1 | /* |
2 | * Copyright (C) 2009, 2013-2017 Apple Inc. All rights reserved. |
3 | * Copyright (C) 2010 Peter Varga ([email protected]), University of Szeged |
4 | * |
5 | * Redistribution and use in source and binary forms, with or without |
6 | * modification, are permitted provided that the following conditions |
7 | * are met: |
8 | * 1. Redistributions of source code must retain the above copyright |
9 | * notice, this list of conditions and the following disclaimer. |
10 | * 2. Redistributions in binary form must reproduce the above copyright |
11 | * notice, this list of conditions and the following disclaimer in the |
12 | * documentation and/or other materials provided with the distribution. |
13 | * |
14 | * THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY |
15 | * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
16 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
17 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL APPLE INC. OR |
18 | * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, |
19 | * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
20 | * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR |
21 | * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY |
22 | * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
23 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
24 | * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
25 | */ |
26 | |
27 | #include "config.h" |
28 | #include "YarrInterpreter.h" |
29 | |
30 | #include "Options.h" |
31 | #include "SuperSampler.h" |
32 | #include "Yarr.h" |
33 | #include "YarrCanonicalize.h" |
34 | #include <wtf/BumpPointerAllocator.h> |
35 | #include <wtf/CheckedArithmetic.h> |
36 | #include <wtf/DataLog.h> |
37 | #include <wtf/text/CString.h> |
38 | #include <wtf/text/WTFString.h> |
39 | |
40 | namespace JSC { namespace Yarr { |
41 | |
42 | template<typename CharType> |
43 | class Interpreter { |
44 | public: |
45 | struct ParenthesesDisjunctionContext; |
46 | |
47 | struct BackTrackInfoParentheses { |
48 | uintptr_t matchAmount; |
49 | ParenthesesDisjunctionContext* lastContext; |
50 | }; |
51 | |
52 | static inline void appendParenthesesDisjunctionContext(BackTrackInfoParentheses* backTrack, ParenthesesDisjunctionContext* context) |
53 | { |
54 | context->next = backTrack->lastContext; |
55 | backTrack->lastContext = context; |
56 | ++backTrack->matchAmount; |
57 | } |
58 | |
59 | static inline void popParenthesesDisjunctionContext(BackTrackInfoParentheses* backTrack) |
60 | { |
61 | RELEASE_ASSERT(backTrack->matchAmount); |
62 | RELEASE_ASSERT(backTrack->lastContext); |
63 | backTrack->lastContext = backTrack->lastContext->next; |
64 | --backTrack->matchAmount; |
65 | } |
66 | |
67 | struct DisjunctionContext |
68 | { |
69 | DisjunctionContext() = default; |
70 | |
71 | void* operator new(size_t, void* where) |
72 | { |
73 | return where; |
74 | } |
75 | |
76 | static size_t allocationSize(unsigned numberOfFrames) |
77 | { |
78 | static_assert(alignof(DisjunctionContext) <= sizeof(void*), "" ); |
79 | size_t rawSize = (sizeof(DisjunctionContext) - sizeof(uintptr_t) + Checked<size_t>(numberOfFrames) * sizeof(uintptr_t)).unsafeGet(); |
80 | size_t roundedSize = roundUpToMultipleOf<sizeof(void*)>(rawSize); |
81 | RELEASE_ASSERT(roundedSize >= rawSize); |
82 | return roundedSize; |
83 | } |
84 | |
85 | int term { 0 }; |
86 | unsigned matchBegin; |
87 | unsigned matchEnd; |
88 | uintptr_t frame[1]; |
89 | }; |
90 | |
91 | DisjunctionContext* allocDisjunctionContext(ByteDisjunction* disjunction) |
92 | { |
93 | size_t size = DisjunctionContext::allocationSize(disjunction->m_frameSize); |
94 | allocatorPool = allocatorPool->ensureCapacity(size); |
95 | RELEASE_ASSERT(allocatorPool); |
96 | return new (allocatorPool->alloc(size)) DisjunctionContext(); |
97 | } |
98 | |
99 | void freeDisjunctionContext(DisjunctionContext* context) |
100 | { |
101 | allocatorPool = allocatorPool->dealloc(context); |
102 | } |
103 | |
104 | struct ParenthesesDisjunctionContext |
105 | { |
106 | ParenthesesDisjunctionContext(unsigned* output, ByteTerm& term) |
107 | { |
108 | unsigned firstSubpatternId = term.atom.subpatternId; |
109 | unsigned numNestedSubpatterns = term.atom.parenthesesDisjunction->m_numSubpatterns; |
110 | |
111 | for (unsigned i = 0; i < (numNestedSubpatterns << 1); ++i) { |
112 | subpatternBackup[i] = output[(firstSubpatternId << 1) + i]; |
113 | output[(firstSubpatternId << 1) + i] = offsetNoMatch; |
114 | } |
115 | |
116 | new (getDisjunctionContext(term)) DisjunctionContext(); |
117 | } |
118 | |
119 | void* operator new(size_t, void* where) |
120 | { |
121 | return where; |
122 | } |
123 | |
124 | void restoreOutput(unsigned* output, unsigned firstSubpatternId, unsigned numNestedSubpatterns) |
125 | { |
126 | for (unsigned i = 0; i < (numNestedSubpatterns << 1); ++i) |
127 | output[(firstSubpatternId << 1) + i] = subpatternBackup[i]; |
128 | } |
129 | |
130 | DisjunctionContext* getDisjunctionContext(ByteTerm& term) |
131 | { |
132 | return bitwise_cast<DisjunctionContext*>(bitwise_cast<uintptr_t>(this) + allocationSize(term.atom.parenthesesDisjunction->m_numSubpatterns)); |
133 | } |
134 | |
135 | static size_t allocationSize(unsigned numberOfSubpatterns) |
136 | { |
137 | static_assert(alignof(ParenthesesDisjunctionContext) <= sizeof(void*), "" ); |
138 | size_t rawSize = (sizeof(ParenthesesDisjunctionContext) - sizeof(unsigned) + (Checked<size_t>(numberOfSubpatterns) * 2U) * sizeof(unsigned)).unsafeGet(); |
139 | size_t roundedSize = roundUpToMultipleOf<sizeof(void*)>(rawSize); |
140 | RELEASE_ASSERT(roundedSize >= rawSize); |
141 | return roundedSize; |
142 | } |
143 | |
144 | ParenthesesDisjunctionContext* next { nullptr }; |
145 | unsigned subpatternBackup[1]; |
146 | }; |
147 | |
148 | ParenthesesDisjunctionContext* allocParenthesesDisjunctionContext(ByteDisjunction* disjunction, unsigned* output, ByteTerm& term) |
149 | { |
150 | size_t size = (Checked<size_t>(ParenthesesDisjunctionContext::allocationSize(term.atom.parenthesesDisjunction->m_numSubpatterns)) + DisjunctionContext::allocationSize(disjunction->m_frameSize)).unsafeGet(); |
151 | allocatorPool = allocatorPool->ensureCapacity(size); |
152 | RELEASE_ASSERT(allocatorPool); |
153 | return new (allocatorPool->alloc(size)) ParenthesesDisjunctionContext(output, term); |
154 | } |
155 | |
156 | void freeParenthesesDisjunctionContext(ParenthesesDisjunctionContext* context) |
157 | { |
158 | allocatorPool = allocatorPool->dealloc(context); |
159 | } |
160 | |
161 | class InputStream { |
162 | public: |
163 | InputStream(const CharType* input, unsigned start, unsigned length, bool decodeSurrogatePairs) |
164 | : input(input) |
165 | , pos(start) |
166 | , length(length) |
167 | , decodeSurrogatePairs(decodeSurrogatePairs) |
168 | { |
169 | } |
170 | |
171 | void next() |
172 | { |
173 | ++pos; |
174 | } |
175 | |
176 | void rewind(unsigned amount) |
177 | { |
178 | ASSERT(pos >= amount); |
179 | pos -= amount; |
180 | } |
181 | |
182 | int read() |
183 | { |
184 | ASSERT(pos < length); |
185 | if (pos < length) |
186 | return input[pos]; |
187 | return -1; |
188 | } |
189 | |
190 | int readPair() |
191 | { |
192 | ASSERT(pos + 1 < length); |
193 | return input[pos] | input[pos + 1] << 16; |
194 | } |
195 | |
196 | int readChecked(unsigned negativePositionOffest) |
197 | { |
198 | RELEASE_ASSERT(pos >= negativePositionOffest); |
199 | unsigned p = pos - negativePositionOffest; |
200 | ASSERT(p < length); |
201 | int result = input[p]; |
202 | if (U16_IS_LEAD(result) && decodeSurrogatePairs && p + 1 < length && U16_IS_TRAIL(input[p + 1])) { |
203 | if (atEnd()) |
204 | return -1; |
205 | |
206 | result = U16_GET_SUPPLEMENTARY(result, input[p + 1]); |
207 | next(); |
208 | } |
209 | return result; |
210 | } |
211 | |
212 | int readSurrogatePairChecked(unsigned negativePositionOffset) |
213 | { |
214 | RELEASE_ASSERT(pos >= negativePositionOffset); |
215 | unsigned p = pos - negativePositionOffset; |
216 | ASSERT(p < length); |
217 | if (p + 1 >= length) |
218 | return -1; |
219 | |
220 | int first = input[p]; |
221 | int second = input[p + 1]; |
222 | if (U16_IS_LEAD(first) && U16_IS_TRAIL(second)) |
223 | return U16_GET_SUPPLEMENTARY(first, second); |
224 | |
225 | return -1; |
226 | } |
227 | |
228 | int reread(unsigned from) |
229 | { |
230 | ASSERT(from < length); |
231 | int result = input[from]; |
232 | if (U16_IS_LEAD(result) && decodeSurrogatePairs && from + 1 < length && U16_IS_TRAIL(input[from + 1])) |
233 | result = U16_GET_SUPPLEMENTARY(result, input[from + 1]); |
234 | return result; |
235 | } |
236 | |
237 | int prev() |
238 | { |
239 | ASSERT(!(pos > length)); |
240 | if (pos && length) |
241 | return input[pos - 1]; |
242 | return -1; |
243 | } |
244 | |
245 | unsigned getPos() |
246 | { |
247 | return pos; |
248 | } |
249 | |
250 | void setPos(unsigned p) |
251 | { |
252 | pos = p; |
253 | } |
254 | |
255 | bool atStart() |
256 | { |
257 | return pos == 0; |
258 | } |
259 | |
260 | bool atEnd() |
261 | { |
262 | return pos == length; |
263 | } |
264 | |
265 | unsigned end() |
266 | { |
267 | return length; |
268 | } |
269 | |
270 | bool checkInput(unsigned count) |
271 | { |
272 | if (((pos + count) <= length) && ((pos + count) >= pos)) { |
273 | pos += count; |
274 | return true; |
275 | } |
276 | return false; |
277 | } |
278 | |
279 | void uncheckInput(unsigned count) |
280 | { |
281 | RELEASE_ASSERT(pos >= count); |
282 | pos -= count; |
283 | } |
284 | |
285 | bool atStart(unsigned negativePositionOffset) |
286 | { |
287 | return pos == negativePositionOffset; |
288 | } |
289 | |
290 | bool atEnd(unsigned negativePositionOffest) |
291 | { |
292 | RELEASE_ASSERT(pos >= negativePositionOffest); |
293 | return (pos - negativePositionOffest) == length; |
294 | } |
295 | |
296 | bool isAvailableInput(unsigned offset) |
297 | { |
298 | return (((pos + offset) <= length) && ((pos + offset) >= pos)); |
299 | } |
300 | |
301 | private: |
302 | const CharType* input; |
303 | unsigned pos; |
304 | unsigned length; |
305 | bool decodeSurrogatePairs; |
306 | }; |
307 | |
308 | bool testCharacterClass(CharacterClass* characterClass, int ch) |
309 | { |
310 | auto linearSearchMatches = [&ch](const Vector<UChar32>& matches) { |
311 | for (unsigned i = 0; i < matches.size(); ++i) { |
312 | if (ch == matches[i]) |
313 | return true; |
314 | } |
315 | |
316 | return false; |
317 | }; |
318 | |
319 | auto binarySearchMatches = [&ch](const Vector<UChar32>& matches) { |
320 | size_t low = 0; |
321 | size_t high = matches.size() - 1; |
322 | |
323 | while (low <= high) { |
324 | size_t mid = low + (high - low) / 2; |
325 | int diff = ch - matches[mid]; |
326 | if (!diff) |
327 | return true; |
328 | |
329 | if (diff < 0) { |
330 | if (mid == low) |
331 | return false; |
332 | high = mid - 1; |
333 | } else |
334 | low = mid + 1; |
335 | } |
336 | return false; |
337 | }; |
338 | |
339 | auto linearSearchRanges = [&ch](const Vector<CharacterRange>& ranges) { |
340 | for (unsigned i = 0; i < ranges.size(); ++i) { |
341 | if ((ch >= ranges[i].begin) && (ch <= ranges[i].end)) |
342 | return true; |
343 | } |
344 | |
345 | return false; |
346 | }; |
347 | |
348 | auto binarySearchRanges = [&ch](const Vector<CharacterRange>& ranges) { |
349 | size_t low = 0; |
350 | size_t high = ranges.size() - 1; |
351 | |
352 | while (low <= high) { |
353 | size_t mid = low + (high - low) / 2; |
354 | int rangeBeginDiff = ch - ranges[mid].begin; |
355 | if (rangeBeginDiff >= 0 && ch <= ranges[mid].end) |
356 | return true; |
357 | |
358 | if (rangeBeginDiff < 0) { |
359 | if (mid == low) |
360 | return false; |
361 | high = mid - 1; |
362 | } else |
363 | low = mid + 1; |
364 | } |
365 | return false; |
366 | }; |
367 | |
368 | if (characterClass->m_anyCharacter) |
369 | return true; |
370 | |
371 | const size_t thresholdForBinarySearch = 6; |
372 | |
373 | if (!isASCII(ch)) { |
374 | if (characterClass->m_matchesUnicode.size()) { |
375 | if (characterClass->m_matchesUnicode.size() > thresholdForBinarySearch) { |
376 | if (binarySearchMatches(characterClass->m_matchesUnicode)) |
377 | return true; |
378 | } else if (linearSearchMatches(characterClass->m_matchesUnicode)) |
379 | return true; |
380 | } |
381 | |
382 | if (characterClass->m_rangesUnicode.size()) { |
