/*!
* XRegExp Unicode Base 4.0.0
* <xregexp.com>
* Steven Levithan (c) 2008-2017 MIT License
*/
export default (XRegExp) => {
/**
* Adds base support for Unicode matching:
* - Adds syntax `\p{..}` for matching Unicode tokens. Tokens can be inverted using `\P{..}` or
* `\p{^..}`. Token names ignore case, spaces, hyphens, and underscores. You can omit the
* braces for token names that are a single letter (e.g. `\pL` or `PL`).
* - Adds flag A (astral), which enables 21-bit Unicode support.
* - Adds the `XRegExp.addUnicodeData` method used by other addons to provide character data.
*
* Unicode Base relies on externally provided Unicode character data. Official addons are
* available to provide data for Unicode categories, scripts, blocks, and properties.
*
* @requires XRegExp
*/
// ==--------------------------==
// Private stuff
// ==--------------------------==
// Storage for Unicode data
const unicode = {};
// Reuse utils
const dec = XRegExp._dec;
const hex = XRegExp._hex;
const pad4 = XRegExp._pad4;
// Generates a token lookup name: lowercase, with hyphens, spaces, and underscores removed
function normalize(name) {
return name.replace(/[- _]+/g, '').toLowerCase();
}
// Gets the decimal code of a literal code unit, \xHH, \uHHHH, or a backslash-escaped literal
function charCode(chr) {
const esc = /^\\[xu](.+)/.exec(chr);
return esc ?
dec(esc[1]) :
chr.charCodeAt(chr[0] === '\\' ? 1 : 0);
}
// Inverts a list of ordered BMP characters and ranges
function invertBmp(range) {
let output = '';
let lastEnd = -1;
XRegExp.forEach(
range,
/(\\x..|\\u....|\\?[\s\S])(?:-(\\x..|\\u....|\\?[\s\S]))?/,
(m) => {
const start = charCode(m[1]);
if (start > (lastEnd + 1)) {
output += `\\u${pad4(hex(lastEnd + 1))}`;
if (start > (lastEnd + 2)) {
output += `-\\u${pad4(hex(start - 1))}`;
}
}
lastEnd = charCode(m[2] || m[1]);
}
);
if (lastEnd < 0xFFFF) {
output += `\\u${pad4(hex(lastEnd + 1))}`;
if (lastEnd < 0xFFFE) {
output += '-\\uFFFF';
}
}
return output;
}
// Generates an inverted BMP range on first use
function cacheInvertedBmp(slug) {
const prop = 'b!';
return (
unicode[slug][prop] ||
(unicode[slug][prop] = invertBmp(unicode[slug].bmp))
);
}
// Combines and optionally negates BMP and astral data
function buildAstral(slug, isNegated) {
const item = unicode[slug];
let combined = '';
if (item.bmp && !item.isBmpLast) {
combined = `[${item.bmp}]${item.astral ? '|' : ''}`;
}
if (item.astral) {
combined += item.astral;
}
if (item.isBmpLast && item.bmp) {
combined += `${item.astral ? '|' : ''}[${item.bmp}]`;
}
// Astral Unicode tokens always match a code point, never a code unit
return isNegated ?
`(?:(?!${combined})(?:[\uD800-\uDBFF][\uDC00-\uDFFF]|[\0-\uFFFF]))` :
`(?:${combined})`;
}
// Builds a complete astral pattern on first use
function cacheAstral(slug, isNegated) {
const prop = isNegated ? 'a!' : 'a=';
return (
unicode[slug][prop] ||
(unicode[slug][prop] = buildAstral(slug, isNegated))
);
}
// ==--------------------------==
// Core functionality
// ==--------------------------==
/*
* Add astral mode (flag A) and Unicode token syntax: `\p{..}`, `\P{..}`, `\p{^..}`, `\pC`.
