OLD | NEW |
(Empty) | |
| 1 // Copyright (c) 2014, the Dart project authors. Please see the AUTHORS file |
| 2 // for details. All rights reserved. Use of this source code is governed by a |
| 3 // BSD-style license that can be found in the LICENSE file. |
| 4 |
| 5 #include "vm/longjump.h" |
| 6 #include "vm/object_store.h" |
| 7 #include "vm/regexp_parser.h" |
| 8 |
| 9 namespace dart { |
| 10 |
| 11 #define I isolate() |
| 12 |
| 13 // Enables possessive quantifier syntax for testing. |
| 14 static const bool FLAG_regexp_possessive_quantifier = false; |
| 15 |
| 16 RegExpBuilder::RegExpBuilder() |
| 17 : isolate_(Isolate::Current()), |
| 18 pending_empty_(false), |
| 19 characters_(NULL), |
| 20 terms_(), |
| 21 text_(), |
| 22 alternatives_() |
| 23 #ifdef DEBUG |
| 24 , last_added_(ADD_NONE) |
| 25 #endif |
| 26 {} |
| 27 |
| 28 |
| 29 void RegExpBuilder::FlushCharacters() { |
| 30 pending_empty_ = false; |
| 31 if (characters_ != NULL) { |
| 32 RegExpTree* atom = new(I) RegExpAtom(characters_); |
| 33 characters_ = NULL; |
| 34 text_.Add(atom); |
| 35 LAST(ADD_ATOM); |
| 36 } |
| 37 } |
| 38 |
| 39 |
| 40 void RegExpBuilder::FlushText() { |
| 41 FlushCharacters(); |
| 42 intptr_t num_text = text_.length(); |
| 43 if (num_text == 0) { |
| 44 return; |
| 45 } else if (num_text == 1) { |
| 46 terms_.Add(text_.Last()); |
| 47 } else { |
| 48 RegExpText* text = new(I) RegExpText(); |
| 49 for (intptr_t i = 0; i < num_text; i++) |
| 50 text_[i]->AppendToText(text); |
| 51 terms_.Add(text); |
| 52 } |
| 53 text_.Clear(); |
| 54 } |
| 55 |
| 56 |
| 57 void RegExpBuilder::AddCharacter(uint16_t c) { |
| 58 pending_empty_ = false; |
| 59 if (characters_ == NULL) { |
| 60 characters_ = new(I) ZoneGrowableArray<uint16_t>(4); |
| 61 } |
| 62 characters_->Add(c); |
| 63 LAST(ADD_CHAR); |
| 64 } |
| 65 |
| 66 |
| 67 void RegExpBuilder::AddEmpty() { |
| 68 pending_empty_ = true; |
| 69 } |
| 70 |
| 71 |
| 72 void RegExpBuilder::AddAtom(RegExpTree* term) { |
| 73 if (term->IsEmpty()) { |
| 74 AddEmpty(); |
| 75 return; |
| 76 } |
| 77 if (term->IsTextElement()) { |
| 78 FlushCharacters(); |
| 79 text_.Add(term); |
| 80 } else { |
| 81 FlushText(); |
| 82 terms_.Add(term); |
| 83 } |
| 84 LAST(ADD_ATOM); |
| 85 } |
| 86 |
| 87 |
| 88 void RegExpBuilder::AddAssertion(RegExpTree* assert) { |
| 89 FlushText(); |
| 90 terms_.Add(assert); |
| 91 LAST(ADD_ASSERT); |
| 92 } |
| 93 |
| 94 |
| 95 void RegExpBuilder::NewAlternative() { |
| 96 FlushTerms(); |
| 97 } |
| 98 |
| 99 |
| 100 void RegExpBuilder::FlushTerms() { |
| 101 FlushText(); |
| 102 intptr_t num_terms = terms_.length(); |
| 103 RegExpTree* alternative; |
| 104 if (num_terms == 0) { |
| 105 alternative = RegExpEmpty::GetInstance(); |
| 106 } else if (num_terms == 1) { |
| 107 alternative = terms_.Last(); |
| 108 } else { |
| 109 ZoneGrowableArray<RegExpTree*>* terms = |
| 110 new(I) ZoneGrowableArray<RegExpTree*>(); |
| 111 for (intptr_t i = 0; i < terms_.length(); i++) { |
| 112 terms->Add(terms_[i]); |
| 113 } |
| 114 alternative = new(I) RegExpAlternative(terms); |
| 115 } |
| 116 alternatives_.Add(alternative); |
| 117 terms_.Clear(); |
| 118 LAST(ADD_NONE); |
| 119 } |
| 120 |
| 121 |
| 122 RegExpTree* RegExpBuilder::ToRegExp() { |
| 123 FlushTerms(); |
| 124 intptr_t num_alternatives = alternatives_.length(); |
| 125 if (num_alternatives == 0) { |
| 126 return RegExpEmpty::GetInstance(); |
| 127 } |
| 128 if (num_alternatives == 1) { |
| 129 return alternatives_.Last(); |
| 130 } |
| 131 ZoneGrowableArray<RegExpTree*>* alternatives = |
| 132 new(I) ZoneGrowableArray<RegExpTree*>(); |
| 133 for (intptr_t i = 0; i < alternatives_.length(); i++) { |
| 134 alternatives->Add(alternatives_[i]); |
| 135 } |
| 136 return new(I) RegExpDisjunction(alternatives); |
| 137 } |
| 138 |
| 139 |
| 140 void RegExpBuilder::AddQuantifierToAtom( |
| 141 intptr_t min, |
| 142 intptr_t max, |
| 143 RegExpQuantifier::QuantifierType quantifier_type) { |
| 144 if (pending_empty_) { |
| 145 pending_empty_ = false; |
| 146 return; |
| 147 } |
| 148 RegExpTree* atom; |
| 149 if (characters_ != NULL) { |