383 | if (characterClass->m_rangesUnicode.size() > thresholdForBinarySearch) { |
384 | if (binarySearchRanges(characterClass->m_rangesUnicode)) |
385 | return true; |
386 | } else if (linearSearchRanges(characterClass->m_rangesUnicode)) |
387 | return true; |
388 | } |
389 | } else { |
390 | if (characterClass->m_matches.size()) { |
391 | if (characterClass->m_matches.size() > thresholdForBinarySearch) { |
392 | if (binarySearchMatches(characterClass->m_matches)) |
393 | return true; |
394 | } else if (linearSearchMatches(characterClass->m_matches)) |
395 | return true; |
396 | } |
397 | |
398 | if (characterClass->m_ranges.size()) { |
399 | if (characterClass->m_ranges.size() > thresholdForBinarySearch) { |
400 | if (binarySearchRanges(characterClass->m_ranges)) |
401 | return true; |
402 | } else if (linearSearchRanges(characterClass->m_ranges)) |
403 | return true; |
404 | } |
405 | } |
406 | |
407 | return false; |
408 | } |
409 | |
410 | bool checkCharacter(int testChar, unsigned negativeInputOffset) |
411 | { |
412 | return testChar == input.readChecked(negativeInputOffset); |
413 | } |
414 | |
415 | bool checkSurrogatePair(int testUnicodeChar, unsigned negativeInputOffset) |
416 | { |
417 | return testUnicodeChar == input.readSurrogatePairChecked(negativeInputOffset); |
418 | } |
419 | |
420 | bool checkCasedCharacter(int loChar, int hiChar, unsigned negativeInputOffset) |
421 | { |
422 | int ch = input.readChecked(negativeInputOffset); |
423 | return (loChar == ch) || (hiChar == ch); |
424 | } |
425 | |
426 | bool checkCharacterClass(CharacterClass* characterClass, bool invert, unsigned negativeInputOffset) |
427 | { |
428 | bool match = testCharacterClass(characterClass, input.readChecked(negativeInputOffset)); |
429 | return invert ? !match : match; |
430 | } |
431 | |
432 | bool checkCharacterClassDontAdvanceInputForNonBMP(CharacterClass* characterClass, unsigned negativeInputOffset) |
433 | { |
434 | int readCharacter = characterClass->hasOnlyNonBMPCharacters() ? input.readSurrogatePairChecked(negativeInputOffset) : input.readChecked(negativeInputOffset); |
435 | return testCharacterClass(characterClass, readCharacter); |
436 | } |
437 | |
438 | bool tryConsumeBackReference(int matchBegin, int matchEnd, unsigned negativeInputOffset) |
439 | { |
440 | unsigned matchSize = (unsigned)(matchEnd - matchBegin); |
441 | |
442 | if (!input.checkInput(matchSize)) |
443 | return false; |
444 | |
445 | for (unsigned i = 0; i < matchSize; ++i) { |
446 | int oldCh = input.reread(matchBegin + i); |
447 | int ch; |
448 | if (!U_IS_BMP(oldCh)) { |
449 | ch = input.readSurrogatePairChecked(negativeInputOffset + matchSize - i); |
450 | ++i; |
451 | } else |
452 | ch = input.readChecked(negativeInputOffset + matchSize - i); |
453 | |
454 | if (oldCh == ch) |
455 | continue; |
456 | |
457 | if (pattern->ignoreCase()) { |
458 | // See ES 6.0, 21.2.2.8.2 for the definition of Canonicalize(). For non-Unicode |
459 | // patterns, Unicode values are never allowed to match against ASCII ones. |
460 | // For Unicode, we need to check all canonical equivalents of a character. |
461 | if (!unicode && (isASCII(oldCh) || isASCII(ch))) { |
462 | if (toASCIIUpper(oldCh) == toASCIIUpper(ch)) |
463 | continue; |
464 | } else if (areCanonicallyEquivalent(oldCh, ch, unicode ? CanonicalMode::Unicode : CanonicalMode::UCS2)) |
465 | continue; |
466 | } |
467 | |
468 | input.uncheckInput(matchSize); |
469 | return false; |
470 | } |
471 | |
472 | return true; |
473 | } |
474 | |
475 | bool matchAssertionBOL(ByteTerm& term) |
476 | { |
477 | return (input.atStart(term.inputPosition)) || (pattern->multiline() && testCharacterClass(pattern->newlineCharacterClass, input.readChecked(term.inputPosition + 1))); |
478 | } |
479 | |
480 | bool matchAssertionEOL(ByteTerm& term) |
481 | { |
482 | if (term.inputPosition) |
483 | return (input.atEnd(term.inputPosition)) || (pattern->multiline() && testCharacterClass(pattern->newlineCharacterClass, input.readChecked(term.inputPosition))); |
484 | |
485 | return (input.atEnd()) || (pattern->multiline() && testCharacterClass(pattern->newlineCharacterClass, input.read())); |
486 | } |
487 | |
488 | bool matchAssertionWordBoundary(ByteTerm& term) |
489 | { |
490 | bool prevIsWordchar = !input.atStart(term.inputPosition) && testCharacterClass(pattern->wordcharCharacterClass, input.readChecked(term.inputPosition + 1)); |
491 | bool readIsWordchar; |
492 | if (term.inputPosition) |
493 | readIsWordchar = !input.atEnd(term.inputPosition) && testCharacterClass(pattern->wordcharCharacterClass, input.readChecked(term.inputPosition)); |
494 | else |
495 | readIsWordchar = !input.atEnd() && testCharacterClass(pattern->wordcharCharacterClass, input.read()); |
496 | |
497 | bool wordBoundary = prevIsWordchar != readIsWordchar; |
498 | return term.invert() ? !wordBoundary : wordBoundary; |
499 | } |
500 | |
501 | bool backtrackPatternCharacter(ByteTerm& term, DisjunctionContext* context) |
502 | { |
503 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + term.frameLocation); |
504 | |
505 | switch (term.atom.quantityType) { |
506 | case QuantifierFixedCount: |
507 | break; |
508 | |
509 | case QuantifierGreedy: |
510 | if (backTrack->matchAmount) { |
511 | --backTrack->matchAmount; |
512 | input.uncheckInput(U16_LENGTH(term.atom.patternCharacter)); |
513 | return true; |
514 | } |
515 | break; |
516 | |
517 | case QuantifierNonGreedy: |
518 | if ((backTrack->matchAmount < term.atom.quantityMaxCount) && input.checkInput(1)) { |
519 | ++backTrack->matchAmount; |
520 | if (checkCharacter(term.atom.patternCharacter, term.inputPosition + 1)) |
521 | return true; |
522 | } |
523 | input.setPos(backTrack->begin); |
524 | break; |
525 | } |
526 | |
527 | return false; |
528 | } |
529 | |
530 | bool backtrackPatternCasedCharacter(ByteTerm& term, DisjunctionContext* context) |
531 | { |
532 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + term.frameLocation); |
533 | |
534 | switch (term.atom.quantityType) { |
535 | case QuantifierFixedCount: |
536 | break; |
537 | |
538 | case QuantifierGreedy: |
539 | if (backTrack->matchAmount) { |
540 | --backTrack->matchAmount; |
541 | input.uncheckInput(1); |
542 | return true; |
543 | } |
544 | break; |
545 | |
546 | case QuantifierNonGreedy: |
547 | if ((backTrack->matchAmount < term.atom.quantityMaxCount) && input.checkInput(1)) { |
548 | ++backTrack->matchAmount; |
549 | if (checkCasedCharacter(term.atom.casedCharacter.lo, term.atom.casedCharacter.hi, term.inputPosition + 1)) |
550 | return true; |
551 | } |
552 | input.uncheckInput(backTrack->matchAmount); |
553 | break; |
554 | } |
555 | |
556 | return false; |
557 | } |
558 | |
559 | bool matchCharacterClass(ByteTerm& term, DisjunctionContext* context) |
560 | { |
561 | ASSERT(term.type == ByteTerm::TypeCharacterClass); |
562 | BackTrackInfoCharacterClass* backTrack = reinterpret_cast<BackTrackInfoCharacterClass*>(context->frame + term.frameLocation); |
563 | |
564 | switch (term.atom.quantityType) { |
565 | case QuantifierFixedCount: { |
566 | if (unicode) { |
567 | CharacterClass* charClass = term.atom.characterClass; |
568 | backTrack->begin = input.getPos(); |
569 | unsigned matchAmount = 0; |
570 | for (matchAmount = 0; matchAmount < term.atom.quantityMaxCount; ++matchAmount) { |
571 | if (term.invert()) { |
572 | if (!checkCharacterClass(charClass, term.invert(), term.inputPosition - matchAmount)) { |
573 | input.setPos(backTrack->begin); |
574 | return false; |
575 | } |
576 | } else { |
577 | unsigned matchOffset = matchAmount * (charClass->hasOnlyNonBMPCharacters() ? 2 : 1); |
578 | if (!checkCharacterClassDontAdvanceInputForNonBMP(charClass, term.inputPosition - matchOffset)) { |
579 | input.setPos(backTrack->begin); |
580 | return false; |
581 | } |
582 | } |
583 | } |
584 | |
585 | return true; |
586 | } |
587 | |
588 | for (unsigned matchAmount = 0; matchAmount < term.atom.quantityMaxCount; ++matchAmount) { |
589 | if (!checkCharacterClass(term.atom.characterClass, term.invert(), term.inputPosition - matchAmount)) |
590 | return false; |
591 | } |
592 | return true; |
593 | } |
594 | |
595 | case QuantifierGreedy: { |
596 | unsigned position = input.getPos(); |
597 | backTrack->begin = position; |
598 | unsigned matchAmount = 0; |
599 | while ((matchAmount < term.atom.quantityMaxCount) && input.checkInput(1)) { |
600 | if (!checkCharacterClass(term.atom.characterClass, term.invert(), term.inputPosition + 1)) { |
601 | input.setPos(position); |
602 | break; |
603 | } |
604 | ++matchAmount; |
605 | position = input.getPos(); |
606 | } |
607 | backTrack->matchAmount = matchAmount; |
608 | |
609 | return true; |
610 | } |
611 | |
612 | case QuantifierNonGreedy: |
613 | backTrack->begin = input.getPos(); |
614 | backTrack->matchAmount = 0; |
615 | return true; |
616 | } |
617 | |
618 | RELEASE_ASSERT_NOT_REACHED(); |
619 | return false; |
620 | } |
621 | |
622 | bool backtrackCharacterClass(ByteTerm& term, DisjunctionContext* context) |
623 | { |
624 | ASSERT(term.type == ByteTerm::TypeCharacterClass); |
625 | BackTrackInfoCharacterClass* backTrack = reinterpret_cast<BackTrackInfoCharacterClass*>(context->frame + term.frameLocation); |
626 | |
627 | switch (term.atom.quantityType) { |
628 | case QuantifierFixedCount: |
629 | if (unicode) |
630 | input.setPos(backTrack->begin); |
631 | break; |
632 | |
633 | case QuantifierGreedy: |
634 | if (backTrack->matchAmount) { |
635 | if (unicode) { |
636 | // Rematch one less match |
637 | input.setPos(backTrack->begin); |
638 | --backTrack->matchAmount; |
639 | for (unsigned matchAmount = 0; (matchAmount < backTrack->matchAmount) && input.checkInput(1); ++matchAmount) { |
640 | if (!checkCharacterClass(term.atom.characterClass, term.invert(), term.inputPosition + 1)) { |
641 | input.uncheckInput(1); |
642 | break; |
643 | } |
644 | } |
645 | return true; |
646 | } |
647 | --backTrack->matchAmount; |
648 | input.uncheckInput(1); |
649 | return true; |
650 | } |
651 | break; |
652 | |
653 | case QuantifierNonGreedy: |
654 | if ((backTrack->matchAmount < term.atom.quantityMaxCount) && input.checkInput(1)) { |
655 | ++backTrack->matchAmount; |
656 | if (checkCharacterClass(term.atom.characterClass, term.invert(), term.inputPosition + 1)) |
657 | return true; |
658 | } |
659 | input.setPos(backTrack->begin); |
660 | break; |
661 | } |
662 | |
663 | return false; |
664 | } |
665 | |
666 | bool matchBackReference(ByteTerm& term, DisjunctionContext* context) |
667 | { |
668 | ASSERT(term.type == ByteTerm::TypeBackReference); |
669 | BackTrackInfoBackReference* backTrack = reinterpret_cast<BackTrackInfoBackReference*>(context->frame + term.frameLocation); |
670 | |
671 | unsigned matchBegin = output[(term.atom.subpatternId << 1)]; |
672 | unsigned matchEnd = output[(term.atom.subpatternId << 1) + 1]; |
673 | |
674 | // If the end position of the referenced match hasn't set yet then the backreference in the same parentheses where it references to that. |
675 | // In this case the result of match is empty string like when it references to a parentheses with zero-width match. |
676 | // Eg.: /(a\1)/ |
677 | if (matchEnd == offsetNoMatch) |
678 | return true; |
679 | |
680 | if (matchBegin == offsetNoMatch) |
681 | return true; |
682 | |
683 | ASSERT(matchBegin <= matchEnd); |
684 | |
685 | if (matchBegin == matchEnd) |
686 | return true; |
687 | |
688 | switch (term.atom.quantityType) { |
689 | case QuantifierFixedCount: { |
690 | backTrack->begin = input.getPos(); |
691 | for (unsigned matchAmount = 0; matchAmount < term.atom.quantityMaxCount; ++matchAmount) { |
692 | if (!tryConsumeBackReference(matchBegin, matchEnd, term.inputPosition)) { |
693 | input.setPos(backTrack->begin); |
694 | return false; |
695 | } |
696 | } |
697 | return true; |
698 | } |
699 | |
700 | case QuantifierGreedy: { |
701 | unsigned matchAmount = 0; |
702 | while ((matchAmount < term.atom.quantityMaxCount) && tryConsumeBackReference(matchBegin, matchEnd, term.inputPosition)) |
703 | ++matchAmount; |
704 | backTrack->matchAmount = matchAmount; |
705 | return true; |
706 | } |
707 | |
708 | case QuantifierNonGreedy: |
709 | backTrack->begin = input.getPos(); |
710 | backTrack->matchAmount = 0; |
711 | return true; |
712 | } |
713 | |
714 | RELEASE_ASSERT_NOT_REACHED(); |
715 | return false; |
716 | } |
717 | |
