracket); |
|
} |
|
|
|
break; |
|
|
|
case "ConditionalExpression": |
|
if (name === "alternate") { |
|
return path.callParent(isFollowedByRightBracket); |
|
} |
|
|
|
break; |
|
|
|
case "UnaryExpression": |
|
if (parent.prefix) { |
|
return path.callParent(isFollowedByRightBracket); |
|
} |
|
|
|
break; |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function shouldWrapFunctionForExportDefault(path, options) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
|
|
if (node.type === "FunctionExpression" || node.type === "ClassExpression") { |
|
return parent.type === "ExportDefaultDeclaration" || // in some cases the function is already wrapped |
|
// (e.g. `export default (function() {})();`) |
|
// in this case we don't need to add extra parens |
|
!needsParens(path, options); |
|
} |
|
|
|
if (!hasNakedLeftSide$1(node) || parent.type !== "ExportDefaultDeclaration" && needsParens(path, options)) { |
|
return false; |
|
} |
|
|
|
return path.call.apply(path, [function (childPath) { |
|
return shouldWrapFunctionForExportDefault(childPath, options); |
|
}].concat(getLeftSidePathName$1(path, node))); |
|
} |
|
|
|
var needsParens_1 = needsParens; |
|
|
|
var _require$$0$builders$6 = doc.builders, |
|
concat$c = _require$$0$builders$6.concat, |
|
join$8 = _require$$0$builders$6.join, |
|
line$6 = _require$$0$builders$6.line; |
|
|
|
function printHtmlBinding(path, options, print) { |
|
var node = path.getValue(); |
|
|
|
if (options.__onHtmlBindingRoot && path.getName() === null) { |
|
options.__onHtmlBindingRoot(node); |
|
} |
|
|
|
if (node.type !== "File") { |
|
return; |
|
} |
|
|
|
if (options.__isVueForBindingLeft) { |
|
return path.call(function (functionDeclarationPath) { |
|
var _functionDeclarationP = functionDeclarationPath.getValue(), |
|
params = _functionDeclarationP.params; |
|
|
|
return concat$c([params.length > 1 ? "(" : "", join$8(concat$c([",", line$6]), functionDeclarationPath.map(print, "params")), params.length > 1 ? ")" : ""]); |
|
}, "program", "body", 0); |
|
} |
|
|
|
if (options.__isVueSlotScope) { |
|
return path.call(function (functionDeclarationPath) { |
|
return join$8(concat$c([",", line$6]), functionDeclarationPath.map(print, "params")); |
|
}, "program", "body", 0); |
|
} |
|
} // based on https://github.com/prettier/prettier/blob/master/src/language-html/syntax-vue.js isVueEventBindingExpression() |
|
|
|
|
|
function isVueEventBindingExpression$2(node) { |
|
switch (node.type) { |
|
case "MemberExpression": |
|
switch (node.property.type) { |
|
case "Identifier": |
|
case "NumericLiteral": |
|
case "StringLiteral": |
|
return isVueEventBindingExpression$2(node.object); |
|
} |
|
|
|
return false; |
|
|
|
case "Identifier": |
|
return true; |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
var htmlBinding = { |
|
isVueEventBindingExpression: isVueEventBindingExpression$2, |
|
printHtmlBinding: printHtmlBinding |
|
}; |
|
|
|
function preprocess$1(ast, options) { |
|
switch (options.parser) { |
|
case "json": |
|
case "json5": |
|
case "json-stringify": |
|
case "__js_expression": |
|
case "__vue_expression": |
|
return Object.assign({}, ast, { |
|
type: options.parser.startsWith("__") ? "JsExpressionRoot" : "JsonRoot", |
|
node: ast, |
|
comments: [] |
|
}); |
|
|
|
default: |
|
return ast; |
|
} |
|
} |
|
|
|
var preprocess_1$1 = preprocess$1; |
|
|
|
var getParentExportDeclaration$1 = util.getParentExportDeclaration, |
|
isExportDeclaration$1 = util.isExportDeclaration, |
|
shouldFlatten$1 = util.shouldFlatten, |
|
getNextNonSpaceNonCommentCharacter$1 = util.getNextNonSpaceNonCommentCharacter, |
|
hasNewline$4 = util.hasNewline, |
|
hasNewlineInRange$2 = util.hasNewlineInRange, |
|
getLast$2 = util.getLast, |
|
getStringWidth$2 = util.getStringWidth, |
|
printString$2 = util.printString, |
|
printNumber$2 = util.printNumber, |
|
hasIgnoreComment$4 = util.hasIgnoreComment, |
|
hasNodeIgnoreComment$2 = util.hasNodeIgnoreComment, |
|
getPenultimate$1 = util.getPenultimate, |
|
startsWithNoLookaheadToken$1 = util.startsWithNoLookaheadToken, |
|
getIndentSize$1 = util.getIndentSize, |
|
matchAncestorTypes$1 = util.matchAncestorTypes, |
|
getPreferredQuote$1 = util.getPreferredQuote; |
|
var isNextLineEmpty$4 = utilShared.isNextLineEmpty, |
|
isNextLineEmptyAfterIndex$1 = utilShared.isNextLineEmptyAfterIndex, |
|
getNextNonSpaceNonCommentCharacterIndex$2 = utilShared.getNextNonSpaceNonCommentCharacterIndex; |
|
var insertPragma$7 = pragma.insertPragma; |
|
var printHtmlBinding$1 = htmlBinding.printHtmlBinding, |
|
isVueEventBindingExpression$3 = htmlBinding.isVueEventBindingExpression; |
|
var classChildNeedsASIProtection$1 = utils$4.classChildNeedsASIProtection, |
|
classPropMayCauseASIProblems$1 = utils$4.classPropMayCauseASIProblems, |
|
conditionalExpressionChainContainsJSX$1 = utils$4.conditionalExpressionChainContainsJSX, |
|
getFlowVariance$1 = utils$4.getFlowVariance, |
|
getLeftSidePathName$2 = utils$4.getLeftSidePathName, |
|
getTypeScriptMappedTypeModifier$1 = utils$4.getTypeScriptMappedTypeModifier, |
|
hasDanglingComments$1 = utils$4.hasDanglingComments, |
|
hasFlowAnnotationComment$1 = utils$4.hasFlowAnnotationComment, |
|
hasFlowShorthandAnnotationComment$2 = utils$4.hasFlowShorthandAnnotationComment, |
|
hasLeadingComment$3 = utils$4.hasLeadingComment, |
|
hasLeadingOwnLineComment$1 = utils$4.hasLeadingOwnLineComment, |
|
hasNakedLeftSide$2 = utils$4.hasNakedLeftSide, |
|
hasNewlineBetweenOrAfterDecorators$1 = utils$4.hasNewlineBetweenOrAfterDecorators, |
|
hasNgSideEffect$1 = utils$4.hasNgSideEffect, |
|
hasPrettierIgnore$3 = utils$4.hasPrettierIgnore, |
|
hasTrailingComment$1 = utils$4.hasTrailingComment, |
|
identity$2 = utils$4.identity, |
|
isBinaryish$1 = utils$4.isBinaryish, |
|
isCallOrOptionalCallExpression$1 = utils$4.isCallOrOptionalCallExpression, |
|
isEmptyJSXElement$1 = utils$4.isEmptyJSXElement, |
|
isFlowAnnotationComment$1 = utils$4.isFlowAnnotationComment, |
|
isFunctionCompositionArgs$1 = utils$4.isFunctionCompositionArgs, |
|
isFunctionNotation$1 = utils$4.isFunctionNotation, |
|
isFunctionOrArrowExpression$1 = utils$4.isFunctionOrArrowExpression, |
|
isGetterOrSetter$1 = utils$4.isGetterOrSetter, |
|
isJestEachTemplateLiteral$1 = utils$4.isJestEachTemplateLiteral, |
|
isJSXNode$1 = utils$4.isJSXNode, |
|
isJSXWhitespaceExpression$1 = utils$4.isJSXWhitespaceExpression, |
|
isLastStatement$1 = utils$4.isLastStatement, |
|
isLiteral$1 = utils$4.isLiteral, |
|
isLongCurriedCallExpression$1 = utils$4.isLongCurriedCallExpression, |
|
isMeaningfulJSXText$1 = utils$4.isMeaningfulJSXText, |
|
isMemberExpressionChain$1 = utils$4.isMemberExpressionChain, |
|
isMemberish$1 = utils$4.isMemberish, |
|
isNgForOf$1 = utils$4.isNgForOf, |
|
isNumericLiteral$1 = utils$4.isNumericLiteral, |
|
isObjectType$1 = utils$4.isObjectType, |
|
isObjectTypePropertyAFunction$1 = utils$4.isObjectTypePropertyAFunction, |
|
isSimpleFlowType$1 = utils$4.isSimpleFlowType, |
|
isSimpleTemplateLiteral$1 = utils$4.isSimpleTemplateLiteral, |
|
isStringLiteral$1 = utils$4.isStringLiteral, |
|
isStringPropSafeToCoerceToIdentifier$1 = utils$4.isStringPropSafeToCoerceToIdentifier, |
|
isTemplateOnItsOwnLine$1 = utils$4.isTemplateOnItsOwnLine, |
|
isTestCall$1 = utils$4.isTestCall, |
|
isTheOnlyJSXElementInMarkdown$1 = utils$4.isTheOnlyJSXElementInMarkdown, |
|
isTSXFile$1 = utils$4.isTSXFile, |
|
isTypeAnnotationAFunction$1 = utils$4.isTypeAnnotationAFunction, |
|
matchJsxWhitespaceRegex$1 = utils$4.matchJsxWhitespaceRegex, |
|
needsHardlineAfterDanglingComment$1 = utils$4.needsHardlineAfterDanglingComment, |
|
rawText$1 = utils$4.rawText, |
|
returnArgumentHasLeadingComment$1 = utils$4.returnArgumentHasLeadingComment; |
|
var needsQuoteProps = new WeakMap(); |
|
var _require$$5$builders = doc.builders, |
|
concat$d = _require$$5$builders.concat, |
|
join$9 = _require$$5$builders.join, |
|
line$7 = _require$$5$builders.line, |
|
hardline$9 = _require$$5$builders.hardline, |
|
softline$6 = _require$$5$builders.softline, |
|
literalline$4 = _require$$5$builders.literalline, |
|
group$b = _require$$5$builders.group, |
|
indent$7 = _require$$5$builders.indent, |
|
align$1 = _require$$5$builders.align, |
|
conditionalGroup$1 = _require$$5$builders.conditionalGroup, |
|
fill$4 = _require$$5$builders.fill, |
|
ifBreak$6 = _require$$5$builders.ifBreak, |
|
breakParent$3 = _require$$5$builders.breakParent, |
|
lineSuffixBoundary$1 = _require$$5$builders.lineSuffixBoundary, |
|
addAlignmentToDoc$2 = _require$$5$builders.addAlignmentToDoc, |
|
dedent$2 = _require$$5$builders.dedent, |
|
_require$$5$utils = doc.utils, |
|
willBreak$1 = _require$$5$utils.willBreak, |
|
isLineNext$1 = _require$$5$utils.isLineNext, |
|
isEmpty$1 = _require$$5$utils.isEmpty, |
|
removeLines$2 = _require$$5$utils.removeLines, |
|
printDocToString$1 = doc.printer.printDocToString; |
|
var uid = 0; |
|
|
|
function shouldPrintComma$1(options, level) { |
|
level = level || "es5"; |
|
|
|
switch (options.trailingComma) { |
|
case "all": |
|
if (level === "all") { |
|
return true; |
|
} |
|
|
|
// fallthrough |
|
|
|
case "es5": |
|
if (level === "es5") { |
|
return true; |
|
} |
|
|
|
// fallthrough |
|
|
|
case "none": |
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function genericPrint$3(path, options, printPath, args) { |
|
var node = path.getValue(); |
|
var needsParens = false; |
|
var linesWithoutParens = printPathNoParens(path, options, printPath, args); |
|
|
|
if (!node || isEmpty$1(linesWithoutParens)) { |
|
return linesWithoutParens; |
|
} |
|
|
|
var parentExportDecl = getParentExportDeclaration$1(path); |
|
var decorators = []; |
|
|
|
if (node.type === "ClassMethod" || node.type === "ClassPrivateMethod" || node.type === "ClassProperty" || node.type === "TSAbstractClassProperty" || node.type === "ClassPrivateProperty" || node.type === "MethodDefinition" || node.type === "TSAbstractMethodDefinition") ; else if (node.decorators && node.decorators.length > 0 && // If the parent node is an export declaration and the decorator |
|
// was written before the export, the export will be responsible |
|
// for printing the decorators. |
|
!(parentExportDecl && options.locStart(parentExportDecl, { |
|
ignoreDecorators: true |
|
}) > options.locStart(node.decorators[0]))) { |
|
var shouldBreak = node.type === "ClassExpression" || node.type === "ClassDeclaration" || hasNewlineBetweenOrAfterDecorators$1(node, options); |
|
var separator = shouldBreak ? hardline$9 : line$7; |
|
path.each(function (decoratorPath) { |
|
var decorator = decoratorPath.getValue(); |
|
|
|
if (decorator.expression) { |
|
decorator = decorator.expression; |
|
} else { |
|
decorator = decorator.callee; |
|
} |
|
|
|
decorators.push(printPath(decoratorPath), separator); |
|
}, "decorators"); |
|
|
|
if (parentExportDecl) { |
|
decorators.unshift(hardline$9); |
|
} |
|
} else if (isExportDeclaration$1(node) && node.declaration && node.declaration.decorators && node.declaration.decorators.length > 0 && // Only print decorators here if they were written before the export, |
|
// otherwise they are printed by the node.declaration |
|
options.locStart(node, { |
|
ignoreDecorators: true |
|
}) > options.locStart(node.declaration.decorators[0])) { |
|
// Export declarations are responsible for printing any decorators |
|
// that logically apply to node.declaration. |
|
path.each(function (decoratorPath) { |
|
var decorator = decoratorPath.getValue(); |
|
var prefix = decorator.type === "Decorator" ? "" : "@"; |
|
decorators.push(prefix, printPath(decoratorPath), hardline$9); |
|
}, "declaration", "decorators"); |
|
} else { |
|
// Nodes with decorators can't have parentheses, so we can avoid |
|
// computing pathNeedsParens() except in this case. |
|
needsParens = needsParens_1(path, options); |
|
} |
|
|
|
var parts = []; |
|
|
|
if (needsParens) { |
|
parts.unshift("("); |
|
} |
|
|
|
parts.push(linesWithoutParens); |
|
|
|
if (needsParens) { |
|
var _node = path.getValue(); |
|
|
|
if (hasFlowShorthandAnnotationComment$2(_node)) { |
|
parts.push(" /*"); |
|
parts.push(_node.trailingComments[0].value.trimLeft()); |
|
parts.push("*/"); |
|
_node.trailingComments[0].printed = true; |
|
} |
|
|
|
parts.push(")"); |
|
} |
|
|
|
if (decorators.length > 0) { |
|
return group$b(concat$d(decorators.concat(parts))); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
function printDecorators(path, options, print) { |
|
var node = path.getValue(); |
|
return group$b(concat$d([join$9(line$7, path.map(print, "decorators")), hasNewlineBetweenOrAfterDecorators$1(node, options) ? hardline$9 : line$7])); |
|
} |
|
/** |
|
* The following is the shared logic for |
|
* ternary operators, namely ConditionalExpression |
|
* and TSConditionalType |
|
* @typedef {Object} OperatorOptions |
|
* @property {() => Array<string | Doc>} beforeParts - Parts to print before the `?`. |
|
* @property {(breakClosingParen: boolean) => Array<string | Doc>} afterParts - Parts to print after the conditional expression. |
|
* @property {boolean} shouldCheckJsx - Whether to check for and print in JSX mode. |
|
* @property {string} conditionalNodeType - The type of the conditional expression node, ie "ConditionalExpression" or "TSConditionalType". |
|
* @property {string} consequentNodePropertyName - The property at which the consequent node can be found on the main node, eg "consequent". |
|
* @property {string} alternateNodePropertyName - The property at which the alternate node can be found on the main node, eg "alternate". |
|
* @property {string} testNodePropertyName - The property at which the test node can be found on the main node, eg "test". |
|
* @property {boolean} breakNested - Whether to break all nested ternaries when one breaks. |
|
* @param {FastPath} path - The path to the ConditionalExpression/TSConditionalType node. |
|
* @param {Options} options - Prettier options |
|
* @param {Function} print - Print function to call recursively |
|
* @param {OperatorOptions} operatorOptions |
|
* @returns Doc |
|
*/ |
|
|
|
|
|
function printTernaryOperator(path, options, print, operatorOptions) { |
|
var node = path.getValue(); |
|
var testNode = node[operatorOptions.testNodePropertyName]; |
|
var consequentNode = node[operatorOptions.consequentNodePropertyName]; |
|
var alternateNode = node[operatorOptions.alternateNodePropertyName]; |
|
var parts = []; // We print a ConditionalExpression in either "JSX mode" or "normal mode". |
|
// See tests/jsx/conditional-expression.js for more info. |
|
|
|
var jsxMode = false; |
|
var parent = path.getParentNode(); |
|
var forceNoIndent = parent.type === operatorOptions.conditionalNodeType; // Find the outermost non-ConditionalExpression parent, and the outermost |
|
// ConditionalExpression parent. We'll use these to determine if we should |
|
// print in JSX mode. |
|
|
|
var currentParent; |
|
var previousParent; |
|
var i = 0; |
|
|
|
do { |
|
previousParent = currentParent || node; |
|
currentParent = path.getParentNode(i); |
|
i++; |
|
} while (currentParent && currentParent.type === operatorOptions.conditionalNodeType); |
|
|
|
var firstNonConditionalParent = currentParent || parent; |
|
var lastConditionalParent = previousParent; |
|
|
|
if (operatorOptions.shouldCheckJsx && (isJSXNode$1(testNode) || isJSXNode$1(consequentNode) || isJSXNode$1(alternateNode) || conditionalExpressionChainContainsJSX$1(lastConditionalParent))) { |
|
jsxMode = true; |
|
forceNoIndent = true; // Even though they don't need parens, we wrap (almost) everything in |
|
// parens when using ?: within JSX, because the parens are analogous to |
|
// curly braces in an if statement. |
|
|
|
var wrap = function wrap(doc) { |
|
return concat$d([ifBreak$6("(", ""), indent$7(concat$d([softline$6, doc])), softline$6, ifBreak$6(")", "")]); |
|
}; // The only things we don't wrap are: |
|
// * Nested conditional expressions in alternates |
|
// * null |
|
|
|
|
|
var isNull = function isNull(node) { |
|
return node.type === "NullLiteral" || node.type === "Literal" && node.value === null; |
|
}; |
|
|
|
parts.push(" ? ", isNull(consequentNode) ? path.call(print, operatorOptions.consequentNodePropertyName) : wrap(path.call(print, operatorOptions.consequentNodePropertyName)), " : ", alternateNode.type === operatorOptions.conditionalNodeType || isNull(alternateNode) ? path.call(print, operatorOptions.alternateNodePropertyName) : wrap(path.call(print, operatorOptions.alternateNodePropertyName))); |
|
} else { |
|
// normal mode |
|
var part = concat$d([line$7, "? ", consequentNode.type === operatorOptions.conditionalNodeType ? ifBreak$6("", "(") : "", align$1(2, path.call(print, operatorOptions.consequentNodePropertyName)), consequentNode.type === operatorOptions.conditionalNodeType ? ifBreak$6("", ")") : "", line$7, ": ", alternateNode.type === operatorOptions.conditionalNodeType ? path.call(print, operatorOptions.alternateNodePropertyName) : align$1(2, path.call(print, operatorOptions.alternateNodePropertyName))]); |
|
parts.push(parent.type !== operatorOptions.conditionalNodeType || parent[operatorOptions.alternateNodePropertyName] === node ? part : options.useTabs ? dedent$2(indent$7(part)) : align$1(Math.max(0, options.tabWidth - 2), part)); |
|
} // We want a whole chain of ConditionalExpressions to all |
|
// break if any of them break. That means we should only group around the |
|
// outer-most ConditionalExpression. |
|
|
|
|
|
var maybeGroup = function maybeGroup(doc) { |
|
return operatorOptions.breakNested ? parent === firstNonConditionalParent ? group$b(doc) : doc : group$b(doc); |
|
}; // Break the closing paren to keep the chain right after it: |
|
// (a |
|
// ? b |
|
// : c |
|
// ).call() |
|
|
|
|
|
var breakClosingParen = !jsxMode && (parent.type === "MemberExpression" || parent.type === "OptionalMemberExpression" || parent.type === "NGPipeExpression" && parent.left === node && operatorOptions.breakNested) && !parent.computed; |
|
return maybeGroup(concat$d([].concat(function (testDoc) { |
|
return ( |
|
/** |
|
* a |
|
* ? b |
|
* : multiline |
|
* test |
|
* node |
|
* ^^ align(2) |
|
* ? d |
|
* : e |
|
*/ |
|
parent.type === operatorOptions.conditionalNodeType && parent[operatorOptions.alternateNodePropertyName] === node ? align$1(2, testDoc) : testDoc |
|
); |
|
}(concat$d(operatorOptions.beforeParts())), forceNoIndent ? concat$d(parts) : indent$7(concat$d(parts)), operatorOptions.afterParts(breakClosingParen)))); |
|
} |
|
|
|
function printPathNoParens(path, options, print, args) { |
|
var n = path.getValue(); |
|
var semi = options.semi ? ";" : ""; |
|
|
|
if (!n) { |
|
return ""; |
|
} |
|
|
|
if (typeof n === "string") { |
|
return n; |
|
} |
|
|
|
var htmlBinding = printHtmlBinding$1(path, options, print); |
|
|
|
if (htmlBinding) { |
|
return htmlBinding; |
|
} |
|
|
|
var parts = []; |
|
|
|
switch (n.type) { |
|
case "JsExpressionRoot": |
|
return path.call(print, "node"); |
|
|
|
case "JsonRoot": |
|
return concat$d([path.call(print, "node"), hardline$9]); |
|
|
|
case "File": |
|
// Print @babel/parser's InterpreterDirective here so that |
|
// leading comments on the `Program` node get printed after the hashbang. |
|
if (n.program && n.program.interpreter) { |
|
parts.push(path.call(function (programPath) { |
|
return programPath.call(print, "interpreter"); |
|
}, "program")); |
|
} |
|
|
|
parts.push(path.call(print, "program")); |
|
return concat$d(parts); |
|
|
|
case "Program": |
|
// Babel 6 |
|
if (n.directives) { |
|
path.each(function (childPath) { |
|
parts.push(print(childPath), semi, hardline$9); |
|
|
|
if (isNextLineEmpty$4(options.originalText, childPath.getValue(), options)) { |
|
parts.push(hardline$9); |
|
} |
|
}, "directives"); |
|
} |
|
|
|
parts.push(path.call(function (bodyPath) { |
|
return printStatementSequence(bodyPath, options, print); |
|
}, "body")); |
|
parts.push(comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true)); // Only force a trailing newline if there were any contents. |
|
|
|
if (n.body.length || n.comments) { |
|
parts.push(hardline$9); |
|
} |
|
|
|
return concat$d(parts); |
|
// Babel extension. |
|
|
|
case "EmptyStatement": |
|
return ""; |
|
|
|
case "ExpressionStatement": |
|
// Detect Flow-parsed directives |
|
if (n.directive) { |
|
return concat$d([nodeStr(n.expression, options, true), semi]); |
|
} |
|
|
|
if (options.parser === "__vue_event_binding") { |
|
var parent = path.getParentNode(); |
|
|
|
if (parent.type === "Program" && parent.body.length === 1 && parent.body[0] === n) { |
|
return concat$d([path.call(print, "expression"), isVueEventBindingExpression$3(n.expression) ? ";" : ""]); |
|
} |
|
} // Do not append semicolon after the only JSX element in a program |
|
|
|
|
|
return concat$d([path.call(print, "expression"), isTheOnlyJSXElementInMarkdown$1(options, path) ? "" : semi]); |
|
// Babel extension. |
|
|
|
case "ParenthesizedExpression": |
|
return concat$d(["(", path.call(print, "expression"), ")"]); |
|
|
|
case "AssignmentExpression": |
|
return printAssignment(n.left, path.call(print, "left"), concat$d([" ", n.operator]), n.right, path.call(print, "right"), options); |
|
|
|
case "BinaryExpression": |
|
case "LogicalExpression": |
|
case "NGPipeExpression": |
|
{ |
|
var _parent = path.getParentNode(); |
|
|
|
var parentParent = path.getParentNode(1); |
|
var isInsideParenthesis = n !== _parent.body && (_parent.type === "IfStatement" || _parent.type === "WhileStatement" || _parent.type === "SwitchStatement" || _parent.type === "DoWhileStatement"); |
|
|
|
var _parts = printBinaryishExpressions(path, print, options, |
|
/* isNested */ |
|
false, isInsideParenthesis); // if ( |
|
// this.hasPlugin("dynamicImports") && this.lookahead().type === tt.parenLeft |
|
// ) { |
|
// |
|
// looks super weird, we want to break the children if the parent breaks |
|
// |
|
// if ( |
|
// this.hasPlugin("dynamicImports") && |
|
// this.lookahead().type === tt.parenLeft |
|
// ) { |
|
|
|
|
|
if (isInsideParenthesis) { |
|
return concat$d(_parts); |
|
} // Break between the parens in |
|
// unaries or in a member or specific call expression, i.e. |
|
// |
|
// ( |
|
// a && |
|
// b && |
|
// c |
|
// ).call() |
|
|
|
|
|
if ((_parent.type === "CallExpression" || _parent.type === "OptionalCallExpression") && _parent.callee === n || _parent.type === "UnaryExpression" || (_parent.type === "MemberExpression" || _parent.type === "OptionalMemberExpression") && !_parent.computed) { |
|
return group$b(concat$d([indent$7(concat$d([softline$6, concat$d(_parts)])), softline$6])); |
|
} // Avoid indenting sub-expressions in some cases where the first sub-expression is already |
|
// indented accordingly. We should indent sub-expressions where the first case isn't indented. |
|
|
|
|
|
var shouldNotIndent = _parent.type === "ReturnStatement" || _parent.type === "JSXExpressionContainer" && parentParent.type === "JSXAttribute" || n.type !== "NGPipeExpression" && (_parent.type === "NGRoot" && options.parser === "__ng_binding" || _parent.type === "NGMicrosyntaxExpression" && parentParent.type === "NGMicrosyntax" && parentParent.body.length === 1) || n === _parent.body && _parent.type === "ArrowFunctionExpression" || n !== _parent.body && _parent.type === "ForStatement" || _parent.type === "ConditionalExpression" && parentParent.type !== "ReturnStatement" && parentParent.type !== "CallExpression" && parentParent.type !== "OptionalCallExpression"; |
|
var shouldIndentIfInlining = _parent.type === "AssignmentExpression" || _parent.type === "VariableDeclarator" || _parent.type === "ClassProperty" || _parent.type === "TSAbstractClassProperty" || _parent.type === "ClassPrivateProperty" || _parent.type === "ObjectProperty" || _parent.type === "Property"; |
|
var samePrecedenceSubExpression = isBinaryish$1(n.left) && shouldFlatten$1(n.operator, n.left.operator); |
|
|
|
if (shouldNotIndent || shouldInlineLogicalExpression(n) && !samePrecedenceSubExpression || !shouldInlineLogicalExpression(n) && shouldIndentIfInlining) { |
|
return group$b(concat$d(_parts)); |
|
} |
|
|
|
if (_parts.length === 0) { |
|
return ""; |
|
} // If the right part is a JSX node, we include it in a separate group to |
|
// prevent it breaking the whole chain, so we can print the expression like: |
|
// |
|
// foo && bar && ( |
|
// <Foo> |
|
// <Bar /> |
|
// </Foo> |
|
// ) |
|
|
|
|
|
var hasJSX = isJSXNode$1(n.right); |
|
var rest = concat$d(hasJSX ? _parts.slice(1, -1) : _parts.slice(1)); |
|
var groupId = Symbol("logicalChain-" + ++uid); |
|
var chain = group$b(concat$d([// Don't include the initial expression in the indentation |
|
// level. The first item is guaranteed to be the first |
|
// left-most expression. |
|
_parts.length > 0 ? _parts[0] : "", indent$7(rest)]), { |
|
id: groupId |
|
}); |
|
|
|
if (!hasJSX) { |
|
return chain; |
|
} |
|
|
|
var jsxPart = getLast$2(_parts); |
|
return group$b(concat$d([chain, ifBreak$6(indent$7(jsxPart), jsxPart, { |
|
groupId: groupId |
|
})])); |
|
} |
|
|
|
case "AssignmentPattern": |
|
return concat$d([path.call(print, "left"), " = ", path.call(print, "right")]); |
|
|
|
case "TSTypeAssertion": |
|
{ |
|
var shouldBreakAfterCast = !(n.expression.type === "ArrayExpression" || n.expression.type === "ObjectExpression"); |
|
var castGroup = group$b(concat$d(["<", indent$7(concat$d([softline$6, path.call(print, "typeAnnotation")])), softline$6, ">"])); |
|
var exprContents = concat$d([ifBreak$6("("), indent$7(concat$d([softline$6, path.call(print, "expression")])), softline$6, ifBreak$6(")")]); |
|
|
|
if (shouldBreakAfterCast) { |
|
return conditionalGroup$1([concat$d([castGroup, path.call(print, "expression")]), concat$d([castGroup, group$b(exprContents, { |
|
shouldBreak: true |
|
})]), concat$d([castGroup, path.call(print, "expression")])]); |
|
} |
|
|
|
return group$b(concat$d([castGroup, path.call(print, "expression")])); |
|
} |
|
|
|
case "OptionalMemberExpression": |
|
case "MemberExpression": |
|
{ |
|
var _parent2 = path.getParentNode(); |
|
|
|
var firstNonMemberParent; |
|
var i = 0; |
|
|
|
do { |
|
firstNonMemberParent = path.getParentNode(i); |
|
i++; |
|
} while (firstNonMemberParent && (firstNonMemberParent.type === "MemberExpression" || firstNonMemberParent.type === "OptionalMemberExpression" || firstNonMemberParent.type === "TSNonNullExpression")); |
|
|
|
var shouldInline = firstNonMemberParent && (firstNonMemberParent.type === "NewExpression" || firstNonMemberParent.type === "BindExpression" || firstNonMemberParent.type === "VariableDeclarator" && firstNonMemberParent.id.type !== "Identifier" || firstNonMemberParent.type === "AssignmentExpression" && firstNonMemberParent.left.type !== "Identifier") || n.computed || n.object.type === "Identifier" && n.property.type === "Identifier" && _parent2.type !== "MemberExpression" && _parent2.type !== "OptionalMemberExpression"; |
|
return concat$d([path.call(print, "object"), shouldInline ? printMemberLookup(path, options, print) : group$b(indent$7(concat$d([softline$6, printMemberLookup(path, options, print)])))]); |
|
} |
|
|
|
case "MetaProperty": |
|
return concat$d([path.call(print, "meta"), ".", path.call(print, "property")]); |
|
|
|
case "BindExpression": |
|
if (n.object) { |
|
parts.push(path.call(print, "object")); |
|
} |
|
|
|
parts.push(group$b(indent$7(concat$d([softline$6, printBindExpressionCallee(path, options, print)])))); |
|
return concat$d(parts); |
|
|
|
case "Identifier": |
|
{ |
|
return concat$d([n.name, printOptionalToken(path), printTypeAnnotation(path, options, print)]); |
|
} |
|
|
|
case "V8IntrinsicIdentifier": |
|
return concat$d(["%", n.name]); |
|
|
|
case "SpreadElement": |
|
case "SpreadElementPattern": |
|
case "RestProperty": |
|
case "SpreadProperty": |
|
case "SpreadPropertyPattern": |
|
case "RestElement": |
|
case "ObjectTypeSpreadProperty": |
|
return concat$d(["...", path.call(print, "argument"), printTypeAnnotation(path, options, print)]); |
|
|
|
case "FunctionDeclaration": |
|
case "FunctionExpression": |
|
parts.push(printFunctionDeclaration(path, print, options)); |
|
|
|
if (!n.body) { |
|
parts.push(semi); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "ArrowFunctionExpression": |
|
{ |
|
if (n.async) { |
|
parts.push("async "); |
|
} |
|
|
|
if (shouldPrintParamsWithoutParens(path, options)) { |
|
parts.push(path.call(print, "params", 0)); |
|
} else { |
|
parts.push(group$b(concat$d([printFunctionParams(path, print, options, |
|
/* expandLast */ |
|
args && (args.expandLastArg || args.expandFirstArg), |
|
/* printTypeParams */ |
|
true), printReturnType(path, print, options)]))); |
|
} |
|
|
|
var dangling = comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true, function (comment) { |
|
var nextCharacter = getNextNonSpaceNonCommentCharacterIndex$2(options.originalText, comment, options); |
|
return options.originalText.substr(nextCharacter, 2) === "=>"; |
|
}); |
|
|
|
if (dangling) { |
|
parts.push(" ", dangling); |
|
} |
|
|
|
parts.push(" =>"); |
|
var body = path.call(function (bodyPath) { |
|
return print(bodyPath, args); |
|
}, "body"); // We want to always keep these types of nodes on the same line |
|
// as the arrow. |
|
|
|
if (!hasLeadingOwnLineComment$1(options.originalText, n.body, options) && (n.body.type === "ArrayExpression" || n.body.type === "ObjectExpression" || n.body.type === "BlockStatement" || isJSXNode$1(n.body) || isTemplateOnItsOwnLine$1(n.body, options.originalText, options) || n.body.type === "ArrowFunctionExpression" || n.body.type === "DoExpression")) { |
|
return group$b(concat$d([concat$d(parts), " ", body])); |
|
} // We handle sequence expressions as the body of arrows specially, |
|
// so that the required parentheses end up on their own lines. |
|
|
|
|
|
if (n.body.type === "SequenceExpression") { |
|
return group$b(concat$d([concat$d(parts), group$b(concat$d([" (", indent$7(concat$d([softline$6, body])), softline$6, ")"]))])); |
|
} // if the arrow function is expanded as last argument, we are adding a |
|
// level of indentation and need to add a softline to align the closing ) |
|
// with the opening (, or if it's inside a JSXExpression (e.g. an attribute) |
|
// we should align the expression's closing } with the line with the opening {. |
|
|
|
|
|
var shouldAddSoftLine = (args && args.expandLastArg || path.getParentNode().type === "JSXExpressionContainer") && !(n.comments && n.comments.length); |
|
var printTrailingComma = args && args.expandLastArg && shouldPrintComma$1(options, "all"); // In order to avoid confusion between |
|
// a => a ? a : a |
|
// a <= a ? a : a |
|
|
|
var shouldAddParens = n.body.type === "ConditionalExpression" && !startsWithNoLookaheadToken$1(n.body, |
|
/* forbidFunctionAndClass */ |
|
false); |
|
return group$b(concat$d([concat$d(parts), group$b(concat$d([indent$7(concat$d([line$7, shouldAddParens ? ifBreak$6("", "(") : "", body, shouldAddParens ? ifBreak$6("", ")") : ""])), shouldAddSoftLine ? concat$d([ifBreak$6(printTrailingComma ? "," : ""), softline$6]) : ""]))])); |
|
} |
|
|
|
case "YieldExpression": |
|
parts.push("yield"); |
|
|
|
if (n.delegate) { |
|
parts.push("*"); |
|
} |
|
|
|
if (n.argument) { |
|
parts.push(" ", path.call(print, "argument")); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "AwaitExpression": |
|
{ |
|
parts.push("await ", path.call(print, "argument")); |
|
|
|
var _parent3 = path.getParentNode(); |
|
|
|
if ((_parent3.type === "CallExpression" || _parent3.type === "OptionalCallExpression") && _parent3.callee === n || (_parent3.type === "MemberExpression" || _parent3.type === "OptionalMemberExpression") && _parent3.object === n) { |
|
return group$b(concat$d([indent$7(concat$d([softline$6, concat$d(parts)])), softline$6])); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
case "ImportSpecifier": |
|
if (n.importKind) { |
|
parts.push(path.call(print, "importKind"), " "); |
|
} |
|
|
|
parts.push(path.call(print, "imported")); |
|
|
|
if (n.local && n.local.name !== n.imported.name) { |
|
parts.push(" as ", path.call(print, "local")); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "ExportSpecifier": |
|
parts.push(path.call(print, "local")); |
|
|
|
if (n.exported && n.exported.name !== n.local.name) { |
|
parts.push(" as ", path.call(print, "exported")); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "ImportNamespaceSpecifier": |
|
parts.push("* as "); |
|
parts.push(path.call(print, "local")); |
|
return concat$d(parts); |
