| 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/regexp.h" |
| 6 | |
| 7 | #include <memory> |
| 8 | |
| 9 | #include "platform/splay-tree-inl.h" |
| 10 | #include "platform/unicode.h" |
| 11 | |
| 12 | #include "unicode/uniset.h" |
| 13 | |
| 14 | #include "vm/dart_entry.h" |
| 15 | #include "vm/regexp_assembler.h" |
| 16 | #include "vm/regexp_assembler_bytecode.h" |
| 17 | #include "vm/regexp_ast.h" |
| 18 | #include "vm/symbols.h" |
| 19 | #include "vm/thread.h" |
| 20 | #include "vm/unibrow-inl.h" |
| 21 | |
| 22 | #if !defined(DART_PRECOMPILED_RUNTIME) |
| 23 | #include "vm/regexp_assembler_ir.h" |
| 24 | #endif // !defined(DART_PRECOMPILED_RUNTIME) |
| 25 | |
| 26 | #define Z (zone()) |
| 27 | |
| 28 | namespace dart { |
| 29 | |
| 30 | // Default to generating optimized regexp code. |
| 31 | static const bool kRegexpOptimization = true; |
| 32 | |
| 33 | // More makes code generation slower, less makes V8 benchmark score lower. |
| 34 | static const intptr_t kMaxLookaheadForBoyerMoore = 8; |
| 35 | |
| 36 | ContainedInLattice AddRange(ContainedInLattice containment, |
| 37 | const int32_t* ranges, |
| 38 | intptr_t ranges_length, |
| 39 | Interval new_range) { |
| 40 | ASSERT((ranges_length & 1) == 1); |
| 41 | ASSERT(ranges[ranges_length - 1] == Utf::kMaxCodePoint + 1); |
| 42 | if (containment == kLatticeUnknown) return containment; |
| 43 | bool inside = false; |
| 44 | int32_t last = 0; |
| 45 | for (intptr_t i = 0; i < ranges_length; |
| 46 | inside = !inside, last = ranges[i], i++) { |
| 47 | // Consider the range from last to ranges[i]. |
| 48 | // We haven't got to the new range yet. |
| 49 | if (ranges[i] <= new_range.from()) continue; |
| 50 | // New range is wholly inside last-ranges[i]. Note that new_range.to() is |
| 51 | // inclusive, but the values in ranges are not. |
| 52 | if (last <= new_range.from() && new_range.to() < ranges[i]) { |
| 53 | return Combine(containment, inside ? kLatticeIn : kLatticeOut); |
| 54 | } |
| 55 | return kLatticeUnknown; |
| 56 | } |
| 57 | return containment; |
| 58 | } |
| 59 | |
| 60 | // ------------------------------------------------------------------- |
| 61 | // Implementation of the Irregexp regular expression engine. |
| 62 | // |
| 63 | // The Irregexp regular expression engine is intended to be a complete |
| 64 | // implementation of ECMAScript regular expressions. It generates |
| 65 | // IR code that is subsequently compiled to native code. |
| 66 | |
| 67 | // The Irregexp regexp engine is structured in three steps. |
| 68 | // 1) The parser generates an abstract syntax tree. See regexp_ast.cc. |
| 69 | // 2) From the AST a node network is created. The nodes are all |
| 70 | // subclasses of RegExpNode. The nodes represent states when |
| 71 | // executing a regular expression. Several optimizations are |
| 72 | // performed on the node network. |
| 73 | // 3) From the nodes we generate IR instructions that can actually |
| 74 | // execute the regular expression (perform the search). The |
| 75 | // code generation step is described in more detail below. |
| 76 | |
| 77 | // Code generation. |
| 78 | // |
| 79 | // The nodes are divided into four main categories. |
| 80 | // * Choice nodes |
| 81 | // These represent places where the regular expression can |
| 82 | // match in more than one way. For example on entry to an |
| 83 | // alternation (foo|bar) or a repetition (*, +, ? or {}). |
| 84 | // * Action nodes |
| 85 | // These represent places where some action should be |
| 86 | // performed. Examples include recording the current position |
| 87 | // in the input string to a register (in order to implement |
| 88 | // captures) or other actions on register for example in order |
| 89 | // to implement the counters needed for {} repetitions. |
| 90 | // * Matching nodes |
| 91 | // These attempt to match some element part of the input string. |
| 92 | // Examples of elements include character classes, plain strings |
| 93 | // or back references. |
| 94 | // * End nodes |
| 95 | // These are used to implement the actions required on finding |
| 96 | // a successful match or failing to find a match. |
| 97 | // |
| 98 | // The code generated maintains some state as it runs. This consists of the |
| 99 | // following elements: |
| 100 | // |
| 101 | // * The capture registers. Used for string captures. |
| 102 | // * Other registers. Used for counters etc. |
| 103 | // * The current position. |
| 104 | // * The stack of backtracking information. Used when a matching node |
| 105 | // fails to find a match and needs to try an alternative. |
| 106 | // |
| 107 | // Conceptual regular expression execution model: |
| 108 | // |
| 109 | // There is a simple conceptual model of regular expression execution |
| 110 | // which will be presented first. The actual code generated is a more |
| 111 | // efficient simulation of the simple conceptual model: |
| 112 | // |
| 113 | // * Choice nodes are implemented as follows: |
| 114 | // For each choice except the last { |
| 115 | // push current position |
| 116 | // push backtrack code location |
| 117 | // <generate code to test for choice> |
| 118 | // backtrack code location: |
| 119 | // pop current position |
| 120 | // } |
| 121 | // <generate code to test for last choice> |
| 122 | // |
| 123 | // * Actions nodes are generated as follows |
| 124 | // <push affected registers on backtrack stack> |
| 125 | // <generate code to perform action> |
| 126 | // push backtrack code location |
| 127 | // <generate code to test for following nodes> |
| 128 | // backtrack code location: |
| 129 | // <pop affected registers to restore their state> |
| 130 | // <pop backtrack location from stack and go to it> |
| 131 | // |
| 132 | // * Matching nodes are generated as follows: |
| 133 | // if input string matches at current position |
| 134 | // update current position |
| 135 | // <generate code to test for following nodes> |
| 136 | // else |
| 137 | // <pop backtrack location from stack and go to it> |
| 138 | // |
| 139 | // Thus it can be seen that the current position is saved and restored |
| 140 | // by the choice nodes, whereas the registers are saved and restored by |
| 141 | // by the action nodes that manipulate them. |
| 142 | // |
| 143 | // The other interesting aspect of this model is that nodes are generated |
| 144 | // at the point where they are needed by a recursive call to Emit(). If |
| 145 | // the node has already been code generated then the Emit() call will |
| 146 | // generate a jump to the previously generated code instead. In order to |
| 147 | // limit recursion it is possible for the Emit() function to put the node |
| 148 | // on a work list for later generation and instead generate a jump. The |
| 149 | // destination of the jump is resolved later when the code is generated. |
| 150 | // |
| 151 | // Actual regular expression code generation. |
| 152 | // |
| 153 | // Code generation is actually more complicated than the above. In order |
| 154 | // to improve the efficiency of the generated code some optimizations are |
| 155 | // performed |
| 156 | // |
| 157 | // * Choice nodes have 1-character lookahead. |
| 158 | // A choice node looks at the following character and eliminates some of |
| 159 | // the choices immediately based on that character. This is not yet |
| 160 | // implemented. |
| 161 | // * Simple greedy loops store reduced backtracking information. |
| 162 | // A quantifier like /.*foo/m will greedily match the whole input. It will |
| 163 | // then need to backtrack to a point where it can match "foo". The naive |
| 164 | // implementation of this would push each character position onto the |
| 165 | // backtracking stack, then pop them off one by one. This would use space |
| 166 | // proportional to the length of the input string. However since the "." |
| 167 | // can only match in one way and always has a constant length (in this case |
| 168 | // of 1) it suffices to store the current position on the top of the stack |
| 169 | // once. Matching now becomes merely incrementing the current position and |
| 170 | // backtracking becomes decrementing the current position and checking the |
| 171 | // result against the stored current position. This is faster and saves |
| 172 | // space. |
| 173 | // * The current state is virtualized. |
| 174 | // This is used to defer expensive operations until it is clear that they |
| 175 | // are needed and to generate code for a node more than once, allowing |
| 176 | // specialized an efficient versions of the code to be created. This is |
| 177 | // explained in the section below. |
| 178 | // |
| 179 | // Execution state virtualization. |
| 180 | // |
| 181 | // Instead of emitting code, nodes that manipulate the state can record their |
| 182 | // manipulation in an object called the Trace. The Trace object can record a |
| 183 | // current position offset, an optional backtrack code location on the top of |
| 184 | // the virtualized backtrack stack and some register changes. When a node is |
| 185 | // to be emitted it can flush the Trace or update it. Flushing the Trace |
| 186 | // will emit code to bring the actual state into line with the virtual state. |
| 187 | // Avoiding flushing the state can postpone some work (e.g. updates of capture |
| 188 | // registers). Postponing work can save time when executing the regular |
| 189 | // expression since it may be found that the work never has to be done as a |
| 190 | // failure to match can occur. In addition it is much faster to jump to a |
