| OLD | NEW |
| 1 // Copyright (c) 2015, the Dart project authors. Please see the AUTHORS file | 1 // Copyright (c) 2015, the Dart project authors. Please see the AUTHORS file |
| 2 // for details. All rights reserved. Use of this source code is governed by a | 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. | 3 // BSD-style license that can be found in the LICENSE file. |
| 4 | 4 |
| 5 #include "vm/regexp_assembler_bytecode.h" | 5 #include "vm/regexp_assembler_bytecode.h" |
| 6 | 6 |
| 7 #include "vm/regexp_assembler_bytecode_inl.h" | 7 #include "vm/regexp_assembler_bytecode_inl.h" |
| 8 #include "vm/exceptions.h" | 8 #include "vm/exceptions.h" |
| 9 #include "vm/object_store.h" | 9 #include "vm/object_store.h" |
| 10 #include "vm/regexp_bytecodes.h" | 10 #include "vm/regexp_bytecodes.h" |
| 11 #include "vm/regexp_assembler.h" | 11 #include "vm/regexp_assembler.h" |
| 12 #include "vm/regexp.h" | 12 #include "vm/regexp.h" |
| 13 #include "vm/regexp_parser.h" | 13 #include "vm/regexp_parser.h" |
| 14 #include "vm/regexp_interpreter.h" | 14 #include "vm/regexp_interpreter.h" |
| 15 #include "vm/timeline.h" | 15 #include "vm/timeline.h" |
| 16 | 16 |
| 17 namespace dart { | 17 namespace dart { |
| 18 | 18 |
| 19 BytecodeRegExpMacroAssembler::BytecodeRegExpMacroAssembler( | 19 BytecodeRegExpMacroAssembler::BytecodeRegExpMacroAssembler( |
| 20 ZoneGrowableArray<uint8_t>* buffer, | 20 ZoneGrowableArray<uint8_t>* buffer, |
| 21 Zone* zone) | 21 Zone* zone) |
| 22 : RegExpMacroAssembler(zone), | 22 : RegExpMacroAssembler(zone), |
| 23 buffer_(buffer), | 23 buffer_(buffer), |
| 24 pc_(0), | 24 pc_(0), |
| 25 advance_current_end_(kInvalidPC) { } | 25 advance_current_end_(kInvalidPC) {} |
| 26 | 26 |
| 27 | 27 |
| 28 BytecodeRegExpMacroAssembler::~BytecodeRegExpMacroAssembler() { | 28 BytecodeRegExpMacroAssembler::~BytecodeRegExpMacroAssembler() { |
| 29 if (backtrack_.is_linked()) backtrack_.Unuse(); | 29 if (backtrack_.is_linked()) backtrack_.Unuse(); |
| 30 } | 30 } |
| 31 | 31 |
| 32 | 32 |
| 33 BytecodeRegExpMacroAssembler::IrregexpImplementation | 33 BytecodeRegExpMacroAssembler::IrregexpImplementation |
| 34 BytecodeRegExpMacroAssembler::Implementation() { | 34 BytecodeRegExpMacroAssembler::Implementation() { |
| 35 return kBytecodeImplementation; | 35 return kBytecodeImplementation; |
| (...skipping 38 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 74 | 74 |
| 75 | 75 |
| 76 void BytecodeRegExpMacroAssembler::PushRegister(intptr_t register_index) { | 76 void BytecodeRegExpMacroAssembler::PushRegister(intptr_t register_index) { |
| 77 ASSERT(register_index >= 0); | 77 ASSERT(register_index >= 0); |
| 78 ASSERT(register_index <= kMaxRegister); | 78 ASSERT(register_index <= kMaxRegister); |
| 79 Emit(BC_PUSH_REGISTER, register_index); | 79 Emit(BC_PUSH_REGISTER, register_index); |
| 80 } | 80 } |
| 81 | 81 |
| 82 | 82 |
| 83 void BytecodeRegExpMacroAssembler::WriteCurrentPositionToRegister( | 83 void BytecodeRegExpMacroAssembler::WriteCurrentPositionToRegister( |
