1 | /*!
|
---|
2 | * regjsgen 0.8.0
|
---|
3 | * Copyright 2014-2023 Benjamin Tan <https://ofcr.se/>
|
---|
4 | * Available under the MIT license <https://github.com/bnjmnt4n/regjsgen/blob/main/LICENSE-MIT.txt>
|
---|
5 | */
|
---|
6 | ;(function() {
|
---|
7 | 'use strict';
|
---|
8 |
|
---|
9 | // Used to determine if values are of the language type `Object`.
|
---|
10 | var objectTypes = {
|
---|
11 | 'function': true,
|
---|
12 | 'object': true
|
---|
13 | };
|
---|
14 |
|
---|
15 | // Used as a reference to the global object.
|
---|
16 | var root = (objectTypes[typeof window] && window) || this;
|
---|
17 |
|
---|
18 | // Detect free variable `exports`.
|
---|
19 | var freeExports = objectTypes[typeof exports] && exports && !exports.nodeType && exports;
|
---|
20 |
|
---|
21 | // Detect free variable `module`.
|
---|
22 | var hasFreeModule = objectTypes[typeof module] && module && !module.nodeType;
|
---|
23 |
|
---|
24 | // Detect free variable `global` from Node.js or Browserified code and use it as `root`.
|
---|
25 | var freeGlobal = freeExports && hasFreeModule && typeof global == 'object' && global;
|
---|
26 | if (freeGlobal && (freeGlobal.global === freeGlobal || freeGlobal.window === freeGlobal || freeGlobal.self === freeGlobal)) {
|
---|
27 | root = freeGlobal;
|
---|
28 | }
|
---|
29 |
|
---|
30 | // Used to check objects for own properties.
|
---|
31 | var hasOwnProperty = Object.prototype.hasOwnProperty;
|
---|
32 |
|
---|
33 | /*--------------------------------------------------------------------------*/
|
---|
34 |
|
---|
35 | // Generates a string based on the given code point.
|
---|
36 | // Based on https://mths.be/fromcodepoint by @mathias.
|
---|
37 | function fromCodePoint() {
|
---|
38 | var codePoint = Number(arguments[0]);
|
---|
39 |
|
---|
40 | if (
|
---|
41 | !isFinite(codePoint) || // `NaN`, `+Infinity`, or `-Infinity`
|
---|
42 | codePoint < 0 || // not a valid Unicode code point
|
---|
43 | codePoint > 0x10FFFF || // not a valid Unicode code point
|
---|
44 | Math.floor(codePoint) != codePoint // not an integer
|
---|
45 | ) {
|
---|
46 | throw RangeError('Invalid code point: ' + codePoint);
|
---|
47 | }
|
---|
48 |
|
---|
49 | if (codePoint <= 0xFFFF) {
|
---|
50 | // BMP code point
|
---|
51 | return String.fromCharCode(codePoint);
|
---|
52 | } else {
|
---|
53 | // Astral code point; split in surrogate halves
|
---|
54 | // http://mathiasbynens.be/notes/javascript-encoding#surrogate-formulae
|
---|
55 | codePoint -= 0x10000;
|
---|
56 | var highSurrogate = (codePoint >> 10) + 0xD800;
|
---|
57 | var lowSurrogate = (codePoint % 0x400) + 0xDC00;
|
---|
58 | return String.fromCharCode(highSurrogate, lowSurrogate);
|
---|
59 | }
|
---|
60 | }
|
---|
61 |
|
---|
62 | /*--------------------------------------------------------------------------*/
|
---|
63 |
|
---|
64 | // Ensures that nodes have the correct types.
