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/StackCheck.h> |
38 | #include <wtf/text/CString.h> |
39 | #include <wtf/text/WTFString.h> |
40 | |
41 | namespace JSC { namespace Yarr { |
42 | |
43 | template<typename CharType> |
44 | class Interpreter { |
45 | public: |
46 | struct ParenthesesDisjunctionContext; |
47 | |
48 | struct BackTrackInfoParentheses { |
49 | uintptr_t matchAmount; |
50 | ParenthesesDisjunctionContext* lastContext; |
51 | }; |
52 | |
53 | static inline void appendParenthesesDisjunctionContext(BackTrackInfoParentheses* backTrack, ParenthesesDisjunctionContext* context) |
54 | { |
55 | context->next = backTrack->lastContext; |
56 | backTrack->lastContext = context; |
57 | ++backTrack->matchAmount; |
58 | } |
59 | |
60 | static inline void popParenthesesDisjunctionContext(BackTrackInfoParentheses* backTrack) |
61 | { |
62 | RELEASE_ASSERT(backTrack->matchAmount); |
63 | RELEASE_ASSERT(backTrack->lastContext); |
64 | backTrack->lastContext = backTrack->lastContext->next; |
65 | --backTrack->matchAmount; |
66 | } |
67 | |
68 | struct DisjunctionContext |
69 | { |
70 | DisjunctionContext() = default; |
71 | |
72 | void* operator new(size_t, void* where) |
73 | { |
74 | return where; |
75 | } |
76 | |
77 | static size_t allocationSize(unsigned numberOfFrames) |
78 | { |
79 | static_assert(alignof(DisjunctionContext) <= sizeof(void*), "" ); |
80 | size_t rawSize = (sizeof(DisjunctionContext) - sizeof(uintptr_t) + Checked<size_t>(numberOfFrames) * sizeof(uintptr_t)).unsafeGet(); |
81 | size_t roundedSize = roundUpToMultipleOf<sizeof(void*)>(rawSize); |
82 | RELEASE_ASSERT(roundedSize >= rawSize); |
83 | return roundedSize; |
84 | } |
85 | |
86 | int term { 0 }; |
87 | unsigned matchBegin; |
88 | unsigned matchEnd; |
89 | uintptr_t frame[1]; |
90 | }; |
91 | |
92 | DisjunctionContext* allocDisjunctionContext(ByteDisjunction* disjunction) |
93 | { |
94 | size_t size = DisjunctionContext::allocationSize(disjunction->m_frameSize); |
95 | allocatorPool = allocatorPool->ensureCapacity(size); |
96 | RELEASE_ASSERT(allocatorPool); |
97 | return new (allocatorPool->alloc(size)) DisjunctionContext(); |
98 | } |
99 | |
100 | void freeDisjunctionContext(DisjunctionContext* context) |
101 | { |
102 | allocatorPool = allocatorPool->dealloc(context); |
103 | } |
104 | |
105 | struct ParenthesesDisjunctionContext |
106 | { |
107 | ParenthesesDisjunctionContext(unsigned* output, ByteTerm& term) |
108 | { |
109 | unsigned firstSubpatternId = term.atom.subpatternId; |
110 | unsigned numNestedSubpatterns = term.atom.parenthesesDisjunction->m_numSubpatterns; |
111 | |
112 | for (unsigned i = 0; i < (numNestedSubpatterns << 1); ++i) { |
113 | subpatternBackup[i] = output[(firstSubpatternId << 1) + i]; |
114 | output[(firstSubpatternId << 1) + i] = offsetNoMatch; |
115 | } |
116 | |
117 | new (getDisjunctionContext(term)) DisjunctionContext(); |
118 | } |
119 | |
120 | void* operator new(size_t, void* where) |
121 | { |
122 | return where; |
123 | } |
124 | |
125 | void restoreOutput(unsigned* output, unsigned firstSubpatternId, unsigned numNestedSubpatterns) |
126 | { |
127 | for (unsigned i = 0; i < (numNestedSubpatterns << 1); ++i) |
128 | output[(firstSubpatternId << 1) + i] = subpatternBackup[i]; |
129 | } |
130 | |
131 | DisjunctionContext* getDisjunctionContext(ByteTerm& term) |
132 | { |
133 | return bitwise_cast<DisjunctionContext*>(bitwise_cast<uintptr_t>(this) + allocationSize(term.atom.parenthesesDisjunction->m_numSubpatterns)); |
134 | } |
135 | |
136 | static size_t allocationSize(unsigned numberOfSubpatterns) |
137 | { |
138 | static_assert(alignof(ParenthesesDisjunctionContext) <= sizeof(void*), "" ); |
139 | size_t rawSize = (sizeof(ParenthesesDisjunctionContext) - sizeof(unsigned) + (Checked<size_t>(numberOfSubpatterns) * 2U) * sizeof(unsigned)).unsafeGet(); |
140 | size_t roundedSize = roundUpToMultipleOf<sizeof(void*)>(rawSize); |
141 | RELEASE_ASSERT(roundedSize >= rawSize); |
142 | return roundedSize; |
143 | } |
144 | |
145 | ParenthesesDisjunctionContext* next { nullptr }; |
146 | unsigned subpatternBackup[1]; |
147 | }; |
148 | |
149 | ParenthesesDisjunctionContext* allocParenthesesDisjunctionContext(ByteDisjunction* disjunction, unsigned* output, ByteTerm& term) |
150 | { |
151 | size_t size = (Checked<size_t>(ParenthesesDisjunctionContext::allocationSize(term.atom.parenthesesDisjunction->m_numSubpatterns)) + DisjunctionContext::allocationSize(disjunction->m_frameSize)).unsafeGet(); |
152 | allocatorPool = allocatorPool->ensureCapacity(size); |
153 | RELEASE_ASSERT(allocatorPool); |
154 | return new (allocatorPool->alloc(size)) ParenthesesDisjunctionContext(output, term); |
155 | } |
156 | |
157 | void freeParenthesesDisjunctionContext(ParenthesesDisjunctionContext* context) |
158 | { |
159 | allocatorPool = allocatorPool->dealloc(context); |
160 | } |
161 | |
162 | class InputStream { |
163 | public: |
164 | InputStream(const CharType* input, unsigned start, unsigned length, bool decodeSurrogatePairs) |
165 | : input(input) |
166 | , pos(start) |
167 | , length(length) |
168 | , decodeSurrogatePairs(decodeSurrogatePairs) |
169 | { |
170 | } |
171 | |
172 | void next() |
173 | { |
174 | ++pos; |
175 | } |
176 | |
177 | void rewind(unsigned amount) |
178 | { |
179 | ASSERT(pos >= amount); |
180 | pos -= amount; |
181 | } |
182 | |
183 | int read() |
184 | { |
185 | ASSERT(pos < length); |
186 | if (pos < length) |
187 | return input[pos]; |
188 | return -1; |
189 | } |
190 | |
191 | int readPair() |
192 | { |
193 | ASSERT(pos + 1 < length); |
194 | return input[pos] | input[pos + 1] << 16; |
195 | } |
196 | |
197 | int readChecked(unsigned negativePositionOffest) |
198 | { |
199 | RELEASE_ASSERT(pos >= negativePositionOffest); |
200 | unsigned p = pos - negativePositionOffest; |
201 | ASSERT(p < length); |
202 | int result = input[p]; |
203 | if (U16_IS_LEAD(result) && decodeSurrogatePairs && p + 1 < length && U16_IS_TRAIL(input[p + 1])) { |
204 | if (atEnd()) |
205 | return -1; |
206 | |
207 | result = U16_GET_SUPPLEMENTARY(result, input[p + 1]); |
208 | next(); |
209 | } |
210 | return result; |
211 | } |
212 | |
213 | int readSurrogatePairChecked(unsigned negativePositionOffset) |
214 | { |
215 | RELEASE_ASSERT(pos >= negativePositionOffset); |
216 | unsigned p = pos - negativePositionOffset; |
217 | ASSERT(p < length); |
218 | if (p + 1 >= length) |
219 | return -1; |
220 | |
221 | int first = input[p]; |
222 | int second = input[p + 1]; |
223 | if (U16_IS_LEAD(first) && U16_IS_TRAIL(second)) |
224 | return U16_GET_SUPPLEMENTARY(first, second); |
225 | |
226 | return -1; |
227 | } |
228 | |
229 | int reread(unsigned from) |
230 | { |
231 | ASSERT(from < length); |
232 | int result = input[from]; |
233 | if (U16_IS_LEAD(result) && decodeSurrogatePairs && from + 1 < length && U16_IS_TRAIL(input[from + 1])) |
234 | result = U16_GET_SUPPLEMENTARY(result, input[from + 1]); |
235 | return result; |
236 | } |
237 | |
238 | int prev() |
239 | { |
240 | ASSERT(!(pos > length)); |
241 | if (pos && length) |
242 | return input[pos - 1]; |
243 | return -1; |
244 | } |
245 | |
246 | unsigned getPos() |
247 | { |
248 | return pos; |
249 | } |
250 | |
251 | void setPos(unsigned p) |
252 | { |
253 | pos = p; |
254 | } |
255 | |
256 | bool atStart() |
257 | { |
258 | return pos == 0; |
259 | } |
260 | |
261 | bool atEnd() |
262 | { |
263 | return pos == length; |
264 | } |
265 | |
266 | unsigned end() |
267 | { |
268 | return length; |
269 | } |
270 | |
271 | bool checkInput(unsigned count) |
272 | { |
273 | if (((pos + count) <= length) && ((pos + count) >= pos)) { |
274 | pos += count; |
275 | return true; |
276 | } |
277 | return false; |
278 | } |
279 | |
280 | void uncheckInput(unsigned count) |
281 | { |
282 | RELEASE_ASSERT(pos >= count); |
283 | pos -= count; |
284 | } |
285 | |
286 | bool atStart(unsigned negativePositionOffset) |
287 | { |
288 | return pos == negativePositionOffset; |
289 | } |
290 | |
291 | bool atEnd(unsigned negativePositionOffest) |
292 | { |
293 | RELEASE_ASSERT(pos >= negativePositionOffest); |
294 | return (pos - negativePositionOffest) == length; |
295 | } |
296 | |
297 | bool isAvailableInput(unsigned offset) |
298 | { |
299 | return (((pos + offset) <= length) && ((pos + offset) >= pos)); |
300 | } |
301 | |
302 | private: |
303 | const CharType* input; |
304 | unsigned pos; |
305 | unsigned length; |
306 | bool decodeSurrogatePairs; |
307 | }; |
308 | |
309 | bool testCharacterClass(CharacterClass* characterClass, int ch) |
310 | { |
311 | auto linearSearchMatches = [&ch](const Vector<UChar32>& matches) { |
312 | for (unsigned i = 0; i < matches.size(); ++i) { |
313 | if (ch == matches[i]) |
314 | return true; |
315 | } |
316 | |
317 | return false; |
318 | }; |
319 | |
320 | auto binarySearchMatches = [&ch](const Vector<UChar32>& matches) { |
321 | size_t low = 0; |
322 | size_t high = matches.size() - 1; |
323 | |
324 | while (low <= high) { |
325 | size_t mid = low + (high - low) / 2; |
326 | int diff = ch - matches[mid]; |
327 | if (!diff) |
328 | return true; |
329 | |
330 | if (diff < 0) { |
331 | if (mid == low) |
332 | return false; |
333 | high = mid - 1; |
334 | } else |
335 | low = mid + 1; |
336 | } |
337 | return false; |
338 | }; |
339 | |
340 | auto linearSearchRanges = [&ch](const Vector<CharacterRange>& ranges) { |
341 | for (unsigned i = 0; i < ranges.size(); ++i) { |
342 | if ((ch >= ranges[i].begin) && (ch <= ranges[i].end)) |
343 | return true; |
344 | } |
345 | |
346 | return false; |
347 | }; |
348 | |
349 | auto binarySearchRanges = [&ch](const Vector<CharacterRange>& ranges) { |
350 | size_t low = 0; |
351 | size_t high = ranges.size() - 1; |
352 | |
353 | while (low <= high) { |
354 | size_t mid = low + (high - low) / 2; |
355 | int rangeBeginDiff = ch - ranges[mid].begin; |
356 | if (rangeBeginDiff >= 0 && ch <= ranges[mid].end) |
357 | return true; |
358 | |
359 | if (rangeBeginDiff < 0) { |
360 | if (mid == low) |
361 | return false; |
362 | high = mid - 1; |
363 | } else |
364 | low = mid + 1; |
365 | } |
366 | return false; |
367 | }; |
368 | |
369 | if (characterClass->m_anyCharacter) |
370 | return true; |
371 | |
372 | const size_t thresholdForBinarySearch = 6; |
373 | |
374 | if (!isASCII(ch)) { |
375 | if (characterClass->m_matchesUnicode.size()) { |
376 | if (characterClass->m_matchesUnicode.size() > thresholdForBinarySearch) { |
377 | if (binarySearchMatches(characterClass->m_matchesUnicode)) |
378 | return true; |
379 | } else if (linearSearchMatches(characterClass->m_matchesUnicode)) |
380 | return true; |
381 | } |
382 | |
383 | if (characterClass->m_rangesUnicode.size()) { |
384 | if (characterClass->m_rangesUnicode.size() > thresholdForBinarySearch) { |
385 | if (binarySearchRanges(characterClass->m_rangesUnicode)) |
386 | return true; |
387 | } else if (linearSearchRanges(characterClass->m_rangesUnicode)) |
388 | return true; |
389 | } |
390 | } else { |
391 | if (characterClass->m_matches.size()) { |
392 | if (characterClass->m_matches.size() > thresholdForBinarySearch) { |
393 | if (binarySearchMatches(characterClass->m_matches)) |
394 | return true; |
395 | } else if (linearSearchMatches(characterClass->m_matches)) |
396 | return true; |
397 | } |
398 | |
399 | if (characterClass->m_ranges.size()) { |
400 | if (characterClass->m_ranges.size() > thresholdForBinarySearch) { |
401 | if (binarySearchRanges(characterClass->m_ranges)) |
402 | return true; |
403 | } else if (linearSearchRanges(characterClass->m_ranges)) |
404 | return true; |
405 | } |
406 | } |
407 | |
408 | return false; |
409 | } |
410 | |
411 | bool checkCharacter(int testChar, unsigned negativeInputOffset) |
412 | { |
413 | return testChar == input.readChecked(negativeInputOffset); |
414 | } |
415 | |
416 | bool checkSurrogatePair(int testUnicodeChar, unsigned negativeInputOffset) |
417 | { |
418 | return testUnicodeChar == input.readSurrogatePairChecked(negativeInputOffset); |
419 | } |
420 | |
421 | bool checkCasedCharacter(int loChar, int hiChar, unsigned negativeInputOffset) |
422 | { |
423 | int ch = input.readChecked(negativeInputOffset); |
424 | return (loChar == ch) || (hiChar == ch); |
425 | } |
426 | |