718 | bool backtrackBackReference(ByteTerm& term, DisjunctionContext* context) |
719 | { |
720 | ASSERT(term.type == ByteTerm::TypeBackReference); |
721 | BackTrackInfoBackReference* backTrack = reinterpret_cast<BackTrackInfoBackReference*>(context->frame + term.frameLocation); |
722 | |
723 | unsigned matchBegin = output[(term.atom.subpatternId << 1)]; |
724 | unsigned matchEnd = output[(term.atom.subpatternId << 1) + 1]; |
725 | |
726 | if (matchBegin == offsetNoMatch) |
727 | return false; |
728 | |
729 | ASSERT(matchBegin <= matchEnd); |
730 | |
731 | if (matchBegin == matchEnd) |
732 | return false; |
733 | |
734 | switch (term.atom.quantityType) { |
735 | case QuantifierFixedCount: |
736 | // for quantityMaxCount == 1, could rewind. |
737 | input.setPos(backTrack->begin); |
738 | break; |
739 | |
740 | case QuantifierGreedy: |
741 | if (backTrack->matchAmount) { |
742 | --backTrack->matchAmount; |
743 | input.rewind(matchEnd - matchBegin); |
744 | return true; |
745 | } |
746 | break; |
747 | |
748 | case QuantifierNonGreedy: |
749 | if ((backTrack->matchAmount < term.atom.quantityMaxCount) && tryConsumeBackReference(matchBegin, matchEnd, term.inputPosition)) { |
750 | ++backTrack->matchAmount; |
751 | return true; |
752 | } |
753 | input.setPos(backTrack->begin); |
754 | break; |
755 | } |
756 | |
757 | return false; |
758 | } |
759 | |
760 | void recordParenthesesMatch(ByteTerm& term, ParenthesesDisjunctionContext* context) |
761 | { |
762 | if (term.capture()) { |
763 | unsigned subpatternId = term.atom.subpatternId; |
764 | output[(subpatternId << 1)] = context->getDisjunctionContext(term)->matchBegin - term.inputPosition; |
765 | output[(subpatternId << 1) + 1] = context->getDisjunctionContext(term)->matchEnd - term.inputPosition; |
766 | } |
767 | } |
768 | void resetMatches(ByteTerm& term, ParenthesesDisjunctionContext* context) |
769 | { |
770 | unsigned firstSubpatternId = term.atom.subpatternId; |
771 | unsigned count = term.atom.parenthesesDisjunction->m_numSubpatterns; |
772 | context->restoreOutput(output, firstSubpatternId, count); |
773 | } |
774 | JSRegExpResult parenthesesDoBacktrack(ByteTerm& term, BackTrackInfoParentheses* backTrack) |
775 | { |
776 | while (backTrack->matchAmount) { |
777 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
778 | |
779 | JSRegExpResult result = matchDisjunction(term.atom.parenthesesDisjunction, context->getDisjunctionContext(term), true); |
780 | if (result == JSRegExpMatch) |
781 | return JSRegExpMatch; |
782 | |
783 | resetMatches(term, context); |
784 | popParenthesesDisjunctionContext(backTrack); |
785 | freeParenthesesDisjunctionContext(context); |
786 | |
787 | if (result != JSRegExpNoMatch) |
788 | return result; |
789 | } |
790 | |
791 | return JSRegExpNoMatch; |
792 | } |
793 | |
794 | bool matchParenthesesOnceBegin(ByteTerm& term, DisjunctionContext* context) |
795 | { |
796 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
797 | ASSERT(term.atom.quantityMaxCount == 1); |
798 | |
799 | BackTrackInfoParenthesesOnce* backTrack = reinterpret_cast<BackTrackInfoParenthesesOnce*>(context->frame + term.frameLocation); |
800 | |
801 | switch (term.atom.quantityType) { |
802 | case QuantifierGreedy: { |
803 | // set this speculatively; if we get to the parens end this will be true. |
804 | backTrack->begin = input.getPos(); |
805 | break; |
806 | } |
807 | case QuantifierNonGreedy: { |
808 | backTrack->begin = notFound; |
809 | context->term += term.atom.parenthesesWidth; |
810 | return true; |
811 | } |
812 | case QuantifierFixedCount: |
813 | break; |
814 | } |
815 | |
816 | if (term.capture()) { |
817 | unsigned subpatternId = term.atom.subpatternId; |
818 | output[(subpatternId << 1)] = input.getPos() - term.inputPosition; |
819 | } |
820 | |
821 | return true; |
822 | } |
823 | |
824 | bool matchParenthesesOnceEnd(ByteTerm& term, DisjunctionContext* context) |
825 | { |
826 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternOnceEnd); |
827 | ASSERT(term.atom.quantityMaxCount == 1); |
828 | |
829 | if (term.capture()) { |
830 | unsigned subpatternId = term.atom.subpatternId; |
831 | output[(subpatternId << 1) + 1] = input.getPos() - term.inputPosition; |
832 | } |
833 | |
834 | if (term.atom.quantityType == QuantifierFixedCount) |
835 | return true; |
836 | |
837 | BackTrackInfoParenthesesOnce* backTrack = reinterpret_cast<BackTrackInfoParenthesesOnce*>(context->frame + term.frameLocation); |
838 | return backTrack->begin != input.getPos(); |
839 | } |
840 | |
841 | bool backtrackParenthesesOnceBegin(ByteTerm& term, DisjunctionContext* context) |
842 | { |
843 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
844 | ASSERT(term.atom.quantityMaxCount == 1); |
845 | |
846 | BackTrackInfoParenthesesOnce* backTrack = reinterpret_cast<BackTrackInfoParenthesesOnce*>(context->frame + term.frameLocation); |
847 | |
848 | if (term.capture()) { |
849 | unsigned subpatternId = term.atom.subpatternId; |
850 | output[(subpatternId << 1)] = offsetNoMatch; |
851 | output[(subpatternId << 1) + 1] = offsetNoMatch; |
852 | } |
853 | |
854 | switch (term.atom.quantityType) { |
855 | case QuantifierGreedy: |
856 | // if we backtrack to this point, there is another chance - try matching nothing. |
857 | ASSERT(backTrack->begin != notFound); |
858 | backTrack->begin = notFound; |
859 | context->term += term.atom.parenthesesWidth; |
860 | return true; |
861 | case QuantifierNonGreedy: |
862 | ASSERT(backTrack->begin != notFound); |
863 | FALLTHROUGH; |
864 | case QuantifierFixedCount: |
865 | break; |
866 | } |
867 | |
868 | return false; |
869 | } |
870 | |
871 | bool backtrackParenthesesOnceEnd(ByteTerm& term, DisjunctionContext* context) |
872 | { |
873 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternOnceEnd); |
874 | ASSERT(term.atom.quantityMaxCount == 1); |
875 | |
876 | BackTrackInfoParenthesesOnce* backTrack = reinterpret_cast<BackTrackInfoParenthesesOnce*>(context->frame + term.frameLocation); |
877 | |
878 | switch (term.atom.quantityType) { |
879 | case QuantifierGreedy: |
880 | if (backTrack->begin == notFound) { |
881 | context->term -= term.atom.parenthesesWidth; |
882 | return false; |
883 | } |
884 | FALLTHROUGH; |
885 | case QuantifierNonGreedy: |
886 | if (backTrack->begin == notFound) { |
887 | backTrack->begin = input.getPos(); |
888 | if (term.capture()) { |
889 | // Technically this access to inputPosition should be accessing the begin term's |
890 | // inputPosition, but for repeats other than fixed these values should be |
891 | // the same anyway! (We don't pre-check for greedy or non-greedy matches.) |
892 | ASSERT((&term - term.atom.parenthesesWidth)->type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
893 | ASSERT((&term - term.atom.parenthesesWidth)->inputPosition == term.inputPosition); |
894 | unsigned subpatternId = term.atom.subpatternId; |
895 | output[subpatternId << 1] = input.getPos() - term.inputPosition; |
896 | } |
897 | context->term -= term.atom.parenthesesWidth; |
898 | return true; |
899 | } |
900 | FALLTHROUGH; |
901 | case QuantifierFixedCount: |
902 | break; |
903 | } |
904 | |
905 | return false; |
906 | } |
907 | |
908 | bool matchParenthesesTerminalBegin(ByteTerm& term, DisjunctionContext* context) |
909 | { |
910 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternTerminalBegin); |
911 | ASSERT(term.atom.quantityType == QuantifierGreedy); |
912 | ASSERT(term.atom.quantityMaxCount == quantifyInfinite); |
913 | ASSERT(!term.capture()); |
914 | |
915 | BackTrackInfoParenthesesTerminal* backTrack = reinterpret_cast<BackTrackInfoParenthesesTerminal*>(context->frame + term.frameLocation); |
916 | backTrack->begin = input.getPos(); |
917 | return true; |
918 | } |
919 | |
920 | bool matchParenthesesTerminalEnd(ByteTerm& term, DisjunctionContext* context) |
921 | { |
922 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternTerminalEnd); |
923 | |
924 | BackTrackInfoParenthesesTerminal* backTrack = reinterpret_cast<BackTrackInfoParenthesesTerminal*>(context->frame + term.frameLocation); |
925 | // Empty match is a failed match. |
926 | if (backTrack->begin == input.getPos()) |
927 | return false; |
928 | |
929 | // Successful match! Okay, what's next? - loop around and try to match more! |
930 | context->term -= (term.atom.parenthesesWidth + 1); |
931 | return true; |
932 | } |
933 | |
934 | bool backtrackParenthesesTerminalBegin(ByteTerm& term, DisjunctionContext* context) |
935 | { |
936 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternTerminalBegin); |
937 | ASSERT(term.atom.quantityType == QuantifierGreedy); |
938 | ASSERT(term.atom.quantityMaxCount == quantifyInfinite); |
939 | ASSERT(!term.capture()); |
940 | |
941 | // If we backtrack to this point, we have failed to match this iteration of the parens. |
942 | // Since this is greedy / zero minimum a failed is also accepted as a match! |
943 | context->term += term.atom.parenthesesWidth; |
944 | return true; |
945 | } |
946 | |
947 | bool backtrackParenthesesTerminalEnd(ByteTerm&, DisjunctionContext*) |
948 | { |
949 | // 'Terminal' parentheses are at the end of the regex, and as such a match past end |
950 | // should always be returned as a successful match - we should never backtrack to here. |
951 | RELEASE_ASSERT_NOT_REACHED(); |
952 | return false; |
953 | } |
954 | |
955 | bool matchParentheticalAssertionBegin(ByteTerm& term, DisjunctionContext* context) |
956 | { |
957 | ASSERT(term.type == ByteTerm::TypeParentheticalAssertionBegin); |
958 | ASSERT(term.atom.quantityMaxCount == 1); |
959 | |
960 | BackTrackInfoParentheticalAssertion* backTrack = reinterpret_cast<BackTrackInfoParentheticalAssertion*>(context->frame + term.frameLocation); |
961 | |
962 | backTrack->begin = input.getPos(); |
963 | return true; |
964 | } |
965 | |
966 | bool matchParentheticalAssertionEnd(ByteTerm& term, DisjunctionContext* context) |
967 | { |
968 | ASSERT(term.type == ByteTerm::TypeParentheticalAssertionEnd); |
969 | ASSERT(term.atom.quantityMaxCount == 1); |
970 | |
971 | BackTrackInfoParentheticalAssertion* backTrack = reinterpret_cast<BackTrackInfoParentheticalAssertion*>(context->frame + term.frameLocation); |
972 | |
973 | input.setPos(backTrack->begin); |
974 | |
975 | // We've reached the end of the parens; if they are inverted, this is failure. |
976 | if (term.invert()) { |
977 | context->term -= term.atom.parenthesesWidth; |
978 | return false; |
979 | } |
980 | |
981 | return true; |
982 | } |
983 | |
984 | bool backtrackParentheticalAssertionBegin(ByteTerm& term, DisjunctionContext* context) |
985 | { |
986 | ASSERT(term.type == ByteTerm::TypeParentheticalAssertionBegin); |
987 | ASSERT(term.atom.quantityMaxCount == 1); |
988 | |
989 | // We've failed to match parens; if they are inverted, this is win! |
990 | if (term.invert()) { |
991 | context->term += term.atom.parenthesesWidth; |
992 | return true; |
993 | } |
994 | |
995 | return false; |
996 | } |
997 | |
998 | bool backtrackParentheticalAssertionEnd(ByteTerm& term, DisjunctionContext* context) |
999 | { |
1000 | ASSERT(term.type == ByteTerm::TypeParentheticalAssertionEnd); |
1001 | ASSERT(term.atom.quantityMaxCount == 1); |
1002 | |
1003 | BackTrackInfoParentheticalAssertion* backTrack = reinterpret_cast<BackTrackInfoParentheticalAssertion*>(context->frame + term.frameLocation); |
1004 | |
1005 | input.setPos(backTrack->begin); |
1006 | |
1007 | context->term -= term.atom.parenthesesWidth; |
1008 | return false; |
1009 | } |
1010 | |
1011 | JSRegExpResult matchParentheses(ByteTerm& term, DisjunctionContext* context) |
1012 | { |
1013 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpattern); |
1014 | |
1015 | BackTrackInfoParentheses* backTrack = reinterpret_cast<BackTrackInfoParentheses*>(context->frame + term.frameLocation); |
1016 | ByteDisjunction* disjunctionBody = term.atom.parenthesesDisjunction; |
1017 | |
1018 | backTrack->matchAmount = 0; |
1019 | backTrack->lastContext = 0; |
1020 | |
1021 | ASSERT(term.atom.quantityType != QuantifierFixedCount || term.atom.quantityMinCount == term.atom.quantityMaxCount); |
1022 | |
1023 | unsigned minimumMatchCount = term.atom.quantityMinCount; |
1024 | JSRegExpResult fixedMatchResult; |
1025 | |
1026 | // Handle fixed matches and the minimum part of a variable length match. |
1027 | if (minimumMatchCount) { |
1028 | // While we haven't yet reached our fixed limit, |