|
---|
65 | var assertTypeRegexMap = {};
|
---|
66 | function assertType(type, expected) {
|
---|
67 | if (expected.indexOf('|') == -1) {
|
---|
68 | if (type == expected) {
|
---|
69 | return;
|
---|
70 | }
|
---|
71 |
|
---|
72 | throw Error('Invalid node type: ' + type + '; expected type: ' + expected);
|
---|
73 | }
|
---|
74 |
|
---|
75 | expected = hasOwnProperty.call(assertTypeRegexMap, expected)
|
---|
76 | ? assertTypeRegexMap[expected]
|
---|
77 | : (assertTypeRegexMap[expected] = RegExp('^(?:' + expected + ')$'));
|
---|
78 |
|
---|
79 | if (expected.test(type)) {
|
---|
80 | return;
|
---|
81 | }
|
---|
82 |
|
---|
83 | throw Error('Invalid node type: ' + type + '; expected types: ' + expected);
|
---|
84 | }
|
---|
85 |
|
---|
86 | /*--------------------------------------------------------------------------*/
|
---|
87 |
|
---|
88 | // Generates a regular expression string based on an AST.
|
---|
89 | function generate(node) {
|
---|
90 | var type = node.type;
|
---|
91 |
|
---|
92 | if (hasOwnProperty.call(generators, type)) {
|
---|
93 | return generators[type](node);
|
---|
94 | }
|
---|
95 |
|
---|
96 | throw Error('Invalid node type: ' + type);
|
---|
97 | }
|
---|
98 |
|
---|
99 | // Constructs a string by concatentating the output of each term.
|
---|
100 | function generateSequence(generator, terms, /* optional */ separator) {
|
---|
101 | var i = -1,
|
---|
102 | length = terms.length,
|
---|
103 | result = '',
|
---|
104 | term;
|
---|
105 |
|
---|
106 | while (++i < length) {
|
---|
107 | term = terms[i];
|
---|
108 |
|
---|
109 | if (separator && i > 0) result += separator;
|
---|
110 |
|
---|
111 | // Ensure that `\0` null escapes followed by number symbols are not
|
---|
112 | // treated as backreferences.
|
---|
113 | if (
|
---|
114 | i + 1 < length &&
|
---|
115 | terms[i].type == 'value' &&
|
---|
116 | terms[i].kind == 'null' &&
|
---|
117 | terms[i + 1].type == 'value' &&
|
---|
118 | terms[i + 1].kind == 'symbol' &&
|
---|
119 | terms[i + 1].codePoint >= 48 &&
|
---|
120 | terms[i + 1].codePoint <= 57
|
---|
121 | ) {
|
---|
122 | result += '\\000';
|
---|
123 | continue;
|
---|
124 | }
|
---|
125 |
|
---|
126 | result += generator(term);
|
---|
127 | }
|
---|
128 |
|
---|
129 | return result;
|
---|
130 | }
|
---|
131 |
|
---|
132 | /*--------------------------------------------------------------------------*/
|
---|
133 |
|
---|
134 | function generateAlternative(node) {
|
---|
135 | assertType(node.type, 'alternative');
|
---|
136 |
|
---|
137 | return generateSequence(generateTerm, node.body);
|
---|
138 | }
|
---|
139 |
|
---|
140 | function generateAnchor(node) {
|
---|
141 | assertType(node.type, 'anchor');
|
---|
142 |
|
---|
143 | switch (node.kind) {
|
---|
144 | case 'start':
|
---|
145 | return '^';
|
---|
146 | case 'end':
|
---|
147 | return '$';
|
---|
148 | case 'boundary':
|
---|
149 | return '\\b';
|
---|
150 | case 'not-boundary':
|
---|
151 | return '\\B';
|
---|
152 | default:
|
---|
153 | throw Error('Invalid assertion');
|
---|
154 | }
|
---|
155 | }
|
---|
156 |
|
---|
157 | var atomType = 'anchor|characterClass|characterClassEscape|dot|group|reference|unicodePropertyEscape|value';
