OLD | NEW |
(Empty) | |
| 1 // Copyright 2016 the V8 project authors. All rights reserved. |
| 2 // Use of this source code is governed by a BSD-style license that can be |
| 3 // found in the LICENSE file. |
| 4 |
| 5 #include "src/regexp/regexp-parser.h" |
| 6 |
| 7 #include "src/char-predicates-inl.h" |
| 8 #include "src/factory.h" |
| 9 #include "src/isolate.h" |
| 10 #include "src/objects-inl.h" |
| 11 #include "src/regexp/jsregexp.h" |
| 12 #include "src/utils.h" |
| 13 |
| 14 namespace v8 { |
| 15 namespace internal { |
| 16 |
| 17 RegExpParser::RegExpParser(FlatStringReader* in, Handle<String>* error, |
| 18 bool multiline, bool unicode, Isolate* isolate, |
| 19 Zone* zone) |
| 20 : isolate_(isolate), |
| 21 zone_(zone), |
| 22 error_(error), |
| 23 captures_(NULL), |
| 24 in_(in), |
| 25 current_(kEndMarker), |
| 26 next_pos_(0), |
| 27 captures_started_(0), |
| 28 capture_count_(0), |
| 29 has_more_(true), |
| 30 multiline_(multiline), |
| 31 unicode_(unicode), |
| 32 simple_(false), |
| 33 contains_anchor_(false), |
| 34 is_scanned_for_captures_(false), |
| 35 failed_(false) { |
| 36 Advance(); |
| 37 } |
| 38 |
| 39 |
| 40 uc32 RegExpParser::Next() { |
| 41 if (has_next()) { |
| 42 return in()->Get(next_pos_); |
| 43 } else { |
| 44 return kEndMarker; |
| 45 } |
| 46 } |
| 47 |
| 48 |
| 49 void RegExpParser::Advance() { |
| 50 if (next_pos_ < in()->length()) { |
| 51 StackLimitCheck check(isolate()); |
| 52 if (check.HasOverflowed()) { |
| 53 ReportError(CStrVector(Isolate::kStackOverflowMessage)); |
| 54 } else if (zone()->excess_allocation()) { |
| 55 ReportError(CStrVector("Regular expression too large")); |
| 56 } else { |
| 57 current_ = in()->Get(next_pos_); |
| 58 next_pos_++; |
| 59 } |
| 60 } else { |
| 61 current_ = kEndMarker; |
| 62 // Advance so that position() points to 1-after-the-last-character. This is |
| 63 // important so that Reset() to this position works correctly. |
| 64 next_pos_ = in()->length() + 1; |
| 65 has_more_ = false; |
| 66 } |
| 67 } |
| 68 |
| 69 |
| 70 void RegExpParser::Reset(int pos) { |
| 71 next_pos_ = pos; |
| 72 has_more_ = (pos < in()->length()); |
| 73 Advance(); |
| 74 } |
| 75 |
| 76 |
| 77 void RegExpParser::Advance(int dist) { |
| 78 next_pos_ += dist - 1; |
| 79 Advance(); |
| 80 } |
| 81 |
| 82 |
| 83 bool RegExpParser::simple() { return simple_; } |
| 84 |
| 85 |
| 86 bool RegExpParser::IsSyntaxCharacter(uc32 c) { |
| 87 return c == '^' || c == '$' || c == '\\' || c == '.' || c == '*' || |
| 88 c == '+' || c == '?' || c == '(' || c == ')' || c == '[' || c == ']' || |
| 89 c == '{' || c == '}' || c == '|'; |
| 90 } |
| 91 |
| 92 |
| 93 RegExpTree* RegExpParser::ReportError(Vector<const char> message) { |
| 94 failed_ = true; |
| 95 *error_ = isolate()->factory()->NewStringFromAscii(message).ToHandleChecked(); |
| 96 // Zip to the end to make sure the no more input is read. |
| 97 current_ = kEndMarker; |
| 98 next_pos_ = in()->length(); |
| 99 return NULL; |
| 100 } |
| 101 |
| 102 |
| 103 #define CHECK_FAILED /**/); \ |
| 104 if (failed_) return NULL; \ |
| 105 ((void)0 |
| 106 |
| 107 |
| 108 // Pattern :: |
| 109 // Disjunction |
| 110 RegExpTree* RegExpParser::ParsePattern() { |
| 111 RegExpTree* result = ParseDisjunction(CHECK_FAILED); |
| 112 DCHECK(!has_more()); |
| 113 // If the result of parsing is a literal string atom, and it has the |
| 114 // same length as the input, then the atom is identical to the input. |
| 115 if (result->IsAtom() && result->AsAtom()->length() == in()->length()) { |
| 116 simple_ = true; |
| 117 } |
| 118 return result; |
| 119 } |
| 120 |
| 121 |
| 122 // Disjunction :: |
| 123 // Alternative |
| 124 // Alternative | Disjunction |
| 125 // Alternative :: |
| 126 // [empty] |
| 127 // Term Alternative |
| 128 // Term :: |
| 129 // Assertion |
| 130 // Atom |
| 131 // Atom Quantifier |
| 132 RegExpTree* RegExpParser::ParseDisjunction() { |
| 133 // Used to store current state while parsing subexpressions. |
| 134 RegExpParserState initial_state(NULL, INITIAL, RegExpLookaround::LOOKAHEAD, 0, |
| 135 zone()); |
| 136 RegExpParserState* state = &initial_state; |
| 137 // Cache the builder in a local variable for quick access. |
| 138 RegExpBuilder* builder = initial_state.builder(); |
| 139 while (true) { |
| 140 switch (current()) { |
| 141 case kEndMarker: |
| 142 if (state->IsSubexpression()) { |
| 143 // Inside a parenthesized group when hitting end of input. |
| 144 ReportError(CStrVector("Unterminated group") CHECK_FAILED); |
