index.js
 1  "use strict"
 2  
 3  // Based on: https://github.com/lodash/lodash/blob/6018350ac10d5ce6a5b7db625140b82aeab804df/.internal/unicodeSize.js
 4  
 5  module.exports = () => {
 6  	// Used to compose unicode character classes.
 7  	const astralRange = "\\ud800-\\udfff"
 8  	const comboMarksRange = "\\u0300-\\u036f"
 9  	const comboHalfMarksRange = "\\ufe20-\\ufe2f"
10  	const comboSymbolsRange = "\\u20d0-\\u20ff"
11  	const comboMarksExtendedRange = "\\u1ab0-\\u1aff"
12  	const comboMarksSupplementRange = "\\u1dc0-\\u1dff"
13  	const comboRange = comboMarksRange + comboHalfMarksRange + comboSymbolsRange + comboMarksExtendedRange + comboMarksSupplementRange
14  	const varRange = "\\ufe0e\\ufe0f"
15  	const familyRange = "\\uD83D\\uDC69\\uD83C\\uDFFB\\u200D\\uD83C\\uDF93"
16  
17  	// Used to compose unicode capture groups.
18  	const astral = `[${astralRange}]`
19  	const combo = `[${comboRange}]`
20  	const fitz = "\\ud83c[\\udffb-\\udfff]"
21  	const modifier = `(?:${combo}|${fitz})`
22  	const nonAstral = `[^${astralRange}]`
23  	const regional = "(?:\\uD83C[\\uDDE6-\\uDDFF]){2}"
24  	const surrogatePair = "[\\ud800-\\udbff][\\udc00-\\udfff]"
25  	const zwj = "\\u200d"
26  	const blackFlag = "(?:\\ud83c\\udff4\\udb40\\udc67\\udb40\\udc62\\udb40(?:\\udc65|\\udc73|\\udc77)\\udb40(?:\\udc6e|\\udc63|\\udc6c)\\udb40(?:\\udc67|\\udc74|\\udc73)\\udb40\\udc7f)"
27  	const family = `[${familyRange}]`
28  
29  	// Used to compose unicode regexes.
30  	const optModifier = `${modifier}?`
31  	const optVar = `[${varRange}]?`
32  	const optJoin = `(?:${zwj}(?:${[nonAstral, regional, surrogatePair].join("|")})${optVar + optModifier})*`
33  	const seq = optVar + optModifier + optJoin
34  	const nonAstralCombo = `${nonAstral}${combo}?`
35  	const symbol = `(?:${[nonAstralCombo, combo, regional, surrogatePair, astral, family].join("|")})`
36  
37  	// Used to match [String symbols](https://mathiasbynens.be/notes/javascript-unicode).
38  	return new RegExp(`${blackFlag}|${fitz}(?=${fitz})|${symbol + seq}`, "g")
39  }