1029 | while (backTrack->matchAmount < minimumMatchCount) { |
1030 | // Try to do a match, and it it succeeds, add it to the list. |
1031 | ParenthesesDisjunctionContext* context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1032 | fixedMatchResult = matchDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1033 | if (fixedMatchResult == JSRegExpMatch) |
1034 | appendParenthesesDisjunctionContext(backTrack, context); |
1035 | else { |
1036 | // The match failed; try to find an alternate point to carry on from. |
1037 | resetMatches(term, context); |
1038 | freeParenthesesDisjunctionContext(context); |
1039 | |
1040 | if (fixedMatchResult != JSRegExpNoMatch) |
1041 | return fixedMatchResult; |
1042 | JSRegExpResult backtrackResult = parenthesesDoBacktrack(term, backTrack); |
1043 | if (backtrackResult != JSRegExpMatch) |
1044 | return backtrackResult; |
1045 | } |
1046 | } |
1047 | |
1048 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1049 | recordParenthesesMatch(term, context); |
1050 | } |
1051 | |
1052 | switch (term.atom.quantityType) { |
1053 | case QuantifierFixedCount: { |
1054 | ASSERT(backTrack->matchAmount == term.atom.quantityMaxCount); |
1055 | return JSRegExpMatch; |
1056 | } |
1057 | |
1058 | case QuantifierGreedy: { |
1059 | while (backTrack->matchAmount < term.atom.quantityMaxCount) { |
1060 | ParenthesesDisjunctionContext* context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1061 | JSRegExpResult result = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1062 | if (result == JSRegExpMatch) |
1063 | appendParenthesesDisjunctionContext(backTrack, context); |
1064 | else { |
1065 | resetMatches(term, context); |
1066 | freeParenthesesDisjunctionContext(context); |
1067 | |
1068 | if (result != JSRegExpNoMatch) |
1069 | return result; |
1070 | |
1071 | break; |
1072 | } |
1073 | } |
1074 | |
1075 | if (backTrack->matchAmount) { |
1076 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1077 | recordParenthesesMatch(term, context); |
1078 | } |
1079 | return JSRegExpMatch; |
1080 | } |
1081 | |
1082 | case QuantifierNonGreedy: |
1083 | return JSRegExpMatch; |
1084 | } |
1085 | |
1086 | RELEASE_ASSERT_NOT_REACHED(); |
1087 | return JSRegExpErrorNoMatch; |
1088 | } |
1089 | |
1090 | // Rules for backtracking differ depending on whether this is greedy or non-greedy. |
1091 | // |
1092 | // Greedy matches never should try just adding more - you should already have done |
1093 | // the 'more' cases. Always backtrack, at least a leetle bit. However cases where |
1094 | // you backtrack an item off the list needs checking, since we'll never have matched |
1095 | // the one less case. Tracking forwards, still add as much as possible. |
1096 | // |
1097 | // Non-greedy, we've already done the one less case, so don't match on popping. |
1098 | // We haven't done the one more case, so always try to add that. |
1099 | // |
1100 | JSRegExpResult backtrackParentheses(ByteTerm& term, DisjunctionContext* context) |
1101 | { |
1102 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpattern); |
1103 | |
1104 | BackTrackInfoParentheses* backTrack = reinterpret_cast<BackTrackInfoParentheses*>(context->frame + term.frameLocation); |
1105 | ByteDisjunction* disjunctionBody = term.atom.parenthesesDisjunction; |
1106 | |
1107 | switch (term.atom.quantityType) { |
1108 | case QuantifierFixedCount: { |
1109 | ASSERT(backTrack->matchAmount == term.atom.quantityMaxCount); |
1110 | |
1111 | ParenthesesDisjunctionContext* context = 0; |
1112 | JSRegExpResult result = parenthesesDoBacktrack(term, backTrack); |
1113 | |
1114 | if (result != JSRegExpMatch) |
1115 | return result; |
1116 | |
1117 | // While we haven't yet reached our fixed limit, |
1118 | while (backTrack->matchAmount < term.atom.quantityMaxCount) { |
1119 | // Try to do a match, and it it succeeds, add it to the list. |
1120 | context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1121 | result = matchDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1122 | |
1123 | if (result == JSRegExpMatch) |
1124 | appendParenthesesDisjunctionContext(backTrack, context); |
1125 | else { |
1126 | // The match failed; try to find an alternate point to carry on from. |
1127 | resetMatches(term, context); |
1128 | freeParenthesesDisjunctionContext(context); |
1129 | |
1130 | if (result != JSRegExpNoMatch) |
1131 | return result; |
1132 | JSRegExpResult backtrackResult = parenthesesDoBacktrack(term, backTrack); |
1133 | if (backtrackResult != JSRegExpMatch) |
1134 | return backtrackResult; |
1135 | } |
1136 | } |
1137 | |
1138 | ASSERT(backTrack->matchAmount == term.atom.quantityMaxCount); |
1139 | context = backTrack->lastContext; |
1140 | recordParenthesesMatch(term, context); |
1141 | return JSRegExpMatch; |
1142 | } |
1143 | |
1144 | case QuantifierGreedy: { |
1145 | if (!backTrack->matchAmount) |
1146 | return JSRegExpNoMatch; |
1147 | |
1148 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1149 | JSRegExpResult result = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term), true); |
1150 | if (result == JSRegExpMatch) { |
1151 | while (backTrack->matchAmount < term.atom.quantityMaxCount) { |
1152 | ParenthesesDisjunctionContext* context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1153 | JSRegExpResult parenthesesResult = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1154 | if (parenthesesResult == JSRegExpMatch) |
1155 | appendParenthesesDisjunctionContext(backTrack, context); |
1156 | else { |
1157 | resetMatches(term, context); |
1158 | freeParenthesesDisjunctionContext(context); |
1159 | |
1160 | if (parenthesesResult != JSRegExpNoMatch) |
1161 | return parenthesesResult; |
1162 | |
1163 | break; |
1164 | } |
1165 | } |
1166 | } else { |
1167 | resetMatches(term, context); |
1168 | popParenthesesDisjunctionContext(backTrack); |
1169 | freeParenthesesDisjunctionContext(context); |
1170 | |
1171 | if (result != JSRegExpNoMatch || backTrack->matchAmount < term.atom.quantityMinCount) |
1172 | return result; |
1173 | } |
1174 | |
1175 | if (backTrack->matchAmount) { |
1176 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1177 | recordParenthesesMatch(term, context); |
1178 | } |
1179 | return JSRegExpMatch; |
1180 | } |
1181 | |
1182 | case QuantifierNonGreedy: { |
1183 | // If we've not reached the limit, try to add one more match. |
1184 | if (backTrack->matchAmount < term.atom.quantityMaxCount) { |
1185 | ParenthesesDisjunctionContext* context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1186 | JSRegExpResult result = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1187 | if (result == JSRegExpMatch) { |
1188 | appendParenthesesDisjunctionContext(backTrack, context); |
1189 | recordParenthesesMatch(term, context); |
1190 | return JSRegExpMatch; |
1191 | } |
1192 | |
1193 | resetMatches(term, context); |
1194 | freeParenthesesDisjunctionContext(context); |
1195 | |
1196 | if (result != JSRegExpNoMatch) |
1197 | return result; |
1198 | } |
1199 | |
1200 | // Nope - okay backtrack looking for an alternative. |
1201 | while (backTrack->matchAmount) { |
1202 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1203 | JSRegExpResult result = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term), true); |
1204 | if (result == JSRegExpMatch) { |
1205 | // successful backtrack! we're back in the game! |
1206 | if (backTrack->matchAmount) { |
1207 | context = backTrack->lastContext; |
1208 | recordParenthesesMatch(term, context); |
1209 | } |
1210 | return JSRegExpMatch; |
1211 | } |
1212 | |
1213 | // pop a match off the stack |
1214 | resetMatches(term, context); |
1215 | popParenthesesDisjunctionContext(backTrack); |
1216 | freeParenthesesDisjunctionContext(context); |
1217 | |
1218 | if (result != JSRegExpNoMatch) |
1219 | return result; |
1220 | } |
1221 | |
1222 | return JSRegExpNoMatch; |
1223 | } |
1224 | } |
1225 | |
1226 | RELEASE_ASSERT_NOT_REACHED(); |
1227 | return JSRegExpErrorNoMatch; |
1228 | } |
1229 | |
1230 | bool matchDotStarEnclosure(ByteTerm& term, DisjunctionContext* context) |
1231 | { |
1232 | UNUSED_PARAM(term); |
1233 | |
1234 | if (pattern->dotAll()) { |
1235 | context->matchBegin = startOffset; |
1236 | context->matchEnd = input.end(); |
1237 | return true; |
1238 | } |
1239 | |
1240 | unsigned matchBegin = context->matchBegin; |
1241 | |
1242 | if (matchBegin > startOffset) { |
1243 | for (matchBegin--; true; matchBegin--) { |
1244 | if (testCharacterClass(pattern->newlineCharacterClass, input.reread(matchBegin))) { |
1245 | ++matchBegin; |
1246 | break; |
1247 | } |
1248 | |
1249 | if (matchBegin == startOffset) |
1250 | break; |
1251 | } |
1252 | } |
1253 | |
1254 | unsigned matchEnd = input.getPos(); |
1255 | |
1256 | for (; (matchEnd != input.end()) |
1257 | && (!testCharacterClass(pattern->newlineCharacterClass, input.reread(matchEnd))); matchEnd++) { } |
1258 | |
1259 | if (((matchBegin && term.anchors.m_bol) |
1260 | || ((matchEnd != input.end()) && term.anchors.m_eol)) |
1261 | && !pattern->multiline()) |
1262 | return false; |
1263 | |
1264 | context->matchBegin = matchBegin; |
1265 | context->matchEnd = matchEnd; |
1266 | return true; |
1267 | } |
1268 | |
1269 | #define MATCH_NEXT() { ++context->term; goto matchAgain; } |
1270 | #define BACKTRACK() { --context->term; goto backtrack; } |
1271 | #define currentTerm() (disjunction->terms[context->term]) |
1272 | JSRegExpResult matchDisjunction(ByteDisjunction* disjunction, DisjunctionContext* context, bool btrack = false) |
1273 | { |
1274 | if (!--remainingMatchCount) |
1275 | return JSRegExpErrorHitLimit; |
1276 | |
1277 | if (btrack) |
1278 | BACKTRACK(); |
1279 | |
1280 | context->matchBegin = input.getPos(); |
1281 | context->term = 0; |
1282 | |
1283 | matchAgain: |
1284 | ASSERT(context->term < static_cast<int>(disjunction->terms.size())); |
1285 | |
1286 | switch (currentTerm().type) { |
1287 | case ByteTerm::TypeSubpatternBegin: |
1288 | MATCH_NEXT(); |
1289 | case ByteTerm::TypeSubpatternEnd: |
1290 | context->matchEnd = input.getPos(); |
1291 | return JSRegExpMatch; |
1292 | |
1293 | case ByteTerm::TypeBodyAlternativeBegin: |
1294 | MATCH_NEXT(); |
1295 | case ByteTerm::TypeBodyAlternativeDisjunction: |
1296 | case ByteTerm::TypeBodyAlternativeEnd: |
1297 | context->matchEnd = input.getPos(); |
1298 | return JSRegExpMatch; |
1299 | |
1300 | case ByteTerm::TypeAlternativeBegin: |
1301 | MATCH_NEXT(); |
1302 | case ByteTerm::TypeAlternativeDisjunction: |
1303 | case ByteTerm::TypeAlternativeEnd: { |
1304 | int offset = currentTerm().alternative.end; |
1305 | BackTrackInfoAlternative* backTrack = reinterpret_cast<BackTrackInfoAlternative*>(context->frame + currentTerm().frameLocation); |
1306 | backTrack->offset = offset; |
1307 | context->term += offset; |
1308 | MATCH_NEXT(); |
1309 | } |
1310 | |
1311 | case ByteTerm::TypeAssertionBOL: |
1312 | if (matchAssertionBOL(currentTerm())) |
1313 | MATCH_NEXT(); |
1314 | BACKTRACK(); |
1315 | case ByteTerm::TypeAssertionEOL: |
1316 | if (matchAssertionEOL(currentTerm())) |
1317 | MATCH_NEXT(); |
1318 | BACKTRACK(); |
1319 | case ByteTerm::TypeAssertionWordBoundary: |
1320 | if (matchAssertionWordBoundary(currentTerm())) |
1321 | MATCH_NEXT(); |
1322 | BACKTRACK(); |
1323 | |
1324 | case ByteTerm::TypePatternCharacterOnce: |
1325 | case ByteTerm::TypePatternCharacterFixed: { |
1326 | if (unicode) { |
1327 | if (!U_IS_BMP(currentTerm().atom.patternCharacter)) { |
1328 | for (unsigned matchAmount = 0; matchAmount < currentTerm().atom.quantityMaxCount; ++matchAmount) { |
1329 | if (!checkSurrogatePair(currentTerm().atom.patternCharacter, currentTerm().inputPosition - 2 * matchAmount)) { |
1330 | BACKTRACK(); |
1331 | } |
1332 | } |
1333 | MATCH_NEXT(); |
1334 | } |
1335 | } |
1336 | unsigned position = input.getPos(); // May need to back out reading a surrogate pair. |
1337 | |
1338 | for (unsigned matchAmount = 0; matchAmount < currentTerm().atom.quantityMaxCount; ++matchAmount) { |
1339 | if (!checkCharacter(currentTerm().atom.patternCharacter, currentTerm().inputPosition - matchAmount)) { |
1340 | input.setPos(position); |
1341 | BACKTRACK(); |
1342 | } |
1343 | } |
1344 | MATCH_NEXT(); |
1345 | } |
1346 | case ByteTerm::TypePatternCharacterGreedy: { |
1347 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + currentTerm().frameLocation); |