| 145 } |
| 146 DCHECK_EQ(INITIAL, state->group_type()); |
| 147 // Parsing completed successfully. |
| 148 return builder->ToRegExp(); |
| 149 case ')': { |
| 150 if (!state->IsSubexpression()) { |
| 151 ReportError(CStrVector("Unmatched ')'") CHECK_FAILED); |
| 152 } |
| 153 DCHECK_NE(INITIAL, state->group_type()); |
| 154 |
| 155 Advance(); |
| 156 // End disjunction parsing and convert builder content to new single |
| 157 // regexp atom. |
| 158 RegExpTree* body = builder->ToRegExp(); |
| 159 |
| 160 int end_capture_index = captures_started(); |
| 161 |
| 162 int capture_index = state->capture_index(); |
| 163 SubexpressionType group_type = state->group_type(); |
| 164 |
| 165 // Build result of subexpression. |
| 166 if (group_type == CAPTURE) { |
| 167 RegExpCapture* capture = GetCapture(capture_index); |
| 168 capture->set_body(body); |
| 169 body = capture; |
| 170 } else if (group_type != GROUPING) { |
| 171 DCHECK(group_type == POSITIVE_LOOKAROUND || |
| 172 group_type == NEGATIVE_LOOKAROUND); |
| 173 bool is_positive = (group_type == POSITIVE_LOOKAROUND); |
| 174 body = new (zone()) RegExpLookaround( |
| 175 body, is_positive, end_capture_index - capture_index, |
| 176 capture_index, state->lookaround_type()); |
| 177 } |
| 178 |
| 179 // Restore previous state. |
| 180 state = state->previous_state(); |
| 181 builder = state->builder(); |
| 182 |
| 183 builder->AddAtom(body); |
| 184 // For compatability with JSC and ES3, we allow quantifiers after |
| 185 // lookaheads, and break in all cases. |
| 186 break; |
| 187 } |
| 188 case '|': { |
| 189 Advance(); |
| 190 builder->NewAlternative(); |
| 191 continue; |
| 192 } |
| 193 case '*': |
| 194 case '+': |
| 195 case '?': |
| 196 return ReportError(CStrVector("Nothing to repeat")); |
| 197 case '^': { |
| 198 Advance(); |
| 199 if (multiline_) { |
| 200 builder->AddAssertion( |
| 201 new (zone()) RegExpAssertion(RegExpAssertion::START_OF_LINE)); |
| 202 } else { |
| 203 builder->AddAssertion( |
| 204 new (zone()) RegExpAssertion(RegExpAssertion::START_OF_INPUT)); |
| 205 set_contains_anchor(); |
| 206 } |
| 207 continue; |
| 208 } |
| 209 case '$': { |
| 210 Advance(); |
| 211 RegExpAssertion::AssertionType assertion_type = |
| 212 multiline_ ? RegExpAssertion::END_OF_LINE |
| 213 : RegExpAssertion::END_OF_INPUT; |
| 214 builder->AddAssertion(new (zone()) RegExpAssertion(assertion_type)); |
| 215 continue; |
| 216 } |
| 217 case '.': { |
| 218 Advance(); |
| 219 // everything except \x0a, \x0d, \u2028 and \u2029 |
| 220 ZoneList<CharacterRange>* ranges = |
| 221 new (zone()) ZoneList<CharacterRange>(2, zone()); |
| 222 CharacterRange::AddClassEscape('.', ranges, zone()); |
| 223 RegExpTree* atom = new (zone()) RegExpCharacterClass(ranges, false); |
| 224 builder->AddAtom(atom); |
| 225 break; |
| 226 } |
| 227 case '(': { |
| 228 SubexpressionType subexpr_type = CAPTURE; |
| 229 RegExpLookaround::Type lookaround_type = state->lookaround_type(); |
| 230 Advance(); |
| 231 if (current() == '?') { |
| 232 switch (Next()) { |
| 233 case ':': |
| 234 subexpr_type = GROUPING; |
| 235 break; |
| 236 case '=': |
| 237 lookaround_type = RegExpLookaround::LOOKAHEAD; |
| 238 subexpr_type = POSITIVE_LOOKAROUND; |
| 239 break; |
| 240 case '!': |
| 241 lookaround_type = RegExpLookaround::LOOKAHEAD; |
| 242 subexpr_type = NEGATIVE_LOOKAROUND; |
| 243 break; |
| 244 case '<': |
| 245 if (FLAG_harmony_regexp_lookbehind) { |
| 246 Advance(); |
| 247 lookaround_type = RegExpLookaround::LOOKBEHIND; |
| 248 if (Next() == '=') { |
| 249 subexpr_type = POSITIVE_LOOKAROUND; |
| 250 break; |
| 251 } else if (Next() == '!') { |
| 252 subexpr_type = NEGATIVE_LOOKAROUND; |
| 253 break; |
| 254 } |
| 255 } |
| 256 // Fall through. |
| 257 default: |
| 258 ReportError(CStrVector("Invalid group") CHECK_FAILED); |
| 259 break; |
| 260 } |
| 261 Advance(2); |
| 262 } else { |
| 263 if (captures_started_ >= kMaxCaptures) { |
| 264 ReportError(CStrVector("Too many captures") CHECK_FAILED); |
| 265 } |
| 266 captures_started_++; |
| 267 } |
| 268 // Store current state and begin new disjunction parsing. |
| 269 state = new (zone()) RegExpParserState( |
| 270 state, subexpr_type, lookaround_type, captures_started_, zone()); |
| 271 builder = state->builder(); |
| 272 continue; |
| 273 } |
| 274 case '[': { |
| 275 RegExpTree* atom = ParseCharacterClass(CHECK_FAILED); |
| 276 builder->AddAtom(atom); |
| 277 break; |
| 278 } |
| 279 // Atom :: |
| 280 // \ AtomEscape |
| 281 case '\\': |
| 282 switch (Next()) { |