|
---|
158 |
|
---|
159 | function generateAtom(node) {
|
---|
160 | assertType(node.type, atomType);
|
---|
161 |
|
---|
162 | return generate(node);
|
---|
163 | }
|
---|
164 |
|
---|
165 | function generateCharacterClass(node) {
|
---|
166 | assertType(node.type, 'characterClass');
|
---|
167 |
|
---|
168 | var kind = node.kind;
|
---|
169 | var separator = kind === 'intersection' ? '&&' : kind === 'subtraction' ? '--' : '';
|
---|
170 |
|
---|
171 | return '[' +
|
---|
172 | (node.negative ? '^' : '') +
|
---|
173 | generateSequence(generateClassAtom, node.body, separator) +
|
---|
174 | ']';
|
---|
175 | }
|
---|
176 |
|
---|
177 | function generateCharacterClassEscape(node) {
|
---|
178 | assertType(node.type, 'characterClassEscape');
|
---|
179 |
|
---|
180 | return '\\' + node.value;
|
---|
181 | }
|
---|
182 |
|
---|
183 | function generateCharacterClassRange(node) {
|
---|
184 | assertType(node.type, 'characterClassRange');
|
---|
185 |
|
---|
186 | var min = node.min,
|
---|
187 | max = node.max;
|
---|
188 |
|
---|
189 | if (min.type == 'characterClassRange' || max.type == 'characterClassRange') {
|
---|
190 | throw Error('Invalid character class range');
|
---|
191 | }
|
---|
192 |
|
---|
193 | return generateClassAtom(min) + '-' + generateClassAtom(max);
|
---|
194 | }
|
---|
195 |
|
---|
196 | function generateClassAtom(node) {
|
---|
197 | assertType(node.type, 'anchor|characterClass|characterClassEscape|characterClassRange|dot|value|unicodePropertyEscape|classStrings');
|
---|
198 |
|
---|
199 | return generate(node);
|
---|
200 | }
|
---|
201 |
|
---|
202 | function generateClassStrings(node) {
|
---|
203 | assertType(node.type, 'classStrings');
|
---|
204 |
|
---|
205 | return '\\q{' + generateSequence(generateClassString, node.strings, '|') + '}';
|
---|
206 | }
|
---|
207 |
|
---|
208 | function generateClassString(node) {
|
---|
209 | assertType(node.type, 'classString');
|
---|
210 |
|
---|
211 | return generateSequence(generate, node.characters);
|
---|
212 | }
|
---|
213 |
|
---|
214 | function generateDisjunction(node) {
|
---|
215 | assertType(node.type, 'disjunction');
|
---|
216 |
|
---|
217 | return generateSequence(generate, node.body, '|');
|
---|
218 | }
|
---|
219 |
|
---|
220 |
|
---|
221 | function generateDot(node) {
|
---|
222 | assertType(node.type, 'dot');
|
---|
223 |
|
---|
224 | return '.';
|
---|
225 | }
|
---|
226 |
|
---|
227 | function generateGroup(node) {
|
---|
228 | assertType(node.type, 'group');
|
---|
229 |
|
---|
230 | var result = '';
|
---|
231 |
|
---|
232 | switch (node.behavior) {
|
---|
233 | case 'normal':
|
---|
234 | if (node.name) {
|
---|
235 | result += '?<' + generateIdentifier(node.name) + '>';
|
---|
236 | }
|
---|
237 | break;
|
---|
238 | case 'ignore':
|
---|
239 | if (node.modifierFlags) {
|
---|
240 | result += '?';
|
---|
241 | if (node.modifierFlags.enabling) result += node.modifierFlags.enabling;
|
---|
242 | if (node.modifierFlags.disabling) result += "-" + node.modifierFlags.disabling;
|
---|
243 | result += ':';
|
---|
244 | } else {
|
---|
245 | result += '?:';
|
---|