*/
XRegExp.addToken(
// Use `*` instead of `+` to avoid capturing `^` as the token name in `\p{^}`
/\\([pP])(?:{(\^?)([^}]*)}|([A-Za-z]))/,
(match, scope, flags) => {
const ERR_DOUBLE_NEG = 'Invalid double negation ';
const ERR_UNKNOWN_NAME = 'Unknown Unicode token ';
const ERR_UNKNOWN_REF = 'Unicode token missing data ';
const ERR_ASTRAL_ONLY = 'Astral mode required for Unicode token ';
const ERR_ASTRAL_IN_CLASS = 'Astral mode does not support Unicode tokens within character classes';
// Negated via \P{..} or \p{^..}
let isNegated = match[1] === 'P' || !!match[2];
// Switch from BMP (0-FFFF) to astral (0-10FFFF) mode via flag A
const isAstralMode = flags.includes('A');
// Token lookup name. Check `[4]` first to avoid passing `undefined` via `\p{}`
let slug = normalize(match[4] || match[3]);
// Token data object
let item = unicode[slug];
if (match[1] === 'P' && match[2]) {
throw new SyntaxError(ERR_DOUBLE_NEG + match[0]);
}
if (!unicode.hasOwnProperty(slug)) {
throw new SyntaxError(ERR_UNKNOWN_NAME + match[0]);
}
// Switch to the negated form of the referenced Unicode token
if (item.inverseOf) {
slug = normalize(item.inverseOf);
if (!unicode.hasOwnProperty(slug)) {
throw new ReferenceError(`${ERR_UNKNOWN_REF + match[0]} -> ${item.inverseOf}`);
}
item = unicode[slug];
isNegated = !isNegated;
}
if (!(item.bmp || isAstralMode)) {
throw new SyntaxError(ERR_ASTRAL_ONLY + match[0]);
}
if (isAstralMode) {
if (scope === 'class') {
throw new SyntaxError(ERR_ASTRAL_IN_CLASS);
}
return cacheAstral(slug, isNegated);
}
return scope === 'class' ?
(isNegated ? cacheInvertedBmp(slug) : item.bmp) :
`${(isNegated ? '[^' : '[') + item.bmp}]`;
},
{
scope: 'all',
optionalFlags: 'A',
leadChar: '\\'
}
);
/**
* Adds to the list of Unicode tokens that XRegExp regexes can match via `\p` or `\P`.
*
* @memberOf XRegExp
* @param {Array} data Objects with named character ranges. Each object may have properties
* `name`, `alias`, `isBmpLast`, `inverseOf`, `bmp`, and `astral`. All but `name` are
* optional, although one of `bmp` or `astral` is required (unless `inverseOf` is set). If
* `astral` is absent, the `bmp` data is used for BMP and astral modes. If `bmp` is absent,
* the name errors in BMP mode but works in astral mode. If both `bmp` and `astral` are
* provided, the `bmp` data only is used in BMP mode, and the combination of `bmp` and
* `astral` data is used in astral mode. `isBmpLast` is needed when a token matches orphan
* high surrogates *and* uses surrogate pairs to match astral code points. The `bmp` and
* `astral` data should be a combination of literal characters and `\xHH` or `\uHHHH` escape
* sequences, with hyphens to create ranges. Any regex metacharacters in the data should be
* escaped, apart from range-creating hyphens. The `astral` data can additionally use
* character classes and alternation, and should use surrogate pairs to represent astral code
* points. `inverseOf` can be used to avoid duplicating character data if a Unicode token is
* defined as the exact inverse of another token.
* @example
*
* // Basic use
* XRegExp.addUnicodeData([{
* name: 'XDigit',
* alias: 'Hexadecimal',
* bmp: '0-9A-Fa-f'
* }]);
* XRegExp('\\p{XDigit}:\\p{Hexadecimal}+').test('0:3D'); // -> true
*/
XRegExp.addUnicodeData = (data) => {
const ERR_NO_NAME = 'Unicode token requires name';
const ERR_NO_DATA = 'Unicode token has no character data ';
let item;
for (let i = 0; i < data.length; ++i) {
item = data[i];
if (!item.name) {
throw new Error(ERR_NO_NAME);
}
if (!(item.inverseOf || item.bmp || item.astral)) {
throw new Error(ERR_NO_DATA + item.name);
}
unicode[normalize(item.name)] = item;
if (item.alias) {
unicode[normalize(item.alias)] = item;
}
}
// Reset the pattern cache used by the `XRegExp` constructor, since the same pattern and
// flags might now produce different results
XRegExp.cache.flush('patterns');
};
/**
* @ignore
*
* Return a reference to the internal Unicode definition structure for the given Unicode
* Property if the given name is a legal Unicode Property for use in XRegExp `\p` or `\P` regex
* constructs.
*
* @memberOf XRegExp
* @param {String} name Name by which the Unicode Property may be recognized (case-insensitive),
* e.g. `'N'` or `'Number'`. The given name is matched against all registered Unicode
* Properties and Property Aliases.
* @returns {Object} Reference to definition structure when the name matches a Unicode Property.
*
* @note
* For more info on Unicode Properties, see also http://unicode.org/reports/tr18/#Categories.
*
* @note
* This method is *not* part of the officially documented API and may change or be removed in
* the future. It is meant for userland code that wishes to reuse the (large) internal Unicode
* structures set up by XRegExp.
*/
XRegExp._getUnicodeProperty = (name) => {
const slug = normalize(name);
return unicode[slug];
};
};