| 283 case kEndMarker: |
| 284 return ReportError(CStrVector("\\ at end of pattern")); |
| 285 case 'b': |
| 286 Advance(2); |
| 287 builder->AddAssertion( |
| 288 new (zone()) RegExpAssertion(RegExpAssertion::BOUNDARY)); |
| 289 continue; |
| 290 case 'B': |
| 291 Advance(2); |
| 292 builder->AddAssertion( |
| 293 new (zone()) RegExpAssertion(RegExpAssertion::NON_BOUNDARY)); |
| 294 continue; |
| 295 // AtomEscape :: |
| 296 // CharacterClassEscape |
| 297 // |
| 298 // CharacterClassEscape :: one of |
| 299 // d D s S w W |
| 300 case 'd': |
| 301 case 'D': |
| 302 case 's': |
| 303 case 'S': |
| 304 case 'w': |
| 305 case 'W': { |
| 306 uc32 c = Next(); |
| 307 Advance(2); |
| 308 ZoneList<CharacterRange>* ranges = |
| 309 new (zone()) ZoneList<CharacterRange>(2, zone()); |
| 310 CharacterRange::AddClassEscape(c, ranges, zone()); |
| 311 RegExpTree* atom = new (zone()) RegExpCharacterClass(ranges, false); |
| 312 builder->AddAtom(atom); |
| 313 break; |
| 314 } |
| 315 case '1': |
| 316 case '2': |
| 317 case '3': |
| 318 case '4': |
| 319 case '5': |
| 320 case '6': |
| 321 case '7': |
| 322 case '8': |
| 323 case '9': { |
| 324 int index = 0; |
| 325 if (ParseBackReferenceIndex(&index)) { |
| 326 if (state->IsInsideCaptureGroup(index)) { |
| 327 // The back reference is inside the capture group it refers to. |
| 328 // Nothing can possibly have been captured yet, so we use empty |
| 329 // instead. This ensures that, when checking a back reference, |
| 330 // the capture registers of the referenced capture are either |
| 331 // both set or both cleared. |
| 332 builder->AddEmpty(); |
| 333 } else { |
| 334 RegExpCapture* capture = GetCapture(index); |
| 335 RegExpTree* atom = new (zone()) RegExpBackReference(capture); |
| 336 builder->AddAtom(atom); |
| 337 } |
| 338 break; |
| 339 } |
| 340 uc32 first_digit = Next(); |
| 341 if (first_digit == '8' || first_digit == '9') { |
| 342 // If the 'u' flag is present, only syntax characters can be |
| 343 // escaped, |
| 344 // no other identity escapes are allowed. If the 'u' flag is not |
| 345 // present, all identity escapes are allowed. |
| 346 if (!FLAG_harmony_unicode_regexps || !unicode_) { |
| 347 builder->AddCharacter(first_digit); |
| 348 Advance(2); |
| 349 } else { |
| 350 return ReportError(CStrVector("Invalid escape")); |
| 351 } |
| 352 break; |
| 353 } |
| 354 } |
| 355 // FALLTHROUGH |
| 356 case '0': { |
| 357 Advance(); |
| 358 uc32 octal = ParseOctalLiteral(); |
| 359 builder->AddCharacter(octal); |
| 360 break; |
| 361 } |
| 362 // ControlEscape :: one of |
| 363 // f n r t v |
| 364 case 'f': |
| 365 Advance(2); |
| 366 builder->AddCharacter('\f'); |
| 367 break; |
| 368 case 'n': |
| 369 Advance(2); |
| 370 builder->AddCharacter('\n'); |
| 371 break; |
| 372 case 'r': |
| 373 Advance(2); |
| 374 builder->AddCharacter('\r'); |
| 375 break; |
| 376 case 't': |
| 377 Advance(2); |
| 378 builder->AddCharacter('\t'); |
| 379 break; |
| 380 case 'v': |
| 381 Advance(2); |
| 382 builder->AddCharacter('\v'); |
| 383 break; |
| 384 case 'c': { |
| 385 Advance(); |
| 386 uc32 controlLetter = Next(); |
| 387 // Special case if it is an ASCII letter. |
| 388 // Convert lower case letters to uppercase. |
| 389 uc32 letter = controlLetter & ~('a' ^ 'A'); |
| 390 if (letter < 'A' || 'Z' < letter) { |
| 391 // controlLetter is not in range 'A'-'Z' or 'a'-'z'. |
| 392 // This is outside the specification. We match JSC in |
| 393 // reading the backslash as a literal character instead |
| 394 // of as starting an escape. |
| 395 builder->AddCharacter('\\'); |
| 396 } else { |
| 397 Advance(2); |
| 398 builder->AddCharacter(controlLetter & 0x1f); |
| 399 } |
| 400 break; |
| 401 } |
| 402 case 'x': { |
| 403 Advance(2); |
| 404 uc32 value; |
| 405 if (ParseHexEscape(2, &value)) { |
| 406 builder->AddCharacter(value); |
| 407 } else if (!FLAG_harmony_unicode_regexps || !unicode_) { |
| 408 builder->AddCharacter('x'); |
| 409 } else { |
| 410 // If the 'u' flag is present, invalid escapes are not treated as |
| 411 // identity escapes. |
| 412 return ReportError(CStrVector("Invalid escape")); |
| 413 } |
| 414 break; |
| 415 } |
| 416 case 'u': { |
| 417 Advance(2); |
| 418 uc32 value; |
| 419 if (ParseUnicodeEscape(&value)) { |
| 420 builder->AddCharacter(value); |
| 421 } else if (!FLAG_harmony_unicode_regexps || !unicode_) { |
| 422 builder->AddCharacter('u'); |
| 423 } else { |
| 424 // If the 'u' flag is present, invalid escapes are not treated as |
| 425 // identity escapes. |