| 191 | // known backtrack code location than it is to pop an unknown backtrack |
| 192 | // location from the stack and jump there. |
| 193 | // |
| 194 | // The virtual state found in the Trace affects code generation. For example |
| 195 | // the virtual state contains the difference between the actual current |
| 196 | // position and the virtual current position, and matching code needs to use |
| 197 | // this offset to attempt a match in the correct location of the input |
| 198 | // string. Therefore code generated for a non-trivial trace is specialized |
| 199 | // to that trace. The code generator therefore has the ability to generate |
| 200 | // code for each node several times. In order to limit the size of the |
| 201 | // generated code there is an arbitrary limit on how many specialized sets of |
| 202 | // code may be generated for a given node. If the limit is reached, the |
| 203 | // trace is flushed and a generic version of the code for a node is emitted. |
| 204 | // This is subsequently used for that node. The code emitted for non-generic |
| 205 | // trace is not recorded in the node and so it cannot currently be reused in |
| 206 | // the event that code generation is requested for an identical trace. |
| 207 | |
| 208 | void RegExpTree::AppendToText(RegExpText* text) { |
| 209 | UNREACHABLE(); |
| 210 | } |
| 211 | |
| 212 | void RegExpAtom::AppendToText(RegExpText* text) { |
| 213 | text->AddElement(TextElement::Atom(this)); |
| 214 | } |
| 215 | |
| 216 | void RegExpCharacterClass::AppendToText(RegExpText* text) { |
| 217 | text->AddElement(TextElement::CharClass(this)); |
| 218 | } |
| 219 | |
| 220 | void RegExpText::AppendToText(RegExpText* text) { |
| 221 | for (intptr_t i = 0; i < elements()->length(); i++) |
| 222 | text->AddElement((*elements())[i]); |
| 223 | } |
| 224 | |
| 225 | TextElement TextElement::Atom(RegExpAtom* atom) { |
| 226 | return TextElement(ATOM, atom); |
| 227 | } |
| 228 | |
| 229 | TextElement TextElement::CharClass(RegExpCharacterClass* char_class) { |
| 230 | return TextElement(CHAR_CLASS, char_class); |
| 231 | } |
| 232 | |
| 233 | intptr_t TextElement::length() const { |
| 234 | switch (text_type()) { |
| 235 | case ATOM: |
| 236 | return atom()->length(); |
| 237 | |
| 238 | case CHAR_CLASS: |
| 239 | return 1; |
| 240 | } |
| 241 | UNREACHABLE(); |
| 242 | return 0; |
| 243 | } |
| 244 | |
| 245 | class FrequencyCollator : public ValueObject { |
| 246 | public: |
| 247 | FrequencyCollator() : total_samples_(0) { |
| 248 | for (intptr_t i = 0; i < RegExpMacroAssembler::kTableSize; i++) { |
| 249 | frequencies_[i] = CharacterFrequency(i); |
| 250 | } |
| 251 | } |
| 252 | |
| 253 | void CountCharacter(intptr_t character) { |
| 254 | intptr_t index = (character & RegExpMacroAssembler::kTableMask); |
| 255 | frequencies_[index].Increment(); |
| 256 | total_samples_++; |
| 257 | } |
| 258 | |
| 259 | // Does not measure in percent, but rather per-128 (the table size from the |
| 260 | // regexp macro assembler). |
| 261 | intptr_t Frequency(intptr_t in_character) { |
| 262 | ASSERT((in_character & RegExpMacroAssembler::kTableMask) == in_character); |
| 263 | if (total_samples_ < 1) return 1; // Division by zero. |
| 264 | intptr_t freq_in_per128 = |
| 265 | (frequencies_[in_character].counter() * 128) / total_samples_; |
| 266 | return freq_in_per128; |
| 267 | } |
| 268 | |
| 269 | private: |
| 270 | class CharacterFrequency { |
| 271 | public: |
| 272 | CharacterFrequency() : counter_(0), character_(-1) {} |
| 273 | explicit CharacterFrequency(intptr_t character) |
| 274 | : counter_(0), character_(character) {} |
| 275 | |
| 276 | void Increment() { counter_++; } |
| 277 | intptr_t counter() { return counter_; } |
| 278 | intptr_t character() { return character_; } |
| 279 | |
| 280 | private: |
| 281 | intptr_t counter_; |
| 282 | intptr_t character_; |
| 283 | |
| 284 | DISALLOW_ALLOCATION(); |
| 285 | }; |
| 286 | |
| 287 | private: |
| 288 | CharacterFrequency frequencies_[RegExpMacroAssembler::kTableSize]; |
| 289 | intptr_t total_samples_; |
| 290 | }; |
| 291 | |
| 292 | class RegExpCompiler : public ValueObject { |
| 293 | public: |
| 294 | RegExpCompiler(intptr_t capture_count, bool is_one_byte); |
| 295 | |
| 296 | intptr_t AllocateRegister() { return next_register_++; } |
| 297 | |
| 298 | // Lookarounds to match lone surrogates for unicode character class matches |
| 299 | // are never nested. We can therefore reuse registers. |
| 300 | intptr_t UnicodeLookaroundStackRegister() { |
| 301 | if (unicode_lookaround_stack_register_ == kNoRegister) { |
| 302 | unicode_lookaround_stack_register_ = AllocateRegister(); |
| 303 | } |
| 304 | return unicode_lookaround_stack_register_; |
| 305 | } |
| 306 | |
| 307 | intptr_t UnicodeLookaroundPositionRegister() { |
| 308 | if (unicode_lookaround_position_register_ == kNoRegister) { |
| 309 | unicode_lookaround_position_register_ = AllocateRegister(); |
| 310 | } |
| 311 | return unicode_lookaround_position_register_; |
| 312 | } |
| 313 | |
| 314 | #if !defined(DART_PRECOMPILED_RUNTIME) |
| 315 | RegExpEngine::CompilationResult Assemble(IRRegExpMacroAssembler* assembler, |
| 316 | RegExpNode* start, |
| 317 | intptr_t capture_count, |
| 318 | const String& pattern); |
| 319 | #endif |
| 320 | |
| 321 | RegExpEngine::CompilationResult Assemble( |
| 322 | BytecodeRegExpMacroAssembler* assembler, |
| 323 | RegExpNode* start, |
| 324 | intptr_t capture_count, |
| 325 | const String& pattern); |
| 326 | |
| 327 | inline void AddWork(RegExpNode* node) { work_list_->Add(node); } |
| 328 | |
| 329 | static const intptr_t kImplementationOffset = 0; |
| 330 | static const intptr_t kNumberOfRegistersOffset = 0; |
| 331 | static const intptr_t kCodeOffset = 1; |
| 332 | |
| 333 | RegExpMacroAssembler* macro_assembler() { return macro_assembler_; } |
| 334 | EndNode* accept() { return accept_; } |
| 335 | |
| 336 | static const intptr_t kMaxRecursion = 100; |
| 337 | inline intptr_t recursion_depth() { return recursion_depth_; } |
| 338 | inline void IncrementRecursionDepth() { recursion_depth_++; } |
| 339 | inline void DecrementRecursionDepth() { recursion_depth_--; } |
| 340 | |
| 341 | void SetRegExpTooBig() { reg_exp_too_big_ = true; } |
| 342 | |
| 343 | inline bool one_byte() const { return is_one_byte_; } |
| 344 | bool read_backward() { return read_backward_; } |
| 345 | void set_read_backward(bool value) { read_backward_ = value; } |
| 346 | FrequencyCollator* frequency_collator() { return &frequency_collator_; } |
| 347 | |
| 348 | intptr_t current_expansion_factor() { return current_expansion_factor_; } |
| 349 | void set_current_expansion_factor(intptr_t value) { |
| 350 | current_expansion_factor_ = value; |
| 351 | } |
| 352 | |
| 353 | Zone* zone() const { return zone_; } |
| 354 | |
| 355 | static const intptr_t kNoRegister = -1; |
| 356 | |
| 357 | private: |
| 358 | EndNode* accept_; |
| 359 | intptr_t next_register_; |
| 360 | intptr_t unicode_lookaround_stack_register_; |
| 361 | intptr_t unicode_lookaround_position_register_; |
| 362 | ZoneGrowableArray<RegExpNode*>* work_list_; |
| 363 | intptr_t recursion_depth_; |
| 364 | RegExpMacroAssembler* macro_assembler_; |
| 365 | bool is_one_byte_; |
| 366 | bool reg_exp_too_big_; |
| 367 | bool read_backward_; |
| 368 | intptr_t current_expansion_factor_; |
| 369 | FrequencyCollator frequency_collator_; |
| 370 | Zone* zone_; |
| 371 | }; |
| 372 | |
| 373 | class RecursionCheck : public ValueObject { |
| 374 | public: |
| 375 | explicit RecursionCheck(RegExpCompiler* compiler) : compiler_(compiler) { |
| 376 | compiler->IncrementRecursionDepth(); |
| 377 | } |
| 378 | ~RecursionCheck() { compiler_->DecrementRecursionDepth(); } |
| 379 | |
| 380 | private: |
| 381 | RegExpCompiler* compiler_; |
| 382 | }; |
| 383 | |
| 384 | static RegExpEngine::CompilationResult IrregexpRegExpTooBig() { |
| 385 | return RegExpEngine::CompilationResult("RegExp too big"); |
| 386 | } |
| 387 | |
| 388 | // Attempts to compile the regexp using an Irregexp code generator. Returns |
| 389 | // a fixed array or a null handle depending on whether it succeeded. |
| 390 | RegExpCompiler::RegExpCompiler(intptr_t capture_count, bool is_one_byte) |
| 391 | : next_register_(2 * (capture_count + 1)), |
| 392 | unicode_lookaround_stack_register_(kNoRegister), |
| 393 | unicode_lookaround_position_register_(kNoRegister), |
| 394 | work_list_(NULL), |
| 395 | recursion_depth_(0), |
| 396 | is_one_byte_(is_one_byte), |
| 397 | reg_exp_too_big_(false), |
| 398 | read_backward_(false), |
| 399 | current_expansion_factor_(1), |
| 400 | zone_(Thread::Current()->zone()) { |
| 401 | accept_ = new (Z) EndNode(EndNode::ACCEPT, Z); |
| 402 | } |
| 403 | |
| 404 | #if !defined(DART_PRECOMPILED_RUNTIME) |
| 405 | RegExpEngine::CompilationResult RegExpCompiler::Assemble( |
| 406 | IRRegExpMacroAssembler* macro_assembler, |
| 407 | RegExpNode* start, |
| 408 | intptr_t capture_count, |
| 409 | const String& pattern) { |
| 410 | macro_assembler->set_slow_safe(false /* use_slow_safe_regexp_compiler */); |
| 411 | macro_assembler_ = macro_assembler; |
| 412 | |
| 413 | ZoneGrowableArray<RegExpNode*> work_list(0); |
| 414 | work_list_ = &work_list; |
| 415 | BlockLabel fail; |
| 416 | macro_assembler_->PushBacktrack(&fail); |
| 417 | Trace new_trace; |
| 418 | start->Emit(this, &new_trace); |
| 419 | macro_assembler_->BindBlock(&fail); |
| 420 | macro_assembler_->Fail(); |
| 421 | while (!work_list.is_empty()) { |
| 422 | work_list.RemoveLast()->Emit(this, &new_trace); |
