mirror of
https://github.com/x1xhlol/system-prompts-and-models-of-ai-tools.git
synced 2025-09-15 20:37:24 +00:00
308 lines
7.5 KiB
JavaScript
308 lines
7.5 KiB
JavaScript
'use strict';
|
|
|
|
const constants = require('./constants');
|
|
const utils = require('./utils');
|
|
|
|
/**
|
|
* Constants
|
|
*/
|
|
|
|
const {
|
|
MAX_LENGTH,
|
|
POSIX_REGEX_SOURCE,
|
|
REGEX_NON_SPECIAL_CHARS,
|
|
REGEX_SPECIAL_CHARS_BACKREF,
|
|
REPLACEMENTS
|
|
} = constants;
|
|
|
|
/**
|
|
* Helpers
|
|
*/
|
|
|
|
const expandRange = (args, options) => {
|
|
if (typeof options.expandRange === 'function') {
|
|
return options.expandRange(...args, options);
|
|
}
|
|
|
|
args.sort();
|
|
const value = `[${args.join('-')}]`;
|
|
|
|
try {
|
|
/* eslint-disable-next-line no-new */
|
|
new RegExp(value);
|
|
} catch (ex) {
|
|
return args.map(v => utils.escapeRegex(v)).join('..');
|
|
}
|
|
|
|
return value;
|
|
};
|
|
|
|
/**
|
|
* Create the message for a syntax error
|
|
*/
|
|
|
|
const syntaxError = (type, char) => {
|
|
return `Missing ${type}: "${char}" - use "\\\\${char}" to match literal characters`;
|
|
};
|
|
|
|
/**
|
|
* Parse the given input string.
|
|
* @param {String} input
|
|
* @param {Object} options
|
|
* @return {Object}
|
|
*/
|
|
|
|
const parse = (input, options) => {
|
|
if (typeof input !== 'string') {
|
|
throw new TypeError('Expected a string');
|
|
}
|
|
|
|
input = REPLACEMENTS[input] || input;
|
|
|
|
const opts = { ...options };
|
|
const max = typeof opts.maxLength === 'number' ? Math.min(MAX_LENGTH, opts.maxLength) : MAX_LENGTH;
|
|
|
|
let len = input.length;
|
|
if (len > max) {
|
|
throw new SyntaxError(`Input length: ${len}, exceeds maximum allowed length: ${max}`);
|
|
}
|
|
|
|
const bos = { type: 'bos', value: '', output: opts.prepend || '' };
|
|
const tokens = [bos];
|
|
|
|
const capture = opts.capture ? '' : '?:';
|
|
const win32 = utils.isWindows(options);
|
|
|
|
// create constants based on platform, for windows or posix
|
|
const PLATFORM_CHARS = constants.globChars(win32);
|
|
const EXTGLOB_CHARS = constants.extglobChars(PLATFORM_CHARS);
|
|
|
|
const {
|
|
DOT_LITERAL,
|
|
PLUS_LITERAL,
|
|
SLASH_LITERAL,
|
|
ONE_CHAR,
|
|
DOTS_SLASH,
|
|
NO_DOT,
|
|
NO_DOT_SLASH,
|
|
NO_DOTS_SLASH,
|
|
QMARK,
|
|
QMARK_NO_DOT,
|
|
STAR,
|
|
START_ANCHOR
|
|
} = PLATFORM_CHARS;
|
|
|
|
const globstar = opts => {
|
|
return `(${capture}(?:(?!${START_ANCHOR}${opts.dot ? DOTS_SLASH : DOT_LITERAL}).)*?)`;
|
|
};
|
|
|
|
const nodot = opts.dot ? '' : NO_DOT;
|
|
const qmarkNoDot = opts.dot ? QMARK : QMARK_NO_DOT;
|
|
let star = opts.bash === true ? globstar(opts) : STAR;
|
|
|
|
if (opts.capture) {
|
|
star = `(${star})`;
|
|
}
|
|
|
|
// minimatch options support
|
|
if (typeof opts.noext === 'boolean') {
|
|
opts.noextglob = opts.noext;
|
|
}
|
|
|
|
const state = {
|
|
input,
|
|
index: -1,
|
|
start: 0,
|
|
dot: opts.dot === true,
|
|
consumed: '',
|
|
output: '',
|
|
prefix: '',
|
|
backtrack: false,
|
|
negated: false,
|
|
brackets: 0,
|
|
braces: 0,
|
|
parens: 0,
|
|
quotes: 0,
|
|
globstar: false,
|
|
tokens
|
|
};
|
|
|
|
input = utils.removePrefix(input, state);
|
|
len = input.length;
|
|
|
|
const extglobs = [];
|
|
const braces = [];
|
|
const stack = [];
|
|
let prev = bos;
|
|
let value;
|
|
|
|
/**
|
|
* Tokenizing helpers
|
|
*/
|
|
|
|
const eos = () => state.index === len - 1;
|
|
const peek = state.peek = (n = 1) => input[state.index + n];
|
|
const advance = state.advance = () => input[++state.index] || '';
|
|
const remaining = () => input.slice(state.index + 1);
|
|
const consume = (value = '', num = 0) => {
|
|
state.consumed += value;
|
|
state.index += num;
|
|
};
|
|
|
|
const append = token => {
|
|
state.output += token.output != null ? token.output : token.value;
|
|
consume(token.value);
|
|
};
|
|
|
|
const negate = () => {
|
|
let count = 1;
|
|
|
|
while (peek() === '!' && (peek(2) !== '(' || peek(3) === '?')) {
|
|
advance();
|
|
state.start++;
|
|
count++;
|
|
}
|
|
|
|
if (count % 2 === 0) {
|
|
return false;
|
|
}
|
|
|
|
state.negated = true;
|
|
state.start++;
|
|
return true;
|
|
};
|
|
|
|
const increment = type => {
|
|
state[type]++;
|
|
stack.push(type);
|
|
};
|
|
|
|
const decrement = type => {
|
|
state[type]--;
|
|
stack.pop();
|
|
};
|
|
|
|
/**
|
|
* Push tokens onto the tokens array. This helper speeds up
|
|
* tokenizing by 1) helping us avoid backtracking as much as possible,
|
|
* and 2) helping us avoid creating extra tokens when consecutive
|
|
* characters are plain text. This improves performance and simplifies
|
|
* lookbehinds.