| 426 return ReportError(CStrVector("Invalid unicode escape")); |
| 427 } |
| 428 break; |
| 429 } |
| 430 default: |
| 431 Advance(); |
| 432 // If the 'u' flag is present, only syntax characters can be |
| 433 // escaped, no |
| 434 // other identity escapes are allowed. If the 'u' flag is not |
| 435 // present, |
| 436 // all identity escapes are allowed. |
| 437 if (!FLAG_harmony_unicode_regexps || !unicode_ || |
| 438 IsSyntaxCharacter(current())) { |
| 439 builder->AddCharacter(current()); |
| 440 Advance(); |
| 441 } else { |
| 442 return ReportError(CStrVector("Invalid escape")); |
| 443 } |
| 444 break; |
| 445 } |
| 446 break; |
| 447 case '{': { |
| 448 int dummy; |
| 449 if (ParseIntervalQuantifier(&dummy, &dummy)) { |
| 450 ReportError(CStrVector("Nothing to repeat") CHECK_FAILED); |
| 451 } |
| 452 // fallthrough |
| 453 } |
| 454 default: |
| 455 builder->AddCharacter(current()); |
| 456 Advance(); |
| 457 break; |
| 458 } // end switch(current()) |
| 459 |
| 460 int min; |
| 461 int max; |
| 462 switch (current()) { |
| 463 // QuantifierPrefix :: |
| 464 // * |
| 465 // + |
| 466 // ? |
| 467 // { |
| 468 case '*': |
| 469 min = 0; |
| 470 max = RegExpTree::kInfinity; |
| 471 Advance(); |
| 472 break; |
| 473 case '+': |
| 474 min = 1; |
| 475 max = RegExpTree::kInfinity; |
| 476 Advance(); |
| 477 break; |
| 478 case '?': |
| 479 min = 0; |
| 480 max = 1; |
| 481 Advance(); |
| 482 break; |
| 483 case '{': |
| 484 if (ParseIntervalQuantifier(&min, &max)) { |
| 485 if (max < min) { |
| 486 ReportError(CStrVector("numbers out of order in {} quantifier.") |
| 487 CHECK_FAILED); |
| 488 } |
| 489 break; |
| 490 } else { |
| 491 continue; |
| 492 } |
| 493 default: |
| 494 continue; |
| 495 } |
| 496 RegExpQuantifier::QuantifierType quantifier_type = RegExpQuantifier::GREEDY; |
| 497 if (current() == '?') { |
| 498 quantifier_type = RegExpQuantifier::NON_GREEDY; |
| 499 Advance(); |
| 500 } else if (FLAG_regexp_possessive_quantifier && current() == '+') { |
| 501 // FLAG_regexp_possessive_quantifier is a debug-only flag. |
| 502 quantifier_type = RegExpQuantifier::POSSESSIVE; |
| 503 Advance(); |
| 504 } |
| 505 builder->AddQuantifierToAtom(min, max, quantifier_type); |
| 506 } |
| 507 } |
| 508 |
| 509 |
| 510 #ifdef DEBUG |
| 511 // Currently only used in an DCHECK. |
| 512 static bool IsSpecialClassEscape(uc32 c) { |
| 513 switch (c) { |
| 514 case 'd': |
| 515 case 'D': |
| 516 case 's': |
| 517 case 'S': |
| 518 case 'w': |
| 519 case 'W': |
| 520 return true; |
| 521 default: |
| 522 return false; |
| 523 } |
| 524 } |
| 525 #endif |
| 526 |
| 527 |
| 528 // In order to know whether an escape is a backreference or not we have to scan |
| 529 // the entire regexp and find the number of capturing parentheses. However we |
| 530 // don't want to scan the regexp twice unless it is necessary. This mini-parser |
| 531 // is called when needed. It can see the difference between capturing and |
| 532 // noncapturing parentheses and can skip character classes and backslash-escaped |
| 533 // characters. |
| 534 void RegExpParser::ScanForCaptures() { |
| 535 // Start with captures started previous to current position |
| 536 int capture_count = captures_started(); |
| 537 // Add count of captures after this position. |
| 538 int n; |
| 539 while ((n = current()) != kEndMarker) { |
| 540 Advance(); |
| 541 switch (n) { |
| 542 case '\\': |
| 543 Advance(); |
| 544 break; |
| 545 case '[': { |
| 546 int c; |
| 547 while ((c = current()) != kEndMarker) { |
| 548 Advance(); |
| 549 if (c == '\\') { |
| 550 Advance(); |
| 551 } else { |
| 552 if (c == ']') break; |
| 553 } |
| 554 } |
| 555 break; |
| 556 } |
| 557 case '(': |
| 558 if (current() != '?') capture_count++; |
| 559 break; |
| 560 } |
| 561 } |
| 562 capture_count_ = capture_count; |
| 563 is_scanned_for_captures_ = true; |
| 564 } |
| 565 |
| 566 |
| 567 bool RegExpParser::ParseBackReferenceIndex(int* index_out) { |
| 568 DCHECK_EQ('\\', current()); |
| 569 DCHECK('1' <= Next() && Next() <= '9'); |
| 570 // Try to parse a decimal literal that is no greater than the total number |
| 571 // of left capturing parentheses in the input. |
| 572 int start = position(); |
| 573 int value = Next() - '0'; |
| 574 Advance(2); |
| 575 while (true) { |
| 576 uc32 c = current(); |
| 577 if (IsDecimalDigit(c)) { |
| 578 value = 10 * value + (c - '0'); |
| 579 if (value > kMaxCaptures) { |
| 580 Reset(start); |
| 581 return false; |
| 582 } |
| 583 Advance(); |