1348 | unsigned matchAmount = 0; |
1349 | unsigned position = input.getPos(); // May need to back out reading a surrogate pair. |
1350 | while ((matchAmount < currentTerm().atom.quantityMaxCount) && input.checkInput(1)) { |
1351 | if (!checkCharacter(currentTerm().atom.patternCharacter, currentTerm().inputPosition + 1)) { |
1352 | input.setPos(position); |
1353 | break; |
1354 | } |
1355 | ++matchAmount; |
1356 | position = input.getPos(); |
1357 | } |
1358 | backTrack->matchAmount = matchAmount; |
1359 | |
1360 | MATCH_NEXT(); |
1361 | } |
1362 | case ByteTerm::TypePatternCharacterNonGreedy: { |
1363 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + currentTerm().frameLocation); |
1364 | backTrack->begin = input.getPos(); |
1365 | backTrack->matchAmount = 0; |
1366 | MATCH_NEXT(); |
1367 | } |
1368 | |
1369 | case ByteTerm::TypePatternCasedCharacterOnce: |
1370 | case ByteTerm::TypePatternCasedCharacterFixed: { |
1371 | if (unicode) { |
1372 | // Case insensitive matching of unicode characters is handled as TypeCharacterClass. |
1373 | ASSERT(U_IS_BMP(currentTerm().atom.patternCharacter)); |
1374 | |
1375 | unsigned position = input.getPos(); // May need to back out reading a surrogate pair. |
1376 | |
1377 | for (unsigned matchAmount = 0; matchAmount < currentTerm().atom.quantityMaxCount; ++matchAmount) { |
1378 | if (!checkCasedCharacter(currentTerm().atom.casedCharacter.lo, currentTerm().atom.casedCharacter.hi, currentTerm().inputPosition - matchAmount)) { |
1379 | input.setPos(position); |
1380 | BACKTRACK(); |
1381 | } |
1382 | } |
1383 | MATCH_NEXT(); |
1384 | } |
1385 | |
1386 | for (unsigned matchAmount = 0; matchAmount < currentTerm().atom.quantityMaxCount; ++matchAmount) { |
1387 | if (!checkCasedCharacter(currentTerm().atom.casedCharacter.lo, currentTerm().atom.casedCharacter.hi, currentTerm().inputPosition - matchAmount)) |
1388 | BACKTRACK(); |
1389 | } |
1390 | MATCH_NEXT(); |
1391 | } |
1392 | case ByteTerm::TypePatternCasedCharacterGreedy: { |
1393 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + currentTerm().frameLocation); |
1394 | |
1395 | // Case insensitive matching of unicode characters is handled as TypeCharacterClass. |
1396 | ASSERT(!unicode || U_IS_BMP(currentTerm().atom.patternCharacter)); |
1397 | |
1398 | unsigned matchAmount = 0; |
1399 | while ((matchAmount < currentTerm().atom.quantityMaxCount) && input.checkInput(1)) { |
1400 | if (!checkCasedCharacter(currentTerm().atom.casedCharacter.lo, currentTerm().atom.casedCharacter.hi, currentTerm().inputPosition + 1)) { |
1401 | input.uncheckInput(1); |
1402 | break; |
1403 | } |
1404 | ++matchAmount; |
1405 | } |
1406 | backTrack->matchAmount = matchAmount; |
1407 | |
1408 | MATCH_NEXT(); |
1409 | } |
1410 | case ByteTerm::TypePatternCasedCharacterNonGreedy: { |
1411 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + currentTerm().frameLocation); |
1412 | |
1413 | // Case insensitive matching of unicode characters is handled as TypeCharacterClass. |
1414 | ASSERT(!unicode || U_IS_BMP(currentTerm().atom.patternCharacter)); |
1415 | |
1416 | backTrack->matchAmount = 0; |
1417 | MATCH_NEXT(); |
1418 | } |
1419 | |
1420 | case ByteTerm::TypeCharacterClass: |
1421 | if (matchCharacterClass(currentTerm(), context)) |
1422 | MATCH_NEXT(); |
1423 | BACKTRACK(); |
1424 | case ByteTerm::TypeBackReference: |
1425 | if (matchBackReference(currentTerm(), context)) |
1426 | MATCH_NEXT(); |
1427 | BACKTRACK(); |
1428 | case ByteTerm::TypeParenthesesSubpattern: { |
1429 | JSRegExpResult result = matchParentheses(currentTerm(), context); |
1430 | |
1431 | if (result == JSRegExpMatch) { |
1432 | MATCH_NEXT(); |
1433 | } else if (result != JSRegExpNoMatch) |
1434 | return result; |
1435 | |
1436 | BACKTRACK(); |
1437 | } |
1438 | case ByteTerm::TypeParenthesesSubpatternOnceBegin: |
1439 | if (matchParenthesesOnceBegin(currentTerm(), context)) |
1440 | MATCH_NEXT(); |
1441 | BACKTRACK(); |
1442 | case ByteTerm::TypeParenthesesSubpatternOnceEnd: |
1443 | if (matchParenthesesOnceEnd(currentTerm(), context)) |
1444 | MATCH_NEXT(); |
1445 | BACKTRACK(); |
1446 | case ByteTerm::TypeParenthesesSubpatternTerminalBegin: |
1447 | if (matchParenthesesTerminalBegin(currentTerm(), context)) |
1448 | MATCH_NEXT(); |
1449 | BACKTRACK(); |
1450 | case ByteTerm::TypeParenthesesSubpatternTerminalEnd: |
1451 | if (matchParenthesesTerminalEnd(currentTerm(), context)) |
1452 | MATCH_NEXT(); |
1453 | BACKTRACK(); |
1454 | case ByteTerm::TypeParentheticalAssertionBegin: |
1455 | if (matchParentheticalAssertionBegin(currentTerm(), context)) |
1456 | MATCH_NEXT(); |
1457 | BACKTRACK(); |
1458 | case ByteTerm::TypeParentheticalAssertionEnd: |
1459 | if (matchParentheticalAssertionEnd(currentTerm(), context)) |
1460 | MATCH_NEXT(); |
1461 | BACKTRACK(); |
1462 | |
1463 | case ByteTerm::TypeCheckInput: |
1464 | if (input.checkInput(currentTerm().checkInputCount)) |
1465 | MATCH_NEXT(); |
1466 | BACKTRACK(); |
1467 | |
1468 | case ByteTerm::TypeUncheckInput: |
1469 | input.uncheckInput(currentTerm().checkInputCount); |
1470 | MATCH_NEXT(); |
1471 | |
1472 | case ByteTerm::TypeDotStarEnclosure: |
1473 | if (matchDotStarEnclosure(currentTerm(), context)) |
1474 | return JSRegExpMatch; |
1475 | BACKTRACK(); |
1476 | } |
1477 | |
1478 | // We should never fall-through to here. |
1479 | RELEASE_ASSERT_NOT_REACHED(); |
1480 | |
1481 | backtrack: |
1482 | ASSERT(context->term < static_cast<int>(disjunction->terms.size())); |
1483 | |
1484 | switch (currentTerm().type) { |
1485 | case ByteTerm::TypeSubpatternBegin: |
1486 | return JSRegExpNoMatch; |
1487 | case ByteTerm::TypeSubpatternEnd: |
1488 | RELEASE_ASSERT_NOT_REACHED(); |
1489 | |
1490 | case ByteTerm::TypeBodyAlternativeBegin: |
1491 | case ByteTerm::TypeBodyAlternativeDisjunction: { |
1492 | int offset = currentTerm().alternative.next; |
1493 | context->term += offset; |
1494 | if (offset > 0) |
1495 | MATCH_NEXT(); |
1496 | |
1497 | if (input.atEnd() || pattern->sticky()) |
1498 | return JSRegExpNoMatch; |
1499 | |
1500 | input.next(); |
1501 | |
1502 | context->matchBegin = input.getPos(); |
1503 | |
1504 | if (currentTerm().alternative.onceThrough) |
1505 | context->term += currentTerm().alternative.next; |
1506 | |
1507 | MATCH_NEXT(); |
1508 | } |
1509 | case ByteTerm::TypeBodyAlternativeEnd: |
1510 | RELEASE_ASSERT_NOT_REACHED(); |
1511 | |
1512 | case ByteTerm::TypeAlternativeBegin: |
1513 | case ByteTerm::TypeAlternativeDisjunction: { |
1514 | int offset = currentTerm().alternative.next; |
1515 | context->term += offset; |
1516 | if (offset > 0) |
1517 | MATCH_NEXT(); |
1518 | BACKTRACK(); |
1519 | } |
1520 | case ByteTerm::TypeAlternativeEnd: { |
1521 | // We should never backtrack back into an alternative of the main body of the regex. |
1522 | BackTrackInfoAlternative* backTrack = reinterpret_cast<BackTrackInfoAlternative*>(context->frame + currentTerm().frameLocation); |
1523 | unsigned offset = backTrack->offset; |
1524 | context->term -= offset; |
1525 | BACKTRACK(); |
1526 | } |
1527 | |
1528 | case ByteTerm::TypeAssertionBOL: |
1529 | case ByteTerm::TypeAssertionEOL: |
1530 | case ByteTerm::TypeAssertionWordBoundary: |
1531 | BACKTRACK(); |
1532 | |
1533 | case ByteTerm::TypePatternCharacterOnce: |
1534 | case ByteTerm::TypePatternCharacterFixed: |
1535 | case ByteTerm::TypePatternCharacterGreedy: |
1536 | case ByteTerm::TypePatternCharacterNonGreedy: |
1537 | if (backtrackPatternCharacter(currentTerm(), context)) |
1538 | MATCH_NEXT(); |
1539 | BACKTRACK(); |
1540 | case ByteTerm::TypePatternCasedCharacterOnce: |
1541 | case ByteTerm::TypePatternCasedCharacterFixed: |
1542 | case ByteTerm::TypePatternCasedCharacterGreedy: |
1543 | case ByteTerm::TypePatternCasedCharacterNonGreedy: |
1544 | if (backtrackPatternCasedCharacter(currentTerm(), context)) |
1545 | MATCH_NEXT(); |
1546 | BACKTRACK(); |
1547 | case ByteTerm::TypeCharacterClass: |
1548 | if (backtrackCharacterClass(currentTerm(), context)) |
1549 | MATCH_NEXT(); |
1550 | BACKTRACK(); |
1551 | case ByteTerm::TypeBackReference: |
1552 | if (backtrackBackReference(currentTerm(), context)) |
1553 | MATCH_NEXT(); |
1554 | BACKTRACK(); |
1555 | case ByteTerm::TypeParenthesesSubpattern: { |
1556 | JSRegExpResult result = backtrackParentheses(currentTerm(), context); |
1557 | |
1558 | if (result == JSRegExpMatch) { |
1559 | MATCH_NEXT(); |
1560 | } else if (result != JSRegExpNoMatch) |
1561 | return result; |
1562 | |
1563 | BACKTRACK(); |
1564 | } |
1565 | case ByteTerm::TypeParenthesesSubpatternOnceBegin: |
1566 | if (backtrackParenthesesOnceBegin(currentTerm(), context)) |
1567 | MATCH_NEXT(); |
1568 | BACKTRACK(); |
1569 | case ByteTerm::TypeParenthesesSubpatternOnceEnd: |
1570 | if (backtrackParenthesesOnceEnd(currentTerm(), context)) |
1571 | MATCH_NEXT(); |
1572 | BACKTRACK(); |
1573 | case ByteTerm::TypeParenthesesSubpatternTerminalBegin: |
1574 | if (backtrackParenthesesTerminalBegin(currentTerm(), context)) |
1575 | MATCH_NEXT(); |
1576 | BACKTRACK(); |
1577 | case ByteTerm::TypeParenthesesSubpatternTerminalEnd: |
1578 | if (backtrackParenthesesTerminalEnd(currentTerm(), context)) |
1579 | MATCH_NEXT(); |
1580 | BACKTRACK(); |
1581 | case ByteTerm::TypeParentheticalAssertionBegin: |
1582 | if (backtrackParentheticalAssertionBegin(currentTerm(), context)) |
1583 | MATCH_NEXT(); |
1584 | BACKTRACK(); |
1585 | case ByteTerm::TypeParentheticalAssertionEnd: |
1586 | if (backtrackParentheticalAssertionEnd(currentTerm(), context)) |
1587 | MATCH_NEXT(); |
1588 | BACKTRACK(); |
1589 | |
1590 | case ByteTerm::TypeCheckInput: |
1591 | input.uncheckInput(currentTerm().checkInputCount); |
1592 | BACKTRACK(); |
1593 | |
1594 | case ByteTerm::TypeUncheckInput: |
1595 | input.checkInput(currentTerm().checkInputCount); |
1596 | BACKTRACK(); |
1597 | |
1598 | case ByteTerm::TypeDotStarEnclosure: |
1599 | RELEASE_ASSERT_NOT_REACHED(); |
1600 | } |
1601 | |
1602 | RELEASE_ASSERT_NOT_REACHED(); |
1603 | return JSRegExpErrorNoMatch; |
1604 | } |
1605 | |
1606 | JSRegExpResult matchNonZeroDisjunction(ByteDisjunction* disjunction, DisjunctionContext* context, bool btrack = false) |
1607 | { |
1608 | JSRegExpResult result = matchDisjunction(disjunction, context, btrack); |
1609 | |
1610 | if (result == JSRegExpMatch) { |
1611 | while (context->matchBegin == context->matchEnd) { |
1612 | result = matchDisjunction(disjunction, context, true); |
1613 | if (result != JSRegExpMatch) |
1614 | return result; |
1615 | } |
1616 | return JSRegExpMatch; |
1617 | } |
1618 | |
1619 | return result; |
1620 | } |
1621 | |
1622 | unsigned interpret() |
1623 | { |
1624 | // FIXME: https://bugs.webkit.org/show_bug.cgi?id=195970 |
1625 | // [Yarr Interpreter] The interpreter doesn't have checks for stack overflow due to deep recursion |
1626 | if (!input.isAvailableInput(0)) |
1627 | return offsetNoMatch; |
1628 | |
1629 | if (pattern->m_lock) |
1630 | pattern->m_lock->lock(); |
1631 | |
1632 | for (unsigned i = 0; i < pattern->m_body->m_numSubpatterns + 1; ++i) |
1633 | output[i << 1] = offsetNoMatch; |
1634 | |
1635 | allocatorPool = pattern->m_allocator->startAllocator(); |
1636 | RELEASE_ASSERT(allocatorPool); |
1637 | |
1638 | DisjunctionContext* context = allocDisjunctionContext(pattern->m_body.get()); |
1639 | |
1640 | JSRegExpResult result = matchDisjunction(pattern->m_body.get(), context, false); |
1641 | if (result == JSRegExpMatch) { |
1642 | output[0] = context->matchBegin; |
1643 | output[1] = context->matchEnd; |
1644 | } |
1645 | |
1646 | freeDisjunctionContext(context); |
1647 | |
1648 | pattern->m_allocator->stopAllocator(); |
1649 | |
1650 | ASSERT((result == JSRegExpMatch) == (output[0] != offsetNoMatch)); |
1651 | |
1652 | if (pattern->m_lock) |
1653 | pattern->m_lock->unlock(); |
1654 | |
1655 | return output[0]; |
1656 | } |
1657 | |
1658 | Interpreter(BytecodePattern* pattern, unsigned* output, const CharType* input, unsigned length, unsigned start) |
1659 | : pattern(pattern) |
1660 | , unicode(pattern->unicode()) |
1661 | , output(output) |
1662 | , input(input, start, length, pattern->unicode()) |
1663 | , startOffset(start) |
1664 | , remainingMatchCount(matchLimit) |
1665 | { |
1666 | } |
1667 | |
1668 | private: |
1669 | BytecodePattern* pattern; |
1670 | bool unicode; |
1671 | unsigned* output; |
1672 | InputStream input; |
1673 | WTF::BumpPointerPool* allocatorPool { nullptr }; |
