Brunobkr/llama.cpp_AlgMor24_github

1// translate the various posix character classes into unicode properties2// this works across all unicode locales3// { <posix class>: [<translation>, /u flag required, negated]4const posixClasses = {5    '[:alnum:]': ['\\p{L}\\p{Nl}\\p{Nd}', true],6    '[:alpha:]': ['\\p{L}\\p{Nl}', true],7    '[:ascii:]': ['\\x' + '00-\\x' + '7f', false],8    '[:blank:]': ['\\p{Zs}\\t', true],9    '[:cntrl:]': ['\\p{Cc}', true],10    '[:digit:]': ['\\p{Nd}', true],11    '[:graph:]': ['\\p{Z}\\p{C}', true, true],12    '[:lower:]': ['\\p{Ll}', true],13    '[:print:]': ['\\p{C}', true],14    '[:punct:]': ['\\p{P}', true],15    '[:space:]': ['\\p{Z}\\t\\r\\n\\v\\f', true],16    '[:upper:]': ['\\p{Lu}', true],17    '[:word:]': ['\\p{L}\\p{Nl}\\p{Nd}\\p{Pc}', true],18    '[:xdigit:]': ['A-Fa-f0-9', false],19};20// only need to escape a few things inside of brace expressions21// escapes: [ \ ] -22const braceEscape = (s) => s.replace(/[[\]\\-]/g, '\\$&');23// escape all regexp magic characters24const regexpEscape = (s) => s.replace(/[-[\]{}()*+?.,\\^$|#\s]/g, '\\$&');25// everything has already been escaped, we just have to join26const rangesToString = (ranges) => ranges.join('');27// takes a glob string at a posix brace expression, and returns28// an equivalent regular expression source, and boolean indicating29// whether the /u flag needs to be applied, and the number of chars30// consumed to parse the character class.31// This also removes out of order ranges, and returns ($.) if the32// entire class just no good.33export const parseClass = (glob, position) => {34    const pos = position;35    /* c8 ignore start */36    if (glob.charAt(pos) !== '[') {37        throw new Error('not in a brace expression');38    }39    /* c8 ignore stop */40    const ranges = [];41    const negs = [];42    let i = pos + 1;43    let sawStart = false;44    let uflag = false;45    let escaping = false;46    let negate = false;47    let endPos = pos;48    let rangeStart = '';49    WHILE: while (i < glob.length) {50        const c = glob.charAt(i);51        if ((c === '!' || c === '^') && i === pos + 1) {52            negate = true;53            i++;54            continue;55        }56        if (c === ']' && sawStart && !escaping) {57            endPos = i + 1;58            break;59        }60        sawStart = true;61        if (c === '\\') {62            if (!escaping) {63                escaping = true;64                i++;65                continue;66            }67            // escaped \ char, fall through and treat like normal char68        }69        if (c === '[' && !escaping) {70            // either a posix class, a collation equivalent, or just a [71            for (const [cls, [unip, u, neg]] of Object.entries(posixClasses)) {72                if (glob.startsWith(cls, i)) {73                    // invalid, [a-[] is fine, but not [a-[:alpha]]74                    if (rangeStart) {75                        return ['$.', false, glob.length - pos, true];76                    }77                    i += cls.length;78                    if (neg)79                        negs.push(unip);80                    else81                        ranges.push(unip);82                    uflag = uflag || u;83                    continue WHILE;84                }85            }86        }87        // now it's just a normal character, effectively88        escaping = false;89        if (rangeStart) {90            // throw this range away if it's not valid, but others91            // can still match.92            if (c > rangeStart) {93                ranges.push(braceEscape(rangeStart) + '-' + braceEscape(c));94            }95            else if (c === rangeStart) {96                ranges.push(braceEscape(c));97            }98            rangeStart = '';99            i++;100            continue;101        }102        // now might be the start of a range.103        // can be either c-d or c-] or c<more...>] or c] at this point104        if (glob.startsWith('-]', i + 1)) {105            ranges.push(braceEscape(c + '-'));106            i += 2;107            continue;108        }109        if (glob.startsWith('-', i + 1)) {110            rangeStart = c;111            i += 2;112            continue;113        }114        // not the start of a range, just a single character115        ranges.push(braceEscape(c));116        i++;117    }118    if (endPos < i) {119        // didn't see the end of the class, not a valid class,120        // but might still be valid as a literal match.121        return ['', false, 0, false];122    }123    // if we got no ranges and no negates, then we have a range that124    // cannot possibly match anything, and that poisons the whole glob125    if (!ranges.length && !negs.length) {126        return ['$.', false, glob.length - pos, true];127    }128    // if we got one positive range, and it's a single character, then that's129    // not actually a magic pattern, it's just that one literal character.130    // we should not treat that as "magic", we should just return the literal131    // character. [_] is a perfectly valid way to escape glob magic chars.132    if (negs.length === 0 &&133        ranges.length === 1 &&134        /^\\?.$/.test(ranges[0]) &&135        !negate) {136        const r = ranges[0].length === 2 ? ranges[0].slice(-1) : ranges[0];137        return [regexpEscape(r), false, endPos - pos, false];138    }139    const sranges = '[' + (negate ? '^' : '') + rangesToString(ranges) + ']';140    const snegs = '[' + (negate ? '' : '^') + rangesToString(negs) + ']';141    const comb = ranges.length && negs.length ? '(' + sranges + '|' + snegs + ')'142        : ranges.length ? sranges143            : snegs;144    return [comb, uflag, endPos - pos, true];145};146//# sourceMappingURL=brace-expressions.js.map