246 | }
|
---|
247 | break;
|
---|
248 | case 'lookahead':
|
---|
249 | result += '?=';
|
---|
250 | break;
|
---|
251 | case 'negativeLookahead':
|
---|
252 | result += '?!';
|
---|
253 | break;
|
---|
254 | case 'lookbehind':
|
---|
255 | result += '?<=';
|
---|
256 | break;
|
---|
257 | case 'negativeLookbehind':
|
---|
258 | result += '?<!';
|
---|
259 | break;
|
---|
260 | default:
|
---|
261 | throw Error('Invalid behaviour: ' + node.behaviour);
|
---|
262 | }
|
---|
263 |
|
---|
264 | result += generateSequence(generate, node.body);
|
---|
265 |
|
---|
266 | return '(' + result + ')';
|
---|
267 | }
|
---|
268 |
|
---|
269 | function generateIdentifier(node) {
|
---|
270 | assertType(node.type, 'identifier');
|
---|
271 |
|
---|
272 | return node.value;
|
---|
273 | }
|
---|
274 |
|
---|
275 | function generateQuantifier(node) {
|
---|
276 | assertType(node.type, 'quantifier');
|
---|
277 |
|
---|
278 | var quantifier = '',
|
---|
279 | min = node.min,
|
---|
280 | max = node.max;
|
---|
281 |
|
---|
282 | if (max == null) {
|
---|
283 | if (min == 0) {
|
---|
284 | quantifier = '*';
|
---|
285 | } else if (min == 1) {
|
---|
286 | quantifier = '+';
|
---|
287 | } else {
|
---|
288 | quantifier = '{' + min + ',}';
|
---|
289 | }
|
---|
290 | } else if (min == max) {
|
---|
291 | quantifier = '{' + min + '}';
|
---|
292 | } else if (min == 0 && max == 1) {
|
---|
293 | quantifier = '?';
|
---|
294 | } else {
|
---|
295 | quantifier = '{' + min + ',' + max + '}';
|
---|
296 | }
|
---|
297 |
|
---|
298 | if (!node.greedy) {
|
---|
299 | quantifier += '?';
|
---|
300 | }
|
---|
301 |
|
---|
302 | return generateAtom(node.body[0]) + quantifier;
|
---|
303 | }
|
---|
304 |
|
---|
305 | function generateReference(node) {
|
---|
306 | assertType(node.type, 'reference');
|
---|
307 |
|
---|
308 | if (node.matchIndex) {
|
---|
309 | return '\\' + node.matchIndex;
|
---|
310 | }
|
---|
311 | if (node.name) {
|
---|
312 | return '\\k<' + generateIdentifier(node.name) + '>';
|
---|
313 | }
|
---|
314 |
|
---|
315 | throw new Error('Unknown reference type');
|
---|
316 | }
|
---|
317 |
|
---|
318 | function generateTerm(node) {
|
---|
319 | assertType(node.type, atomType + '|empty|quantifier');
|
---|
320 |
|
---|
321 | return generate(node);
|
---|
322 | }
|
---|
323 |
|
---|
324 | function generateUnicodePropertyEscape(node) {
|
---|
325 | assertType(node.type, 'unicodePropertyEscape');
|
---|
326 |
|
---|
327 | return '\\' + (node.negative ? 'P' : 'p') + '{' + node.value + '}';
|
---|
328 | }
|
---|
329 |
|
---|
330 | function generateValue(node) {
|
---|
331 | assertType(node.type, 'value');
|
---|
332 |
|
---|
333 | var kind = node.kind,
|
---|
334 | codePoint = node.codePoint;
|
---|
335 |
|
---|
336 | if (typeof codePoint != 'number') {
|
---|
337 | throw new Error('Invalid code point: ' + codePoint);
|
---|
338 | }
|
---|
339 |
|
---|
340 | switch (kind) {
|
---|
341 | case 'controlLetter':
|
---|
342 | return '\\c' + fromCodePoint(codePoint + 64);
|
---|
343 | case 'hexadecimalEscape':
|
---|
344 | return '\\x' + ('00' + codePoint.toString(16).toUpperCase()).slice(-2);
|
---|
345 | case 'identifier':