427 | bool checkCharacterClass(CharacterClass* characterClass, bool invert, unsigned negativeInputOffset) |
428 | { |
429 | bool match = testCharacterClass(characterClass, input.readChecked(negativeInputOffset)); |
430 | return invert ? !match : match; |
431 | } |
432 | |
433 | bool checkCharacterClassDontAdvanceInputForNonBMP(CharacterClass* characterClass, unsigned negativeInputOffset) |
434 | { |
435 | int readCharacter = characterClass->hasOnlyNonBMPCharacters() ? input.readSurrogatePairChecked(negativeInputOffset) : input.readChecked(negativeInputOffset); |
436 | return testCharacterClass(characterClass, readCharacter); |
437 | } |
438 | |
439 | bool tryConsumeBackReference(int matchBegin, int matchEnd, unsigned negativeInputOffset) |
440 | { |
441 | unsigned matchSize = (unsigned)(matchEnd - matchBegin); |
442 | |
443 | if (!input.checkInput(matchSize)) |
444 | return false; |
445 | |
446 | for (unsigned i = 0; i < matchSize; ++i) { |
447 | int oldCh = input.reread(matchBegin + i); |
448 | int ch; |
449 | if (!U_IS_BMP(oldCh)) { |
450 | ch = input.readSurrogatePairChecked(negativeInputOffset + matchSize - i); |
451 | ++i; |
452 | } else |
453 | ch = input.readChecked(negativeInputOffset + matchSize - i); |
454 | |
455 | if (oldCh == ch) |
456 | continue; |
457 | |
458 | if (pattern->ignoreCase()) { |
459 | // See ES 6.0, 21.2.2.8.2 for the definition of Canonicalize(). For non-Unicode |
460 | // patterns, Unicode values are never allowed to match against ASCII ones. |
461 | // For Unicode, we need to check all canonical equivalents of a character. |
462 | if (!unicode && (isASCII(oldCh) || isASCII(ch))) { |
463 | if (toASCIIUpper(oldCh) == toASCIIUpper(ch)) |
464 | continue; |
465 | } else if (areCanonicallyEquivalent(oldCh, ch, unicode ? CanonicalMode::Unicode : CanonicalMode::UCS2)) |
466 | continue; |
467 | } |
468 | |
469 | input.uncheckInput(matchSize); |
470 | return false; |
471 | } |
472 | |
473 | return true; |
474 | } |
475 | |
476 | bool matchAssertionBOL(ByteTerm& term) |
477 | { |
478 | return (input.atStart(term.inputPosition)) || (pattern->multiline() && testCharacterClass(pattern->newlineCharacterClass, input.readChecked(term.inputPosition + 1))); |
479 | } |
480 | |
481 | bool matchAssertionEOL(ByteTerm& term) |
482 | { |
483 | if (term.inputPosition) |
484 | return (input.atEnd(term.inputPosition)) || (pattern->multiline() && testCharacterClass(pattern->newlineCharacterClass, input.readChecked(term.inputPosition))); |
485 | |
486 | return (input.atEnd()) || (pattern->multiline() && testCharacterClass(pattern->newlineCharacterClass, input.read())); |
487 | } |
488 | |
489 | bool matchAssertionWordBoundary(ByteTerm& term) |
490 | { |
491 | bool prevIsWordchar = !input.atStart(term.inputPosition) && testCharacterClass(pattern->wordcharCharacterClass, input.readChecked(term.inputPosition + 1)); |
492 | bool readIsWordchar; |
493 | if (term.inputPosition) |
494 | readIsWordchar = !input.atEnd(term.inputPosition) && testCharacterClass(pattern->wordcharCharacterClass, input.readChecked(term.inputPosition)); |
495 | else |
496 | readIsWordchar = !input.atEnd() && testCharacterClass(pattern->wordcharCharacterClass, input.read()); |
497 | |
498 | bool wordBoundary = prevIsWordchar != readIsWordchar; |
499 | return term.invert() ? !wordBoundary : wordBoundary; |
500 | } |
501 | |
502 | bool backtrackPatternCharacter(ByteTerm& term, DisjunctionContext* context) |
503 | { |
504 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + term.frameLocation); |
505 | |
506 | switch (term.atom.quantityType) { |
507 | case QuantifierFixedCount: |
508 | break; |
509 | |
510 | case QuantifierGreedy: |
511 | if (backTrack->matchAmount) { |
512 | --backTrack->matchAmount; |
513 | input.uncheckInput(U16_LENGTH(term.atom.patternCharacter)); |
514 | return true; |
515 | } |
516 | break; |
517 | |
518 | case QuantifierNonGreedy: |
519 | if ((backTrack->matchAmount < term.atom.quantityMaxCount) && input.checkInput(1)) { |
520 | ++backTrack->matchAmount; |
521 | if (checkCharacter(term.atom.patternCharacter, term.inputPosition + 1)) |
522 | return true; |
523 | } |
524 | input.setPos(backTrack->begin); |
525 | break; |
526 | } |
527 | |
528 | return false; |
529 | } |
530 | |
531 | bool backtrackPatternCasedCharacter(ByteTerm& term, DisjunctionContext* context) |
532 | { |
533 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + term.frameLocation); |
534 | |
535 | switch (term.atom.quantityType) { |
536 | case QuantifierFixedCount: |
537 | break; |
538 | |
539 | case QuantifierGreedy: |
540 | if (backTrack->matchAmount) { |
541 | --backTrack->matchAmount; |
542 | input.uncheckInput(1); |
543 | return true; |
544 | } |
545 | break; |
546 | |
547 | case QuantifierNonGreedy: |
548 | if ((backTrack->matchAmount < term.atom.quantityMaxCount) && input.checkInput(1)) { |
549 | ++backTrack->matchAmount; |
550 | if (checkCasedCharacter(term.atom.casedCharacter.lo, term.atom.casedCharacter.hi, term.inputPosition + 1)) |
551 | return true; |
552 | } |
553 | input.uncheckInput(backTrack->matchAmount); |
554 | break; |
555 | } |
556 | |
557 | return false; |
558 | } |
559 | |
560 | bool matchCharacterClass(ByteTerm& term, DisjunctionContext* context) |
561 | { |
562 | ASSERT(term.type == ByteTerm::TypeCharacterClass); |
563 | BackTrackInfoCharacterClass* backTrack = reinterpret_cast<BackTrackInfoCharacterClass*>(context->frame + term.frameLocation); |
564 | |
565 | switch (term.atom.quantityType) { |
566 | case QuantifierFixedCount: { |
567 | if (unicode) { |
568 | CharacterClass* charClass = term.atom.characterClass; |
569 | backTrack->begin = input.getPos(); |
570 | unsigned matchAmount = 0; |
571 | for (matchAmount = 0; matchAmount < term.atom.quantityMaxCount; ++matchAmount) { |
572 | if (term.invert()) { |
573 | if (!checkCharacterClass(charClass, term.invert(), term.inputPosition - matchAmount)) { |
574 | input.setPos(backTrack->begin); |
575 | return false; |
576 | } |
577 | } else { |
578 | unsigned matchOffset = matchAmount * (charClass->hasOnlyNonBMPCharacters() ? 2 : 1); |
579 | if (!checkCharacterClassDontAdvanceInputForNonBMP(charClass, term.inputPosition - matchOffset)) { |
580 | input.setPos(backTrack->begin); |
581 | return false; |
582 | } |
583 | } |
584 | } |
585 | |
586 | return true; |
587 | } |
588 | |
589 | for (unsigned matchAmount = 0; matchAmount < term.atom.quantityMaxCount; ++matchAmount) { |
590 | if (!checkCharacterClass(term.atom.characterClass, term.invert(), term.inputPosition - matchAmount)) |
591 | return false; |
592 | } |
593 | return true; |
594 | } |
595 | |
596 | case QuantifierGreedy: { |
597 | unsigned position = input.getPos(); |
598 | backTrack->begin = position; |
599 | unsigned matchAmount = 0; |
600 | while ((matchAmount < term.atom.quantityMaxCount) && input.checkInput(1)) { |
601 | if (!checkCharacterClass(term.atom.characterClass, term.invert(), term.inputPosition + 1)) { |
602 | input.setPos(position); |
603 | break; |
604 | } |
605 | ++matchAmount; |
606 | position = input.getPos(); |
607 | } |
608 | backTrack->matchAmount = matchAmount; |
609 | |
610 | return true; |
611 | } |
612 | |
613 | case QuantifierNonGreedy: |
614 | backTrack->begin = input.getPos(); |
615 | backTrack->matchAmount = 0; |
616 | return true; |
617 | } |
618 | |
619 | RELEASE_ASSERT_NOT_REACHED(); |
620 | return false; |
621 | } |
622 | |
623 | bool backtrackCharacterClass(ByteTerm& term, DisjunctionContext* context) |
624 | { |
625 | ASSERT(term.type == ByteTerm::TypeCharacterClass); |
626 | BackTrackInfoCharacterClass* backTrack = reinterpret_cast<BackTrackInfoCharacterClass*>(context->frame + term.frameLocation); |
627 | |
628 | switch (term.atom.quantityType) { |
629 | case QuantifierFixedCount: |
630 | if (unicode) |
631 | input.setPos(backTrack->begin); |
632 | break; |
633 | |
634 | case QuantifierGreedy: |
635 | if (backTrack->matchAmount) { |
636 | if (unicode) { |
637 | // Rematch one less match |
638 | input.setPos(backTrack->begin); |
639 | --backTrack->matchAmount; |
640 | for (unsigned matchAmount = 0; (matchAmount < backTrack->matchAmount) && input.checkInput(1); ++matchAmount) { |
641 | if (!checkCharacterClass(term.atom.characterClass, term.invert(), term.inputPosition + 1)) { |
642 | input.uncheckInput(1); |
643 | break; |
644 | } |
645 | } |
646 | return true; |
647 | } |
648 | --backTrack->matchAmount; |
649 | input.uncheckInput(1); |
650 | return true; |
651 | } |
652 | break; |
653 | |
654 | case QuantifierNonGreedy: |
655 | if ((backTrack->matchAmount < term.atom.quantityMaxCount) && input.checkInput(1)) { |
656 | ++backTrack->matchAmount; |
657 | if (checkCharacterClass(term.atom.characterClass, term.invert(), term.inputPosition + 1)) |
658 | return true; |
659 | } |
660 | input.setPos(backTrack->begin); |
661 | break; |
662 | } |
663 | |
664 | return false; |
665 | } |
666 | |
667 | bool matchBackReference(ByteTerm& term, DisjunctionContext* context) |
668 | { |
669 | ASSERT(term.type == ByteTerm::TypeBackReference); |
670 | BackTrackInfoBackReference* backTrack = reinterpret_cast<BackTrackInfoBackReference*>(context->frame + term.frameLocation); |
671 | |
672 | unsigned matchBegin = output[(term.atom.subpatternId << 1)]; |
673 | unsigned matchEnd = output[(term.atom.subpatternId << 1) + 1]; |
674 | |
675 | // If the end position of the referenced match hasn't set yet then the backreference in the same parentheses where it references to that. |
676 | // In this case the result of match is empty string like when it references to a parentheses with zero-width match. |
677 | // Eg.: /(a\1)/ |
678 | if (matchEnd == offsetNoMatch) |
679 | return true; |
680 | |
681 | if (matchBegin == offsetNoMatch) |
682 | return true; |
683 | |
684 | ASSERT(matchBegin <= matchEnd); |
685 | |
686 | if (matchBegin == matchEnd) |
687 | return true; |
688 | |
689 | switch (term.atom.quantityType) { |
690 | case QuantifierFixedCount: { |
691 | backTrack->begin = input.getPos(); |
692 | for (unsigned matchAmount = 0; matchAmount < term.atom.quantityMaxCount; ++matchAmount) { |
693 | if (!tryConsumeBackReference(matchBegin, matchEnd, term.inputPosition)) { |
694 | input.setPos(backTrack->begin); |
695 | return false; |
696 | } |
697 | } |
698 | return true; |
699 | } |
700 | |
701 | case QuantifierGreedy: { |
702 | unsigned matchAmount = 0; |
703 | while ((matchAmount < term.atom.quantityMaxCount) && tryConsumeBackReference(matchBegin, matchEnd, term.inputPosition)) |
704 | ++matchAmount; |
705 | backTrack->matchAmount = matchAmount; |
706 | return true; |
707 | } |
708 | |
709 | case QuantifierNonGreedy: |
710 | backTrack->begin = input.getPos(); |
711 | backTrack->matchAmount = 0; |
712 | return true; |
713 | } |
714 | |
715 | RELEASE_ASSERT_NOT_REACHED(); |
716 | return false; |
717 | } |
718 | |
719 | bool backtrackBackReference(ByteTerm& term, DisjunctionContext* context) |
720 | { |
721 | ASSERT(term.type == ByteTerm::TypeBackReference); |
722 | BackTrackInfoBackReference* backTrack = reinterpret_cast<BackTrackInfoBackReference*>(context->frame + term.frameLocation); |
723 | |
724 | unsigned matchBegin = output[(term.atom.subpatternId << 1)]; |
725 | unsigned matchEnd = output[(term.atom.subpatternId << 1) + 1]; |
726 | |
727 | if (matchBegin == offsetNoMatch) |
728 | return false; |
729 | |
730 | ASSERT(matchBegin <= matchEnd); |
731 | |
732 | if (matchBegin == matchEnd) |
733 | return false; |
734 | |
735 | switch (term.atom.quantityType) { |
736 | case QuantifierFixedCount: |
737 | // for quantityMaxCount == 1, could rewind. |
738 | input.setPos(backTrack->begin); |
739 | break; |
740 | |
741 | case QuantifierGreedy: |
742 | if (backTrack->matchAmount) { |
743 | --backTrack->matchAmount; |
744 | input.rewind(matchEnd - matchBegin); |
745 | return true; |
746 | } |
747 | break; |
748 | |
749 | case QuantifierNonGreedy: |
750 | if ((backTrack->matchAmount < term.atom.quantityMaxCount) && tryConsumeBackReference(matchBegin, matchEnd, term.inputPosition)) { |
751 | ++backTrack->matchAmount; |
752 | return true; |
753 | } |
754 | input.setPos(backTrack->begin); |
755 | break; |
756 | } |
757 | |
758 | return false; |
759 | } |
760 | |
761 | void recordParenthesesMatch(ByteTerm& term, ParenthesesDisjunctionContext* context) |