| 584 } else { |
| 585 break; |
| 586 } |
| 587 } |
| 588 if (value > captures_started()) { |
| 589 if (!is_scanned_for_captures_) { |
| 590 int saved_position = position(); |
| 591 ScanForCaptures(); |
| 592 Reset(saved_position); |
| 593 } |
| 594 if (value > capture_count_) { |
| 595 Reset(start); |
| 596 return false; |
| 597 } |
| 598 } |
| 599 *index_out = value; |
| 600 return true; |
| 601 } |
| 602 |
| 603 |
| 604 RegExpCapture* RegExpParser::GetCapture(int index) { |
| 605 // The index for the capture groups are one-based. Its index in the list is |
| 606 // zero-based. |
| 607 int know_captures = |
| 608 is_scanned_for_captures_ ? capture_count_ : captures_started_; |
| 609 DCHECK(index <= know_captures); |
| 610 if (captures_ == NULL) { |
| 611 captures_ = new (zone()) ZoneList<RegExpCapture*>(know_captures, zone()); |
| 612 } |
| 613 while (captures_->length() < know_captures) { |
| 614 captures_->Add(new (zone()) RegExpCapture(captures_->length() + 1), zone()); |
| 615 } |
| 616 return captures_->at(index - 1); |
| 617 } |
| 618 |
| 619 |
| 620 bool RegExpParser::RegExpParserState::IsInsideCaptureGroup(int index) { |
| 621 for (RegExpParserState* s = this; s != NULL; s = s->previous_state()) { |
| 622 if (s->group_type() != CAPTURE) continue; |
| 623 // Return true if we found the matching capture index. |
| 624 if (index == s->capture_index()) return true; |
| 625 // Abort if index is larger than what has been parsed up till this state. |
| 626 if (index > s->capture_index()) return false; |
| 627 } |
| 628 return false; |
| 629 } |
| 630 |
| 631 |
| 632 // QuantifierPrefix :: |
| 633 // { DecimalDigits } |
| 634 // { DecimalDigits , } |
| 635 // { DecimalDigits , DecimalDigits } |
| 636 // |
| 637 // Returns true if parsing succeeds, and set the min_out and max_out |
| 638 // values. Values are truncated to RegExpTree::kInfinity if they overflow. |
| 639 bool RegExpParser::ParseIntervalQuantifier(int* min_out, int* max_out) { |
| 640 DCHECK_EQ(current(), '{'); |
| 641 int start = position(); |
| 642 Advance(); |
| 643 int min = 0; |
| 644 if (!IsDecimalDigit(current())) { |
| 645 Reset(start); |
| 646 return false; |
| 647 } |
| 648 while (IsDecimalDigit(current())) { |
| 649 int next = current() - '0'; |
| 650 if (min > (RegExpTree::kInfinity - next) / 10) { |
| 651 // Overflow. Skip past remaining decimal digits and return -1. |
| 652 do { |
| 653 Advance(); |
| 654 } while (IsDecimalDigit(current())); |
| 655 min = RegExpTree::kInfinity; |
| 656 break; |
| 657 } |
| 658 min = 10 * min + next; |
| 659 Advance(); |
| 660 } |
| 661 int max = 0; |
| 662 if (current() == '}') { |
| 663 max = min; |
| 664 Advance(); |
| 665 } else if (current() == ',') { |
| 666 Advance(); |
| 667 if (current() == '}') { |
| 668 max = RegExpTree::kInfinity; |
| 669 Advance(); |
| 670 } else { |
| 671 while (IsDecimalDigit(current())) { |
| 672 int next = current() - '0'; |
| 673 if (max > (RegExpTree::kInfinity - next) / 10) { |
| 674 do { |
| 675 Advance(); |
| 676 } while (IsDecimalDigit(current())); |
| 677 max = RegExpTree::kInfinity; |
| 678 break; |
| 679 } |
| 680 max = 10 * max + next; |
| 681 Advance(); |
| 682 } |
| 683 if (current() != '}') { |
| 684 Reset(start); |
| 685 return false; |
| 686 } |
| 687 Advance(); |
| 688 } |
| 689 } else { |
| 690 Reset(start); |
| 691 return false; |
| 692 } |
| 693 *min_out = min; |
| 694 *max_out = max; |
| 695 return true; |
| 696 } |
| 697 |
| 698 |
| 699 uc32 RegExpParser::ParseOctalLiteral() { |
| 700 DCHECK(('0' <= current() && current() <= '7') || current() == kEndMarker); |
| 701 // For compatibility with some other browsers (not all), we parse |
| 702 // up to three octal digits with a value below 256. |
| 703 uc32 value = current() - '0'; |
| 704 Advance(); |
| 705 if ('0' <= current() && current() <= '7') { |
| 706 value = value * 8 + current() - '0'; |
| 707 Advance(); |
| 708 if (value < 32 && '0' <= current() && current() <= '7') { |
| 709 value = value * 8 + current() - '0'; |
| 710 Advance(); |
| 711 } |
| 712 } |
| 713 return value; |
| 714 } |
| 715 |
| 716 |
| 717 bool RegExpParser::ParseHexEscape(int length, uc32* value) { |
| 718 int start = position(); |
| 719 uc32 val = 0; |
| 720 for (int i = 0; i < length; ++i) { |
| 721 uc32 c = current(); |
| 722 int d = HexValue(c); |
| 723 if (d < 0) { |
| 724 Reset(start); |
| 725 return false; |
| 726 } |
| 727 val = val * 16 + d; |
| 728 Advance(); |
| 729 } |
| 730 *value = val; |
| 731 return true; |
| 732 } |
| 733 |
| 734 |