| 150 DEBUG_ASSERT(last_added_ == ADD_CHAR); |
| 151 // Last atom was character. |
| 152 |
| 153 ZoneGrowableArray<uint16_t> *char_vector = |
| 154 new(I) ZoneGrowableArray<uint16_t>(); |
| 155 char_vector->AddArray(*characters_); |
| 156 intptr_t num_chars = char_vector->length(); |
| 157 if (num_chars > 1) { |
| 158 ZoneGrowableArray<uint16_t> *prefix = |
| 159 new(I) ZoneGrowableArray<uint16_t>(); |
| 160 for (intptr_t i = 0; i < num_chars - 1; i++) { |
| 161 prefix->Add(char_vector->At(i)); |
| 162 } |
| 163 text_.Add(new(I) RegExpAtom(prefix)); |
| 164 ZoneGrowableArray<uint16_t> *tail = new(I) ZoneGrowableArray<uint16_t>(); |
| 165 tail->Add(char_vector->At(num_chars - 1)); |
| 166 char_vector = tail; |
| 167 } |
| 168 characters_ = NULL; |
| 169 atom = new(I) RegExpAtom(char_vector); |
| 170 FlushText(); |
| 171 } else if (text_.length() > 0) { |
| 172 DEBUG_ASSERT(last_added_ == ADD_ATOM); |
| 173 atom = text_.RemoveLast(); |
| 174 FlushText(); |
| 175 } else if (terms_.length() > 0) { |
| 176 DEBUG_ASSERT(last_added_ == ADD_ATOM); |
| 177 atom = terms_.RemoveLast(); |
| 178 if (atom->max_match() == 0) { |
| 179 // Guaranteed to only match an empty string. |
| 180 LAST(ADD_TERM); |
| 181 if (min == 0) { |
| 182 return; |
| 183 } |
| 184 terms_.Add(atom); |
| 185 return; |
| 186 } |
| 187 } else { |
| 188 // Only call immediately after adding an atom or character! |
| 189 UNREACHABLE(); |
| 190 return; |
| 191 } |
| 192 terms_.Add(new(I) RegExpQuantifier(min, max, quantifier_type, atom)); |
| 193 LAST(ADD_TERM); |
| 194 } |
| 195 |
| 196 // ---------------------------------------------------------------------------- |
| 197 // Implementation of Parser |
| 198 |
| 199 RegExpParser::RegExpParser(const String& in, |
| 200 String* error, |
| 201 bool multiline) |
| 202 : isolate_(Isolate::Current()), |
| 203 error_(error), |
| 204 captures_(NULL), |
| 205 in_(in), |
| 206 current_(kEndMarker), |
| 207 next_pos_(0), |
| 208 capture_count_(0), |
| 209 has_more_(true), |
| 210 multiline_(multiline), |
| 211 simple_(false), |
| 212 contains_anchor_(false), |
| 213 is_scanned_for_captures_(false), |
| 214 failed_(false) { |
| 215 Advance(); |
| 216 } |
| 217 |
| 218 |
| 219 uint32_t RegExpParser::Next() { |
| 220 if (has_next()) { |
| 221 return in().CharAt(next_pos_); |
| 222 } else { |
| 223 return kEndMarker; |
| 224 } |
| 225 } |
| 226 |
| 227 |
| 228 void RegExpParser::Advance() { |
| 229 if (next_pos_ < in().Length()) { |
| 230 current_ = in().CharAt(next_pos_); |
| 231 next_pos_++; |
| 232 } else { |
| 233 current_ = kEndMarker; |
| 234 has_more_ = false; |
| 235 } |
| 236 } |
| 237 |
| 238 |
| 239 void RegExpParser::Advance(intptr_t dist) { |
| 240 next_pos_ += dist - 1; |
| 241 Advance(); |
| 242 } |
| 243 |
| 244 |
| 245 void RegExpParser::Reset(intptr_t pos) { |
| 246 next_pos_ = pos; |
| 247 has_more_ = (pos < in().Length()); |
| 248 Advance(); |
| 249 } |
| 250 |
| 251 |
| 252 bool RegExpParser::ParseFunction(ParsedFunction *parsed_function) { |
| 253 Isolate* isolate = parsed_function->isolate(); |
| 254 JSRegExp& regexp = JSRegExp::Handle(parsed_function->function().regexp()); |
| 255 |
| 256 const String& pattern = String::Handle(regexp.pattern()); |
| 257 const bool multiline = regexp.is_multi_line(); |
| 258 |
| 259 RegExpCompileData* compile_data = new(isolate) RegExpCompileData(); |
| 260 if (!RegExpParser::ParseRegExp(pattern, multiline, compile_data)) { |
| 261 // Parsing failures are handled in the JSRegExp factory constructor. |
| 262 UNREACHABLE(); |
| 263 } |
| 264 |
| 265 regexp.set_num_bracket_expressions(compile_data->capture_count); |
| 266 if (compile_data->simple) { |
| 267 regexp.set_is_simple(); |
| 268 } else { |
| 269 regexp.set_is_complex(); |
| 270 } |
| 271 |
| 272 parsed_function->SetRegExpCompileData(compile_data); |
| 273 |
| 274 return true; |
| 275 } |
| 276 |
| 277 |
| 278 bool RegExpParser::ParseRegExp(const String& input, |
| 279 bool multiline, |
| 280 RegExpCompileData* result) { |
| 281 ASSERT(result != NULL); |
| 282 LongJumpScope jump; |