| 84 intptr_t register_index, intptr_t cp_offset) { | 84 intptr_t register_index, |
| 85 intptr_t cp_offset) { |
| 85 ASSERT(register_index >= 0); | 86 ASSERT(register_index >= 0); |
| 86 ASSERT(register_index <= kMaxRegister); | 87 ASSERT(register_index <= kMaxRegister); |
| 87 Emit(BC_SET_REGISTER_TO_CP, register_index); | 88 Emit(BC_SET_REGISTER_TO_CP, register_index); |
| 88 Emit32(cp_offset); // Current position offset. | 89 Emit32(cp_offset); // Current position offset. |
| 89 } | 90 } |
| 90 | 91 |
| 91 | 92 |
| 92 void BytecodeRegExpMacroAssembler::ClearRegisters(intptr_t reg_from, | 93 void BytecodeRegExpMacroAssembler::ClearRegisters(intptr_t reg_from, |
| 93 intptr_t reg_to) { | 94 intptr_t reg_to) { |
| 94 ASSERT(reg_from <= reg_to); | 95 ASSERT(reg_from <= reg_to); |
| (...skipping 102 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 197 ASSERT(by >= kMinCPOffset); | 198 ASSERT(by >= kMinCPOffset); |
| 198 ASSERT(by <= kMaxCPOffset); | 199 ASSERT(by <= kMaxCPOffset); |
| 199 advance_current_start_ = pc_; | 200 advance_current_start_ = pc_; |
| 200 advance_current_offset_ = by; | 201 advance_current_offset_ = by; |
| 201 Emit(BC_ADVANCE_CP, by); | 202 Emit(BC_ADVANCE_CP, by); |
| 202 advance_current_end_ = pc_; | 203 advance_current_end_ = pc_; |
| 203 } | 204 } |
| 204 | 205 |
| 205 | 206 |
| 206 void BytecodeRegExpMacroAssembler::CheckGreedyLoop( | 207 void BytecodeRegExpMacroAssembler::CheckGreedyLoop( |
| 207 BlockLabel* on_tos_equals_current_position) { | 208 BlockLabel* on_tos_equals_current_position) { |
| 208 Emit(BC_CHECK_GREEDY, 0); | 209 Emit(BC_CHECK_GREEDY, 0); |
| 209 EmitOrLink(on_tos_equals_current_position); | 210 EmitOrLink(on_tos_equals_current_position); |
| 210 } | 211 } |
| 211 | 212 |
| 212 | 213 |
| 213 void BytecodeRegExpMacroAssembler::LoadCurrentCharacter(intptr_t cp_offset, | 214 void BytecodeRegExpMacroAssembler::LoadCurrentCharacter(intptr_t cp_offset, |
| 214 BlockLabel* on_failure, | 215 BlockLabel* on_failure, |
| 215 bool check_bounds, | 216 bool check_bounds, |
| 216 intptr_t characters) { | 217 intptr_t characters) { |
| 217 ASSERT(cp_offset >= kMinCPOffset); | 218 ASSERT(cp_offset >= kMinCPOffset); |
| (...skipping 131 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 349 uint16_t from, | 350 uint16_t from, |
| 350 uint16_t to, | 351 uint16_t to, |
| 351 BlockLabel* on_not_in_range) { | 352 BlockLabel* on_not_in_range) { |
| 352 Emit(BC_CHECK_CHAR_NOT_IN_RANGE, 0); | 353 Emit(BC_CHECK_CHAR_NOT_IN_RANGE, 0); |
| 353 Emit16(from); | 354 Emit16(from); |
| 354 Emit16(to); | 355 Emit16(to); |
| 355 EmitOrLink(on_not_in_range); | 356 EmitOrLink(on_not_in_range); |
| 356 } | 357 } |
| 357 | 358 |
| 358 | 359 |
| 359 void BytecodeRegExpMacroAssembler::CheckBitInTable( | 360 void BytecodeRegExpMacroAssembler::CheckBitInTable(const TypedData& table, |
| 360 const TypedData& table, BlockLabel* on_bit_set) { | 361 BlockLabel* on_bit_set) { |
| 361 Emit(BC_CHECK_BIT_IN_TABLE, 0); | 362 Emit(BC_CHECK_BIT_IN_TABLE, 0); |
| 362 EmitOrLink(on_bit_set); | 363 EmitOrLink(on_bit_set); |