| 735 bool RegExpParser::ParseUnicodeEscape(uc32* value) { |
| 736 // Accept both \uxxxx and \u{xxxxxx} (if harmony unicode escapes are |
| 737 // allowed). In the latter case, the number of hex digits between { } is |
| 738 // arbitrary. \ and u have already been read. |
| 739 if (current() == '{' && FLAG_harmony_unicode_regexps && unicode_) { |
| 740 int start = position(); |
| 741 Advance(); |
| 742 if (ParseUnlimitedLengthHexNumber(0x10ffff, value)) { |
| 743 if (current() == '}') { |
| 744 Advance(); |
| 745 return true; |
| 746 } |
| 747 } |
| 748 Reset(start); |
| 749 return false; |
| 750 } |
| 751 // \u but no {, or \u{...} escapes not allowed. |
| 752 return ParseHexEscape(4, value); |
| 753 } |
| 754 |
| 755 |
| 756 bool RegExpParser::ParseUnlimitedLengthHexNumber(int max_value, uc32* value) { |
| 757 uc32 x = 0; |
| 758 int d = HexValue(current()); |
| 759 if (d < 0) { |
| 760 return false; |
| 761 } |
| 762 while (d >= 0) { |
| 763 x = x * 16 + d; |
| 764 if (x > max_value) { |
| 765 return false; |
| 766 } |
| 767 Advance(); |
| 768 d = HexValue(current()); |
| 769 } |
| 770 *value = x; |
| 771 return true; |
| 772 } |
| 773 |
| 774 |
| 775 uc32 RegExpParser::ParseClassCharacterEscape() { |
| 776 DCHECK(current() == '\\'); |
| 777 DCHECK(has_next() && !IsSpecialClassEscape(Next())); |
| 778 Advance(); |
| 779 switch (current()) { |
| 780 case 'b': |
| 781 Advance(); |
| 782 return '\b'; |
| 783 // ControlEscape :: one of |
| 784 // f n r t v |
| 785 case 'f': |
| 786 Advance(); |
| 787 return '\f'; |
| 788 case 'n': |
| 789 Advance(); |
| 790 return '\n'; |
| 791 case 'r': |
| 792 Advance(); |
| 793 return '\r'; |
| 794 case 't': |
| 795 Advance(); |
| 796 return '\t'; |
| 797 case 'v': |
| 798 Advance(); |
| 799 return '\v'; |
| 800 case 'c': { |
| 801 uc32 controlLetter = Next(); |
| 802 uc32 letter = controlLetter & ~('A' ^ 'a'); |
| 803 // For compatibility with JSC, inside a character class |
| 804 // we also accept digits and underscore as control characters. |
| 805 if ((controlLetter >= '0' && controlLetter <= '9') || |
| 806 controlLetter == '_' || (letter >= 'A' && letter <= 'Z')) { |
| 807 Advance(2); |
| 808 // Control letters mapped to ASCII control characters in the range |
| 809 // 0x00-0x1f. |
| 810 return controlLetter & 0x1f; |
| 811 } |
| 812 // We match JSC in reading the backslash as a literal |
| 813 // character instead of as starting an escape. |
| 814 return '\\'; |
| 815 } |
| 816 case '0': |
| 817 case '1': |
| 818 case '2': |
| 819 case '3': |
| 820 case '4': |
| 821 case '5': |
| 822 case '6': |
| 823 case '7': |
| 824 // For compatibility, we interpret a decimal escape that isn't |
| 825 // a back reference (and therefore either \0 or not valid according |
| 826 // to the specification) as a 1..3 digit octal character code. |
| 827 return ParseOctalLiteral(); |
| 828 case 'x': { |
| 829 Advance(); |
| 830 uc32 value; |
| 831 if (ParseHexEscape(2, &value)) { |
| 832 return value; |
| 833 } |
| 834 if (!FLAG_harmony_unicode_regexps || !unicode_) { |
| 835 // If \x is not followed by a two-digit hexadecimal, treat it |
| 836 // as an identity escape. |
| 837 return 'x'; |
| 838 } |
| 839 // If the 'u' flag is present, invalid escapes are not treated as |
| 840 // identity escapes. |
| 841 ReportError(CStrVector("Invalid escape")); |
| 842 return 0; |
| 843 } |
| 844 case 'u': { |
| 845 Advance(); |
| 846 uc32 value; |
| 847 if (ParseUnicodeEscape(&value)) { |
| 848 return value; |
| 849 } |
| 850 if (!FLAG_harmony_unicode_regexps || !unicode_) { |
| 851 return 'u'; |
| 852 } |
| 853 // If the 'u' flag is present, invalid escapes are not treated as |
| 854 // identity escapes. |
| 855 ReportError(CStrVector("Invalid unicode escape")); |
| 856 return 0; |
| 857 } |
| 858 default: { |
| 859 uc32 result = current(); |
| 860 // If the 'u' flag is present, only syntax characters can be escaped, no |
| 861 // other identity escapes are allowed. If the 'u' flag is not present, all |
| 862 // identity escapes are allowed. |
| 863 if (!FLAG_harmony_unicode_regexps || !unicode_ || |
| 864 IsSyntaxCharacter(result)) { |
| 865 Advance(); |
| 866 return result; |
| 867 } |
| 868 ReportError(CStrVector("Invalid escape")); |
| 869 return 0; |
| 870 } |
| 871 } |
| 872 return 0; |
| 873 } |
| 874 |
| 875 |
| 876 CharacterRange RegExpParser::ParseClassAtom(uc16* char_class) { |
| 877 DCHECK_EQ(0, *char_class); |
| 878 uc32 first = current(); |
| 879 if (first == '\\') { |
| 880 switch (Next()) { |
| 881 case 'w': |