| 423 | } |
| 424 | if (reg_exp_too_big_) return IrregexpRegExpTooBig(); |
| 425 | |
| 426 | macro_assembler->GenerateBacktrackBlock(); |
| 427 | macro_assembler->FinalizeRegistersArray(); |
| 428 | |
| 429 | return RegExpEngine::CompilationResult( |
| 430 | macro_assembler->backtrack_goto(), macro_assembler->graph_entry(), |
| 431 | macro_assembler->num_blocks(), macro_assembler->num_stack_locals(), |
| 432 | next_register_); |
| 433 | } |
| 434 | #endif |
| 435 | |
| 436 | RegExpEngine::CompilationResult RegExpCompiler::Assemble( |
| 437 | BytecodeRegExpMacroAssembler* macro_assembler, |
| 438 | RegExpNode* start, |
| 439 | intptr_t capture_count, |
| 440 | const String& pattern) { |
| 441 | macro_assembler->set_slow_safe(false /* use_slow_safe_regexp_compiler */); |
| 442 | macro_assembler_ = macro_assembler; |
| 443 | |
| 444 | ZoneGrowableArray<RegExpNode*> work_list(0); |
| 445 | work_list_ = &work_list; |
| 446 | BlockLabel fail; |
| 447 | macro_assembler_->PushBacktrack(&fail); |
| 448 | Trace new_trace; |
| 449 | start->Emit(this, &new_trace); |
| 450 | macro_assembler_->BindBlock(&fail); |
| 451 | macro_assembler_->Fail(); |
| 452 | while (!work_list.is_empty()) { |
| 453 | work_list.RemoveLast()->Emit(this, &new_trace); |
| 454 | } |
| 455 | if (reg_exp_too_big_) return IrregexpRegExpTooBig(); |
| 456 | |
| 457 | TypedData& bytecode = TypedData::ZoneHandle(macro_assembler->GetBytecode()); |
| 458 | return RegExpEngine::CompilationResult(&bytecode, next_register_); |
| 459 | } |
| 460 | |
| 461 | bool Trace::DeferredAction::Mentions(intptr_t that) { |
| 462 | if (action_type() == ActionNode::CLEAR_CAPTURES) { |
| 463 | Interval range = static_cast<DeferredClearCaptures*>(this)->range(); |
| 464 | return range.Contains(that); |
| 465 | } else { |
| 466 | return reg() == that; |
| 467 | } |
| 468 | } |
| 469 | |
| 470 | bool Trace::mentions_reg(intptr_t reg) { |
| 471 | for (DeferredAction* action = actions_; action != NULL; |
| 472 | action = action->next()) { |
| 473 | if (action->Mentions(reg)) return true; |
| 474 | } |
| 475 | return false; |
| 476 | } |
| 477 | |
| 478 | bool Trace::GetStoredPosition(intptr_t reg, intptr_t* cp_offset) { |
| 479 | ASSERT(*cp_offset == 0); |
| 480 | for (DeferredAction* action = actions_; action != NULL; |
| 481 | action = action->next()) { |
| 482 | if (action->Mentions(reg)) { |
| 483 | if (action->action_type() == ActionNode::STORE_POSITION) { |
| 484 | *cp_offset = static_cast<DeferredCapture*>(action)->cp_offset(); |
| 485 | return true; |
| 486 | } else { |
| 487 | return false; |
| 488 | } |
| 489 | } |
| 490 | } |
| 491 | return false; |
| 492 | } |
| 493 | |
| 494 | // This is called as we come into a loop choice node and some other tricky |
| 495 | // nodes. It normalizes the state of the code generator to ensure we can |
| 496 | // generate generic code. |
| 497 | intptr_t Trace::FindAffectedRegisters(OutSet* affected_registers, Zone* zone) { |
| 498 | intptr_t max_register = RegExpCompiler::kNoRegister; |
| 499 | for (DeferredAction* action = actions_; action != NULL; |
| 500 | action = action->next()) { |
| 501 | if (action->action_type() == ActionNode::CLEAR_CAPTURES) { |
| 502 | Interval range = static_cast<DeferredClearCaptures*>(action)->range(); |
| 503 | for (intptr_t i = range.from(); i <= range.to(); i++) |
| 504 | affected_registers->Set(i, zone); |
| 505 | if (range.to() > max_register) max_register = range.to(); |
| 506 | } else { |
| 507 | affected_registers->Set(action->reg(), zone); |
| 508 | if (action->reg() > max_register) max_register = action->reg(); |
| 509 | } |
| 510 | } |
| 511 | return max_register; |
| 512 | } |
| 513 | |
| 514 | void Trace::RestoreAffectedRegisters(RegExpMacroAssembler* assembler, |
| 515 | intptr_t max_register, |
| 516 | const OutSet& registers_to_pop, |
| 517 | const OutSet& registers_to_clear) { |
| 518 | for (intptr_t reg = max_register; reg >= 0; reg--) { |
| 519 | if (registers_to_pop.Get(reg)) { |
| 520 | assembler->PopRegister(reg); |
| 521 | } else if (registers_to_clear.Get(reg)) { |
| 522 | intptr_t clear_to = reg; |
| 523 | while (reg > 0 && registers_to_clear.Get(reg - 1)) { |
| 524 | reg--; |
| 525 | } |
| 526 | assembler->ClearRegisters(reg, clear_to); |
| 527 | } |
| 528 | } |
| 529 | } |
| 530 | |
| 531 | void Trace::PerformDeferredActions(RegExpMacroAssembler* assembler, |
| 532 | intptr_t max_register, |
| 533 | const OutSet& affected_registers, |
| 534 | OutSet* registers_to_pop, |
| 535 | OutSet* registers_to_clear, |
| 536 | Zone* zone) { |
| 537 | for (intptr_t reg = 0; reg <= max_register; reg++) { |
| 538 | if (!affected_registers.Get(reg)) { |
| 539 | continue; |
| 540 | } |
| 541 | |
| 542 | // The chronologically first deferred action in the trace |
| 543 | // is used to infer the action needed to restore a register |
| 544 | // to its previous state (or not, if it's safe to ignore it). |
| 545 | enum DeferredActionUndoType { ACTION_IGNORE, ACTION_RESTORE, ACTION_CLEAR }; |
| 546 | DeferredActionUndoType undo_action = ACTION_IGNORE; |
| 547 | |
| 548 | intptr_t value = 0; |
| 549 | bool absolute = false; |
| 550 | bool clear = false; |
| 551 | static const intptr_t kNoStore = kMinInt32; |
| 552 | intptr_t store_position = kNoStore; |
| 553 | // This is a little tricky because we are scanning the actions in reverse |
| 554 | // historical order (newest first). |
| 555 | for (DeferredAction* action = actions_; action != NULL; |
| 556 | action = action->next()) { |
| 557 | if (action->Mentions(reg)) { |
| 558 | switch (action->action_type()) { |
| 559 | case ActionNode::SET_REGISTER: { |
| 560 | Trace::DeferredSetRegister* psr = |
| 561 | static_cast<Trace::DeferredSetRegister*>(action); |
| 562 | if (!absolute) { |
| 563 | value += psr->value(); |
| 564 | absolute = true; |
| 565 | } |
| 566 | // SET_REGISTER is currently only used for newly introduced loop |
| 567 | // counters. They can have a significant previous value if they |
| 568 | // occour in a loop. TODO(lrn): Propagate this information, so we |
| 569 | // can set undo_action to ACTION_IGNORE if we know there is no |
| 570 | // value to restore. |
| 571 | undo_action = ACTION_RESTORE; |
| 572 | ASSERT(store_position == kNoStore); |
| 573 | ASSERT(!clear); |
| 574 | break; |
| 575 | } |
| 576 | case ActionNode::INCREMENT_REGISTER: |
| 577 | if (!absolute) { |
| 578 | value++; |
| 579 | } |
| 580 | ASSERT(store_position == kNoStore); |
| 581 | ASSERT(!clear); |
| 582 | undo_action = ACTION_RESTORE; |
| 583 | break; |
| 584 | case ActionNode::STORE_POSITION: { |
| 585 | Trace::DeferredCapture* pc = |
| 586 | static_cast<Trace::DeferredCapture*>(action); |
| 587 | if (!clear && store_position == kNoStore) { |
| 588 | store_position = pc->cp_offset(); |
| 589 | } |
| 590 | |
| 591 | // For captures we know that stores and clears alternate. |
| 592 | // Other register, are never cleared, and if the occur |
| 593 | // inside a loop, they might be assigned more than once. |
| 594 | if (reg <= 1) { |
| 595 | // Registers zero and one, aka "capture zero", is |
| 596 | // always set correctly if we succeed. There is no |
| 597 | // need to undo a setting on backtrack, because we |
| 598 | // will set it again or fail. |
| 599 | undo_action = ACTION_IGNORE; |
| 600 | } else { |
| 601 | undo_action = pc->is_capture() ? ACTION_CLEAR : ACTION_RESTORE; |
| 602 | } |
| 603 | ASSERT(!absolute); |
| 604 | ASSERT(value == 0); |
| 605 | break; |
| 606 | } |
| 607 | case ActionNode::CLEAR_CAPTURES: { |
| 608 | // Since we're scanning in reverse order, if we've already |
| 609 | // set the position we have to ignore historically earlier |
| 610 | // clearing operations. |
| 611 | if (store_position == kNoStore) { |
| 612 | clear = true; |
| 613 | } |
| 614 | undo_action = ACTION_RESTORE; |
| 615 | ASSERT(!absolute); |
| 616 | ASSERT(value == 0); |
| 617 | break; |
| 618 | } |
| 619 | default: |
| 620 | UNREACHABLE(); |
| 621 | break; |
| 622 | } |
| 623 | } |
| 624 | } |
| 625 | // Prepare for the undo-action (e.g., push if it's going to be popped). |
| 626 | if (undo_action == ACTION_RESTORE) { |
| 627 | assembler->PushRegister(reg); |
| 628 | registers_to_pop->Set(reg, zone); |
| 629 | } else if (undo_action == ACTION_CLEAR) { |
| 630 | registers_to_clear->Set(reg, zone); |
| 631 | } |
| 632 | // Perform the chronologically last action (or accumulated increment) |
| 633 | // for the register. |
| 634 | if (store_position != kNoStore) { |
| 635 | assembler->WriteCurrentPositionToRegister(reg, store_position); |
| 636 | } else if (clear) { |
| 637 | assembler->ClearRegisters(reg, reg); |
| 638 | } else if (absolute) { |
| 639 | assembler->SetRegister(reg, value); |
| 640 | } else if (value != 0) { |
| 641 | assembler->AdvanceRegister(reg, value); |
| 642 | } |
| 643 | } |
| 644 | } |
| 645 | |
| 646 | // This is called as we come into a loop choice node and some other tricky |
| 647 | // nodes. It normalizes the state of the code generator to ensure we can |
| 648 | // generate generic code. |
| 649 | void Trace::Flush(RegExpCompiler* compiler, RegExpNode* successor) { |
| 650 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 651 | |
| 652 | ASSERT(!is_trivial()); |
| 653 | |
| 654 | if (actions_ == NULL && backtrack() == NULL) { |
| 655 | // Here we just have some deferred cp advances to fix and we are back to |
| 656 | // a normal situation. We may also have to forget some information gained |
| 657 | // through a quick check that was already performed. |
| 658 | if (cp_offset_ != 0) assembler->AdvanceCurrentPosition(cp_offset_); |