| 363 for (int i = 0; i < kTableSize; i += kBitsPerByte) { | 364 for (int i = 0; i < kTableSize; i += kBitsPerByte) { |
| 364 int byte = 0; | 365 int byte = 0; |
| 365 for (int j = 0; j < kBitsPerByte; j++) { | 366 for (int j = 0; j < kBitsPerByte; j++) { |
| 366 if (table.GetUint8(i + j) != 0) byte |= 1 << j; | 367 if (table.GetUint8(i + j) != 0) byte |= 1 << j; |
| 367 } | 368 } |
| 368 Emit8(byte); | 369 Emit8(byte); |
| 369 } | 370 } |
| 370 } | 371 } |
| (...skipping 71 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 442 } | 443 } |
| 443 | 444 |
| 444 | 445 |
| 445 void BytecodeRegExpMacroAssembler::Expand() { | 446 void BytecodeRegExpMacroAssembler::Expand() { |
| 446 // BOGUS | 447 // BOGUS |
| 447 buffer_->Add(0); | 448 buffer_->Add(0); |
| 448 buffer_->Add(0); | 449 buffer_->Add(0); |
| 449 buffer_->Add(0); | 450 buffer_->Add(0); |
| 450 buffer_->Add(0); | 451 buffer_->Add(0); |
| 451 intptr_t x = buffer_->length(); | 452 intptr_t x = buffer_->length(); |
| 452 for (intptr_t i = 0; i < x; i++) buffer_->Add(0); | 453 for (intptr_t i = 0; i < x; i++) |
| 454 buffer_->Add(0); |
| 453 } | 455 } |
| 454 | 456 |
| 455 | 457 |
| 456 static intptr_t Prepare(const RegExp& regexp, | 458 static intptr_t Prepare(const RegExp& regexp, |
| 457 const String& subject, | 459 const String& subject, |
| 458 Zone* zone) { | 460 Zone* zone) { |
| 459 bool is_one_byte = subject.IsOneByteString() || | 461 bool is_one_byte = |
| 460 subject.IsExternalOneByteString(); | 462 subject.IsOneByteString() || subject.IsExternalOneByteString(); |
| 461 | 463 |
| 462 if (regexp.bytecode(is_one_byte) == TypedData::null()) { | 464 if (regexp.bytecode(is_one_byte) == TypedData::null()) { |
| 463 const String& pattern = String::Handle(zone, regexp.pattern()); | 465 const String& pattern = String::Handle(zone, regexp.pattern()); |
| 464 NOT_IN_PRODUCT(TimelineDurationScope tds(Thread::Current(), | 466 #if !defined(PRODUCT) |
| 465 Timeline::GetCompilerStream(), | 467 TimelineDurationScope tds(Thread::Current(), Timeline::GetCompilerStream(), |
| 466 "CompileIrregexpBytecode"); | 468 "CompileIrregexpBytecode"); |
| 467 if (tds.enabled()) { | 469 if (tds.enabled()) { |
| 468 tds.SetNumArguments(1); | 470 tds.SetNumArguments(1); |
| 469 tds.CopyArgument(0, "pattern", pattern.ToCString()); | 471 tds.CopyArgument(0, "pattern", pattern.ToCString()); |
| 470 }); // !PRODUCT | 472 } |
| 473 #endif // !defined(PRODUCT) |
| 471 | 474 |
| 472 const bool multiline = regexp.is_multi_line(); | 475 const bool multiline = regexp.is_multi_line(); |
| 473 RegExpCompileData* compile_data = new(zone) RegExpCompileData(); | 476 RegExpCompileData* compile_data = new (zone) RegExpCompileData(); |
| 474 if (!RegExpParser::ParseRegExp(pattern, multiline, compile_data)) { | 477 if (!RegExpParser::ParseRegExp(pattern, multiline, compile_data)) { |
| 475 // Parsing failures are handled in the RegExp factory constructor. | 478 // Parsing failures are handled in the RegExp factory constructor. |
| 476 UNREACHABLE(); | 479 UNREACHABLE(); |
| 477 } | 480 } |
| 478 | 481 |