1674 | unsigned startOffset; |
1675 | unsigned remainingMatchCount; |
1676 | }; |
1677 | |
1678 | class ByteCompiler { |
1679 | struct ParenthesesStackEntry { |
1680 | unsigned beginTerm; |
1681 | unsigned savedAlternativeIndex; |
1682 | ParenthesesStackEntry(unsigned beginTerm, unsigned savedAlternativeIndex/*, unsigned subpatternId, bool capture = false*/) |
1683 | : beginTerm(beginTerm) |
1684 | , savedAlternativeIndex(savedAlternativeIndex) |
1685 | { |
1686 | } |
1687 | }; |
1688 | |
1689 | public: |
1690 | ByteCompiler(YarrPattern& pattern) |
1691 | : m_pattern(pattern) |
1692 | { |
1693 | } |
1694 | |
1695 | std::unique_ptr<BytecodePattern> compile(BumpPointerAllocator* allocator, ConcurrentJSLock* lock, ErrorCode& errorCode) |
1696 | { |
1697 | regexBegin(m_pattern.m_numSubpatterns, m_pattern.m_body->m_callFrameSize, m_pattern.m_body->m_alternatives[0]->onceThrough()); |
1698 | if (auto error = emitDisjunction(m_pattern.m_body, 0, 0)) { |
1699 | errorCode = error.value(); |
1700 | return nullptr; |
1701 | } |
1702 | regexEnd(); |
1703 | |
1704 | #ifndef NDEBUG |
1705 | if (Options::dumpCompiledRegExpPatterns()) |
1706 | dumpDisjunction(m_bodyDisjunction.get()); |
1707 | #endif |
1708 | |
1709 | return std::make_unique<BytecodePattern>(WTFMove(m_bodyDisjunction), m_allParenthesesInfo, m_pattern, allocator, lock); |
1710 | } |
1711 | |
1712 | void checkInput(unsigned count) |
1713 | { |
1714 | m_bodyDisjunction->terms.append(ByteTerm::CheckInput(count)); |
1715 | } |
1716 | |
1717 | void uncheckInput(unsigned count) |
1718 | { |
1719 | m_bodyDisjunction->terms.append(ByteTerm::UncheckInput(count)); |
1720 | } |
1721 | |
1722 | void assertionBOL(unsigned inputPosition) |
1723 | { |
1724 | m_bodyDisjunction->terms.append(ByteTerm::BOL(inputPosition)); |
1725 | } |
1726 | |
1727 | void assertionEOL(unsigned inputPosition) |
1728 | { |
1729 | m_bodyDisjunction->terms.append(ByteTerm::EOL(inputPosition)); |
1730 | } |
1731 | |
1732 | void assertionWordBoundary(bool invert, unsigned inputPosition) |
1733 | { |
1734 | m_bodyDisjunction->terms.append(ByteTerm::WordBoundary(invert, inputPosition)); |
1735 | } |
1736 | |
1737 | void atomPatternCharacter(UChar32 ch, unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1738 | { |
1739 | if (m_pattern.ignoreCase()) { |
1740 | UChar32 lo = u_tolower(ch); |
1741 | UChar32 hi = u_toupper(ch); |
1742 | |
1743 | if (lo != hi) { |
1744 | m_bodyDisjunction->terms.append(ByteTerm(lo, hi, inputPosition, frameLocation, quantityMaxCount, quantityType)); |
1745 | return; |
1746 | } |
1747 | } |
1748 | |
1749 | m_bodyDisjunction->terms.append(ByteTerm(ch, inputPosition, frameLocation, quantityMaxCount, quantityType)); |
1750 | } |
1751 | |
1752 | void atomCharacterClass(CharacterClass* characterClass, bool invert, unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1753 | { |
1754 | m_bodyDisjunction->terms.append(ByteTerm(characterClass, invert, inputPosition)); |
1755 | |
1756 | m_bodyDisjunction->terms.last().atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1757 | m_bodyDisjunction->terms.last().atom.quantityType = quantityType; |
1758 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1759 | } |
1760 | |
1761 | void atomBackReference(unsigned subpatternId, unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1762 | { |
1763 | ASSERT(subpatternId); |
1764 | |
1765 | m_bodyDisjunction->terms.append(ByteTerm::BackReference(subpatternId, inputPosition)); |
1766 | |
1767 | m_bodyDisjunction->terms.last().atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1768 | m_bodyDisjunction->terms.last().atom.quantityType = quantityType; |
1769 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1770 | } |
1771 | |
1772 | void atomParenthesesOnceBegin(unsigned subpatternId, bool capture, unsigned inputPosition, unsigned frameLocation, unsigned alternativeFrameLocation) |
1773 | { |
1774 | unsigned beginTerm = m_bodyDisjunction->terms.size(); |
1775 | |
1776 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternOnceBegin, subpatternId, capture, false, inputPosition)); |
1777 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1778 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeBegin()); |
1779 | m_bodyDisjunction->terms.last().frameLocation = alternativeFrameLocation; |
1780 | |
1781 | m_parenthesesStack.append(ParenthesesStackEntry(beginTerm, m_currentAlternativeIndex)); |
1782 | m_currentAlternativeIndex = beginTerm + 1; |
1783 | } |
1784 | |
1785 | void atomParenthesesTerminalBegin(unsigned subpatternId, bool capture, unsigned inputPosition, unsigned frameLocation, unsigned alternativeFrameLocation) |
1786 | { |
1787 | unsigned beginTerm = m_bodyDisjunction->terms.size(); |
1788 | |
1789 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternTerminalBegin, subpatternId, capture, false, inputPosition)); |
1790 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1791 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeBegin()); |
1792 | m_bodyDisjunction->terms.last().frameLocation = alternativeFrameLocation; |
1793 | |
1794 | m_parenthesesStack.append(ParenthesesStackEntry(beginTerm, m_currentAlternativeIndex)); |
1795 | m_currentAlternativeIndex = beginTerm + 1; |
1796 | } |
1797 | |
1798 | void atomParenthesesSubpatternBegin(unsigned subpatternId, bool capture, unsigned inputPosition, unsigned frameLocation, unsigned alternativeFrameLocation) |
1799 | { |
1800 | // Errrk! - this is a little crazy, we initially generate as a TypeParenthesesSubpatternOnceBegin, |
1801 | // then fix this up at the end! - simplifying this should make it much clearer. |
1802 | // https://bugs.webkit.org/show_bug.cgi?id=50136 |
1803 | |
1804 | unsigned beginTerm = m_bodyDisjunction->terms.size(); |
1805 | |
1806 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternOnceBegin, subpatternId, capture, false, inputPosition)); |
1807 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1808 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeBegin()); |
1809 | m_bodyDisjunction->terms.last().frameLocation = alternativeFrameLocation; |
1810 | |
1811 | m_parenthesesStack.append(ParenthesesStackEntry(beginTerm, m_currentAlternativeIndex)); |
1812 | m_currentAlternativeIndex = beginTerm + 1; |
1813 | } |
1814 | |
1815 | void atomParentheticalAssertionBegin(unsigned subpatternId, bool invert, unsigned frameLocation, unsigned alternativeFrameLocation) |
1816 | { |
1817 | unsigned beginTerm = m_bodyDisjunction->terms.size(); |
1818 | |
1819 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParentheticalAssertionBegin, subpatternId, false, invert, 0)); |
1820 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1821 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeBegin()); |
1822 | m_bodyDisjunction->terms.last().frameLocation = alternativeFrameLocation; |
1823 | |
1824 | m_parenthesesStack.append(ParenthesesStackEntry(beginTerm, m_currentAlternativeIndex)); |
1825 | m_currentAlternativeIndex = beginTerm + 1; |
1826 | } |
1827 | |
1828 | void atomParentheticalAssertionEnd(unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1829 | { |
1830 | unsigned beginTerm = popParenthesesStack(); |
1831 | closeAlternative(beginTerm + 1); |
1832 | unsigned endTerm = m_bodyDisjunction->terms.size(); |
1833 | |
1834 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeParentheticalAssertionBegin); |
1835 | |
1836 | bool invert = m_bodyDisjunction->terms[beginTerm].invert(); |
1837 | unsigned subpatternId = m_bodyDisjunction->terms[beginTerm].atom.subpatternId; |
1838 | |
1839 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParentheticalAssertionEnd, subpatternId, false, invert, inputPosition)); |
1840 | m_bodyDisjunction->terms[beginTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1841 | m_bodyDisjunction->terms[endTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1842 | m_bodyDisjunction->terms[endTerm].frameLocation = frameLocation; |
1843 | |
1844 | m_bodyDisjunction->terms[beginTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1845 | m_bodyDisjunction->terms[beginTerm].atom.quantityType = quantityType; |
1846 | m_bodyDisjunction->terms[endTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1847 | m_bodyDisjunction->terms[endTerm].atom.quantityType = quantityType; |
1848 | } |
1849 | |
1850 | void assertionDotStarEnclosure(bool bolAnchored, bool eolAnchored) |
1851 | { |
1852 | m_bodyDisjunction->terms.append(ByteTerm::DotStarEnclosure(bolAnchored, eolAnchored)); |
1853 | } |
1854 | |
1855 | unsigned popParenthesesStack() |
1856 | { |
1857 | ASSERT(m_parenthesesStack.size()); |
1858 | unsigned beginTerm = m_parenthesesStack.last().beginTerm; |
1859 | m_currentAlternativeIndex = m_parenthesesStack.last().savedAlternativeIndex; |
1860 | m_parenthesesStack.removeLast(); |
1861 | |
1862 | ASSERT(beginTerm < m_bodyDisjunction->terms.size()); |
1863 | ASSERT(m_currentAlternativeIndex < m_bodyDisjunction->terms.size()); |
1864 | |
1865 | return beginTerm; |
1866 | } |
1867 | |
1868 | void closeAlternative(unsigned beginTerm) |
1869 | { |
1870 | unsigned origBeginTerm = beginTerm; |
1871 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeAlternativeBegin); |
1872 | unsigned endIndex = m_bodyDisjunction->terms.size(); |
1873 | |
1874 | unsigned frameLocation = m_bodyDisjunction->terms[beginTerm].frameLocation; |
1875 | |
1876 | if (!m_bodyDisjunction->terms[beginTerm].alternative.next) |
1877 | m_bodyDisjunction->terms.remove(beginTerm); |
1878 | else { |
1879 | while (m_bodyDisjunction->terms[beginTerm].alternative.next) { |
1880 | beginTerm += m_bodyDisjunction->terms[beginTerm].alternative.next; |
1881 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeAlternativeDisjunction); |
1882 | m_bodyDisjunction->terms[beginTerm].alternative.end = endIndex - beginTerm; |
1883 | m_bodyDisjunction->terms[beginTerm].frameLocation = frameLocation; |
1884 | } |
1885 | |
1886 | m_bodyDisjunction->terms[beginTerm].alternative.next = origBeginTerm - beginTerm; |
1887 | |
1888 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeEnd()); |
1889 | m_bodyDisjunction->terms[endIndex].frameLocation = frameLocation; |
1890 | } |
1891 | } |
1892 | |
1893 | void closeBodyAlternative() |
1894 | { |
1895 | unsigned beginTerm = 0; |
1896 | unsigned origBeginTerm = 0; |
1897 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeBodyAlternativeBegin); |
1898 | unsigned endIndex = m_bodyDisjunction->terms.size(); |
1899 | |
1900 | unsigned frameLocation = m_bodyDisjunction->terms[beginTerm].frameLocation; |
1901 | |
1902 | while (m_bodyDisjunction->terms[beginTerm].alternative.next) { |
1903 | beginTerm += m_bodyDisjunction->terms[beginTerm].alternative.next; |
1904 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeBodyAlternativeDisjunction); |
1905 | m_bodyDisjunction->terms[beginTerm].alternative.end = endIndex - beginTerm; |
1906 | m_bodyDisjunction->terms[beginTerm].frameLocation = frameLocation; |
1907 | } |
1908 | |
1909 | m_bodyDisjunction->terms[beginTerm].alternative.next = origBeginTerm - beginTerm; |
1910 | |
1911 | m_bodyDisjunction->terms.append(ByteTerm::BodyAlternativeEnd()); |
1912 | m_bodyDisjunction->terms[endIndex].frameLocation = frameLocation; |
1913 | } |
1914 | |
1915 | void atomParenthesesSubpatternEnd(unsigned lastSubpatternId, unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMinCount, Checked<unsigned> quantityMaxCount, QuantifierType quantityType, unsigned callFrameSize = 0) |
1916 | { |
1917 | unsigned beginTerm = popParenthesesStack(); |
1918 | closeAlternative(beginTerm + 1); |
1919 | unsigned endTerm = m_bodyDisjunction->terms.size(); |
1920 | |
1921 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
1922 | |
1923 | ByteTerm& parenthesesBegin = m_bodyDisjunction->terms[beginTerm]; |
1924 | |
1925 | bool capture = parenthesesBegin.capture(); |
1926 | unsigned subpatternId = parenthesesBegin.atom.subpatternId; |
1927 | |
1928 | unsigned numSubpatterns = lastSubpatternId - subpatternId + 1; |
1929 | auto parenthesesDisjunction = std::make_unique<ByteDisjunction>(numSubpatterns, callFrameSize); |
1930 | |
1931 | unsigned firstTermInParentheses = beginTerm + 1; |
1932 | parenthesesDisjunction->terms.reserveInitialCapacity(endTerm - firstTermInParentheses + 2); |
1933 | |
1934 | parenthesesDisjunction->terms.append(ByteTerm::SubpatternBegin()); |