762 | { |
763 | if (term.capture()) { |
764 | unsigned subpatternId = term.atom.subpatternId; |
765 | output[(subpatternId << 1)] = context->getDisjunctionContext(term)->matchBegin - term.inputPosition; |
766 | output[(subpatternId << 1) + 1] = context->getDisjunctionContext(term)->matchEnd - term.inputPosition; |
767 | } |
768 | } |
769 | void resetMatches(ByteTerm& term, ParenthesesDisjunctionContext* context) |
770 | { |
771 | unsigned firstSubpatternId = term.atom.subpatternId; |
772 | unsigned count = term.atom.parenthesesDisjunction->m_numSubpatterns; |
773 | context->restoreOutput(output, firstSubpatternId, count); |
774 | } |
775 | JSRegExpResult parenthesesDoBacktrack(ByteTerm& term, BackTrackInfoParentheses* backTrack) |
776 | { |
777 | while (backTrack->matchAmount) { |
778 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
779 | |
780 | JSRegExpResult result = matchDisjunction(term.atom.parenthesesDisjunction, context->getDisjunctionContext(term), true); |
781 | if (result == JSRegExpMatch) |
782 | return JSRegExpMatch; |
783 | |
784 | resetMatches(term, context); |
785 | popParenthesesDisjunctionContext(backTrack); |
786 | freeParenthesesDisjunctionContext(context); |
787 | |
788 | if (result != JSRegExpNoMatch) |
789 | return result; |
790 | } |
791 | |
792 | return JSRegExpNoMatch; |
793 | } |
794 | |
795 | bool matchParenthesesOnceBegin(ByteTerm& term, DisjunctionContext* context) |
796 | { |
797 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
798 | ASSERT(term.atom.quantityMaxCount == 1); |
799 | |
800 | BackTrackInfoParenthesesOnce* backTrack = reinterpret_cast<BackTrackInfoParenthesesOnce*>(context->frame + term.frameLocation); |
801 | |
802 | switch (term.atom.quantityType) { |
803 | case QuantifierGreedy: { |
804 | // set this speculatively; if we get to the parens end this will be true. |
805 | backTrack->begin = input.getPos(); |
806 | break; |
807 | } |
808 | case QuantifierNonGreedy: { |
809 | backTrack->begin = notFound; |
810 | context->term += term.atom.parenthesesWidth; |
811 | return true; |
812 | } |
813 | case QuantifierFixedCount: |
814 | break; |
815 | } |
816 | |
817 | if (term.capture()) { |
818 | unsigned subpatternId = term.atom.subpatternId; |
819 | output[(subpatternId << 1)] = input.getPos() - term.inputPosition; |
820 | } |
821 | |
822 | return true; |
823 | } |
824 | |
825 | bool matchParenthesesOnceEnd(ByteTerm& term, DisjunctionContext* context) |
826 | { |
827 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternOnceEnd); |
828 | ASSERT(term.atom.quantityMaxCount == 1); |
829 | |
830 | if (term.capture()) { |
831 | unsigned subpatternId = term.atom.subpatternId; |
832 | output[(subpatternId << 1) + 1] = input.getPos() - term.inputPosition; |
833 | } |
834 | |
835 | if (term.atom.quantityType == QuantifierFixedCount) |
836 | return true; |
837 | |
838 | BackTrackInfoParenthesesOnce* backTrack = reinterpret_cast<BackTrackInfoParenthesesOnce*>(context->frame + term.frameLocation); |
839 | return backTrack->begin != input.getPos(); |
840 | } |
841 | |
842 | bool backtrackParenthesesOnceBegin(ByteTerm& term, DisjunctionContext* context) |
843 | { |
844 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
845 | ASSERT(term.atom.quantityMaxCount == 1); |
846 | |
847 | BackTrackInfoParenthesesOnce* backTrack = reinterpret_cast<BackTrackInfoParenthesesOnce*>(context->frame + term.frameLocation); |
848 | |
849 | if (term.capture()) { |
850 | unsigned subpatternId = term.atom.subpatternId; |
851 | output[(subpatternId << 1)] = offsetNoMatch; |
852 | output[(subpatternId << 1) + 1] = offsetNoMatch; |
853 | } |
854 | |
855 | switch (term.atom.quantityType) { |
856 | case QuantifierGreedy: |
857 | // if we backtrack to this point, there is another chance - try matching nothing. |
858 | ASSERT(backTrack->begin != notFound); |
859 | backTrack->begin = notFound; |
860 | context->term += term.atom.parenthesesWidth; |
861 | return true; |
862 | case QuantifierNonGreedy: |
863 | ASSERT(backTrack->begin != notFound); |
864 | FALLTHROUGH; |
865 | case QuantifierFixedCount: |
866 | break; |
867 | } |
868 | |
869 | return false; |
870 | } |
871 | |
872 | bool backtrackParenthesesOnceEnd(ByteTerm& term, DisjunctionContext* context) |
873 | { |
874 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternOnceEnd); |
875 | ASSERT(term.atom.quantityMaxCount == 1); |
876 | |
877 | BackTrackInfoParenthesesOnce* backTrack = reinterpret_cast<BackTrackInfoParenthesesOnce*>(context->frame + term.frameLocation); |
878 | |
879 | switch (term.atom.quantityType) { |
880 | case QuantifierGreedy: |
881 | if (backTrack->begin == notFound) { |
882 | context->term -= term.atom.parenthesesWidth; |
883 | return false; |
884 | } |
885 | FALLTHROUGH; |
886 | case QuantifierNonGreedy: |
887 | if (backTrack->begin == notFound) { |
888 | backTrack->begin = input.getPos(); |
889 | if (term.capture()) { |
890 | // Technically this access to inputPosition should be accessing the begin term's |
891 | // inputPosition, but for repeats other than fixed these values should be |
892 | // the same anyway! (We don't pre-check for greedy or non-greedy matches.) |
893 | ASSERT((&term - term.atom.parenthesesWidth)->type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
894 | ASSERT((&term - term.atom.parenthesesWidth)->inputPosition == term.inputPosition); |
895 | unsigned subpatternId = term.atom.subpatternId; |
896 | output[subpatternId << 1] = input.getPos() - term.inputPosition; |
897 | } |
898 | context->term -= term.atom.parenthesesWidth; |
899 | return true; |
900 | } |
901 | FALLTHROUGH; |
902 | case QuantifierFixedCount: |
903 | break; |
904 | } |
905 | |
906 | return false; |
907 | } |
908 | |
909 | bool matchParenthesesTerminalBegin(ByteTerm& term, DisjunctionContext* context) |
910 | { |
911 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternTerminalBegin); |
912 | ASSERT(term.atom.quantityType == QuantifierGreedy); |
913 | ASSERT(term.atom.quantityMaxCount == quantifyInfinite); |
914 | ASSERT(!term.capture()); |
915 | |
916 | BackTrackInfoParenthesesTerminal* backTrack = reinterpret_cast<BackTrackInfoParenthesesTerminal*>(context->frame + term.frameLocation); |
917 | backTrack->begin = input.getPos(); |
918 | return true; |
919 | } |
920 | |
921 | bool matchParenthesesTerminalEnd(ByteTerm& term, DisjunctionContext* context) |
922 | { |
923 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternTerminalEnd); |
924 | |
925 | BackTrackInfoParenthesesTerminal* backTrack = reinterpret_cast<BackTrackInfoParenthesesTerminal*>(context->frame + term.frameLocation); |
926 | // Empty match is a failed match. |
927 | if (backTrack->begin == input.getPos()) |
928 | return false; |
929 | |
930 | // Successful match! Okay, what's next? - loop around and try to match more! |
931 | context->term -= (term.atom.parenthesesWidth + 1); |
932 | return true; |
933 | } |
934 | |
935 | bool backtrackParenthesesTerminalBegin(ByteTerm& term, DisjunctionContext* context) |
936 | { |
937 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpatternTerminalBegin); |
938 | ASSERT(term.atom.quantityType == QuantifierGreedy); |
939 | ASSERT(term.atom.quantityMaxCount == quantifyInfinite); |
940 | ASSERT(!term.capture()); |
941 | |
942 | // If we backtrack to this point, we have failed to match this iteration of the parens. |
943 | // Since this is greedy / zero minimum a failed is also accepted as a match! |
944 | context->term += term.atom.parenthesesWidth; |
945 | return true; |
946 | } |
947 | |
948 | bool backtrackParenthesesTerminalEnd(ByteTerm&, DisjunctionContext*) |
949 | { |
950 | // 'Terminal' parentheses are at the end of the regex, and as such a match past end |
951 | // should always be returned as a successful match - we should never backtrack to here. |
952 | RELEASE_ASSERT_NOT_REACHED(); |
953 | return false; |
954 | } |
955 | |
956 | bool matchParentheticalAssertionBegin(ByteTerm& term, DisjunctionContext* context) |
957 | { |
958 | ASSERT(term.type == ByteTerm::TypeParentheticalAssertionBegin); |
959 | ASSERT(term.atom.quantityMaxCount == 1); |
960 | |
961 | BackTrackInfoParentheticalAssertion* backTrack = reinterpret_cast<BackTrackInfoParentheticalAssertion*>(context->frame + term.frameLocation); |
962 | |
963 | backTrack->begin = input.getPos(); |
964 | return true; |
965 | } |
966 | |
967 | bool matchParentheticalAssertionEnd(ByteTerm& term, DisjunctionContext* context) |
968 | { |
969 | ASSERT(term.type == ByteTerm::TypeParentheticalAssertionEnd); |
970 | ASSERT(term.atom.quantityMaxCount == 1); |
971 | |
972 | BackTrackInfoParentheticalAssertion* backTrack = reinterpret_cast<BackTrackInfoParentheticalAssertion*>(context->frame + term.frameLocation); |
973 | |
974 | input.setPos(backTrack->begin); |
975 | |
976 | // We've reached the end of the parens; if they are inverted, this is failure. |
977 | if (term.invert()) { |
978 | context->term -= term.atom.parenthesesWidth; |
979 | return false; |
980 | } |
981 | |
982 | return true; |
983 | } |
984 | |
985 | bool backtrackParentheticalAssertionBegin(ByteTerm& term, DisjunctionContext* context) |
986 | { |
987 | ASSERT(term.type == ByteTerm::TypeParentheticalAssertionBegin); |
988 | ASSERT(term.atom.quantityMaxCount == 1); |
989 | |
990 | // We've failed to match parens; if they are inverted, this is win! |
991 | if (term.invert()) { |
992 | context->term += term.atom.parenthesesWidth; |
993 | return true; |
994 | } |
995 | |
996 | return false; |
997 | } |
998 | |
999 | bool backtrackParentheticalAssertionEnd(ByteTerm& term, DisjunctionContext* context) |
1000 | { |
1001 | ASSERT(term.type == ByteTerm::TypeParentheticalAssertionEnd); |
1002 | ASSERT(term.atom.quantityMaxCount == 1); |
1003 | |
1004 | BackTrackInfoParentheticalAssertion* backTrack = reinterpret_cast<BackTrackInfoParentheticalAssertion*>(context->frame + term.frameLocation); |
1005 | |
1006 | input.setPos(backTrack->begin); |
1007 | |
1008 | context->term -= term.atom.parenthesesWidth; |
1009 | return false; |
1010 | } |
1011 | |
1012 | JSRegExpResult matchParentheses(ByteTerm& term, DisjunctionContext* context) |
1013 | { |
1014 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpattern); |
1015 | |
1016 | BackTrackInfoParentheses* backTrack = reinterpret_cast<BackTrackInfoParentheses*>(context->frame + term.frameLocation); |
1017 | ByteDisjunction* disjunctionBody = term.atom.parenthesesDisjunction; |
1018 | |
1019 | backTrack->matchAmount = 0; |
1020 | backTrack->lastContext = 0; |
1021 | |
1022 | ASSERT(term.atom.quantityType != QuantifierFixedCount || term.atom.quantityMinCount == term.atom.quantityMaxCount); |
1023 | |
1024 | unsigned minimumMatchCount = term.atom.quantityMinCount; |
1025 | JSRegExpResult fixedMatchResult; |
1026 | |
1027 | // Handle fixed matches and the minimum part of a variable length match. |
1028 | if (minimumMatchCount) { |
1029 | // While we haven't yet reached our fixed limit, |
1030 | while (backTrack->matchAmount < minimumMatchCount) { |
1031 | // Try to do a match, and it it succeeds, add it to the list. |
1032 | ParenthesesDisjunctionContext* context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1033 | fixedMatchResult = matchDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1034 | if (fixedMatchResult == JSRegExpMatch) |
1035 | appendParenthesesDisjunctionContext(backTrack, context); |
1036 | else { |
1037 | // The match failed; try to find an alternate point to carry on from. |
1038 | resetMatches(term, context); |
1039 | freeParenthesesDisjunctionContext(context); |
1040 | |
1041 | if (fixedMatchResult != JSRegExpNoMatch) |
1042 | return fixedMatchResult; |
1043 | JSRegExpResult backtrackResult = parenthesesDoBacktrack(term, backTrack); |
1044 | if (backtrackResult != JSRegExpMatch) |
1045 | return backtrackResult; |
1046 | } |
1047 | } |
1048 | |
1049 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1050 | recordParenthesesMatch(term, context); |
1051 | } |
1052 | |
1053 | switch (term.atom.quantityType) { |
1054 | case QuantifierFixedCount: { |
1055 | ASSERT(backTrack->matchAmount == term.atom.quantityMaxCount); |
1056 | return JSRegExpMatch; |
1057 | } |
1058 | |