| 283 RegExpParser parser(input, &result->error, multiline); |
| 284 if (setjmp(*jump.Set()) == 0) { |
| 285 RegExpTree* tree = parser.ParsePattern(); |
| 286 ASSERT(tree != NULL); |
| 287 ASSERT(result->error.IsNull()); |
| 288 result->tree = tree; |
| 289 intptr_t capture_count = parser.captures_started(); |
| 290 result->simple = tree->IsAtom() && parser.simple() && capture_count == 0; |
| 291 result->contains_anchor = parser.contains_anchor(); |
| 292 result->capture_count = capture_count; |
| 293 } else { |
| 294 ASSERT(!result->error.IsNull()); |
| 295 Isolate::Current()->object_store()->clear_sticky_error(); |
| 296 |
| 297 // Throw a FormatException on parsing failures. |
| 298 const String& message = String::Handle( |
| 299 String::Concat(result->error, input)); |
| 300 const Array& args = Array::Handle(Array::New(1)); |
| 301 args.SetAt(0, message); |
| 302 |
| 303 Exceptions::ThrowByType(Exceptions::kFormat, args); |
| 304 } |
| 305 return !parser.failed(); |
| 306 } |
| 307 |
| 308 |
| 309 void RegExpParser::ReportError(const char* message) { |
| 310 failed_ = true; |
| 311 *error_ = String::New(message); |
| 312 // Zip to the end to make sure the no more input is read. |
| 313 current_ = kEndMarker; |
| 314 next_pos_ = in().Length(); |
| 315 |
| 316 const Error& error = Error::Handle(LanguageError::New(*error_)); |
| 317 Report::LongJump(error); |
| 318 UNREACHABLE(); |
| 319 } |
| 320 |
| 321 |
| 322 // Pattern :: |
| 323 // Disjunction |
| 324 RegExpTree* RegExpParser::ParsePattern() { |
| 325 RegExpTree* result = ParseDisjunction(); |
| 326 ASSERT(!has_more()); |
| 327 // If the result of parsing is a literal string atom, and it has the |
| 328 // same length as the input, then the atom is identical to the input. |
| 329 if (result->IsAtom() && result->AsAtom()->length() == in().Length()) { |
| 330 simple_ = true; |
| 331 } |
| 332 return result; |
| 333 } |
| 334 |
| 335 |
| 336 // Disjunction :: |
| 337 // Alternative |
| 338 // Alternative | Disjunction |
| 339 // Alternative :: |
| 340 // [empty] |
| 341 // Term Alternative |
| 342 // Term :: |
| 343 // Assertion |
| 344 // Atom |
| 345 // Atom Quantifier |
| 346 RegExpTree* RegExpParser::ParseDisjunction() { |
| 347 // Used to store current state while parsing subexpressions. |
| 348 RegExpParserState initial_state(NULL, INITIAL, 0, I); |
| 349 RegExpParserState* stored_state = &initial_state; |
| 350 // Cache the builder in a local variable for quick access. |
| 351 RegExpBuilder* builder = initial_state.builder(); |
| 352 while (true) { |
| 353 switch (current()) { |
| 354 case kEndMarker: |
| 355 if (stored_state->IsSubexpression()) { |
| 356 // Inside a parenthesized group when hitting end of input. |
| 357 ReportError("Unterminated group"); |
| 358 UNREACHABLE(); |
| 359 } |
| 360 ASSERT(INITIAL == stored_state->group_type()); |
| 361 // Parsing completed successfully. |
| 362 return builder->ToRegExp(); |
| 363 case ')': { |
| 364 if (!stored_state->IsSubexpression()) { |
| 365 ReportError("Unmatched ')'"); |
| 366 UNREACHABLE(); |
| 367 } |
| 368 ASSERT(INITIAL != stored_state->group_type()); |
| 369 |
| 370 Advance(); |
| 371 // End disjunction parsing and convert builder content to new single |
| 372 // regexp atom. |
| 373 RegExpTree* body = builder->ToRegExp(); |
| 374 |
| 375 intptr_t end_capture_index = captures_started(); |
| 376 |
| 377 intptr_t capture_index = stored_state->capture_index(); |
| 378 SubexpressionType group_type = stored_state->group_type(); |
| 379 |
| 380 // Restore previous state. |
| 381 stored_state = stored_state->previous_state(); |
| 382 builder = stored_state->builder(); |
| 383 |
| 384 // Build result of subexpression. |
| 385 if (group_type == CAPTURE) { |
| 386 RegExpCapture* capture = new(I) RegExpCapture(body, capture_index); |
| 387 (*captures_)[capture_index - 1] = capture; |
| 388 body = capture; |
| 389 } else if (group_type != GROUPING) { |
| 390 ASSERT(group_type == POSITIVE_LOOKAHEAD || |
| 391 group_type == NEGATIVE_LOOKAHEAD); |
| 392 bool is_positive = (group_type == POSITIVE_LOOKAHEAD); |