|
|
|
case "ImportDefaultSpecifier": |
|
return path.call(print, "local"); |
|
|
|
case "TSExportAssignment": |
|
return concat$d(["export = ", path.call(print, "expression"), semi]); |
|
|
|
case "ExportDefaultDeclaration": |
|
case "ExportNamedDeclaration": |
|
return printExportDeclaration(path, options, print); |
|
|
|
case "ExportAllDeclaration": |
|
parts.push("export "); |
|
|
|
if (n.exportKind === "type") { |
|
parts.push("type "); |
|
} |
|
|
|
parts.push("* from ", path.call(print, "source"), semi); |
|
return concat$d(parts); |
|
|
|
case "ExportNamespaceSpecifier": |
|
case "ExportDefaultSpecifier": |
|
return path.call(print, "exported"); |
|
|
|
case "ImportDeclaration": |
|
{ |
|
parts.push("import "); |
|
|
|
if (n.importKind && n.importKind !== "value") { |
|
parts.push(n.importKind + " "); |
|
} |
|
|
|
var standalones = []; |
|
var grouped = []; |
|
|
|
if (n.specifiers && n.specifiers.length > 0) { |
|
path.each(function (specifierPath) { |
|
var value = specifierPath.getValue(); |
|
|
|
if (value.type === "ImportDefaultSpecifier" || value.type === "ImportNamespaceSpecifier") { |
|
standalones.push(print(specifierPath)); |
|
} else { |
|
grouped.push(print(specifierPath)); |
|
} |
|
}, "specifiers"); |
|
|
|
if (standalones.length > 0) { |
|
parts.push(join$9(", ", standalones)); |
|
} |
|
|
|
if (standalones.length > 0 && grouped.length > 0) { |
|
parts.push(", "); |
|
} |
|
|
|
if (grouped.length === 1 && standalones.length === 0 && n.specifiers && !n.specifiers.some(function (node) { |
|
return node.comments; |
|
})) { |
|
parts.push(concat$d(["{", options.bracketSpacing ? " " : "", concat$d(grouped), options.bracketSpacing ? " " : "", "}"])); |
|
} else if (grouped.length >= 1) { |
|
parts.push(group$b(concat$d(["{", indent$7(concat$d([options.bracketSpacing ? line$7 : softline$6, join$9(concat$d([",", line$7]), grouped)])), ifBreak$6(shouldPrintComma$1(options) ? "," : ""), options.bracketSpacing ? line$7 : softline$6, "}"]))); |
|
} |
|
|
|
parts.push(" from "); |
|
} else if (n.importKind && n.importKind === "type" || // import {} from 'x' |
|
/{\s*}/.test(options.originalText.slice(options.locStart(n), options.locStart(n.source)))) { |
|
parts.push("{} from "); |
|
} |
|
|
|
parts.push(path.call(print, "source"), semi); |
|
return concat$d(parts); |
|
} |
|
|
|
case "Import": |
|
return "import"; |
|
|
|
case "TSModuleBlock": |
|
case "BlockStatement": |
|
{ |
|
var naked = path.call(function (bodyPath) { |
|
return printStatementSequence(bodyPath, options, print); |
|
}, "body"); |
|
var hasContent = n.body.find(function (node) { |
|
return node.type !== "EmptyStatement"; |
|
}); |
|
var hasDirectives = n.directives && n.directives.length > 0; |
|
|
|
var _parent4 = path.getParentNode(); |
|
|
|
var _parentParent = path.getParentNode(1); |
|
|
|
if (!hasContent && !hasDirectives && !hasDanglingComments$1(n) && (_parent4.type === "ArrowFunctionExpression" || _parent4.type === "FunctionExpression" || _parent4.type === "FunctionDeclaration" || _parent4.type === "ObjectMethod" || _parent4.type === "ClassMethod" || _parent4.type === "ClassPrivateMethod" || _parent4.type === "ForStatement" || _parent4.type === "WhileStatement" || _parent4.type === "DoWhileStatement" || _parent4.type === "DoExpression" || _parent4.type === "CatchClause" && !_parentParent.finalizer || _parent4.type === "TSModuleDeclaration")) { |
|
return "{}"; |
|
} |
|
|
|
parts.push("{"); // Babel 6 |
|
|
|
if (hasDirectives) { |
|
path.each(function (childPath) { |
|
parts.push(indent$7(concat$d([hardline$9, print(childPath), semi]))); |
|
|
|
if (isNextLineEmpty$4(options.originalText, childPath.getValue(), options)) { |
|
parts.push(hardline$9); |
|
} |
|
}, "directives"); |
|
} |
|
|
|
if (hasContent) { |
|
parts.push(indent$7(concat$d([hardline$9, naked]))); |
|
} |
|
|
|
parts.push(comments.printDanglingComments(path, options)); |
|
parts.push(hardline$9, "}"); |
|
return concat$d(parts); |
|
} |
|
|
|
case "ReturnStatement": |
|
parts.push("return"); |
|
|
|
if (n.argument) { |
|
if (returnArgumentHasLeadingComment$1(options, n.argument)) { |
|
parts.push(concat$d([" (", indent$7(concat$d([hardline$9, path.call(print, "argument")])), hardline$9, ")"])); |
|
} else if (n.argument.type === "LogicalExpression" || n.argument.type === "BinaryExpression" || n.argument.type === "SequenceExpression") { |
|
parts.push(group$b(concat$d([ifBreak$6(" (", " "), indent$7(concat$d([softline$6, path.call(print, "argument")])), softline$6, ifBreak$6(")")]))); |
|
} else { |
|
parts.push(" ", path.call(print, "argument")); |
|
} |
|
} |
|
|
|
if (hasDanglingComments$1(n)) { |
|
parts.push(" ", comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true)); |
|
} |
|
|
|
parts.push(semi); |
|
return concat$d(parts); |
|
|
|
case "NewExpression": |
|
case "OptionalCallExpression": |
|
case "CallExpression": |
|
{ |
|
var isNew = n.type === "NewExpression"; |
|
var optional = printOptionalToken(path); |
|
|
|
if ( // We want to keep CommonJS- and AMD-style require calls, and AMD-style |
|
// define calls, as a unit. |
|
// e.g. `define(["some/lib", (lib) => {` |
|
!isNew && n.callee.type === "Identifier" && (n.callee.name === "require" || n.callee.name === "define") || // Template literals as single arguments |
|
n.arguments.length === 1 && isTemplateOnItsOwnLine$1(n.arguments[0], options.originalText, options) || // Keep test declarations on a single line |
|
// e.g. `it('long name', () => {` |
|
!isNew && isTestCall$1(n, path.getParentNode())) { |
|
return concat$d([isNew ? "new " : "", path.call(print, "callee"), optional, printFunctionTypeParameters(path, options, print), concat$d(["(", join$9(", ", path.map(print, "arguments")), ")"])]); |
|
} // Inline Flow annotation comments following Identifiers in Call nodes need to |
|
// stay with the Identifier. For example: |
|
// |
|
// foo /*:: <SomeGeneric> */(bar); |
|
// |
|
// Here, we ensure that such comments stay between the Identifier and the Callee. |
|
|
|
|
|
var isIdentifierWithFlowAnnotation = n.callee.type === "Identifier" && hasFlowAnnotationComment$1(n.callee.trailingComments); |
|
|
|
if (isIdentifierWithFlowAnnotation) { |
|
n.callee.trailingComments[0].printed = true; |
|
} // We detect calls on member lookups and possibly print them in a |
|
// special chain format. See `printMemberChain` for more info. |
|
|
|
|
|
if (!isNew && isMemberish$1(n.callee)) { |
|
return printMemberChain(path, options, print); |
|
} |
|
|
|
var contents = concat$d([isNew ? "new " : "", path.call(print, "callee"), optional, isIdentifierWithFlowAnnotation ? "/*:: ".concat(n.callee.trailingComments[0].value.substring(2).trim(), " */") : "", printFunctionTypeParameters(path, options, print), printArgumentsList(path, options, print)]); // We group here when the callee is itself a call expression. |
|
// See `isLongCurriedCallExpression` for more info. |
|
|
|
if (isCallOrOptionalCallExpression$1(n.callee)) { |
|
return group$b(contents); |
|
} |
|
|
|
return contents; |
|
} |
|
|
|
case "TSInterfaceDeclaration": |
|
if (n.declare) { |
|
parts.push("declare "); |
|
} |
|
|
|
parts.push(n.abstract ? "abstract " : "", printTypeScriptModifiers(path, options, print), "interface ", path.call(print, "id"), n.typeParameters ? path.call(print, "typeParameters") : "", " "); |
|
|
|
if (n.extends && n.extends.length) { |
|
parts.push(group$b(indent$7(concat$d([softline$6, "extends ", (n.extends.length === 1 ? identity$2 : indent$7)(join$9(concat$d([",", line$7]), path.map(print, "extends"))), " "])))); |
|
} |
|
|
|
parts.push(path.call(print, "body")); |
|
return concat$d(parts); |
|
|
|
case "ObjectTypeInternalSlot": |
|
return concat$d([n.static ? "static " : "", "[[", path.call(print, "id"), "]]", printOptionalToken(path), n.method ? "" : ": ", path.call(print, "value")]); |
|
|
|
case "ObjectExpression": |
|
case "ObjectPattern": |
|
case "ObjectTypeAnnotation": |
|
case "TSInterfaceBody": |
|
case "TSTypeLiteral": |
|
{ |
|
var propertiesField; |
|
|
|
if (n.type === "TSTypeLiteral") { |
|
propertiesField = "members"; |
|
} else if (n.type === "TSInterfaceBody") { |
|
propertiesField = "body"; |
|
} else { |
|
propertiesField = "properties"; |
|
} |
|
|
|
var isTypeAnnotation = n.type === "ObjectTypeAnnotation"; |
|
var fields = []; |
|
|
|
if (isTypeAnnotation) { |
|
fields.push("indexers", "callProperties", "internalSlots"); |
|
} |
|
|
|
fields.push(propertiesField); |
|
var firstProperty = fields.map(function (field) { |
|
return n[field][0]; |
|
}).sort(function (a, b) { |
|
return options.locStart(a) - options.locStart(b); |
|
})[0]; |
|
|
|
var _parent5 = path.getParentNode(0); |
|
|
|
var isFlowInterfaceLikeBody = isTypeAnnotation && _parent5 && (_parent5.type === "InterfaceDeclaration" || _parent5.type === "DeclareInterface" || _parent5.type === "DeclareClass") && path.getName() === "body"; |
|
var shouldBreak = n.type === "TSInterfaceBody" || isFlowInterfaceLikeBody || n.type === "ObjectPattern" && _parent5.type !== "FunctionDeclaration" && _parent5.type !== "FunctionExpression" && _parent5.type !== "ArrowFunctionExpression" && _parent5.type !== "ObjectMethod" && _parent5.type !== "ClassMethod" && _parent5.type !== "ClassPrivateMethod" && _parent5.type !== "AssignmentPattern" && _parent5.type !== "CatchClause" && n.properties.some(function (property) { |
|
return property.value && (property.value.type === "ObjectPattern" || property.value.type === "ArrayPattern"); |
|
}) || n.type !== "ObjectPattern" && firstProperty && hasNewlineInRange$2(options.originalText, options.locStart(n), options.locStart(firstProperty)); |
|
var separator = isFlowInterfaceLikeBody ? ";" : n.type === "TSInterfaceBody" || n.type === "TSTypeLiteral" ? ifBreak$6(semi, ";") : ","; |
|
var leftBrace = n.exact ? "{|" : "{"; |
|
var rightBrace = n.exact ? "|}" : "}"; // Unfortunately, things are grouped together in the ast can be |
|
// interleaved in the source code. So we need to reorder them before |
|
// printing them. |
|
|
|
var propsAndLoc = []; |
|
fields.forEach(function (field) { |
|
path.each(function (childPath) { |
|
var node = childPath.getValue(); |
|
propsAndLoc.push({ |
|
node: node, |
|
printed: print(childPath), |
|
loc: options.locStart(node) |
|
}); |
|
}, field); |
|
}); |
|
var separatorParts = []; |
|
var props = propsAndLoc.sort(function (a, b) { |
|
return a.loc - b.loc; |
|
}).map(function (prop) { |
|
var result = concat$d(separatorParts.concat(group$b(prop.printed))); |
|
separatorParts = [separator, line$7]; |
|
|
|
if ((prop.node.type === "TSPropertySignature" || prop.node.type === "TSMethodSignature" || prop.node.type === "TSConstructSignatureDeclaration") && hasNodeIgnoreComment$2(prop.node)) { |
|
separatorParts.shift(); |
|
} |
|
|
|
if (isNextLineEmpty$4(options.originalText, prop.node, options)) { |
|
separatorParts.push(hardline$9); |
|
} |
|
|
|
return result; |
|
}); |
|
|
|
if (n.inexact) { |
|
props.push(concat$d(separatorParts.concat(group$b("...")))); |
|
} |
|
|
|
var lastElem = getLast$2(n[propertiesField]); |
|
var canHaveTrailingSeparator = !(lastElem && (lastElem.type === "RestProperty" || lastElem.type === "RestElement" || hasNodeIgnoreComment$2(lastElem) || n.inexact)); |
|
var content; |
|
|
|
if (props.length === 0) { |
|
if (!hasDanglingComments$1(n)) { |
|
return concat$d([leftBrace, rightBrace, printTypeAnnotation(path, options, print)]); |
|
} |
|
|
|
content = group$b(concat$d([leftBrace, comments.printDanglingComments(path, options), softline$6, rightBrace, printOptionalToken(path), printTypeAnnotation(path, options, print)])); |
|
} else { |
|
content = concat$d([leftBrace, indent$7(concat$d([options.bracketSpacing ? line$7 : softline$6, concat$d(props)])), ifBreak$6(canHaveTrailingSeparator && (separator !== "," || shouldPrintComma$1(options)) ? separator : ""), concat$d([options.bracketSpacing ? line$7 : softline$6, rightBrace]), printOptionalToken(path), printTypeAnnotation(path, options, print)]); |
|
} // If we inline the object as first argument of the parent, we don't want |
|
// to create another group so that the object breaks before the return |
|
// type |
|
|
|
|
|
var parentParentParent = path.getParentNode(2); |
|
|
|
if (n.type === "ObjectPattern" && _parent5 && shouldHugArguments(_parent5) && !n.decorators && _parent5.params[0] === n || shouldHugType(n) && parentParentParent && shouldHugArguments(parentParentParent) && parentParentParent.params[0].typeAnnotation && parentParentParent.params[0].typeAnnotation.typeAnnotation === n) { |
|
return content; |
|
} |
|
|
|
return group$b(content, { |
|
shouldBreak: shouldBreak |
|
}); |
|
} |
|
// Babel 6 |
|
|
|
case "ObjectProperty": // Non-standard AST node type. |
|
|
|
case "Property": |
|
if (n.method || n.kind === "get" || n.kind === "set") { |
|
return printMethod(path, options, print); |
|
} |
|
|
|
if (n.shorthand) { |
|
parts.push(path.call(print, "value")); |
|
} else { |
|
parts.push(printAssignment(n.key, printPropertyKey(path, options, print), ":", n.value, path.call(print, "value"), options)); |
|
} |
|
|
|
return concat$d(parts); |
|
// Babel 6 |
|
|
|
case "ClassMethod": |
|
case "ClassPrivateMethod": |
|
case "MethodDefinition": |
|
case "TSAbstractMethodDefinition": |
|
if (n.decorators && n.decorators.length !== 0) { |
|
parts.push(printDecorators(path, options, print)); |
|
} |
|
|
|
if (n.accessibility) { |
|
parts.push(n.accessibility + " "); |
|
} |
|
|
|
if (n.static) { |
|
parts.push("static "); |
|
} |
|
|
|
if (n.type === "TSAbstractMethodDefinition") { |
|
parts.push("abstract "); |
|
} |
|
|
|
parts.push(printMethod(path, options, print)); |
|
return concat$d(parts); |
|
|
|
case "ObjectMethod": |
|
return printMethod(path, options, print); |
|
|
|
case "Decorator": |
|
return concat$d(["@", path.call(print, "expression"), path.call(print, "callee")]); |
|
|
|
case "ArrayExpression": |
|
case "ArrayPattern": |
|
if (n.elements.length === 0) { |
|
if (!hasDanglingComments$1(n)) { |
|
parts.push("[]"); |
|
} else { |
|
parts.push(group$b(concat$d(["[", comments.printDanglingComments(path, options), softline$6, "]"]))); |
|
} |
|
} else { |
|
var _lastElem = getLast$2(n.elements); |
|
|
|
var canHaveTrailingComma = !(_lastElem && _lastElem.type === "RestElement"); // JavaScript allows you to have empty elements in an array which |
|
// changes its length based on the number of commas. The algorithm |
|
// is that if the last argument is null, we need to force insert |
|
// a comma to ensure JavaScript recognizes it. |
|
// [,].length === 1 |
|
// [1,].length === 1 |
|
// [1,,].length === 2 |
|
// |
|
// Note that getLast returns null if the array is empty, but |
|
// we already check for an empty array just above so we are safe |
|
|
|
var needsForcedTrailingComma = canHaveTrailingComma && _lastElem === null; |
|
|
|
var _shouldBreak = n.elements.length > 1 && n.elements.every(function (element, i, elements) { |
|
var elementType = element && element.type; |
|
|
|
if (elementType !== "ArrayExpression" && elementType !== "ObjectExpression") { |
|
return false; |
|
} |
|
|
|
var nextElement = elements[i + 1]; |
|
|
|
if (nextElement && elementType !== nextElement.type) { |
|
return false; |
|
} |
|
|
|
var itemsKey = elementType === "ArrayExpression" ? "elements" : "properties"; |
|
return element[itemsKey] && element[itemsKey].length > 1; |
|
}); |
|
|
|
parts.push(group$b(concat$d(["[", indent$7(concat$d([softline$6, printArrayItems(path, options, "elements", print)])), needsForcedTrailingComma ? "," : "", ifBreak$6(canHaveTrailingComma && !needsForcedTrailingComma && shouldPrintComma$1(options) ? "," : ""), comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true), softline$6, "]"]), { |
|
shouldBreak: _shouldBreak |
|
})); |
|
} |
|
|
|
parts.push(printOptionalToken(path), printTypeAnnotation(path, options, print)); |
|
return concat$d(parts); |
|
|
|
case "SequenceExpression": |
|
{ |
|
var _parent6 = path.getParentNode(0); |
|
|
|
if (_parent6.type === "ExpressionStatement" || _parent6.type === "ForStatement") { |
|
// For ExpressionStatements and for-loop heads, which are among |
|
// the few places a SequenceExpression appears unparenthesized, we want |
|
// to indent expressions after the first. |
|
var _parts2 = []; |
|
path.each(function (p) { |
|
if (p.getName() === 0) { |
|
_parts2.push(print(p)); |
|
} else { |
|
_parts2.push(",", indent$7(concat$d([line$7, print(p)]))); |
|
} |
|
}, "expressions"); |
|
return group$b(concat$d(_parts2)); |
|
} |
|
|
|
return group$b(concat$d([join$9(concat$d([",", line$7]), path.map(print, "expressions"))])); |
|
} |
|
|
|
case "ThisExpression": |
|
return "this"; |
|
|
|
case "Super": |
|
return "super"; |
|
|
|
case "NullLiteral": |
|
// Babel 6 Literal split |
|
return "null"; |
|
|
|
case "RegExpLiteral": |
|
// Babel 6 Literal split |
|
return printRegex(n); |
|
|
|
case "NumericLiteral": |
|
// Babel 6 Literal split |
|
return printNumber$2(n.extra.raw); |
|
|
|
case "BigIntLiteral": |
|
// babel: n.extra.raw, typescript: n.raw, flow: n.bigint |
|
return (n.bigint || (n.extra ? n.extra.raw : n.raw)).toLowerCase(); |
|
|
|
case "BooleanLiteral": // Babel 6 Literal split |
|
|
|
case "StringLiteral": // Babel 6 Literal split |
|
|
|
case "Literal": |
|
{ |
|
if (n.regex) { |
|
return printRegex(n.regex); |
|
} |
|
|
|
if (typeof n.value === "number") { |
|
return printNumber$2(n.raw); |
|
} |
|
|
|
if (typeof n.value !== "string") { |
|
return "" + n.value; |
|
} // TypeScript workaround for https://github.com/JamesHenry/typescript-estree/issues/2 |
|
// See corresponding workaround in needs-parens.js |
|
|
|
|
|
var grandParent = path.getParentNode(1); |
|
var isTypeScriptDirective = options.parser === "typescript" && typeof n.value === "string" && grandParent && (grandParent.type === "Program" || grandParent.type === "BlockStatement"); |
|
return nodeStr(n, options, isTypeScriptDirective); |
|
} |
|
|
|
case "Directive": |
|
return path.call(print, "value"); |
|
// Babel 6 |
|
|
|
case "DirectiveLiteral": |
|
return nodeStr(n, options); |
|
|
|
case "UnaryExpression": |
|
parts.push(n.operator); |
|
|
|
if (/[a-z]$/.test(n.operator)) { |
|
parts.push(" "); |
|
} |
|
|
|
if (n.argument.comments && n.argument.comments.length > 0) { |
|
parts.push(group$b(concat$d(["(", indent$7(concat$d([softline$6, path.call(print, "argument")])), softline$6, ")"]))); |
|
} else { |
|
parts.push(path.call(print, "argument")); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "UpdateExpression": |
|
parts.push(path.call(print, "argument"), n.operator); |
|
|
|
if (n.prefix) { |
|
parts.reverse(); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "ConditionalExpression": |
|
return printTernaryOperator(path, options, print, { |
|
beforeParts: function beforeParts() { |
|
return [path.call(print, "test")]; |
|
}, |
|
afterParts: function afterParts(breakClosingParen) { |
|
return [breakClosingParen ? softline$6 : ""]; |
|
}, |
|
shouldCheckJsx: true, |
|
conditionalNodeType: "ConditionalExpression", |
|
consequentNodePropertyName: "consequent", |
|
alternateNodePropertyName: "alternate", |
|
testNodePropertyName: "test", |
|
breakNested: true |
|
}); |
|
|
|
case "VariableDeclaration": |
|
{ |
|
var printed = path.map(function (childPath) { |
|
return print(childPath); |
|
}, "declarations"); // We generally want to terminate all variable declarations with a |
|
// semicolon, except when they in the () part of for loops. |
|
|
|
var parentNode = path.getParentNode(); |
|
var isParentForLoop = parentNode.type === "ForStatement" || parentNode.type === "ForInStatement" || parentNode.type === "ForOfStatement" || parentNode.type === "ForAwaitStatement"; |
|
var hasValue = n.declarations.some(function (decl) { |
|
return decl.init; |
|
}); |
|
var firstVariable; |
|
|
|
if (printed.length === 1 && !n.declarations[0].comments) { |
|
firstVariable = printed[0]; |
|
} else if (printed.length > 0) { |
|
// Indent first var to comply with eslint one-var rule |
|
firstVariable = indent$7(printed[0]); |
|
} |
|
|
|
parts = [n.declare ? "declare " : "", n.kind, firstVariable ? concat$d([" ", firstVariable]) : "", indent$7(concat$d(printed.slice(1).map(function (p) { |
|
return concat$d([",", hasValue && !isParentForLoop ? hardline$9 : line$7, p]); |
|
})))]; |
|
|
|
if (!(isParentForLoop && parentNode.body !== n)) { |
|
parts.push(semi); |
|
} |
|
|
|
return group$b(concat$d(parts)); |
|
} |
|
|
|
case "TSTypeAliasDeclaration": |
|
{ |
|
if (n.declare) { |
|
parts.push("declare "); |
|
} |
|
|
|
var _printed = printAssignmentRight(n.id, n.typeAnnotation, n.typeAnnotation && path.call(print, "typeAnnotation"), options); |
|
|
|
parts.push("type ", path.call(print, "id"), path.call(print, "typeParameters"), " =", _printed, semi); |
|
return group$b(concat$d(parts)); |
|
} |
|
|
|
case "VariableDeclarator": |
|
return printAssignment(n.id, path.call(print, "id"), " =", n.init, n.init && path.call(print, "init"), options); |
|
|
|
case "WithStatement": |
|
return group$b(concat$d(["with (", path.call(print, "object"), ")", adjustClause(n.body, path.call(print, "body"))])); |
|
|
|
case "IfStatement": |
|
{ |
|
var con = adjustClause(n.consequent, path.call(print, "consequent")); |
|
var opening = group$b(concat$d(["if (", group$b(concat$d([indent$7(concat$d([softline$6, path.call(print, "test")])), softline$6])), ")", con])); |
|
parts.push(opening); |
|
|
|
if (n.alternate) { |
|
var commentOnOwnLine = hasTrailingComment$1(n.consequent) && n.consequent.comments.some(function (comment) { |
|
return comment.trailing && !comments$1.isBlockComment(comment); |
|
}) || needsHardlineAfterDanglingComment$1(n); |
|
var elseOnSameLine = n.consequent.type === "BlockStatement" && !commentOnOwnLine; |
|
parts.push(elseOnSameLine ? " " : hardline$9); |
|
|
|
if (hasDanglingComments$1(n)) { |
|
parts.push(comments.printDanglingComments(path, options, true), commentOnOwnLine ? hardline$9 : " "); |
|
} |
|
|
|
parts.push("else", group$b(adjustClause(n.alternate, path.call(print, "alternate"), n.alternate.type === "IfStatement"))); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
case "ForStatement": |
|
{ |
|
var _body = adjustClause(n.body, path.call(print, "body")); // We want to keep dangling comments above the loop to stay consistent. |
|
// Any comment positioned between the for statement and the parentheses |
|
// is going to be printed before the statement. |
|
|
|
|
|
var _dangling = comments.printDanglingComments(path, options, |
|
/* sameLine */ |
|
true); |
|
|
|
var printedComments = _dangling ? concat$d([_dangling, softline$6]) : ""; |
|
|
|
if (!n.init && !n.test && !n.update) { |
|
return concat$d([printedComments, group$b(concat$d(["for (;;)", _body]))]); |
|
} |
|
|
|
return concat$d([printedComments, group$b(concat$d(["for (", group$b(concat$d([indent$7(concat$d([softline$6, path.call(print, "init"), ";", line$7, path.call(print, "test"), ";", line$7, path.call(print, "update")])), softline$6])), ")", _body]))]); |
|
} |
|
|
|
case "WhileStatement": |
|
return group$b(concat$d(["while (", group$b(concat$d([indent$7(concat$d([softline$6, path.call(print, "test")])), softline$6])), ")", adjustClause(n.body, path.call(print, "body"))])); |
|
|
|
case "ForInStatement": |
|
// Note: esprima can't actually parse "for each (". |
|
return group$b(concat$d([n.each ? "for each (" : "for (", path.call(print, "left"), " in ", path.call(print, "right"), ")", adjustClause(n.body, path.call(print, "body"))])); |
|
|
|
case "ForOfStatement": |
|
case "ForAwaitStatement": |
|
{ |
|
// Babel 7 removed ForAwaitStatement in favor of ForOfStatement |
|
// with `"await": true`: |
|
// https://github.com/estree/estree/pull/138 |
|
var isAwait = n.type === "ForAwaitStatement" || n.await; |
|
return group$b(concat$d(["for", isAwait ? " await" : "", " (", path.call(print, "left"), " of ", path.call(print, "right"), ")", adjustClause(n.body, path.call(print, "body"))])); |
|
} |
|
|
|
case "DoWhileStatement": |
|
{ |
|
var clause = adjustClause(n.body, path.call(print, "body")); |
|
var doBody = group$b(concat$d(["do", clause])); |
|
parts = [doBody]; |
|
|
|
if (n.body.type === "BlockStatement") { |
|
parts.push(" "); |
|
} else { |
|
parts.push(hardline$9); |
|
} |
|
|
|
parts.push("while ("); |
|
parts.push(group$b(concat$d([indent$7(concat$d([softline$6, path.call(print, "test")])), softline$6])), ")", semi); |
|
return concat$d(parts); |
|
} |
|
|
|
case "DoExpression": |
|
return concat$d(["do ", path.call(print, "body")]); |
|
|
|
case "BreakStatement": |
|
parts.push("break"); |
|
|
|
if (n.label) { |
|
parts.push(" ", path.call(print, "label")); |
|
} |
|
|
|
parts.push(semi); |
|
return concat$d(parts); |
|
|
|
case "ContinueStatement": |
|
parts.push("continue"); |
|
|
|
if (n.label) { |
|
parts.push(" ", path.call(print, "label")); |
|
} |
|
|
|
parts.push(semi); |
|
return concat$d(parts); |
|
|
|
case "LabeledStatement": |
|
if (n.body.type === "EmptyStatement") { |
|
return concat$d([path.call(print, "label"), ":;"]); |
|
} |
|
|
|
return concat$d([path.call(print, "label"), ": ", path.call(print, "body")]); |
|
|
|
case "TryStatement": |
|
return concat$d(["try ", path.call(print, "block"), n.handler ? concat$d([" ", path.call(print, "handler")]) : "", n.finalizer ? concat$d([" finally ", path.call(print, "finalizer")]) : ""]); |
|
|
|
case "CatchClause": |
|
if (n.param) { |
|
var hasComments = n.param.comments && n.param.comments.some(function (comment) { |
|
return !comments$1.isBlockComment(comment) || comment.leading && hasNewline$4(options.originalText, options.locEnd(comment)) || comment.trailing && hasNewline$4(options.originalText, options.locStart(comment), { |
|
backwards: true |
|
}); |
|
}); |
|
var param = path.call(print, "param"); |
|
return concat$d(["catch ", hasComments ? concat$d(["(", indent$7(concat$d([softline$6, param])), softline$6, ") "]) : concat$d(["(", param, ") "]), path.call(print, "body")]); |
|
} |
|
|
|
return concat$d(["catch ", path.call(print, "body")]); |
|
|
|
case "ThrowStatement": |
|
return concat$d(["throw ", path.call(print, "argument"), semi]); |
|
// Note: ignoring n.lexical because it has no printing consequences. |
|
|
|
case "SwitchStatement": |
|
return concat$d([group$b(concat$d(["switch (", indent$7(concat$d([softline$6, path.call(print, "discriminant")])), softline$6, ")"])), " {", n.cases.length > 0 ? indent$7(concat$d([hardline$9, join$9(hardline$9, path.map(function (casePath) { |
|
var caseNode = casePath.getValue(); |
|
return concat$d([casePath.call(print), n.cases.indexOf(caseNode) !== n.cases.length - 1 && isNextLineEmpty$4(options.originalText, caseNode, options) ? hardline$9 : ""]); |
|
}, "cases"))])) : "", hardline$9, "}"]); |
|
|
|
case "SwitchCase": |
|
{ |
|
if (n.test) { |
|
parts.push("case ", path.call(print, "test"), ":"); |
|
} else { |
|
parts.push("default:"); |
|
} |
|
|
|
var consequent = n.consequent.filter(function (node) { |
|
return node.type !== "EmptyStatement"; |
|
}); |
|
|
|
if (consequent.length > 0) { |
|
var cons = path.call(function (consequentPath) { |
|
return printStatementSequence(consequentPath, options, print); |
|
}, "consequent"); |
|
parts.push(consequent.length === 1 && consequent[0].type === "BlockStatement" ? concat$d([" ", cons]) : indent$7(concat$d([hardline$9, cons]))); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
// JSX extensions below. |
|
|
|
case "DebuggerStatement": |
|
return concat$d(["debugger", semi]); |
|
|
|
case "JSXAttribute": |
|
parts.push(path.call(print, "name")); |
|
|
|
if (n.value) { |
|
var res; |
|
|
|
if (isStringLiteral$1(n.value)) { |
|
var raw = rawText$1(n.value); // Unescape all quotes so we get an accurate preferred quote |
|
|
|
var final = raw.replace(/'/g, "'").replace(/"/g, '"'); |
|
var quote = getPreferredQuote$1(final, options.jsxSingleQuote ? "'" : '"'); |
|
|
|
var _escape = quote === "'" ? "'" : """; |
|
|
|
final = final.slice(1, -1).replace(new RegExp(quote, "g"), _escape); |
|
res = concat$d([quote, final, quote]); |
|
} else { |
|
res = path.call(print, "value"); |
|
} |
|
|
|
parts.push("=", res); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "JSXIdentifier": |
|
return "" + n.name; |
|
|
|
case "JSXNamespacedName": |
|
return join$9(":", [path.call(print, "namespace"), path.call(print, "name")]); |
|
|
|
case "JSXMemberExpression": |
|
return join$9(".", [path.call(print, "object"), path.call(print, "property")]); |
|
|
|
case "TSQualifiedName": |
|
return join$9(".", [path.call(print, "left"), path.call(print, "right")]); |
|
|
|
case "JSXSpreadAttribute": |
|
case "JSXSpreadChild": |
|
{ |
|
return concat$d(["{", path.call(function (p) { |
|
var printed = concat$d(["...", print(p)]); |
|
var n = p.getValue(); |
|
|
|
if (!n.comments || !n.comments.length) { |
|
return printed; |
|
} |
|
|
|
return concat$d([indent$7(concat$d([softline$6, comments.printComments(p, function () { |
|
return printed; |
|
}, options)])), softline$6]); |
|
}, n.type === "JSXSpreadAttribute" ? "argument" : "expression"), "}"]); |
|
} |
|
|
|
case "JSXExpressionContainer": |
|
{ |
|
var _parent7 = path.getParentNode(0); |
|
|
|
var preventInline = _parent7.type === "JSXAttribute" && n.expression.comments && n.expression.comments.length > 0; |
|
|
|
var _shouldInline = !preventInline && (n.expression.type === "ArrayExpression" || n.expression.type === "ObjectExpression" || n.expression.type === "ArrowFunctionExpression" || n.expression.type === "CallExpression" || n.expression.type === "OptionalCallExpression" || n.expression.type === "FunctionExpression" || n.expression.type === "JSXEmptyExpression" || n.expression.type === "TemplateLiteral" || n.expression.type === "TaggedTemplateExpression" || n.expression.type === "DoExpression" || isJSXNode$1(_parent7) && (n.expression.type === "ConditionalExpression" || isBinaryish$1(n.expression))); |
|
|
|
if (_shouldInline) { |
|
return group$b(concat$d(["{", path.call(print, "expression"), lineSuffixBoundary$1, "}"])); |
|
} |
|
|
|
return group$b(concat$d(["{", indent$7(concat$d([softline$6, path.call(print, "expression")])), softline$6, lineSuffixBoundary$1, "}"])); |
|
} |
|
|
|
case "JSXFragment": |
|
case "JSXElement": |
|
{ |
|
var elem = comments.printComments(path, function () { |
|
return printJSXElement(path, options, print); |
|
}, options); |
|
return maybeWrapJSXElementInParens(path, elem, options); |
|
} |
|
|
|
case "JSXOpeningElement": |
|
{ |
|
var _n = path.getValue(); |
|
|
|
var nameHasComments = _n.name && _n.name.comments && _n.name.comments.length > 0 || _n.typeParameters && _n.typeParameters.comments && _n.typeParameters.comments.length > 0; // Don't break self-closing elements with no attributes and no comments |
|
|
|
if (_n.selfClosing && !_n.attributes.length && !nameHasComments) { |
|
return concat$d(["<", path.call(print, "name"), path.call(print, "typeParameters"), " />"]); |
|
} // don't break up opening elements with a single long text attribute |
|
|
|
|
|
if (_n.attributes && _n.attributes.length === 1 && _n.attributes[0].value && isStringLiteral$1(_n.attributes[0].value) && !_n.attributes[0].value.value.includes("\n") && // We should break for the following cases: |
|
// <div |
|
// // comment |
|
// attr="value" |
|
// > |
|
// <div |
|
// attr="value" |
|
// // comment |
|
// > |
|
!nameHasComments && (!_n.attributes[0].comments || !_n.attributes[0].comments.length)) { |
|
return group$b(concat$d(["<", path.call(print, "name"), path.call(print, "typeParameters"), " ", concat$d(path.map(print, "attributes")), _n.selfClosing ? " />" : ">"])); |
|
} |
|
|
|
var lastAttrHasTrailingComments = _n.attributes.length && hasTrailingComment$1(getLast$2(_n.attributes)); |
|
var bracketSameLine = // Simple tags (no attributes and no comment in tag name) should be |
|
// kept unbroken regardless of `jsxBracketSameLine` |
|
!_n.attributes.length && !nameHasComments || options.jsxBracketSameLine && ( // We should print the bracket in a new line for the following cases: |
|
// <div |
|
// // comment |
|
// > |
|
// <div |
|
// attr // comment |
|
// > |
|
!nameHasComments || _n.attributes.length) && !lastAttrHasTrailingComments; // We should print the opening element expanded if any prop value is a |
|
// string literal with newlines |
|
|
|
var _shouldBreak2 = _n.attributes && _n.attributes.some(function (attr) { |
|