1059 | case QuantifierGreedy: { |
1060 | while (backTrack->matchAmount < term.atom.quantityMaxCount) { |
1061 | ParenthesesDisjunctionContext* context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1062 | JSRegExpResult result = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1063 | if (result == JSRegExpMatch) |
1064 | appendParenthesesDisjunctionContext(backTrack, context); |
1065 | else { |
1066 | resetMatches(term, context); |
1067 | freeParenthesesDisjunctionContext(context); |
1068 | |
1069 | if (result != JSRegExpNoMatch) |
1070 | return result; |
1071 | |
1072 | break; |
1073 | } |
1074 | } |
1075 | |
1076 | if (backTrack->matchAmount) { |
1077 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1078 | recordParenthesesMatch(term, context); |
1079 | } |
1080 | return JSRegExpMatch; |
1081 | } |
1082 | |
1083 | case QuantifierNonGreedy: |
1084 | return JSRegExpMatch; |
1085 | } |
1086 | |
1087 | RELEASE_ASSERT_NOT_REACHED(); |
1088 | return JSRegExpErrorNoMatch; |
1089 | } |
1090 | |
1091 | // Rules for backtracking differ depending on whether this is greedy or non-greedy. |
1092 | // |
1093 | // Greedy matches never should try just adding more - you should already have done |
1094 | // the 'more' cases. Always backtrack, at least a leetle bit. However cases where |
1095 | // you backtrack an item off the list needs checking, since we'll never have matched |
1096 | // the one less case. Tracking forwards, still add as much as possible. |
1097 | // |
1098 | // Non-greedy, we've already done the one less case, so don't match on popping. |
1099 | // We haven't done the one more case, so always try to add that. |
1100 | // |
1101 | JSRegExpResult backtrackParentheses(ByteTerm& term, DisjunctionContext* context) |
1102 | { |
1103 | ASSERT(term.type == ByteTerm::TypeParenthesesSubpattern); |
1104 | |
1105 | BackTrackInfoParentheses* backTrack = reinterpret_cast<BackTrackInfoParentheses*>(context->frame + term.frameLocation); |
1106 | ByteDisjunction* disjunctionBody = term.atom.parenthesesDisjunction; |
1107 | |
1108 | switch (term.atom.quantityType) { |
1109 | case QuantifierFixedCount: { |
1110 | ASSERT(backTrack->matchAmount == term.atom.quantityMaxCount); |
1111 | |
1112 | ParenthesesDisjunctionContext* context = 0; |
1113 | JSRegExpResult result = parenthesesDoBacktrack(term, backTrack); |
1114 | |
1115 | if (result != JSRegExpMatch) |
1116 | return result; |
1117 | |
1118 | // While we haven't yet reached our fixed limit, |
1119 | while (backTrack->matchAmount < term.atom.quantityMaxCount) { |
1120 | // Try to do a match, and it it succeeds, add it to the list. |
1121 | context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1122 | result = matchDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1123 | |
1124 | if (result == JSRegExpMatch) |
1125 | appendParenthesesDisjunctionContext(backTrack, context); |
1126 | else { |
1127 | // The match failed; try to find an alternate point to carry on from. |
1128 | resetMatches(term, context); |
1129 | freeParenthesesDisjunctionContext(context); |
1130 | |
1131 | if (result != JSRegExpNoMatch) |
1132 | return result; |
1133 | JSRegExpResult backtrackResult = parenthesesDoBacktrack(term, backTrack); |
1134 | if (backtrackResult != JSRegExpMatch) |
1135 | return backtrackResult; |
1136 | } |
1137 | } |
1138 | |
1139 | ASSERT(backTrack->matchAmount == term.atom.quantityMaxCount); |
1140 | context = backTrack->lastContext; |
1141 | recordParenthesesMatch(term, context); |
1142 | return JSRegExpMatch; |
1143 | } |
1144 | |
1145 | case QuantifierGreedy: { |
1146 | if (!backTrack->matchAmount) |
1147 | return JSRegExpNoMatch; |
1148 | |
1149 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1150 | JSRegExpResult result = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term), true); |
1151 | if (result == JSRegExpMatch) { |
1152 | while (backTrack->matchAmount < term.atom.quantityMaxCount) { |
1153 | ParenthesesDisjunctionContext* context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1154 | JSRegExpResult parenthesesResult = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1155 | if (parenthesesResult == JSRegExpMatch) |
1156 | appendParenthesesDisjunctionContext(backTrack, context); |
1157 | else { |
1158 | resetMatches(term, context); |
1159 | freeParenthesesDisjunctionContext(context); |
1160 | |
1161 | if (parenthesesResult != JSRegExpNoMatch) |
1162 | return parenthesesResult; |
1163 | |
1164 | break; |
1165 | } |
1166 | } |
1167 | } else { |
1168 | resetMatches(term, context); |
1169 | popParenthesesDisjunctionContext(backTrack); |
1170 | freeParenthesesDisjunctionContext(context); |
1171 | |
1172 | if (result != JSRegExpNoMatch || backTrack->matchAmount < term.atom.quantityMinCount) |
1173 | return result; |
1174 | } |
1175 | |
1176 | if (backTrack->matchAmount) { |
1177 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1178 | recordParenthesesMatch(term, context); |
1179 | } |
1180 | return JSRegExpMatch; |
1181 | } |
1182 | |
1183 | case QuantifierNonGreedy: { |
1184 | // If we've not reached the limit, try to add one more match. |
1185 | if (backTrack->matchAmount < term.atom.quantityMaxCount) { |
1186 | ParenthesesDisjunctionContext* context = allocParenthesesDisjunctionContext(disjunctionBody, output, term); |
1187 | JSRegExpResult result = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term)); |
1188 | if (result == JSRegExpMatch) { |
1189 | appendParenthesesDisjunctionContext(backTrack, context); |
1190 | recordParenthesesMatch(term, context); |
1191 | return JSRegExpMatch; |
1192 | } |
1193 | |
1194 | resetMatches(term, context); |
1195 | freeParenthesesDisjunctionContext(context); |
1196 | |
1197 | if (result != JSRegExpNoMatch) |
1198 | return result; |
1199 | } |
1200 | |
1201 | // Nope - okay backtrack looking for an alternative. |
1202 | while (backTrack->matchAmount) { |
1203 | ParenthesesDisjunctionContext* context = backTrack->lastContext; |
1204 | JSRegExpResult result = matchNonZeroDisjunction(disjunctionBody, context->getDisjunctionContext(term), true); |
1205 | if (result == JSRegExpMatch) { |
1206 | // successful backtrack! we're back in the game! |
1207 | if (backTrack->matchAmount) { |
1208 | context = backTrack->lastContext; |
1209 | recordParenthesesMatch(term, context); |
1210 | } |
1211 | return JSRegExpMatch; |
1212 | } |
1213 | |
1214 | // pop a match off the stack |
1215 | resetMatches(term, context); |
1216 | popParenthesesDisjunctionContext(backTrack); |
1217 | freeParenthesesDisjunctionContext(context); |
1218 | |
1219 | if (result != JSRegExpNoMatch) |
1220 | return result; |
1221 | } |
1222 | |
1223 | return JSRegExpNoMatch; |
1224 | } |
1225 | } |
1226 | |
1227 | RELEASE_ASSERT_NOT_REACHED(); |
1228 | return JSRegExpErrorNoMatch; |
1229 | } |
1230 | |
1231 | bool matchDotStarEnclosure(ByteTerm& term, DisjunctionContext* context) |
1232 | { |
1233 | UNUSED_PARAM(term); |
1234 | |
1235 | if (pattern->dotAll()) { |
1236 | context->matchBegin = startOffset; |
1237 | context->matchEnd = input.end(); |
1238 | return true; |
1239 | } |
1240 | |
1241 | unsigned matchBegin = context->matchBegin; |
1242 | |
1243 | if (matchBegin > startOffset) { |
1244 | for (matchBegin--; true; matchBegin--) { |
1245 | if (testCharacterClass(pattern->newlineCharacterClass, input.reread(matchBegin))) { |
1246 | ++matchBegin; |
1247 | break; |
1248 | } |
1249 | |
1250 | if (matchBegin == startOffset) |
1251 | break; |
1252 | } |
1253 | } |
1254 | |
1255 | unsigned matchEnd = input.getPos(); |
1256 | |
1257 | for (; (matchEnd != input.end()) |
1258 | && (!testCharacterClass(pattern->newlineCharacterClass, input.reread(matchEnd))); matchEnd++) { } |
1259 | |
1260 | if (((matchBegin && term.anchors.m_bol) |
1261 | || ((matchEnd != input.end()) && term.anchors.m_eol)) |
1262 | && !pattern->multiline()) |
1263 | return false; |
1264 | |
1265 | context->matchBegin = matchBegin; |
1266 | context->matchEnd = matchEnd; |
1267 | return true; |
1268 | } |
1269 | |
1270 | #define MATCH_NEXT() { ++context->term; goto matchAgain; } |
1271 | #define BACKTRACK() { --context->term; goto backtrack; } |
1272 | #define currentTerm() (disjunction->terms[context->term]) |
1273 | JSRegExpResult matchDisjunction(ByteDisjunction* disjunction, DisjunctionContext* context, bool btrack = false) |
1274 | { |
1275 | if (!--remainingMatchCount) |
1276 | return JSRegExpErrorHitLimit; |
1277 | |
1278 | if (btrack) |
1279 | BACKTRACK(); |
1280 | |
1281 | context->matchBegin = input.getPos(); |
1282 | context->term = 0; |
1283 | |
1284 | matchAgain: |
1285 | ASSERT(context->term < static_cast<int>(disjunction->terms.size())); |
1286 | |
1287 | switch (currentTerm().type) { |
1288 | case ByteTerm::TypeSubpatternBegin: |
1289 | MATCH_NEXT(); |
1290 | case ByteTerm::TypeSubpatternEnd: |
1291 | context->matchEnd = input.getPos(); |
1292 | return JSRegExpMatch; |
1293 | |
1294 | case ByteTerm::TypeBodyAlternativeBegin: |
1295 | MATCH_NEXT(); |
1296 | case ByteTerm::TypeBodyAlternativeDisjunction: |
1297 | case ByteTerm::TypeBodyAlternativeEnd: |
1298 | context->matchEnd = input.getPos(); |
1299 | return JSRegExpMatch; |
1300 | |
1301 | case ByteTerm::TypeAlternativeBegin: |
1302 | MATCH_NEXT(); |
1303 | case ByteTerm::TypeAlternativeDisjunction: |
1304 | case ByteTerm::TypeAlternativeEnd: { |
1305 | int offset = currentTerm().alternative.end; |
1306 | BackTrackInfoAlternative* backTrack = reinterpret_cast<BackTrackInfoAlternative*>(context->frame + currentTerm().frameLocation); |
1307 | backTrack->offset = offset; |
1308 | context->term += offset; |
1309 | MATCH_NEXT(); |
1310 | } |
1311 | |
1312 | case ByteTerm::TypeAssertionBOL: |
1313 | if (matchAssertionBOL(currentTerm())) |
1314 | MATCH_NEXT(); |
1315 | BACKTRACK(); |
1316 | case ByteTerm::TypeAssertionEOL: |
1317 | if (matchAssertionEOL(currentTerm())) |
1318 | MATCH_NEXT(); |
1319 | BACKTRACK(); |
1320 | case ByteTerm::TypeAssertionWordBoundary: |
1321 | if (matchAssertionWordBoundary(currentTerm())) |
1322 | MATCH_NEXT(); |
1323 | BACKTRACK(); |
1324 | |
1325 | case ByteTerm::TypePatternCharacterOnce: |
1326 | case ByteTerm::TypePatternCharacterFixed: { |
1327 | if (unicode) { |
1328 | if (!U_IS_BMP(currentTerm().atom.patternCharacter)) { |
1329 | for (unsigned matchAmount = 0; matchAmount < currentTerm().atom.quantityMaxCount; ++matchAmount) { |
1330 | if (!checkSurrogatePair(currentTerm().atom.patternCharacter, currentTerm().inputPosition - 2 * matchAmount)) { |
1331 | BACKTRACK(); |
1332 | } |
1333 | } |
1334 | MATCH_NEXT(); |
1335 | } |
1336 | } |
1337 | unsigned position = input.getPos(); // May need to back out reading a surrogate pair. |
1338 | |
1339 | for (unsigned matchAmount = 0; matchAmount < currentTerm().atom.quantityMaxCount; ++matchAmount) { |
1340 | if (!checkCharacter(currentTerm().atom.patternCharacter, currentTerm().inputPosition - matchAmount)) { |
1341 | input.setPos(position); |
1342 | BACKTRACK(); |
1343 | } |
1344 | } |
1345 | MATCH_NEXT(); |
1346 | } |
1347 | case ByteTerm::TypePatternCharacterGreedy: { |
1348 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + currentTerm().frameLocation); |
1349 | unsigned matchAmount = 0; |
1350 | unsigned position = input.getPos(); // May need to back out reading a surrogate pair. |
1351 | while ((matchAmount < currentTerm().atom.quantityMaxCount) && input.checkInput(1)) { |
1352 | if (!checkCharacter(currentTerm().atom.patternCharacter, currentTerm().inputPosition + 1)) { |
1353 | input.setPos(position); |
1354 | break; |
1355 | } |
1356 | ++matchAmount; |
1357 | position = input.getPos(); |
1358 | } |
1359 | backTrack->matchAmount = matchAmount; |
1360 | |
1361 | MATCH_NEXT(); |
1362 | } |
1363 | case ByteTerm::TypePatternCharacterNonGreedy: { |
1364 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + currentTerm().frameLocation); |
1365 | backTrack->begin = input.getPos(); |
1366 | backTrack->matchAmount = 0; |
1367 | MATCH_NEXT(); |
1368 | } |
1369 | |
1370 | case ByteTerm::TypePatternCasedCharacterOnce: |
1371 | case ByteTerm::TypePatternCasedCharacterFixed: { |
1372 | if (unicode) { |
1373 | // Case insensitive matching of unicode characters is handled as TypeCharacterClass. |