| 393 body = new(I) RegExpLookahead(body, |
| 394 is_positive, |
| 395 end_capture_index - capture_index, |
| 396 capture_index); |
| 397 } |
| 398 builder->AddAtom(body); |
| 399 // For compatibility with JSC and ES3, we allow quantifiers after |
| 400 // lookaheads, and break in all cases. |
| 401 break; |
| 402 } |
| 403 case '|': { |
| 404 Advance(); |
| 405 builder->NewAlternative(); |
| 406 continue; |
| 407 } |
| 408 case '*': |
| 409 case '+': |
| 410 case '?': |
| 411 ReportError("Nothing to repeat"); |
| 412 UNREACHABLE(); |
| 413 case '^': { |
| 414 Advance(); |
| 415 if (multiline_) { |
| 416 builder->AddAssertion( |
| 417 new(I) RegExpAssertion(RegExpAssertion::START_OF_LINE)); |
| 418 } else { |
| 419 builder->AddAssertion( |
| 420 new(I) RegExpAssertion(RegExpAssertion::START_OF_INPUT)); |
| 421 set_contains_anchor(); |
| 422 } |
| 423 continue; |
| 424 } |
| 425 case '$': { |
| 426 Advance(); |
| 427 RegExpAssertion::AssertionType assertion_type = |
| 428 multiline_ ? RegExpAssertion::END_OF_LINE : |
| 429 RegExpAssertion::END_OF_INPUT; |
| 430 builder->AddAssertion(new RegExpAssertion(assertion_type)); |
| 431 continue; |
| 432 } |
| 433 case '.': { |
| 434 Advance(); |
| 435 // everything except \x0a, \x0d, \u2028 and \u2029 |
| 436 ZoneGrowableArray<CharacterRange>* ranges = |
| 437 new ZoneGrowableArray<CharacterRange>(2); |
| 438 CharacterRange::AddClassEscape('.', ranges); |
| 439 RegExpTree* atom = new RegExpCharacterClass(ranges, false); |
| 440 builder->AddAtom(atom); |
| 441 break; |
| 442 } |
| 443 case '(': { |
| 444 SubexpressionType subexpr_type = CAPTURE; |
| 445 Advance(); |
| 446 if (current() == '?') { |
| 447 switch (Next()) { |
| 448 case ':': |
| 449 subexpr_type = GROUPING; |
| 450 break; |
| 451 case '=': |
| 452 subexpr_type = POSITIVE_LOOKAHEAD; |
| 453 break; |
| 454 case '!': |
| 455 subexpr_type = NEGATIVE_LOOKAHEAD; |
| 456 break; |
| 457 default: |
| 458 ReportError("Invalid group"); |
| 459 UNREACHABLE(); |
| 460 } |
| 461 Advance(2); |
| 462 } else { |
| 463 if (captures_ == NULL) { |
| 464 captures_ = new ZoneGrowableArray<RegExpCapture*>(2); |
| 465 } |
| 466 if (captures_started() >= kMaxCaptures) { |
| 467 ReportError("Too many captures"); |
| 468 UNREACHABLE(); |
| 469 } |
| 470 captures_->Add(NULL); |
| 471 } |
| 472 // Store current state and begin new disjunction parsing. |
| 473 stored_state = new RegExpParserState(stored_state, subexpr_type, |
| 474 captures_started(), I); |
| 475 builder = stored_state->builder(); |
| 476 continue; |
| 477 } |
| 478 case '[': { |
| 479 RegExpTree* atom = ParseCharacterClass(); |
| 480 builder->AddAtom(atom); |
| 481 break; |
| 482 } |
| 483 // Atom :: |
| 484 // \ AtomEscape |
| 485 case '\\': |
| 486 switch (Next()) { |
| 487 case kEndMarker: |
| 488 ReportError("\\ at end of pattern"); |
| 489 UNREACHABLE(); |
| 490 case 'b': |
| 491 Advance(2); |
| 492 builder->AddAssertion( |
| 493 new RegExpAssertion(RegExpAssertion::BOUNDARY)); |
| 494 continue; |
| 495 case 'B': |
| 496 Advance(2); |
| 497 builder->AddAssertion( |
| 498 new RegExpAssertion(RegExpAssertion::NON_BOUNDARY)); |
| 499 continue; |
| 500 // AtomEscape :: |
| 501 // CharacterClassEscape |
| 502 // |
| 503 // CharacterClassEscape :: one of |
| 504 // d D s S w W |
| 505 case 'd': case 'D': case 's': case 'S': case 'w': case 'W': { |
| 506 uint32_t c = Next(); |
| 507 Advance(2); |
| 508 ZoneGrowableArray<CharacterRange>* ranges = |
| 509 new ZoneGrowableArray<CharacterRange>(2); |
| 510 CharacterRange::AddClassEscape(c, ranges); |
| 511 RegExpTree* atom = new RegExpCharacterClass(ranges, false); |
| 512 builder->AddAtom(atom); |
| 513 break; |
| 514 } |
| 515 case '1': case '2': case '3': case '4': case '5': case '6': |
| 516 case '7': case '8': case '9': { |
| 517 intptr_t index = 0; |
| 518 if (ParseBackReferenceIndex(&index)) { |
| 519 RegExpCapture* capture = NULL; |
| 520 if (captures_ != NULL && index <= captures_->length()) { |
| 521 capture = captures_->At(index - 1); |
| 522 } |
| 523 if (capture == NULL) { |
| 524 builder->AddEmpty(); |
| 525 break; |
| 526 } |
