unicode-base.js

/*!
 * XRegExp Unicode Base 4.4.1
 * <xregexp.com>
 * Steven Levithan (c) 2008-present 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 = {};
    const unicodeTypes = {};

    // 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])(?:{(\^?)(?:(\w+)=)?([^}]*)}|([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';
            const [
                fullToken,
                pPrefix,
                caretNegation,
                typePrefix,
                tokenName,
                tokenSingleCharName
            ] = match;
            // Negated via \P{..} or \p{^..}
            let isNegated = pPrefix === 'P' || !!caretNegation;
            // Switch from BMP (0-FFFF) to astral (0-10FFFF) mode via flag A
            const isAstralMode = flags.includes('A');
            // Token lookup name. Check `tokenSingleCharName` first to avoid passing `undefined`
            // via `\p{}`
            let slug = normalize(tokenSingleCharName || tokenName);
            // Token data object
            let item = unicode[slug];

            if (pPrefix === 'P' && caretNegation) {
                throw new SyntaxError(ERR_DOUBLE_NEG + fullToken);
            }
            if (!unicode.hasOwnProperty(slug)) {
                throw new SyntaxError(ERR_UNKNOWN_NAME + fullToken);
            }

            if (typePrefix) {
                if (!(unicodeTypes[typePrefix] && unicodeTypes[typePrefix][slug])) {
                    throw new SyntaxError(ERR_UNKNOWN_NAME + fullToken);
                }
            }

            // 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 + fullToken} -> ${item.inverseOf}`);
                }
                item = unicode[slug];
                isNegated = !isNegated;
            }

            if (!(item.bmp || isAstralMode)) {
                throw new SyntaxError(ERR_ASTRAL_ONLY + fullToken);
            }
            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.
     * @param {String} [typePrefix] Enables optionally using this type as a prefix for all of the
     *   provided Unicode tokens, e.g. if given `'Type'`, then `\p{TokenName}` can also be written
     *   as `\p{Type=TokenName}`.
     * @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, typePrefix) => {
        const ERR_NO_NAME = 'Unicode token requires name';
        const ERR_NO_DATA = 'Unicode token has no character data ';

        if (typePrefix) {
            // Case sensitive to match ES2018
            unicodeTypes[typePrefix] = {};
        }

        for (const item of data) {
            if (!item.name) {
                throw new Error(ERR_NO_NAME);
            }
            if (!(item.inverseOf || item.bmp || item.astral)) {
                throw new Error(ERR_NO_DATA + item.name);
            }

            const normalizedName = normalize(item.name);
            unicode[normalizedName] = item;
            if (typePrefix) {
                unicodeTypes[typePrefix][normalizedName] = true;
            }

            if (item.alias) {
                const normalizedAlias = normalize(item.alias);
                unicode[normalizedAlias] = item;
                if (typePrefix) {
                    unicodeTypes[typePrefix][normalizedAlias] = true;
                }
            }
        }

        // 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];
    };
};