| 659 | // Create a new trivial state and generate the node with that. |
| 660 | Trace new_state; |
| 661 | successor->Emit(compiler, &new_state); |
| 662 | return; |
| 663 | } |
| 664 | |
| 665 | // Generate deferred actions here along with code to undo them again. |
| 666 | OutSet affected_registers; |
| 667 | |
| 668 | if (backtrack() != NULL) { |
| 669 | // Here we have a concrete backtrack location. These are set up by choice |
| 670 | // nodes and so they indicate that we have a deferred save of the current |
| 671 | // position which we may need to emit here. |
| 672 | assembler->PushCurrentPosition(); |
| 673 | } |
| 674 | Zone* zone = successor->zone(); |
| 675 | intptr_t max_register = FindAffectedRegisters(&affected_registers, zone); |
| 676 | OutSet registers_to_pop; |
| 677 | OutSet registers_to_clear; |
| 678 | PerformDeferredActions(assembler, max_register, affected_registers, |
| 679 | ®isters_to_pop, ®isters_to_clear, zone); |
| 680 | if (cp_offset_ != 0) { |
| 681 | assembler->AdvanceCurrentPosition(cp_offset_); |
| 682 | } |
| 683 | |
| 684 | // Create a new trivial state and generate the node with that. |
| 685 | BlockLabel undo; |
| 686 | assembler->PushBacktrack(&undo); |
| 687 | Trace new_state; |
| 688 | successor->Emit(compiler, &new_state); |
| 689 | |
| 690 | // On backtrack we need to restore state. |
| 691 | assembler->BindBlock(&undo); |
| 692 | RestoreAffectedRegisters(assembler, max_register, registers_to_pop, |
| 693 | registers_to_clear); |
| 694 | if (backtrack() == NULL) { |
| 695 | assembler->Backtrack(); |
| 696 | } else { |
| 697 | assembler->PopCurrentPosition(); |
| 698 | assembler->GoTo(backtrack()); |
| 699 | } |
| 700 | } |
| 701 | |
| 702 | void NegativeSubmatchSuccess::Emit(RegExpCompiler* compiler, Trace* trace) { |
| 703 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 704 | |
| 705 | // Omit flushing the trace. We discard the entire stack frame anyway. |
| 706 | |
| 707 | if (!label()->is_bound()) { |
| 708 | // We are completely independent of the trace, since we ignore it, |
| 709 | // so this code can be used as the generic version. |
| 710 | assembler->BindBlock(label()); |
| 711 | } |
| 712 | |
| 713 | // Throw away everything on the backtrack stack since the start |
| 714 | // of the negative submatch and restore the character position. |
| 715 | assembler->ReadCurrentPositionFromRegister(current_position_register_); |
| 716 | assembler->ReadStackPointerFromRegister(stack_pointer_register_); |
| 717 | if (clear_capture_count_ > 0) { |
| 718 | // Clear any captures that might have been performed during the success |
| 719 | // of the body of the negative look-ahead. |
| 720 | int clear_capture_end = clear_capture_start_ + clear_capture_count_ - 1; |
| 721 | assembler->ClearRegisters(clear_capture_start_, clear_capture_end); |
| 722 | } |
| 723 | // Now that we have unwound the stack we find at the top of the stack the |
| 724 | // backtrack that the BeginSubmatch node got. |
| 725 | assembler->Backtrack(); |
| 726 | } |
| 727 | |
| 728 | void EndNode::Emit(RegExpCompiler* compiler, Trace* trace) { |
| 729 | if (!trace->is_trivial()) { |
| 730 | trace->Flush(compiler, this); |
| 731 | return; |
| 732 | } |
| 733 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 734 | if (!label()->is_bound()) { |
| 735 | assembler->BindBlock(label()); |
| 736 | } |
| 737 | switch (action_) { |
| 738 | case ACCEPT: |
| 739 | assembler->Succeed(); |
| 740 | return; |
| 741 | case BACKTRACK: |
| 742 | assembler->GoTo(trace->backtrack()); |
| 743 | return; |
| 744 | case NEGATIVE_SUBMATCH_SUCCESS: |
| 745 | // This case is handled in a different virtual method. |
| 746 | UNREACHABLE(); |
| 747 | } |
| 748 | UNIMPLEMENTED(); |
| 749 | } |
| 750 | |
| 751 | void GuardedAlternative::AddGuard(Guard* guard, Zone* zone) { |
| 752 | if (guards_ == NULL) guards_ = new (zone) ZoneGrowableArray<Guard*>(1); |
| 753 | guards_->Add(guard); |
| 754 | } |
| 755 | |
| 756 | ActionNode* ActionNode::SetRegister(intptr_t reg, |
| 757 | intptr_t val, |
| 758 | RegExpNode* on_success) { |
| 759 | ActionNode* result = |
| 760 | new (on_success->zone()) ActionNode(SET_REGISTER, on_success); |
| 761 | result->data_.u_store_register.reg = reg; |
| 762 | result->data_.u_store_register.value = val; |
| 763 | return result; |
| 764 | } |
| 765 | |
| 766 | ActionNode* ActionNode::IncrementRegister(intptr_t reg, |
| 767 | RegExpNode* on_success) { |
| 768 | ActionNode* result = |
| 769 | new (on_success->zone()) ActionNode(INCREMENT_REGISTER, on_success); |
| 770 | result->data_.u_increment_register.reg = reg; |
| 771 | return result; |
| 772 | } |
| 773 | |
| 774 | ActionNode* ActionNode::StorePosition(intptr_t reg, |
| 775 | bool is_capture, |
| 776 | RegExpNode* on_success) { |
| 777 | ActionNode* result = |
| 778 | new (on_success->zone()) ActionNode(STORE_POSITION, on_success); |
| 779 | result->data_.u_position_register.reg = reg; |
| 780 | result->data_.u_position_register.is_capture = is_capture; |
| 781 | return result; |
| 782 | } |
| 783 | |
| 784 | ActionNode* ActionNode::ClearCaptures(Interval range, RegExpNode* on_success) { |
| 785 | ActionNode* result = |
| 786 | new (on_success->zone()) ActionNode(CLEAR_CAPTURES, on_success); |
| 787 | result->data_.u_clear_captures.range_from = range.from(); |
| 788 | result->data_.u_clear_captures.range_to = range.to(); |
| 789 | return result; |
| 790 | } |
| 791 | |
| 792 | ActionNode* ActionNode::BeginSubmatch(intptr_t stack_reg, |
| 793 | intptr_t position_reg, |
| 794 | RegExpNode* on_success) { |
| 795 | ActionNode* result = |
| 796 | new (on_success->zone()) ActionNode(BEGIN_SUBMATCH, on_success); |
| 797 | result->data_.u_submatch.stack_pointer_register = stack_reg; |
| 798 | result->data_.u_submatch.current_position_register = position_reg; |
| 799 | return result; |
| 800 | } |
| 801 | |
| 802 | ActionNode* ActionNode::PositiveSubmatchSuccess(intptr_t stack_reg, |
| 803 | intptr_t position_reg, |
| 804 | intptr_t clear_register_count, |
| 805 | intptr_t clear_register_from, |
| 806 | RegExpNode* on_success) { |
| 807 | ActionNode* result = new (on_success->zone()) |
| 808 | ActionNode(POSITIVE_SUBMATCH_SUCCESS, on_success); |
| 809 | result->data_.u_submatch.stack_pointer_register = stack_reg; |
| 810 | result->data_.u_submatch.current_position_register = position_reg; |
| 811 | result->data_.u_submatch.clear_register_count = clear_register_count; |
| 812 | result->data_.u_submatch.clear_register_from = clear_register_from; |
| 813 | return result; |
| 814 | } |
| 815 | |
| 816 | ActionNode* ActionNode::EmptyMatchCheck(intptr_t start_register, |
| 817 | intptr_t repetition_register, |
| 818 | intptr_t repetition_limit, |
| 819 | RegExpNode* on_success) { |
| 820 | ActionNode* result = |
| 821 | new (on_success->zone()) ActionNode(EMPTY_MATCH_CHECK, on_success); |
| 822 | result->data_.u_empty_match_check.start_register = start_register; |
| 823 | result->data_.u_empty_match_check.repetition_register = repetition_register; |
| 824 | result->data_.u_empty_match_check.repetition_limit = repetition_limit; |
| 825 | return result; |
| 826 | } |
| 827 | |
| 828 | #define DEFINE_ACCEPT(Type) \ |
| 829 | void Type##Node::Accept(NodeVisitor* visitor) { visitor->Visit##Type(this); } |
| 830 | FOR_EACH_NODE_TYPE(DEFINE_ACCEPT) |
| 831 | #undef DEFINE_ACCEPT |
| 832 | |
| 833 | void LoopChoiceNode::Accept(NodeVisitor* visitor) { |
| 834 | visitor->VisitLoopChoice(this); |
| 835 | } |
| 836 | |
| 837 | // ------------------------------------------------------------------- |
| 838 | // Emit code. |
| 839 | |
| 840 | void ChoiceNode::GenerateGuard(RegExpMacroAssembler* macro_assembler, |
| 841 | Guard* guard, |
| 842 | Trace* trace) { |
| 843 | switch (guard->op()) { |
| 844 | case Guard::LT: |
| 845 | ASSERT(!trace->mentions_reg(guard->reg())); |
| 846 | macro_assembler->IfRegisterGE(guard->reg(), guard->value(), |
| 847 | trace->backtrack()); |
| 848 | break; |
| 849 | case Guard::GEQ: |
| 850 | ASSERT(!trace->mentions_reg(guard->reg())); |
| 851 | macro_assembler->IfRegisterLT(guard->reg(), guard->value(), |
| 852 | trace->backtrack()); |
| 853 | break; |
| 854 | } |
| 855 | } |
| 856 | |
| 857 | // Returns the number of characters in the equivalence class, omitting those |
| 858 | // that cannot occur in the source string because it is ASCII. |
| 859 | static intptr_t GetCaseIndependentLetters(uint16_t character, |
| 860 | bool one_byte_subject, |
| 861 | int32_t* letters) { |
| 862 | unibrow::Mapping<unibrow::Ecma262UnCanonicalize> jsregexp_uncanonicalize; |
| 863 | intptr_t length = jsregexp_uncanonicalize.get(character, '\0', letters); |
| 864 | // Unibrow returns 0 or 1 for characters where case independence is |
| 865 | // trivial. |
| 866 | if (length == 0) { |
| 867 | letters[0] = character; |
| 868 | length = 1; |
| 869 | } |
| 870 | if (!one_byte_subject || character <= Symbols::kMaxOneCharCodeSymbol) { |
| 871 | return length; |
| 872 | } |
| 873 | |
| 874 | // The standard requires that non-ASCII characters cannot have ASCII |
| 875 | // character codes in their equivalence class. |
| 876 | // TODO(dcarney): issue 3550 this is not actually true for Latin1 anymore, |
| 877 | // is it? For example, \u00C5 is equivalent to \u212B. |
| 878 | return 0; |
| 879 | } |
| 880 | |
| 881 | static inline bool EmitSimpleCharacter(Zone* zone, |
| 882 | RegExpCompiler* compiler, |
| 883 | uint16_t c, |
| 884 | BlockLabel* on_failure, |
| 885 | intptr_t cp_offset, |
| 886 | bool check, |
| 887 | bool preloaded) { |