return attr.value && isStringLiteral$1(attr.value) && attr.value.value.includes("\n"); |
|
}); |
|
|
|
return group$b(concat$d(["<", path.call(print, "name"), path.call(print, "typeParameters"), concat$d([indent$7(concat$d(path.map(function (attr) { |
|
return concat$d([line$7, print(attr)]); |
|
}, "attributes"))), _n.selfClosing ? line$7 : bracketSameLine ? ">" : softline$6]), _n.selfClosing ? "/>" : bracketSameLine ? "" : ">"]), { |
|
shouldBreak: _shouldBreak2 |
|
}); |
|
} |
|
|
|
case "JSXClosingElement": |
|
return concat$d(["</", path.call(print, "name"), ">"]); |
|
|
|
case "JSXOpeningFragment": |
|
case "JSXClosingFragment": |
|
{ |
|
var hasComment = n.comments && n.comments.length; |
|
var hasOwnLineComment = hasComment && !n.comments.every(comments$1.isBlockComment); |
|
var isOpeningFragment = n.type === "JSXOpeningFragment"; |
|
return concat$d([isOpeningFragment ? "<" : "</", indent$7(concat$d([hasOwnLineComment ? hardline$9 : hasComment && !isOpeningFragment ? " " : "", comments.printDanglingComments(path, options, true)])), hasOwnLineComment ? hardline$9 : "", ">"]); |
|
} |
|
|
|
case "JSXText": |
|
/* istanbul ignore next */ |
|
throw new Error("JSXTest should be handled by JSXElement"); |
|
|
|
case "JSXEmptyExpression": |
|
{ |
|
var requiresHardline = n.comments && !n.comments.every(comments$1.isBlockComment); |
|
return concat$d([comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
!requiresHardline), requiresHardline ? hardline$9 : ""]); |
|
} |
|
|
|
case "ClassBody": |
|
if (!n.comments && n.body.length === 0) { |
|
return "{}"; |
|
} |
|
|
|
return concat$d(["{", n.body.length > 0 ? indent$7(concat$d([hardline$9, path.call(function (bodyPath) { |
|
return printStatementSequence(bodyPath, options, print); |
|
}, "body")])) : comments.printDanglingComments(path, options), hardline$9, "}"]); |
|
|
|
case "ClassProperty": |
|
case "TSAbstractClassProperty": |
|
case "ClassPrivateProperty": |
|
{ |
|
if (n.decorators && n.decorators.length !== 0) { |
|
parts.push(printDecorators(path, options, print)); |
|
} |
|
|
|
if (n.accessibility) { |
|
parts.push(n.accessibility + " "); |
|
} |
|
|
|
if (n.declare) { |
|
parts.push("declare "); |
|
} |
|
|
|
if (n.static) { |
|
parts.push("static "); |
|
} |
|
|
|
if (n.type === "TSAbstractClassProperty") { |
|
parts.push("abstract "); |
|
} |
|
|
|
if (n.readonly) { |
|
parts.push("readonly "); |
|
} |
|
|
|
var variance = getFlowVariance$1(n); |
|
|
|
if (variance) { |
|
parts.push(variance); |
|
} |
|
|
|
parts.push(printPropertyKey(path, options, print), printOptionalToken(path), printTypeAnnotation(path, options, print)); |
|
|
|
if (n.value) { |
|
parts.push(" =", printAssignmentRight(n.key, n.value, path.call(print, "value"), options)); |
|
} |
|
|
|
parts.push(semi); |
|
return group$b(concat$d(parts)); |
|
} |
|
|
|
case "ClassDeclaration": |
|
case "ClassExpression": |
|
if (n.declare) { |
|
parts.push("declare "); |
|
} |
|
|
|
parts.push(concat$d(printClass(path, options, print))); |
|
return concat$d(parts); |
|
|
|
case "TSInterfaceHeritage": |
|
parts.push(path.call(print, "expression")); |
|
|
|
if (n.typeParameters) { |
|
parts.push(path.call(print, "typeParameters")); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "TemplateElement": |
|
return join$9(literalline$4, n.value.raw.split(/\r?\n/g)); |
|
|
|
case "TemplateLiteral": |
|
{ |
|
var expressions = path.map(print, "expressions"); |
|
|
|
var _parentNode = path.getParentNode(); |
|
|
|
if (isJestEachTemplateLiteral$1(n, _parentNode)) { |
|
var _printed2 = printJestEachTemplateLiteral(n, expressions, options); |
|
|
|
if (_printed2) { |
|
return _printed2; |
|
} |
|
} |
|
|
|
var isSimple = isSimpleTemplateLiteral$1(n); |
|
|
|
if (isSimple) { |
|
expressions = expressions.map(function (doc) { |
|
return printDocToString$1(doc, Object.assign({}, options, { |
|
printWidth: Infinity |
|
})).formatted; |
|
}); |
|
} |
|
|
|
parts.push(lineSuffixBoundary$1, "`"); |
|
path.each(function (childPath) { |
|
var i = childPath.getName(); |
|
parts.push(print(childPath)); |
|
|
|
if (i < expressions.length) { |
|
// For a template literal of the following form: |
|
// `someQuery { |
|
// ${call({ |
|
// a, |
|
// b, |
|
// })} |
|
// }` |
|
// the expression is on its own line (there is a \n in the previous |
|
// quasi literal), therefore we want to indent the JavaScript |
|
// expression inside at the beginning of ${ instead of the beginning |
|
// of the `. |
|
var tabWidth = options.tabWidth; |
|
var quasi = childPath.getValue(); |
|
var indentSize = getIndentSize$1(quasi.value.raw, tabWidth); |
|
var _printed3 = expressions[i]; |
|
|
|
if (!isSimple) { |
|
// Breaks at the template element boundaries (${ and }) are preferred to breaking |
|
// in the middle of a MemberExpression |
|
if (n.expressions[i].comments && n.expressions[i].comments.length || n.expressions[i].type === "MemberExpression" || n.expressions[i].type === "OptionalMemberExpression" || n.expressions[i].type === "ConditionalExpression") { |
|
_printed3 = concat$d([indent$7(concat$d([softline$6, _printed3])), softline$6]); |
|
} |
|
} |
|
|
|
var aligned = indentSize === 0 && quasi.value.raw.endsWith("\n") ? align$1(-Infinity, _printed3) : addAlignmentToDoc$2(_printed3, indentSize, tabWidth); |
|
parts.push(group$b(concat$d(["${", aligned, lineSuffixBoundary$1, "}"]))); |
|
} |
|
}, "quasis"); |
|
parts.push("`"); |
|
return concat$d(parts); |
|
} |
|
// These types are unprintable because they serve as abstract |
|
// supertypes for other (printable) types. |
|
|
|
case "TaggedTemplateExpression": |
|
return concat$d([path.call(print, "tag"), path.call(print, "typeParameters"), path.call(print, "quasi")]); |
|
|
|
case "Node": |
|
case "Printable": |
|
case "SourceLocation": |
|
case "Position": |
|
case "Statement": |
|
case "Function": |
|
case "Pattern": |
|
case "Expression": |
|
case "Declaration": |
|
case "Specifier": |
|
case "NamedSpecifier": |
|
case "Comment": |
|
case "MemberTypeAnnotation": // Flow |
|
|
|
case "Type": |
|
/* istanbul ignore next */ |
|
throw new Error("unprintable type: " + JSON.stringify(n.type)); |
|
// Type Annotations for Facebook Flow, typically stripped out or |
|
// transformed away before printing. |
|
|
|
case "TypeAnnotation": |
|
case "TSTypeAnnotation": |
|
if (n.typeAnnotation) { |
|
return path.call(print, "typeAnnotation"); |
|
} |
|
/* istanbul ignore next */ |
|
|
|
|
|
return ""; |
|
|
|
case "TSTupleType": |
|
case "TupleTypeAnnotation": |
|
{ |
|
var typesField = n.type === "TSTupleType" ? "elementTypes" : "types"; |
|
return group$b(concat$d(["[", indent$7(concat$d([softline$6, printArrayItems(path, options, typesField, print)])), ifBreak$6(shouldPrintComma$1(options, "all") ? "," : ""), comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true), softline$6, "]"])); |
|
} |
|
|
|
case "ExistsTypeAnnotation": |
|
return "*"; |
|
|
|
case "EmptyTypeAnnotation": |
|
return "empty"; |
|
|
|
case "AnyTypeAnnotation": |
|
return "any"; |
|
|
|
case "MixedTypeAnnotation": |
|
return "mixed"; |
|
|
|
case "ArrayTypeAnnotation": |
|
return concat$d([path.call(print, "elementType"), "[]"]); |
|
|
|
case "BooleanTypeAnnotation": |
|
return "boolean"; |
|
|
|
case "BooleanLiteralTypeAnnotation": |
|
return "" + n.value; |
|
|
|
case "DeclareClass": |
|
return printFlowDeclaration(path, printClass(path, options, print)); |
|
|
|
case "TSDeclareFunction": |
|
// For TypeScript the TSDeclareFunction node shares the AST |
|
// structure with FunctionDeclaration |
|
return concat$d([n.declare ? "declare " : "", printFunctionDeclaration(path, print, options), semi]); |
|
|
|
case "DeclareFunction": |
|
return printFlowDeclaration(path, ["function ", path.call(print, "id"), n.predicate ? " " : "", path.call(print, "predicate"), semi]); |
|
|
|
case "DeclareModule": |
|
return printFlowDeclaration(path, ["module ", path.call(print, "id"), " ", path.call(print, "body")]); |
|
|
|
case "DeclareModuleExports": |
|
return printFlowDeclaration(path, ["module.exports", ": ", path.call(print, "typeAnnotation"), semi]); |
|
|
|
case "DeclareVariable": |
|
return printFlowDeclaration(path, ["var ", path.call(print, "id"), semi]); |
|
|
|
case "DeclareExportAllDeclaration": |
|
return concat$d(["declare export * from ", path.call(print, "source")]); |
|
|
|
case "DeclareExportDeclaration": |
|
return concat$d(["declare ", printExportDeclaration(path, options, print)]); |
|
|
|
case "DeclareOpaqueType": |
|
case "OpaqueType": |
|
{ |
|
parts.push("opaque type ", path.call(print, "id"), path.call(print, "typeParameters")); |
|
|
|
if (n.supertype) { |
|
parts.push(": ", path.call(print, "supertype")); |
|
} |
|
|
|
if (n.impltype) { |
|
parts.push(" = ", path.call(print, "impltype")); |
|
} |
|
|
|
parts.push(semi); |
|
|
|
if (n.type === "DeclareOpaqueType") { |
|
return printFlowDeclaration(path, parts); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
case "EnumDeclaration": |
|
return concat$d(["enum ", path.call(print, "id"), " ", path.call(print, "body")]); |
|
|
|
case "EnumBooleanBody": |
|
case "EnumNumberBody": |
|
case "EnumStringBody": |
|
case "EnumSymbolBody": |
|
{ |
|
if (n.type === "EnumSymbolBody" || n.explicitType) { |
|
var type = null; |
|
|
|
switch (n.type) { |
|
case "EnumBooleanBody": |
|
type = "boolean"; |
|
break; |
|
|
|
case "EnumNumberBody": |
|
type = "number"; |
|
break; |
|
|
|
case "EnumStringBody": |
|
type = "string"; |
|
break; |
|
|
|
case "EnumSymbolBody": |
|
type = "symbol"; |
|
break; |
|
} |
|
|
|
parts.push("of ", type, " "); |
|
} |
|
|
|
if (n.members.length === 0) { |
|
parts.push(group$b(concat$d(["{", comments.printDanglingComments(path, options), softline$6, "}"]))); |
|
} else { |
|
parts.push(group$b(concat$d(["{", indent$7(concat$d([hardline$9, printArrayItems(path, options, "members", print), shouldPrintComma$1(options) ? "," : ""])), comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true), hardline$9, "}"]))); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
case "EnumBooleanMember": |
|
case "EnumNumberMember": |
|
case "EnumStringMember": |
|
return concat$d([path.call(print, "id"), " = ", _typeof(n.init) === "object" ? path.call(print, "init") : String(n.init)]); |
|
|
|
case "EnumDefaultedMember": |
|
return path.call(print, "id"); |
|
|
|
case "FunctionTypeAnnotation": |
|
case "TSFunctionType": |
|
{ |
|
// FunctionTypeAnnotation is ambiguous: |
|
// declare function foo(a: B): void; OR |
|
// var A: (a: B) => void; |
|
var _parent8 = path.getParentNode(0); |
|
|
|
var _parentParent2 = path.getParentNode(1); |
|
|
|
var _parentParentParent = path.getParentNode(2); |
|
|
|
var isArrowFunctionTypeAnnotation = n.type === "TSFunctionType" || !((_parent8.type === "ObjectTypeProperty" || _parent8.type === "ObjectTypeInternalSlot") && !getFlowVariance$1(_parent8) && !_parent8.optional && options.locStart(_parent8) === options.locStart(n) || _parent8.type === "ObjectTypeCallProperty" || _parentParentParent && _parentParentParent.type === "DeclareFunction"); |
|
var needsColon = isArrowFunctionTypeAnnotation && (_parent8.type === "TypeAnnotation" || _parent8.type === "TSTypeAnnotation"); // Sadly we can't put it inside of FastPath::needsColon because we are |
|
// printing ":" as part of the expression and it would put parenthesis |
|
// around :( |
|
|
|
var needsParens = needsColon && isArrowFunctionTypeAnnotation && (_parent8.type === "TypeAnnotation" || _parent8.type === "TSTypeAnnotation") && _parentParent2.type === "ArrowFunctionExpression"; |
|
|
|
if (isObjectTypePropertyAFunction$1(_parent8, options)) { |
|
isArrowFunctionTypeAnnotation = true; |
|
needsColon = true; |
|
} |
|
|
|
if (needsParens) { |
|
parts.push("("); |
|
} |
|
|
|
parts.push(printFunctionParams(path, print, options, |
|
/* expandArg */ |
|
false, |
|
/* printTypeParams */ |
|
true)); // The returnType is not wrapped in a TypeAnnotation, so the colon |
|
// needs to be added separately. |
|
|
|
if (n.returnType || n.predicate || n.typeAnnotation) { |
|
parts.push(isArrowFunctionTypeAnnotation ? " => " : ": ", path.call(print, "returnType"), path.call(print, "predicate"), path.call(print, "typeAnnotation")); |
|
} |
|
|
|
if (needsParens) { |
|
parts.push(")"); |
|
} |
|
|
|
return group$b(concat$d(parts)); |
|
} |
|
|
|
case "TSRestType": |
|
return concat$d(["...", path.call(print, "typeAnnotation")]); |
|
|
|
case "TSOptionalType": |
|
return concat$d([path.call(print, "typeAnnotation"), "?"]); |
|
|
|
case "FunctionTypeParam": |
|
return concat$d([path.call(print, "name"), printOptionalToken(path), n.name ? ": " : "", path.call(print, "typeAnnotation")]); |
|
|
|
case "GenericTypeAnnotation": |
|
return concat$d([path.call(print, "id"), path.call(print, "typeParameters")]); |
|
|
|
case "DeclareInterface": |
|
case "InterfaceDeclaration": |
|
case "InterfaceTypeAnnotation": |
|
{ |
|
if (n.type === "DeclareInterface" || n.declare) { |
|
parts.push("declare "); |
|
} |
|
|
|
parts.push("interface"); |
|
|
|
if (n.type === "DeclareInterface" || n.type === "InterfaceDeclaration") { |
|
parts.push(" ", path.call(print, "id"), path.call(print, "typeParameters")); |
|
} |
|
|
|
if (n["extends"].length > 0) { |
|
parts.push(group$b(indent$7(concat$d([line$7, "extends ", (n.extends.length === 1 ? identity$2 : indent$7)(join$9(concat$d([",", line$7]), path.map(print, "extends")))])))); |
|
} |
|
|
|
parts.push(" ", path.call(print, "body")); |
|
return group$b(concat$d(parts)); |
|
} |
|
|
|
case "ClassImplements": |
|
case "InterfaceExtends": |
|
return concat$d([path.call(print, "id"), path.call(print, "typeParameters")]); |
|
|
|
case "TSClassImplements": |
|
return concat$d([path.call(print, "expression"), path.call(print, "typeParameters")]); |
|
|
|
case "TSIntersectionType": |
|
case "IntersectionTypeAnnotation": |
|
{ |
|
var types = path.map(print, "types"); |
|
var result = []; |
|
var wasIndented = false; |
|
|
|
for (var _i = 0; _i < types.length; ++_i) { |
|
if (_i === 0) { |
|
result.push(types[_i]); |
|
} else if (isObjectType$1(n.types[_i - 1]) && isObjectType$1(n.types[_i])) { |
|
// If both are objects, don't indent |
|
result.push(concat$d([" & ", wasIndented ? indent$7(types[_i]) : types[_i]])); |
|
} else if (!isObjectType$1(n.types[_i - 1]) && !isObjectType$1(n.types[_i])) { |
|
// If no object is involved, go to the next line if it breaks |
|
result.push(indent$7(concat$d([" &", line$7, types[_i]]))); |
|
} else { |
|
// If you go from object to non-object or vis-versa, then inline it |
|
if (_i > 1) { |
|
wasIndented = true; |
|
} |
|
|
|
result.push(" & ", _i > 1 ? indent$7(types[_i]) : types[_i]); |
|
} |
|
} |
|
|
|
return group$b(concat$d(result)); |
|
} |
|
|
|
case "TSUnionType": |
|
case "UnionTypeAnnotation": |
|
{ |
|
// single-line variation |
|
// A | B | C |
|
// multi-line variation |
|
// | A |
|
// | B |
|
// | C |
|
var _parent9 = path.getParentNode(); // If there's a leading comment, the parent is doing the indentation |
|
|
|
|
|
var shouldIndent = _parent9.type !== "TypeParameterInstantiation" && _parent9.type !== "TSTypeParameterInstantiation" && _parent9.type !== "GenericTypeAnnotation" && _parent9.type !== "TSTypeReference" && _parent9.type !== "TSTypeAssertion" && _parent9.type !== "TupleTypeAnnotation" && _parent9.type !== "TSTupleType" && !(_parent9.type === "FunctionTypeParam" && !_parent9.name) && !((_parent9.type === "TypeAlias" || _parent9.type === "VariableDeclarator" || _parent9.type === "TSTypeAliasDeclaration") && hasLeadingOwnLineComment$1(options.originalText, n, options)); // { |
|
// a: string |
|
// } | null | void |
|
// should be inlined and not be printed in the multi-line variant |
|
|
|
var shouldHug = shouldHugType(n); // We want to align the children but without its comment, so it looks like |
|
// | child1 |
|
// // comment |
|
// | child2 |
|
|
|
var _printed4 = path.map(function (typePath) { |
|
var printedType = typePath.call(print); |
|
|
|
if (!shouldHug) { |
|
printedType = align$1(2, printedType); |
|
} |
|
|
|
return comments.printComments(typePath, function () { |
|
return printedType; |
|
}, options); |
|
}, "types"); |
|
|
|
if (shouldHug) { |
|
return join$9(" | ", _printed4); |
|
} |
|
|
|
var shouldAddStartLine = shouldIndent && !hasLeadingOwnLineComment$1(options.originalText, n, options); |
|
var code = concat$d([ifBreak$6(concat$d([shouldAddStartLine ? line$7 : "", "| "])), join$9(concat$d([line$7, "| "]), _printed4)]); |
|
|
|
if (needsParens_1(path, options)) { |
|
return group$b(concat$d([indent$7(code), softline$6])); |
|
} |
|
|
|
if (_parent9.type === "TupleTypeAnnotation" && _parent9.types.length > 1 || _parent9.type === "TSTupleType" && _parent9.elementTypes.length > 1) { |
|
return group$b(concat$d([indent$7(concat$d([ifBreak$6(concat$d(["(", softline$6])), code])), softline$6, ifBreak$6(")")])); |
|
} |
|
|
|
return group$b(shouldIndent ? indent$7(code) : code); |
|
} |
|
|
|
case "NullableTypeAnnotation": |
|
return concat$d(["?", path.call(print, "typeAnnotation")]); |
|
|
|
case "TSNullKeyword": |
|
case "NullLiteralTypeAnnotation": |
|
return "null"; |
|
|
|
case "ThisTypeAnnotation": |
|
return "this"; |
|
|
|
case "NumberTypeAnnotation": |
|
return "number"; |
|
|
|
case "ObjectTypeCallProperty": |
|
if (n.static) { |
|
parts.push("static "); |
|
} |
|
|
|
parts.push(path.call(print, "value")); |
|
return concat$d(parts); |
|
|
|
case "ObjectTypeIndexer": |
|
{ |
|
var _variance = getFlowVariance$1(n); |
|
|
|
return concat$d([_variance || "", "[", path.call(print, "id"), n.id ? ": " : "", path.call(print, "key"), "]: ", path.call(print, "value")]); |
|
} |
|
|
|
case "ObjectTypeProperty": |
|
{ |
|
var _variance2 = getFlowVariance$1(n); |
|
|
|
var modifier = ""; |
|
|
|
if (n.proto) { |
|
modifier = "proto "; |
|
} else if (n.static) { |
|
modifier = "static "; |
|
} |
|
|
|
return concat$d([modifier, isGetterOrSetter$1(n) ? n.kind + " " : "", _variance2 || "", printPropertyKey(path, options, print), printOptionalToken(path), isFunctionNotation$1(n, options) ? "" : ": ", path.call(print, "value")]); |
|
} |
|
|
|
case "QualifiedTypeIdentifier": |
|
return concat$d([path.call(print, "qualification"), ".", path.call(print, "id")]); |
|
|
|
case "StringLiteralTypeAnnotation": |
|
return nodeStr(n, options); |
|
|
|
case "NumberLiteralTypeAnnotation": |
|
assert.strictEqual(_typeof(n.value), "number"); |
|
|
|
if (n.extra != null) { |
|
return printNumber$2(n.extra.raw); |
|
} |
|
|
|
return printNumber$2(n.raw); |
|
|
|
case "StringTypeAnnotation": |
|
return "string"; |
|
|
|
case "DeclareTypeAlias": |
|
case "TypeAlias": |
|
{ |
|
if (n.type === "DeclareTypeAlias" || n.declare) { |
|
parts.push("declare "); |
|
} |
|
|
|
var _printed5 = printAssignmentRight(n.id, n.right, path.call(print, "right"), options); |
|
|
|
parts.push("type ", path.call(print, "id"), path.call(print, "typeParameters"), " =", _printed5, semi); |
|
return group$b(concat$d(parts)); |
|
} |
|
|
|
case "TypeCastExpression": |
|
{ |
|
var value = path.getValue(); // Flow supports a comment syntax for specifying type annotations: https://flow.org/en/docs/types/comments/. |
|
// Unfortunately, its parser doesn't differentiate between comment annotations and regular |
|
// annotations when producing an AST. So to preserve parentheses around type casts that use |
|
// the comment syntax, we need to hackily read the source itself to see if the code contains |
|
// a type annotation comment. |
|
// |
|
// Note that we're able to use the normal whitespace regex here because the Flow parser has |
|
// already deemed this AST node to be a type cast. Only the Babel parser needs the |
|
// non-line-break whitespace regex, which is why hasFlowShorthandAnnotationComment() is |
|
// implemented differently. |
|
|
|
var commentSyntax = value && value.typeAnnotation && value.typeAnnotation.range && options.originalText.substring(value.typeAnnotation.range[0]).match(/^\/\*\s*:/); |
|
return concat$d(["(", path.call(print, "expression"), commentSyntax ? " /*" : "", ": ", path.call(print, "typeAnnotation"), commentSyntax ? " */" : "", ")"]); |
|
} |
|
|
|
case "TypeParameterDeclaration": |
|
case "TypeParameterInstantiation": |
|
{ |
|
var _value = path.getValue(); |
|
|
|
var commentStart = _value.range ? options.originalText.substring(0, _value.range[0]).lastIndexOf("/*") : -1; // As noted in the TypeCastExpression comments above, we're able to use a normal whitespace regex here |
|
// because we know for sure that this is a type definition. |
|
|
|
var _commentSyntax = commentStart >= 0 && options.originalText.substring(commentStart).match(/^\/\*\s*::/); |
|
|
|
if (_commentSyntax) { |
|
return concat$d(["/*:: ", printTypeParameters(path, options, print, "params"), " */"]); |
|
} |
|
|
|
return printTypeParameters(path, options, print, "params"); |
|
} |
|
|
|
case "TSTypeParameterDeclaration": |
|
case "TSTypeParameterInstantiation": |
|
return printTypeParameters(path, options, print, "params"); |
|
|
|
case "TSTypeParameter": |
|
case "TypeParameter": |
|
{ |
|
var _parent10 = path.getParentNode(); |
|
|
|
if (_parent10.type === "TSMappedType") { |
|
parts.push("[", path.call(print, "name")); |
|
|
|
if (n.constraint) { |
|
parts.push(" in ", path.call(print, "constraint")); |
|
} |
|
|
|
parts.push("]"); |
|
return concat$d(parts); |
|
} |
|
|
|
var _variance3 = getFlowVariance$1(n); |
|
|
|
if (_variance3) { |
|
parts.push(_variance3); |
|
} |
|
|
|
parts.push(path.call(print, "name")); |
|
|
|
if (n.bound) { |
|
parts.push(": "); |
|
parts.push(path.call(print, "bound")); |
|
} |
|
|
|
if (n.constraint) { |
|
parts.push(" extends ", path.call(print, "constraint")); |
|
} |
|
|
|
if (n["default"]) { |
|
parts.push(" = ", path.call(print, "default")); |
|
} // Keep comma if the file extension is .tsx and |
|
// has one type parameter that isn't extend with any types. |
|
// Because, otherwise formatted result will be invalid as tsx. |
|
|
|
|
|
var _grandParent = path.getNode(2); |
|
|
|
if (_parent10.params && _parent10.params.length === 1 && isTSXFile$1(options) && !n.constraint && _grandParent.type === "ArrowFunctionExpression") { |
|
parts.push(","); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
case "TypeofTypeAnnotation": |
|
return concat$d(["typeof ", path.call(print, "argument")]); |
|
|
|
case "VoidTypeAnnotation": |
|
return "void"; |
|
|
|
case "InferredPredicate": |
|
return "%checks"; |
|
// Unhandled types below. If encountered, nodes of these types should |
|
// be either left alone or desugared into AST types that are fully |
|
// supported by the pretty-printer. |
|
|
|
case "DeclaredPredicate": |
|
return concat$d(["%checks(", path.call(print, "value"), ")"]); |
|
|
|
case "TSAbstractKeyword": |
|
return "abstract"; |
|
|
|
case "TSAnyKeyword": |
|
return "any"; |
|
|
|
case "TSAsyncKeyword": |
|
return "async"; |
|
|
|
case "TSBooleanKeyword": |
|
return "boolean"; |
|
|
|
case "TSBigIntKeyword": |
|
return "bigint"; |
|
|
|
case "TSConstKeyword": |
|
return "const"; |
|
|
|
case "TSDeclareKeyword": |
|
return "declare"; |
|
|
|
case "TSExportKeyword": |
|
return "export"; |
|
|
|
case "TSNeverKeyword": |
|
return "never"; |
|
|
|
case "TSNumberKeyword": |
|
return "number"; |
|
|
|
case "TSObjectKeyword": |
|
return "object"; |
|
|
|
case "TSProtectedKeyword": |
|
return "protected"; |
|
|
|
case "TSPrivateKeyword": |
|
return "private"; |
|
|
|
case "TSPublicKeyword": |
|
return "public"; |
|
|
|
case "TSReadonlyKeyword": |
|
return "readonly"; |
|
|
|
case "TSSymbolKeyword": |
|
return "symbol"; |
|
|
|
case "TSStaticKeyword": |
|
return "static"; |
|
|
|
case "TSStringKeyword": |
|
return "string"; |
|
|
|
case "TSUndefinedKeyword": |
|
return "undefined"; |
|
|
|
case "TSUnknownKeyword": |
|
return "unknown"; |
|
|
|
case "TSVoidKeyword": |
|
return "void"; |
|
|
|
case "TSAsExpression": |
|
return concat$d([path.call(print, "expression"), " as ", path.call(print, "typeAnnotation")]); |
|
|
|
case "TSArrayType": |
|
return concat$d([path.call(print, "elementType"), "[]"]); |
|
|
|
case "TSPropertySignature": |
|
{ |
|
if (n.export) { |
|
parts.push("export "); |
|
} |
|
|
|
if (n.accessibility) { |
|
parts.push(n.accessibility + " "); |
|
} |
|
|
|
if (n.static) { |
|
parts.push("static "); |
|
} |
|
|
|
if (n.readonly) { |
|
parts.push("readonly "); |
|
} |
|
|
|
parts.push(printPropertyKey(path, options, print), printOptionalToken(path)); |
|
|
|
if (n.typeAnnotation) { |
|
parts.push(": "); |
|
parts.push(path.call(print, "typeAnnotation")); |
|
} // This isn't valid semantically, but it's in the AST so we can print it. |
|
|
|
|
|
if (n.initializer) { |
|
parts.push(" = ", path.call(print, "initializer")); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
case "TSParameterProperty": |
|
if (n.accessibility) { |
|
parts.push(n.accessibility + " "); |
|
} |
|
|
|
if (n.export) { |
|
parts.push("export "); |
|
} |
|
|
|
if (n.static) { |
|
parts.push("static "); |
|
} |
|
|
|
if (n.readonly) { |
|
parts.push("readonly "); |
|
} |
|
|
|
parts.push(path.call(print, "parameter")); |
|
return concat$d(parts); |
|
|
|
case "TSTypeReference": |
|
return concat$d([path.call(print, "typeName"), printTypeParameters(path, options, print, "typeParameters")]); |
|
|
|
case "TSTypeQuery": |
|
return concat$d(["typeof ", path.call(print, "exprName")]); |
|
|
|
case "TSIndexSignature": |
|
{ |
|
var _parent11 = path.getParentNode(); |
|
|
|
return concat$d([n.export ? "export " : "", n.accessibility ? concat$d([n.accessibility, " "]) : "", n.static ? "static " : "", n.readonly ? "readonly " : "", "[", n.parameters ? concat$d(path.map(print, "parameters")) : "", "]: ", path.call(print, "typeAnnotation"), _parent11.type === "ClassBody" ? semi : ""]); |
|
} |
|
|
|
case "TSTypePredicate": |
|
return concat$d([n.asserts ? "asserts " : "", path.call(print, "parameterName"), n.typeAnnotation ? concat$d([" is ", path.call(print, "typeAnnotation")]) : ""]); |
|
|
|
case "TSNonNullExpression": |
|
return concat$d([path.call(print, "expression"), "!"]); |
|
|
|
case "TSThisType": |
|
return "this"; |
|
|
|
case "TSImportType": |
|
return concat$d([!n.isTypeOf ? "" : "typeof ", "import(", path.call(print, "parameter"), ")", !n.qualifier ? "" : concat$d([".", path.call(print, "qualifier")]), printTypeParameters(path, options, print, "typeParameters")]); |
|
|
|
case "TSLiteralType": |
|
return path.call(print, "literal"); |
|
|
|
case "TSIndexedAccessType": |
|
return concat$d([path.call(print, "objectType"), "[", path.call(print, "indexType"), "]"]); |
|
|
|
case "TSConstructSignatureDeclaration": |
|
case "TSCallSignatureDeclaration": |
|
case "TSConstructorType": |
|
{ |
|
if (n.type !== "TSCallSignatureDeclaration") { |
|
parts.push("new "); |
|
} |
|
|
|
parts.push(group$b(printFunctionParams(path, print, options, |
|
/* expandArg */ |
|
false, |
|
/* printTypeParams */ |
|
true))); |
|
|
|
if (n.returnType) { |
|
var isType = n.type === "TSConstructorType"; |
|
parts.push(isType ? " => " : ": ", path.call(print, "returnType")); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
case "TSTypeOperator": |
|
return concat$d([n.operator, " ", path.call(print, "typeAnnotation")]); |
|
|
|
case "TSMappedType": |
|
{ |
|
var _shouldBreak3 = hasNewlineInRange$2(options.originalText, options.locStart(n), options.locEnd(n)); |
|
|
|
return group$b(concat$d(["{", indent$7(concat$d([options.bracketSpacing ? line$7 : softline$6, n.readonly ? concat$d([getTypeScriptMappedTypeModifier$1(n.readonly, "readonly"), " "]) : "", printTypeScriptModifiers(path, options, print), path.call(print, "typeParameter"), n.optional ? getTypeScriptMappedTypeModifier$1(n.optional, "?") : "", ": ", path.call(print, "typeAnnotation"), ifBreak$6(semi, "")])), comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true), options.bracketSpacing ? line$7 : softline$6, "}"]), { |
|
shouldBreak: _shouldBreak3 |
|
}); |
|
} |
|
|
|
case "TSMethodSignature": |
|
parts.push(n.accessibility ? concat$d([n.accessibility, " "]) : "", n.export ? "export " : "", n.static ? "static " : "", n.readonly ? "readonly " : "", n.computed ? "[" : "", path.call(print, "key"), n.computed ? "]" : "", printOptionalToken(path), printFunctionParams(path, print, options, |
|
/* expandArg */ |
|
false, |
|
/* printTypeParams */ |
|
true)); |
|
|
|
if (n.returnType) { |
|
parts.push(": ", path.call(print, "returnType")); |
|
} |
|
|
|
return group$b(concat$d(parts)); |
|
|
|
case "TSNamespaceExportDeclaration": |
|
parts.push("export as namespace ", path.call(print, "id")); |
|
|
|
if (options.semi) { |
|
parts.push(";"); |
|
} |
|
|
|
return group$b(concat$d(parts)); |
|
|
|
case "TSEnumDeclaration": |
|
if (n.declare) { |
|
parts.push("declare "); |
|
} |
|
|
|
if (n.modifiers) { |
|
parts.push(printTypeScriptModifiers(path, options, print)); |
|
} |
|
|
|
if (n.const) { |
|
parts.push("const "); |
|
} |
|
|
|
parts.push("enum ", path.call(print, "id"), " "); |
|
|
|
if (n.members.length === 0) { |
|
parts.push(group$b(concat$d(["{", comments.printDanglingComments(path, options), softline$6, "}"]))); |
|
} else { |
|
parts.push(group$b(concat$d(["{", indent$7(concat$d([hardline$9, printArrayItems(path, options, "members", print), shouldPrintComma$1(options, "es5") ? "," : ""])), comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true), hardline$9, "}"]))); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "TSEnumMember": |
|
parts.push(path.call(print, "id")); |
|
|
|
if (n.initializer) { |
|
parts.push(" = ", path.call(print, "initializer")); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "TSImportEqualsDeclaration": |
|
if (n.isExport) { |
|
parts.push("export "); |
|
} |
|
|
|
parts.push("import ", path.call(print, "id"), " = ", path.call(print, "moduleReference")); |
|
|
|
if (options.semi) { |
|
parts.push(";"); |
|
} |
|
|
|
return group$b(concat$d(parts)); |
|
|
|
case "TSExternalModuleReference": |
|
return concat$d(["require(", path.call(print, "expression"), ")"]); |
|
|
|
case "TSModuleDeclaration": |
|
{ |
|
var _parent12 = path.getParentNode(); |
|
|
|
var isExternalModule = isLiteral$1(n.id); |
|
var parentIsDeclaration = _parent12.type === "TSModuleDeclaration"; |
|
var bodyIsDeclaration = n.body && n.body.type === "TSModuleDeclaration"; |
|
|
|
if (parentIsDeclaration) { |
|
parts.push("."); |
|
} else { |
|
if (n.declare) { |
|
parts.push("declare "); |
|
} |
|
|
|
parts.push(printTypeScriptModifiers(path, options, print)); |
|
var textBetweenNodeAndItsId = options.originalText.slice(options.locStart(n), options.locStart(n.id)); // Global declaration looks like this: |
|
// (declare)? global { ... } |
|
|
|
var isGlobalDeclaration = n.id.type === "Identifier" && n.id.name === "global" && !/namespace|module/.test(textBetweenNodeAndItsId); |
|
|
|
if (!isGlobalDeclaration) { |
|
parts.push(isExternalModule || /(^|\s)module(\s|$)/.test(textBetweenNodeAndItsId) ? "module " : "namespace "); |
|
} |
|
} |
|
|
|
parts.push(path.call(print, "id")); |
|
|
|
if (bodyIsDeclaration) { |
|
parts.push(path.call(print, "body")); |
|
} else if (n.body) { |
|
parts.push(" ", group$b(path.call(print, "body"))); |
|
} else { |
|
parts.push(semi); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
case "PrivateName": |
|
return concat$d(["#", path.call(print, "id")]); |
|
|
|
case "TSConditionalType": |
|
return printTernaryOperator(path, options, print, { |
|
beforeParts: function beforeParts() { |
|
return [path.call(print, "checkType"), " ", "extends", " ", path.call(print, "extendsType")]; |
|
}, |
|
afterParts: function afterParts() { |
|
return []; |
|
}, |
|
shouldCheckJsx: false, |
|
conditionalNodeType: "TSConditionalType", |
|
consequentNodePropertyName: "trueType", |
|
alternateNodePropertyName: "falseType", |
|
testNodePropertyName: "checkType", |
|
breakNested: true |
|
}); |
|
|
|
case "TSInferType": |
|
return concat$d(["infer", " ", path.call(print, "typeParameter")]); |
|
|
|
case "InterpreterDirective": |
|
parts.push("#!", n.value, hardline$9); |
|
|
|
if (isNextLineEmpty$4(options.originalText, n, options)) { |
|
parts.push(hardline$9); |
|
} |
|
|
|
return concat$d(parts); |
|
|
|
case "NGRoot": |
|
return concat$d([].concat(path.call(print, "node"), !n.node.comments || n.node.comments.length === 0 ? [] : concat$d([" //", n.node.comments[0].value.trimRight()]))); |
|
|
|
case "NGChainedExpression": |
|