1374 | ASSERT(U_IS_BMP(currentTerm().atom.patternCharacter)); |
1375 | |
1376 | unsigned position = input.getPos(); // May need to back out reading a surrogate pair. |
1377 | |
1378 | for (unsigned matchAmount = 0; matchAmount < currentTerm().atom.quantityMaxCount; ++matchAmount) { |
1379 | if (!checkCasedCharacter(currentTerm().atom.casedCharacter.lo, currentTerm().atom.casedCharacter.hi, currentTerm().inputPosition - matchAmount)) { |
1380 | input.setPos(position); |
1381 | BACKTRACK(); |
1382 | } |
1383 | } |
1384 | MATCH_NEXT(); |
1385 | } |
1386 | |
1387 | for (unsigned matchAmount = 0; matchAmount < currentTerm().atom.quantityMaxCount; ++matchAmount) { |
1388 | if (!checkCasedCharacter(currentTerm().atom.casedCharacter.lo, currentTerm().atom.casedCharacter.hi, currentTerm().inputPosition - matchAmount)) |
1389 | BACKTRACK(); |
1390 | } |
1391 | MATCH_NEXT(); |
1392 | } |
1393 | case ByteTerm::TypePatternCasedCharacterGreedy: { |
1394 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + currentTerm().frameLocation); |
1395 | |
1396 | // Case insensitive matching of unicode characters is handled as TypeCharacterClass. |
1397 | ASSERT(!unicode || U_IS_BMP(currentTerm().atom.patternCharacter)); |
1398 | |
1399 | unsigned matchAmount = 0; |
1400 | while ((matchAmount < currentTerm().atom.quantityMaxCount) && input.checkInput(1)) { |
1401 | if (!checkCasedCharacter(currentTerm().atom.casedCharacter.lo, currentTerm().atom.casedCharacter.hi, currentTerm().inputPosition + 1)) { |
1402 | input.uncheckInput(1); |
1403 | break; |
1404 | } |
1405 | ++matchAmount; |
1406 | } |
1407 | backTrack->matchAmount = matchAmount; |
1408 | |
1409 | MATCH_NEXT(); |
1410 | } |
1411 | case ByteTerm::TypePatternCasedCharacterNonGreedy: { |
1412 | BackTrackInfoPatternCharacter* backTrack = reinterpret_cast<BackTrackInfoPatternCharacter*>(context->frame + currentTerm().frameLocation); |
1413 | |
1414 | // Case insensitive matching of unicode characters is handled as TypeCharacterClass. |
1415 | ASSERT(!unicode || U_IS_BMP(currentTerm().atom.patternCharacter)); |
1416 | |
1417 | backTrack->matchAmount = 0; |
1418 | MATCH_NEXT(); |
1419 | } |
1420 | |
1421 | case ByteTerm::TypeCharacterClass: |
1422 | if (matchCharacterClass(currentTerm(), context)) |
1423 | MATCH_NEXT(); |
1424 | BACKTRACK(); |
1425 | case ByteTerm::TypeBackReference: |
1426 | if (matchBackReference(currentTerm(), context)) |
1427 | MATCH_NEXT(); |
1428 | BACKTRACK(); |
1429 | case ByteTerm::TypeParenthesesSubpattern: { |
1430 | JSRegExpResult result = matchParentheses(currentTerm(), context); |
1431 | |
1432 | if (result == JSRegExpMatch) { |
1433 | MATCH_NEXT(); |
1434 | } else if (result != JSRegExpNoMatch) |
1435 | return result; |
1436 | |
1437 | BACKTRACK(); |
1438 | } |
1439 | case ByteTerm::TypeParenthesesSubpatternOnceBegin: |
1440 | if (matchParenthesesOnceBegin(currentTerm(), context)) |
1441 | MATCH_NEXT(); |
1442 | BACKTRACK(); |
1443 | case ByteTerm::TypeParenthesesSubpatternOnceEnd: |
1444 | if (matchParenthesesOnceEnd(currentTerm(), context)) |
1445 | MATCH_NEXT(); |
1446 | BACKTRACK(); |
1447 | case ByteTerm::TypeParenthesesSubpatternTerminalBegin: |
1448 | if (matchParenthesesTerminalBegin(currentTerm(), context)) |
1449 | MATCH_NEXT(); |
1450 | BACKTRACK(); |
1451 | case ByteTerm::TypeParenthesesSubpatternTerminalEnd: |
1452 | if (matchParenthesesTerminalEnd(currentTerm(), context)) |
1453 | MATCH_NEXT(); |
1454 | BACKTRACK(); |
1455 | case ByteTerm::TypeParentheticalAssertionBegin: |
1456 | if (matchParentheticalAssertionBegin(currentTerm(), context)) |
1457 | MATCH_NEXT(); |
1458 | BACKTRACK(); |
1459 | case ByteTerm::TypeParentheticalAssertionEnd: |
1460 | if (matchParentheticalAssertionEnd(currentTerm(), context)) |
1461 | MATCH_NEXT(); |
1462 | BACKTRACK(); |
1463 | |
1464 | case ByteTerm::TypeCheckInput: |
1465 | if (input.checkInput(currentTerm().checkInputCount)) |
1466 | MATCH_NEXT(); |
1467 | BACKTRACK(); |
1468 | |
1469 | case ByteTerm::TypeUncheckInput: |
1470 | input.uncheckInput(currentTerm().checkInputCount); |
1471 | MATCH_NEXT(); |
1472 | |
1473 | case ByteTerm::TypeDotStarEnclosure: |
1474 | if (matchDotStarEnclosure(currentTerm(), context)) |
1475 | return JSRegExpMatch; |
1476 | BACKTRACK(); |
1477 | } |
1478 | |
1479 | // We should never fall-through to here. |
1480 | RELEASE_ASSERT_NOT_REACHED(); |
1481 | |
1482 | backtrack: |
1483 | ASSERT(context->term < static_cast<int>(disjunction->terms.size())); |
1484 | |
1485 | switch (currentTerm().type) { |
1486 | case ByteTerm::TypeSubpatternBegin: |
1487 | return JSRegExpNoMatch; |
1488 | case ByteTerm::TypeSubpatternEnd: |
1489 | RELEASE_ASSERT_NOT_REACHED(); |
1490 | |
1491 | case ByteTerm::TypeBodyAlternativeBegin: |
1492 | case ByteTerm::TypeBodyAlternativeDisjunction: { |
1493 | int offset = currentTerm().alternative.next; |
1494 | context->term += offset; |
1495 | if (offset > 0) |
1496 | MATCH_NEXT(); |
1497 | |
1498 | if (input.atEnd() || pattern->sticky()) |
1499 | return JSRegExpNoMatch; |
1500 | |
1501 | input.next(); |
1502 | |
1503 | context->matchBegin = input.getPos(); |
1504 | |
1505 | if (currentTerm().alternative.onceThrough) |
1506 | context->term += currentTerm().alternative.next; |
1507 | |
1508 | MATCH_NEXT(); |
1509 | } |
1510 | case ByteTerm::TypeBodyAlternativeEnd: |
1511 | RELEASE_ASSERT_NOT_REACHED(); |
1512 | |
1513 | case ByteTerm::TypeAlternativeBegin: |
1514 | case ByteTerm::TypeAlternativeDisjunction: { |
1515 | int offset = currentTerm().alternative.next; |
1516 | context->term += offset; |
1517 | if (offset > 0) |
1518 | MATCH_NEXT(); |
1519 | BACKTRACK(); |
1520 | } |
1521 | case ByteTerm::TypeAlternativeEnd: { |
1522 | // We should never backtrack back into an alternative of the main body of the regex. |
1523 | BackTrackInfoAlternative* backTrack = reinterpret_cast<BackTrackInfoAlternative*>(context->frame + currentTerm().frameLocation); |
1524 | unsigned offset = backTrack->offset; |
1525 | context->term -= offset; |
1526 | BACKTRACK(); |
1527 | } |
1528 | |
1529 | case ByteTerm::TypeAssertionBOL: |
1530 | case ByteTerm::TypeAssertionEOL: |
1531 | case ByteTerm::TypeAssertionWordBoundary: |
1532 | BACKTRACK(); |
1533 | |
1534 | case ByteTerm::TypePatternCharacterOnce: |
1535 | case ByteTerm::TypePatternCharacterFixed: |
1536 | case ByteTerm::TypePatternCharacterGreedy: |
1537 | case ByteTerm::TypePatternCharacterNonGreedy: |
1538 | if (backtrackPatternCharacter(currentTerm(), context)) |
1539 | MATCH_NEXT(); |
1540 | BACKTRACK(); |
1541 | case ByteTerm::TypePatternCasedCharacterOnce: |
1542 | case ByteTerm::TypePatternCasedCharacterFixed: |
1543 | case ByteTerm::TypePatternCasedCharacterGreedy: |
1544 | case ByteTerm::TypePatternCasedCharacterNonGreedy: |
1545 | if (backtrackPatternCasedCharacter(currentTerm(), context)) |
1546 | MATCH_NEXT(); |
1547 | BACKTRACK(); |
1548 | case ByteTerm::TypeCharacterClass: |
1549 | if (backtrackCharacterClass(currentTerm(), context)) |
1550 | MATCH_NEXT(); |
1551 | BACKTRACK(); |
1552 | case ByteTerm::TypeBackReference: |
1553 | if (backtrackBackReference(currentTerm(), context)) |
1554 | MATCH_NEXT(); |
1555 | BACKTRACK(); |
1556 | case ByteTerm::TypeParenthesesSubpattern: { |
1557 | JSRegExpResult result = backtrackParentheses(currentTerm(), context); |
1558 | |
1559 | if (result == JSRegExpMatch) { |
1560 | MATCH_NEXT(); |
1561 | } else if (result != JSRegExpNoMatch) |
1562 | return result; |
1563 | |
1564 | BACKTRACK(); |
1565 | } |
1566 | case ByteTerm::TypeParenthesesSubpatternOnceBegin: |
1567 | if (backtrackParenthesesOnceBegin(currentTerm(), context)) |
1568 | MATCH_NEXT(); |
1569 | BACKTRACK(); |
1570 | case ByteTerm::TypeParenthesesSubpatternOnceEnd: |
1571 | if (backtrackParenthesesOnceEnd(currentTerm(), context)) |
1572 | MATCH_NEXT(); |
1573 | BACKTRACK(); |
1574 | case ByteTerm::TypeParenthesesSubpatternTerminalBegin: |
1575 | if (backtrackParenthesesTerminalBegin(currentTerm(), context)) |
1576 | MATCH_NEXT(); |
1577 | BACKTRACK(); |
1578 | case ByteTerm::TypeParenthesesSubpatternTerminalEnd: |
1579 | if (backtrackParenthesesTerminalEnd(currentTerm(), context)) |
1580 | MATCH_NEXT(); |
1581 | BACKTRACK(); |
1582 | case ByteTerm::TypeParentheticalAssertionBegin: |
1583 | if (backtrackParentheticalAssertionBegin(currentTerm(), context)) |
1584 | MATCH_NEXT(); |
1585 | BACKTRACK(); |
1586 | case ByteTerm::TypeParentheticalAssertionEnd: |
1587 | if (backtrackParentheticalAssertionEnd(currentTerm(), context)) |
1588 | MATCH_NEXT(); |
1589 | BACKTRACK(); |
1590 | |
1591 | case ByteTerm::TypeCheckInput: |
1592 | input.uncheckInput(currentTerm().checkInputCount); |
1593 | BACKTRACK(); |
1594 | |
1595 | case ByteTerm::TypeUncheckInput: |
1596 | input.checkInput(currentTerm().checkInputCount); |
1597 | BACKTRACK(); |
1598 | |
1599 | case ByteTerm::TypeDotStarEnclosure: |
1600 | RELEASE_ASSERT_NOT_REACHED(); |
1601 | } |
1602 | |
1603 | RELEASE_ASSERT_NOT_REACHED(); |
1604 | return JSRegExpErrorNoMatch; |
1605 | } |
1606 | |
1607 | JSRegExpResult matchNonZeroDisjunction(ByteDisjunction* disjunction, DisjunctionContext* context, bool btrack = false) |
1608 | { |
1609 | JSRegExpResult result = matchDisjunction(disjunction, context, btrack); |
1610 | |
1611 | if (result == JSRegExpMatch) { |
1612 | while (context->matchBegin == context->matchEnd) { |
1613 | result = matchDisjunction(disjunction, context, true); |
1614 | if (result != JSRegExpMatch) |
1615 | return result; |
1616 | } |
1617 | return JSRegExpMatch; |
1618 | } |
1619 | |
1620 | return result; |
1621 | } |
1622 | |
1623 | unsigned interpret() |
1624 | { |
1625 | // FIXME: https://bugs.webkit.org/show_bug.cgi?id=195970 |
1626 | // [Yarr Interpreter] The interpreter doesn't have checks for stack overflow due to deep recursion |
1627 | if (!input.isAvailableInput(0)) |
1628 | return offsetNoMatch; |
1629 | |
1630 | if (pattern->m_lock) |
1631 | pattern->m_lock->lock(); |
1632 | |
1633 | for (unsigned i = 0; i < pattern->m_body->m_numSubpatterns + 1; ++i) |
1634 | output[i << 1] = offsetNoMatch; |
1635 | |
1636 | allocatorPool = pattern->m_allocator->startAllocator(); |
1637 | RELEASE_ASSERT(allocatorPool); |
1638 | |
1639 | DisjunctionContext* context = allocDisjunctionContext(pattern->m_body.get()); |
1640 | |
1641 | JSRegExpResult result = matchDisjunction(pattern->m_body.get(), context, false); |
1642 | if (result == JSRegExpMatch) { |
1643 | output[0] = context->matchBegin; |
1644 | output[1] = context->matchEnd; |
1645 | } |
1646 | |
1647 | freeDisjunctionContext(context); |
1648 | |
1649 | pattern->m_allocator->stopAllocator(); |
1650 | |
1651 | ASSERT((result == JSRegExpMatch) == (output[0] != offsetNoMatch)); |
1652 | |
1653 | if (pattern->m_lock) |
1654 | pattern->m_lock->unlock(); |
1655 | |
1656 | return output[0]; |
1657 | } |
1658 | |
1659 | Interpreter(BytecodePattern* pattern, unsigned* output, const CharType* input, unsigned length, unsigned start) |
1660 | : pattern(pattern) |
1661 | , unicode(pattern->unicode()) |
1662 | , output(output) |
1663 | , input(input, start, length, pattern->unicode()) |
1664 | , startOffset(start) |
1665 | , remainingMatchCount(matchLimit) |
1666 | { |
1667 | } |
1668 | |
1669 | private: |
1670 | BytecodePattern* pattern; |
1671 | bool unicode; |
1672 | unsigned* output; |
1673 | InputStream input; |
1674 | WTF::BumpPointerPool* allocatorPool { nullptr }; |
1675 | unsigned startOffset; |
1676 | unsigned remainingMatchCount; |
1677 | }; |
1678 | |
1679 | class ByteCompiler { |
1680 | struct ParenthesesStackEntry { |
1681 | unsigned beginTerm; |
1682 | unsigned savedAlternativeIndex; |
1683 | ParenthesesStackEntry(unsigned beginTerm, unsigned savedAlternativeIndex/*, unsigned subpatternId, bool capture = false*/) |
1684 | : beginTerm(beginTerm) |
1685 | , savedAlternativeIndex(savedAlternativeIndex) |
1686 | { |
1687 | } |
1688 | }; |
1689 | |
1690 | public: |
1691 | ByteCompiler(YarrPattern& pattern) |
1692 | : m_pattern(pattern) |
1693 | { |
1694 | } |
1695 | |
1696 | std::unique_ptr<BytecodePattern> compile(BumpPointerAllocator* allocator, ConcurrentJSLock* lock, ErrorCode& errorCode) |
1697 | { |
1698 | if (UNLIKELY(!isSafeToRecurse())) { |
1699 | errorCode = ErrorCode::TooManyDisjunctions; |