1935 | for (unsigned termInParentheses = firstTermInParentheses; termInParentheses < endTerm; ++termInParentheses) |
1936 | parenthesesDisjunction->terms.append(m_bodyDisjunction->terms[termInParentheses]); |
1937 | parenthesesDisjunction->terms.append(ByteTerm::SubpatternEnd()); |
1938 | |
1939 | m_bodyDisjunction->terms.shrink(beginTerm); |
1940 | |
1941 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpattern, subpatternId, parenthesesDisjunction.get(), capture, inputPosition)); |
1942 | m_allParenthesesInfo.append(WTFMove(parenthesesDisjunction)); |
1943 | |
1944 | m_bodyDisjunction->terms[beginTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1945 | m_bodyDisjunction->terms[beginTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1946 | m_bodyDisjunction->terms[beginTerm].atom.quantityType = quantityType; |
1947 | m_bodyDisjunction->terms[beginTerm].frameLocation = frameLocation; |
1948 | } |
1949 | |
1950 | void atomParenthesesOnceEnd(unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMinCount, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1951 | { |
1952 | unsigned beginTerm = popParenthesesStack(); |
1953 | closeAlternative(beginTerm + 1); |
1954 | unsigned endTerm = m_bodyDisjunction->terms.size(); |
1955 | |
1956 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
1957 | |
1958 | bool capture = m_bodyDisjunction->terms[beginTerm].capture(); |
1959 | unsigned subpatternId = m_bodyDisjunction->terms[beginTerm].atom.subpatternId; |
1960 | |
1961 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternOnceEnd, subpatternId, capture, false, inputPosition)); |
1962 | m_bodyDisjunction->terms[beginTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1963 | m_bodyDisjunction->terms[endTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1964 | m_bodyDisjunction->terms[endTerm].frameLocation = frameLocation; |
1965 | |
1966 | m_bodyDisjunction->terms[beginTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1967 | m_bodyDisjunction->terms[beginTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1968 | m_bodyDisjunction->terms[beginTerm].atom.quantityType = quantityType; |
1969 | m_bodyDisjunction->terms[endTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1970 | m_bodyDisjunction->terms[endTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1971 | m_bodyDisjunction->terms[endTerm].atom.quantityType = quantityType; |
1972 | } |
1973 | |
1974 | void atomParenthesesTerminalEnd(unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMinCount, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1975 | { |
1976 | unsigned beginTerm = popParenthesesStack(); |
1977 | closeAlternative(beginTerm + 1); |
1978 | unsigned endTerm = m_bodyDisjunction->terms.size(); |
1979 | |
1980 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeParenthesesSubpatternTerminalBegin); |
1981 | |
1982 | bool capture = m_bodyDisjunction->terms[beginTerm].capture(); |
1983 | unsigned subpatternId = m_bodyDisjunction->terms[beginTerm].atom.subpatternId; |
1984 | |
1985 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternTerminalEnd, subpatternId, capture, false, inputPosition)); |
1986 | m_bodyDisjunction->terms[beginTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1987 | m_bodyDisjunction->terms[endTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1988 | m_bodyDisjunction->terms[endTerm].frameLocation = frameLocation; |
1989 | |
1990 | m_bodyDisjunction->terms[beginTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1991 | m_bodyDisjunction->terms[beginTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1992 | m_bodyDisjunction->terms[beginTerm].atom.quantityType = quantityType; |
1993 | m_bodyDisjunction->terms[endTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1994 | m_bodyDisjunction->terms[endTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1995 | m_bodyDisjunction->terms[endTerm].atom.quantityType = quantityType; |
1996 | } |
1997 | |
1998 | void regexBegin(unsigned numSubpatterns, unsigned callFrameSize, bool onceThrough) |
1999 | { |
2000 | m_bodyDisjunction = std::make_unique<ByteDisjunction>(numSubpatterns, callFrameSize); |
2001 | m_bodyDisjunction->terms.append(ByteTerm::BodyAlternativeBegin(onceThrough)); |
2002 | m_bodyDisjunction->terms[0].frameLocation = 0; |
2003 | m_currentAlternativeIndex = 0; |
2004 | } |
2005 | |
2006 | void regexEnd() |
2007 | { |
2008 | closeBodyAlternative(); |
2009 | } |
2010 | |
2011 | void alternativeBodyDisjunction(bool onceThrough) |
2012 | { |
2013 | unsigned newAlternativeIndex = m_bodyDisjunction->terms.size(); |
2014 | m_bodyDisjunction->terms[m_currentAlternativeIndex].alternative.next = newAlternativeIndex - m_currentAlternativeIndex; |
2015 | m_bodyDisjunction->terms.append(ByteTerm::BodyAlternativeDisjunction(onceThrough)); |
2016 | |
2017 | m_currentAlternativeIndex = newAlternativeIndex; |
2018 | } |
2019 | |
2020 | void alternativeDisjunction() |
2021 | { |
2022 | unsigned newAlternativeIndex = m_bodyDisjunction->terms.size(); |
2023 | m_bodyDisjunction->terms[m_currentAlternativeIndex].alternative.next = newAlternativeIndex - m_currentAlternativeIndex; |
2024 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeDisjunction()); |
2025 | |
2026 | m_currentAlternativeIndex = newAlternativeIndex; |
2027 | } |
2028 | |
2029 | Optional<ErrorCode> WARN_UNUSED_RETURN emitDisjunction(PatternDisjunction* disjunction, Checked<unsigned, RecordOverflow> inputCountAlreadyChecked, unsigned parenthesesInputCountAlreadyChecked) |
2030 | { |
2031 | for (unsigned alt = 0; alt < disjunction->m_alternatives.size(); ++alt) { |
2032 | auto currentCountAlreadyChecked = inputCountAlreadyChecked; |
2033 | |
2034 | PatternAlternative* alternative = disjunction->m_alternatives[alt].get(); |
2035 | |
2036 | if (alt) { |
2037 | if (disjunction == m_pattern.m_body) |
2038 | alternativeBodyDisjunction(alternative->onceThrough()); |
2039 | else |
2040 | alternativeDisjunction(); |
2041 | } |
2042 | |
2043 | unsigned minimumSize = alternative->m_minimumSize; |
2044 | ASSERT(minimumSize >= parenthesesInputCountAlreadyChecked); |
2045 | unsigned countToCheck = minimumSize - parenthesesInputCountAlreadyChecked; |
2046 | |
2047 | if (countToCheck) { |
2048 | checkInput(countToCheck); |
2049 | currentCountAlreadyChecked += countToCheck; |
2050 | if (currentCountAlreadyChecked.hasOverflowed()) |
2051 | return ErrorCode::OffsetTooLarge; |
2052 | } |
2053 | |
2054 | for (auto& term : alternative->m_terms) { |
2055 | switch (term.type) { |
2056 | case PatternTerm::TypeAssertionBOL: |
2057 | assertionBOL((currentCountAlreadyChecked - term.inputPosition).unsafeGet()); |
2058 | break; |
2059 | |
2060 | case PatternTerm::TypeAssertionEOL: |
2061 | assertionEOL((currentCountAlreadyChecked - term.inputPosition).unsafeGet()); |
2062 | break; |
2063 | |
2064 | case PatternTerm::TypeAssertionWordBoundary: |
2065 | assertionWordBoundary(term.invert(), (currentCountAlreadyChecked - term.inputPosition).unsafeGet()); |
2066 | break; |
2067 | |
2068 | case PatternTerm::TypePatternCharacter: |
2069 | atomPatternCharacter(term.patternCharacter, (currentCountAlreadyChecked - term.inputPosition).unsafeGet(), term.frameLocation, term.quantityMaxCount, term.quantityType); |
2070 | break; |
2071 | |
2072 | case PatternTerm::TypeCharacterClass: |
2073 | atomCharacterClass(term.characterClass, term.invert(), (currentCountAlreadyChecked - term.inputPosition).unsafeGet(), term.frameLocation, term.quantityMaxCount, term.quantityType); |
2074 | break; |
2075 | |
2076 | case PatternTerm::TypeBackReference: |
2077 | atomBackReference(term.backReferenceSubpatternId, (currentCountAlreadyChecked - term.inputPosition).unsafeGet(), term.frameLocation, term.quantityMaxCount, term.quantityType); |
2078 | break; |
2079 | |
2080 | case PatternTerm::TypeForwardReference: |
2081 | break; |
2082 | |
2083 | case PatternTerm::TypeParenthesesSubpattern: { |
2084 | unsigned disjunctionAlreadyCheckedCount = 0; |
2085 | if (term.quantityMaxCount == 1 && !term.parentheses.isCopy) { |
2086 | unsigned alternativeFrameLocation = term.frameLocation; |
2087 | // For QuantifierFixedCount we pre-check the minimum size; for greedy/non-greedy we reserve a slot in the frame. |
2088 | if (term.quantityType == QuantifierFixedCount) |
2089 | disjunctionAlreadyCheckedCount = term.parentheses.disjunction->m_minimumSize; |
2090 | else |
2091 | alternativeFrameLocation += YarrStackSpaceForBackTrackInfoParenthesesOnce; |
2092 | unsigned delegateEndInputOffset = (currentCountAlreadyChecked - term.inputPosition).unsafeGet(); |
2093 | atomParenthesesOnceBegin(term.parentheses.subpatternId, term.capture(), disjunctionAlreadyCheckedCount + delegateEndInputOffset, term.frameLocation, alternativeFrameLocation); |
2094 | if (auto error = emitDisjunction(term.parentheses.disjunction, currentCountAlreadyChecked, disjunctionAlreadyCheckedCount)) |
2095 | return error; |
2096 | atomParenthesesOnceEnd(delegateEndInputOffset, term.frameLocation, term.quantityMinCount, term.quantityMaxCount, term.quantityType); |
2097 | } else if (term.parentheses.isTerminal) { |
2098 | unsigned delegateEndInputOffset = (currentCountAlreadyChecked - term.inputPosition).unsafeGet(); |
2099 | atomParenthesesTerminalBegin(term.parentheses.subpatternId, term.capture(), disjunctionAlreadyCheckedCount + delegateEndInputOffset, term.frameLocation, term.frameLocation + YarrStackSpaceForBackTrackInfoParenthesesTerminal); |
2100 | if (auto error = emitDisjunction(term.parentheses.disjunction, currentCountAlreadyChecked, disjunctionAlreadyCheckedCount)) |
2101 | return error; |
2102 | atomParenthesesTerminalEnd(delegateEndInputOffset, term.frameLocation, term.quantityMinCount, term.quantityMaxCount, term.quantityType); |
2103 | } else { |
2104 | unsigned delegateEndInputOffset = (currentCountAlreadyChecked - term.inputPosition).unsafeGet(); |
2105 | atomParenthesesSubpatternBegin(term.parentheses.subpatternId, term.capture(), disjunctionAlreadyCheckedCount + delegateEndInputOffset, term.frameLocation, 0); |
2106 | if (auto error = emitDisjunction(term.parentheses.disjunction, currentCountAlreadyChecked, 0)) |
2107 | return error; |
2108 | atomParenthesesSubpatternEnd(term.parentheses.lastSubpatternId, delegateEndInputOffset, term.frameLocation, term.quantityMinCount, term.quantityMaxCount, term.quantityType, term.parentheses.disjunction->m_callFrameSize); |
2109 | } |
2110 | break; |
2111 | } |
2112 | |
2113 | case PatternTerm::TypeParentheticalAssertion: { |
2114 | unsigned alternativeFrameLocation = term.frameLocation + YarrStackSpaceForBackTrackInfoParentheticalAssertion; |
2115 | unsigned positiveInputOffset = (currentCountAlreadyChecked - term.inputPosition).unsafeGet(); |
2116 | unsigned uncheckAmount = 0; |
2117 | if (positiveInputOffset > term.parentheses.disjunction->m_minimumSize) { |
2118 | uncheckAmount = positiveInputOffset - term.parentheses.disjunction->m_minimumSize; |
2119 | uncheckInput(uncheckAmount); |
2120 | currentCountAlreadyChecked -= uncheckAmount; |
2121 | if (currentCountAlreadyChecked.hasOverflowed()) |
2122 | return ErrorCode::OffsetTooLarge; |
2123 | } |
2124 | |
2125 | atomParentheticalAssertionBegin(term.parentheses.subpatternId, term.invert(), term.frameLocation, alternativeFrameLocation); |
2126 | if (auto error = emitDisjunction(term.parentheses.disjunction, currentCountAlreadyChecked, positiveInputOffset - uncheckAmount)) |
2127 | return error; |
2128 | atomParentheticalAssertionEnd(0, term.frameLocation, term.quantityMaxCount, term.quantityType); |
2129 | if (uncheckAmount) { |
2130 | checkInput(uncheckAmount); |
2131 | currentCountAlreadyChecked += uncheckAmount; |
2132 | if (currentCountAlreadyChecked.hasOverflowed()) |
2133 | return ErrorCode::OffsetTooLarge; |
2134 | } |
2135 | break; |
2136 | } |
2137 | |
2138 | case PatternTerm::TypeDotStarEnclosure: |
2139 | assertionDotStarEnclosure(term.anchors.bolAnchor, term.anchors.eolAnchor); |
2140 | break; |
2141 | } |
2142 | } |
2143 | } |
2144 | return WTF::nullopt; |
2145 | } |
2146 | #ifndef NDEBUG |
2147 | void dumpDisjunction(ByteDisjunction* disjunction, unsigned nesting = 0) |
2148 | { |
2149 | PrintStream& out = WTF::dataFile(); |
2150 | |
2151 | unsigned termIndexNest = 0; |
2152 | |
2153 | if (!nesting) { |
2154 | out.printf("ByteDisjunction(%p):\n" , disjunction); |