return group$b(join$9(concat$d([";", line$7]), path.map(function (childPath) { |
|
return hasNgSideEffect$1(childPath) ? print(childPath) : concat$d(["(", print(childPath), ")"]); |
|
}, "expressions"))); |
|
|
|
case "NGEmptyExpression": |
|
return ""; |
|
|
|
case "NGQuotedExpression": |
|
return concat$d([n.prefix, ":", n.value]); |
|
|
|
case "NGMicrosyntax": |
|
return concat$d(path.map(function (childPath, index) { |
|
return concat$d([index === 0 ? "" : isNgForOf$1(childPath.getValue(), index, n) ? " " : concat$d([";", line$7]), print(childPath)]); |
|
}, "body")); |
|
|
|
case "NGMicrosyntaxKey": |
|
return /^[a-z_$][a-z0-9_$]*(-[a-z_$][a-z0-9_$])*$/i.test(n.name) ? n.name : JSON.stringify(n.name); |
|
|
|
case "NGMicrosyntaxExpression": |
|
return concat$d([path.call(print, "expression"), n.alias === null ? "" : concat$d([" as ", path.call(print, "alias")])]); |
|
|
|
case "NGMicrosyntaxKeyedExpression": |
|
{ |
|
var index = path.getName(); |
|
|
|
var _parentNode2 = path.getParentNode(); |
|
|
|
var shouldNotPrintColon = isNgForOf$1(n, index, _parentNode2) || (index === 1 && (n.key.name === "then" || n.key.name === "else") || index === 2 && n.key.name === "else" && _parentNode2.body[index - 1].type === "NGMicrosyntaxKeyedExpression" && _parentNode2.body[index - 1].key.name === "then") && _parentNode2.body[0].type === "NGMicrosyntaxExpression"; |
|
return concat$d([path.call(print, "key"), shouldNotPrintColon ? " " : ": ", path.call(print, "expression")]); |
|
} |
|
|
|
case "NGMicrosyntaxLet": |
|
return concat$d(["let ", path.call(print, "key"), n.value === null ? "" : concat$d([" = ", path.call(print, "value")])]); |
|
|
|
case "NGMicrosyntaxAs": |
|
return concat$d([path.call(print, "key"), " as ", path.call(print, "alias")]); |
|
|
|
case "ArgumentPlaceholder": |
|
return "?"; |
|
|
|
default: |
|
/* istanbul ignore next */ |
|
throw new Error("unknown type: " + JSON.stringify(n.type)); |
|
} |
|
} |
|
|
|
function printStatementSequence(path, options, print) { |
|
var printed = []; |
|
var bodyNode = path.getNode(); |
|
var isClass = bodyNode.type === "ClassBody"; |
|
path.map(function (stmtPath, i) { |
|
var stmt = stmtPath.getValue(); // Just in case the AST has been modified to contain falsy |
|
// "statements," it's safer simply to skip them. |
|
|
|
/* istanbul ignore if */ |
|
|
|
if (!stmt) { |
|
return; |
|
} // Skip printing EmptyStatement nodes to avoid leaving stray |
|
// semicolons lying around. |
|
|
|
|
|
if (stmt.type === "EmptyStatement") { |
|
return; |
|
} |
|
|
|
var stmtPrinted = print(stmtPath); |
|
var text = options.originalText; |
|
var parts = []; // in no-semi mode, prepend statement with semicolon if it might break ASI |
|
// don't prepend the only JSX element in a program with semicolon |
|
|
|
if (!options.semi && !isClass && !isTheOnlyJSXElementInMarkdown$1(options, stmtPath) && stmtNeedsASIProtection(stmtPath, options)) { |
|
if (stmt.comments && stmt.comments.some(function (comment) { |
|
return comment.leading; |
|
})) { |
|
parts.push(print(stmtPath, { |
|
needsSemi: true |
|
})); |
|
} else { |
|
parts.push(";", stmtPrinted); |
|
} |
|
} else { |
|
parts.push(stmtPrinted); |
|
} |
|
|
|
if (!options.semi && isClass) { |
|
if (classPropMayCauseASIProblems$1(stmtPath)) { |
|
parts.push(";"); |
|
} else if (stmt.type === "ClassProperty") { |
|
var nextChild = bodyNode.body[i + 1]; |
|
|
|
if (classChildNeedsASIProtection$1(nextChild)) { |
|
parts.push(";"); |
|
} |
|
} |
|
} |
|
|
|
if (isNextLineEmpty$4(text, stmt, options) && !isLastStatement$1(stmtPath)) { |
|
parts.push(hardline$9); |
|
} |
|
|
|
printed.push(concat$d(parts)); |
|
}); |
|
return join$9(hardline$9, printed); |
|
} |
|
|
|
function printPropertyKey(path, options, print) { |
|
var node = path.getNode(); |
|
|
|
if (node.computed) { |
|
return concat$d(["[", path.call(print, "key"), "]"]); |
|
} |
|
|
|
var parent = path.getParentNode(); |
|
var key = node.key; |
|
|
|
if (options.quoteProps === "consistent" && !needsQuoteProps.has(parent)) { |
|
var objectHasStringProp = (parent.properties || parent.body || parent.members).some(function (prop) { |
|
return !prop.computed && prop.key && isStringLiteral$1(prop.key) && !isStringPropSafeToCoerceToIdentifier$1(prop, options); |
|
}); |
|
needsQuoteProps.set(parent, objectHasStringProp); |
|
} |
|
|
|
if (key.type === "Identifier" && (options.parser === "json" || options.quoteProps === "consistent" && needsQuoteProps.get(parent))) { |
|
// a -> "a" |
|
var prop = printString$2(JSON.stringify(key.name), options); |
|
return path.call(function (keyPath) { |
|
return comments.printComments(keyPath, function () { |
|
return prop; |
|
}, options); |
|
}, "key"); |
|
} |
|
|
|
if (isStringPropSafeToCoerceToIdentifier$1(node, options) && (options.quoteProps === "as-needed" || options.quoteProps === "consistent" && !needsQuoteProps.get(parent))) { |
|
// 'a' -> a |
|
return path.call(function (keyPath) { |
|
return comments.printComments(keyPath, function () { |
|
return key.value; |
|
}, options); |
|
}, "key"); |
|
} |
|
|
|
return path.call(print, "key"); |
|
} |
|
|
|
function printMethod(path, options, print) { |
|
var node = path.getNode(); |
|
var kind = node.kind; |
|
var value = node.value || node; |
|
var parts = []; |
|
|
|
if (!kind || kind === "init" || kind === "method" || kind === "constructor") { |
|
if (value.async) { |
|
parts.push("async "); |
|
} |
|
|
|
if (value.generator) { |
|
parts.push("*"); |
|
} |
|
} else { |
|
assert.ok(kind === "get" || kind === "set"); |
|
parts.push(kind, " "); |
|
} |
|
|
|
parts.push(printPropertyKey(path, options, print), node.optional || node.key.optional ? "?" : "", node === value ? printMethodInternal(path, options, print) : path.call(function (path) { |
|
return printMethodInternal(path, options, print); |
|
}, "value")); |
|
return concat$d(parts); |
|
} |
|
|
|
function printMethodInternal(path, options, print) { |
|
var parts = [printFunctionTypeParameters(path, options, print), group$b(concat$d([printFunctionParams(path, print, options), printReturnType(path, print, options)]))]; |
|
|
|
if (path.getNode().body) { |
|
parts.push(" ", path.call(print, "body")); |
|
} else { |
|
parts.push(options.semi ? ";" : ""); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
function couldGroupArg(arg) { |
|
return arg.type === "ObjectExpression" && (arg.properties.length > 0 || arg.comments) || arg.type === "ArrayExpression" && (arg.elements.length > 0 || arg.comments) || arg.type === "TSTypeAssertion" && couldGroupArg(arg.expression) || arg.type === "TSAsExpression" && couldGroupArg(arg.expression) || arg.type === "FunctionExpression" || arg.type === "ArrowFunctionExpression" && ( // we want to avoid breaking inside composite return types but not simple keywords |
|
// https://github.com/prettier/prettier/issues/4070 |
|
// export class Thing implements OtherThing { |
|
// do: (type: Type) => Provider<Prop> = memoize( |
|
// (type: ObjectType): Provider<Opts> => {} |
|
// ); |
|
// } |
|
// https://github.com/prettier/prettier/issues/6099 |
|
// app.get("/", (req, res): void => { |
|
// res.send("Hello World!"); |
|
// }); |
|
!arg.returnType || !arg.returnType.typeAnnotation || arg.returnType.typeAnnotation.type !== "TSTypeReference") && (arg.body.type === "BlockStatement" || arg.body.type === "ArrowFunctionExpression" || arg.body.type === "ObjectExpression" || arg.body.type === "ArrayExpression" || arg.body.type === "CallExpression" || arg.body.type === "OptionalCallExpression" || arg.body.type === "ConditionalExpression" || isJSXNode$1(arg.body)); |
|
} |
|
|
|
function shouldGroupLastArg(args) { |
|
var lastArg = getLast$2(args); |
|
var penultimateArg = getPenultimate$1(args); |
|
return !hasLeadingComment$3(lastArg) && !hasTrailingComment$1(lastArg) && couldGroupArg(lastArg) && ( // If the last two arguments are of the same type, |
|
// disable last element expansion. |
|
!penultimateArg || penultimateArg.type !== lastArg.type); |
|
} |
|
|
|
function shouldGroupFirstArg(args) { |
|
if (args.length !== 2) { |
|
return false; |
|
} |
|
|
|
var firstArg = args[0]; |
|
var secondArg = args[1]; |
|
return (!firstArg.comments || !firstArg.comments.length) && (firstArg.type === "FunctionExpression" || firstArg.type === "ArrowFunctionExpression" && firstArg.body.type === "BlockStatement") && secondArg.type !== "FunctionExpression" && secondArg.type !== "ArrowFunctionExpression" && secondArg.type !== "ConditionalExpression" && !couldGroupArg(secondArg); |
|
} |
|
|
|
function printJestEachTemplateLiteral(node, expressions, options) { |
|
/** |
|
* a | b | expected |
|
* ${1} | ${1} | ${2} |
|
* ${1} | ${2} | ${3} |
|
* ${2} | ${1} | ${3} |
|
*/ |
|
var headerNames = node.quasis[0].value.raw.trim().split(/\s*\|\s*/); |
|
|
|
if (headerNames.length > 1 || headerNames.some(function (headerName) { |
|
return headerName.length !== 0; |
|
})) { |
|
var parts = []; |
|
var stringifiedExpressions = expressions.map(function (doc) { |
|
return "${" + printDocToString$1(doc, Object.assign({}, options, { |
|
printWidth: Infinity, |
|
endOfLine: "lf" |
|
})).formatted + "}"; |
|
}); |
|
var tableBody = [{ |
|
hasLineBreak: false, |
|
cells: [] |
|
}]; |
|
|
|
for (var i = 1; i < node.quasis.length; i++) { |
|
var row = tableBody[tableBody.length - 1]; |
|
var correspondingExpression = stringifiedExpressions[i - 1]; |
|
row.cells.push(correspondingExpression); |
|
|
|
if (correspondingExpression.indexOf("\n") !== -1) { |
|
row.hasLineBreak = true; |
|
} |
|
|
|
if (node.quasis[i].value.raw.indexOf("\n") !== -1) { |
|
tableBody.push({ |
|
hasLineBreak: false, |
|
cells: [] |
|
}); |
|
} |
|
} |
|
|
|
var maxColumnCount = tableBody.reduce(function (maxColumnCount, row) { |
|
return Math.max(maxColumnCount, row.cells.length); |
|
}, headerNames.length); |
|
var maxColumnWidths = Array.from(new Array(maxColumnCount), function () { |
|
return 0; |
|
}); |
|
var table = [{ |
|
cells: headerNames |
|
}].concat(tableBody.filter(function (row) { |
|
return row.cells.length !== 0; |
|
})); |
|
table.filter(function (row) { |
|
return !row.hasLineBreak; |
|
}).forEach(function (row) { |
|
row.cells.forEach(function (cell, index) { |
|
maxColumnWidths[index] = Math.max(maxColumnWidths[index], getStringWidth$2(cell)); |
|
}); |
|
}); |
|
parts.push(lineSuffixBoundary$1, "`", indent$7(concat$d([hardline$9, join$9(hardline$9, table.map(function (row) { |
|
return join$9(" | ", row.cells.map(function (cell, index) { |
|
return row.hasLineBreak ? cell : cell + " ".repeat(maxColumnWidths[index] - getStringWidth$2(cell)); |
|
})); |
|
}))])), hardline$9, "`"); |
|
return concat$d(parts); |
|
} |
|
} |
|
|
|
function printArgumentsList(path, options, print) { |
|
var node = path.getValue(); |
|
var args = node.arguments; |
|
|
|
if (args.length === 0) { |
|
return concat$d(["(", comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true), ")"]); |
|
} // useEffect(() => { ... }, [foo, bar, baz]) |
|
|
|
|
|
if (args.length === 2 && args[0].type === "ArrowFunctionExpression" && args[0].params.length === 0 && args[0].body.type === "BlockStatement" && args[1].type === "ArrayExpression" && !args.find(function (arg) { |
|
return arg.comments; |
|
})) { |
|
return concat$d(["(", path.call(print, "arguments", 0), ", ", path.call(print, "arguments", 1), ")"]); |
|
} // func( |
|
// ({ |
|
// a, |
|
// b |
|
// }) => {} |
|
// ); |
|
|
|
|
|
function shouldBreakForArrowFunctionInArguments(arg, argPath) { |
|
if (!arg || arg.type !== "ArrowFunctionExpression" || !arg.body || arg.body.type !== "BlockStatement" || !arg.params || arg.params.length < 1) { |
|
return false; |
|
} |
|
|
|
var shouldBreak = false; |
|
argPath.each(function (paramPath) { |
|
var printed = concat$d([print(paramPath)]); |
|
shouldBreak = shouldBreak || willBreak$1(printed); |
|
}, "params"); |
|
return shouldBreak; |
|
} |
|
|
|
var anyArgEmptyLine = false; |
|
var shouldBreakForArrowFunction = false; |
|
var hasEmptyLineFollowingFirstArg = false; |
|
var lastArgIndex = args.length - 1; |
|
var printedArguments = path.map(function (argPath, index) { |
|
var arg = argPath.getNode(); |
|
var parts = [print(argPath)]; |
|
|
|
if (index === lastArgIndex) ; else if (isNextLineEmpty$4(options.originalText, arg, options)) { |
|
if (index === 0) { |
|
hasEmptyLineFollowingFirstArg = true; |
|
} |
|
|
|
anyArgEmptyLine = true; |
|
parts.push(",", hardline$9, hardline$9); |
|
} else { |
|
parts.push(",", line$7); |
|
} |
|
|
|
shouldBreakForArrowFunction = shouldBreakForArrowFunctionInArguments(arg, argPath); |
|
return concat$d(parts); |
|
}, "arguments"); |
|
var maybeTrailingComma = // Dynamic imports cannot have trailing commas |
|
!(node.callee && node.callee.type === "Import") && shouldPrintComma$1(options, "all") ? "," : ""; |
|
|
|
function allArgsBrokenOut() { |
|
return group$b(concat$d(["(", indent$7(concat$d([line$7, concat$d(printedArguments)])), maybeTrailingComma, line$7, ")"]), { |
|
shouldBreak: true |
|
}); |
|
} |
|
|
|
if (isFunctionCompositionArgs$1(args)) { |
|
return allArgsBrokenOut(); |
|
} |
|
|
|
var shouldGroupFirst = shouldGroupFirstArg(args); |
|
var shouldGroupLast = shouldGroupLastArg(args); |
|
|
|
if (shouldGroupFirst || shouldGroupLast) { |
|
var shouldBreak = (shouldGroupFirst ? printedArguments.slice(1).some(willBreak$1) : printedArguments.slice(0, -1).some(willBreak$1)) || anyArgEmptyLine || shouldBreakForArrowFunction; // We want to print the last argument with a special flag |
|
|
|
var printedExpanded; |
|
var i = 0; |
|
path.each(function (argPath) { |
|
if (shouldGroupFirst && i === 0) { |
|
printedExpanded = [concat$d([argPath.call(function (p) { |
|
return print(p, { |
|
expandFirstArg: true |
|
}); |
|
}), printedArguments.length > 1 ? "," : "", hasEmptyLineFollowingFirstArg ? hardline$9 : line$7, hasEmptyLineFollowingFirstArg ? hardline$9 : ""])].concat(printedArguments.slice(1)); |
|
} |
|
|
|
if (shouldGroupLast && i === args.length - 1) { |
|
printedExpanded = printedArguments.slice(0, -1).concat(argPath.call(function (p) { |
|
return print(p, { |
|
expandLastArg: true |
|
}); |
|
})); |
|
} |
|
|
|
i++; |
|
}, "arguments"); |
|
var somePrintedArgumentsWillBreak = printedArguments.some(willBreak$1); |
|
var simpleConcat = concat$d(["(", concat$d(printedExpanded), ")"]); |
|
return concat$d([somePrintedArgumentsWillBreak ? breakParent$3 : "", conditionalGroup$1([!somePrintedArgumentsWillBreak && !node.typeArguments && !node.typeParameters ? simpleConcat : ifBreak$6(allArgsBrokenOut(), simpleConcat), shouldGroupFirst ? concat$d(["(", group$b(printedExpanded[0], { |
|
shouldBreak: true |
|
}), concat$d(printedExpanded.slice(1)), ")"]) : concat$d(["(", concat$d(printedArguments.slice(0, -1)), group$b(getLast$2(printedExpanded), { |
|
shouldBreak: true |
|
}), ")"]), allArgsBrokenOut()], { |
|
shouldBreak: shouldBreak |
|
})]); |
|
} |
|
|
|
var contents = concat$d(["(", indent$7(concat$d([softline$6, concat$d(printedArguments)])), ifBreak$6(maybeTrailingComma), softline$6, ")"]); |
|
|
|
if (isLongCurriedCallExpression$1(path)) { |
|
// By not wrapping the arguments in a group, the printer prioritizes |
|
// breaking up these arguments rather than the args of the parent call. |
|
return contents; |
|
} |
|
|
|
return group$b(contents, { |
|
shouldBreak: printedArguments.some(willBreak$1) || anyArgEmptyLine |
|
}); |
|
} |
|
|
|
function printTypeAnnotation(path, options, print) { |
|
var node = path.getValue(); |
|
|
|
if (!node.typeAnnotation) { |
|
return ""; |
|
} |
|
|
|
var parentNode = path.getParentNode(); |
|
var isDefinite = node.definite || parentNode && parentNode.type === "VariableDeclarator" && parentNode.definite; |
|
var isFunctionDeclarationIdentifier = parentNode.type === "DeclareFunction" && parentNode.id === node; |
|
|
|
if (isFlowAnnotationComment$1(options.originalText, node.typeAnnotation, options)) { |
|
return concat$d([" /*: ", path.call(print, "typeAnnotation"), " */"]); |
|
} |
|
|
|
return concat$d([isFunctionDeclarationIdentifier ? "" : isDefinite ? "!: " : ": ", path.call(print, "typeAnnotation")]); |
|
} |
|
|
|
function printFunctionTypeParameters(path, options, print) { |
|
var fun = path.getValue(); |
|
|
|
if (fun.typeArguments) { |
|
return path.call(print, "typeArguments"); |
|
} |
|
|
|
if (fun.typeParameters) { |
|
return path.call(print, "typeParameters"); |
|
} |
|
|
|
return ""; |
|
} |
|
|
|
function printFunctionParams(path, print, options, expandArg, printTypeParams) { |
|
var fun = path.getValue(); |
|
var parent = path.getParentNode(); |
|
var paramsField = fun.parameters ? "parameters" : "params"; |
|
var isParametersInTestCall = isTestCall$1(parent); |
|
var shouldHugParameters = shouldHugArguments(fun); |
|
var shouldExpandParameters = expandArg && !(fun[paramsField] && fun[paramsField].some(function (n) { |
|
return n.comments; |
|
})); |
|
var typeParams = printTypeParams ? printFunctionTypeParameters(path, options, print) : ""; |
|
var printed = []; |
|
|
|
if (fun[paramsField]) { |
|
var lastArgIndex = fun[paramsField].length - 1; |
|
printed = path.map(function (childPath, index) { |
|
var parts = []; |
|
var param = childPath.getValue(); |
|
parts.push(print(childPath)); |
|
|
|
if (index === lastArgIndex) { |
|
if (fun.rest) { |
|
parts.push(",", line$7); |
|
} |
|
} else if (isParametersInTestCall || shouldHugParameters || shouldExpandParameters) { |
|
parts.push(", "); |
|
} else if (isNextLineEmpty$4(options.originalText, param, options)) { |
|
parts.push(",", hardline$9, hardline$9); |
|
} else { |
|
parts.push(",", line$7); |
|
} |
|
|
|
return concat$d(parts); |
|
}, paramsField); |
|
} |
|
|
|
if (fun.rest) { |
|
printed.push(concat$d(["...", path.call(print, "rest")])); |
|
} |
|
|
|
if (printed.length === 0) { |
|
return concat$d([typeParams, "(", comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true, function (comment) { |
|
return getNextNonSpaceNonCommentCharacter$1(options.originalText, comment, options.locEnd) === ")"; |
|
}), ")"]); |
|
} |
|
|
|
var lastParam = getLast$2(fun[paramsField]); // If the parent is a call with the first/last argument expansion and this is the |
|
// params of the first/last argument, we don't want the arguments to break and instead |
|
// want the whole expression to be on a new line. |
|
// |
|
// Good: Bad: |
|
// verylongcall( verylongcall(( |
|
// (a, b) => { a, |
|
// } b, |
|
// }) ) => { |
|
// }) |
|
|
|
if (shouldExpandParameters) { |
|
return group$b(concat$d([removeLines$2(typeParams), "(", concat$d(printed.map(removeLines$2)), ")"])); |
|
} // Single object destructuring should hug |
|
// |
|
// function({ |
|
// a, |
|
// b, |
|
// c |
|
// }) {} |
|
|
|
|
|
var hasNotParameterDecorator = fun[paramsField].every(function (param) { |
|
return !param.decorators; |
|
}); |
|
|
|
if (shouldHugParameters && hasNotParameterDecorator) { |
|
return concat$d([typeParams, "(", concat$d(printed), ")"]); |
|
} // don't break in specs, eg; `it("should maintain parens around done even when long", (done) => {})` |
|
|
|
|
|
if (isParametersInTestCall) { |
|
return concat$d([typeParams, "(", concat$d(printed), ")"]); |
|
} |
|
|
|
var isFlowShorthandWithOneArg = (isObjectTypePropertyAFunction$1(parent, options) || isTypeAnnotationAFunction$1(parent, options) || parent.type === "TypeAlias" || parent.type === "UnionTypeAnnotation" || parent.type === "TSUnionType" || parent.type === "IntersectionTypeAnnotation" || parent.type === "FunctionTypeAnnotation" && parent.returnType === fun) && fun[paramsField].length === 1 && fun[paramsField][0].name === null && fun[paramsField][0].typeAnnotation && fun.typeParameters === null && isSimpleFlowType$1(fun[paramsField][0].typeAnnotation) && !fun.rest; |
|
|
|
if (isFlowShorthandWithOneArg) { |
|
if (options.arrowParens === "always") { |
|
return concat$d(["(", concat$d(printed), ")"]); |
|
} |
|
|
|
return concat$d(printed); |
|
} |
|
|
|
var canHaveTrailingComma = !(lastParam && lastParam.type === "RestElement") && !fun.rest; |
|
return concat$d([typeParams, "(", indent$7(concat$d([softline$6, concat$d(printed)])), ifBreak$6(canHaveTrailingComma && shouldPrintComma$1(options, "all") ? "," : ""), softline$6, ")"]); |
|
} |
|
|
|
function shouldPrintParamsWithoutParens(path, options) { |
|
if (options.arrowParens === "always") { |
|
return false; |
|
} |
|
|
|
if (options.arrowParens === "avoid") { |
|
var node = path.getValue(); |
|
return canPrintParamsWithoutParens(node); |
|
} // Fallback default; should be unreachable |
|
|
|
|
|
return false; |
|
} |
|
|
|
function canPrintParamsWithoutParens(node) { |
|
return node.params.length === 1 && !node.rest && !node.typeParameters && !hasDanglingComments$1(node) && node.params[0].type === "Identifier" && !node.params[0].typeAnnotation && !node.params[0].comments && !node.params[0].optional && !node.predicate && !node.returnType; |
|
} |
|
|
|
function printFunctionDeclaration(path, print, options) { |
|
var n = path.getValue(); |
|
var parts = []; |
|
|
|
if (n.async) { |
|
parts.push("async "); |
|
} |
|
|
|
parts.push("function"); |
|
|
|
if (n.generator) { |
|
parts.push("*"); |
|
} |
|
|
|
if (n.id) { |
|
parts.push(" ", path.call(print, "id")); |
|
} |
|
|
|
parts.push(printFunctionTypeParameters(path, options, print), group$b(concat$d([printFunctionParams(path, print, options), printReturnType(path, print, options)])), n.body ? " " : "", path.call(print, "body")); |
|
return concat$d(parts); |
|
} |
|
|
|
function printReturnType(path, print, options) { |
|
var n = path.getValue(); |
|
var returnType = path.call(print, "returnType"); |
|
|
|
if (n.returnType && isFlowAnnotationComment$1(options.originalText, n.returnType, options)) { |
|
return concat$d([" /*: ", returnType, " */"]); |
|
} |
|
|
|
var parts = [returnType]; // prepend colon to TypeScript type annotation |
|
|
|
if (n.returnType && n.returnType.typeAnnotation) { |
|
parts.unshift(": "); |
|
} |
|
|
|
if (n.predicate) { |
|
// The return type will already add the colon, but otherwise we |
|
// need to do it ourselves |
|
parts.push(n.returnType ? " " : ": ", path.call(print, "predicate")); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
function printExportDeclaration(path, options, print) { |
|
var decl = path.getValue(); |
|
var semi = options.semi ? ";" : ""; |
|
var parts = ["export "]; |
|
var isDefault = decl["default"] || decl.type === "ExportDefaultDeclaration"; |
|
|
|
if (isDefault) { |
|
parts.push("default "); |
|
} |
|
|
|
parts.push(comments.printDanglingComments(path, options, |
|
/* sameIndent */ |
|
true)); |
|
|
|
if (needsHardlineAfterDanglingComment$1(decl)) { |
|
parts.push(hardline$9); |
|
} |
|
|
|
if (decl.declaration) { |
|
parts.push(path.call(print, "declaration")); |
|
|
|
if (isDefault && decl.declaration.type !== "ClassDeclaration" && decl.declaration.type !== "FunctionDeclaration" && decl.declaration.type !== "TSInterfaceDeclaration" && decl.declaration.type !== "DeclareClass" && decl.declaration.type !== "DeclareFunction" && decl.declaration.type !== "TSDeclareFunction") { |
|
parts.push(semi); |
|
} |
|
} else { |
|
if (decl.specifiers && decl.specifiers.length > 0) { |
|
var specifiers = []; |
|
var defaultSpecifiers = []; |
|
var namespaceSpecifiers = []; |
|
path.each(function (specifierPath) { |
|
var specifierType = path.getValue().type; |
|
|
|
if (specifierType === "ExportSpecifier") { |
|
specifiers.push(print(specifierPath)); |
|
} else if (specifierType === "ExportDefaultSpecifier") { |
|
defaultSpecifiers.push(print(specifierPath)); |
|
} else if (specifierType === "ExportNamespaceSpecifier") { |
|
namespaceSpecifiers.push(concat$d(["* as ", print(specifierPath)])); |
|
} |
|
}, "specifiers"); |
|
var isNamespaceFollowed = namespaceSpecifiers.length !== 0 && specifiers.length !== 0; |
|
var isDefaultFollowed = defaultSpecifiers.length !== 0 && (namespaceSpecifiers.length !== 0 || specifiers.length !== 0); |
|
var canBreak = specifiers.length > 1 || defaultSpecifiers.length > 0 || decl.specifiers && decl.specifiers.some(function (node) { |
|
return node.comments; |
|
}); |
|
var printed = ""; |
|
|
|
if (specifiers.length !== 0) { |
|
if (canBreak) { |
|
printed = group$b(concat$d(["{", indent$7(concat$d([options.bracketSpacing ? line$7 : softline$6, join$9(concat$d([",", line$7]), specifiers)])), ifBreak$6(shouldPrintComma$1(options) ? "," : ""), options.bracketSpacing ? line$7 : softline$6, "}"])); |
|
} else { |
|
printed = concat$d(["{", options.bracketSpacing ? " " : "", concat$d(specifiers), options.bracketSpacing ? " " : "", "}"]); |
|
} |
|
} |
|
|
|
parts.push(decl.exportKind === "type" ? "type " : "", concat$d(defaultSpecifiers), concat$d([isDefaultFollowed ? ", " : ""]), concat$d(namespaceSpecifiers), concat$d([isNamespaceFollowed ? ", " : ""]), printed); |
|
} else { |
|
parts.push("{}"); |
|
} |
|
|
|
if (decl.source) { |
|
parts.push(" from ", path.call(print, "source")); |
|
} |
|
|
|
parts.push(semi); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
function printFlowDeclaration(path, parts) { |
|
var parentExportDecl = getParentExportDeclaration$1(path); |
|
|
|
if (parentExportDecl) { |
|
assert.strictEqual(parentExportDecl.type, "DeclareExportDeclaration"); |
|
} else { |
|
// If the parent node has type DeclareExportDeclaration, then it |
|
// will be responsible for printing the "declare" token. Otherwise |
|
// it needs to be printed with this non-exported declaration node. |
|
parts.unshift("declare "); |
|
} |
|
|
|
return concat$d(parts); |
|
} |
|
|
|
function printTypeScriptModifiers(path, options, print) { |
|
var n = path.getValue(); |
|
|
|
if (!n.modifiers || !n.modifiers.length) { |
|
return ""; |
|
} |
|
|
|
return concat$d([join$9(" ", path.map(print, "modifiers")), " "]); |
|
} |
|
|
|
function printTypeParameters(path, options, print, paramsKey) { |
|
var n = path.getValue(); |
|
|
|
if (!n[paramsKey]) { |
|
return ""; |
|
} // for TypeParameterDeclaration typeParameters is a single node |
|
|
|
|
|
if (!Array.isArray(n[paramsKey])) { |
|
return path.call(print, paramsKey); |
|
} |
|
|
|
var grandparent = path.getNode(2); |
|
var greatGreatGrandParent = path.getNode(4); |
|
var isParameterInTestCall = grandparent != null && isTestCall$1(grandparent); |
|
var shouldInline = isParameterInTestCall || n[paramsKey].length === 0 || n[paramsKey].length === 1 && (shouldHugType(n[paramsKey][0]) || n[paramsKey][0].type === "GenericTypeAnnotation" && shouldHugType(n[paramsKey][0].id) || n[paramsKey][0].type === "TSTypeReference" && shouldHugType(n[paramsKey][0].typeName) || n[paramsKey][0].type === "NullableTypeAnnotation" || // See https://github.com/prettier/prettier/pull/6467 for the context. |
|
greatGreatGrandParent && greatGreatGrandParent.type === "VariableDeclarator" && grandparent && grandparent.type === "TSTypeAnnotation" && n[paramsKey][0].type !== "TSUnionType" && n[paramsKey][0].type !== "UnionTypeAnnotation" && n[paramsKey][0].type !== "TSConditionalType" && n[paramsKey][0].type !== "TSMappedType"); |
|
|
|
if (shouldInline) { |
|
return concat$d(["<", join$9(", ", path.map(print, paramsKey)), ">"]); |
|
} |
|
|
|
return group$b(concat$d(["<", indent$7(concat$d([softline$6, join$9(concat$d([",", line$7]), path.map(print, paramsKey))])), ifBreak$6(options.parser !== "typescript" && shouldPrintComma$1(options, "all") ? "," : ""), softline$6, ">"])); |
|
} |
|
|
|
function printClass(path, options, print) { |
|
var n = path.getValue(); |
|
var parts = []; |
|
|
|
if (n.abstract) { |
|
parts.push("abstract "); |
|
} |
|
|
|
parts.push("class"); |
|
|
|
if (n.id) { |
|
parts.push(" ", path.call(print, "id")); |
|
} |
|
|
|
parts.push(path.call(print, "typeParameters")); |
|
var partsGroup = []; |
|
|
|
if (n.superClass) { |
|
var printed = concat$d(["extends ", path.call(print, "superClass"), path.call(print, "superTypeParameters")]); // Keep old behaviour of extends in same line |
|
// If there is only on extends and there are not comments |
|
|
|
if ((!n.implements || n.implements.length === 0) && (!n.superClass.comments || n.superClass.comments.length === 0)) { |
|
parts.push(concat$d([" ", path.call(function (superClass) { |
|
return comments.printComments(superClass, function () { |
|
return printed; |
|
}, options); |
|
}, "superClass")])); |
|
} else { |
|
partsGroup.push(group$b(concat$d([line$7, path.call(function (superClass) { |
|
return comments.printComments(superClass, function () { |
|
return printed; |
|
}, options); |
|
}, "superClass")]))); |
|
} |
|
} else if (n.extends && n.extends.length > 0) { |
|
parts.push(" extends ", join$9(", ", path.map(print, "extends"))); |
|
} |
|
|
|
if (n["mixins"] && n["mixins"].length > 0) { |
|
partsGroup.push(line$7, "mixins ", group$b(indent$7(join$9(concat$d([",", line$7]), path.map(print, "mixins"))))); |
|
} |
|
|
|
if (n["implements"] && n["implements"].length > 0) { |
|
partsGroup.push(line$7, "implements", group$b(indent$7(concat$d([line$7, join$9(concat$d([",", line$7]), path.map(print, "implements"))])))); |
|
} |
|
|
|
if (partsGroup.length > 0) { |
|
parts.push(group$b(indent$7(concat$d(partsGroup)))); |
|
} |
|
|
|
if (n.body && n.body.comments && hasLeadingOwnLineComment$1(options.originalText, n.body, options)) { |
|
parts.push(hardline$9); |
|
} else { |
|
parts.push(" "); |
|
} |
|
|
|
parts.push(path.call(print, "body")); |
|
return parts; |
|
} |
|
|
|
function printOptionalToken(path) { |
|
var node = path.getValue(); |
|
|
|
if (!node.optional || // It's an optional computed method parsed by typescript-estree. |
|
// "?" is printed in `printMethod`. |
|
node.type === "Identifier" && node === path.getParentNode().key) { |
|
return ""; |
|
} |
|
|
|
if (node.type === "OptionalCallExpression" || node.type === "OptionalMemberExpression" && node.computed) { |
|
return "?."; |
|
} |
|
|
|
return "?"; |
|
} |
|
|
|
function printMemberLookup(path, options, print) { |
|
var property = path.call(print, "property"); |
|
var n = path.getValue(); |
|
var optional = printOptionalToken(path); |
|
|
|
if (!n.computed) { |
|
return concat$d([optional, ".", property]); |
|
} |
|
|
|
if (!n.property || isNumericLiteral$1(n.property)) { |
|
return concat$d([optional, "[", property, "]"]); |
|
} |
|
|
|
return group$b(concat$d([optional, "[", indent$7(concat$d([softline$6, property])), softline$6, "]"])); |
|
} |
|
|
|
function printBindExpressionCallee(path, options, print) { |
|
return concat$d(["::", path.call(print, "callee")]); |
|
} // We detect calls on member expressions specially to format a |
|
// common pattern better. The pattern we are looking for is this: |
|
// |
|
// arr |
|
// .map(x => x + 1) |
|
// .filter(x => x > 10) |
|
// .some(x => x % 2) |
|
// |
|
// The way it is structured in the AST is via a nested sequence of |
|
// MemberExpression and CallExpression. We need to traverse the AST |
|
// and make groups out of it to print it in the desired way. |
|
|
|
|
|
function printMemberChain(path, options, print) { |
|
// The first phase is to linearize the AST by traversing it down. |
|
// |
|
// a().b() |
|
// has the following AST structure: |
|
// CallExpression(MemberExpression(CallExpression(Identifier))) |
|
// and we transform it into |
|
// [Identifier, CallExpression, MemberExpression, CallExpression] |
|
var printedNodes = []; // Here we try to retain one typed empty line after each call expression or |
|
// the first group whether it is in parentheses or not |
|
|
|
function shouldInsertEmptyLineAfter(node) { |
|
var originalText = options.originalText; |
|
var nextCharIndex = getNextNonSpaceNonCommentCharacterIndex$2(originalText, node, options); |
|
var nextChar = originalText.charAt(nextCharIndex); // if it is cut off by a parenthesis, we only account for one typed empty |
|
// line after that parenthesis |
|
|
|
if (nextChar == ")") { |
|
return isNextLineEmptyAfterIndex$1(originalText, nextCharIndex + 1, options); |
|
} |
|
|
|
return isNextLineEmpty$4(originalText, node, options); |
|
} |
|
|
|
function rec(path) { |
|
var node = path.getValue(); |
|
|
|
if ((node.type === "CallExpression" || node.type === "OptionalCallExpression") && (isMemberish$1(node.callee) || node.callee.type === "CallExpression" || node.callee.type === "OptionalCallExpression")) { |
|
printedNodes.unshift({ |
|
node: node, |
|
printed: concat$d([comments.printComments(path, function () { |
|