| 527 RegExpTree* atom = new RegExpBackReference(capture); |
| 528 builder->AddAtom(atom); |
| 529 break; |
| 530 } |
| 531 uint32_t first_digit = Next(); |
| 532 if (first_digit == '8' || first_digit == '9') { |
| 533 // Treat as identity escape |
| 534 builder->AddCharacter(first_digit); |
| 535 Advance(2); |
| 536 break; |
| 537 } |
| 538 } |
| 539 // FALLTHROUGH |
| 540 case '0': { |
| 541 Advance(); |
| 542 uint32_t octal = ParseOctalLiteral(); |
| 543 builder->AddCharacter(octal); |
| 544 break; |
| 545 } |
| 546 // ControlEscape :: one of |
| 547 // f n r t v |
| 548 case 'f': |
| 549 Advance(2); |
| 550 builder->AddCharacter('\f'); |
| 551 break; |
| 552 case 'n': |
| 553 Advance(2); |
| 554 builder->AddCharacter('\n'); |
| 555 break; |
| 556 case 'r': |
| 557 Advance(2); |
| 558 builder->AddCharacter('\r'); |
| 559 break; |
| 560 case 't': |
| 561 Advance(2); |
| 562 builder->AddCharacter('\t'); |
| 563 break; |
| 564 case 'v': |
| 565 Advance(2); |
| 566 builder->AddCharacter('\v'); |
| 567 break; |
| 568 case 'c': { |
| 569 Advance(); |
| 570 uint32_t controlLetter = Next(); |
| 571 // Special case if it is an ASCII letter. |
| 572 // Convert lower case letters to uppercase. |
| 573 uint32_t letter = controlLetter & ~('a' ^ 'A'); |
| 574 if (letter < 'A' || 'Z' < letter) { |
| 575 // controlLetter is not in range 'A'-'Z' or 'a'-'z'. |
| 576 // This is outside the specification. We match JSC in |
| 577 // reading the backslash as a literal character instead |
| 578 // of as starting an escape. |
| 579 builder->AddCharacter('\\'); |
| 580 } else { |
| 581 Advance(2); |
| 582 builder->AddCharacter(controlLetter & 0x1f); |
| 583 } |
| 584 break; |
| 585 } |
| 586 case 'x': { |
| 587 Advance(2); |
| 588 uint32_t value; |
| 589 if (ParseHexEscape(2, &value)) { |
| 590 builder->AddCharacter(value); |
| 591 } else { |
| 592 builder->AddCharacter('x'); |
| 593 } |
| 594 break; |
| 595 } |
| 596 case 'u': { |
| 597 Advance(2); |
| 598 uint32_t value; |
| 599 if (ParseHexEscape(4, &value)) { |
| 600 builder->AddCharacter(value); |
| 601 } else { |
| 602 builder->AddCharacter('u'); |
| 603 } |
| 604 break; |
| 605 } |
| 606 default: |
| 607 // Identity escape. |
| 608 builder->AddCharacter(Next()); |
| 609 Advance(2); |
| 610 break; |
| 611 } |
| 612 break; |
| 613 case '{': { |
| 614 intptr_t dummy; |
| 615 if (ParseIntervalQuantifier(&dummy, &dummy)) { |
| 616 ReportError("Nothing to repeat"); |
| 617 UNREACHABLE(); |
| 618 } |
| 619 // fallthrough |
| 620 } |
| 621 default: |
| 622 builder->AddCharacter(current()); |
| 623 Advance(); |
| 624 break; |
| 625 } // end switch(current()) |
| 626 |
| 627 intptr_t min; |
| 628 intptr_t max; |
| 629 switch (current()) { |
| 630 // QuantifierPrefix :: |
| 631 // * |
| 632 // + |
| 633 // ? |
| 634 // { |
| 635 case '*': |
| 636 min = 0; |
| 637 max = RegExpTree::kInfinity; |
| 638 Advance(); |
| 639 break; |
| 640 case '+': |
| 641 min = 1; |
| 642 max = RegExpTree::kInfinity; |
| 643 Advance(); |
| 644 break; |
| 645 case '?': |
| 646 min = 0; |
| 647 max = 1; |
| 648 Advance(); |
| 649 break; |
| 650 case '{': |
| 651 if (ParseIntervalQuantifier(&min, &max)) { |
| 652 if (max < min) { |
| 653 ReportError("numbers out of order in {} quantifier."); |
| 654 UNREACHABLE(); |
| 655 } |
| 656 break; |
| 657 } else { |
| 658 continue; |
| 659 } |
| 660 default: |
| 661 continue; |
| 662 } |
| 663 RegExpQuantifier::QuantifierType quantifier_type = RegExpQuantifier::GREEDY; |
| 664 if (current() == '?') { |
| 665 quantifier_type = RegExpQuantifier::NON_GREEDY; |
| 666 Advance(); |
| 667 } else if (FLAG_regexp_possessive_quantifier && current() == '+') { |
| 668 // FLAG_regexp_possessive_quantifier is a debug-only flag. |
| 669 quantifier_type = RegExpQuantifier::POSSESSIVE; |
| 670 Advance(); |
| 671 } |
| 672 builder->AddQuantifierToAtom(min, max, quantifier_type); |
| 673 } |
| 674 } |
| 675 |
| 676 |
| 677 static const uint16_t kNoCharClass = 0; |
| 678 |
| 679 // Adds range or pre-defined character class to character ranges. |
| 680 // If char_class is not kInvalidClass, it's interpreted as a class |