| 479 regexp.set_num_bracket_expressions(compile_data->capture_count); | 482 regexp.set_num_bracket_expressions(compile_data->capture_count); |
| 480 if (compile_data->simple) { | 483 if (compile_data->simple) { |
| 481 regexp.set_is_simple(); | 484 regexp.set_is_simple(); |
| 482 } else { | 485 } else { |
| 483 regexp.set_is_complex(); | 486 regexp.set_is_complex(); |
| (...skipping 14 matching lines...) Expand all Loading... |
| 498 (Smi::Value(regexp.num_bracket_expressions()) + 1) * 2; | 501 (Smi::Value(regexp.num_bracket_expressions()) + 1) * 2; |
| 499 } | 502 } |
| 500 | 503 |
| 501 | 504 |
| 502 static IrregexpInterpreter::IrregexpResult ExecRaw(const RegExp& regexp, | 505 static IrregexpInterpreter::IrregexpResult ExecRaw(const RegExp& regexp, |
| 503 const String& subject, | 506 const String& subject, |
| 504 intptr_t index, | 507 intptr_t index, |
| 505 int32_t* output, | 508 int32_t* output, |
| 506 intptr_t output_size, | 509 intptr_t output_size, |
| 507 Zone* zone) { | 510 Zone* zone) { |
| 508 bool is_one_byte = subject.IsOneByteString() || | 511 bool is_one_byte = |
| 509 subject.IsExternalOneByteString(); | 512 subject.IsOneByteString() || subject.IsExternalOneByteString(); |
| 510 | 513 |
| 511 ASSERT(regexp.num_bracket_expressions() != Smi::null()); | 514 ASSERT(regexp.num_bracket_expressions() != Smi::null()); |
| 512 | 515 |
| 513 // We must have done EnsureCompiledIrregexp, so we can get the number of | 516 // We must have done EnsureCompiledIrregexp, so we can get the number of |
| 514 // registers. | 517 // registers. |
| 515 int number_of_capture_registers = | 518 int number_of_capture_registers = |
| 516 (Smi::Value(regexp.num_bracket_expressions()) + 1) * 2; | 519 (Smi::Value(regexp.num_bracket_expressions()) + 1) * 2; |
| 517 int32_t* raw_output = &output[number_of_capture_registers]; | 520 int32_t* raw_output = &output[number_of_capture_registers]; |
| 518 | 521 |
| 519 // We do not touch the actual capture result registers until we know there | 522 // We do not touch the actual capture result registers until we know there |
| 520 // has been a match so that we can use those capture results to set the | 523 // has been a match so that we can use those capture results to set the |
| 521 // last match info. | 524 // last match info. |
| 522 for (int i = number_of_capture_registers - 1; i >= 0; i--) { | 525 for (int i = number_of_capture_registers - 1; i >= 0; i--) { |
| 523 raw_output[i] = -1; | 526 raw_output[i] = -1; |
| 524 } | 527 } |
| 525 | 528 |
| 526 const TypedData& bytecode = | 529 const TypedData& bytecode = |
| (...skipping 24 matching lines...) Expand all Loading... |
| 551 Zone* zone) { | 554 Zone* zone) { |
| 552 intptr_t required_registers = Prepare(regexp, subject, zone); | 555 intptr_t required_registers = Prepare(regexp, subject, zone); |
| 553 if (required_registers < 0) { | 556 if (required_registers < 0) { |
| 554 // Compiling failed with an exception. | 557 // Compiling failed with an exception. |
| 555 UNREACHABLE(); | 558 UNREACHABLE(); |
| 556 } | 559 } |
| 557 | 560 |