| 888 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 889 | bool bound_checked = false; |
| 890 | if (!preloaded) { |
| 891 | assembler->LoadCurrentCharacter(cp_offset, on_failure, check); |
| 892 | bound_checked = true; |
| 893 | } |
| 894 | assembler->CheckNotCharacter(c, on_failure); |
| 895 | return bound_checked; |
| 896 | } |
| 897 | |
| 898 | // Only emits non-letters (things that don't have case). Only used for case |
| 899 | // independent matches. |
| 900 | static inline bool EmitAtomNonLetter(Zone* zone, |
| 901 | RegExpCompiler* compiler, |
| 902 | uint16_t c, |
| 903 | BlockLabel* on_failure, |
| 904 | intptr_t cp_offset, |
| 905 | bool check, |
| 906 | bool preloaded) { |
| 907 | RegExpMacroAssembler* macro_assembler = compiler->macro_assembler(); |
| 908 | bool one_byte = compiler->one_byte(); |
| 909 | int32_t chars[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 910 | intptr_t length = GetCaseIndependentLetters(c, one_byte, chars); |
| 911 | if (length < 1) { |
| 912 | // This can't match. Must be an one-byte subject and a non-one-byte |
| 913 | // character. We do not need to do anything since the one-byte pass |
| 914 | // already handled this. |
| 915 | return false; // Bounds not checked. |
| 916 | } |
| 917 | bool checked = false; |
| 918 | // We handle the length > 1 case in a later pass. |
| 919 | if (length == 1) { |
| 920 | if (one_byte && c > Symbols::kMaxOneCharCodeSymbol) { |
| 921 | // Can't match - see above. |
| 922 | return false; // Bounds not checked. |
| 923 | } |
| 924 | if (!preloaded) { |
| 925 | macro_assembler->LoadCurrentCharacter(cp_offset, on_failure, check); |
| 926 | checked = check; |
| 927 | } |
| 928 | macro_assembler->CheckNotCharacter(c, on_failure); |
| 929 | } |
| 930 | return checked; |
| 931 | } |
| 932 | |
| 933 | static bool ShortCutEmitCharacterPair(RegExpMacroAssembler* macro_assembler, |
| 934 | bool one_byte, |
| 935 | uint16_t c1, |
| 936 | uint16_t c2, |
| 937 | BlockLabel* on_failure) { |
| 938 | uint16_t char_mask; |
| 939 | if (one_byte) { |
| 940 | char_mask = Symbols::kMaxOneCharCodeSymbol; |
| 941 | } else { |
| 942 | char_mask = Utf16::kMaxCodeUnit; |
| 943 | } |
| 944 | uint16_t exor = c1 ^ c2; |
| 945 | // Check whether exor has only one bit set. |
| 946 | if (((exor - 1) & exor) == 0) { |
| 947 | // If c1 and c2 differ only by one bit. |
| 948 | // Ecma262UnCanonicalize always gives the highest number last. |
| 949 | ASSERT(c2 > c1); |
| 950 | uint16_t mask = char_mask ^ exor; |
| 951 | macro_assembler->CheckNotCharacterAfterAnd(c1, mask, on_failure); |
| 952 | return true; |
| 953 | } |
| 954 | ASSERT(c2 > c1); |
| 955 | uint16_t diff = c2 - c1; |
| 956 | if (((diff - 1) & diff) == 0 && c1 >= diff) { |
| 957 | // If the characters differ by 2^n but don't differ by one bit then |
| 958 | // subtract the difference from the found character, then do the or |
| 959 | // trick. We avoid the theoretical case where negative numbers are |
| 960 | // involved in order to simplify code generation. |
| 961 | uint16_t mask = char_mask ^ diff; |
| 962 | macro_assembler->CheckNotCharacterAfterMinusAnd(c1 - diff, diff, mask, |
| 963 | on_failure); |
| 964 | return true; |
| 965 | } |
| 966 | return false; |
| 967 | } |
| 968 | |
| 969 | typedef bool EmitCharacterFunction(Zone* zone, |
| 970 | RegExpCompiler* compiler, |
| 971 | uint16_t c, |
| 972 | BlockLabel* on_failure, |
| 973 | intptr_t cp_offset, |
| 974 | bool check, |
| 975 | bool preloaded); |
| 976 | |
| 977 | // Only emits letters (things that have case). Only used for case independent |
| 978 | // matches. |
| 979 | static inline bool EmitAtomLetter(Zone* zone, |
| 980 | RegExpCompiler* compiler, |
| 981 | uint16_t c, |
| 982 | BlockLabel* on_failure, |
| 983 | intptr_t cp_offset, |
| 984 | bool check, |
| 985 | bool preloaded) { |
| 986 | RegExpMacroAssembler* macro_assembler = compiler->macro_assembler(); |
| 987 | bool one_byte = compiler->one_byte(); |
| 988 | int32_t chars[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 989 | intptr_t length = GetCaseIndependentLetters(c, one_byte, chars); |
| 990 | if (length <= 1) return false; |
| 991 | // We may not need to check against the end of the input string |
| 992 | // if this character lies before a character that matched. |
| 993 | if (!preloaded) { |
| 994 | macro_assembler->LoadCurrentCharacter(cp_offset, on_failure, check); |
| 995 | } |
| 996 | BlockLabel ok; |
| 997 | ASSERT(unibrow::Ecma262UnCanonicalize::kMaxWidth == 4); |
| 998 | switch (length) { |
| 999 | case 2: { |
| 1000 | if (ShortCutEmitCharacterPair(macro_assembler, one_byte, chars[0], |
| 1001 | chars[1], on_failure)) { |
| 1002 | } else { |
| 1003 | macro_assembler->CheckCharacter(chars[0], &ok); |
| 1004 | macro_assembler->CheckNotCharacter(chars[1], on_failure); |
| 1005 | macro_assembler->BindBlock(&ok); |
| 1006 | } |
| 1007 | break; |
| 1008 | } |
| 1009 | case 4: |
| 1010 | macro_assembler->CheckCharacter(chars[3], &ok); |
| 1011 | FALL_THROUGH; |
| 1012 | case 3: |
| 1013 | macro_assembler->CheckCharacter(chars[0], &ok); |
| 1014 | macro_assembler->CheckCharacter(chars[1], &ok); |
| 1015 | macro_assembler->CheckNotCharacter(chars[2], on_failure); |
| 1016 | macro_assembler->BindBlock(&ok); |
| 1017 | break; |
| 1018 | default: |
| 1019 | UNREACHABLE(); |
| 1020 | break; |
| 1021 | } |
| 1022 | return true; |
| 1023 | } |
| 1024 | |
| 1025 | static void EmitBoundaryTest(RegExpMacroAssembler* masm, |
| 1026 | uint16_t border, |
| 1027 | BlockLabel* fall_through, |
| 1028 | BlockLabel* above_or_equal, |
| 1029 | BlockLabel* below) { |
| 1030 | if (below != fall_through) { |
| 1031 | masm->CheckCharacterLT(border, below); |
| 1032 | if (above_or_equal != fall_through) masm->GoTo(above_or_equal); |
| 1033 | } else { |
| 1034 | masm->CheckCharacterGT(border - 1, above_or_equal); |
| 1035 | } |
| 1036 | } |
| 1037 | |
| 1038 | static void EmitDoubleBoundaryTest(RegExpMacroAssembler* masm, |
| 1039 | uint16_t first, |
| 1040 | uint16_t last, |
| 1041 | BlockLabel* fall_through, |
| 1042 | BlockLabel* in_range, |
| 1043 | BlockLabel* out_of_range) { |
| 1044 | if (in_range == fall_through) { |
| 1045 | if (first == last) { |
| 1046 | masm->CheckNotCharacter(first, out_of_range); |
| 1047 | } else { |
| 1048 | masm->CheckCharacterNotInRange(first, last, out_of_range); |
| 1049 | } |
| 1050 | } else { |
| 1051 | if (first == last) { |
| 1052 | masm->CheckCharacter(first, in_range); |
| 1053 | } else { |
| 1054 | masm->CheckCharacterInRange(first, last, in_range); |
| 1055 | } |
| 1056 | if (out_of_range != fall_through) masm->GoTo(out_of_range); |
| 1057 | } |
| 1058 | } |
| 1059 | |
| 1060 | // even_label is for ranges[i] to ranges[i + 1] where i - start_index is even. |
| 1061 | // odd_label is for ranges[i] to ranges[i + 1] where i - start_index is odd. |
| 1062 | static void EmitUseLookupTable(RegExpMacroAssembler* masm, |
| 1063 | ZoneGrowableArray<uint16_t>* ranges, |
| 1064 | intptr_t start_index, |
| 1065 | intptr_t end_index, |
| 1066 | uint16_t min_char, |
| 1067 | BlockLabel* fall_through, |
| 1068 | BlockLabel* even_label, |
| 1069 | BlockLabel* odd_label) { |
| 1070 | static const intptr_t kSize = RegExpMacroAssembler::kTableSize; |
| 1071 | static const intptr_t kMask = RegExpMacroAssembler::kTableMask; |
| 1072 | |
| 1073 | intptr_t base = (min_char & ~kMask); |
| 1074 | |
| 1075 | // Assert that everything is on one kTableSize page. |
| 1076 | for (intptr_t i = start_index; i <= end_index; i++) { |
| 1077 | ASSERT((ranges->At(i) & ~kMask) == base); |
| 1078 | } |
| 1079 | ASSERT(start_index == 0 || (ranges->At(start_index - 1) & ~kMask) <= base); |
| 1080 | |
| 1081 | char templ[kSize]; |
| 1082 | BlockLabel* on_bit_set; |
| 1083 | BlockLabel* on_bit_clear; |
| 1084 | intptr_t bit; |
| 1085 | if (even_label == fall_through) { |
| 1086 | on_bit_set = odd_label; |
| 1087 | on_bit_clear = even_label; |
| 1088 | bit = 1; |
| 1089 | } else { |
| 1090 | on_bit_set = even_label; |
| 1091 | on_bit_clear = odd_label; |
| 1092 | bit = 0; |
| 1093 | } |
| 1094 | for (intptr_t i = 0; i < (ranges->At(start_index) & kMask) && i < kSize; |
| 1095 | i++) { |
| 1096 | templ[i] = bit; |
| 1097 | } |
| 1098 | intptr_t j = 0; |
| 1099 | bit ^= 1; |
| 1100 | for (intptr_t i = start_index; i < end_index; i++) { |
| 1101 | for (j = (ranges->At(i) & kMask); j < (ranges->At(i + 1) & kMask); j++) { |
| 1102 | templ[j] = bit; |
| 1103 | } |
| 1104 | bit ^= 1; |
| 1105 | } |
| 1106 | for (intptr_t i = j; i < kSize; i++) { |
| 1107 | templ[i] = bit; |
| 1108 | } |
| 1109 | // TODO(erikcorry): Cache these. |
| 1110 | const TypedData& ba = TypedData::ZoneHandle( |
| 1111 | masm->zone(), TypedData::New(kTypedDataUint8ArrayCid, kSize, Heap::kOld)); |
| 1112 | for (intptr_t i = 0; i < kSize; i++) { |
| 1113 | ba.SetUint8(i, templ[i]); |
| 1114 | } |
| 1115 | masm->CheckBitInTable(ba, on_bit_set); |
| 1116 | if (on_bit_clear != fall_through) masm->GoTo(on_bit_clear); |
| 1117 | } |
| 1118 | |
| 1119 | static void CutOutRange(RegExpMacroAssembler* masm, |
| 1120 | ZoneGrowableArray<uint16_t>* ranges, |
| 1121 | intptr_t start_index, |
| 1122 | intptr_t end_index, |
| 1123 | intptr_t cut_index, |
| 1124 | BlockLabel* even_label, |
| 1125 | BlockLabel* odd_label) { |
| 1126 | bool odd = (((cut_index - start_index) & 1) == 1); |
| 1127 | BlockLabel* in_range_label = odd ? odd_label : even_label; |
| 1128 | BlockLabel dummy; |
| 1129 | EmitDoubleBoundaryTest(masm, ranges->At(cut_index), |
| 1130 | ranges->At(cut_index + 1) - 1, &dummy, in_range_label, |
| 1131 | &dummy); |