| 681 // escape (i.e., 's' means whitespace, from '\s'). |
| 682 static inline void AddRangeOrEscape(ZoneGrowableArray<CharacterRange>* ranges, |
| 683 uint16_t char_class, |
| 684 CharacterRange range) { |
| 685 if (char_class != kNoCharClass) { |
| 686 CharacterRange::AddClassEscape(char_class, ranges); |
| 687 } else { |
| 688 ranges->Add(range); |
| 689 } |
| 690 } |
| 691 |
| 692 |
| 693 RegExpTree* RegExpParser::ParseCharacterClass() { |
| 694 static const char* kUnterminated = "Unterminated character class"; |
| 695 static const char* kRangeOutOfOrder = "Range out of order in character class"; |
| 696 |
| 697 ASSERT(current() == '['); |
| 698 Advance(); |
| 699 bool is_negated = false; |
| 700 if (current() == '^') { |
| 701 is_negated = true; |
| 702 Advance(); |
| 703 } |
| 704 ZoneGrowableArray<CharacterRange>* ranges = |
| 705 new(I) ZoneGrowableArray<CharacterRange>(2); |
| 706 while (has_more() && current() != ']') { |
| 707 uint16_t char_class = kNoCharClass; |
| 708 CharacterRange first = ParseClassAtom(&char_class); |
| 709 if (current() == '-') { |
| 710 Advance(); |
| 711 if (current() == kEndMarker) { |
| 712 // If we reach the end we break out of the loop and let the |
| 713 // following code report an error. |
| 714 break; |
| 715 } else if (current() == ']') { |
| 716 AddRangeOrEscape(ranges, char_class, first); |
| 717 ranges->Add(CharacterRange::Singleton('-')); |
| 718 break; |
| 719 } |
| 720 uint16_t char_class_2 = kNoCharClass; |
| 721 CharacterRange next = ParseClassAtom(&char_class_2); |
| 722 if (char_class != kNoCharClass || char_class_2 != kNoCharClass) { |
| 723 // Either end is an escaped character class. Treat the '-' verbatim. |
| 724 AddRangeOrEscape(ranges, char_class, first); |
| 725 ranges->Add(CharacterRange::Singleton('-')); |
| 726 AddRangeOrEscape(ranges, char_class_2, next); |
| 727 continue; |
| 728 } |
| 729 if (first.from() > next.to()) { |
| 730 ReportError(kRangeOutOfOrder); |
| 731 UNREACHABLE(); |
| 732 } |
| 733 ranges->Add(CharacterRange::Range(first.from(), next.to())); |
| 734 } else { |
| 735 AddRangeOrEscape(ranges, char_class, first); |
| 736 } |
| 737 } |
| 738 if (!has_more()) { |
| 739 ReportError(kUnterminated); |
| 740 UNREACHABLE(); |
| 741 } |
| 742 Advance(); |
| 743 if (ranges->length() == 0) { |
| 744 ranges->Add(CharacterRange::Everything()); |
| 745 is_negated = !is_negated; |
| 746 } |
| 747 return new(I) RegExpCharacterClass(ranges, is_negated); |
| 748 } |
| 749 |
| 750 |
| 751 #ifdef DEBUG |
| 752 // Currently only used in an ASSERT. |
| 753 static bool IsSpecialClassEscape(uint32_t c) { |
| 754 switch (c) { |
| 755 case 'd': case 'D': |
| 756 case 's': case 'S': |
| 757 case 'w': case 'W': |
| 758 return true; |
| 759 default: |
| 760 return false; |
| 761 } |
| 762 } |
| 763 #endif |
| 764 |
| 765 |
| 766 // In order to know whether an escape is a backreference or not we have to scan |
| 767 // the entire regexp and find the number of capturing parentheses. However we |
| 768 // don't want to scan the regexp twice unless it is necessary. This mini-parser |
| 769 // is called when needed. It can see the difference between capturing and |
| 770 // noncapturing parentheses and can skip character classes and backslash-escaped |
| 771 // characters. |
| 772 void RegExpParser::ScanForCaptures() { |
| 773 // Start with captures started previous to current position |
| 774 intptr_t capture_count = captures_started(); |
| 775 // Add count of captures after this position. |
| 776 intptr_t n; |
| 777 while ((n = current()) != kEndMarker) { |
| 778 Advance(); |
| 779 switch (n) { |
| 780 case '\\': |
| 781 Advance(); |
| 782 break; |
| 783 case '[': { |
| 784 intptr_t c; |
| 785 while ((c = current()) != kEndMarker) { |
| 786 Advance(); |
| 787 if (c == '\\') { |
| 788 Advance(); |
| 789 } else { |
| 790 if (c == ']') break; |
| 791 } |
| 792 } |
| 793 break; |
| 794 } |
| 795 case '(': |
| 796 if (current() != '?') capture_count++; |
| 797 break; |
| 798 } |
| 799 } |
| 800 capture_count_ = capture_count; |