| 882 case 'W': |
| 883 case 'd': |
| 884 case 'D': |
| 885 case 's': |
| 886 case 'S': { |
| 887 *char_class = Next(); |
| 888 Advance(2); |
| 889 return CharacterRange::Singleton(0); // Return dummy value. |
| 890 } |
| 891 case kEndMarker: |
| 892 return ReportError(CStrVector("\\ at end of pattern")); |
| 893 default: |
| 894 uc32 c = ParseClassCharacterEscape(CHECK_FAILED); |
| 895 return CharacterRange::Singleton(c); |
| 896 } |
| 897 } else { |
| 898 Advance(); |
| 899 return CharacterRange::Singleton(first); |
| 900 } |
| 901 } |
| 902 |
| 903 |
| 904 static const uc16 kNoCharClass = 0; |
| 905 |
| 906 // Adds range or pre-defined character class to character ranges. |
| 907 // If char_class is not kInvalidClass, it's interpreted as a class |
| 908 // escape (i.e., 's' means whitespace, from '\s'). |
| 909 static inline void AddRangeOrEscape(ZoneList<CharacterRange>* ranges, |
| 910 uc16 char_class, CharacterRange range, |
| 911 Zone* zone) { |
| 912 if (char_class != kNoCharClass) { |
| 913 CharacterRange::AddClassEscape(char_class, ranges, zone); |
| 914 } else { |
| 915 ranges->Add(range, zone); |
| 916 } |
| 917 } |
| 918 |
| 919 |
| 920 RegExpTree* RegExpParser::ParseCharacterClass() { |
| 921 static const char* kUnterminated = "Unterminated character class"; |
| 922 static const char* kRangeOutOfOrder = "Range out of order in character class"; |
| 923 |
| 924 DCHECK_EQ(current(), '['); |
| 925 Advance(); |
| 926 bool is_negated = false; |
| 927 if (current() == '^') { |
| 928 is_negated = true; |
| 929 Advance(); |
| 930 } |
| 931 ZoneList<CharacterRange>* ranges = |
| 932 new (zone()) ZoneList<CharacterRange>(2, zone()); |
| 933 while (has_more() && current() != ']') { |
| 934 uc16 char_class = kNoCharClass; |
| 935 CharacterRange first = ParseClassAtom(&char_class CHECK_FAILED); |
| 936 if (current() == '-') { |
| 937 Advance(); |
| 938 if (current() == kEndMarker) { |
| 939 // If we reach the end we break out of the loop and let the |
| 940 // following code report an error. |
| 941 break; |
| 942 } else if (current() == ']') { |
| 943 AddRangeOrEscape(ranges, char_class, first, zone()); |
| 944 ranges->Add(CharacterRange::Singleton('-'), zone()); |
| 945 break; |
| 946 } |
| 947 uc16 char_class_2 = kNoCharClass; |
| 948 CharacterRange next = ParseClassAtom(&char_class_2 CHECK_FAILED); |
| 949 if (char_class != kNoCharClass || char_class_2 != kNoCharClass) { |
| 950 // Either end is an escaped character class. Treat the '-' verbatim. |
| 951 AddRangeOrEscape(ranges, char_class, first, zone()); |
| 952 ranges->Add(CharacterRange::Singleton('-'), zone()); |
| 953 AddRangeOrEscape(ranges, char_class_2, next, zone()); |
| 954 continue; |
| 955 } |
| 956 if (first.from() > next.to()) { |
| 957 return ReportError(CStrVector(kRangeOutOfOrder) CHECK_FAILED); |
| 958 } |
| 959 ranges->Add(CharacterRange::Range(first.from(), next.to()), zone()); |
| 960 } else { |
| 961 AddRangeOrEscape(ranges, char_class, first, zone()); |
| 962 } |
| 963 } |
| 964 if (!has_more()) { |
| 965 return ReportError(CStrVector(kUnterminated) CHECK_FAILED); |
| 966 } |
| 967 Advance(); |
| 968 if (ranges->length() == 0) { |
| 969 ranges->Add(CharacterRange::Everything(), zone()); |
| 970 is_negated = !is_negated; |
| 971 } |
| 972 return new (zone()) RegExpCharacterClass(ranges, is_negated); |
| 973 } |
| 974 |
| 975 |
| 976 #undef CHECK_FAILED |
| 977 |
| 978 |
| 979 bool RegExpParser::ParseRegExp(Isolate* isolate, Zone* zone, |
| 980 FlatStringReader* input, bool multiline, |
| 981 bool unicode, RegExpCompileData* result) { |
| 982 DCHECK(result != NULL); |
| 983 RegExpParser parser(input, &result->error, multiline, unicode, isolate, zone); |
| 984 RegExpTree* tree = parser.ParsePattern(); |
| 985 if (parser.failed()) { |
| 986 DCHECK(tree == NULL); |
| 987 DCHECK(!result->error.is_null()); |
| 988 } else { |
| 989 DCHECK(tree != NULL); |
| 990 DCHECK(result->error.is_null()); |
| 991 result->tree = tree; |
| 992 int capture_count = parser.captures_started(); |
| 993 result->simple = tree->IsAtom() && parser.simple() && capture_count == 0; |
| 994 result->contains_anchor = parser.contains_anchor(); |
| 995 result->capture_count = capture_count; |
| 996 } |
| 997 return !parser.failed(); |
| 998 } |
| 999 |
| 1000 |
| 1001 RegExpBuilder::RegExpBuilder(Zone* zone) |
| 1002 : zone_(zone), |
| 1003 pending_empty_(false), |
| 1004 characters_(NULL), |
| 1005 terms_(), |
| 1006 alternatives_() |
| 1007 #ifdef DEBUG |
| 1008 , |
| 1009 last_added_(ADD_NONE) |