2155 | nesting = 1; |
2156 | } else { |
2157 | termIndexNest = nesting - 1; |
2158 | nesting = 2; |
2159 | } |
2160 | |
2161 | auto outputTermIndexAndNest = [&](size_t index, unsigned termNesting) { |
2162 | for (unsigned nestingDepth = 0; nestingDepth < termIndexNest; nestingDepth++) |
2163 | out.print(" " ); |
2164 | out.printf("%4zu" , index); |
2165 | for (unsigned nestingDepth = 0; nestingDepth < termNesting; nestingDepth++) |
2166 | out.print(" " ); |
2167 | }; |
2168 | |
2169 | auto dumpQuantity = [&](ByteTerm& term) { |
2170 | if (term.atom.quantityType == QuantifierFixedCount && term.atom.quantityMinCount == 1 && term.atom.quantityMaxCount == 1) |
2171 | return; |
2172 | |
2173 | out.print(" {" , term.atom.quantityMinCount); |
2174 | if (term.atom.quantityMinCount != term.atom.quantityMaxCount) { |
2175 | if (term.atom.quantityMaxCount == UINT_MAX) |
2176 | out.print(",inf" ); |
2177 | else |
2178 | out.print("," , term.atom.quantityMaxCount); |
2179 | } |
2180 | out.print("}" ); |
2181 | if (term.atom.quantityType == QuantifierGreedy) |
2182 | out.print(" greedy" ); |
2183 | else if (term.atom.quantityType == QuantifierNonGreedy) |
2184 | out.print(" non-greedy" ); |
2185 | }; |
2186 | |
2187 | auto dumpCaptured = [&](ByteTerm& term) { |
2188 | if (term.capture()) |
2189 | out.print(" captured (#" , term.atom.subpatternId, ")" ); |
2190 | }; |
2191 | |
2192 | auto dumpInverted = [&](ByteTerm& term) { |
2193 | if (term.invert()) |
2194 | out.print(" inverted" ); |
2195 | }; |
2196 | |
2197 | auto dumpInputPosition = [&](ByteTerm& term) { |
2198 | out.printf(" inputPosition %u" , term.inputPosition); |
2199 | }; |
2200 | |
2201 | auto dumpFrameLocation = [&](ByteTerm& term) { |
2202 | out.printf(" frameLocation %u" , term.frameLocation); |
2203 | }; |
2204 | |
2205 | auto dumpCharacter = [&](ByteTerm& term) { |
2206 | out.print(" " ); |
2207 | dumpUChar32(out, term.atom.patternCharacter); |
2208 | }; |
2209 | |
2210 | auto dumpCharClass = [&](ByteTerm& term) { |
2211 | out.print(" " ); |
2212 | dumpCharacterClass(out, &m_pattern, term.atom.characterClass); |
2213 | }; |
2214 | |
2215 | for (size_t idx = 0; idx < disjunction->terms.size(); ++idx) { |
2216 | ByteTerm term = disjunction->terms[idx]; |
2217 | |
2218 | bool outputNewline = true; |
2219 | |
2220 | switch (term.type) { |
2221 | case ByteTerm::TypeBodyAlternativeBegin: |
2222 | outputTermIndexAndNest(idx, nesting++); |
2223 | out.print("BodyAlternativeBegin" ); |
2224 | if (term.alternative.onceThrough) |
2225 | out.print(" onceThrough" ); |
2226 | dumpFrameLocation(term); |
2227 | break; |
2228 | case ByteTerm::TypeBodyAlternativeDisjunction: |
2229 | outputTermIndexAndNest(idx, nesting - 1); |
2230 | out.print("BodyAlternativeDisjunction" ); |
2231 | dumpFrameLocation(term); |
2232 | break; |
2233 | case ByteTerm::TypeBodyAlternativeEnd: |
2234 | outputTermIndexAndNest(idx, --nesting); |
2235 | out.print("BodyAlternativeEnd" ); |
2236 | dumpFrameLocation(term); |
2237 | break; |
2238 | case ByteTerm::TypeAlternativeBegin: |
2239 | outputTermIndexAndNest(idx, nesting++); |
2240 | out.print("AlternativeBegin" ); |
2241 | dumpFrameLocation(term); |
2242 | break; |
2243 | case ByteTerm::TypeAlternativeDisjunction: |
2244 | outputTermIndexAndNest(idx, nesting - 1); |
2245 | out.print("AlternativeDisjunction" ); |
2246 | dumpFrameLocation(term); |
2247 | break; |
2248 | case ByteTerm::TypeAlternativeEnd: |
2249 | outputTermIndexAndNest(idx, --nesting); |
2250 | out.print("AlternativeEnd" ); |
2251 | dumpFrameLocation(term); |
2252 | break; |
2253 | case ByteTerm::TypeSubpatternBegin: |
2254 | outputTermIndexAndNest(idx, nesting++); |
2255 | out.print("SubpatternBegin" ); |
2256 | break; |
2257 | case ByteTerm::TypeSubpatternEnd: |
2258 | outputTermIndexAndNest(idx, --nesting); |
2259 | out.print("SubpatternEnd" ); |
2260 | break; |
2261 | case ByteTerm::TypeAssertionBOL: |
2262 | outputTermIndexAndNest(idx, nesting); |
2263 | out.print("AssertionBOL" ); |
2264 | break; |
2265 | case ByteTerm::TypeAssertionEOL: |
2266 | outputTermIndexAndNest(idx, nesting); |
2267 | out.print("AssertionEOL" ); |
2268 | break; |
2269 | case ByteTerm::TypeAssertionWordBoundary: |
2270 | outputTermIndexAndNest(idx, nesting); |
2271 | out.print("AssertionWordBoundary" ); |
2272 | break; |
2273 | case ByteTerm::TypePatternCharacterOnce: |
2274 | outputTermIndexAndNest(idx, nesting); |
2275 | out.print("PatternCharacterOnce" ); |
2276 | dumpInverted(term); |
2277 | dumpInputPosition(term); |
2278 | dumpFrameLocation(term); |
2279 | dumpCharacter(term); |
2280 | dumpQuantity(term); |
2281 | break; |
2282 | case ByteTerm::TypePatternCharacterFixed: |
2283 | outputTermIndexAndNest(idx, nesting); |
2284 | out.print("PatternCharacterFixed" ); |
2285 | dumpInverted(term); |
2286 | dumpInputPosition(term); |
2287 | dumpFrameLocation(term); |
2288 | dumpCharacter(term); |
2289 | out.print(" {" , term.atom.quantityMinCount, "}" ); |
2290 | break; |
2291 | case ByteTerm::TypePatternCharacterGreedy: |
2292 | outputTermIndexAndNest(idx, nesting); |
2293 | out.print("PatternCharacterGreedy" ); |
2294 | dumpInverted(term); |
2295 | dumpInputPosition(term); |
2296 | dumpFrameLocation(term); |
2297 | dumpCharacter(term); |
2298 | dumpQuantity(term); |
2299 | break; |
2300 | case ByteTerm::TypePatternCharacterNonGreedy: |
2301 | outputTermIndexAndNest(idx, nesting); |
2302 | out.print("PatternCharacterNonGreedy" ); |
2303 | dumpInverted(term); |
2304 | dumpInputPosition(term); |
2305 | dumpFrameLocation(term); |
2306 | dumpCharacter(term); |
2307 | dumpQuantity(term); |
2308 | break; |
2309 | case ByteTerm::TypePatternCasedCharacterOnce: |
2310 | outputTermIndexAndNest(idx, nesting); |
2311 | out.print("PatternCasedCharacterOnce" ); |
2312 | break; |
2313 | case ByteTerm::TypePatternCasedCharacterFixed: |
2314 | outputTermIndexAndNest(idx, nesting); |
2315 | out.print("PatternCasedCharacterFixed" ); |
2316 | break; |
2317 | case ByteTerm::TypePatternCasedCharacterGreedy: |
2318 | outputTermIndexAndNest(idx, nesting); |
2319 | out.print("PatternCasedCharacterGreedy" ); |
2320 | break; |
2321 | case ByteTerm::TypePatternCasedCharacterNonGreedy: |
2322 | outputTermIndexAndNest(idx, nesting); |
2323 | out.print("PatternCasedCharacterNonGreedy" ); |
2324 | break; |
2325 | case ByteTerm::TypeCharacterClass: |
2326 | outputTermIndexAndNest(idx, nesting); |
2327 | out.print("CharacterClass" ); |
2328 | dumpInverted(term); |
2329 | dumpInputPosition(term); |
2330 | dumpFrameLocation(term); |
2331 | dumpCharClass(term); |
2332 | dumpQuantity(term); |
2333 | break; |
2334 | case ByteTerm::TypeBackReference: |
2335 | outputTermIndexAndNest(idx, nesting); |
2336 | out.print("BackReference #" , term.atom.subpatternId); |
2337 | dumpQuantity(term); |
2338 | break; |
2339 | case ByteTerm::TypeParenthesesSubpattern: |
2340 | outputTermIndexAndNest(idx, nesting); |
2341 | out.print("ParenthesesSubpattern" ); |
2342 | dumpCaptured(term); |
2343 | dumpInverted(term); |
2344 | dumpInputPosition(term); |
2345 | dumpFrameLocation(term); |
2346 | dumpQuantity(term); |
2347 | out.print("\n" ); |
2348 | outputNewline = false; |
2349 | dumpDisjunction(term.atom.parenthesesDisjunction, nesting); |
2350 | break; |
2351 | case ByteTerm::TypeParenthesesSubpatternOnceBegin: |
2352 | outputTermIndexAndNest(idx, nesting++); |
2353 | out.print("ParenthesesSubpatternOnceBegin" ); |
2354 | dumpCaptured(term); |
2355 | dumpInverted(term); |
2356 | dumpInputPosition(term); |
2357 | dumpFrameLocation(term); |
2358 | break; |
2359 | case ByteTerm::TypeParenthesesSubpatternOnceEnd: |
2360 | outputTermIndexAndNest(idx, --nesting); |
2361 | out.print("ParenthesesSubpatternOnceEnd" ); |
2362 | dumpFrameLocation(term); |
2363 | break; |
2364 | case ByteTerm::TypeParenthesesSubpatternTerminalBegin: |
2365 | outputTermIndexAndNest(idx, nesting++); |
2366 | out.print("ParenthesesSubpatternTerminalBegin" ); |
2367 | dumpInverted(term); |
2368 | dumpInputPosition(term); |
2369 | dumpFrameLocation(term); |
2370 | break; |
2371 | case ByteTerm::TypeParenthesesSubpatternTerminalEnd: |
2372 | outputTermIndexAndNest(idx, --nesting); |
2373 | out.print("ParenthesesSubpatternTerminalEnd" ); |
2374 | dumpFrameLocation(term); |
2375 | break; |
2376 | case ByteTerm::TypeParentheticalAssertionBegin: |
2377 | outputTermIndexAndNest(idx, nesting++); |
2378 | out.print("ParentheticalAssertionBegin" ); |
2379 | dumpInverted(term); |
2380 | dumpInputPosition(term); |
2381 | dumpFrameLocation(term); |
2382 | break; |
2383 | case ByteTerm::TypeParentheticalAssertionEnd: |
2384 | outputTermIndexAndNest(idx, --nesting); |
2385 | out.print("ParentheticalAssertionEnd" ); |
2386 | dumpFrameLocation(term); |
2387 | break; |
2388 | case ByteTerm::TypeCheckInput: |
2389 | outputTermIndexAndNest(idx, nesting); |
2390 | out.print("CheckInput " , term.checkInputCount); |
2391 | break; |
2392 | case ByteTerm::TypeUncheckInput: |
2393 | outputTermIndexAndNest(idx, nesting); |
2394 | out.print("UncheckInput " , term.checkInputCount); |
2395 | break; |
2396 | case ByteTerm::TypeDotStarEnclosure: |
2397 | outputTermIndexAndNest(idx, nesting); |
2398 | out.print("DotStarEnclosure" ); |
2399 | break; |
2400 | } |
2401 | if (outputNewline) |
2402 | out.print("\n" ); |
2403 | } |
2404 | } |
2405 | #endif |
2406 | |
2407 | private: |
2408 | YarrPattern& m_pattern; |
2409 | std::unique_ptr<ByteDisjunction> m_bodyDisjunction; |
2410 | unsigned m_currentAlternativeIndex { 0 }; |
2411 | Vector<ParenthesesStackEntry> m_parenthesesStack; |
2412 | Vector<std::unique_ptr<ByteDisjunction>> m_allParenthesesInfo; |
2413 | }; |
2414 | |
2415 | std::unique_ptr<BytecodePattern> byteCompile(YarrPattern& pattern, BumpPointerAllocator* allocator, ErrorCode& errorCode, ConcurrentJSLock* lock) |
2416 | { |
2417 | return ByteCompiler(pattern).compile(allocator, lock, errorCode); |
2418 | } |
2419 | |
2420 | unsigned interpret(BytecodePattern* bytecode, const String& input, unsigned start, unsigned* output) |
2421 | { |
2422 | SuperSamplerScope superSamplerScope(false); |
2423 | if (input.is8Bit()) |
2424 | return Interpreter<LChar>(bytecode, output, input.characters8(), input.length(), start).interpret(); |
2425 | return Interpreter<UChar>(bytecode, output, input.characters16(), input.length(), start).interpret(); |
2426 | } |
2427 | |
2428 | unsigned interpret(BytecodePattern* bytecode, const LChar* input, unsigned length, unsigned start, unsigned* output) |
2429 | { |
2430 | SuperSamplerScope superSamplerScope(false); |
2431 | return Interpreter<LChar>(bytecode, output, input, length, start).interpret(); |
2432 | } |
2433 | |
2434 | unsigned interpret(BytecodePattern* bytecode, const UChar* input, unsigned length, unsigned start, unsigned* output) |
2435 | { |
2436 | SuperSamplerScope superSamplerScope(false); |
2437 | return Interpreter<UChar>(bytecode, output, input, length, start).interpret(); |
2438 | } |
2439 | |
2440 | // These should be the same for both UChar & LChar. |
2441 | COMPILE_ASSERT(sizeof(BackTrackInfoPatternCharacter) == (YarrStackSpaceForBackTrackInfoPatternCharacter * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoPatternCharacter); |
2442 | COMPILE_ASSERT(sizeof(BackTrackInfoCharacterClass) == (YarrStackSpaceForBackTrackInfoCharacterClass * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoCharacterClass); |
2443 | COMPILE_ASSERT(sizeof(BackTrackInfoBackReference) == (YarrStackSpaceForBackTrackInfoBackReference * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoBackReference); |
2444 | COMPILE_ASSERT(sizeof(BackTrackInfoAlternative) == (YarrStackSpaceForBackTrackInfoAlternative * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoAlternative); |
2445 | COMPILE_ASSERT(sizeof(BackTrackInfoParentheticalAssertion) == (YarrStackSpaceForBackTrackInfoParentheticalAssertion * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoParentheticalAssertion); |
2446 | COMPILE_ASSERT(sizeof(BackTrackInfoParenthesesOnce) == (YarrStackSpaceForBackTrackInfoParenthesesOnce * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoParenthesesOnce); |
2447 | COMPILE_ASSERT(sizeof(Interpreter<UChar>::BackTrackInfoParentheses) <= (YarrStackSpaceForBackTrackInfoParentheses * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoParentheses); |
2448 | |
2449 | |
2450 | } } |
2451 | |