1700 | return nullptr; |
1701 | } |
1702 | |
1703 | regexBegin(m_pattern.m_numSubpatterns, m_pattern.m_body->m_callFrameSize, m_pattern.m_body->m_alternatives[0]->onceThrough()); |
1704 | if (auto error = emitDisjunction(m_pattern.m_body, 0, 0)) { |
1705 | errorCode = error.value(); |
1706 | return nullptr; |
1707 | } |
1708 | regexEnd(); |
1709 | |
1710 | #ifndef NDEBUG |
1711 | if (Options::dumpCompiledRegExpPatterns()) |
1712 | dumpDisjunction(m_bodyDisjunction.get()); |
1713 | #endif |
1714 | |
1715 | return makeUnique<BytecodePattern>(WTFMove(m_bodyDisjunction), m_allParenthesesInfo, m_pattern, allocator, lock); |
1716 | } |
1717 | |
1718 | void checkInput(unsigned count) |
1719 | { |
1720 | m_bodyDisjunction->terms.append(ByteTerm::CheckInput(count)); |
1721 | } |
1722 | |
1723 | void uncheckInput(unsigned count) |
1724 | { |
1725 | m_bodyDisjunction->terms.append(ByteTerm::UncheckInput(count)); |
1726 | } |
1727 | |
1728 | void assertionBOL(unsigned inputPosition) |
1729 | { |
1730 | m_bodyDisjunction->terms.append(ByteTerm::BOL(inputPosition)); |
1731 | } |
1732 | |
1733 | void assertionEOL(unsigned inputPosition) |
1734 | { |
1735 | m_bodyDisjunction->terms.append(ByteTerm::EOL(inputPosition)); |
1736 | } |
1737 | |
1738 | void assertionWordBoundary(bool invert, unsigned inputPosition) |
1739 | { |
1740 | m_bodyDisjunction->terms.append(ByteTerm::WordBoundary(invert, inputPosition)); |
1741 | } |
1742 | |
1743 | void atomPatternCharacter(UChar32 ch, unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1744 | { |
1745 | if (m_pattern.ignoreCase()) { |
1746 | UChar32 lo = u_tolower(ch); |
1747 | UChar32 hi = u_toupper(ch); |
1748 | |
1749 | if (lo != hi) { |
1750 | m_bodyDisjunction->terms.append(ByteTerm(lo, hi, inputPosition, frameLocation, quantityMaxCount, quantityType)); |
1751 | return; |
1752 | } |
1753 | } |
1754 | |
1755 | m_bodyDisjunction->terms.append(ByteTerm(ch, inputPosition, frameLocation, quantityMaxCount, quantityType)); |
1756 | } |
1757 | |
1758 | void atomCharacterClass(CharacterClass* characterClass, bool invert, unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1759 | { |
1760 | m_bodyDisjunction->terms.append(ByteTerm(characterClass, invert, inputPosition)); |
1761 | |
1762 | m_bodyDisjunction->terms.last().atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1763 | m_bodyDisjunction->terms.last().atom.quantityType = quantityType; |
1764 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1765 | } |
1766 | |
1767 | void atomBackReference(unsigned subpatternId, unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1768 | { |
1769 | ASSERT(subpatternId); |
1770 | |
1771 | m_bodyDisjunction->terms.append(ByteTerm::BackReference(subpatternId, inputPosition)); |
1772 | |
1773 | m_bodyDisjunction->terms.last().atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1774 | m_bodyDisjunction->terms.last().atom.quantityType = quantityType; |
1775 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1776 | } |
1777 | |
1778 | void atomParenthesesOnceBegin(unsigned subpatternId, bool capture, unsigned inputPosition, unsigned frameLocation, unsigned alternativeFrameLocation) |
1779 | { |
1780 | unsigned beginTerm = m_bodyDisjunction->terms.size(); |
1781 | |
1782 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternOnceBegin, subpatternId, capture, false, inputPosition)); |
1783 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1784 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeBegin()); |
1785 | m_bodyDisjunction->terms.last().frameLocation = alternativeFrameLocation; |
1786 | |
1787 | m_parenthesesStack.append(ParenthesesStackEntry(beginTerm, m_currentAlternativeIndex)); |
1788 | m_currentAlternativeIndex = beginTerm + 1; |
1789 | } |
1790 | |
1791 | void atomParenthesesTerminalBegin(unsigned subpatternId, bool capture, unsigned inputPosition, unsigned frameLocation, unsigned alternativeFrameLocation) |
1792 | { |
1793 | unsigned beginTerm = m_bodyDisjunction->terms.size(); |
1794 | |
1795 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternTerminalBegin, subpatternId, capture, false, inputPosition)); |
1796 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1797 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeBegin()); |
1798 | m_bodyDisjunction->terms.last().frameLocation = alternativeFrameLocation; |
1799 | |
1800 | m_parenthesesStack.append(ParenthesesStackEntry(beginTerm, m_currentAlternativeIndex)); |
1801 | m_currentAlternativeIndex = beginTerm + 1; |
1802 | } |
1803 | |
1804 | void atomParenthesesSubpatternBegin(unsigned subpatternId, bool capture, unsigned inputPosition, unsigned frameLocation, unsigned alternativeFrameLocation) |
1805 | { |
1806 | // Errrk! - this is a little crazy, we initially generate as a TypeParenthesesSubpatternOnceBegin, |
1807 | // then fix this up at the end! - simplifying this should make it much clearer. |
1808 | // https://bugs.webkit.org/show_bug.cgi?id=50136 |
1809 | |
1810 | unsigned beginTerm = m_bodyDisjunction->terms.size(); |
1811 | |
1812 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternOnceBegin, subpatternId, capture, false, inputPosition)); |
1813 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1814 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeBegin()); |
1815 | m_bodyDisjunction->terms.last().frameLocation = alternativeFrameLocation; |
1816 | |
1817 | m_parenthesesStack.append(ParenthesesStackEntry(beginTerm, m_currentAlternativeIndex)); |
1818 | m_currentAlternativeIndex = beginTerm + 1; |
1819 | } |
1820 | |
1821 | void atomParentheticalAssertionBegin(unsigned subpatternId, bool invert, unsigned frameLocation, unsigned alternativeFrameLocation) |
1822 | { |
1823 | unsigned beginTerm = m_bodyDisjunction->terms.size(); |
1824 | |
1825 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParentheticalAssertionBegin, subpatternId, false, invert, 0)); |
1826 | m_bodyDisjunction->terms.last().frameLocation = frameLocation; |
1827 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeBegin()); |
1828 | m_bodyDisjunction->terms.last().frameLocation = alternativeFrameLocation; |
1829 | |
1830 | m_parenthesesStack.append(ParenthesesStackEntry(beginTerm, m_currentAlternativeIndex)); |
1831 | m_currentAlternativeIndex = beginTerm + 1; |
1832 | } |
1833 | |
1834 | void atomParentheticalAssertionEnd(unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1835 | { |
1836 | unsigned beginTerm = popParenthesesStack(); |
1837 | closeAlternative(beginTerm + 1); |
1838 | unsigned endTerm = m_bodyDisjunction->terms.size(); |
1839 | |
1840 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeParentheticalAssertionBegin); |
1841 | |
1842 | bool invert = m_bodyDisjunction->terms[beginTerm].invert(); |
1843 | unsigned subpatternId = m_bodyDisjunction->terms[beginTerm].atom.subpatternId; |
1844 | |
1845 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParentheticalAssertionEnd, subpatternId, false, invert, inputPosition)); |
1846 | m_bodyDisjunction->terms[beginTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1847 | m_bodyDisjunction->terms[endTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1848 | m_bodyDisjunction->terms[endTerm].frameLocation = frameLocation; |
1849 | |
1850 | m_bodyDisjunction->terms[beginTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1851 | m_bodyDisjunction->terms[beginTerm].atom.quantityType = quantityType; |
1852 | m_bodyDisjunction->terms[endTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1853 | m_bodyDisjunction->terms[endTerm].atom.quantityType = quantityType; |
1854 | } |
1855 | |
1856 | void assertionDotStarEnclosure(bool bolAnchored, bool eolAnchored) |
1857 | { |
1858 | m_bodyDisjunction->terms.append(ByteTerm::DotStarEnclosure(bolAnchored, eolAnchored)); |
1859 | } |
1860 | |
1861 | unsigned popParenthesesStack() |
1862 | { |
1863 | ASSERT(m_parenthesesStack.size()); |
1864 | unsigned beginTerm = m_parenthesesStack.last().beginTerm; |
1865 | m_currentAlternativeIndex = m_parenthesesStack.last().savedAlternativeIndex; |
1866 | m_parenthesesStack.removeLast(); |
1867 | |
1868 | ASSERT(beginTerm < m_bodyDisjunction->terms.size()); |
1869 | ASSERT(m_currentAlternativeIndex < m_bodyDisjunction->terms.size()); |
1870 | |
1871 | return beginTerm; |
1872 | } |
1873 | |
1874 | void closeAlternative(unsigned beginTerm) |
1875 | { |
1876 | unsigned origBeginTerm = beginTerm; |
1877 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeAlternativeBegin); |
1878 | unsigned endIndex = m_bodyDisjunction->terms.size(); |
1879 | |
1880 | unsigned frameLocation = m_bodyDisjunction->terms[beginTerm].frameLocation; |
1881 | |
1882 | if (!m_bodyDisjunction->terms[beginTerm].alternative.next) |
1883 | m_bodyDisjunction->terms.remove(beginTerm); |
1884 | else { |
1885 | while (m_bodyDisjunction->terms[beginTerm].alternative.next) { |
1886 | beginTerm += m_bodyDisjunction->terms[beginTerm].alternative.next; |
1887 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeAlternativeDisjunction); |
1888 | m_bodyDisjunction->terms[beginTerm].alternative.end = endIndex - beginTerm; |
1889 | m_bodyDisjunction->terms[beginTerm].frameLocation = frameLocation; |
1890 | } |
1891 | |
1892 | m_bodyDisjunction->terms[beginTerm].alternative.next = origBeginTerm - beginTerm; |
1893 | |
1894 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeEnd()); |
1895 | m_bodyDisjunction->terms[endIndex].frameLocation = frameLocation; |
1896 | } |
1897 | } |
1898 | |
1899 | void closeBodyAlternative() |
1900 | { |
1901 | unsigned beginTerm = 0; |
1902 | unsigned origBeginTerm = 0; |
1903 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeBodyAlternativeBegin); |
1904 | unsigned endIndex = m_bodyDisjunction->terms.size(); |
1905 | |
1906 | unsigned frameLocation = m_bodyDisjunction->terms[beginTerm].frameLocation; |
1907 | |
1908 | while (m_bodyDisjunction->terms[beginTerm].alternative.next) { |
1909 | beginTerm += m_bodyDisjunction->terms[beginTerm].alternative.next; |
1910 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeBodyAlternativeDisjunction); |
1911 | m_bodyDisjunction->terms[beginTerm].alternative.end = endIndex - beginTerm; |
1912 | m_bodyDisjunction->terms[beginTerm].frameLocation = frameLocation; |
1913 | } |
1914 | |
1915 | m_bodyDisjunction->terms[beginTerm].alternative.next = origBeginTerm - beginTerm; |
1916 | |
1917 | m_bodyDisjunction->terms.append(ByteTerm::BodyAlternativeEnd()); |
1918 | m_bodyDisjunction->terms[endIndex].frameLocation = frameLocation; |
1919 | } |
1920 | |
1921 | void atomParenthesesSubpatternEnd(unsigned lastSubpatternId, unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMinCount, Checked<unsigned> quantityMaxCount, QuantifierType quantityType, unsigned callFrameSize = 0) |
1922 | { |
1923 | unsigned beginTerm = popParenthesesStack(); |
1924 | closeAlternative(beginTerm + 1); |
1925 | unsigned endTerm = m_bodyDisjunction->terms.size(); |
1926 | |
1927 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
1928 | |
1929 | ByteTerm& parenthesesBegin = m_bodyDisjunction->terms[beginTerm]; |
1930 | |
1931 | bool capture = parenthesesBegin.capture(); |
1932 | unsigned subpatternId = parenthesesBegin.atom.subpatternId; |
1933 | |
1934 | unsigned numSubpatterns = lastSubpatternId - subpatternId + 1; |
1935 | auto parenthesesDisjunction = makeUnique<ByteDisjunction>(numSubpatterns, callFrameSize); |
1936 | |
1937 | unsigned firstTermInParentheses = beginTerm + 1; |
1938 | parenthesesDisjunction->terms.reserveInitialCapacity(endTerm - firstTermInParentheses + 2); |
1939 | |
1940 | parenthesesDisjunction->terms.append(ByteTerm::SubpatternBegin()); |
1941 | for (unsigned termInParentheses = firstTermInParentheses; termInParentheses < endTerm; ++termInParentheses) |
1942 | parenthesesDisjunction->terms.append(m_bodyDisjunction->terms[termInParentheses]); |
1943 | parenthesesDisjunction->terms.append(ByteTerm::SubpatternEnd()); |
1944 | |
1945 | m_bodyDisjunction->terms.shrink(beginTerm); |
1946 | |
1947 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpattern, subpatternId, parenthesesDisjunction.get(), capture, inputPosition)); |
1948 | m_allParenthesesInfo.append(WTFMove(parenthesesDisjunction)); |
1949 | |