return concat$d([printOptionalToken(path), printFunctionTypeParameters(path, options, print), printArgumentsList(path, options, print)]); |
|
}, options), shouldInsertEmptyLineAfter(node) ? hardline$9 : ""]) |
|
}); |
|
path.call(function (callee) { |
|
return rec(callee); |
|
}, "callee"); |
|
} else if (isMemberish$1(node)) { |
|
printedNodes.unshift({ |
|
node: node, |
|
needsParens: needsParens_1(path, options), |
|
printed: comments.printComments(path, function () { |
|
return node.type === "OptionalMemberExpression" || node.type === "MemberExpression" ? printMemberLookup(path, options, print) : printBindExpressionCallee(path, options, print); |
|
}, options) |
|
}); |
|
path.call(function (object) { |
|
return rec(object); |
|
}, "object"); |
|
} else if (node.type === "TSNonNullExpression") { |
|
printedNodes.unshift({ |
|
node: node, |
|
printed: comments.printComments(path, function () { |
|
return "!"; |
|
}, options) |
|
}); |
|
path.call(function (expression) { |
|
return rec(expression); |
|
}, "expression"); |
|
} else { |
|
printedNodes.unshift({ |
|
node: node, |
|
printed: path.call(print) |
|
}); |
|
} |
|
} // Note: the comments of the root node have already been printed, so we |
|
// need to extract this first call without printing them as they would |
|
// if handled inside of the recursive call. |
|
|
|
|
|
var node = path.getValue(); |
|
printedNodes.unshift({ |
|
node: node, |
|
printed: concat$d([printOptionalToken(path), printFunctionTypeParameters(path, options, print), printArgumentsList(path, options, print)]) |
|
}); |
|
path.call(function (callee) { |
|
return rec(callee); |
|
}, "callee"); // Once we have a linear list of printed nodes, we want to create groups out |
|
// of it. |
|
// |
|
// a().b.c().d().e |
|
// will be grouped as |
|
// [ |
|
// [Identifier, CallExpression], |
|
// [MemberExpression, MemberExpression, CallExpression], |
|
// [MemberExpression, CallExpression], |
|
// [MemberExpression], |
|
// ] |
|
// so that we can print it as |
|
// a() |
|
// .b.c() |
|
// .d() |
|
// .e |
|
// The first group is the first node followed by |
|
// - as many CallExpression as possible |
|
// < fn()()() >.something() |
|
// - as many array accessors as possible |
|
// < fn()[0][1][2] >.something() |
|
// - then, as many MemberExpression as possible but the last one |
|
// < this.items >.something() |
|
|
|
var groups = []; |
|
var currentGroup = [printedNodes[0]]; |
|
var i = 1; |
|
|
|
for (; i < printedNodes.length; ++i) { |
|
if (printedNodes[i].node.type === "TSNonNullExpression" || printedNodes[i].node.type === "OptionalCallExpression" || printedNodes[i].node.type === "CallExpression" || (printedNodes[i].node.type === "MemberExpression" || printedNodes[i].node.type === "OptionalMemberExpression") && printedNodes[i].node.computed && isNumericLiteral$1(printedNodes[i].node.property)) { |
|
currentGroup.push(printedNodes[i]); |
|
} else { |
|
break; |
|
} |
|
} |
|
|
|
if (printedNodes[0].node.type !== "CallExpression" && printedNodes[0].node.type !== "OptionalCallExpression") { |
|
for (; i + 1 < printedNodes.length; ++i) { |
|
if (isMemberish$1(printedNodes[i].node) && isMemberish$1(printedNodes[i + 1].node)) { |
|
currentGroup.push(printedNodes[i]); |
|
} else { |
|
break; |
|
} |
|
} |
|
} |
|
|
|
groups.push(currentGroup); |
|
currentGroup = []; // Then, each following group is a sequence of MemberExpression followed by |
|
// a sequence of CallExpression. To compute it, we keep adding things to the |
|
// group until we has seen a CallExpression in the past and reach a |
|
// MemberExpression |
|
|
|
var hasSeenCallExpression = false; |
|
|
|
for (; i < printedNodes.length; ++i) { |
|
if (hasSeenCallExpression && isMemberish$1(printedNodes[i].node)) { |
|
// [0] should be appended at the end of the group instead of the |
|
// beginning of the next one |
|
if (printedNodes[i].node.computed && isNumericLiteral$1(printedNodes[i].node.property)) { |
|
currentGroup.push(printedNodes[i]); |
|
continue; |
|
} |
|
|
|
groups.push(currentGroup); |
|
currentGroup = []; |
|
hasSeenCallExpression = false; |
|
} |
|
|
|
if (printedNodes[i].node.type === "CallExpression" || printedNodes[i].node.type === "OptionalCallExpression") { |
|
hasSeenCallExpression = true; |
|
} |
|
|
|
currentGroup.push(printedNodes[i]); |
|
|
|
if (printedNodes[i].node.comments && printedNodes[i].node.comments.some(function (comment) { |
|
return comment.trailing; |
|
})) { |
|
groups.push(currentGroup); |
|
currentGroup = []; |
|
hasSeenCallExpression = false; |
|
} |
|
} |
|
|
|
if (currentGroup.length > 0) { |
|
groups.push(currentGroup); |
|
} // There are cases like Object.keys(), Observable.of(), _.values() where |
|
// they are the subject of all the chained calls and therefore should |
|
// be kept on the same line: |
|
// |
|
// Object.keys(items) |
|
// .filter(x => x) |
|
// .map(x => x) |
|
// |
|
// In order to detect those cases, we use an heuristic: if the first |
|
// node is an identifier with the name starting with a capital |
|
// letter or just a sequence of _$. The rationale is that they are |
|
// likely to be factories. |
|
|
|
|
|
function isFactory(name) { |
|
return /^[A-Z]|^[_$]+$/.test(name); |
|
} // In case the Identifier is shorter than tab width, we can keep the |
|
// first call in a single line, if it's an ExpressionStatement. |
|
// |
|
// d3.scaleLinear() |
|
// .domain([0, 100]) |
|
// .range([0, width]); |
|
// |
|
|
|
|
|
function isShort(name) { |
|
return name.length <= options.tabWidth; |
|
} |
|
|
|
function shouldNotWrap(groups) { |
|
var parent = path.getParentNode(); |
|
var isExpression = parent && parent.type === "ExpressionStatement"; |
|
var hasComputed = groups[1].length && groups[1][0].node.computed; |
|
|
|
if (groups[0].length === 1) { |
|
var firstNode = groups[0][0].node; |
|
return firstNode.type === "ThisExpression" || firstNode.type === "Identifier" && (isFactory(firstNode.name) || isExpression && isShort(firstNode.name) || hasComputed); |
|
} |
|
|
|
var lastNode = getLast$2(groups[0]).node; |
|
return (lastNode.type === "MemberExpression" || lastNode.type === "OptionalMemberExpression") && lastNode.property.type === "Identifier" && (isFactory(lastNode.property.name) || hasComputed); |
|
} |
|
|
|
var shouldMerge = groups.length >= 2 && !groups[1][0].node.comments && shouldNotWrap(groups); |
|
|
|
function printGroup(printedGroup) { |
|
var printed = printedGroup.map(function (tuple) { |
|
return tuple.printed; |
|
}); // Checks if the last node (i.e. the parent node) needs parens and print |
|
// accordingly |
|
|
|
if (printedGroup.length > 0 && printedGroup[printedGroup.length - 1].needsParens) { |
|
return concat$d(["("].concat(_toConsumableArray(printed), [")"])); |
|
} |
|
|
|
return concat$d(printed); |
|
} |
|
|
|
function printIndentedGroup(groups) { |
|
if (groups.length === 0) { |
|
return ""; |
|
} |
|
|
|
return indent$7(group$b(concat$d([hardline$9, join$9(hardline$9, groups.map(printGroup))]))); |
|
} |
|
|
|
var printedGroups = groups.map(printGroup); |
|
var oneLine = concat$d(printedGroups); |
|
var cutoff = shouldMerge ? 3 : 2; |
|
var flatGroups = groups.slice(0, cutoff).reduce(function (res, group) { |
|
return res.concat(group); |
|
}, []); |
|
var hasComment = flatGroups.slice(1, -1).some(function (node) { |
|
return hasLeadingComment$3(node.node); |
|
}) || flatGroups.slice(0, -1).some(function (node) { |
|
return hasTrailingComment$1(node.node); |
|
}) || groups[cutoff] && hasLeadingComment$3(groups[cutoff][0].node); // If we only have a single `.`, we shouldn't do anything fancy and just |
|
// render everything concatenated together. |
|
|
|
if (groups.length <= cutoff && !hasComment) { |
|
if (isLongCurriedCallExpression$1(path)) { |
|
return oneLine; |
|
} |
|
|
|
return group$b(oneLine); |
|
} // Find out the last node in the first group and check if it has an |
|
// empty line after |
|
|
|
|
|
var lastNodeBeforeIndent = getLast$2(shouldMerge ? groups.slice(1, 2)[0] : groups[0]).node; |
|
var shouldHaveEmptyLineBeforeIndent = lastNodeBeforeIndent.type !== "CallExpression" && lastNodeBeforeIndent.type !== "OptionalCallExpression" && shouldInsertEmptyLineAfter(lastNodeBeforeIndent); |
|
var expanded = concat$d([printGroup(groups[0]), shouldMerge ? concat$d(groups.slice(1, 2).map(printGroup)) : "", shouldHaveEmptyLineBeforeIndent ? hardline$9 : "", printIndentedGroup(groups.slice(shouldMerge ? 2 : 1))]); |
|
var callExpressions = printedNodes.map(function (_ref) { |
|
var node = _ref.node; |
|
return node; |
|
}).filter(isCallOrOptionalCallExpression$1); // We don't want to print in one line if there's: |
|
// * A comment. |
|
// * 3 or more chained calls. |
|
// * Any group but the last one has a hard line. |
|
// If the last group is a function it's okay to inline if it fits. |
|
|
|
if (hasComment || callExpressions.length >= 3 || printedGroups.slice(0, -1).some(willBreak$1) || |
|
/** |
|
* scopes.filter(scope => scope.value !== '').map((scope, i) => { |
|
* // multi line content |
|
* }) |
|
*/ |
|
function (lastGroupDoc, lastGroupNode) { |
|
return isCallOrOptionalCallExpression$1(lastGroupNode) && willBreak$1(lastGroupDoc); |
|
}(getLast$2(printedGroups), getLast$2(getLast$2(groups)).node) && callExpressions.slice(0, -1).some(function (n) { |
|
return n.arguments.some(isFunctionOrArrowExpression$1); |
|
})) { |
|
return group$b(expanded); |
|
} |
|
|
|
return concat$d([// We only need to check `oneLine` because if `expanded` is chosen |
|
// that means that the parent group has already been broken |
|
// naturally |
|
willBreak$1(oneLine) || shouldHaveEmptyLineBeforeIndent ? breakParent$3 : "", conditionalGroup$1([oneLine, expanded])]); |
|
} |
|
|
|
function separatorNoWhitespace(isFacebookTranslationTag, child, childNode, nextNode) { |
|
if (isFacebookTranslationTag) { |
|
return ""; |
|
} |
|
|
|
if (childNode.type === "JSXElement" && !childNode.closingElement || nextNode && nextNode.type === "JSXElement" && !nextNode.closingElement) { |
|
return child.length === 1 ? softline$6 : hardline$9; |
|
} |
|
|
|
return softline$6; |
|
} |
|
|
|
function separatorWithWhitespace(isFacebookTranslationTag, child, childNode, nextNode) { |
|
if (isFacebookTranslationTag) { |
|
return hardline$9; |
|
} |
|
|
|
if (child.length === 1) { |
|
return childNode.type === "JSXElement" && !childNode.closingElement || nextNode && nextNode.type === "JSXElement" && !nextNode.closingElement ? hardline$9 : softline$6; |
|
} |
|
|
|
return hardline$9; |
|
} // JSX Children are strange, mostly for two reasons: |
|
// 1. JSX reads newlines into string values, instead of skipping them like JS |
|
// 2. up to one whitespace between elements within a line is significant, |
|
// but not between lines. |
|
// |
|
// Leading, trailing, and lone whitespace all need to |
|
// turn themselves into the rather ugly `{' '}` when breaking. |
|
// |
|
// We print JSX using the `fill` doc primitive. |
|
// This requires that we give it an array of alternating |
|
// content and whitespace elements. |
|
// To ensure this we add dummy `""` content elements as needed. |
|
|
|
|
|
function printJSXChildren(path, options, print, jsxWhitespace, isFacebookTranslationTag) { |
|
var n = path.getValue(); |
|
var children = []; // using `map` instead of `each` because it provides `i` |
|
|
|
path.map(function (childPath, i) { |
|
var child = childPath.getValue(); |
|
|
|
if (isLiteral$1(child)) { |
|
var text = rawText$1(child); // Contains a non-whitespace character |
|
|
|
if (isMeaningfulJSXText$1(child)) { |
|
var words = text.split(matchJsxWhitespaceRegex$1); // Starts with whitespace |
|
|
|
if (words[0] === "") { |
|
children.push(""); |
|
words.shift(); |
|
|
|
if (/\n/.test(words[0])) { |
|
var next = n.children[i + 1]; |
|
children.push(separatorWithWhitespace(isFacebookTranslationTag, words[1], child, next)); |
|
} else { |
|
children.push(jsxWhitespace); |
|
} |
|
|
|
words.shift(); |
|
} |
|
|
|
var endWhitespace; // Ends with whitespace |
|
|
|
if (getLast$2(words) === "") { |
|
words.pop(); |
|
endWhitespace = words.pop(); |
|
} // This was whitespace only without a new line. |
|
|
|
|
|
if (words.length === 0) { |
|
return; |
|
} |
|
|
|
words.forEach(function (word, i) { |
|
if (i % 2 === 1) { |
|
children.push(line$7); |
|
} else { |
|
children.push(word); |
|
} |
|
}); |
|
|
|
if (endWhitespace !== undefined) { |
|
if (/\n/.test(endWhitespace)) { |
|
var _next = n.children[i + 1]; |
|
children.push(separatorWithWhitespace(isFacebookTranslationTag, getLast$2(children), child, _next)); |
|
} else { |
|
children.push(jsxWhitespace); |
|
} |
|
} else { |
|
var _next2 = n.children[i + 1]; |
|
children.push(separatorNoWhitespace(isFacebookTranslationTag, getLast$2(children), child, _next2)); |
|
} |
|
} else if (/\n/.test(text)) { |
|
// Keep (up to one) blank line between tags/expressions/text. |
|
// Note: We don't keep blank lines between text elements. |
|
if (text.match(/\n/g).length > 1) { |
|
children.push(""); |
|
children.push(hardline$9); |
|
} |
|
} else { |
|
children.push(""); |
|
children.push(jsxWhitespace); |
|
} |
|
} else { |
|
var printedChild = print(childPath); |
|
children.push(printedChild); |
|
var _next3 = n.children[i + 1]; |
|
|
|
var directlyFollowedByMeaningfulText = _next3 && isMeaningfulJSXText$1(_next3); |
|
|
|
if (directlyFollowedByMeaningfulText) { |
|
var firstWord = rawText$1(_next3).trim().split(matchJsxWhitespaceRegex$1)[0]; |
|
children.push(separatorNoWhitespace(isFacebookTranslationTag, firstWord, child, _next3)); |
|
} else { |
|
children.push(hardline$9); |
|
} |
|
} |
|
}, "children"); |
|
return children; |
|
} // JSX expands children from the inside-out, instead of the outside-in. |
|
// This is both to break children before attributes, |
|
// and to ensure that when children break, their parents do as well. |
|
// |
|
// Any element that is written without any newlines and fits on a single line |
|
// is left that way. |
|
// Not only that, any user-written-line containing multiple JSX siblings |
|
// should also be kept on one line if possible, |
|
// so each user-written-line is wrapped in its own group. |
|
// |
|
// Elements that contain newlines or don't fit on a single line (recursively) |
|
// are fully-split, using hardline and shouldBreak: true. |
|
// |
|
// To support that case properly, all leading and trailing spaces |
|
// are stripped from the list of children, and replaced with a single hardline. |
|
|
|
|
|
function printJSXElement(path, options, print) { |
|
var n = path.getValue(); |
|
|
|
if (n.type === "JSXElement" && isEmptyJSXElement$1(n)) { |
|
return concat$d([path.call(print, "openingElement"), path.call(print, "closingElement")]); |
|
} |
|
|
|
var openingLines = n.type === "JSXElement" ? path.call(print, "openingElement") : path.call(print, "openingFragment"); |
|
var closingLines = n.type === "JSXElement" ? path.call(print, "closingElement") : path.call(print, "closingFragment"); |
|
|
|
if (n.children.length === 1 && n.children[0].type === "JSXExpressionContainer" && (n.children[0].expression.type === "TemplateLiteral" || n.children[0].expression.type === "TaggedTemplateExpression")) { |
|
return concat$d([openingLines, concat$d(path.map(print, "children")), closingLines]); |
|
} // Convert `{" "}` to text nodes containing a space. |
|
// This makes it easy to turn them into `jsxWhitespace` which |
|
// can then print as either a space or `{" "}` when breaking. |
|
|
|
|
|
n.children = n.children.map(function (child) { |
|
if (isJSXWhitespaceExpression$1(child)) { |
|
return { |
|
type: "JSXText", |
|
value: " ", |
|
raw: " " |
|
}; |
|
} |
|
|
|
return child; |
|
}); |
|
var containsTag = n.children.filter(isJSXNode$1).length > 0; |
|
var containsMultipleExpressions = n.children.filter(function (child) { |
|
return child.type === "JSXExpressionContainer"; |
|
}).length > 1; |
|
var containsMultipleAttributes = n.type === "JSXElement" && n.openingElement.attributes.length > 1; // Record any breaks. Should never go from true to false, only false to true. |
|
|
|
var forcedBreak = willBreak$1(openingLines) || containsTag || containsMultipleAttributes || containsMultipleExpressions; |
|
var rawJsxWhitespace = options.singleQuote ? "{' '}" : '{" "}'; |
|
var jsxWhitespace = ifBreak$6(concat$d([rawJsxWhitespace, softline$6]), " "); |
|
var isFacebookTranslationTag = n.openingElement && n.openingElement.name && n.openingElement.name.name === "fbt"; |
|
var children = printJSXChildren(path, options, print, jsxWhitespace, isFacebookTranslationTag); |
|
var containsText = n.children.filter(function (child) { |
|
return isMeaningfulJSXText$1(child); |
|
}).length > 0; // We can end up we multiple whitespace elements with empty string |
|
// content between them. |
|
// We need to remove empty whitespace and softlines before JSX whitespace |
|
// to get the correct output. |
|
|
|
for (var i = children.length - 2; i >= 0; i--) { |
|
var isPairOfEmptyStrings = children[i] === "" && children[i + 1] === ""; |
|
var isPairOfHardlines = children[i] === hardline$9 && children[i + 1] === "" && children[i + 2] === hardline$9; |
|
var isLineFollowedByJSXWhitespace = (children[i] === softline$6 || children[i] === hardline$9) && children[i + 1] === "" && children[i + 2] === jsxWhitespace; |
|
var isJSXWhitespaceFollowedByLine = children[i] === jsxWhitespace && children[i + 1] === "" && (children[i + 2] === softline$6 || children[i + 2] === hardline$9); |
|
var isDoubleJSXWhitespace = children[i] === jsxWhitespace && children[i + 1] === "" && children[i + 2] === jsxWhitespace; |
|
var isPairOfHardOrSoftLines = children[i] === softline$6 && children[i + 1] === "" && children[i + 2] === hardline$9 || children[i] === hardline$9 && children[i + 1] === "" && children[i + 2] === softline$6; |
|
|
|
if (isPairOfHardlines && containsText || isPairOfEmptyStrings || isLineFollowedByJSXWhitespace || isDoubleJSXWhitespace || isPairOfHardOrSoftLines) { |
|
children.splice(i, 2); |
|
} else if (isJSXWhitespaceFollowedByLine) { |
|
children.splice(i + 1, 2); |
|
} |
|
} // Trim trailing lines (or empty strings) |
|
|
|
|
|
while (children.length && (isLineNext$1(getLast$2(children)) || isEmpty$1(getLast$2(children)))) { |
|
children.pop(); |
|
} // Trim leading lines (or empty strings) |
|
|
|
|
|
while (children.length && (isLineNext$1(children[0]) || isEmpty$1(children[0])) && (isLineNext$1(children[1]) || isEmpty$1(children[1]))) { |
|
children.shift(); |
|
children.shift(); |
|
} // Tweak how we format children if outputting this element over multiple lines. |
|
// Also detect whether we will force this element to output over multiple lines. |
|
|
|
|
|
var multilineChildren = []; |
|
children.forEach(function (child, i) { |
|
// There are a number of situations where we need to ensure we display |
|
// whitespace as `{" "}` when outputting this element over multiple lines. |
|
if (child === jsxWhitespace) { |
|
if (i === 1 && children[i - 1] === "") { |
|
if (children.length === 2) { |
|
// Solitary whitespace |
|
multilineChildren.push(rawJsxWhitespace); |
|
return; |
|
} // Leading whitespace |
|
|
|
|
|
multilineChildren.push(concat$d([rawJsxWhitespace, hardline$9])); |
|
return; |
|
} else if (i === children.length - 1) { |
|
// Trailing whitespace |
|
multilineChildren.push(rawJsxWhitespace); |
|
return; |
|
} else if (children[i - 1] === "" && children[i - 2] === hardline$9) { |
|
// Whitespace after line break |
|
multilineChildren.push(rawJsxWhitespace); |
|
return; |
|
} |
|
} |
|
|
|
multilineChildren.push(child); |
|
|
|
if (willBreak$1(child)) { |
|
forcedBreak = true; |
|
} |
|
}); // If there is text we use `fill` to fit as much onto each line as possible. |
|
// When there is no text (just tags and expressions) we use `group` |
|
// to output each on a separate line. |
|
|
|
var content = containsText ? fill$4(multilineChildren) : group$b(concat$d(multilineChildren), { |
|
shouldBreak: true |
|
}); |
|
var multiLineElem = group$b(concat$d([openingLines, indent$7(concat$d([hardline$9, content])), hardline$9, closingLines])); |
|
|
|
if (forcedBreak) { |
|
return multiLineElem; |
|
} |
|
|
|
return conditionalGroup$1([group$b(concat$d([openingLines, concat$d(children), closingLines])), multiLineElem]); |
|
} |
|
|
|
function maybeWrapJSXElementInParens(path, elem, options) { |
|
var parent = path.getParentNode(); |
|
|
|
if (!parent) { |
|
return elem; |
|
} |
|
|
|
var NO_WRAP_PARENTS = { |
|
ArrayExpression: true, |
|
JSXAttribute: true, |
|
JSXElement: true, |
|
JSXExpressionContainer: true, |
|
JSXFragment: true, |
|
ExpressionStatement: true, |
|
CallExpression: true, |
|
OptionalCallExpression: true, |
|
ConditionalExpression: true, |
|
JsExpressionRoot: true |
|
}; |
|
|
|
if (NO_WRAP_PARENTS[parent.type]) { |
|
return elem; |
|
} |
|
|
|
var shouldBreak = matchAncestorTypes$1(path, ["ArrowFunctionExpression", "CallExpression", "JSXExpressionContainer"]) || matchAncestorTypes$1(path, ["ArrowFunctionExpression", "OptionalCallExpression", "JSXExpressionContainer"]); |
|
var needsParens = needsParens_1(path, options); |
|
return group$b(concat$d([needsParens ? "" : ifBreak$6("("), indent$7(concat$d([softline$6, elem])), softline$6, needsParens ? "" : ifBreak$6(")")]), { |
|
shouldBreak: shouldBreak |
|
}); |
|
} |
|
|
|
function shouldInlineLogicalExpression(node) { |
|
if (node.type !== "LogicalExpression") { |
|
return false; |
|
} |
|
|
|
if (node.right.type === "ObjectExpression" && node.right.properties.length !== 0) { |
|
return true; |
|
} |
|
|
|
if (node.right.type === "ArrayExpression" && node.right.elements.length !== 0) { |
|
return true; |
|
} |
|
|
|
if (isJSXNode$1(node.right)) { |
|
return true; |
|
} |
|
|
|
return false; |
|
} // For binary expressions to be consistent, we need to group |
|
// subsequent operators with the same precedence level under a single |
|
// group. Otherwise they will be nested such that some of them break |
|
// onto new lines but not all. Operators with the same precedence |
|
// level should either all break or not. Because we group them by |
|
// precedence level and the AST is structured based on precedence |
|
// level, things are naturally broken up correctly, i.e. `&&` is |
|
// broken before `+`. |
|
|
|
|
|
function printBinaryishExpressions(path, print, options, isNested, isInsideParenthesis) { |
|
var parts = []; |
|
var node = path.getValue(); // We treat BinaryExpression and LogicalExpression nodes the same. |
|
|
|
if (isBinaryish$1(node)) { |
|
// Put all operators with the same precedence level in the same |
|
// group. The reason we only need to do this with the `left` |
|
// expression is because given an expression like `1 + 2 - 3`, it |
|
// is always parsed like `((1 + 2) - 3)`, meaning the `left` side |
|
// is where the rest of the expression will exist. Binary |
|
// expressions on the right side mean they have a difference |
|
// precedence level and should be treated as a separate group, so |
|
// print them normally. (This doesn't hold for the `**` operator, |
|
// which is unique in that it is right-associative.) |
|
if (shouldFlatten$1(node.operator, node.left.operator)) { |
|
// Flatten them out by recursively calling this function. |
|
parts = parts.concat(path.call(function (left) { |
|
return printBinaryishExpressions(left, print, options, |
|
/* isNested */ |
|
true, isInsideParenthesis); |
|
}, "left")); |
|
} else { |
|
parts.push(path.call(print, "left")); |
|
} |
|
|
|
var shouldInline = shouldInlineLogicalExpression(node); |
|
var lineBeforeOperator = (node.operator === "|>" || node.type === "NGPipeExpression" || node.operator === "|" && options.parser === "__vue_expression") && !hasLeadingOwnLineComment$1(options.originalText, node.right, options); |
|
var operator = node.type === "NGPipeExpression" ? "|" : node.operator; |
|
var rightSuffix = node.type === "NGPipeExpression" && node.arguments.length !== 0 ? group$b(indent$7(concat$d([softline$6, ": ", join$9(concat$d([softline$6, ":", ifBreak$6(" ")]), path.map(print, "arguments").map(function (arg) { |
|
return align$1(2, group$b(arg)); |
|
}))]))) : ""; |
|
var right = shouldInline ? concat$d([operator, " ", path.call(print, "right"), rightSuffix]) : concat$d([lineBeforeOperator ? softline$6 : "", operator, lineBeforeOperator ? " " : line$7, path.call(print, "right"), rightSuffix]); // If there's only a single binary expression, we want to create a group |
|
// in order to avoid having a small right part like -1 be on its own line. |
|
|
|
var parent = path.getParentNode(); |
|
var shouldGroup = !(isInsideParenthesis && node.type === "LogicalExpression") && parent.type !== node.type && node.left.type !== node.type && node.right.type !== node.type; |
|
parts.push(" ", shouldGroup ? group$b(right) : right); // The root comments are already printed, but we need to manually print |
|
// the other ones since we don't call the normal print on BinaryExpression, |
|
// only for the left and right parts |
|
|
|
if (isNested && node.comments) { |
|
parts = comments.printComments(path, function () { |
|
return concat$d(parts); |
|
}, options); |
|
} |
|
} else { |
|
// Our stopping case. Simply print the node normally. |
|
parts.push(path.call(print)); |
|
} |
|
|
|
return parts; |
|
} |
|
|
|
function printAssignmentRight(leftNode, rightNode, printedRight, options) { |
|
if (hasLeadingOwnLineComment$1(options.originalText, rightNode, options)) { |
|
return indent$7(concat$d([hardline$9, printedRight])); |
|
} |
|
|
|
var canBreak = isBinaryish$1(rightNode) && !shouldInlineLogicalExpression(rightNode) || rightNode.type === "ConditionalExpression" && isBinaryish$1(rightNode.test) && !shouldInlineLogicalExpression(rightNode.test) || rightNode.type === "StringLiteralTypeAnnotation" || rightNode.type === "ClassExpression" && rightNode.decorators && rightNode.decorators.length || (leftNode.type === "Identifier" || isStringLiteral$1(leftNode) || leftNode.type === "MemberExpression") && (isStringLiteral$1(rightNode) || isMemberExpressionChain$1(rightNode)) && // do not put values on a separate line from the key in json |
|
options.parser !== "json" && options.parser !== "json5" || rightNode.type === "SequenceExpression"; |
|
|
|
if (canBreak) { |
|
return group$b(indent$7(concat$d([line$7, printedRight]))); |
|
} |
|
|
|
return concat$d([" ", printedRight]); |
|
} |
|
|
|
function printAssignment(leftNode, printedLeft, operator, rightNode, printedRight, options) { |
|
if (!rightNode) { |
|
return printedLeft; |
|
} |
|
|
|
var printed = printAssignmentRight(leftNode, rightNode, printedRight, options); |
|
return group$b(concat$d([printedLeft, operator, printed])); |
|
} |
|
|
|
function adjustClause(node, clause, forceSpace) { |
|
if (node.type === "EmptyStatement") { |
|
return ";"; |
|
} |
|
|
|
if (node.type === "BlockStatement" || forceSpace) { |
|
return concat$d([" ", clause]); |
|
} |
|
|
|
return indent$7(concat$d([line$7, clause])); |
|
} |
|
|
|
function nodeStr(node, options, isFlowOrTypeScriptDirectiveLiteral) { |
|
var raw = rawText$1(node); |
|
var isDirectiveLiteral = isFlowOrTypeScriptDirectiveLiteral || node.type === "DirectiveLiteral"; |
|
return printString$2(raw, options, isDirectiveLiteral); |
|
} |
|
|
|
function printRegex(node) { |
|
var flags = node.flags.split("").sort().join(""); |
|
return "/".concat(node.pattern, "/").concat(flags); |
|
} |
|
|
|
function exprNeedsASIProtection(path, options) { |
|
var node = path.getValue(); |
|
var maybeASIProblem = needsParens_1(path, options) || node.type === "ParenthesizedExpression" || node.type === "TypeCastExpression" || node.type === "ArrowFunctionExpression" && !shouldPrintParamsWithoutParens(path, options) || node.type === "ArrayExpression" || node.type === "ArrayPattern" || node.type === "UnaryExpression" && node.prefix && (node.operator === "+" || node.operator === "-") || node.type === "TemplateLiteral" || node.type === "TemplateElement" || isJSXNode$1(node) || node.type === "BindExpression" && !node.object || node.type === "RegExpLiteral" || node.type === "Literal" && node.pattern || node.type === "Literal" && node.regex; |
|
|
|
if (maybeASIProblem) { |
|
return true; |
|
} |
|
|
|
if (!hasNakedLeftSide$2(node)) { |
|
return false; |
|
} |
|
|
|
return path.call.apply(path, [function (childPath) { |
|
return exprNeedsASIProtection(childPath, options); |
|
}].concat(getLeftSidePathName$2(path, node))); |
|
} |
|
|
|
function stmtNeedsASIProtection(path, options) { |
|
var node = path.getNode(); |
|
|
|
if (node.type !== "ExpressionStatement") { |
|
return false; |
|
} |
|
|
|
return path.call(function (childPath) { |
|
return exprNeedsASIProtection(childPath, options); |
|
}, "expression"); |
|
} |
|
|
|
function shouldHugType(node) { |
|
if (isSimpleFlowType$1(node) || isObjectType$1(node)) { |
|
return true; |
|
} |
|
|
|
if (node.type === "UnionTypeAnnotation" || node.type === "TSUnionType") { |
|
var voidCount = node.types.filter(function (n) { |
|
return n.type === "VoidTypeAnnotation" || n.type === "TSVoidKeyword" || n.type === "NullLiteralTypeAnnotation" || n.type === "TSNullKeyword"; |
|
}).length; |
|
var objectCount = node.types.filter(function (n) { |
|
return n.type === "ObjectTypeAnnotation" || n.type === "TSTypeLiteral" || // This is a bit aggressive but captures Array<{x}> |
|
n.type === "GenericTypeAnnotation" || n.type === "TSTypeReference"; |
|
}).length; |
|
|
|
if (node.types.length - 1 === voidCount && objectCount > 0) { |
|
return true; |
|
} |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function shouldHugArguments(fun) { |
|
return fun && fun.params && fun.params.length === 1 && !fun.params[0].comments && (fun.params[0].type === "ObjectPattern" || fun.params[0].type === "ArrayPattern" || fun.params[0].type === "Identifier" && fun.params[0].typeAnnotation && (fun.params[0].typeAnnotation.type === "TypeAnnotation" || fun.params[0].typeAnnotation.type === "TSTypeAnnotation") && isObjectType$1(fun.params[0].typeAnnotation.typeAnnotation) || fun.params[0].type === "FunctionTypeParam" && isObjectType$1(fun.params[0].typeAnnotation) || fun.params[0].type === "AssignmentPattern" && (fun.params[0].left.type === "ObjectPattern" || fun.params[0].left.type === "ArrayPattern") && (fun.params[0].right.type === "Identifier" || fun.params[0].right.type === "ObjectExpression" && fun.params[0].right.properties.length === 0 || fun.params[0].right.type === "ArrayExpression" && fun.params[0].right.elements.length === 0)) && !fun.rest; |
|
} |
|
|
|
function printArrayItems(path, options, printPath, print) { |
|
var printedElements = []; |
|
var separatorParts = []; |
|
path.each(function (childPath) { |
|
printedElements.push(concat$d(separatorParts)); |
|
printedElements.push(group$b(print(childPath))); |
|
separatorParts = [",", line$7]; |
|
|
|
if (childPath.getValue() && isNextLineEmpty$4(options.originalText, childPath.getValue(), options)) { |
|
separatorParts.push(softline$6); |
|
} |
|
}, printPath); |
|
return concat$d(printedElements); |
|
} |
|
|
|
function willPrintOwnComments(path |
|
/*, options */ |
|
) { |
|
var node = path.getValue(); |
|
var parent = path.getParentNode(); |
|
return (node && (isJSXNode$1(node) || hasFlowShorthandAnnotationComment$2(node) || parent && (parent.type === "CallExpression" || parent.type === "OptionalCallExpression") && (hasFlowAnnotationComment$1(node.leadingComments) || hasFlowAnnotationComment$1(node.trailingComments))) || parent && (parent.type === "JSXSpreadAttribute" || parent.type === "JSXSpreadChild" || parent.type === "UnionTypeAnnotation" || parent.type === "TSUnionType" || (parent.type === "ClassDeclaration" || parent.type === "ClassExpression") && parent.superClass === node)) && !hasIgnoreComment$4(path); |
|
} |
|
|
|
function canAttachComment$1(node) { |
|
return node.type && node.type !== "CommentBlock" && node.type !== "CommentLine" && node.type !== "Line" && node.type !== "Block" && node.type !== "EmptyStatement" && node.type !== "TemplateElement" && node.type !== "Import"; |
|
} |
|
|
|
function printComment$2(commentPath, options) { |
|
var comment = commentPath.getValue(); |
|
|
|
switch (comment.type) { |
|
case "CommentBlock": |
|
case "Block": |
|
{ |
|
if (isIndentableBlockComment(comment)) { |
|
var printed = printIndentableBlockComment(comment); // We need to prevent an edge case of a previous trailing comment |
|
// printed as a `lineSuffix` which causes the comments to be |