| 558 // V8 uses a shared copy on the isolate when smaller than some threshold. | 561 // V8 uses a shared copy on the isolate when smaller than some threshold. |
| 559 int32_t* output_registers = zone->Alloc<int32_t>(required_registers); | 562 int32_t* output_registers = zone->Alloc<int32_t>(required_registers); |
| 560 | 563 |
| 561 IrregexpInterpreter::IrregexpResult result = ExecRaw(regexp, | 564 IrregexpInterpreter::IrregexpResult result = |
| 562 subject, | 565 ExecRaw(regexp, subject, start_index.Value(), output_registers, |
| 563 start_index.Value(), | 566 required_registers, zone); |
| 564 output_registers, | |
| 565 required_registers, | |
| 566 zone); | |
| 567 | 567 |
| 568 if (result == IrregexpInterpreter::RE_SUCCESS) { | 568 if (result == IrregexpInterpreter::RE_SUCCESS) { |
| 569 intptr_t capture_count = Smi::Value(regexp.num_bracket_expressions()); | 569 intptr_t capture_count = Smi::Value(regexp.num_bracket_expressions()); |
| 570 intptr_t capture_register_count = (capture_count + 1) * 2; | 570 intptr_t capture_register_count = (capture_count + 1) * 2; |
| 571 ASSERT(required_registers >= capture_register_count); | 571 ASSERT(required_registers >= capture_register_count); |
| 572 | 572 |
| 573 const TypedData& result = | 573 const TypedData& result = TypedData::Handle( |
| 574 TypedData::Handle(TypedData::New(kTypedDataInt32ArrayCid, | 574 TypedData::New(kTypedDataInt32ArrayCid, capture_register_count)); |
| 575 capture_register_count)); | |
| 576 { | 575 { |
| 577 #ifdef DEBUG | 576 #ifdef DEBUG |
| 578 // These indices will be used with substring operations that don't check | 577 // These indices will be used with substring operations that don't check |
| 579 // bounds, so sanity check them here. | 578 // bounds, so sanity check them here. |
| 580 for (intptr_t i = 0; i < capture_register_count; i++) { | 579 for (intptr_t i = 0; i < capture_register_count; i++) { |
| 581 int32_t val = output_registers[i]; | 580 int32_t val = output_registers[i]; |
| 582 ASSERT(val == -1 || (val >= 0 && val <= subject.Length())); | 581 ASSERT(val == -1 || (val >= 0 && val <= subject.Length())); |
| 583 } | 582 } |
| 584 #endif | 583 #endif |
| 585 | 584 |
| 586 NoSafepointScope no_safepoint; | 585 NoSafepointScope no_safepoint; |
| 587 memmove(result.DataAddr(0), | 586 memmove(result.DataAddr(0), output_registers, |
| 588 output_registers, | |
| 589 capture_register_count * sizeof(int32_t)); | 587 capture_register_count * sizeof(int32_t)); |
| 590 } | 588 } |
| 591 | 589 |
| 592 return result.raw(); | 590 return result.raw(); |
| 593 } | 591 } |
| 594 if (result == IrregexpInterpreter::RE_EXCEPTION) { | 592 if (result == IrregexpInterpreter::RE_EXCEPTION) { |
| 595 UNREACHABLE(); | 593 UNREACHABLE(); |
| 596 } | 594 } |
| 597 ASSERT(result == IrregexpInterpreter::RE_FAILURE); | 595 ASSERT(result == IrregexpInterpreter::RE_FAILURE); |
| 598 return Instance::null(); | 596 return Instance::null(); |
| 599 } | 597 } |
| 600 | 598 |
| 601 | 599 |
| 602 } // namespace dart | 600 } // namespace dart |
| OLD | NEW |