1950 | m_bodyDisjunction->terms[beginTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1951 | m_bodyDisjunction->terms[beginTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1952 | m_bodyDisjunction->terms[beginTerm].atom.quantityType = quantityType; |
1953 | m_bodyDisjunction->terms[beginTerm].frameLocation = frameLocation; |
1954 | } |
1955 | |
1956 | void atomParenthesesOnceEnd(unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMinCount, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1957 | { |
1958 | unsigned beginTerm = popParenthesesStack(); |
1959 | closeAlternative(beginTerm + 1); |
1960 | unsigned endTerm = m_bodyDisjunction->terms.size(); |
1961 | |
1962 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeParenthesesSubpatternOnceBegin); |
1963 | |
1964 | bool capture = m_bodyDisjunction->terms[beginTerm].capture(); |
1965 | unsigned subpatternId = m_bodyDisjunction->terms[beginTerm].atom.subpatternId; |
1966 | |
1967 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternOnceEnd, subpatternId, capture, false, inputPosition)); |
1968 | m_bodyDisjunction->terms[beginTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1969 | m_bodyDisjunction->terms[endTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1970 | m_bodyDisjunction->terms[endTerm].frameLocation = frameLocation; |
1971 | |
1972 | m_bodyDisjunction->terms[beginTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1973 | m_bodyDisjunction->terms[beginTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1974 | m_bodyDisjunction->terms[beginTerm].atom.quantityType = quantityType; |
1975 | m_bodyDisjunction->terms[endTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1976 | m_bodyDisjunction->terms[endTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1977 | m_bodyDisjunction->terms[endTerm].atom.quantityType = quantityType; |
1978 | } |
1979 | |
1980 | void atomParenthesesTerminalEnd(unsigned inputPosition, unsigned frameLocation, Checked<unsigned> quantityMinCount, Checked<unsigned> quantityMaxCount, QuantifierType quantityType) |
1981 | { |
1982 | unsigned beginTerm = popParenthesesStack(); |
1983 | closeAlternative(beginTerm + 1); |
1984 | unsigned endTerm = m_bodyDisjunction->terms.size(); |
1985 | |
1986 | ASSERT(m_bodyDisjunction->terms[beginTerm].type == ByteTerm::TypeParenthesesSubpatternTerminalBegin); |
1987 | |
1988 | bool capture = m_bodyDisjunction->terms[beginTerm].capture(); |
1989 | unsigned subpatternId = m_bodyDisjunction->terms[beginTerm].atom.subpatternId; |
1990 | |
1991 | m_bodyDisjunction->terms.append(ByteTerm(ByteTerm::TypeParenthesesSubpatternTerminalEnd, subpatternId, capture, false, inputPosition)); |
1992 | m_bodyDisjunction->terms[beginTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1993 | m_bodyDisjunction->terms[endTerm].atom.parenthesesWidth = endTerm - beginTerm; |
1994 | m_bodyDisjunction->terms[endTerm].frameLocation = frameLocation; |
1995 | |
1996 | m_bodyDisjunction->terms[beginTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
1997 | m_bodyDisjunction->terms[beginTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
1998 | m_bodyDisjunction->terms[beginTerm].atom.quantityType = quantityType; |
1999 | m_bodyDisjunction->terms[endTerm].atom.quantityMinCount = quantityMinCount.unsafeGet(); |
2000 | m_bodyDisjunction->terms[endTerm].atom.quantityMaxCount = quantityMaxCount.unsafeGet(); |
2001 | m_bodyDisjunction->terms[endTerm].atom.quantityType = quantityType; |
2002 | } |
2003 | |
2004 | void regexBegin(unsigned numSubpatterns, unsigned callFrameSize, bool onceThrough) |
2005 | { |
2006 | m_bodyDisjunction = makeUnique<ByteDisjunction>(numSubpatterns, callFrameSize); |
2007 | m_bodyDisjunction->terms.append(ByteTerm::BodyAlternativeBegin(onceThrough)); |
2008 | m_bodyDisjunction->terms[0].frameLocation = 0; |
2009 | m_currentAlternativeIndex = 0; |
2010 | } |
2011 | |
2012 | void regexEnd() |
2013 | { |
2014 | closeBodyAlternative(); |
2015 | } |
2016 | |
2017 | void alternativeBodyDisjunction(bool onceThrough) |
2018 | { |
2019 | unsigned newAlternativeIndex = m_bodyDisjunction->terms.size(); |
2020 | m_bodyDisjunction->terms[m_currentAlternativeIndex].alternative.next = newAlternativeIndex - m_currentAlternativeIndex; |
2021 | m_bodyDisjunction->terms.append(ByteTerm::BodyAlternativeDisjunction(onceThrough)); |
2022 | |
2023 | m_currentAlternativeIndex = newAlternativeIndex; |
2024 | } |
2025 | |
2026 | void alternativeDisjunction() |
2027 | { |
2028 | unsigned newAlternativeIndex = m_bodyDisjunction->terms.size(); |
2029 | m_bodyDisjunction->terms[m_currentAlternativeIndex].alternative.next = newAlternativeIndex - m_currentAlternativeIndex; |
2030 | m_bodyDisjunction->terms.append(ByteTerm::AlternativeDisjunction()); |
2031 | |
2032 | m_currentAlternativeIndex = newAlternativeIndex; |
2033 | } |
2034 | |
2035 | Optional<ErrorCode> WARN_UNUSED_RETURN emitDisjunction(PatternDisjunction* disjunction, Checked<unsigned, RecordOverflow> inputCountAlreadyChecked, unsigned parenthesesInputCountAlreadyChecked) |
2036 | { |
2037 | if (UNLIKELY(!isSafeToRecurse())) |
2038 | return ErrorCode::TooManyDisjunctions; |
2039 | |
2040 | for (unsigned alt = 0; alt < disjunction->m_alternatives.size(); ++alt) { |
2041 | auto currentCountAlreadyChecked = inputCountAlreadyChecked; |
2042 | |
2043 | PatternAlternative* alternative = disjunction->m_alternatives[alt].get(); |
2044 | |
2045 | if (alt) { |
2046 | if (disjunction == m_pattern.m_body) |
2047 | alternativeBodyDisjunction(alternative->onceThrough()); |
2048 | else |
2049 | alternativeDisjunction(); |
2050 | } |
2051 | |
2052 | unsigned minimumSize = alternative->m_minimumSize; |
2053 | ASSERT(minimumSize >= parenthesesInputCountAlreadyChecked); |
2054 | unsigned countToCheck = minimumSize - parenthesesInputCountAlreadyChecked; |
2055 | |
2056 | if (countToCheck) { |
2057 | checkInput(countToCheck); |
2058 | currentCountAlreadyChecked += countToCheck; |
2059 | if (currentCountAlreadyChecked.hasOverflowed()) |
2060 | return ErrorCode::OffsetTooLarge; |
2061 | } |
2062 | |
2063 | for (auto& term : alternative->m_terms) { |
2064 | switch (term.type) { |
2065 | case PatternTerm::TypeAssertionBOL: |
2066 | assertionBOL((currentCountAlreadyChecked - term.inputPosition).unsafeGet()); |
2067 | break; |
2068 | |
2069 | case PatternTerm::TypeAssertionEOL: |
2070 | assertionEOL((currentCountAlreadyChecked - term.inputPosition).unsafeGet()); |
2071 | break; |
2072 | |
2073 | case PatternTerm::TypeAssertionWordBoundary: |
2074 | assertionWordBoundary(term.invert(), (currentCountAlreadyChecked - term.inputPosition).unsafeGet()); |
2075 | break; |
2076 | |
2077 | case PatternTerm::TypePatternCharacter: |
2078 | atomPatternCharacter(term.patternCharacter, (currentCountAlreadyChecked - term.inputPosition).unsafeGet(), term.frameLocation, term.quantityMaxCount, term.quantityType); |
2079 | break; |
2080 | |
2081 | case PatternTerm::TypeCharacterClass: |
2082 | atomCharacterClass(term.characterClass, term.invert(), (currentCountAlreadyChecked - term.inputPosition).unsafeGet(), term.frameLocation, term.quantityMaxCount, term.quantityType); |
2083 | break; |
2084 | |
2085 | case PatternTerm::TypeBackReference: |
2086 | atomBackReference(term.backReferenceSubpatternId, (currentCountAlreadyChecked - term.inputPosition).unsafeGet(), term.frameLocation, term.quantityMaxCount, term.quantityType); |
2087 | break; |
2088 | |
2089 | case PatternTerm::TypeForwardReference: |
2090 | break; |
2091 | |
2092 | case PatternTerm::TypeParenthesesSubpattern: { |
2093 | unsigned disjunctionAlreadyCheckedCount = 0; |
2094 | if (term.quantityMaxCount == 1 && !term.parentheses.isCopy) { |
2095 | unsigned alternativeFrameLocation = term.frameLocation; |
2096 | // For QuantifierFixedCount we pre-check the minimum size; for greedy/non-greedy we reserve a slot in the frame. |
2097 | if (term.quantityType == QuantifierFixedCount) |
2098 | disjunctionAlreadyCheckedCount = term.parentheses.disjunction->m_minimumSize; |
2099 | else |
2100 | alternativeFrameLocation += YarrStackSpaceForBackTrackInfoParenthesesOnce; |
2101 | unsigned delegateEndInputOffset = (currentCountAlreadyChecked - term.inputPosition).unsafeGet(); |
2102 | atomParenthesesOnceBegin(term.parentheses.subpatternId, term.capture(), disjunctionAlreadyCheckedCount + delegateEndInputOffset, term.frameLocation, alternativeFrameLocation); |
2103 | if (auto error = emitDisjunction(term.parentheses.disjunction, currentCountAlreadyChecked, disjunctionAlreadyCheckedCount)) |
2104 | return error; |
2105 | atomParenthesesOnceEnd(delegateEndInputOffset, term.frameLocation, term.quantityMinCount, term.quantityMaxCount, term.quantityType); |
2106 | } else if (term.parentheses.isTerminal) { |
2107 | unsigned delegateEndInputOffset = (currentCountAlreadyChecked - term.inputPosition).unsafeGet(); |
2108 | atomParenthesesTerminalBegin(term.parentheses.subpatternId, term.capture(), disjunctionAlreadyCheckedCount + delegateEndInputOffset, term.frameLocation, term.frameLocation + YarrStackSpaceForBackTrackInfoParenthesesTerminal); |
2109 | if (auto error = emitDisjunction(term.parentheses.disjunction, currentCountAlreadyChecked, disjunctionAlreadyCheckedCount)) |
2110 | return error; |
2111 | atomParenthesesTerminalEnd(delegateEndInputOffset, term.frameLocation, term.quantityMinCount, term.quantityMaxCount, term.quantityType); |
2112 | } else { |
2113 | unsigned delegateEndInputOffset = (currentCountAlreadyChecked - term.inputPosition).unsafeGet(); |
2114 | atomParenthesesSubpatternBegin(term.parentheses.subpatternId, term.capture(), disjunctionAlreadyCheckedCount + delegateEndInputOffset, term.frameLocation, 0); |
2115 | if (auto error = emitDisjunction(term.parentheses.disjunction, currentCountAlreadyChecked, 0)) |
2116 | return error; |
2117 | atomParenthesesSubpatternEnd(term.parentheses.lastSubpatternId, delegateEndInputOffset, term.frameLocation, term.quantityMinCount, term.quantityMaxCount, term.quantityType, term.parentheses.disjunction->m_callFrameSize); |
2118 | } |
2119 | break; |
2120 | } |
2121 | |
2122 | case PatternTerm::TypeParentheticalAssertion: { |
2123 | unsigned alternativeFrameLocation = term.frameLocation + YarrStackSpaceForBackTrackInfoParentheticalAssertion; |
2124 | unsigned positiveInputOffset = (currentCountAlreadyChecked - term.inputPosition).unsafeGet(); |
2125 | unsigned uncheckAmount = 0; |
2126 | if (positiveInputOffset > term.parentheses.disjunction->m_minimumSize) { |
2127 | uncheckAmount = positiveInputOffset - term.parentheses.disjunction->m_minimumSize; |
2128 | uncheckInput(uncheckAmount); |
2129 | currentCountAlreadyChecked -= uncheckAmount; |
2130 | if (currentCountAlreadyChecked.hasOverflowed()) |
2131 | return ErrorCode::OffsetTooLarge; |
2132 | } |
2133 | |
2134 | atomParentheticalAssertionBegin(term.parentheses.subpatternId, term.invert(), term.frameLocation, alternativeFrameLocation); |
2135 | if (auto error = emitDisjunction(term.parentheses.disjunction, currentCountAlreadyChecked, positiveInputOffset - uncheckAmount)) |
2136 | return error; |
2137 | atomParentheticalAssertionEnd(0, term.frameLocation, term.quantityMaxCount, term.quantityType); |
2138 | if (uncheckAmount) { |
2139 | checkInput(uncheckAmount); |
2140 | currentCountAlreadyChecked += uncheckAmount; |
2141 | if (currentCountAlreadyChecked.hasOverflowed()) |
2142 | return ErrorCode::OffsetTooLarge; |
2143 | } |
2144 | break; |
2145 | } |
2146 | |
2147 | case PatternTerm::TypeDotStarEnclosure: |
2148 | assertionDotStarEnclosure(term.anchors.bolAnchor, term.anchors.eolAnchor); |
2149 | break; |
2150 | } |
2151 | } |
2152 | } |
2153 | return WTF::nullopt; |
2154 | } |
2155 | #ifndef NDEBUG |
2156 | void dumpDisjunction(ByteDisjunction* disjunction, unsigned nesting = 0) |
2157 | { |
2158 | PrintStream& out = WTF::dataFile(); |
2159 | |
2160 | unsigned termIndexNest = 0; |
2161 | |
2162 | if (!nesting) { |
2163 | out.printf("ByteDisjunction(%p):\n" , disjunction); |
2164 | nesting = 1; |
2165 | } else { |
2166 | termIndexNest = nesting - 1; |
2167 | nesting = 2; |
2168 | } |
2169 | |
2170 | auto outputTermIndexAndNest = [&](size_t index, unsigned termNesting) { |
2171 | for (unsigned nestingDepth = 0; nestingDepth < termIndexNest; nestingDepth++) |