|
// interleaved. See https://github.com/prettier/prettier/issues/4412 |
|
|
|
if (comment.trailing && !hasNewline$4(options.originalText, options.locStart(comment), { |
|
backwards: true |
|
})) { |
|
return concat$d([hardline$9, printed]); |
|
} |
|
|
|
return printed; |
|
} |
|
|
|
var isInsideFlowComment = options.originalText.substr(options.locEnd(comment) - 3, 3) === "*-/"; |
|
return "/*" + comment.value + (isInsideFlowComment ? "*-/" : "*/"); |
|
} |
|
|
|
case "CommentLine": |
|
case "Line": |
|
// Print shebangs with the proper comment characters |
|
if (options.originalText.slice(options.locStart(comment)).startsWith("#!")) { |
|
return "#!" + comment.value.trimRight(); |
|
} |
|
|
|
return "//" + comment.value.trimRight(); |
|
|
|
default: |
|
throw new Error("Not a comment: " + JSON.stringify(comment)); |
|
} |
|
} |
|
|
|
function isIndentableBlockComment(comment) { |
|
// If the comment has multiple lines and every line starts with a star |
|
// we can fix the indentation of each line. The stars in the `/*` and |
|
// `*/` delimiters are not included in the comment value, so add them |
|
// back first. |
|
var lines = "*".concat(comment.value, "*").split("\n"); |
|
return lines.length > 1 && lines.every(function (line) { |
|
return line.trim()[0] === "*"; |
|
}); |
|
} |
|
|
|
function printIndentableBlockComment(comment) { |
|
var lines = comment.value.split("\n"); |
|
return concat$d(["/*", join$9(hardline$9, lines.map(function (line, index) { |
|
return index === 0 ? line.trimRight() : " " + (index < lines.length - 1 ? line.trim() : line.trimLeft()); |
|
})), "*/"]); |
|
} |
|
|
|
var printerEstree = { |
|
preprocess: preprocess_1$1, |
|
print: genericPrint$3, |
|
embed: embed_1$1, |
|
insertPragma: insertPragma$7, |
|
massageAstNode: clean_1$1, |
|
hasPrettierIgnore: hasPrettierIgnore$3, |
|
willPrintOwnComments: willPrintOwnComments, |
|
canAttachComment: canAttachComment$1, |
|
printComment: printComment$2, |
|
isBlockComment: comments$1.isBlockComment, |
|
handleComments: { |
|
ownLine: comments$1.handleOwnLineComment, |
|
endOfLine: comments$1.handleEndOfLineComment, |
|
remaining: comments$1.handleRemainingComment |
|
} |
|
}; |
|
|
|
var _require$$0$builders$7 = doc.builders, |
|
concat$e = _require$$0$builders$7.concat, |
|
hardline$a = _require$$0$builders$7.hardline, |
|
indent$8 = _require$$0$builders$7.indent, |
|
join$a = _require$$0$builders$7.join; |
|
|
|
function genericPrint$4(path, options, print) { |
|
var node = path.getValue(); |
|
|
|
switch (node.type) { |
|
case "JsonRoot": |
|
return concat$e([path.call(print, "node"), hardline$a]); |
|
|
|
case "ArrayExpression": |
|
return node.elements.length === 0 ? "[]" : concat$e(["[", indent$8(concat$e([hardline$a, join$a(concat$e([",", hardline$a]), path.map(print, "elements"))])), hardline$a, "]"]); |
|
|
|
case "ObjectExpression": |
|
return node.properties.length === 0 ? "{}" : concat$e(["{", indent$8(concat$e([hardline$a, join$a(concat$e([",", hardline$a]), path.map(print, "properties"))])), hardline$a, "}"]); |
|
|
|
case "ObjectProperty": |
|
return concat$e([path.call(print, "key"), ": ", path.call(print, "value")]); |
|
|
|
case "UnaryExpression": |
|
return concat$e([node.operator === "+" ? "" : node.operator, path.call(print, "argument")]); |
|
|
|
case "NullLiteral": |
|
return "null"; |
|
|
|
case "BooleanLiteral": |
|
return node.value ? "true" : "false"; |
|
|
|
case "StringLiteral": |
|
case "NumericLiteral": |
|
return JSON.stringify(node.value); |
|
|
|
case "Identifier": |
|
return JSON.stringify(node.name); |
|
|
|
default: |
|
/* istanbul ignore next */ |
|
throw new Error("unknown type: " + JSON.stringify(node.type)); |
|
} |
|
} |
|
|
|
function clean$5(node, newNode |
|
/*, parent*/ |
|
) { |
|
delete newNode.start; |
|
delete newNode.end; |
|
delete newNode.extra; |
|
delete newNode.loc; |
|
delete newNode.comments; |
|
delete newNode.errors; |
|
|
|
if (node.type === "Identifier") { |
|
return { |
|
type: "StringLiteral", |
|
value: node.name |
|
}; |
|
} |
|
|
|
if (node.type === "UnaryExpression" && node.operator === "+") { |
|
return newNode.argument; |
|
} |
|
} |
|
|
|
var printerEstreeJson = { |
|
preprocess: preprocess_1$1, |
|
print: genericPrint$4, |
|
massageAstNode: clean$5 |
|
}; |
|
|
|
var CATEGORY_JAVASCRIPT = "JavaScript"; // format based on https://github.com/prettier/prettier/blob/master/src/main/core-options.js |
|
|
|
var options$5 = { |
|
arrowParens: { |
|
since: "1.9.0", |
|
category: CATEGORY_JAVASCRIPT, |
|
type: "choice", |
|
default: "avoid", |
|
description: "Include parentheses around a sole arrow function parameter.", |
|
choices: [{ |
|
value: "avoid", |
|
description: "Omit parens when possible. Example: `x => x`" |
|
}, { |
|
value: "always", |
|
description: "Always include parens. Example: `(x) => x`" |
|
}] |
|
}, |
|
bracketSpacing: commonOptions.bracketSpacing, |
|
jsxBracketSameLine: { |
|
since: "0.17.0", |
|
category: CATEGORY_JAVASCRIPT, |
|
type: "boolean", |
|
default: false, |
|
description: "Put > on the last line instead of at a new line." |
|
}, |
|
semi: { |
|
since: "1.0.0", |
|
category: CATEGORY_JAVASCRIPT, |
|
type: "boolean", |
|
default: true, |
|
description: "Print semicolons.", |
|
oppositeDescription: "Do not print semicolons, except at the beginning of lines which may need them." |
|
}, |
|
singleQuote: commonOptions.singleQuote, |
|
jsxSingleQuote: { |
|
since: "1.15.0", |
|
category: CATEGORY_JAVASCRIPT, |
|
type: "boolean", |
|
default: false, |
|
description: "Use single quotes in JSX." |
|
}, |
|
quoteProps: { |
|
since: "1.17.0", |
|
category: CATEGORY_JAVASCRIPT, |
|
type: "choice", |
|
default: "as-needed", |
|
description: "Change when properties in objects are quoted.", |
|
choices: [{ |
|
value: "as-needed", |
|
description: "Only add quotes around object properties where required." |
|
}, { |
|
value: "consistent", |
|
description: "If at least one property in an object requires quotes, quote all properties." |
|
}, { |
|
value: "preserve", |
|
description: "Respect the input use of quotes in object properties." |
|
}] |
|
}, |
|
trailingComma: { |
|
since: "0.0.0", |
|
category: CATEGORY_JAVASCRIPT, |
|
type: "choice", |
|
default: [{ |
|
since: "0.0.0", |
|
value: false |
|
}, { |
|
since: "0.19.0", |
|
value: "none" |
|
}], |
|
description: "Print trailing commas wherever possible when multi-line.", |
|
choices: [{ |
|
value: "none", |
|
description: "No trailing commas." |
|
}, { |
|
value: "es5", |
|
description: "Trailing commas where valid in ES5 (objects, arrays, etc.)" |
|
}, { |
|
value: "all", |
|
description: "Trailing commas wherever possible (including function arguments)." |
|
}, { |
|
value: true, |
|
deprecated: "0.19.0", |
|
redirect: "es5" |
|
}, { |
|
value: false, |
|
deprecated: "0.19.0", |
|
redirect: "none" |
|
}] |
|
} |
|
}; |
|
|
|
var name$9 = "JavaScript"; |
|
var type$8 = "programming"; |
|
var tmScope$8 = "source.js"; |
|
var aceMode$8 = "javascript"; |
|
var codemirrorMode$4 = "javascript"; |
|
var codemirrorMimeType$4 = "text/javascript"; |
|
var color$3 = "#f1e05a"; |
|
var aliases$2 = [ |
|
"js", |
|
"node" |
|
]; |
|
var extensions$8 = [ |
|
".js", |
|
"._js", |
|
".bones", |
|
".es", |
|
".es6", |
|
".frag", |
|
".gs", |
|
".jake", |
|
".jsb", |
|
".jscad", |
|
".jsfl", |
|
".jsm", |
|
".jss", |
|
".mjs", |
|
".njs", |
|
".pac", |
|
".sjs", |
|
".ssjs", |
|
".xsjs", |
|
".xsjslib" |
|
]; |
|
var filenames = [ |
|
"Jakefile" |
|
]; |
|
var interpreters = [ |
|
"chakra", |
|
"d8", |
|
"js", |
|
"node", |
|
"rhino", |
|
"v8", |
|
"v8-shell" |
|
]; |
|
var languageId$8 = 183; |
|
var JavaScript = { |
|
name: name$9, |
|
type: type$8, |
|
tmScope: tmScope$8, |
|
aceMode: aceMode$8, |
|
codemirrorMode: codemirrorMode$4, |
|
codemirrorMimeType: codemirrorMimeType$4, |
|
color: color$3, |
|
aliases: aliases$2, |
|
extensions: extensions$8, |
|
filenames: filenames, |
|
interpreters: interpreters, |
|
languageId: languageId$8 |
|
}; |
|
|
|
var JavaScript$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$9, |
|
type: type$8, |
|
tmScope: tmScope$8, |
|
aceMode: aceMode$8, |
|
codemirrorMode: codemirrorMode$4, |
|
codemirrorMimeType: codemirrorMimeType$4, |
|
color: color$3, |
|
aliases: aliases$2, |
|
extensions: extensions$8, |
|
filenames: filenames, |
|
interpreters: interpreters, |
|
languageId: languageId$8, |
|
'default': JavaScript |
|
}); |
|
|
|
var name$a = "JSX"; |
|
var type$9 = "programming"; |
|
var group$c = "JavaScript"; |
|
var extensions$9 = [ |
|
".jsx" |
|
]; |
|
var tmScope$9 = "source.js.jsx"; |
|
var aceMode$9 = "javascript"; |
|
var codemirrorMode$5 = "jsx"; |
|
var codemirrorMimeType$5 = "text/jsx"; |
|
var languageId$9 = 178; |
|
var JSX = { |
|
name: name$a, |
|
type: type$9, |
|
group: group$c, |
|
extensions: extensions$9, |
|
tmScope: tmScope$9, |
|
aceMode: aceMode$9, |
|
codemirrorMode: codemirrorMode$5, |
|
codemirrorMimeType: codemirrorMimeType$5, |
|
languageId: languageId$9 |
|
}; |
|
|
|
var JSX$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$a, |
|
type: type$9, |
|
group: group$c, |
|
extensions: extensions$9, |
|
tmScope: tmScope$9, |
|
aceMode: aceMode$9, |
|
codemirrorMode: codemirrorMode$5, |
|
codemirrorMimeType: codemirrorMimeType$5, |
|
languageId: languageId$9, |
|
'default': JSX |
|
}); |
|
|
|
var name$b = "TypeScript"; |
|
var type$a = "programming"; |
|
var color$4 = "#2b7489"; |
|
var aliases$3 = [ |
|
"ts" |
|
]; |
|
var interpreters$1 = [ |
|
"deno", |
|
"ts-node" |
|
]; |
|
var extensions$a = [ |
|
".ts" |
|
]; |
|
var tmScope$a = "source.ts"; |
|
var aceMode$a = "typescript"; |
|
var codemirrorMode$6 = "javascript"; |
|
var codemirrorMimeType$6 = "application/typescript"; |
|
var languageId$a = 378; |
|
var TypeScript = { |
|
name: name$b, |
|
type: type$a, |
|
color: color$4, |
|
aliases: aliases$3, |
|
interpreters: interpreters$1, |
|
extensions: extensions$a, |
|
tmScope: tmScope$a, |
|
aceMode: aceMode$a, |
|
codemirrorMode: codemirrorMode$6, |
|
codemirrorMimeType: codemirrorMimeType$6, |
|
languageId: languageId$a |
|
}; |
|
|
|
var TypeScript$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$b, |
|
type: type$a, |
|
color: color$4, |
|
aliases: aliases$3, |
|
interpreters: interpreters$1, |
|
extensions: extensions$a, |
|
tmScope: tmScope$a, |
|
aceMode: aceMode$a, |
|
codemirrorMode: codemirrorMode$6, |
|
codemirrorMimeType: codemirrorMimeType$6, |
|
languageId: languageId$a, |
|
'default': TypeScript |
|
}); |
|
|
|
var name$c = "TSX"; |
|
var type$b = "programming"; |
|
var group$d = "TypeScript"; |
|
var extensions$b = [ |
|
".tsx" |
|
]; |
|
var tmScope$b = "source.tsx"; |
|
var aceMode$b = "javascript"; |
|
var codemirrorMode$7 = "jsx"; |
|
var codemirrorMimeType$7 = "text/jsx"; |
|
var languageId$b = 94901924; |
|
var TSX = { |
|
name: name$c, |
|
type: type$b, |
|
group: group$d, |
|
extensions: extensions$b, |
|
tmScope: tmScope$b, |
|
aceMode: aceMode$b, |
|
codemirrorMode: codemirrorMode$7, |
|
codemirrorMimeType: codemirrorMimeType$7, |
|
languageId: languageId$b |
|
}; |
|
|
|
var TSX$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$c, |
|
type: type$b, |
|
group: group$d, |
|
extensions: extensions$b, |
|
tmScope: tmScope$b, |
|
aceMode: aceMode$b, |
|
codemirrorMode: codemirrorMode$7, |
|
codemirrorMimeType: codemirrorMimeType$7, |
|
languageId: languageId$b, |
|
'default': TSX |
|
}); |
|
|
|
var name$d = "JSON"; |
|
var type$c = "data"; |
|
var tmScope$c = "source.json"; |
|
var aceMode$c = "json"; |
|
var codemirrorMode$8 = "javascript"; |
|
var codemirrorMimeType$8 = "application/json"; |
|
var searchable = false; |
|
var extensions$c = [ |
|
".json", |
|
".avsc", |
|
".geojson", |
|
".gltf", |
|
".har", |
|
".ice", |
|
".JSON-tmLanguage", |
|
".jsonl", |
|
".mcmeta", |
|
".tfstate", |
|
".tfstate.backup", |
|
".topojson", |
|
".webapp", |
|
".webmanifest", |
|
".yy", |
|
".yyp" |
|
]; |
|
var filenames$1 = [ |
|
".arcconfig", |
|
".htmlhintrc", |
|
".tern-config", |
|
".tern-project", |
|
".watchmanconfig", |
|
"composer.lock", |
|
"mcmod.info" |
|
]; |
|
var languageId$c = 174; |
|
var _JSON = { |
|
name: name$d, |
|
type: type$c, |
|
tmScope: tmScope$c, |
|
aceMode: aceMode$c, |
|
codemirrorMode: codemirrorMode$8, |
|
codemirrorMimeType: codemirrorMimeType$8, |
|
searchable: searchable, |
|
extensions: extensions$c, |
|
filenames: filenames$1, |
|
languageId: languageId$c |
|
}; |
|
|
|
var _JSON$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$d, |
|
type: type$c, |
|
tmScope: tmScope$c, |
|
aceMode: aceMode$c, |
|
codemirrorMode: codemirrorMode$8, |
|
codemirrorMimeType: codemirrorMimeType$8, |
|
searchable: searchable, |
|
extensions: extensions$c, |
|
filenames: filenames$1, |
|
languageId: languageId$c, |
|
'default': _JSON |
|
}); |
|
|
|
var name$e = "JSON with Comments"; |
|
var type$d = "data"; |
|
var group$e = "JSON"; |
|
var tmScope$d = "source.js"; |
|
var aceMode$d = "javascript"; |
|
var codemirrorMode$9 = "javascript"; |
|
var codemirrorMimeType$9 = "text/javascript"; |
|
var aliases$4 = [ |
|
"jsonc" |
|
]; |
|
var extensions$d = [ |
|
".sublime-build", |
|
".sublime-commands", |
|
".sublime-completions", |
|
".sublime-keymap", |
|
".sublime-macro", |
|
".sublime-menu", |
|
".sublime-mousemap", |
|
".sublime-project", |
|
".sublime-settings", |
|
".sublime-theme", |
|
".sublime-workspace", |
|
".sublime_metrics", |
|
".sublime_session" |
|
]; |
|
var filenames$2 = [ |
|
".babelrc", |
|
".eslintrc.json", |
|
".jscsrc", |
|
".jshintrc", |
|
".jslintrc", |
|
"jsconfig.json", |
|
"language-configuration.json", |
|
"tsconfig.json" |
|
]; |
|
var languageId$d = 423; |
|
var JSON_with_Comments = { |
|
name: name$e, |
|
type: type$d, |
|
group: group$e, |
|
tmScope: tmScope$d, |
|
aceMode: aceMode$d, |
|
codemirrorMode: codemirrorMode$9, |
|
codemirrorMimeType: codemirrorMimeType$9, |
|
aliases: aliases$4, |
|
extensions: extensions$d, |
|
filenames: filenames$2, |
|
languageId: languageId$d |
|
}; |
|
|
|
var JSON_with_Comments$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$e, |
|
type: type$d, |
|
group: group$e, |
|
tmScope: tmScope$d, |
|
aceMode: aceMode$d, |
|
codemirrorMode: codemirrorMode$9, |
|
codemirrorMimeType: codemirrorMimeType$9, |
|
aliases: aliases$4, |
|
extensions: extensions$d, |
|
filenames: filenames$2, |
|
languageId: languageId$d, |
|
'default': JSON_with_Comments |
|
}); |
|
|
|
var name$f = "JSON5"; |
|
var type$e = "data"; |
|
var extensions$e = [ |
|
".json5" |
|
]; |
|
var tmScope$e = "source.js"; |
|
var aceMode$e = "javascript"; |
|
var codemirrorMode$a = "javascript"; |
|
var codemirrorMimeType$a = "application/json"; |
|
var languageId$e = 175; |
|
var JSON5 = { |
|
name: name$f, |
|
type: type$e, |
|
extensions: extensions$e, |
|
tmScope: tmScope$e, |
|
aceMode: aceMode$e, |
|
codemirrorMode: codemirrorMode$a, |
|
codemirrorMimeType: codemirrorMimeType$a, |
|
languageId: languageId$e |
|
}; |
|
|
|
var JSON5$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$f, |
|
type: type$e, |
|
extensions: extensions$e, |
|
tmScope: tmScope$e, |
|
aceMode: aceMode$e, |
|
codemirrorMode: codemirrorMode$a, |
|
codemirrorMimeType: codemirrorMimeType$a, |
|
languageId: languageId$e, |
|
'default': JSON5 |
|
}); |
|
|
|
var require$$0$6 = getCjsExportFromNamespace(JavaScript$1); |
|
|
|
var require$$1$2 = getCjsExportFromNamespace(JSX$1); |
|
|
|
var require$$2$1 = getCjsExportFromNamespace(TypeScript$1); |
|
|
|
var require$$3$1 = getCjsExportFromNamespace(TSX$1); |
|
|
|
var require$$4 = getCjsExportFromNamespace(_JSON$1); |
|
|
|
var require$$5 = getCjsExportFromNamespace(JSON_with_Comments$1); |
|
|
|
var require$$6 = getCjsExportFromNamespace(JSON5$1); |
|
|
|
var languages$4 = [createLanguage(require$$0$6, function (data) { |
|
return Object.assign(data, { |
|
since: "0.0.0", |
|
parsers: ["babel", "flow"], |
|
vscodeLanguageIds: ["javascript", "mongo"], |
|
interpreters: data.interpreters.concat(["nodejs"]) |
|
}); |
|
}), createLanguage(require$$0$6, function (data) { |
|
return Object.assign(data, { |
|
name: "Flow", |
|
since: "0.0.0", |
|
parsers: ["babel", "flow"], |
|
vscodeLanguageIds: ["javascript"], |
|
aliases: [], |
|
filenames: [], |
|
extensions: [".js.flow"] |
|
}); |
|
}), createLanguage(require$$1$2, function (data) { |
|
return Object.assign(data, { |
|
since: "0.0.0", |
|
parsers: ["babel", "flow"], |
|
vscodeLanguageIds: ["javascriptreact"] |
|
}); |
|
}), createLanguage(require$$2$1, function (data) { |
|
return Object.assign(data, { |
|
since: "1.4.0", |
|
parsers: ["typescript"], |
|
vscodeLanguageIds: ["typescript"] |
|
}); |
|
}), createLanguage(require$$3$1, function (data) { |
|
return Object.assign(data, { |
|
since: "1.4.0", |
|
parsers: ["typescript"], |
|
vscodeLanguageIds: ["typescriptreact"] |
|
}); |
|
}), createLanguage(require$$4, function (data) { |
|
return Object.assign(data, { |
|
name: "JSON.stringify", |
|
since: "1.13.0", |
|
parsers: ["json-stringify"], |
|
vscodeLanguageIds: ["json"], |
|
extensions: [], |
|
// .json file defaults to json instead of json-stringify |
|
filenames: ["package.json", "package-lock.json", "composer.json"] |
|
}); |
|
}), createLanguage(require$$4, function (data) { |
|
return Object.assign(data, { |
|
since: "1.5.0", |
|
parsers: ["json"], |
|
vscodeLanguageIds: ["json"], |
|
filenames: data.filenames.concat([".prettierrc"]) |
|
}); |
|
}), createLanguage(require$$5, function (data) { |
|
return Object.assign(data, { |
|
since: "1.5.0", |
|
parsers: ["json"], |
|
vscodeLanguageIds: ["jsonc"], |
|
filenames: data.filenames.concat([".eslintrc"]) |
|
}); |
|
}), createLanguage(require$$6, function (data) { |
|
return Object.assign(data, { |
|
since: "1.13.0", |
|
parsers: ["json5"], |
|
vscodeLanguageIds: ["json5"] |
|
}); |
|
})]; |
|
var printers$4 = { |
|
estree: printerEstree, |
|
"estree-json": printerEstreeJson |
|
}; |
|
var languageJs = { |
|
languages: languages$4, |
|
options: options$5, |
|
printers: printers$4 |
|
}; |
|
|
|
var json$1 = { |
|
"cjkPattern": "[\\u02ea-\\u02eb\\u1100-\\u11ff\\u2e80-\\u2e99\\u2e9b-\\u2ef3\\u2f00-\\u2fd5\\u3000-\\u303f\\u3041-\\u3096\\u3099-\\u309f\\u30a1-\\u30fa\\u30fc-\\u30ff\\u3105-\\u312f\\u3131-\\u318e\\u3190-\\u3191\\u3196-\\u31ba\\u31c0-\\u31e3\\u31f0-\\u321e\\u322a-\\u3247\\u3260-\\u327e\\u328a-\\u32b0\\u32c0-\\u32cb\\u32d0-\\u3370\\u337b-\\u337f\\u33e0-\\u33fe\\u3400-\\u4db5\\u4e00-\\u9fef\\ua960-\\ua97c\\uac00-\\ud7a3\\ud7b0-\\ud7c6\\ud7cb-\\ud7fb\\uf900-\\ufa6d\\ufa70-\\ufad9\\ufe10-\\ufe1f\\ufe30-\\ufe6f\\uff00-\\uffef]|[\\ud840-\\ud868\\ud86a-\\ud86c\\ud86f-\\ud872\\ud874-\\ud879][\\udc00-\\udfff]|\\ud82c[\\udc00-\\udd1e\\udd50-\\udd52\\udd64-\\udd67]|\\ud83c[\\ude00\\ude50-\\ude51]|\\ud869[\\udc00-\\uded6\\udf00-\\udfff]|\\ud86d[\\udc00-\\udf34\\udf40-\\udfff]|\\ud86e[\\udc00-\\udc1d\\udc20-\\udfff]|\\ud873[\\udc00-\\udea1\\udeb0-\\udfff]|\\ud87a[\\udc00-\\udfe0]|\\ud87e[\\udc00-\\ude1d]", |
|
"kPattern": "[\\u1100-\\u11ff\\u3001-\\u3003\\u3008-\\u3011\\u3013-\\u301f\\u302e-\\u3030\\u3037\\u30fb\\u3131-\\u318e\\u3200-\\u321e\\u3260-\\u327e\\ua960-\\ua97c\\uac00-\\ud7a3\\ud7b0-\\ud7c6\\ud7cb-\\ud7fb\\ufe45-\\ufe46\\uff61-\\uff65\\uffa0-\\uffbe\\uffc2-\\uffc7\\uffca-\\uffcf\\uffd2-\\uffd7\\uffda-\\uffdc]", |
|
"punctuationPattern": "[\\u0021-\\u002f\\u003a-\\u0040\\u005b-\\u0060\\u007b-\\u007e\\u00a1\\u00a7\\u00ab\\u00b6-\\u00b7\\u00bb\\u00bf\\u037e\\u0387\\u055a-\\u055f\\u0589-\\u058a\\u05be\\u05c0\\u05c3\\u05c6\\u05f3-\\u05f4\\u0609-\\u060a\\u060c-\\u060d\\u061b\\u061e-\\u061f\\u066a-\\u066d\\u06d4\\u0700-\\u070d\\u07f7-\\u07f9\\u0830-\\u083e\\u085e\\u0964-\\u0965\\u0970\\u09fd\\u0a76\\u0af0\\u0c77\\u0c84\\u0df4\\u0e4f\\u0e5a-\\u0e5b\\u0f04-\\u0f12\\u0f14\\u0f3a-\\u0f3d\\u0f85\\u0fd0-\\u0fd4\\u0fd9-\\u0fda\\u104a-\\u104f\\u10fb\\u1360-\\u1368\\u1400\\u166e\\u169b-\\u169c\\u16eb-\\u16ed\\u1735-\\u1736\\u17d4-\\u17d6\\u17d8-\\u17da\\u1800-\\u180a\\u1944-\\u1945\\u1a1e-\\u1a1f\\u1aa0-\\u1aa6\\u1aa8-\\u1aad\\u1b5a-\\u1b60\\u1bfc-\\u1bff\\u1c3b-\\u1c3f\\u1c7e-\\u1c7f\\u1cc0-\\u1cc7\\u1cd3\\u2010-\\u2027\\u2030-\\u2043\\u2045-\\u2051\\u2053-\\u205e\\u207d-\\u207e\\u208d-\\u208e\\u2308-\\u230b\\u2329-\\u232a\\u2768-\\u2775\\u27c5-\\u27c6\\u27e6-\\u27ef\\u2983-\\u2998\\u29d8-\\u29db\\u29fc-\\u29fd\\u2cf9-\\u2cfc\\u2cfe-\\u2cff\\u2d70\\u2e00-\\u2e2e\\u2e30-\\u2e4f\\u3001-\\u3003\\u3008-\\u3011\\u3014-\\u301f\\u3030\\u303d\\u30a0\\u30fb\\ua4fe-\\ua4ff\\ua60d-\\ua60f\\ua673\\ua67e\\ua6f2-\\ua6f7\\ua874-\\ua877\\ua8ce-\\ua8cf\\ua8f8-\\ua8fa\\ua8fc\\ua92e-\\ua92f\\ua95f\\ua9c1-\\ua9cd\\ua9de-\\ua9df\\uaa5c-\\uaa5f\\uaade-\\uaadf\\uaaf0-\\uaaf1\\uabeb\\ufd3e-\\ufd3f\\ufe10-\\ufe19\\ufe30-\\ufe52\\ufe54-\\ufe61\\ufe63\\ufe68\\ufe6a-\\ufe6b\\uff01-\\uff03\\uff05-\\uff0a\\uff0c-\\uff0f\\uff1a-\\uff1b\\uff1f-\\uff20\\uff3b-\\uff3d\\uff3f\\uff5b\\uff5d\\uff5f-\\uff65]|\\ud800[\\udd00-\\udd02\\udf9f\\udfd0]|\\ud801[\\udd6f]|\\ud802[\\udc57\\udd1f\\udd3f\\ude50-\\ude58\\ude7f\\udef0-\\udef6\\udf39-\\udf3f\\udf99-\\udf9c]|\\ud803[\\udf55-\\udf59]|\\ud804[\\udc47-\\udc4d\\udcbb-\\udcbc\\udcbe-\\udcc1\\udd40-\\udd43\\udd74-\\udd75\\uddc5-\\uddc8\\uddcd\\udddb\\udddd-\\udddf\\ude38-\\ude3d\\udea9]|\\ud805[\\udc4b-\\udc4f\\udc5b\\udc5d\\udcc6\\uddc1-\\uddd7\\ude41-\\ude43\\ude60-\\ude6c\\udf3c-\\udf3e]|\\ud806[\\udc3b\\udde2\\ude3f-\\ude46\\ude9a-\\ude9c\\ude9e-\\udea2]|\\ud807[\\udc41-\\udc45\\udc70-\\udc71\\udef7-\\udef8\\udfff]|\\ud809[\\udc70-\\udc74]|\\ud81a[\\ude6e-\\ude6f\\udef5\\udf37-\\udf3b\\udf44]|\\ud81b[\\ude97-\\ude9a\\udfe2]|\\ud82f[\\udc9f]|\\ud836[\\ude87-\\ude8b]|\\ud83a[\\udd5e-\\udd5f]" |
|
}; |
|
|
|
var cjkPattern = json$1.cjkPattern, |
|
kPattern = json$1.kPattern, |
|
punctuationPattern = json$1.punctuationPattern; |
|
var getLast$3 = util.getLast; |
|
var INLINE_NODE_TYPES = ["liquidNode", "inlineCode", "emphasis", "strong", "delete", "link", "linkReference", "image", "imageReference", "footnote", "footnoteReference", "sentence", "whitespace", "word", "break", "inlineMath"]; |
|
var INLINE_NODE_WRAPPER_TYPES = INLINE_NODE_TYPES.concat(["tableCell", "paragraph", "heading"]); |
|
var kRegex = new RegExp(kPattern); |
|
var punctuationRegex = new RegExp(punctuationPattern); |
|
/** |
|
* split text into whitespaces and words |
|
* @param {string} text |
|
* @return {Array<{ type: "whitespace", value: " " | "\n" | "" } | { type: "word", value: string }>} |
|
*/ |
|
|
|
function splitText(text, options) { |
|
var KIND_NON_CJK = "non-cjk"; |
|
var KIND_CJ_LETTER = "cj-letter"; |
|
var KIND_K_LETTER = "k-letter"; |
|
var KIND_CJK_PUNCTUATION = "cjk-punctuation"; |
|
var nodes = []; |
|
(options.proseWrap === "preserve" ? text : text.replace(new RegExp("(".concat(cjkPattern, ")\n(").concat(cjkPattern, ")"), "g"), "$1$2")).split(/([ \t\n]+)/).forEach(function (token, index, tokens) { |
|
// whitespace |
|
if (index % 2 === 1) { |
|
nodes.push({ |
|
type: "whitespace", |
|
value: /\n/.test(token) ? "\n" : " " |
|
}); |
|
return; |
|
} // word separated by whitespace |
|
|
|
|
|
if ((index === 0 || index === tokens.length - 1) && token === "") { |
|
return; |
|
} |
|
|
|
token.split(new RegExp("(".concat(cjkPattern, ")"))).forEach(function (innerToken, innerIndex, innerTokens) { |
|
if ((innerIndex === 0 || innerIndex === innerTokens.length - 1) && innerToken === "") { |
|
return; |
|
} // non-CJK word |
|
|
|
|
|
if (innerIndex % 2 === 0) { |
|
if (innerToken !== "") { |
|
appendNode({ |
|
type: "word", |
|
value: innerToken, |
|
kind: KIND_NON_CJK, |
|
hasLeadingPunctuation: punctuationRegex.test(innerToken[0]), |
|
hasTrailingPunctuation: punctuationRegex.test(getLast$3(innerToken)) |
|
}); |
|
} |
|
|
|
return; |
|
} // CJK character |
|
|
|
|
|
appendNode(punctuationRegex.test(innerToken) ? { |
|
type: "word", |
|
value: innerToken, |
|
kind: KIND_CJK_PUNCTUATION, |
|
hasLeadingPunctuation: true, |
|
hasTrailingPunctuation: true |
|
} : { |
|
type: "word", |
|
value: innerToken, |
|
kind: kRegex.test(innerToken) ? KIND_K_LETTER : KIND_CJ_LETTER, |
|
hasLeadingPunctuation: false, |
|
hasTrailingPunctuation: false |
|
}); |
|
}); |
|
}); |
|
return nodes; |
|
|
|
function appendNode(node) { |
|
var lastNode = getLast$3(nodes); |
|
|
|
if (lastNode && lastNode.type === "word") { |
|
if (lastNode.kind === KIND_NON_CJK && node.kind === KIND_CJ_LETTER && !lastNode.hasTrailingPunctuation || lastNode.kind === KIND_CJ_LETTER && node.kind === KIND_NON_CJK && !node.hasLeadingPunctuation) { |
|
nodes.push({ |
|
type: "whitespace", |
|
value: " " |
|
}); |
|
} else if (!isBetween(KIND_NON_CJK, KIND_CJK_PUNCTUATION) && // disallow leading/trailing full-width whitespace |
|
![lastNode.value, node.value].some(function (value) { |
|
return /\u3000/.test(value); |
|
})) { |
|
nodes.push({ |
|
type: "whitespace", |
|
value: "" |
|
}); |
|
} |
|
} |
|
|
|
nodes.push(node); |
|
|
|
function isBetween(kind1, kind2) { |
|
return lastNode.kind === kind1 && node.kind === kind2 || lastNode.kind === kind2 && node.kind === kind1; |
|
} |
|
} |
|
} |
|
|
|
function getOrderedListItemInfo(orderListItem, originalText) { |
|
var _originalText$slice$m = originalText.slice(orderListItem.position.start.offset, orderListItem.position.end.offset).match(/^\s*(\d+)(\.|\))(\s*)/), |
|
_originalText$slice$m2 = _slicedToArray(_originalText$slice$m, 4), |
|
numberText = _originalText$slice$m2[1], |
|
marker = _originalText$slice$m2[2], |
|
leadingSpaces = _originalText$slice$m2[3]; |
|
|
|
return { |
|
numberText: numberText, |
|
marker: marker, |
|
leadingSpaces: leadingSpaces |
|
}; |
|
} // workaround for https://github.com/remarkjs/remark/issues/351 |
|
// leading and trailing newlines are stripped by remark |
|
|
|
|
|
function getFencedCodeBlockValue(node, originalText) { |
|
var text = originalText.slice(node.position.start.offset, node.position.end.offset); |
|
var leadingSpaceCount = text.match(/^\s*/)[0].length; |
|
var replaceRegex = new RegExp("^\\s{0,".concat(leadingSpaceCount, "}")); |
|
var lineContents = text.split("\n"); |
|
var markerStyle = text[leadingSpaceCount]; // ` or ~ |
|
|
|
var marker = text.slice(leadingSpaceCount).match(new RegExp("^[".concat(markerStyle, "]+")))[0]; // https://spec.commonmark.org/0.28/#example-104: Closing fences may be indented by 0-3 spaces |
|
// https://spec.commonmark.org/0.28/#example-93: The closing code fence must be at least as long as the opening fence |
|
|
|
var hasEndMarker = new RegExp("^\\s{0,3}".concat(marker)).test(lineContents[lineContents.length - 1].slice(getIndent(lineContents.length - 1))); |
|
return lineContents.slice(1, hasEndMarker ? -1 : undefined).map(function (x, i) { |
|
return x.slice(getIndent(i + 1)).replace(replaceRegex, ""); |
|
}).join("\n"); |
|
|
|
function getIndent(lineIndex) { |
|
return node.position.indent[lineIndex - 1] - 1; |
|
} |
|
} |
|
|
|
function mapAst(ast, handler) { |
|
return function preorder(node, index, parentStack) { |
|
parentStack = parentStack || []; |
|
var newNode = handler(node, index, parentStack); |
|
|
|
if (Array.isArray(newNode)) { |
|
return newNode; |
|
} |
|
|
|
newNode = Object.assign({}, newNode); |
|
|
|
if (newNode.children) { |
|
newNode.children = newNode.children.reduce(function (nodes, child, index) { |
|
var newNodes = preorder(child, index, [newNode].concat(parentStack)); |
|
|
|
if (!Array.isArray(newNodes)) { |
|
newNodes = [newNodes]; |
|
} |
|
|
|
nodes.push.apply(nodes, newNodes); |
|
return nodes; |
|
}, []); |
|
} |
|
|
|
return newNode; |
|
}(ast, null, null); |
|
} |
|
|
|
var utils$5 = { |
|
mapAst: mapAst, |
|
splitText: splitText, |
|
punctuationPattern: punctuationPattern, |
|
getFencedCodeBlockValue: getFencedCodeBlockValue, |
|
getOrderedListItemInfo: getOrderedListItemInfo, |
|
INLINE_NODE_TYPES: INLINE_NODE_TYPES, |
|
INLINE_NODE_WRAPPER_TYPES: INLINE_NODE_WRAPPER_TYPES |
|
}; |
|
|
|
var _require$$0$builders$8 = doc.builders, |
|
hardline$b = _require$$0$builders$8.hardline, |
|
literalline$5 = _require$$0$builders$8.literalline, |
|
concat$f = _require$$0$builders$8.concat, |
|
markAsRoot$3 = _require$$0$builders$8.markAsRoot, |
|
mapDoc$6 = doc.utils.mapDoc; |
|
var getFencedCodeBlockValue$1 = utils$5.getFencedCodeBlockValue; |
|
|
|
function embed$4(path, print, textToDoc, options) { |
|
var node = path.getValue(); |
|
|
|
if (node.type === "code" && node.lang !== null) { |
|
// only look for the first string so as to support [markdown-preview-enhanced](https://shd101wyy.github.io/markdown-preview-enhanced/#/code-chunk) |
|
var langMatch = node.lang.match(/^[A-Za-z0-9_-]+/); |
|
var lang = langMatch ? langMatch[0] : ""; |
|
var parser = getParserName(lang); |
|
|
|
if (parser) { |
|
var styleUnit = options.__inJsTemplate ? "~" : "`"; |
|
var style = styleUnit.repeat(Math.max(3, util.getMaxContinuousCount(node.value, styleUnit) + 1)); |
|
var doc = textToDoc(getFencedCodeBlockValue$1(node, options.originalText), { |
|
parser: parser |
|
}); |
|
return markAsRoot$3(concat$f([style, node.lang, hardline$b, replaceNewlinesWithLiterallines(doc), style])); |
|
} |
|
} |
|
|
|
if (node.type === "yaml") { |
|
return markAsRoot$3(concat$f(["---", hardline$b, node.value && node.value.trim() ? replaceNewlinesWithLiterallines(textToDoc(node.value, { |
|
parser: "yaml" |
|
})) : "", "---"])); |
|
} // MDX |
|
|
|
|
|
switch (node.type) { |
|
case "importExport": |
|
return textToDoc(node.value, { |
|
parser: "babel" |
|
}); |
|
|
|
case "jsx": |
|
return textToDoc(node.value, { |
|
parser: "__js_expression" |
|
}); |
|
} |
|
|
|
return null; |
|
|
|
function getParserName(lang) { |
|
var supportInfo = support.getSupportInfo(null, { |
|
plugins: options.plugins |
|
}); |
|
var language = supportInfo.languages.find(function (language) { |
|
return language.name.toLowerCase() === lang || language.aliases && language.aliases.indexOf(lang) !== -1 || language.extensions && language.extensions.find(function (ext) { |
|
return ext.substring(1) === lang; |
|
}); |
|
}); |
|
|
|
if (language) { |
|
return language.parsers[0]; |
|
} |
|
|
|
return null; |
|
} |
|
|
|
function replaceNewlinesWithLiterallines(doc) { |
|
return mapDoc$6(doc, function (currentDoc) { |
|
return typeof currentDoc === "string" && currentDoc.includes("\n") ? concat$f(currentDoc.split(/(\n)/g).map(function (v, i) { |
|
return i % 2 === 0 ? v : literalline$5; |
|
})) : currentDoc; |
|
}); |
|
} |
|
} |
|
|
|
var embed_1$2 = embed$4; |
|
|
|
var pragmas = ["format", "prettier"]; |
|
|
|
function startWithPragma(text) { |
|
var pragma = "@(".concat(pragmas.join("|"), ")"); |
|
var regex = new RegExp(["<!--\\s*".concat(pragma, "\\s*-->"), "<!--.*\r?\n[\\s\\S]*(^|\n)[^\\S\n]*".concat(pragma, "[^\\S\n]*($|\n)[\\s\\S]*\n.*-->")].join("|"), "m"); |
|
var matched = text.match(regex); |
|
return matched && matched.index === 0; |
|
} |
|
|
|
var pragma$4 = { |
|
startWithPragma: startWithPragma, |
|
hasPragma: function hasPragma(text) { |
|
return startWithPragma(frontMatter(text).content.trimLeft()); |
|
}, |
|
insertPragma: function insertPragma(text) { |
|
var extracted = frontMatter(text); |
|
var pragma = "<!-- @".concat(pragmas[0], " -->"); |
|
return extracted.frontMatter ? "".concat(extracted.frontMatter.raw, "\n\n").concat(pragma, "\n\n").concat(extracted.content) : "".concat(pragma, "\n\n").concat(extracted.content); |
|
} |
|
}; |
|
|
|
var getOrderedListItemInfo$1 = utils$5.getOrderedListItemInfo, |
|
mapAst$1 = utils$5.mapAst, |
|
splitText$1 = utils$5.splitText; // 0x0 ~ 0x10ffff |
|
// eslint-disable-next-line no-control-regex |
|
|
|
var isSingleCharRegex = /^([\u0000-\uffff]|[\ud800-\udbff][\udc00-\udfff])$/; |
|
|
|
function preprocess$2(ast, options) { |
|
ast = restoreUnescapedCharacter(ast, options); |
|
ast = mergeContinuousTexts(ast); |
|
ast = transformInlineCode(ast); |
|
ast = transformIndentedCodeblockAndMarkItsParentList(ast, options); |
|
ast = markAlignedList(ast, options); |
|
ast = splitTextIntoSentences(ast, options); |
|
ast = transformImportExport(ast); |
|
ast = mergeContinuousImportExport(ast); |
|
return ast; |
|
} |
|
|
|
function transformImportExport(ast) { |
|
return mapAst$1(ast, function (node) { |
|
if (node.type !== "import" && node.type !== "export") { |
|
return node; |
|
} |
|
|
|
return Object.assign({}, node, { |
|
type: "importExport" |
|
}); |
|
}); |
|
} |
|
|
|
function transformInlineCode(ast) { |
|
return mapAst$1(ast, function (node) { |
|
if (node.type !== "inlineCode") { |
|
return node; |
|
} |
|
|
|