| 1132 | ASSERT(!dummy.is_linked()); |
| 1133 | // Cut out the single range by rewriting the array. This creates a new |
| 1134 | // range that is a merger of the two ranges on either side of the one we |
| 1135 | // are cutting out. The oddity of the labels is preserved. |
| 1136 | for (intptr_t j = cut_index; j > start_index; j--) { |
| 1137 | (*ranges)[j] = ranges->At(j - 1); |
| 1138 | } |
| 1139 | for (intptr_t j = cut_index + 1; j < end_index; j++) { |
| 1140 | (*ranges)[j] = ranges->At(j + 1); |
| 1141 | } |
| 1142 | } |
| 1143 | |
| 1144 | // Unicode case. Split the search space into kSize spaces that are handled |
| 1145 | // with recursion. |
| 1146 | static void SplitSearchSpace(ZoneGrowableArray<uint16_t>* ranges, |
| 1147 | intptr_t start_index, |
| 1148 | intptr_t end_index, |
| 1149 | intptr_t* new_start_index, |
| 1150 | intptr_t* new_end_index, |
| 1151 | uint16_t* border) { |
| 1152 | static const intptr_t kSize = RegExpMacroAssembler::kTableSize; |
| 1153 | static const intptr_t kMask = RegExpMacroAssembler::kTableMask; |
| 1154 | |
| 1155 | uint16_t first = ranges->At(start_index); |
| 1156 | uint16_t last = ranges->At(end_index) - 1; |
| 1157 | |
| 1158 | *new_start_index = start_index; |
| 1159 | *border = (ranges->At(start_index) & ~kMask) + kSize; |
| 1160 | while (*new_start_index < end_index) { |
| 1161 | if (ranges->At(*new_start_index) > *border) break; |
| 1162 | (*new_start_index)++; |
| 1163 | } |
| 1164 | // new_start_index is the index of the first edge that is beyond the |
| 1165 | // current kSize space. |
| 1166 | |
| 1167 | // For very large search spaces we do a binary chop search of the non-Latin1 |
| 1168 | // space instead of just going to the end of the current kSize space. The |
| 1169 | // heuristics are complicated a little by the fact that any 128-character |
| 1170 | // encoding space can be quickly tested with a table lookup, so we don't |
| 1171 | // wish to do binary chop search at a smaller granularity than that. A |
| 1172 | // 128-character space can take up a lot of space in the ranges array if, |
| 1173 | // for example, we only want to match every second character (eg. the lower |
| 1174 | // case characters on some Unicode pages). |
| 1175 | intptr_t binary_chop_index = (end_index + start_index) / 2; |
| 1176 | // The first test ensures that we get to the code that handles the Latin1 |
| 1177 | // range with a single not-taken branch, speeding up this important |
| 1178 | // character range (even non-Latin1 charset-based text has spaces and |
| 1179 | // punctuation). |
| 1180 | if (*border - 1 > Symbols::kMaxOneCharCodeSymbol && // Latin1 case. |
| 1181 | end_index - start_index > (*new_start_index - start_index) * 2 && |
| 1182 | last - first > kSize * 2 && binary_chop_index > *new_start_index && |
| 1183 | ranges->At(binary_chop_index) >= first + 2 * kSize) { |
| 1184 | intptr_t scan_forward_for_section_border = binary_chop_index; |
| 1185 | intptr_t new_border = (ranges->At(binary_chop_index) | kMask) + 1; |
| 1186 | |
| 1187 | while (scan_forward_for_section_border < end_index) { |
| 1188 | if (ranges->At(scan_forward_for_section_border) > new_border) { |
| 1189 | *new_start_index = scan_forward_for_section_border; |
| 1190 | *border = new_border; |
| 1191 | break; |
| 1192 | } |
| 1193 | scan_forward_for_section_border++; |
| 1194 | } |
| 1195 | } |
| 1196 | |
| 1197 | ASSERT(*new_start_index > start_index); |
| 1198 | *new_end_index = *new_start_index - 1; |
| 1199 | if (ranges->At(*new_end_index) == *border) { |
| 1200 | (*new_end_index)--; |
| 1201 | } |
| 1202 | if (*border >= ranges->At(end_index)) { |
| 1203 | *border = ranges->At(end_index); |
| 1204 | *new_start_index = end_index; // Won't be used. |
| 1205 | *new_end_index = end_index - 1; |
| 1206 | } |
| 1207 | } |
| 1208 | |
| 1209 | // Gets a series of segment boundaries representing a character class. If the |
| 1210 | // character is in the range between an even and an odd boundary (counting from |
| 1211 | // start_index) then go to even_label, otherwise go to odd_label. We already |
| 1212 | // know that the character is in the range of min_char to max_char inclusive. |
| 1213 | // Either label can be NULL indicating backtracking. Either label can also be |
| 1214 | // equal to the fall_through label. |
| 1215 | static void GenerateBranches(RegExpMacroAssembler* masm, |
| 1216 | ZoneGrowableArray<uint16_t>* ranges, |
| 1217 | intptr_t start_index, |
| 1218 | intptr_t end_index, |
| 1219 | uint16_t min_char, |
| 1220 | uint16_t max_char, |
| 1221 | BlockLabel* fall_through, |
| 1222 | BlockLabel* even_label, |
| 1223 | BlockLabel* odd_label) { |
| 1224 | uint16_t first = ranges->At(start_index); |
| 1225 | uint16_t last = ranges->At(end_index) - 1; |
| 1226 | |
| 1227 | ASSERT(min_char < first); |
| 1228 | |
| 1229 | // Just need to test if the character is before or on-or-after |
| 1230 | // a particular character. |
| 1231 | if (start_index == end_index) { |
| 1232 | EmitBoundaryTest(masm, first, fall_through, even_label, odd_label); |
| 1233 | return; |
| 1234 | } |
| 1235 | |
| 1236 | // Another almost trivial case: There is one interval in the middle that is |
| 1237 | // different from the end intervals. |
| 1238 | if (start_index + 1 == end_index) { |
| 1239 | EmitDoubleBoundaryTest(masm, first, last, fall_through, even_label, |
| 1240 | odd_label); |
| 1241 | return; |
| 1242 | } |
| 1243 | |
| 1244 | // It's not worth using table lookup if there are very few intervals in the |
| 1245 | // character class. |
| 1246 | if (end_index - start_index <= 6) { |
| 1247 | // It is faster to test for individual characters, so we look for those |
| 1248 | // first, then try arbitrary ranges in the second round. |
| 1249 | static intptr_t kNoCutIndex = -1; |
| 1250 | intptr_t cut = kNoCutIndex; |
| 1251 | for (intptr_t i = start_index; i < end_index; i++) { |
| 1252 | if (ranges->At(i) == ranges->At(i + 1) - 1) { |
| 1253 | cut = i; |
| 1254 | break; |
| 1255 | } |
| 1256 | } |
| 1257 | if (cut == kNoCutIndex) cut = start_index; |
| 1258 | CutOutRange(masm, ranges, start_index, end_index, cut, even_label, |
| 1259 | odd_label); |
| 1260 | ASSERT(end_index - start_index >= 2); |
| 1261 | GenerateBranches(masm, ranges, start_index + 1, end_index - 1, min_char, |
| 1262 | max_char, fall_through, even_label, odd_label); |
| 1263 | return; |
| 1264 | } |
| 1265 | |
| 1266 | // If there are a lot of intervals in the regexp, then we will use tables to |
| 1267 | // determine whether the character is inside or outside the character class. |
| 1268 | static const intptr_t kBits = RegExpMacroAssembler::kTableSizeBits; |
| 1269 | |
| 1270 | if ((max_char >> kBits) == (min_char >> kBits)) { |
| 1271 | EmitUseLookupTable(masm, ranges, start_index, end_index, min_char, |
| 1272 | fall_through, even_label, odd_label); |
| 1273 | return; |
| 1274 | } |
| 1275 | |
| 1276 | if ((min_char >> kBits) != (first >> kBits)) { |
| 1277 | masm->CheckCharacterLT(first, odd_label); |
| 1278 | GenerateBranches(masm, ranges, start_index + 1, end_index, first, max_char, |
| 1279 | fall_through, odd_label, even_label); |
| 1280 | return; |
| 1281 | } |
| 1282 | |
| 1283 | intptr_t new_start_index = 0; |
| 1284 | intptr_t new_end_index = 0; |
| 1285 | uint16_t border = 0; |
| 1286 | |
| 1287 | SplitSearchSpace(ranges, start_index, end_index, &new_start_index, |
| 1288 | &new_end_index, &border); |
| 1289 | |
| 1290 | BlockLabel handle_rest; |
| 1291 | BlockLabel* above = &handle_rest; |
| 1292 | if (border == last + 1) { |
| 1293 | // We didn't find any section that started after the limit, so everything |
| 1294 | // above the border is one of the terminal labels. |
| 1295 | above = (end_index & 1) != (start_index & 1) ? odd_label : even_label; |
| 1296 | ASSERT(new_end_index == end_index - 1); |
| 1297 | } |
| 1298 | |
| 1299 | ASSERT(start_index <= new_end_index); |
| 1300 | ASSERT(new_start_index <= end_index); |
| 1301 | ASSERT(start_index < new_start_index); |
| 1302 | ASSERT(new_end_index < end_index); |
| 1303 | ASSERT(new_end_index + 1 == new_start_index || |
| 1304 | (new_end_index + 2 == new_start_index && |
| 1305 | border == ranges->At(new_end_index + 1))); |
| 1306 | ASSERT(min_char < border - 1); |
| 1307 | ASSERT(border < max_char); |
| 1308 | ASSERT(ranges->At(new_end_index) < border); |
| 1309 | ASSERT(border < ranges->At(new_start_index) || |
| 1310 | (border == ranges->At(new_start_index) && |
| 1311 | new_start_index == end_index && new_end_index == end_index - 1 && |
| 1312 | border == last + 1)); |
| 1313 | ASSERT(new_start_index == 0 || border >= ranges->At(new_start_index - 1)); |
| 1314 | |
| 1315 | masm->CheckCharacterGT(border - 1, above); |
| 1316 | BlockLabel dummy; |
| 1317 | GenerateBranches(masm, ranges, start_index, new_end_index, min_char, |
| 1318 | border - 1, &dummy, even_label, odd_label); |
| 1319 | |
| 1320 | if (handle_rest.is_linked()) { |
| 1321 | masm->BindBlock(&handle_rest); |
| 1322 | bool flip = (new_start_index & 1) != (start_index & 1); |
| 1323 | GenerateBranches(masm, ranges, new_start_index, end_index, border, max_char, |
| 1324 | &dummy, flip ? odd_label : even_label, |
| 1325 | flip ? even_label : odd_label); |
| 1326 | } |
| 1327 | } |
| 1328 | |
| 1329 | static void EmitCharClass(RegExpMacroAssembler* macro_assembler, |
| 1330 | RegExpCharacterClass* cc, |
| 1331 | bool one_byte, |
| 1332 | BlockLabel* on_failure, |
| 1333 | intptr_t cp_offset, |
| 1334 | bool check_offset, |
| 1335 | bool preloaded, |
| 1336 | Zone* zone) { |