| 1010 #endif |
| 1011 { |
| 1012 } |
| 1013 |
| 1014 |
| 1015 void RegExpBuilder::FlushCharacters() { |
| 1016 pending_empty_ = false; |
| 1017 if (characters_ != NULL) { |
| 1018 RegExpTree* atom = new (zone()) RegExpAtom(characters_->ToConstVector()); |
| 1019 characters_ = NULL; |
| 1020 text_.Add(atom, zone()); |
| 1021 LAST(ADD_ATOM); |
| 1022 } |
| 1023 } |
| 1024 |
| 1025 |
| 1026 void RegExpBuilder::FlushText() { |
| 1027 FlushCharacters(); |
| 1028 int num_text = text_.length(); |
| 1029 if (num_text == 0) { |
| 1030 return; |
| 1031 } else if (num_text == 1) { |
| 1032 terms_.Add(text_.last(), zone()); |
| 1033 } else { |
| 1034 RegExpText* text = new (zone()) RegExpText(zone()); |
| 1035 for (int i = 0; i < num_text; i++) text_.Get(i)->AppendToText(text, zone()); |
| 1036 terms_.Add(text, zone()); |
| 1037 } |
| 1038 text_.Clear(); |
| 1039 } |
| 1040 |
| 1041 |
| 1042 void RegExpBuilder::AddCharacter(uc16 c) { |
| 1043 pending_empty_ = false; |
| 1044 if (characters_ == NULL) { |
| 1045 characters_ = new (zone()) ZoneList<uc16>(4, zone()); |
| 1046 } |
| 1047 characters_->Add(c, zone()); |
| 1048 LAST(ADD_CHAR); |
| 1049 } |
| 1050 |
| 1051 |
| 1052 void RegExpBuilder::AddEmpty() { pending_empty_ = true; } |
| 1053 |
| 1054 |
| 1055 void RegExpBuilder::AddAtom(RegExpTree* term) { |
| 1056 if (term->IsEmpty()) { |
| 1057 AddEmpty(); |
| 1058 return; |
| 1059 } |
| 1060 if (term->IsTextElement()) { |
| 1061 FlushCharacters(); |
| 1062 text_.Add(term, zone()); |
| 1063 } else { |
| 1064 FlushText(); |
| 1065 terms_.Add(term, zone()); |
| 1066 } |
| 1067 LAST(ADD_ATOM); |
| 1068 } |
| 1069 |
| 1070 |
| 1071 void RegExpBuilder::AddAssertion(RegExpTree* assert) { |
| 1072 FlushText(); |
| 1073 terms_.Add(assert, zone()); |
| 1074 LAST(ADD_ASSERT); |
| 1075 } |
| 1076 |
| 1077 |
| 1078 void RegExpBuilder::NewAlternative() { FlushTerms(); } |
| 1079 |
| 1080 |
| 1081 void RegExpBuilder::FlushTerms() { |
| 1082 FlushText(); |
| 1083 int num_terms = terms_.length(); |
| 1084 RegExpTree* alternative; |
| 1085 if (num_terms == 0) { |
| 1086 alternative = new (zone()) RegExpEmpty(); |
| 1087 } else if (num_terms == 1) { |
| 1088 alternative = terms_.last(); |
| 1089 } else { |
| 1090 alternative = new (zone()) RegExpAlternative(terms_.GetList(zone())); |
| 1091 } |
| 1092 alternatives_.Add(alternative, zone()); |
| 1093 terms_.Clear(); |
| 1094 LAST(ADD_NONE); |
| 1095 } |
| 1096 |
| 1097 |
| 1098 RegExpTree* RegExpBuilder::ToRegExp() { |
| 1099 FlushTerms(); |
| 1100 int num_alternatives = alternatives_.length(); |
| 1101 if (num_alternatives == 0) return new (zone()) RegExpEmpty(); |
| 1102 if (num_alternatives == 1) return alternatives_.last(); |
| 1103 return new (zone()) RegExpDisjunction(alternatives_.GetList(zone())); |
| 1104 } |
| 1105 |
| 1106 |
| 1107 void RegExpBuilder::AddQuantifierToAtom( |
| 1108 int min, int max, RegExpQuantifier::QuantifierType quantifier_type) { |
| 1109 if (pending_empty_) { |
| 1110 pending_empty_ = false; |
| 1111 return; |
| 1112 } |
| 1113 RegExpTree* atom; |
| 1114 if (characters_ != NULL) { |
| 1115 DCHECK(last_added_ == ADD_CHAR); |
| 1116 // Last atom was character. |
| 1117 Vector<const uc16> char_vector = characters_->ToConstVector(); |
| 1118 int num_chars = char_vector.length(); |
| 1119 if (num_chars > 1) { |
| 1120 Vector<const uc16> prefix = char_vector.SubVector(0, num_chars - 1); |
| 1121 text_.Add(new (zone()) RegExpAtom(prefix), zone()); |
| 1122 char_vector = char_vector.SubVector(num_chars - 1, num_chars); |
| 1123 } |
| 1124 characters_ = NULL; |
| 1125 atom = new (zone()) RegExpAtom(char_vector); |
| 1126 FlushText(); |
| 1127 } else if (text_.length() > 0) { |
| 1128 DCHECK(last_added_ == ADD_ATOM); |
| 1129 atom = text_.RemoveLast(); |
| 1130 FlushText(); |
| 1131 } else if (terms_.length() > 0) { |
| 1132 DCHECK(last_added_ == ADD_ATOM); |
| 1133 atom = terms_.RemoveLast(); |
| 1134 if (atom->max_match() == 0) { |
| 1135 // Guaranteed to only match an empty string. |
| 1136 LAST(ADD_TERM); |
| 1137 if (min == 0) { |
| 1138 return; |
| 1139 } |
| 1140 terms_.Add(atom, zone()); |
| 1141 return; |
| 1142 } |
| 1143 } else { |
| 1144 // Only call immediately after adding an atom or character! |
| 1145 UNREACHABLE(); |
| 1146 return; |
| 1147 } |
| 1148 terms_.Add(new (zone()) RegExpQuantifier(min, max, quantifier_type, atom), |
| 1149 zone()); |
| 1150 LAST(ADD_TERM); |
| 1151 } |
| 1152 |
| 1153 } // namespace internal |
| 1154 } // namespace v8 |
OLD | NEW |