return Object.assign({}, node, { |
|
value: node.value.replace(/\s+/g, " ") |
|
}); |
|
}); |
|
} |
|
|
|
function restoreUnescapedCharacter(ast, options) { |
|
return mapAst$1(ast, function (node) { |
|
return node.type !== "text" ? node : Object.assign({}, node, { |
|
value: node.value !== "*" && node.value !== "_" && node.value !== "$" && // handle these cases in printer |
|
isSingleCharRegex.test(node.value) && node.position.end.offset - node.position.start.offset !== node.value.length ? options.originalText.slice(node.position.start.offset, node.position.end.offset) : node.value |
|
}); |
|
}); |
|
} |
|
|
|
function mergeContinuousImportExport(ast) { |
|
return mergeChildren(ast, function (prevNode, node) { |
|
return prevNode.type === "importExport" && node.type === "importExport"; |
|
}, function (prevNode, node) { |
|
return { |
|
type: "importExport", |
|
value: prevNode.value + "\n\n" + node.value, |
|
position: { |
|
start: prevNode.position.start, |
|
end: node.position.end |
|
} |
|
}; |
|
}); |
|
} |
|
|
|
function mergeChildren(ast, shouldMerge, mergeNode) { |
|
return mapAst$1(ast, function (node) { |
|
if (!node.children) { |
|
return node; |
|
} |
|
|
|
var children = node.children.reduce(function (current, child) { |
|
var lastChild = current[current.length - 1]; |
|
|
|
if (lastChild && shouldMerge(lastChild, child)) { |
|
current.splice(-1, 1, mergeNode(lastChild, child)); |
|
} else { |
|
current.push(child); |
|
} |
|
|
|
return current; |
|
}, []); |
|
return Object.assign({}, node, { |
|
children: children |
|
}); |
|
}); |
|
} |
|
|
|
function mergeContinuousTexts(ast) { |
|
return mergeChildren(ast, function (prevNode, node) { |
|
return prevNode.type === "text" && node.type === "text"; |
|
}, function (prevNode, node) { |
|
return { |
|
type: "text", |
|
value: prevNode.value + node.value, |
|
position: { |
|
start: prevNode.position.start, |
|
end: node.position.end |
|
} |
|
}; |
|
}); |
|
} |
|
|
|
function splitTextIntoSentences(ast, options) { |
|
return mapAst$1(ast, function (node, index, _ref) { |
|
var _ref2 = _slicedToArray(_ref, 1), |
|
parentNode = _ref2[0]; |
|
|
|
if (node.type !== "text") { |
|
return node; |
|
} |
|
|
|
var value = node.value; |
|
|
|
if (parentNode.type === "paragraph") { |
|
if (index === 0) { |
|
value = value.trimLeft(); |
|
} |
|
|
|
if (index === parentNode.children.length - 1) { |
|
value = value.trimRight(); |
|
} |
|
} |
|
|
|
return { |
|
type: "sentence", |
|
position: node.position, |
|
children: splitText$1(value, options) |
|
}; |
|
}); |
|
} |
|
|
|
function transformIndentedCodeblockAndMarkItsParentList(ast, options) { |
|
return mapAst$1(ast, function (node, index, parentStack) { |
|
if (node.type === "code") { |
|
// the first char may point to `\n`, e.g. `\n\t\tbar`, just ignore it |
|
var isIndented = /^\n?( {4,}|\t)/.test(options.originalText.slice(node.position.start.offset, node.position.end.offset)); |
|
node.isIndented = isIndented; |
|
|
|
if (isIndented) { |
|
for (var i = 0; i < parentStack.length; i++) { |
|
var parent = parentStack[i]; // no need to check checked items |
|
|
|
if (parent.hasIndentedCodeblock) { |
|
break; |
|
} |
|
|
|
if (parent.type === "list") { |
|
parent.hasIndentedCodeblock = true; |
|
} |
|
} |
|
} |
|
} |
|
|
|
return node; |
|
}); |
|
} |
|
|
|
function markAlignedList(ast, options) { |
|
return mapAst$1(ast, function (node, index, parentStack) { |
|
if (node.type === "list" && node.children.length !== 0) { |
|
// if one of its parents is not aligned, it's not possible to be aligned in sub-lists |
|
for (var i = 0; i < parentStack.length; i++) { |
|
var parent = parentStack[i]; |
|
|
|
if (parent.type === "list" && !parent.isAligned) { |
|
node.isAligned = false; |
|
return node; |
|
} |
|
} |
|
|
|
node.isAligned = isAligned(node); |
|
} |
|
|
|
return node; |
|
}); |
|
|
|
function getListItemStart(listItem) { |
|
return listItem.children.length === 0 ? -1 : listItem.children[0].position.start.column - 1; |
|
} |
|
|
|
function isAligned(list) { |
|
if (!list.ordered) { |
|
/** |
|
* - 123 |
|
* - 123 |
|
*/ |
|
return true; |
|
} |
|
|
|
var _list$children = _slicedToArray(list.children, 2), |
|
firstItem = _list$children[0], |
|
secondItem = _list$children[1]; |
|
|
|
var firstInfo = getOrderedListItemInfo$1(firstItem, options.originalText); |
|
|
|
if (firstInfo.leadingSpaces.length > 1) { |
|
/** |
|
* 1. 123 |
|
* |
|
* 1. 123 |
|
* 1. 123 |
|
*/ |
|
return true; |
|
} |
|
|
|
var firstStart = getListItemStart(firstItem); |
|
|
|
if (firstStart === -1) { |
|
/** |
|
* 1. |
|
* |
|
* 1. |
|
* 1. |
|
*/ |
|
return false; |
|
} |
|
|
|
if (list.children.length === 1) { |
|
/** |
|
* aligned: |
|
* |
|
* 11. 123 |
|
* |
|
* not aligned: |
|
* |
|
* 1. 123 |
|
*/ |
|
return firstStart % options.tabWidth === 0; |
|
} |
|
|
|
var secondStart = getListItemStart(secondItem); |
|
|
|
if (firstStart !== secondStart) { |
|
/** |
|
* 11. 123 |
|
* 1. 123 |
|
* |
|
* 1. 123 |
|
* 11. 123 |
|
*/ |
|
return false; |
|
} |
|
|
|
if (firstStart % options.tabWidth === 0) { |
|
/** |
|
* 11. 123 |
|
* 12. 123 |
|
*/ |
|
return true; |
|
} |
|
/** |
|
* aligned: |
|
* |
|
* 11. 123 |
|
* 1. 123 |
|
* |
|
* not aligned: |
|
* |
|
* 1. 123 |
|
* 2. 123 |
|
*/ |
|
|
|
|
|
var secondInfo = getOrderedListItemInfo$1(secondItem, options.originalText); |
|
return secondInfo.leadingSpaces.length > 1; |
|
} |
|
} |
|
|
|
var preprocess_1$2 = preprocess$2; |
|
|
|
var _require$$0$builders$9 = doc.builders, |
|
breakParent$4 = _require$$0$builders$9.breakParent, |
|
concat$g = _require$$0$builders$9.concat, |
|
join$b = _require$$0$builders$9.join, |
|
line$8 = _require$$0$builders$9.line, |
|
literalline$6 = _require$$0$builders$9.literalline, |
|
markAsRoot$4 = _require$$0$builders$9.markAsRoot, |
|
hardline$c = _require$$0$builders$9.hardline, |
|
softline$7 = _require$$0$builders$9.softline, |
|
ifBreak$7 = _require$$0$builders$9.ifBreak, |
|
fill$5 = _require$$0$builders$9.fill, |
|
align$2 = _require$$0$builders$9.align, |
|
indent$9 = _require$$0$builders$9.indent, |
|
group$f = _require$$0$builders$9.group, |
|
mapDoc$7 = doc.utils.mapDoc, |
|
printDocToString$2 = doc.printer.printDocToString; |
|
var getFencedCodeBlockValue$2 = utils$5.getFencedCodeBlockValue, |
|
getOrderedListItemInfo$2 = utils$5.getOrderedListItemInfo, |
|
splitText$2 = utils$5.splitText, |
|
punctuationPattern$1 = utils$5.punctuationPattern, |
|
INLINE_NODE_TYPES$1 = utils$5.INLINE_NODE_TYPES, |
|
INLINE_NODE_WRAPPER_TYPES$1 = utils$5.INLINE_NODE_WRAPPER_TYPES; |
|
var replaceEndOfLineWith$2 = util.replaceEndOfLineWith; |
|
var TRAILING_HARDLINE_NODES = ["importExport"]; |
|
var SINGLE_LINE_NODE_TYPES = ["heading", "tableCell", "link"]; |
|
var SIBLING_NODE_TYPES = ["listItem", "definition", "footnoteDefinition", "jsx"]; |
|
|
|
function genericPrint$5(path, options, print) { |
|
var node = path.getValue(); |
|
|
|
if (shouldRemainTheSameContent(path)) { |
|
return concat$g(splitText$2(options.originalText.slice(node.position.start.offset, node.position.end.offset), options).map(function (node) { |
|
return node.type === "word" ? node.value : node.value === "" ? "" : printLine(path, node.value, options); |
|
})); |
|
} |
|
|
|
switch (node.type) { |
|
case "root": |
|
if (node.children.length === 0) { |
|
return ""; |
|
} |
|
|
|
return concat$g([normalizeDoc(printRoot(path, options, print)), TRAILING_HARDLINE_NODES.indexOf(getLastDescendantNode(node).type) === -1 ? hardline$c : ""]); |
|
|
|
case "paragraph": |
|
return printChildren$2(path, options, print, { |
|
postprocessor: fill$5 |
|
}); |
|
|
|
case "sentence": |
|
return printChildren$2(path, options, print); |
|
|
|
case "word": |
|
return node.value.replace(/[*$]/g, "\\$&") // escape all `*` and `$` (math) |
|
.replace(new RegExp(["(^|".concat(punctuationPattern$1, ")(_+)"), "(_+)(".concat(punctuationPattern$1, "|$)")].join("|"), "g"), function (_, text1, underscore1, underscore2, text2) { |
|
return (underscore1 ? "".concat(text1).concat(underscore1) : "".concat(underscore2).concat(text2)).replace(/_/g, "\\_"); |
|
}); |
|
// escape all `_` except concating with non-punctuation, e.g. `1_2_3` is not considered emphasis |
|
|
|
case "whitespace": |
|
{ |
|
var parentNode = path.getParentNode(); |
|
var index = parentNode.children.indexOf(node); |
|
var nextNode = parentNode.children[index + 1]; |
|
var proseWrap = // leading char that may cause different syntax |
|
nextNode && /^>|^([-+*]|#{1,6}|[0-9]+[.)])$/.test(nextNode.value) ? "never" : options.proseWrap; |
|
return printLine(path, node.value, { |
|
proseWrap: proseWrap |
|
}); |
|
} |
|
|
|
case "emphasis": |
|
{ |
|
var _parentNode = path.getParentNode(); |
|
|
|
var _index = _parentNode.children.indexOf(node); |
|
|
|
var prevNode = _parentNode.children[_index - 1]; |
|
var _nextNode = _parentNode.children[_index + 1]; |
|
var hasPrevOrNextWord = // `1*2*3` is considered emphasis but `1_2_3` is not |
|
prevNode && prevNode.type === "sentence" && prevNode.children.length > 0 && util.getLast(prevNode.children).type === "word" && !util.getLast(prevNode.children).hasTrailingPunctuation || _nextNode && _nextNode.type === "sentence" && _nextNode.children.length > 0 && _nextNode.children[0].type === "word" && !_nextNode.children[0].hasLeadingPunctuation; |
|
var style = hasPrevOrNextWord || getAncestorNode$2(path, "emphasis") ? "*" : "_"; |
|
return concat$g([style, printChildren$2(path, options, print), style]); |
|
} |
|
|
|
case "strong": |
|
return concat$g(["**", printChildren$2(path, options, print), "**"]); |
|
|
|
case "delete": |
|
return concat$g(["~~", printChildren$2(path, options, print), "~~"]); |
|
|
|
case "inlineCode": |
|
{ |
|
var backtickCount = util.getMinNotPresentContinuousCount(node.value, "`"); |
|
|
|
var _style = "`".repeat(backtickCount || 1); |
|
|
|
var gap = backtickCount ? " " : ""; |
|
return concat$g([_style, gap, node.value, gap, _style]); |
|
} |
|
|
|
case "link": |
|
switch (options.originalText[node.position.start.offset]) { |
|
case "<": |
|
{ |
|
var mailto = "mailto:"; |
|
var url = // <hello@example.com> is parsed as { url: "mailto:hello@example.com" } |
|
node.url.startsWith(mailto) && options.originalText.slice(node.position.start.offset + 1, node.position.start.offset + 1 + mailto.length) !== mailto ? node.url.slice(mailto.length) : node.url; |
|
return concat$g(["<", url, ">"]); |
|
} |
|
|
|
case "[": |
|
return concat$g(["[", printChildren$2(path, options, print), "](", printUrl(node.url, ")"), printTitle(node.title, options), ")"]); |
|
|
|
default: |
|
return options.originalText.slice(node.position.start.offset, node.position.end.offset); |
|
} |
|
|
|
case "image": |
|
return concat$g([""), printTitle(node.title, options), ")"]); |
|
|
|
case "blockquote": |
|
return concat$g(["> ", align$2("> ", printChildren$2(path, options, print))]); |
|
|
|
case "heading": |
|
return concat$g(["#".repeat(node.depth) + " ", printChildren$2(path, options, print)]); |
|
|
|
case "code": |
|
{ |
|
if (node.isIndented) { |
|
// indented code block |
|
var alignment = " ".repeat(4); |
|
return align$2(alignment, concat$g([alignment, concat$g(replaceEndOfLineWith$2(node.value, hardline$c))])); |
|
} // fenced code block |
|
|
|
|
|
var styleUnit = options.__inJsTemplate ? "~" : "`"; |
|
|
|
var _style2 = styleUnit.repeat(Math.max(3, util.getMaxContinuousCount(node.value, styleUnit) + 1)); |
|
|
|
return concat$g([_style2, node.lang || "", hardline$c, concat$g(replaceEndOfLineWith$2(getFencedCodeBlockValue$2(node, options.originalText), hardline$c)), hardline$c, _style2]); |
|
} |
|
|
|
case "yaml": |
|
case "toml": |
|
return options.originalText.slice(node.position.start.offset, node.position.end.offset); |
|
|
|
case "html": |
|
{ |
|
var _parentNode2 = path.getParentNode(); |
|
|
|
var value = _parentNode2.type === "root" && util.getLast(_parentNode2.children) === node ? node.value.trimRight() : node.value; |
|
var isHtmlComment = /^<!--[\s\S]*-->$/.test(value); |
|
return concat$g(replaceEndOfLineWith$2(value, isHtmlComment ? hardline$c : markAsRoot$4(literalline$6))); |
|
} |
|
|
|
case "list": |
|
{ |
|
var nthSiblingIndex = getNthListSiblingIndex(node, path.getParentNode()); |
|
var isGitDiffFriendlyOrderedList = node.ordered && node.children.length > 1 && +getOrderedListItemInfo$2(node.children[1], options.originalText).numberText === 1; |
|
return printChildren$2(path, options, print, { |
|
processor: function processor(childPath, index) { |
|
var prefix = getPrefix(); |
|
return concat$g([prefix, align$2(" ".repeat(prefix.length), printListItem(childPath, options, print, prefix))]); |
|
|
|
function getPrefix() { |
|
var rawPrefix = node.ordered ? (index === 0 ? node.start : isGitDiffFriendlyOrderedList ? 1 : node.start + index) + (nthSiblingIndex % 2 === 0 ? ". " : ") ") : nthSiblingIndex % 2 === 0 ? "- " : "* "; |
|
return node.isAligned || |
|
/* workaround for https://github.com/remarkjs/remark/issues/315 */ |
|
node.hasIndentedCodeblock ? alignListPrefix(rawPrefix, options) : rawPrefix; |
|
} |
|
} |
|
}); |
|
} |
|
|
|
case "thematicBreak": |
|
{ |
|
var counter = getAncestorCounter$1(path, "list"); |
|
|
|
if (counter === -1) { |
|
return "---"; |
|
} |
|
|
|
var _nthSiblingIndex = getNthListSiblingIndex(path.getParentNode(counter), path.getParentNode(counter + 1)); |
|
|
|
return _nthSiblingIndex % 2 === 0 ? "***" : "---"; |
|
} |
|
|
|
case "linkReference": |
|
return concat$g(["[", printChildren$2(path, options, print), "]", node.referenceType === "full" ? concat$g(["[", node.identifier, "]"]) : node.referenceType === "collapsed" ? "[]" : ""]); |
|
|
|
case "imageReference": |
|
switch (node.referenceType) { |
|
case "full": |
|
return concat$g(["![", node.alt || "", "][", node.identifier, "]"]); |
|
|
|
default: |
|
return concat$g(["![", node.alt, "]", node.referenceType === "collapsed" ? "[]" : ""]); |
|
} |
|
|
|
case "definition": |
|
{ |
|
var lineOrSpace = options.proseWrap === "always" ? line$8 : " "; |
|
return group$f(concat$g([concat$g(["[", node.identifier, "]:"]), indent$9(concat$g([lineOrSpace, printUrl(node.url), node.title === null ? "" : concat$g([lineOrSpace, printTitle(node.title, options, false)])]))])); |
|
} |
|
|
|
case "footnote": |
|
return concat$g(["[^", printChildren$2(path, options, print), "]"]); |
|
|
|
case "footnoteReference": |
|
return concat$g(["[^", node.identifier, "]"]); |
|
|
|
case "footnoteDefinition": |
|
{ |
|
var _nextNode2 = path.getParentNode().children[path.getName() + 1]; |
|
var shouldInlineFootnote = node.children.length === 1 && node.children[0].type === "paragraph" && (options.proseWrap === "never" || options.proseWrap === "preserve" && node.children[0].position.start.line === node.children[0].position.end.line); |
|
return concat$g(["[^", node.identifier, "]: ", shouldInlineFootnote ? printChildren$2(path, options, print) : group$f(concat$g([align$2(" ".repeat(options.tabWidth), printChildren$2(path, options, print, { |
|
processor: function processor(childPath, index) { |
|
return index === 0 ? group$f(concat$g([softline$7, softline$7, childPath.call(print)])) : childPath.call(print); |
|
} |
|
})), _nextNode2 && _nextNode2.type === "footnoteDefinition" ? softline$7 : ""]))]); |
|
} |
|
|
|
case "table": |
|
return printTable(path, options, print); |
|
|
|
case "tableCell": |
|
return printChildren$2(path, options, print); |
|
|
|
case "break": |
|
return /\s/.test(options.originalText[node.position.start.offset]) ? concat$g([" ", markAsRoot$4(literalline$6)]) : concat$g(["\\", hardline$c]); |
|
|
|
case "liquidNode": |
|
return concat$g(replaceEndOfLineWith$2(node.value, hardline$c)); |
|
// MDX |
|
|
|
case "importExport": |
|
case "jsx": |
|
return node.value; |
|
// fallback to the original text if multiparser failed |
|
|
|
case "math": |
|
return concat$g(["$$", hardline$c, node.value ? concat$g([concat$g(replaceEndOfLineWith$2(node.value, hardline$c)), hardline$c]) : "", "$$"]); |
|
|
|
case "inlineMath": |
|
{ |
|
// remark-math trims content but we don't want to remove whitespaces |
|
// since it's very possible that it's recognized as math accidentally |
|
return options.originalText.slice(options.locStart(node), options.locEnd(node)); |
|
} |
|
|
|
case "tableRow": // handled in "table" |
|
|
|
case "listItem": // handled in "list" |
|
|
|
default: |
|
throw new Error("Unknown markdown type ".concat(JSON.stringify(node.type))); |
|
} |
|
} |
|
|
|
function printListItem(path, options, print, listPrefix) { |
|
var node = path.getValue(); |
|
var prefix = node.checked === null ? "" : node.checked ? "[x] " : "[ ] "; |
|
return concat$g([prefix, printChildren$2(path, options, print, { |
|
processor: function processor(childPath, index) { |
|
if (index === 0 && childPath.getValue().type !== "list") { |
|
return align$2(" ".repeat(prefix.length), childPath.call(print)); |
|
} |
|
|
|
var alignment = " ".repeat(clamp(options.tabWidth - listPrefix.length, 0, 3) // 4+ will cause indented code block |
|
); |
|
return concat$g([alignment, align$2(alignment, childPath.call(print))]); |
|
} |
|
})]); |
|
} |
|
|
|
function alignListPrefix(prefix, options) { |
|
var additionalSpaces = getAdditionalSpaces(); |
|
return prefix + " ".repeat(additionalSpaces >= 4 ? 0 : additionalSpaces // 4+ will cause indented code block |
|
); |
|
|
|
function getAdditionalSpaces() { |
|
var restSpaces = prefix.length % options.tabWidth; |
|
return restSpaces === 0 ? 0 : options.tabWidth - restSpaces; |
|
} |
|
} |
|
|
|
function getNthListSiblingIndex(node, parentNode) { |
|
return getNthSiblingIndex(node, parentNode, function (siblingNode) { |
|
return siblingNode.ordered === node.ordered; |
|
}); |
|
} |
|
|
|
function getNthSiblingIndex(node, parentNode, condition) { |
|
condition = condition || function () { |
|
return true; |
|
}; |
|
|
|
var index = -1; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = parentNode.children[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var childNode = _step.value; |
|
|
|
if (childNode.type === node.type && condition(childNode)) { |
|
index++; |
|
} else { |
|
index = -1; |
|
} |
|
|
|
if (childNode === node) { |
|
return index; |
|
} |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
} |
|
|
|
function getAncestorCounter$1(path, typeOrTypes) { |
|
var types = [].concat(typeOrTypes); |
|
var counter = -1; |
|
var ancestorNode; |
|
|
|
while (ancestorNode = path.getParentNode(++counter)) { |
|
if (types.indexOf(ancestorNode.type) !== -1) { |
|
return counter; |
|
} |
|
} |
|
|
|
return -1; |
|
} |
|
|
|
function getAncestorNode$2(path, typeOrTypes) { |
|
var counter = getAncestorCounter$1(path, typeOrTypes); |
|
return counter === -1 ? null : path.getParentNode(counter); |
|
} |
|
|
|
function printLine(path, value, options) { |
|
if (options.proseWrap === "preserve" && value === "\n") { |
|
return hardline$c; |
|
} |
|
|
|
var isBreakable = options.proseWrap === "always" && !getAncestorNode$2(path, SINGLE_LINE_NODE_TYPES); |
|
return value !== "" ? isBreakable ? line$8 : " " : isBreakable ? softline$7 : ""; |
|
} |
|
|
|
function printTable(path, options, print) { |
|
var hardlineWithoutBreakParent = hardline$c.parts[0]; |
|
var node = path.getValue(); |
|
var contents = []; // { [rowIndex: number]: { [columnIndex: number]: string } } |
|
|
|
path.map(function (rowPath) { |
|
var rowContents = []; |
|
rowPath.map(function (cellPath) { |
|
rowContents.push(printDocToString$2(cellPath.call(print), options).formatted); |
|
}, "children"); |
|
contents.push(rowContents); |
|
}, "children"); // Get the width of each column |
|
|
|
var columnMaxWidths = contents.reduce(function (currentWidths, rowContents) { |
|
return currentWidths.map(function (width, columnIndex) { |
|
return Math.max(width, util.getStringWidth(rowContents[columnIndex])); |
|
}); |
|
}, contents[0].map(function () { |
|
return 3; |
|
}) // minimum width = 3 (---, :--, :-:, --:) |
|
); |
|
var alignedTable = join$b(hardlineWithoutBreakParent, [printRow(contents[0]), printSeparator(), join$b(hardlineWithoutBreakParent, contents.slice(1).map(function (rowContents) { |
|
return printRow(rowContents); |
|
}))]); |
|
|
|
if (options.proseWrap !== "never") { |
|
return concat$g([breakParent$4, alignedTable]); |
|
} // Only if the --prose-wrap never is set and it exceeds the print width. |
|
|
|
|
|
var compactTable = join$b(hardlineWithoutBreakParent, [printRow(contents[0], |
|
/* isCompact */ |
|
true), printSeparator( |
|
/* isCompact */ |
|
true), join$b(hardlineWithoutBreakParent, contents.slice(1).map(function (rowContents) { |
|
return printRow(rowContents, |
|
/* isCompact */ |
|
true); |
|
}))]); |
|
return concat$g([breakParent$4, group$f(ifBreak$7(compactTable, alignedTable))]); |
|
|
|
function printSeparator(isCompact) { |
|
return concat$g(["| ", join$b(" | ", columnMaxWidths.map(function (width, index) { |
|
var spaces = isCompact ? 3 : width; |
|
|
|
switch (node.align[index]) { |
|
case "left": |
|
return ":" + "-".repeat(spaces - 1); |
|
|
|
case "right": |
|
return "-".repeat(spaces - 1) + ":"; |
|
|
|
case "center": |
|
return ":" + "-".repeat(spaces - 2) + ":"; |
|
|
|
default: |
|
return "-".repeat(spaces); |
|
} |
|
})), " |"]); |
|
} |
|
|
|
function printRow(rowContents, isCompact) { |
|
return concat$g(["| ", join$b(" | ", isCompact ? rowContents : rowContents.map(function (rowContent, columnIndex) { |
|
switch (node.align[columnIndex]) { |
|
case "right": |
|
return alignRight(rowContent, columnMaxWidths[columnIndex]); |
|
|
|
case "center": |
|
return alignCenter(rowContent, columnMaxWidths[columnIndex]); |
|
|
|
default: |
|
return alignLeft(rowContent, columnMaxWidths[columnIndex]); |
|
} |
|
})), " |"]); |
|
} |
|
|
|
function alignLeft(text, width) { |
|
var spaces = width - util.getStringWidth(text); |
|
return concat$g([text, " ".repeat(spaces)]); |
|
} |
|
|
|
function alignRight(text, width) { |
|
var spaces = width - util.getStringWidth(text); |
|
return concat$g([" ".repeat(spaces), text]); |
|
} |
|
|
|
function alignCenter(text, width) { |
|
var spaces = width - util.getStringWidth(text); |
|
var left = Math.floor(spaces / 2); |
|
var right = spaces - left; |
|
return concat$g([" ".repeat(left), text, " ".repeat(right)]); |
|
} |
|
} |
|
|
|
function printRoot(path, options, print) { |
|
/** @typedef {{ index: number, offset: number }} IgnorePosition */ |
|
|
|
/** @type {Array<{start: IgnorePosition, end: IgnorePosition}>} */ |
|
var ignoreRanges = []; |
|
/** @type {IgnorePosition | null} */ |
|
|
|
var ignoreStart = null; |
|
var children = path.getValue().children; |
|
children.forEach(function (childNode, index) { |
|
switch (isPrettierIgnore$1(childNode)) { |
|
case "start": |
|
if (ignoreStart === null) { |
|
ignoreStart = { |
|
index: index, |
|
offset: childNode.position.end.offset |
|
}; |
|
} |
|
|
|
break; |
|
|
|
case "end": |
|
if (ignoreStart !== null) { |
|
ignoreRanges.push({ |
|
start: ignoreStart, |
|
end: { |
|
index: index, |
|
offset: childNode.position.start.offset |
|
} |
|
}); |
|
ignoreStart = null; |
|
} |
|
|
|
break; |
|
} |
|
}); |
|
return printChildren$2(path, options, print, { |
|
processor: function processor(childPath, index) { |
|
if (ignoreRanges.length !== 0) { |
|
var ignoreRange = ignoreRanges[0]; |
|
|
|
if (index === ignoreRange.start.index) { |
|
return concat$g([children[ignoreRange.start.index].value, options.originalText.slice(ignoreRange.start.offset, ignoreRange.end.offset), children[ignoreRange.end.index].value]); |
|
} |
|
|
|
if (ignoreRange.start.index < index && index < ignoreRange.end.index) { |
|
return false; |
|
} |
|
|
|
if (index === ignoreRange.end.index) { |
|
ignoreRanges.shift(); |
|
return false; |
|
} |
|
} |
|
|
|
return childPath.call(print); |
|
} |
|
}); |
|
} |
|
|
|
function printChildren$2(path, options, print, events) { |
|
events = events || {}; |
|
var postprocessor = events.postprocessor || concat$g; |
|
|
|
var processor = events.processor || function (childPath) { |
|
return childPath.call(print); |
|
}; |
|
|
|
var node = path.getValue(); |
|
var parts = []; |
|
var lastChildNode; |
|
path.map(function (childPath, index) { |
|
var childNode = childPath.getValue(); |
|
var result = processor(childPath, index); |
|
|
|
if (result !== false) { |
|
var data = { |
|
parts: parts, |
|
prevNode: lastChildNode, |
|
parentNode: node, |
|
options: options |
|
}; |
|
|
|
if (!shouldNotPrePrintHardline(childNode, data)) { |
|
parts.push(hardline$c); |
|
|
|
if (lastChildNode && TRAILING_HARDLINE_NODES.indexOf(lastChildNode.type) !== -1) { |
|
if (shouldPrePrintTripleHardline(childNode, data)) { |
|
parts.push(hardline$c); |
|
} |
|
} else { |
|
if (shouldPrePrintDoubleHardline(childNode, data) || shouldPrePrintTripleHardline(childNode, data)) { |
|
parts.push(hardline$c); |
|
} |
|
|
|
if (shouldPrePrintTripleHardline(childNode, data)) { |
|
parts.push(hardline$c); |
|
} |
|
} |
|
} |
|
|
|
parts.push(result); |
|
lastChildNode = childNode; |
|
} |
|
}, "children"); |
|
return postprocessor(parts); |
|
} |
|
|
|
function getLastDescendantNode(node) { |
|
var current = node; |
|
|
|
while (current.children && current.children.length !== 0) { |
|
current = current.children[current.children.length - 1]; |
|
} |
|
|
|
return current; |
|
} |
|
/** @return {false | 'next' | 'start' | 'end'} */ |
|
|
|
|
|
function isPrettierIgnore$1(node) { |
|
if (node.type !== "html") { |
|
return false; |
|
} |
|
|
|
var match = node.value.match(/^<!--\s*prettier-ignore(?:-(start|end))?\s*-->$/); |
|
return match === null ? false : match[1] ? match[1] : "next"; |
|
} |
|
|
|
function isInlineNode(node) { |
|
return node && INLINE_NODE_TYPES$1.indexOf(node.type) !== -1; |
|
} |
|
|
|
function isEndsWithHardLine(node) { |
|
return node && /\n+$/.test(node.value); |
|
} |
|
|
|
function last(nodes) { |
|
return nodes && nodes[nodes.length - 1]; |
|
} |
|
|
|
function shouldNotPrePrintHardline(node, _ref) { |
|
var parentNode = _ref.parentNode, |
|
parts = _ref.parts, |
|
prevNode = _ref.prevNode; |
|
var isFirstNode = parts.length === 0; |
|
var isInlineHTML = node.type === "html" && INLINE_NODE_WRAPPER_TYPES$1.indexOf(parentNode.type) !== -1; |
|
var isAfterHardlineNode = prevNode && (isEndsWithHardLine(prevNode) || isEndsWithHardLine(last(prevNode.children))); |
|
return isFirstNode || isInlineNode(node) || isInlineHTML || isAfterHardlineNode; |
|
} |
|
|
|
function shouldPrePrintDoubleHardline(node, _ref2) { |
|
var parentNode = _ref2.parentNode, |
|
prevNode = _ref2.prevNode; |
|
var prevNodeType = prevNode && prevNode.type; |
|
var nodeType = node.type; |
|
var isSequence = prevNodeType === nodeType; |
|
var isSiblingNode = isSequence && SIBLING_NODE_TYPES.indexOf(nodeType) !== -1; |
|
var isInTightListItem = parentNode.type === "listItem" && !parentNode.loose; |
|
var isPrevNodeLooseListItem = prevNodeType === "listItem" && prevNode.loose; |
|
var isPrevNodePrettierIgnore = isPrettierIgnore$1(prevNode) === "next"; |
|
var isBlockHtmlWithoutBlankLineBetweenPrevHtml = nodeType === "html" && prevNodeType === "html" && prevNode.position.end.line + 1 === node.position.start.line; |
|
var isJsxInlineSibling = prevNodeType === "jsx" && isInlineNode(node) || nodeType === "jsx" && isInlineNode(prevNode); |
|
return isPrevNodeLooseListItem || !(isSiblingNode || isInTightListItem || isPrevNodePrettierIgnore || isBlockHtmlWithoutBlankLineBetweenPrevHtml || isJsxInlineSibling); |
|
} |
|
|
|
function shouldPrePrintTripleHardline(node, data) { |
|
var isPrevNodeList = data.prevNode && data.prevNode.type === "list"; |
|
var isIndentedCode = node.type === "code" && node.isIndented; |
|
return isPrevNodeList && isIndentedCode; |
|
} |
|
|
|
function shouldRemainTheSameContent(path) { |
|
var ancestorNode = getAncestorNode$2(path, ["linkReference", "imageReference"]); |
|
return ancestorNode && (ancestorNode.type !== "linkReference" || ancestorNode.referenceType !== "full"); |
|
} |
|
|
|
function normalizeDoc(doc) { |
|
return mapDoc$7(doc, function (currentDoc) { |
|
if (!currentDoc.parts) { |
|
return currentDoc; |
|
} |
|
|
|
if (currentDoc.type === "concat" && currentDoc.parts.length === 1) { |
|
return currentDoc.parts[0]; |
|
} |
|
|
|
var parts = []; |
|
currentDoc.parts.forEach(function (part) { |
|
if (part.type === "concat") { |
|
parts.push.apply(parts, part.parts); |
|
} else if (part !== "") { |
|
parts.push(part); |
|
} |
|
}); |
|
return Object.assign({}, currentDoc, { |
|
parts: normalizeParts$2(parts) |
|
}); |
|
}); |
|
} |
|
|
|
function printUrl(url, dangerousCharOrChars) { |
|
var dangerousChars = [" "].concat(dangerousCharOrChars || []); |
|
return new RegExp(dangerousChars.map(function (x) { |
|
return "\\".concat(x); |
|
}).join("|")).test(url) ? "<".concat(url, ">") : url; |
|
} |
|
|
|
function printTitle(title, options, printSpace) { |
|
if (printSpace == null) { |
|
printSpace = true; |
|
} |
|
|
|
if (!title) { |
|
return ""; |
|
} |
|
|
|
if (printSpace) { |
|
return " " + printTitle(title, options, false); |
|
} |
|
|
|
if (title.includes('"') && title.includes("'") && !title.includes(")")) { |
|
return "(".concat(title, ")"); // avoid escaped quotes |
|
} // faster than using RegExps: https://jsperf.com/performance-of-match-vs-split |
|
|
|
|
|
var singleCount = title.split("'").length - 1; |
|
var doubleCount = title.split('"').length - 1; |
|
var quote = singleCount > doubleCount ? '"' : doubleCount > singleCount ? "'" : options.singleQuote ? "'" : '"'; |
|
title = title.replace(new RegExp("(".concat(quote, ")"), "g"), "\\$1"); |
|
return "".concat(quote).concat(title).concat(quote); |
|
} |
|
|
|
function normalizeParts$2(parts) { |
|
return parts.reduce(function (current, part) { |
|
var lastPart = util.getLast(current); |
|
|
|
if (typeof lastPart === "string" && typeof part === "string") { |
|
current.splice(-1, 1, lastPart + part); |
|
} else { |
|
current.push(part); |
|
} |
|
|
|
return current; |
|
}, []); |
|
} |
|
|
|
function clamp(value, min, max) { |
|
return value < min ? min : value > max ? max : value; |
|
} |
|
|
|
function clean$6(ast, newObj, parent) { |
|
delete newObj.position; |
|
delete newObj.raw; // front-matter |
|
// for codeblock |
|
|
|
if (ast.type === "code" || ast.type === "yaml" || ast.type === "import" || ast.type === "export" || ast.type === "jsx") { |
|
delete newObj.value; |
|
} |
|
|
|
if (ast.type === "list") { |
|
delete newObj.isAligned; |
|
} // texts can be splitted or merged |
|
|
|
|
|
if (ast.type === "text") { |
|
return null; |
|
} |
|
|
|
if (ast.type === "inlineCode") { |
|
newObj.value = ast.value.replace(/[ \t\n]+/g, " "); |
|
} // for insert pragma |
|
|
|
|
|
if (parent && parent.type === "root" && parent.children.length > 0 && (parent.children[0] === ast || (parent.children[0].type === "yaml" || parent.children[0].type === "toml") && parent.children[1] === ast) && ast.type === "html" && pragma$4.startWithPragma(ast.value)) { |
|
return null; |
|
} |
|
} |
|
|
|
function hasPrettierIgnore$4(path) { |
|
var index = +path.getName(); |
|
|
|
if (index === 0) { |
|
return false; |
|
} |
|
|
|
var prevNode = path.getParentNode().children[index - 1]; |
|
return isPrettierIgnore$1(prevNode) === "next"; |
|
} |
|
|
|
var printerMarkdown = { |
|
preprocess: preprocess_1$2, |
|
print: genericPrint$5, |
|
embed: embed_1$2, |
|
massageAstNode: clean$6, |
|
hasPrettierIgnore: hasPrettierIgnore$4, |
|
insertPragma: pragma$4.insertPragma |
|
}; |
|
|
|
var options$6 = { |
|
proseWrap: commonOptions.proseWrap, |
|
singleQuote: commonOptions.singleQuote |
|
}; |
|
|
|
var name$g = "Markdown"; |
|
var type$f = "prose"; |
|
var aliases$5 = [ |
|
"pandoc" |
|
]; |
|
var aceMode$f = "markdown"; |
|
var codemirrorMode$b = "gfm"; |
|
var codemirrorMimeType$b = "text/x-gfm"; |
|
var wrap = true; |
|
var extensions$f = [ |
|
".md", |
|
".markdown", |
|
".mdown", |
|
".mdwn", |
|
".mdx", |
|
".mkd", |
|
".mkdn", |
|
".mkdown", |
|
".ronn", |
|
".workbook" |
|
]; |
|
var filenames$3 = [ |
|
"contents.lr" |
|
]; |
|
var tmScope$f = "source.gfm"; |
|
var languageId$f = 222; |
|
var Markdown = { |
|
name: name$g, |
|
type: type$f, |
|
aliases: aliases$5, |
|
aceMode: aceMode$f, |
|
codemirrorMode: codemirrorMode$b, |
|
codemirrorMimeType: codemirrorMimeType$b, |
|
wrap: wrap, |
|
extensions: extensions$f, |
|
filenames: filenames$3, |
|
tmScope: tmScope$f, |
|
languageId: languageId$f |
|
}; |
|
|
|