| 801 is_scanned_for_captures_ = true; |
| 802 } |
| 803 |
| 804 |
| 805 static inline bool IsDecimalDigit(int32_t c) { |
| 806 return '0' <= c && c <= '9'; |
| 807 } |
| 808 |
| 809 |
| 810 bool RegExpParser::ParseBackReferenceIndex(intptr_t* index_out) { |
| 811 ASSERT('\\' == current()); |
| 812 ASSERT('1' <= Next() && Next() <= '9'); |
| 813 // Try to parse a decimal literal that is no greater than the total number |
| 814 // of left capturing parentheses in the input. |
| 815 intptr_t start = position(); |
| 816 intptr_t value = Next() - '0'; |
| 817 Advance(2); |
| 818 while (true) { |
| 819 uint32_t c = current(); |
| 820 if (IsDecimalDigit(c)) { |
| 821 value = 10 * value + (c - '0'); |
| 822 if (value > kMaxCaptures) { |
| 823 Reset(start); |
| 824 return false; |
| 825 } |
| 826 Advance(); |
| 827 } else { |
| 828 break; |
| 829 } |
| 830 } |
| 831 if (value > captures_started()) { |
| 832 if (!is_scanned_for_captures_) { |
| 833 intptr_t saved_position = position(); |
| 834 ScanForCaptures(); |
| 835 Reset(saved_position); |
| 836 } |
| 837 if (value > capture_count_) { |
| 838 Reset(start); |
| 839 return false; |
| 840 } |
| 841 } |
| 842 *index_out = value; |
| 843 return true; |
| 844 } |
| 845 |
| 846 |
| 847 // QuantifierPrefix :: |
| 848 // { DecimalDigits } |
| 849 // { DecimalDigits , } |
| 850 // { DecimalDigits , DecimalDigits } |
| 851 // |
| 852 // Returns true if parsing succeeds, and set the min_out and max_out |
| 853 // values. Values are truncated to RegExpTree::kInfinity if they overflow. |
| 854 bool RegExpParser::ParseIntervalQuantifier(intptr_t* min_out, |
| 855 intptr_t* max_out) { |
| 856 ASSERT(current() == '{'); |
| 857 intptr_t start = position(); |
| 858 Advance(); |
| 859 intptr_t min = 0; |
| 860 if (!IsDecimalDigit(current())) { |
| 861 Reset(start); |
| 862 return false; |
| 863 } |
| 864 while (IsDecimalDigit(current())) { |
| 865 intptr_t next = current() - '0'; |
| 866 if (min > (RegExpTree::kInfinity - next) / 10) { |
| 867 // Overflow. Skip past remaining decimal digits and return -1. |
| 868 do { |
| 869 Advance(); |
| 870 } while (IsDecimalDigit(current())); |
| 871 min = RegExpTree::kInfinity; |
| 872 break; |
| 873 } |
| 874 min = 10 * min + next; |
| 875 Advance(); |
| 876 } |
| 877 intptr_t max = 0; |
| 878 if (current() == '}') { |
| 879 max = min; |
| 880 Advance(); |
| 881 } else if (current() == ',') { |
| 882 Advance(); |
| 883 if (current() == '}') { |
| 884 max = RegExpTree::kInfinity; |
| 885 Advance(); |
| 886 } else { |
| 887 while (IsDecimalDigit(current())) { |
| 888 intptr_t next = current() - '0'; |
| 889 if (max > (RegExpTree::kInfinity - next) / 10) { |
| 890 do { |
| 891 Advance(); |
| 892 } while (IsDecimalDigit(current())); |
| 893 max = RegExpTree::kInfinity; |
| 894 break; |
| 895 } |
| 896 max = 10 * max + next; |
| 897 Advance(); |
| 898 } |
| 899 if (current() != '}') { |
| 900 Reset(start); |
| 901 return false; |
| 902 } |
| 903 Advance(); |
| 904 } |
| 905 } else { |
| 906 Reset(start); |
| 907 return false; |
| 908 } |
| 909 *min_out = min; |
| 910 *max_out = max; |
| 911 return true; |
| 912 } |
| 913 |
| 914 |
| 915 uint32_t RegExpParser::ParseOctalLiteral() { |
| 916 ASSERT(('0' <= current() && current() <= '7') || current() == kEndMarker); |
| 917 // For compatibility with some other browsers (not all), we parse |
| 918 // up to three octal digits with a value below 256. |
| 919 uint32_t value = current() - '0'; |
| 920 Advance(); |
| 921 if ('0' <= current() && current() <= '7') { |
| 922 value = value * 8 + current() - '0'; |
| 923 Advance(); |
| 924 if (value < 32 && '0' <= current() && current() <= '7') { |
| 925 value = value * 8 + current() - '0'; |
| 926 Advance(); |
| 927 } |
| 928 } |
| 929 return value; |
| 930 } |
| 931 |
| 932 |
| 933 // Returns the value (0 .. 15) of a hexadecimal character c. |
| 934 // If c is not a legal hexadecimal character, returns a value < 0. |
| 935 static inline intptr_t HexValue(uint32_t c) { |
| 936 c -= '0'; |
| 937 if (static_cast<unsigned>(c) <= 9) return c; |