|
---|
346 | return '\\' + fromCodePoint(codePoint);
|
---|
347 | case 'null':
|
---|
348 | return '\\' + codePoint;
|
---|
349 | case 'octal':
|
---|
350 | return '\\' + ('000' + codePoint.toString(8)).slice(-3);
|
---|
351 | case 'singleEscape':
|
---|
352 | switch (codePoint) {
|
---|
353 | case 0x0008:
|
---|
354 | return '\\b';
|
---|
355 | case 0x0009:
|
---|
356 | return '\\t';
|
---|
357 | case 0x000A:
|
---|
358 | return '\\n';
|
---|
359 | case 0x000B:
|
---|
360 | return '\\v';
|
---|
361 | case 0x000C:
|
---|
362 | return '\\f';
|
---|
363 | case 0x000D:
|
---|
364 | return '\\r';
|
---|
365 | case 0x002D:
|
---|
366 | return '\\-';
|
---|
367 | default:
|
---|
368 | throw Error('Invalid code point: ' + codePoint);
|
---|
369 | }
|
---|
370 | case 'symbol':
|
---|
371 | return fromCodePoint(codePoint);
|
---|
372 | case 'unicodeEscape':
|
---|
373 | return '\\u' + ('0000' + codePoint.toString(16).toUpperCase()).slice(-4);
|
---|
374 | case 'unicodeCodePointEscape':
|
---|
375 | return '\\u{' + codePoint.toString(16).toUpperCase() + '}';
|
---|
376 | default:
|
---|
377 | throw Error('Unsupported node kind: ' + kind);
|
---|
378 | }
|
---|
379 | }
|
---|
380 |
|
---|
381 | /*--------------------------------------------------------------------------*/
|
---|
382 |
|
---|
383 | // Used to generate strings for each node type.
|
---|
384 | var generators = {
|
---|
385 | 'alternative': generateAlternative,
|
---|
386 | 'anchor': generateAnchor,
|
---|
387 | 'characterClass': generateCharacterClass,
|
---|
388 | 'characterClassEscape': generateCharacterClassEscape,
|
---|
389 | 'characterClassRange': generateCharacterClassRange,
|
---|
390 | 'classStrings': generateClassStrings,
|
---|
391 | 'disjunction': generateDisjunction,
|
---|
392 | 'dot': generateDot,
|
---|
393 | 'group': generateGroup,
|
---|
394 | 'quantifier': generateQuantifier,
|
---|
395 | 'reference': generateReference,
|
---|
396 | 'unicodePropertyEscape': generateUnicodePropertyEscape,
|
---|
397 | 'value': generateValue
|
---|
398 | };
|
---|
399 |
|
---|
400 | /*--------------------------------------------------------------------------*/
|
---|
401 |
|
---|
402 | // Export regjsgen.
|
---|
403 | var regjsgen = {
|
---|
404 | 'generate': generate
|
---|
405 | };
|
---|
406 |
|
---|
407 | // Some AMD build optimizers, like r.js, check for condition patterns like the following:
|
---|
408 | if (typeof define == 'function' && typeof define.amd == 'object' && define.amd) {
|
---|
409 | // Define as an anonymous module so it can be aliased through path mapping.
|
---|
410 | define(function() {
|
---|
411 | return regjsgen;
|
---|
412 | });
|
---|
413 |
|
---|
414 | root.regjsgen = regjsgen;
|
---|
415 | }
|
---|
416 | // Check for `exports` after `define` in case a build optimizer adds an `exports` object.
|
---|
417 | else if (freeExports && hasFreeModule) {
|
---|
418 | // Export for CommonJS support.
|
---|
419 | freeExports.generate = generate;
|
---|
420 | }
|
---|
421 | else {
|
---|
422 | // Export to the global object.
|
---|
423 | root.regjsgen = regjsgen;
|
---|
424 | }
|
---|
425 | }.call(this));
|
---|