var Markdown$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$g, |
|
type: type$f, |
|
aliases: aliases$5, |
|
aceMode: aceMode$f, |
|
codemirrorMode: codemirrorMode$b, |
|
codemirrorMimeType: codemirrorMimeType$b, |
|
wrap: wrap, |
|
extensions: extensions$f, |
|
filenames: filenames$3, |
|
tmScope: tmScope$f, |
|
languageId: languageId$f, |
|
'default': Markdown |
|
}); |
|
|
|
var require$$0$7 = getCjsExportFromNamespace(Markdown$1); |
|
|
|
var languages$5 = [createLanguage(require$$0$7, function (data) { |
|
return Object.assign(data, { |
|
since: "1.8.0", |
|
parsers: ["remark"], |
|
vscodeLanguageIds: ["markdown"], |
|
filenames: data.filenames.concat(["README"]), |
|
extensions: data.extensions.filter(function (extension) { |
|
return extension !== ".mdx"; |
|
}) |
|
}); |
|
}), createLanguage(require$$0$7, function (data) { |
|
return Object.assign(data, { |
|
name: "MDX", |
|
since: "1.15.0", |
|
parsers: ["mdx"], |
|
vscodeLanguageIds: ["mdx"], |
|
filenames: [], |
|
extensions: [".mdx"] |
|
}); |
|
})]; |
|
var printers$5 = { |
|
mdast: printerMarkdown |
|
}; |
|
var languageMarkdown = { |
|
languages: languages$5, |
|
options: options$6, |
|
printers: printers$5 |
|
}; |
|
|
|
function isPragma(text) { |
|
return /^\s*@(prettier|format)\s*$/.test(text); |
|
} |
|
|
|
function hasPragma$4(text) { |
|
return /^\s*#[^\n\S]*@(prettier|format)\s*?(\n|$)/.test(text); |
|
} |
|
|
|
function insertPragma$8(text) { |
|
return "# @format\n\n".concat(text); |
|
} |
|
|
|
var pragma$5 = { |
|
isPragma: isPragma, |
|
hasPragma: hasPragma$4, |
|
insertPragma: insertPragma$8 |
|
}; |
|
|
|
var getLast$4 = util.getLast; |
|
|
|
function getAncestorCount(path, filter) { |
|
var counter = 0; |
|
var pathStackLength = path.stack.length - 1; |
|
|
|
for (var i = 0; i < pathStackLength; i++) { |
|
var value = path.stack[i]; |
|
|
|
if (isNode(value) && filter(value)) { |
|
counter++; |
|
} |
|
} |
|
|
|
return counter; |
|
} |
|
/** |
|
* @param {any} value |
|
* @param {string[]=} types |
|
*/ |
|
|
|
|
|
function isNode(value, types) { |
|
return value && typeof value.type === "string" && (!types || types.indexOf(value.type) !== -1); |
|
} |
|
|
|
function mapNode(node, callback, parent) { |
|
return callback("children" in node ? Object.assign({}, node, { |
|
children: node.children.map(function (childNode) { |
|
return mapNode(childNode, callback, node); |
|
}) |
|
}) : node, parent); |
|
} |
|
|
|
function defineShortcut(x, key, getter) { |
|
Object.defineProperty(x, key, { |
|
get: getter, |
|
enumerable: false |
|
}); |
|
} |
|
|
|
function isNextLineEmpty$5(node, text) { |
|
var newlineCount = 0; |
|
var textLength = text.length; |
|
|
|
for (var i = node.position.end.offset - 1; i < textLength; i++) { |
|
var char = text[i]; |
|
|
|
if (char === "\n") { |
|
newlineCount++; |
|
} |
|
|
|
if (newlineCount === 1 && /\S/.test(char)) { |
|
return false; |
|
} |
|
|
|
if (newlineCount === 2) { |
|
return true; |
|
} |
|
} |
|
|
|
return false; |
|
} |
|
|
|
function isLastDescendantNode(path) { |
|
var node = path.getValue(); |
|
|
|
switch (node.type) { |
|
case "tag": |
|
case "anchor": |
|
case "comment": |
|
return false; |
|
} |
|
|
|
var pathStackLength = path.stack.length; |
|
|
|
for (var i = 1; i < pathStackLength; i++) { |
|
var item = path.stack[i]; |
|
var parentItem = path.stack[i - 1]; |
|
|
|
if (Array.isArray(parentItem) && typeof item === "number" && item !== parentItem.length - 1) { |
|
return false; |
|
} |
|
} |
|
|
|
return true; |
|
} |
|
|
|
function getLastDescendantNode$1(node) { |
|
return "children" in node && node.children.length !== 0 ? getLastDescendantNode$1(getLast$4(node.children)) : node; |
|
} |
|
|
|
function isPrettierIgnore$2(comment) { |
|
return comment.value.trim() === "prettier-ignore"; |
|
} |
|
|
|
function hasPrettierIgnore$5(path) { |
|
var node = path.getValue(); |
|
|
|
if (node.type === "documentBody") { |
|
var document = path.getParentNode(); |
|
return hasEndComments(document.head) && isPrettierIgnore$2(getLast$4(document.head.endComments)); |
|
} |
|
|
|
return hasLeadingComments(node) && isPrettierIgnore$2(getLast$4(node.leadingComments)); |
|
} |
|
|
|
function isEmptyNode(node) { |
|
return (!node.children || node.children.length === 0) && !hasComments(node); |
|
} |
|
|
|
function hasComments(node) { |
|
return hasLeadingComments(node) || hasMiddleComments(node) || hasIndicatorComment(node) || hasTrailingComment$2(node) || hasEndComments(node); |
|
} |
|
|
|
function hasLeadingComments(node) { |
|
return node && node.leadingComments && node.leadingComments.length !== 0; |
|
} |
|
|
|
function hasMiddleComments(node) { |
|
return node && node.middleComments && node.middleComments.length !== 0; |
|
} |
|
|
|
function hasIndicatorComment(node) { |
|
return node && node.indicatorComment; |
|
} |
|
|
|
function hasTrailingComment$2(node) { |
|
return node && node.trailingComment; |
|
} |
|
|
|
function hasEndComments(node) { |
|
return node && node.endComments && node.endComments.length !== 0; |
|
} |
|
/** |
|
* " a b c d e f " -> [" a b", "c d", "e f "] |
|
*/ |
|
|
|
|
|
function splitWithSingleSpace(text) { |
|
var parts = []; |
|
var lastPart = undefined; |
|
var _iteratorNormalCompletion = true; |
|
var _didIteratorError = false; |
|
var _iteratorError = undefined; |
|
|
|
try { |
|
for (var _iterator = text.split(/( +)/g)[Symbol.iterator](), _step; !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true) { |
|
var part = _step.value; |
|
|
|
if (part !== " ") { |
|
if (lastPart === " ") { |
|
parts.push(part); |
|
} else { |
|
parts.push((parts.pop() || "") + part); |
|
} |
|
} else if (lastPart === undefined) { |
|
parts.unshift(""); |
|
} |
|
|
|
lastPart = part; |
|
} |
|
} catch (err) { |
|
_didIteratorError = true; |
|
_iteratorError = err; |
|
} finally { |
|
try { |
|
if (!_iteratorNormalCompletion && _iterator.return != null) { |
|
_iterator.return(); |
|
} |
|
} finally { |
|
if (_didIteratorError) { |
|
throw _iteratorError; |
|
} |
|
} |
|
} |
|
|
|
if (lastPart === " ") { |
|
parts.push((parts.pop() || "") + " "); |
|
} |
|
|
|
if (parts[0] === "") { |
|
parts.shift(); |
|
parts.unshift(" " + (parts.shift() || "")); |
|
} |
|
|
|
return parts; |
|
} |
|
|
|
function getFlowScalarLineContents(nodeType, content, options) { |
|
var rawLineContents = content.split("\n").map(function (lineContent, index, lineContents) { |
|
return index === 0 && index === lineContents.length - 1 ? lineContent : index !== 0 && index !== lineContents.length - 1 ? lineContent.trim() : index === 0 ? lineContent.trimRight() : lineContent.trimLeft(); |
|
}); |
|
|
|
if (options.proseWrap === "preserve") { |
|
return rawLineContents.map(function (lineContent) { |
|
return lineContent.length === 0 ? [] : [lineContent]; |
|
}); |
|
} |
|
|
|
return rawLineContents.map(function (lineContent) { |
|
return lineContent.length === 0 ? [] : splitWithSingleSpace(lineContent); |
|
}).reduce(function (reduced, lineContentWords, index) { |
|
return index !== 0 && rawLineContents[index - 1].length !== 0 && lineContentWords.length !== 0 && !( // trailing backslash in quoteDouble should be preserved |
|
nodeType === "quoteDouble" && getLast$4(getLast$4(reduced)).endsWith("\\")) ? reduced.concat([reduced.pop().concat(lineContentWords)]) : reduced.concat([lineContentWords]); |
|
}, []).map(function (lineContentWords) { |
|
return options.proseWrap === "never" ? [lineContentWords.join(" ")] : lineContentWords; |
|
}); |
|
} |
|
|
|
function getBlockValueLineContents(node, _ref) { |
|
var parentIndent = _ref.parentIndent, |
|
isLastDescendant = _ref.isLastDescendant, |
|
options = _ref.options; |
|
var content = node.position.start.line === node.position.end.line ? "" : options.originalText.slice(node.position.start.offset, node.position.end.offset) // exclude open line `>` or `|` |
|
.match(/^[^\n]*?\n([\s\S]*)$/)[1]; |
|
var leadingSpaceCount = node.indent === null ? function (match) { |
|
return match ? match[1].length : Infinity; |
|
}(content.match(/^( *)\S/m)) : node.indent - 1 + parentIndent; |
|
var rawLineContents = content.split("\n").map(function (lineContent) { |
|
return lineContent.slice(leadingSpaceCount); |
|
}); |
|
|
|
if (options.proseWrap === "preserve" || node.type === "blockLiteral") { |
|
return removeUnnecessaryTrailingNewlines(rawLineContents.map(function (lineContent) { |
|
return lineContent.length === 0 ? [] : [lineContent]; |
|
})); |
|
} |
|
|
|
return removeUnnecessaryTrailingNewlines(rawLineContents.map(function (lineContent) { |
|
return lineContent.length === 0 ? [] : splitWithSingleSpace(lineContent); |
|
}).reduce(function (reduced, lineContentWords, index) { |
|
return index !== 0 && rawLineContents[index - 1].length !== 0 && lineContentWords.length !== 0 && !/^\s/.test(lineContentWords[0]) && !/^\s|\s$/.test(getLast$4(reduced)) ? reduced.concat([reduced.pop().concat(lineContentWords)]) : reduced.concat([lineContentWords]); |
|
}, []).map(function (lineContentWords) { |
|
return lineContentWords.reduce(function (reduced, word) { |
|
return (// disallow trailing spaces |
|
reduced.length !== 0 && /\s$/.test(getLast$4(reduced)) ? reduced.concat(reduced.pop() + " " + word) : reduced.concat(word) |
|
); |
|
}, []); |
|
}).map(function (lineContentWords) { |
|
return options.proseWrap === "never" ? [lineContentWords.join(" ")] : lineContentWords; |
|
})); |
|
|
|
function removeUnnecessaryTrailingNewlines(lineContents) { |
|
if (node.chomping === "keep") { |
|
return getLast$4(lineContents).length === 0 ? lineContents.slice(0, -1) : lineContents; |
|
} |
|
|
|
var trailingNewlineCount = 0; |
|
|
|
for (var i = lineContents.length - 1; i >= 0; i--) { |
|
if (lineContents[i].length === 0) { |
|
trailingNewlineCount++; |
|
} else { |
|
break; |
|
} |
|
} |
|
|
|
return trailingNewlineCount === 0 ? lineContents : trailingNewlineCount >= 2 && !isLastDescendant ? // next empty line |
|
lineContents.slice(0, -(trailingNewlineCount - 1)) : lineContents.slice(0, -trailingNewlineCount); |
|
} |
|
} |
|
|
|
var utils$6 = { |
|
getLast: getLast$4, |
|
getAncestorCount: getAncestorCount, |
|
isNode: isNode, |
|
isEmptyNode: isEmptyNode, |
|
mapNode: mapNode, |
|
defineShortcut: defineShortcut, |
|
isNextLineEmpty: isNextLineEmpty$5, |
|
isLastDescendantNode: isLastDescendantNode, |
|
getBlockValueLineContents: getBlockValueLineContents, |
|
getFlowScalarLineContents: getFlowScalarLineContents, |
|
getLastDescendantNode: getLastDescendantNode$1, |
|
hasPrettierIgnore: hasPrettierIgnore$5, |
|
hasLeadingComments: hasLeadingComments, |
|
hasMiddleComments: hasMiddleComments, |
|
hasIndicatorComment: hasIndicatorComment, |
|
hasTrailingComment: hasTrailingComment$2, |
|
hasEndComments: hasEndComments |
|
}; |
|
|
|
var insertPragma$9 = pragma$5.insertPragma, |
|
isPragma$1 = pragma$5.isPragma; |
|
var getAncestorCount$1 = utils$6.getAncestorCount, |
|
getBlockValueLineContents$1 = utils$6.getBlockValueLineContents, |
|
getFlowScalarLineContents$1 = utils$6.getFlowScalarLineContents, |
|
getLast$5 = utils$6.getLast, |
|
getLastDescendantNode$2 = utils$6.getLastDescendantNode, |
|
hasLeadingComments$1 = utils$6.hasLeadingComments, |
|
hasMiddleComments$1 = utils$6.hasMiddleComments, |
|
hasIndicatorComment$1 = utils$6.hasIndicatorComment, |
|
hasTrailingComment$3 = utils$6.hasTrailingComment, |
|
hasEndComments$1 = utils$6.hasEndComments, |
|
hasPrettierIgnore$6 = utils$6.hasPrettierIgnore, |
|
isLastDescendantNode$1 = utils$6.isLastDescendantNode, |
|
isNextLineEmpty$6 = utils$6.isNextLineEmpty, |
|
isNode$1 = utils$6.isNode, |
|
isEmptyNode$1 = utils$6.isEmptyNode, |
|
defineShortcut$1 = utils$6.defineShortcut, |
|
mapNode$1 = utils$6.mapNode; |
|
var docBuilders$2 = doc.builders; |
|
var conditionalGroup$2 = docBuilders$2.conditionalGroup, |
|
breakParent$5 = docBuilders$2.breakParent, |
|
concat$h = docBuilders$2.concat, |
|
dedent$3 = docBuilders$2.dedent, |
|
dedentToRoot$3 = docBuilders$2.dedentToRoot, |
|
fill$6 = docBuilders$2.fill, |
|
group$g = docBuilders$2.group, |
|
hardline$d = docBuilders$2.hardline, |
|
ifBreak$8 = docBuilders$2.ifBreak, |
|
join$c = docBuilders$2.join, |
|
line$9 = docBuilders$2.line, |
|
lineSuffix$2 = docBuilders$2.lineSuffix, |
|
literalline$7 = docBuilders$2.literalline, |
|
markAsRoot$5 = docBuilders$2.markAsRoot, |
|
softline$8 = docBuilders$2.softline; |
|
var replaceEndOfLineWith$3 = util.replaceEndOfLineWith; |
|
|
|
function preprocess$3(ast) { |
|
return mapNode$1(ast, defineShortcuts); |
|
} |
|
|
|
function defineShortcuts(node) { |
|
switch (node.type) { |
|
case "document": |
|
defineShortcut$1(node, "head", function () { |
|
return node.children[0]; |
|
}); |
|
defineShortcut$1(node, "body", function () { |
|
return node.children[1]; |
|
}); |
|
break; |
|
|
|
case "documentBody": |
|
case "sequenceItem": |
|
case "flowSequenceItem": |
|
case "mappingKey": |
|
case "mappingValue": |
|
defineShortcut$1(node, "content", function () { |
|
return node.children[0]; |
|
}); |
|
break; |
|
|
|
case "mappingItem": |
|
case "flowMappingItem": |
|
defineShortcut$1(node, "key", function () { |
|
return node.children[0]; |
|
}); |
|
defineShortcut$1(node, "value", function () { |
|
return node.children[1]; |
|
}); |
|
break; |
|
} |
|
|
|
return node; |
|
} |
|
|
|
function genericPrint$6(path, options, print) { |
|
var node = path.getValue(); |
|
var parentNode = path.getParentNode(); |
|
var tag = !node.tag ? "" : path.call(print, "tag"); |
|
var anchor = !node.anchor ? "" : path.call(print, "anchor"); |
|
var nextEmptyLine = isNode$1(node, ["mapping", "sequence", "comment", "directive", "mappingItem", "sequenceItem"]) && !isLastDescendantNode$1(path) ? printNextEmptyLine(path, options.originalText) : ""; |
|
return concat$h([node.type !== "mappingValue" && hasLeadingComments$1(node) ? concat$h([join$c(hardline$d, path.map(print, "leadingComments")), hardline$d]) : "", tag, tag && anchor ? " " : "", anchor, tag || anchor ? isNode$1(node, ["sequence", "mapping"]) && !hasMiddleComments$1(node) ? hardline$d : " " : "", hasMiddleComments$1(node) ? concat$h([node.middleComments.length === 1 ? "" : hardline$d, join$c(hardline$d, path.map(print, "middleComments")), hardline$d]) : "", hasPrettierIgnore$6(path) ? concat$h(replaceEndOfLineWith$3(options.originalText.slice(node.position.start.offset, node.position.end.offset), literalline$7)) : group$g(_print(node, parentNode, path, options, print)), hasTrailingComment$3(node) && !isNode$1(node, ["document", "documentHead"]) ? lineSuffix$2(concat$h([node.type === "mappingValue" && !node.content ? "" : " ", parentNode.type === "mappingKey" && path.getParentNode(2).type === "mapping" && isInlineNode$1(node) ? "" : breakParent$5, path.call(print, "trailingComment")])) : "", nextEmptyLine, hasEndComments$1(node) && !isNode$1(node, ["documentHead", "documentBody"]) ? align$3(node.type === "sequenceItem" ? 2 : 0, concat$h([hardline$d, join$c(hardline$d, path.map(print, "endComments"))])) : ""]); |
|
} |
|
|
|
function _print(node, parentNode, path, options, print) { |
|
switch (node.type) { |
|
case "root": |
|
return concat$h([join$c(hardline$d, path.map(function (childPath, index) { |
|
var document = node.children[index]; |
|
var nextDocument = node.children[index + 1]; |
|
return concat$h([print(childPath), shouldPrintDocumentEndMarker(document, nextDocument) ? concat$h([hardline$d, "...", hasTrailingComment$3(document) ? concat$h([" ", path.call(print, "trailingComment")]) : ""]) : !nextDocument || hasTrailingComment$3(nextDocument.head) ? "" : concat$h([hardline$d, "---"])]); |
|
}, "children")), node.children.length === 0 || function (lastDescendantNode) { |
|
return isNode$1(lastDescendantNode, ["blockLiteral", "blockFolded"]) && lastDescendantNode.chomping === "keep"; |
|
}(getLastDescendantNode$2(node)) ? "" : hardline$d]); |
|
|
|
case "document": |
|
{ |
|
var nextDocument = parentNode.children[path.getName() + 1]; |
|
return join$c(hardline$d, [shouldPrintDocumentHeadEndMarker(node, nextDocument, parentNode, options) === "head" ? join$c(hardline$d, [node.head.children.length === 0 && node.head.endComments.length === 0 ? "" : path.call(print, "head"), concat$h(["---", hasTrailingComment$3(node.head) ? concat$h([" ", path.call(print, "head", "trailingComment")]) : ""])].filter(Boolean)) : "", shouldPrintDocumentBody(node) ? path.call(print, "body") : ""].filter(Boolean)); |
|
} |
|
|
|
case "documentHead": |
|
return join$c(hardline$d, [].concat(path.map(print, "children"), path.map(print, "endComments"))); |
|
|
|
case "documentBody": |
|
{ |
|
var children = join$c(hardline$d, path.map(print, "children")).parts; |
|
var endComments = join$c(hardline$d, path.map(print, "endComments")).parts; |
|
var separator = children.length === 0 || endComments.length === 0 ? "" : function (lastDescendantNode) { |
|
return isNode$1(lastDescendantNode, ["blockFolded", "blockLiteral"]) ? lastDescendantNode.chomping === "keep" ? // there's already a newline printed at the end of blockValue (chomping=keep, lastDescendant=true) |
|
"" : // an extra newline for better readability |
|
concat$h([hardline$d, hardline$d]) : hardline$d; |
|
}(getLastDescendantNode$2(node)); |
|
return concat$h([].concat(children, separator, endComments)); |
|
} |
|
|
|
case "directive": |
|
return concat$h(["%", join$c(" ", [node.name].concat(node.parameters))]); |
|
|
|
case "comment": |
|
return concat$h(["#", node.value]); |
|
|
|
case "alias": |
|
return concat$h(["*", node.value]); |
|
|
|
case "tag": |
|
return options.originalText.slice(node.position.start.offset, node.position.end.offset); |
|
|
|
case "anchor": |
|
return concat$h(["&", node.value]); |
|
|
|
case "plain": |
|
return printFlowScalarContent(node.type, options.originalText.slice(node.position.start.offset, node.position.end.offset), options); |
|
|
|
case "quoteDouble": |
|
case "quoteSingle": |
|
{ |
|
var singleQuote = "'"; |
|
var doubleQuote = '"'; |
|
var raw = options.originalText.slice(node.position.start.offset + 1, node.position.end.offset - 1); |
|
|
|
if (node.type === "quoteSingle" && raw.includes("\\") || node.type === "quoteDouble" && /\\[^"]/.test(raw)) { |
|
// only quoteDouble can use escape chars |
|
// and quoteSingle do not need to escape backslashes |
|
var originalQuote = node.type === "quoteDouble" ? doubleQuote : singleQuote; |
|
return concat$h([originalQuote, printFlowScalarContent(node.type, raw, options), originalQuote]); |
|
} else if (raw.includes(doubleQuote)) { |
|
return concat$h([singleQuote, printFlowScalarContent(node.type, node.type === "quoteDouble" ? raw // double quote needs to be escaped by backslash in quoteDouble |
|
.replace(/\\"/g, doubleQuote).replace(/'/g, singleQuote.repeat(2)) : raw, options), singleQuote]); |
|
} |
|
|
|
if (raw.includes(singleQuote)) { |
|
return concat$h([doubleQuote, printFlowScalarContent(node.type, node.type === "quoteSingle" ? // single quote needs to be escaped by 2 single quotes in quoteSingle |
|
raw.replace(/''/g, singleQuote) : raw, options), doubleQuote]); |
|
} |
|
|
|
var quote = options.singleQuote ? singleQuote : doubleQuote; |
|
return concat$h([quote, printFlowScalarContent(node.type, raw, options), quote]); |
|
} |
|
|
|
case "blockFolded": |
|
case "blockLiteral": |
|
{ |
|
var parentIndent = getAncestorCount$1(path, function (ancestorNode) { |
|
return isNode$1(ancestorNode, ["sequence", "mapping"]); |
|
}); |
|
var isLastDescendant = isLastDescendantNode$1(path); |
|
return concat$h([node.type === "blockFolded" ? ">" : "|", node.indent === null ? "" : node.indent.toString(), node.chomping === "clip" ? "" : node.chomping === "keep" ? "+" : "-", hasIndicatorComment$1(node) ? concat$h([" ", path.call(print, "indicatorComment")]) : "", (node.indent === null ? dedent$3 : dedentToRoot$3)(align$3(node.indent === null ? options.tabWidth : node.indent - 1 + parentIndent, concat$h(getBlockValueLineContents$1(node, { |
|
parentIndent: parentIndent, |
|
isLastDescendant: isLastDescendant, |
|
options: options |
|
}).reduce(function (reduced, lineWords, index, lineContents) { |
|
return reduced.concat(index === 0 ? hardline$d : "", fill$6(join$c(line$9, lineWords).parts), index !== lineContents.length - 1 ? lineWords.length === 0 ? hardline$d : markAsRoot$5(literalline$7) : node.chomping === "keep" && isLastDescendant ? lineWords.length === 0 ? dedentToRoot$3(hardline$d) : dedentToRoot$3(literalline$7) : ""); |
|
}, []))))]); |
|
} |
|
|
|
case "sequence": |
|
return join$c(hardline$d, path.map(print, "children")); |
|
|
|
case "sequenceItem": |
|
return concat$h(["- ", align$3(2, !node.content ? "" : path.call(print, "content"))]); |
|
|
|
case "mappingKey": |
|
return !node.content ? "" : path.call(print, "content"); |
|
|
|
case "mappingValue": |
|
return !node.content ? "" : path.call(print, "content"); |
|
|
|
case "mapping": |
|
return join$c(hardline$d, path.map(print, "children")); |
|
|
|
case "mappingItem": |
|
case "flowMappingItem": |
|
{ |
|
var isEmptyMappingKey = isEmptyNode$1(node.key); |
|
var isEmptyMappingValue = isEmptyNode$1(node.value); |
|
|
|
if (isEmptyMappingKey && isEmptyMappingValue) { |
|
return concat$h([": "]); |
|
} |
|
|
|
var key = path.call(print, "key"); |
|
var value = path.call(print, "value"); |
|
|
|
if (isEmptyMappingValue) { |
|
return node.type === "flowMappingItem" && parentNode.type === "flowMapping" ? key : node.type === "mappingItem" && isAbsolutelyPrintedAsSingleLineNode(node.key.content, options) && !hasTrailingComment$3(node.key.content) && (!parentNode.tag || parentNode.tag.value !== "tag:yaml.org,2002:set") ? concat$h([key, needsSpaceInFrontOfMappingValue(node) ? " " : "", ":"]) : concat$h(["? ", align$3(2, key)]); |
|
} |
|
|
|
if (isEmptyMappingKey) { |
|
return concat$h([": ", align$3(2, value)]); |
|
} |
|
|
|
var groupId = Symbol("mappingKey"); |
|
var forceExplicitKey = hasLeadingComments$1(node.value) || !isInlineNode$1(node.key.content); |
|
return forceExplicitKey ? concat$h(["? ", align$3(2, key), hardline$d, join$c("", path.map(print, "value", "leadingComments").map(function (comment) { |
|
return concat$h([comment, hardline$d]); |
|
})), ": ", align$3(2, value)]) : // force singleline |
|
isSingleLineNode(node.key.content) && !hasLeadingComments$1(node.key.content) && !hasMiddleComments$1(node.key.content) && !hasTrailingComment$3(node.key.content) && !hasEndComments$1(node.key) && !hasLeadingComments$1(node.value.content) && !hasMiddleComments$1(node.value.content) && !hasEndComments$1(node.value) && isAbsolutelyPrintedAsSingleLineNode(node.value.content, options) ? concat$h([key, needsSpaceInFrontOfMappingValue(node) ? " " : "", ": ", value]) : conditionalGroup$2([concat$h([group$g(concat$h([ifBreak$8("? "), group$g(align$3(2, key), { |
|
id: groupId |
|
})])), ifBreak$8(concat$h([hardline$d, ": ", align$3(2, value)]), indent(concat$h([needsSpaceInFrontOfMappingValue(node) ? " " : "", ":", hasLeadingComments$1(node.value.content) || hasEndComments$1(node.value) && node.value.content && !isNode$1(node.value.content, ["mapping", "sequence"]) || parentNode.type === "mapping" && hasTrailingComment$3(node.key.content) && isInlineNode$1(node.value.content) || isNode$1(node.value.content, ["mapping", "sequence"]) && node.value.content.tag === null && node.value.content.anchor === null ? hardline$d : !node.value.content ? "" : line$9, value])), { |
|
groupId: groupId |
|
})])]); |
|
} |
|
|
|
case "flowMapping": |
|
case "flowSequence": |
|
{ |
|
var openMarker = node.type === "flowMapping" ? "{" : "["; |
|
var closeMarker = node.type === "flowMapping" ? "}" : "]"; |
|
var bracketSpacing = node.type === "flowMapping" && node.children.length !== 0 && options.bracketSpacing ? line$9 : softline$8; |
|
|
|
var isLastItemEmptyMappingItem = node.children.length !== 0 && function (lastItem) { |
|
return lastItem.type === "flowMappingItem" && isEmptyNode$1(lastItem.key) && isEmptyNode$1(lastItem.value); |
|
}(getLast$5(node.children)); |
|
|
|
return concat$h([openMarker, indent(concat$h([bracketSpacing, concat$h(path.map(function (childPath, index) { |
|
return concat$h([print(childPath), index === node.children.length - 1 ? "" : concat$h([",", line$9, node.children[index].position.start.line !== node.children[index + 1].position.start.line ? printNextEmptyLine(childPath, options.originalText) : ""])]); |
|
}, "children")), ifBreak$8(",", "")])), isLastItemEmptyMappingItem ? "" : bracketSpacing, closeMarker]); |
|
} |
|
|
|
case "flowSequenceItem": |
|
return path.call(print, "content"); |
|
// istanbul ignore next |
|
|
|
default: |
|
throw new Error("Unexpected node type ".concat(node.type)); |
|
} |
|
|
|
function indent(doc) { |
|
return docBuilders$2.align(" ".repeat(options.tabWidth), doc); |
|
} |
|
} |
|
|
|
function align$3(n, doc) { |
|
return typeof n === "number" && n > 0 ? docBuilders$2.align(" ".repeat(n), doc) : docBuilders$2.align(n, doc); |
|
} |
|
|
|
function isInlineNode$1(node) { |
|
if (!node) { |
|
return true; |
|
} |
|
|
|
switch (node.type) { |
|
case "plain": |
|
case "quoteDouble": |
|
case "quoteSingle": |
|
case "alias": |
|
case "flowMapping": |
|
case "flowSequence": |
|
return true; |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function isSingleLineNode(node) { |
|
if (!node) { |
|
return true; |
|
} |
|
|
|
switch (node.type) { |
|
case "plain": |
|
case "quoteDouble": |
|
case "quoteSingle": |
|
return node.position.start.line === node.position.end.line; |
|
|
|
case "alias": |
|
return true; |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function shouldPrintDocumentBody(document) { |
|
return document.body.children.length !== 0 || hasEndComments$1(document.body); |
|
} |
|
|
|
function shouldPrintDocumentEndMarker(document, nextDocument) { |
|
return ( |
|
/** |
|
*... # trailingComment |
|
*/ |
|
hasTrailingComment$3(document) || nextDocument && ( |
|
/** |
|
* ... |
|
* %DIRECTIVE |
|
* --- |
|
*/ |
|
nextDocument.head.children.length !== 0 || |
|
/** |
|
* ... |
|
* # endComment |
|
* --- |
|
*/ |
|
hasEndComments$1(nextDocument.head)) |
|
); |
|
} |
|
|
|
function shouldPrintDocumentHeadEndMarker(document, nextDocument, root, options) { |
|
if ( |
|
/** |
|
* --- |
|
* preserve the first document head end marker |
|
*/ |
|
root.children[0] === document && /---(\s|$)/.test(options.originalText.slice(options.locStart(document), options.locStart(document) + 4)) || |
|
/** |
|
* %DIRECTIVE |
|
* --- |
|
*/ |
|
document.head.children.length !== 0 || |
|
/** |
|
* # end comment |
|
* --- |
|
*/ |
|
hasEndComments$1(document.head) || |
|
/** |
|
* --- # trailing comment |
|
*/ |
|
hasTrailingComment$3(document.head)) { |
|
return "head"; |
|
} |
|
|
|
if (shouldPrintDocumentEndMarker(document, nextDocument)) { |
|
return false; |
|
} |
|
|
|
return nextDocument ? "root" : false; |
|
} |
|
|
|
function isAbsolutelyPrintedAsSingleLineNode(node, options) { |
|
if (!node) { |
|
return true; |
|
} |
|
|
|
switch (node.type) { |
|
case "plain": |
|
case "quoteSingle": |
|
case "quoteDouble": |
|
break; |
|
|
|
case "alias": |
|
return true; |
|
|
|
default: |
|
return false; |
|
} |
|
|
|
if (options.proseWrap === "preserve") { |
|
return node.position.start.line === node.position.end.line; |
|
} |
|
|
|
if ( // backslash-newline |
|
/\\$/m.test(options.originalText.slice(node.position.start.offset, node.position.end.offset))) { |
|
return false; |
|
} |
|
|
|
switch (options.proseWrap) { |
|
case "never": |
|
return node.value.indexOf("\n") === -1; |
|
|
|
case "always": |
|
return !/[\n ]/.test(node.value); |
|
// istanbul ignore next |
|
|
|
default: |
|
return false; |
|
} |
|
} |
|
|
|
function needsSpaceInFrontOfMappingValue(node) { |
|
return node.key.content && node.key.content.type === "alias"; |
|
} |
|
|
|
function printNextEmptyLine(path, originalText) { |
|
var node = path.getValue(); |
|
var root = path.stack[0]; |
|
root.isNextEmptyLinePrintedChecklist = root.isNextEmptyLinePrintedChecklist || []; |
|
|
|
if (!root.isNextEmptyLinePrintedChecklist[node.position.end.line]) { |
|
if (isNextLineEmpty$6(node, originalText)) { |
|
root.isNextEmptyLinePrintedChecklist[node.position.end.line] = true; |
|
return softline$8; |
|
} |
|
} |
|
|
|
return ""; |
|
} |
|
|
|
function printFlowScalarContent(nodeType, content, options) { |
|
var lineContents = getFlowScalarLineContents$1(nodeType, content, options); |
|
return join$c(hardline$d, lineContents.map(function (lineContentWords) { |
|
return fill$6(join$c(line$9, lineContentWords).parts); |
|
})); |
|
} |
|
|
|
function clean$7(node, newNode |
|
/*, parent */ |
|
) { |
|
if (isNode$1(newNode)) { |
|
delete newNode.position; |
|
|
|
switch (newNode.type) { |
|
case "comment": |
|
// insert pragma |
|
if (isPragma$1(newNode.value)) { |
|
return null; |
|
} |
|
|
|
break; |
|
|
|
case "quoteDouble": |
|
case "quoteSingle": |
|
newNode.type = "quote"; |
|
break; |
|
} |
|
} |
|
} |
|
|
|
var printerYaml = { |
|
preprocess: preprocess$3, |
|
print: genericPrint$6, |
|
massageAstNode: clean$7, |
|
insertPragma: insertPragma$9 |
|
}; |
|
|
|
var options$7 = { |
|
bracketSpacing: commonOptions.bracketSpacing, |
|
singleQuote: commonOptions.singleQuote, |
|
proseWrap: commonOptions.proseWrap |
|
}; |
|
|
|
var name$h = "YAML"; |
|
var type$g = "data"; |
|
var tmScope$g = "source.yaml"; |
|
var aliases$6 = [ |
|
"yml" |
|
]; |
|
var extensions$g = [ |
|
".yml", |
|
".mir", |
|
".reek", |
|
".rviz", |
|
".sublime-syntax", |
|
".syntax", |
|
".yaml", |
|
".yaml-tmlanguage", |
|
".yml.mysql" |
|
]; |
|
var filenames$4 = [ |
|
".clang-format", |
|
".clang-tidy", |
|
".gemrc", |
|
"glide.lock" |
|
]; |
|
var aceMode$g = "yaml"; |
|
var codemirrorMode$c = "yaml"; |
|
var codemirrorMimeType$c = "text/x-yaml"; |
|
var languageId$g = 407; |
|
var YAML = { |
|
name: name$h, |
|
type: type$g, |
|
tmScope: tmScope$g, |
|
aliases: aliases$6, |
|
extensions: extensions$g, |
|
filenames: filenames$4, |
|
aceMode: aceMode$g, |
|
codemirrorMode: codemirrorMode$c, |
|
codemirrorMimeType: codemirrorMimeType$c, |
|
languageId: languageId$g |
|
}; |
|
|
|
var YAML$1 = /*#__PURE__*/Object.freeze({ |
|
__proto__: null, |
|
name: name$h, |
|
type: type$g, |
|
tmScope: tmScope$g, |
|
aliases: aliases$6, |
|
extensions: extensions$g, |
|
filenames: filenames$4, |
|
aceMode: aceMode$g, |
|
codemirrorMode: codemirrorMode$c, |
|
codemirrorMimeType: codemirrorMimeType$c, |
|
languageId: languageId$g, |
|
'default': YAML |
|
}); |
|
|
|
var require$$0$8 = getCjsExportFromNamespace(YAML$1); |
|
|
|
var languages$6 = [createLanguage(require$$0$8, function (data) { |
|
return Object.assign(data, { |
|
since: "1.14.0", |
|
parsers: ["yaml"], |
|
vscodeLanguageIds: ["yaml"] |
|
}); |
|
})]; |
|
var languageYaml = { |
|
languages: languages$6, |
|
printers: { |
|
yaml: printerYaml |
|
}, |
|
options: options$7 |
|
}; |
|
|
|
var version$2 = require$$0.version; |
|
var getSupportInfo$2 = support.getSupportInfo; |
|
var internalPlugins = [languageCss, languageGraphql, languageHandlebars, languageHtml, languageJs, languageMarkdown, languageYaml]; |
|
|
|
var isArray$1 = Array.isArray || function (arr) { |
|
return Object.prototype.toString.call(arr) === "[object Array]"; |
|
}; // Luckily `opts` is always the 2nd argument |
|
|
|
|
|
function withPlugins(fn) { |
|
return function () { |
|
var args = Array.from(arguments); |
|
var plugins = args[1] && args[1].plugins || []; |
|
|
|
if (!isArray$1(plugins)) { |
|
plugins = Object.values(plugins); |
|
} |
|
|
|
args[1] = Object.assign({}, args[1], { |
|
plugins: internalPlugins.concat(plugins) |
|
}); |
|
return fn.apply(null, args); |
|
}; |
|
} |
|
|
|
var formatWithCursor = withPlugins(core.formatWithCursor); |
|
var standalone = { |
|
formatWithCursor: formatWithCursor, |
|
format: function format(text, opts) { |
|
return formatWithCursor(text, opts).formatted; |
|
}, |
|
check: function check(text, opts) { |
|
var formatted = formatWithCursor(text, opts).formatted; |
|
return formatted === text; |
|
}, |
|
doc: doc, |
|
getSupportInfo: withPlugins(getSupportInfo$2), |
|
version: version$2, |
|
util: utilShared, |
|
__debug: { |
|
parse: withPlugins(core.parse), |
|
formatAST: withPlugins(core.formatAST), |
|
formatDoc: withPlugins(core.formatDoc), |
|
printToDoc: withPlugins(core.printToDoc), |
|
printDocToString: withPlugins(core.printDocToString) |
|
} |
|
}; |
|
|
|
var standalone$1 = standalone; |
|
|
|
return standalone$1; |
|
|
|
})));
|