| 938 c = (c | 0x20) - ('a' - '0'); // detect 0x11..0x16 and 0x31..0x36. |
| 939 if (static_cast<unsigned>(c) <= 5) return c + 10; |
| 940 return -1; |
| 941 } |
| 942 |
| 943 |
| 944 bool RegExpParser::ParseHexEscape(intptr_t length, uint32_t *value) { |
| 945 intptr_t start = position(); |
| 946 uint32_t val = 0; |
| 947 bool done = false; |
| 948 for (intptr_t i = 0; !done; i++) { |
| 949 uint32_t c = current(); |
| 950 intptr_t d = HexValue(c); |
| 951 if (d < 0) { |
| 952 Reset(start); |
| 953 return false; |
| 954 } |
| 955 val = val * 16 + d; |
| 956 Advance(); |
| 957 if (i == length - 1) { |
| 958 done = true; |
| 959 } |
| 960 } |
| 961 *value = val; |
| 962 return true; |
| 963 } |
| 964 |
| 965 |
| 966 uint32_t RegExpParser::ParseClassCharacterEscape() { |
| 967 ASSERT(current() == '\\'); |
| 968 DEBUG_ASSERT(has_next() && !IsSpecialClassEscape(Next())); |
| 969 Advance(); |
| 970 switch (current()) { |
| 971 case 'b': |
| 972 Advance(); |
| 973 return '\b'; |
| 974 // ControlEscape :: one of |
| 975 // f n r t v |
| 976 case 'f': |
| 977 Advance(); |
| 978 return '\f'; |
| 979 case 'n': |
| 980 Advance(); |
| 981 return '\n'; |
| 982 case 'r': |
| 983 Advance(); |
| 984 return '\r'; |
| 985 case 't': |
| 986 Advance(); |
| 987 return '\t'; |
| 988 case 'v': |
| 989 Advance(); |
| 990 return '\v'; |
| 991 case 'c': { |
| 992 uint32_t controlLetter = Next(); |
| 993 uint32_t letter = controlLetter & ~('A' ^ 'a'); |
| 994 // For compatibility with JSC, inside a character class |
| 995 // we also accept digits and underscore as control characters. |
| 996 if ((controlLetter >= '0' && controlLetter <= '9') || |
| 997 controlLetter == '_' || |
| 998 (letter >= 'A' && letter <= 'Z')) { |
| 999 Advance(2); |
| 1000 // Control letters mapped to ASCII control characters in the range |
| 1001 // 0x00-0x1f. |
| 1002 return controlLetter & 0x1f; |
| 1003 } |
| 1004 // We match JSC in reading the backslash as a literal |
| 1005 // character instead of as starting an escape. |
| 1006 return '\\'; |
| 1007 } |
| 1008 case '0': case '1': case '2': case '3': case '4': case '5': |
| 1009 case '6': case '7': |
| 1010 // For compatibility, we interpret a decimal escape that isn't |
| 1011 // a back reference (and therefore either \0 or not valid according |
| 1012 // to the specification) as a 1..3 digit octal character code. |
| 1013 return ParseOctalLiteral(); |
| 1014 case 'x': { |
| 1015 Advance(); |
| 1016 uint32_t value; |
| 1017 if (ParseHexEscape(2, &value)) { |
| 1018 return value; |
| 1019 } |
| 1020 // If \x is not followed by a two-digit hexadecimal, treat it |
| 1021 // as an identity escape. |
| 1022 return 'x'; |
| 1023 } |
| 1024 case 'u': { |
| 1025 Advance(); |
| 1026 uint32_t value; |
| 1027 if (ParseHexEscape(4, &value)) { |
| 1028 return value; |
| 1029 } |
| 1030 // If \u is not followed by a four-digit hexadecimal, treat it |
| 1031 // as an identity escape. |
| 1032 return 'u'; |
| 1033 } |
| 1034 default: { |
| 1035 // Extended identity escape. We accept any character that hasn't |
| 1036 // been matched by a more specific case, not just the subset required |
| 1037 // by the ECMAScript specification. |
| 1038 uint32_t result = current(); |
| 1039 Advance(); |
| 1040 return result; |
| 1041 } |
| 1042 } |
| 1043 return 0; |
| 1044 } |
| 1045 |
| 1046 |
| 1047 CharacterRange RegExpParser::ParseClassAtom(uint16_t* char_class) { |
| 1048 ASSERT(0 == *char_class); |
| 1049 uint32_t first = current(); |
| 1050 if (first == '\\') { |
| 1051 switch (Next()) { |
| 1052 case 'w': case 'W': case 'd': case 'D': case 's': case 'S': { |
| 1053 *char_class = Next(); |
| 1054 Advance(2); |
| 1055 return CharacterRange::Singleton(0); // Return dummy value. |
| 1056 } |
| 1057 case kEndMarker: |
| 1058 ReportError("\\ at end of pattern"); |
| 1059 UNREACHABLE(); |
| 1060 default: |
| 1061 uint32_t c = ParseClassCharacterEscape(); |
| 1062 return CharacterRange::Singleton(c); |
| 1063 } |
| 1064 } else { |
| 1065 Advance(); |
| 1066 return CharacterRange::Singleton(first); |
| 1067 } |
| 1068 } |
| 1069 |
| 1070 } // namespace dart |
OLD | NEW |