|
|
*/
|
|
|
|
const push = tok => {
|
|
if (prev.type === 'globstar') {
|
|
const isBrace = state.braces > 0 && (tok.type === 'comma' || tok.type === 'brace');
|
|
const isExtglob = tok.extglob === true || (extglobs.length && (tok.type === 'pipe' || tok.type === 'paren'));
|
|
|
|
if (tok.type !== 'slash' && tok.type !== 'paren' && !isBrace && !isExtglob) {
|
|
state.output = state.output.slice(0, -prev.output.length);
|
|
prev.type = 'star';
|
|
prev.value = '*';
|
|
prev.output = star;
|
|
state.output += prev.output;
|
|
}
|
|
}
|
|
|
|
if (extglobs.length && tok.type !== 'paren') {
|
|
extglobs[extglobs.length - 1].inner += tok.value;
|
|
}
|
|
|
|
if (tok.value || tok.output) append(tok);
|
|
if (prev && prev.type === 'text' && tok.type === 'text') {
|
|
prev.value += tok.value;
|
|
prev.output = (prev.output || '') + tok.value;
|
|
return;
|
|
}
|
|
|
|
tok.prev = prev;
|
|
tokens.push(tok);
|
|
prev = tok;
|
|
};
|
|
|
|
const extglobOpen = (type, value) => {
|
|
const token = { ...EXTGLOB_CHARS[value], conditions: 1, inner: '' };
|
|
|
|
token.prev = prev;
|
|
token.parens = state.parens;
|
|
token.output = state.output;
|
|
const output = (opts.capture ? '(' : '') + token.open;
|
|
|
|
increment('parens');
|
|
push({ type, value, output: state.output ? '' : ONE_CHAR });
|
|
push({ type: 'paren', extglob: true, value: advance(), output });
|
|
extglobs.push(token);
|
|
};
|
|
|
|
const extglobClose = token => {
|
|
let output = token.close + (opts.capture ? ')' : '');
|
|
let rest;
|
|
|
|
if (token.type === 'negate') {
|
|
let extglobStar = star;
|
|
|
|
if (token.inner && token.inner.length > 1 && token.inner.includes('/')) {
|
|
extglobStar = globstar(opts);
|
|
}
|
|
|
|
if (extglobStar !== star || eos() || /^\)+$/.test(remaining())) {
|
|
output = token.close = `)$))${extglobStar}`;
|
|
}
|
|
|
|
if (token.inner.includes('*') && (rest = remaining()) && /^\.[^\\/.]+$/.test(rest)) {
|
|
// Any non-magical string (`.ts`) or even nested expression (`.{ts,tsx}`) can follow after the closing parenthesis.
|
|
// In this case, we need to parse the string and use it in the output of the original pattern.
|
|
// Suitable patterns: `/!(*.d).ts`, `/!(*.d).{ts,tsx}`, `**/!(*-dbg).@(js)`.
|
|
//
|
|
// Disabling the `fastpaths` option due to a problem with parsing strings as `.ts` in the pattern like `**/!(*.d).ts`.
|
|
const expression = parse(rest, { ...options, fastpaths: false }).output;
|
|
|
|
output = token.close = `)${expression})${extglobStar})`;
|
|
}
|
|
|
|
if (token.prev.type === 'bos') {
|
|
state.negatedExtglob = true;
|
|
}
|
|
}
|
|
|
|
push({ type: 'paren', extglob: true, value, output });
|
|
decrement('parens');
|
|
};
|
|
|
|
/**
|
|
* Fast paths
|
|
*/
|
|
|
|
if (opts.fastpaths !== false && !/(^[*!]|[/()[\]{}"])/.test(input)) {
|
|
let backslashes = false;
|
|
|
|
let output = input.replace(REGEX_SPECIAL_CHARS_BACKREF, (m, esc, chars, first, rest, index) => {
|
|
if (first === '\\') {
|
|
backslashes = true;
|
|
return m;
|
|
}
|
|
|
|
if (first === '?') {
|
|
if (esc) {
|
|
return esc + first + (rest ? QMARK.repeat(rest.length) : '');
|
|
}
|
|
if (index === 0) {
|
|
return qmarkNoDot + (rest ? QMARK.repeat(rest.length) : '');
|
|
}
|
|
return QMARK.repeat(chars.length);
|
|
}
|
|
|
|
if (first === '.') {
|
|
return DOT_LITERAL.repeat(chars.length);
|
|
}
|
|
|
|
if (first === '*') {
|
|
if (esc) {
|
|
return esc + first + (rest ? star : '');
|
|
}
|
|
return star;
|
|
}
|
|
return esc ? m : `\\${m}`;
|
|
});
|
|
|
|
if |