2172 | out.print(" " ); |
2173 | out.printf("%4zu" , index); |
2174 | for (unsigned nestingDepth = 0; nestingDepth < termNesting; nestingDepth++) |
2175 | out.print(" " ); |
2176 | }; |
2177 | |
2178 | auto dumpQuantity = [&](ByteTerm& term) { |
2179 | if (term.atom.quantityType == QuantifierFixedCount && term.atom.quantityMinCount == 1 && term.atom.quantityMaxCount == 1) |
2180 | return; |
2181 | |
2182 | out.print(" {" , term.atom.quantityMinCount); |
2183 | if (term.atom.quantityMinCount != term.atom.quantityMaxCount) { |
2184 | if (term.atom.quantityMaxCount == UINT_MAX) |
2185 | out.print(",inf" ); |
2186 | else |
2187 | out.print("," , term.atom.quantityMaxCount); |
2188 | } |
2189 | out.print("}" ); |
2190 | if (term.atom.quantityType == QuantifierGreedy) |
2191 | out.print(" greedy" ); |
2192 | else if (term.atom.quantityType == QuantifierNonGreedy) |
2193 | out.print(" non-greedy" ); |
2194 | }; |
2195 | |
2196 | auto dumpCaptured = [&](ByteTerm& term) { |
2197 | if (term.capture()) |
2198 | out.print(" captured (#" , term.atom.subpatternId, ")" ); |
2199 | }; |
2200 | |
2201 | auto dumpInverted = [&](ByteTerm& term) { |
2202 | if (term.invert()) |
2203 | out.print(" inverted" ); |
2204 | }; |
2205 | |
2206 | auto dumpInputPosition = [&](ByteTerm& term) { |
2207 | out.printf(" inputPosition %u" , term.inputPosition); |
2208 | }; |
2209 | |
2210 | auto dumpFrameLocation = [&](ByteTerm& term) { |
2211 | out.printf(" frameLocation %u" , term.frameLocation); |
2212 | }; |
2213 | |
2214 | auto dumpCharacter = [&](ByteTerm& term) { |
2215 | out.print(" " ); |
2216 | dumpUChar32(out, term.atom.patternCharacter); |
2217 | }; |
2218 | |
2219 | auto dumpCharClass = [&](ByteTerm& term) { |
2220 | out.print(" " ); |
2221 | dumpCharacterClass(out, &m_pattern, term.atom.characterClass); |
2222 | }; |
2223 | |
2224 | for (size_t idx = 0; idx < disjunction->terms.size(); ++idx) { |
2225 | ByteTerm term = disjunction->terms[idx]; |
2226 | |
2227 | bool outputNewline = true; |
2228 | |
2229 | switch (term.type) { |
2230 | case ByteTerm::TypeBodyAlternativeBegin: |
2231 | outputTermIndexAndNest(idx, nesting++); |
2232 | out.print("BodyAlternativeBegin" ); |
2233 | if (term.alternative.onceThrough) |
2234 | out.print(" onceThrough" ); |
2235 | dumpFrameLocation(term); |
2236 | break; |
2237 | case ByteTerm::TypeBodyAlternativeDisjunction: |
2238 | outputTermIndexAndNest(idx, nesting - 1); |
2239 | out.print("BodyAlternativeDisjunction" ); |
2240 | dumpFrameLocation(term); |
2241 | break; |
2242 | case ByteTerm::TypeBodyAlternativeEnd: |
2243 | outputTermIndexAndNest(idx, --nesting); |
2244 | out.print("BodyAlternativeEnd" ); |
2245 | dumpFrameLocation(term); |
2246 | break; |
2247 | case ByteTerm::TypeAlternativeBegin: |
2248 | outputTermIndexAndNest(idx, nesting++); |
2249 | out.print("AlternativeBegin" ); |
2250 | dumpFrameLocation(term); |
2251 | break; |
2252 | case ByteTerm::TypeAlternativeDisjunction: |
2253 | outputTermIndexAndNest(idx, nesting - 1); |
2254 | out.print("AlternativeDisjunction" ); |
2255 | dumpFrameLocation(term); |
2256 | break; |
2257 | case ByteTerm::TypeAlternativeEnd: |
2258 | outputTermIndexAndNest(idx, --nesting); |
2259 | out.print("AlternativeEnd" ); |
2260 | dumpFrameLocation(term); |
2261 | break; |
2262 | case ByteTerm::TypeSubpatternBegin: |
2263 | outputTermIndexAndNest(idx, nesting++); |
2264 | out.print("SubpatternBegin" ); |
2265 | break; |
2266 | case ByteTerm::TypeSubpatternEnd: |
2267 | outputTermIndexAndNest(idx, --nesting); |
2268 | out.print("SubpatternEnd" ); |
2269 | break; |
2270 | case ByteTerm::TypeAssertionBOL: |
2271 | outputTermIndexAndNest(idx, nesting); |
2272 | out.print("AssertionBOL" ); |
2273 | break; |
2274 | case ByteTerm::TypeAssertionEOL: |
2275 | outputTermIndexAndNest(idx, nesting); |
2276 | out.print("AssertionEOL" ); |
2277 | break; |
2278 | case ByteTerm::TypeAssertionWordBoundary: |
2279 | outputTermIndexAndNest(idx, nesting); |
2280 | out.print("AssertionWordBoundary" ); |
2281 | break; |
2282 | case ByteTerm::TypePatternCharacterOnce: |
2283 | outputTermIndexAndNest(idx, nesting); |
2284 | out.print("PatternCharacterOnce" ); |
2285 | dumpInverted(term); |
2286 | dumpInputPosition(term); |
2287 | dumpFrameLocation(term); |
2288 | dumpCharacter(term); |
2289 | dumpQuantity(term); |
2290 | break; |
2291 | case ByteTerm::TypePatternCharacterFixed: |
2292 | outputTermIndexAndNest(idx, nesting); |
2293 | out.print("PatternCharacterFixed" ); |
2294 | dumpInverted(term); |
2295 | dumpInputPosition(term); |
2296 | dumpFrameLocation(term); |
2297 | dumpCharacter(term); |
2298 | out.print(" {" , term.atom.quantityMinCount, "}" ); |
2299 | break; |
2300 | case ByteTerm::TypePatternCharacterGreedy: |
2301 | outputTermIndexAndNest(idx, nesting); |
2302 | out.print("PatternCharacterGreedy" ); |
2303 | dumpInverted(term); |
2304 | dumpInputPosition(term); |
2305 | dumpFrameLocation(term); |
2306 | dumpCharacter(term); |
2307 | dumpQuantity(term); |
2308 | break; |
2309 | case ByteTerm::TypePatternCharacterNonGreedy: |
2310 | outputTermIndexAndNest(idx, nesting); |
2311 | out.print("PatternCharacterNonGreedy" ); |
2312 | dumpInverted(term); |
2313 | dumpInputPosition(term); |
2314 | dumpFrameLocation(term); |
2315 | dumpCharacter(term); |
2316 | dumpQuantity(term); |
2317 | break; |
2318 | case ByteTerm::TypePatternCasedCharacterOnce: |
2319 | outputTermIndexAndNest(idx, nesting); |
2320 | out.print("PatternCasedCharacterOnce" ); |
2321 | break; |
2322 | case ByteTerm::TypePatternCasedCharacterFixed: |
2323 | outputTermIndexAndNest(idx, nesting); |
2324 | out.print("PatternCasedCharacterFixed" ); |
2325 | break; |
2326 | case ByteTerm::TypePatternCasedCharacterGreedy: |
2327 | outputTermIndexAndNest(idx, nesting); |
2328 | out.print("PatternCasedCharacterGreedy" ); |
2329 | break; |
2330 | case ByteTerm::TypePatternCasedCharacterNonGreedy: |
2331 | outputTermIndexAndNest(idx, nesting); |
2332 | out.print("PatternCasedCharacterNonGreedy" ); |
2333 | break; |
2334 | case ByteTerm::TypeCharacterClass: |
2335 | outputTermIndexAndNest(idx, nesting); |
2336 | out.print("CharacterClass" ); |
2337 | dumpInverted(term); |
2338 | dumpInputPosition(term); |
2339 | dumpFrameLocation(term); |
2340 | dumpCharClass(term); |
2341 | dumpQuantity(term); |
2342 | break; |
2343 | case ByteTerm::TypeBackReference: |
2344 | outputTermIndexAndNest(idx, nesting); |
2345 | out.print("BackReference #" , term.atom.subpatternId); |
2346 | dumpQuantity(term); |
2347 | break; |
2348 | case ByteTerm::TypeParenthesesSubpattern: |
2349 | outputTermIndexAndNest(idx, nesting); |
2350 | out.print("ParenthesesSubpattern" ); |
2351 | dumpCaptured(term); |
2352 | dumpInverted(term); |
2353 | dumpInputPosition(term); |
2354 | dumpFrameLocation(term); |
2355 | dumpQuantity(term); |
2356 | out.print("\n" ); |
2357 | outputNewline = false; |
2358 | dumpDisjunction(term.atom.parenthesesDisjunction, nesting); |
2359 | break; |
2360 | case ByteTerm::TypeParenthesesSubpatternOnceBegin: |
2361 | outputTermIndexAndNest(idx, nesting++); |
2362 | out.print("ParenthesesSubpatternOnceBegin" ); |
2363 | dumpCaptured(term); |
2364 | dumpInverted(term); |
2365 | dumpInputPosition(term); |
2366 | dumpFrameLocation(term); |
2367 | break; |
2368 | case ByteTerm::TypeParenthesesSubpatternOnceEnd: |
2369 | outputTermIndexAndNest(idx, --nesting); |
2370 | out.print("ParenthesesSubpatternOnceEnd" ); |
2371 | dumpFrameLocation(term); |
2372 | break; |
2373 | case ByteTerm::TypeParenthesesSubpatternTerminalBegin: |
2374 | outputTermIndexAndNest(idx, nesting++); |
2375 | out.print("ParenthesesSubpatternTerminalBegin" ); |
2376 | dumpInverted(term); |
2377 | dumpInputPosition(term); |
2378 | dumpFrameLocation(term); |
2379 | break; |
2380 | case ByteTerm::TypeParenthesesSubpatternTerminalEnd: |
2381 | outputTermIndexAndNest(idx, --nesting); |
2382 | out.print("ParenthesesSubpatternTerminalEnd" ); |
2383 | dumpFrameLocation(term); |
2384 | break; |
2385 | case ByteTerm::TypeParentheticalAssertionBegin: |
2386 | outputTermIndexAndNest(idx, nesting++); |
2387 | out.print("ParentheticalAssertionBegin" ); |
2388 | dumpInverted(term); |
2389 | dumpInputPosition(term); |
2390 | dumpFrameLocation(term); |
2391 | break; |
2392 | case ByteTerm::TypeParentheticalAssertionEnd: |
2393 | outputTermIndexAndNest(idx, --nesting); |
2394 | out.print("ParentheticalAssertionEnd" ); |
2395 | dumpFrameLocation(term); |
2396 | break; |
2397 | case ByteTerm::TypeCheckInput: |
2398 | outputTermIndexAndNest(idx, nesting); |
2399 | out.print("CheckInput " , term.checkInputCount); |
2400 | break; |
2401 | case ByteTerm::TypeUncheckInput: |
2402 | outputTermIndexAndNest(idx, nesting); |
2403 | out.print("UncheckInput " , term.checkInputCount); |
2404 | break; |
2405 | case ByteTerm::TypeDotStarEnclosure: |
2406 | outputTermIndexAndNest(idx, nesting); |
2407 | out.print("DotStarEnclosure" ); |
2408 | break; |
2409 | } |
2410 | if (outputNewline) |
2411 | out.print("\n" ); |
2412 | } |
2413 | } |
2414 | #endif |
2415 | |
2416 | private: |
2417 | inline bool isSafeToRecurse() { return m_stackCheck.isSafeToRecurse(); } |
2418 | |
2419 | YarrPattern& m_pattern; |
2420 | std::unique_ptr<ByteDisjunction> m_bodyDisjunction; |
2421 | StackCheck m_stackCheck; |
2422 | unsigned m_currentAlternativeIndex { 0 }; |
2423 | Vector<ParenthesesStackEntry> m_parenthesesStack; |
2424 | Vector<std::unique_ptr<ByteDisjunction>> m_allParenthesesInfo; |
2425 | }; |
2426 | |
2427 | std::unique_ptr<BytecodePattern> byteCompile(YarrPattern& pattern, BumpPointerAllocator* allocator, ErrorCode& errorCode, ConcurrentJSLock* lock) |
2428 | { |
2429 | return ByteCompiler(pattern).compile(allocator, lock, errorCode); |
2430 | } |
2431 | |
2432 | unsigned interpret(BytecodePattern* bytecode, const String& input, unsigned start, unsigned* output) |
2433 | { |
2434 | SuperSamplerScope superSamplerScope(false); |
2435 | if (input.is8Bit()) |
2436 | return Interpreter<LChar>(bytecode, output, input.characters8(), input.length(), start).interpret(); |
2437 | return Interpreter<UChar>(bytecode, output, input.characters16(), input.length(), start).interpret(); |
2438 | } |
2439 | |
2440 | unsigned interpret(BytecodePattern* bytecode, const LChar* input, unsigned length, unsigned start, unsigned* output) |
2441 | { |
2442 | SuperSamplerScope superSamplerScope(false); |
2443 | return Interpreter<LChar>(bytecode, output, input, length, start).interpret(); |
2444 | } |
2445 | |
2446 | unsigned interpret(BytecodePattern* bytecode, const UChar* input, unsigned length, unsigned start, unsigned* output) |
2447 | { |
2448 | SuperSamplerScope superSamplerScope(false); |
2449 | return Interpreter<UChar>(bytecode, output, input, length, start).interpret(); |
2450 | } |
2451 | |
2452 | // These should be the same for both UChar & LChar. |
2453 | COMPILE_ASSERT(sizeof(BackTrackInfoPatternCharacter) == (YarrStackSpaceForBackTrackInfoPatternCharacter * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoPatternCharacter); |
2454 | COMPILE_ASSERT(sizeof(BackTrackInfoCharacterClass) == (YarrStackSpaceForBackTrackInfoCharacterClass * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoCharacterClass); |
2455 | COMPILE_ASSERT(sizeof(BackTrackInfoBackReference) == (YarrStackSpaceForBackTrackInfoBackReference * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoBackReference); |
2456 | COMPILE_ASSERT(sizeof(BackTrackInfoAlternative) == (YarrStackSpaceForBackTrackInfoAlternative * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoAlternative); |
2457 | COMPILE_ASSERT(sizeof(BackTrackInfoParentheticalAssertion) == (YarrStackSpaceForBackTrackInfoParentheticalAssertion * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoParentheticalAssertion); |
2458 | COMPILE_ASSERT(sizeof(BackTrackInfoParenthesesOnce) == (YarrStackSpaceForBackTrackInfoParenthesesOnce * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoParenthesesOnce); |
2459 | COMPILE_ASSERT(sizeof(Interpreter<UChar>::BackTrackInfoParentheses) <= (YarrStackSpaceForBackTrackInfoParentheses * sizeof(uintptr_t)), CheckYarrStackSpaceForBackTrackInfoParentheses); |
2460 | |
2461 | |
2462 | } } |
2463 | |