| 1337 | ZoneGrowableArray<CharacterRange>* ranges = cc->ranges(); |
| 1338 | if (!CharacterRange::IsCanonical(ranges)) { |
| 1339 | CharacterRange::Canonicalize(ranges); |
| 1340 | } |
| 1341 | |
| 1342 | uint16_t max_char; |
| 1343 | if (one_byte) { |
| 1344 | max_char = Symbols::kMaxOneCharCodeSymbol; |
| 1345 | } else { |
| 1346 | max_char = Utf16::kMaxCodeUnit; |
| 1347 | } |
| 1348 | |
| 1349 | intptr_t range_count = ranges->length(); |
| 1350 | |
| 1351 | intptr_t last_valid_range = range_count - 1; |
| 1352 | while (last_valid_range >= 0) { |
| 1353 | const CharacterRange& range = ranges->At(last_valid_range); |
| 1354 | if (range.from() <= max_char) { |
| 1355 | break; |
| 1356 | } |
| 1357 | last_valid_range--; |
| 1358 | } |
| 1359 | |
| 1360 | if (last_valid_range < 0) { |
| 1361 | if (!cc->is_negated()) { |
| 1362 | macro_assembler->GoTo(on_failure); |
| 1363 | } |
| 1364 | if (check_offset) { |
| 1365 | macro_assembler->CheckPosition(cp_offset, on_failure); |
| 1366 | } |
| 1367 | return; |
| 1368 | } |
| 1369 | |
| 1370 | if (last_valid_range == 0 && ranges->At(0).IsEverything(max_char)) { |
| 1371 | if (cc->is_negated()) { |
| 1372 | macro_assembler->GoTo(on_failure); |
| 1373 | } else { |
| 1374 | // This is a common case hit by non-anchored expressions. |
| 1375 | if (check_offset) { |
| 1376 | macro_assembler->CheckPosition(cp_offset, on_failure); |
| 1377 | } |
| 1378 | } |
| 1379 | return; |
| 1380 | } |
| 1381 | |
| 1382 | if (!preloaded) { |
| 1383 | macro_assembler->LoadCurrentCharacter(cp_offset, on_failure, check_offset); |
| 1384 | } |
| 1385 | |
| 1386 | if (cc->is_standard() && macro_assembler->CheckSpecialCharacterClass( |
| 1387 | cc->standard_type(), on_failure)) { |
| 1388 | return; |
| 1389 | } |
| 1390 | |
| 1391 | // A new list with ascending entries. Each entry is a code unit |
| 1392 | // where there is a boundary between code units that are part of |
| 1393 | // the class and code units that are not. Normally we insert an |
| 1394 | // entry at zero which goes to the failure label, but if there |
| 1395 | // was already one there we fall through for success on that entry. |
| 1396 | // Subsequent entries have alternating meaning (success/failure). |
| 1397 | ZoneGrowableArray<uint16_t>* range_boundaries = |
| 1398 | new (zone) ZoneGrowableArray<uint16_t>(last_valid_range); |
| 1399 | |
| 1400 | bool zeroth_entry_is_failure = !cc->is_negated(); |
| 1401 | |
| 1402 | for (intptr_t i = 0; i <= last_valid_range; i++) { |
| 1403 | const CharacterRange& range = ranges->At(i); |
| 1404 | if (range.from() == 0) { |
| 1405 | ASSERT(i == 0); |
| 1406 | zeroth_entry_is_failure = !zeroth_entry_is_failure; |
| 1407 | } else { |
| 1408 | range_boundaries->Add(range.from()); |
| 1409 | } |
| 1410 | if (range.to() + 1 <= max_char) { |
| 1411 | range_boundaries->Add(range.to() + 1); |
| 1412 | } |
| 1413 | } |
| 1414 | intptr_t end_index = range_boundaries->length() - 1; |
| 1415 | |
| 1416 | BlockLabel fall_through; |
| 1417 | GenerateBranches(macro_assembler, range_boundaries, |
| 1418 | 0, // start_index. |
| 1419 | end_index, |
| 1420 | 0, // min_char. |
| 1421 | max_char, &fall_through, |
| 1422 | zeroth_entry_is_failure ? &fall_through : on_failure, |
| 1423 | zeroth_entry_is_failure ? on_failure : &fall_through); |
| 1424 | macro_assembler->BindBlock(&fall_through); |
| 1425 | } |
| 1426 | |
| 1427 | RegExpNode::~RegExpNode() {} |
| 1428 | |
| 1429 | RegExpNode::LimitResult RegExpNode::LimitVersions(RegExpCompiler* compiler, |
| 1430 | Trace* trace) { |
| 1431 | // If we are generating a greedy loop then don't stop and don't reuse code. |
| 1432 | if (trace->stop_node() != NULL) { |
| 1433 | return CONTINUE; |
| 1434 | } |
| 1435 | |
| 1436 | RegExpMacroAssembler* macro_assembler = compiler->macro_assembler(); |
| 1437 | if (trace->is_trivial()) { |
| 1438 | if (label_.is_bound()) { |
| 1439 | // We are being asked to generate a generic version, but that's already |
| 1440 | // been done so just go to it. |
| 1441 | macro_assembler->GoTo(&label_); |
| 1442 | return DONE; |
| 1443 | } |
| 1444 | if (compiler->recursion_depth() >= RegExpCompiler::kMaxRecursion) { |
| 1445 | // To avoid too deep recursion we push the node to the work queue and just |
| 1446 | // generate a goto here. |
| 1447 | compiler->AddWork(this); |
| 1448 | macro_assembler->GoTo(&label_); |
| 1449 | return DONE; |
| 1450 | } |
| 1451 | // Generate generic version of the node and bind the label for later use. |
| 1452 | macro_assembler->BindBlock(&label_); |
| 1453 | return CONTINUE; |
| 1454 | } |
| 1455 | |
| 1456 | // We are being asked to make a non-generic version. Keep track of how many |
| 1457 | // non-generic versions we generate so as not to overdo it. |
| 1458 | trace_count_++; |
| 1459 | if (kRegexpOptimization && trace_count_ < kMaxCopiesCodeGenerated && |
| 1460 | compiler->recursion_depth() <= RegExpCompiler::kMaxRecursion) { |
| 1461 | return CONTINUE; |
| 1462 | } |
| 1463 | |
| 1464 | // If we get here code has been generated for this node too many times or |
| 1465 | // recursion is too deep. Time to switch to a generic version. The code for |
| 1466 | // generic versions above can handle deep recursion properly. |
| 1467 | trace->Flush(compiler, this); |
| 1468 | return DONE; |
| 1469 | } |
| 1470 | |
| 1471 | intptr_t ActionNode::EatsAtLeast(intptr_t still_to_find, |
| 1472 | intptr_t budget, |
| 1473 | bool not_at_start) { |
| 1474 | if (budget <= 0) return 0; |
| 1475 | if (action_type_ == POSITIVE_SUBMATCH_SUCCESS) return 0; // Rewinds input! |
| 1476 | return on_success()->EatsAtLeast(still_to_find, budget - 1, not_at_start); |
| 1477 | } |
| 1478 | |
| 1479 | void ActionNode::FillInBMInfo(intptr_t offset, |
| 1480 | intptr_t budget, |
| 1481 | BoyerMooreLookahead* bm, |
| 1482 | bool not_at_start) { |
| 1483 | if (action_type_ == BEGIN_SUBMATCH) { |
| 1484 | bm->SetRest(offset); |
| 1485 | } else if (action_type_ != POSITIVE_SUBMATCH_SUCCESS) { |
| 1486 | on_success()->FillInBMInfo(offset, budget - 1, bm, not_at_start); |
| 1487 | } |
| 1488 | SaveBMInfo(bm, not_at_start, offset); |
| 1489 | } |
| 1490 | |
| 1491 | intptr_t AssertionNode::EatsAtLeast(intptr_t still_to_find, |
| 1492 | intptr_t budget, |
| 1493 | bool not_at_start) { |
| 1494 | if (budget <= 0) return 0; |
| 1495 | // If we know we are not at the start and we are asked "how many characters |
| 1496 | // will you match if you succeed?" then we can answer anything since false |
| 1497 | // implies false. So lets just return the max answer (still_to_find) since |
| 1498 | // that won't prevent us from preloading a lot of characters for the other |
| 1499 | // branches in the node graph. |
| 1500 | if (assertion_type() == AT_START && not_at_start) return still_to_find; |
| 1501 | return on_success()->EatsAtLeast(still_to_find, budget - 1, not_at_start); |
| 1502 | } |
| 1503 | |
| 1504 | void AssertionNode::FillInBMInfo(intptr_t offset, |
| 1505 | intptr_t budget, |
| 1506 | BoyerMooreLookahead* bm, |
| 1507 | bool not_at_start) { |
| 1508 | // Match the behaviour of EatsAtLeast on this node. |
| 1509 | if (assertion_type() == AT_START && not_at_start) return; |
| 1510 | on_success()->FillInBMInfo(offset, budget - 1, bm, not_at_start); |
| 1511 | SaveBMInfo(bm, not_at_start, offset); |
| 1512 | } |
| 1513 | |
| 1514 | intptr_t BackReferenceNode::EatsAtLeast(intptr_t still_to_find, |
| 1515 | intptr_t budget, |
| 1516 | bool not_at_start) { |
| 1517 | if (read_backward()) return 0; |
| 1518 | if (budget <= 0) return 0; |
| 1519 | return on_success()->EatsAtLeast(still_to_find, budget - 1, not_at_start); |
| 1520 | } |
| 1521 | |
| 1522 | intptr_t TextNode::EatsAtLeast(intptr_t still_to_find, |
| 1523 | intptr_t budget, |
| 1524 | bool not_at_start) { |
| 1525 | if (read_backward()) return 0; |
| 1526 | intptr_t answer = Length(); |
| 1527 | if (answer >= still_to_find) return answer; |
| 1528 | if (budget <= 0) return answer; |
| 1529 | // We are not at start after this node so we set the last argument to 'true'. |
| 1530 | return answer + |
| 1531 | on_success()->EatsAtLeast(still_to_find - answer, budget - 1, true); |
| 1532 | } |
| 1533 | |
| 1534 | intptr_t NegativeLookaroundChoiceNode::EatsAtLeast(intptr_t still_to_find, |
| 1535 | intptr_t budget, |
| 1536 | bool not_at_start) { |
| 1537 | if (budget <= 0) return 0; |
| 1538 | // Alternative 0 is the negative lookahead, alternative 1 is what comes |
| 1539 | // afterwards. |
| 1540 | RegExpNode* node = (*alternatives_)[1].node(); |
| 1541 | return node->EatsAtLeast(still_to_find, budget - 1, not_at_start); |
| 1542 | } |
| 1543 | |
| 1544 | void NegativeLookaroundChoiceNode::GetQuickCheckDetails( |
| 1545 | QuickCheckDetails* details, |
| 1546 | RegExpCompiler* compiler, |
| 1547 | intptr_t filled_in, |
| 1548 | bool not_at_start) { |
| 1549 | // Alternative 0 is the negative lookahead, alternative 1 is what comes |
| 1550 | // afterwards. |
| 1551 | RegExpNode* node = (*alternatives_)[1].node(); |
| 1552 | return node->GetQuickCheckDetails(details, compiler, filled_in, not_at_start); |
| 1553 | } |
| 1554 | |
| 1555 | intptr_t ChoiceNode::EatsAtLeastHelper(intptr_t still_to_find, |
| 1556 | intptr_t budget, |
| 1557 | RegExpNode* ignore_this_node, |
| 1558 | bool not_at_start) { |
| 1559 | if (budget <= 0) return 0; |
| 1560 | intptr_t min = 100; |
| 1561 | intptr_t choice_count = alternatives_->length(); |
| 1562 | budget = (budget - 1) / choice_count; |
| 1563 | for (intptr_t i = 0; i < choice_count; i++) { |
| 1564 | RegExpNode* node = (*alternatives_)[i].node(); |
| 1565 | if (node == ignore_this_node) continue; |
| 1566 | intptr_t node_eats_at_least = |
| 1567 | node->EatsAtLeast(still_to_find, budget, not_at_start); |
| 1568 | if (node_eats_at_least < min) min = node_eats_at_least; |
| 1569 | if (min == 0) return 0; |
| 1570 | } |
| 1571 | return min; |
| 1572 | } |
| 1573 | |
| 1574 | intptr_t LoopChoiceNode::EatsAtLeast(intptr_t still_to_find, |
| 1575 | intptr_t budget, |
| 1576 | bool not_at_start) { |
| 1577 | return EatsAtLeastHelper(still_to_find, budget - 1, loop_node_, not_at_start); |
| 1578 | } |
| 1579 | |
| 1580 | intptr_t ChoiceNode::EatsAtLeast(intptr_t still_to_find, |
| 1581 | intptr_t budget, |
| 1582 | bool not_at_start) { |
| 1583 | return EatsAtLeastHelper(still_to_find, budget, NULL, not_at_start); |
| 1584 | } |
| 1585 | |
| 1586 | // Takes the left-most 1-bit and smears it out, setting all bits to its right. |
| 1587 | static inline uint32_t SmearBitsRight(uint32_t v) { |
| 1588 | v |= v >> 1; |
| 1589 | v |= v >> 2; |
| 1590 | v |= v >> 4; |
| 1591 | v |= v >> 8; |
| 1592 | v |= v >> 16; |
| 1593 | return v; |
| 1594 | } |
| 1595 | |
| 1596 | bool QuickCheckDetails::Rationalize(bool asc) { |
| 1597 | bool found_useful_op = false; |
| 1598 | uint32_t char_mask; |
| 1599 | if (asc) { |
| 1600 | char_mask = Symbols::kMaxOneCharCodeSymbol; |
| 1601 | } else { |
| 1602 | char_mask = Utf16::kMaxCodeUnit; |
| 1603 | } |
| 1604 | mask_ = 0; |
| 1605 | value_ = 0; |
| 1606 | intptr_t char_shift = 0; |
| 1607 | for (intptr_t i = 0; i < characters_; i++) { |
| 1608 | Position* pos = &positions_[i]; |
| 1609 | if ((pos->mask & Symbols::kMaxOneCharCodeSymbol) != 0) { |
| 1610 | found_useful_op = true; |
| 1611 | } |
| 1612 | mask_ |= (pos->mask & char_mask) << char_shift; |
| 1613 | value_ |= (pos->value & char_mask) << char_shift; |
| 1614 | char_shift += asc ? 8 : 16; |
| 1615 | } |
| 1616 | return found_useful_op; |
| 1617 | } |
| 1618 | |
| 1619 | bool RegExpNode::EmitQuickCheck(RegExpCompiler* compiler, |
| 1620 | Trace* bounds_check_trace, |
| 1621 | Trace* trace, |
| 1622 | bool preload_has_checked_bounds, |
| 1623 | BlockLabel* on_possible_success, |
| 1624 | QuickCheckDetails* details, |
| 1625 | bool fall_through_on_failure) { |
| 1626 | if (details->characters() == 0) return false; |
| 1627 | GetQuickCheckDetails(details, compiler, 0, |
| 1628 | trace->at_start() == Trace::FALSE_VALUE); |
| 1629 | if (details->cannot_match()) return false; |
| 1630 | if (!details->Rationalize(compiler->one_byte())) return false; |
| 1631 | ASSERT(details->characters() == 1 || |
| 1632 | compiler->macro_assembler()->CanReadUnaligned()); |
| 1633 | uint32_t mask = details->mask(); |
| 1634 | uint32_t value = details->value(); |
| 1635 | |
| 1636 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 1637 | |
| 1638 | if (trace->characters_preloaded() != details->characters()) { |
| 1639 | ASSERT(trace->cp_offset() == bounds_check_trace->cp_offset()); |
| 1640 | // We are attempting to preload the minimum number of characters |
| 1641 | // any choice would eat, so if the bounds check fails, then none of the |
| 1642 | // choices can succeed, so we can just immediately backtrack, rather |
| 1643 | // than go to the next choice. |
| 1644 | assembler->LoadCurrentCharacter( |
| 1645 | trace->cp_offset(), bounds_check_trace->backtrack(), |
| 1646 | !preload_has_checked_bounds, details->characters()); |
| 1647 | } |
| 1648 | |
| 1649 | bool need_mask = true; |
| 1650 | |
| 1651 | if (details->characters() == 1) { |
| 1652 | // If number of characters preloaded is 1 then we used a byte or 16 bit |
| 1653 | // load so the value is already masked down. |
| 1654 | uint32_t char_mask; |
| 1655 | if (compiler->one_byte()) { |
| 1656 | char_mask = Symbols::kMaxOneCharCodeSymbol; |
| 1657 | } else { |
| 1658 | char_mask = Utf16::kMaxCodeUnit; |
| 1659 | } |
| 1660 | if ((mask & char_mask) == char_mask) need_mask = false; |
| 1661 | mask &= char_mask; |
| 1662 | } else { |
| 1663 | // For 2-character preloads in one-byte mode or 1-character preloads in |
| 1664 | // two-byte mode we also use a 16 bit load with zero extend. |
| 1665 | if (details->characters() == 2 && compiler->one_byte()) { |
| 1666 | if ((mask & 0xffff) == 0xffff) need_mask = false; |
| 1667 | } else if (details->characters() == 1 && !compiler->one_byte()) { |
| 1668 | if ((mask & 0xffff) == 0xffff) need_mask = false; |
| 1669 | } else { |
| 1670 | if (mask == 0xffffffff) need_mask = false; |
| 1671 | } |
| 1672 | } |
| 1673 | |
| 1674 | if (fall_through_on_failure) { |
| 1675 | if (need_mask) { |
| 1676 | assembler->CheckCharacterAfterAnd(value, mask, on_possible_success); |
| 1677 | } else { |
| 1678 | assembler->CheckCharacter(value, on_possible_success); |
| 1679 | } |
| 1680 | } else { |
| 1681 | if (need_mask) { |
| 1682 | assembler->CheckNotCharacterAfterAnd(value, mask, trace->backtrack()); |
| 1683 | } else { |
| 1684 | assembler->CheckNotCharacter(value, trace->backtrack()); |
| 1685 | } |
| 1686 | } |
| 1687 | return true; |
| 1688 | } |
| 1689 | |
| 1690 | // Here is the meat of GetQuickCheckDetails (see also the comment on the |
| 1691 | // super-class in the .h file). |
| 1692 | // |
| 1693 | // We iterate along the text object, building up for each character a |
| 1694 | // mask and value that can be used to test for a quick failure to match. |
| 1695 | // The masks and values for the positions will be combined into a single |
| 1696 | // machine word for the current character width in order to be used in |
| 1697 | // generating a quick check. |
| 1698 | void TextNode::GetQuickCheckDetails(QuickCheckDetails* details, |
| 1699 | RegExpCompiler* compiler, |
| 1700 | intptr_t characters_filled_in, |
| 1701 | bool not_at_start) { |
| 1702 | #if defined(__GNUC__) && defined(__BYTE_ORDER__) |
| 1703 | // TODO(zerny): Make the combination code byte-order independent. |
| 1704 | ASSERT(details->characters() == 1 || |
| 1705 | (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)); |
| 1706 | #endif |
| 1707 | // Do not collect any quick check details if the text node reads backward, |
| 1708 | // since it reads in the opposite direction than we use for quick checks. |
| 1709 | if (read_backward()) return; |
| 1710 | ASSERT(characters_filled_in < details->characters()); |
| 1711 | intptr_t characters = details->characters(); |
| 1712 | int32_t char_mask; |
| 1713 | if (compiler->one_byte()) { |
| 1714 | char_mask = Symbols::kMaxOneCharCodeSymbol; |
| 1715 | } else { |
| 1716 | char_mask = Utf16::kMaxCodeUnit; |
| 1717 | } |
| 1718 | for (intptr_t k = 0; k < elms_->length(); k++) { |
| 1719 | TextElement elm = elms_->At(k); |
| 1720 | if (elm.text_type() == TextElement::ATOM) { |
| 1721 | ZoneGrowableArray<uint16_t>* quarks = elm.atom()->data(); |
| 1722 | for (intptr_t i = 0; i < characters && i < quarks->length(); i++) { |
| 1723 | QuickCheckDetails::Position* pos = |
| 1724 | details->positions(characters_filled_in); |
| 1725 | uint16_t c = quarks->At(i); |
| 1726 | if (c > char_mask) { |
| 1727 | // If we expect a non-Latin1 character from an one-byte string, |
| 1728 | // there is no way we can match. Not even case independent |
| 1729 | // matching can turn an Latin1 character into non-Latin1 or |
| 1730 | // vice versa. |
| 1731 | // TODO(dcarney): issue 3550. Verify that this works as expected. |
| 1732 | // For example, \u0178 is uppercase of \u00ff (y-umlaut). |
| 1733 | details->set_cannot_match(); |
| 1734 | pos->determines_perfectly = false; |
| 1735 | return; |
| 1736 | } |
| 1737 | if (elm.atom()->ignore_case()) { |
| 1738 | int32_t chars[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 1739 | intptr_t length = |
| 1740 | GetCaseIndependentLetters(c, compiler->one_byte(), chars); |
| 1741 | ASSERT(length != 0); // Can only happen if c > char_mask (see above). |
| 1742 | if (length == 1) { |
| 1743 | // This letter has no case equivalents, so it's nice and simple |
| 1744 | // and the mask-compare will determine definitely whether we have |
| 1745 | // a match at this character position. |
| 1746 | pos->mask = char_mask; |
| 1747 | pos->value = c; |
| 1748 | pos->determines_perfectly = true; |
| 1749 | } else { |
| 1750 | uint32_t common_bits = char_mask; |
| 1751 | uint32_t bits = chars[0]; |
| 1752 | for (intptr_t j = 1; j < length; j++) { |
| 1753 | uint32_t differing_bits = ((chars[j] & common_bits) ^ bits); |
| 1754 | common_bits ^= differing_bits; |
| 1755 | bits &= common_bits; |
| 1756 | } |
| 1757 | // If length is 2 and common bits has only one zero in it then |
| 1758 | // our mask and compare instruction will determine definitely |
| 1759 | // whether we have a match at this character position. Otherwise |
| 1760 | // it can only be an approximate check. |
| 1761 | uint32_t one_zero = (common_bits | ~char_mask); |
| 1762 | if (length == 2 && ((~one_zero) & ((~one_zero) - 1)) == 0) { |
| 1763 | pos->determines_perfectly = true; |
| 1764 | } |
| 1765 | pos->mask = common_bits; |
| 1766 | pos->value = bits; |
| 1767 | } |
| 1768 | } else { |
| 1769 | // Don't ignore case. Nice simple case where the mask-compare will |
| 1770 | // determine definitely whether we have a match at this character |
| 1771 | // position. |
| 1772 | pos->mask = char_mask; |
| 1773 | pos->value = c; |
| 1774 | pos->determines_perfectly = true; |
| 1775 | } |
| 1776 | characters_filled_in++; |
| 1777 | ASSERT(characters_filled_in <= details->characters()); |
| 1778 | if (characters_filled_in == details->characters()) { |
| 1779 | return; |
| 1780 | } |
| 1781 | } |
| 1782 | } else { |
| 1783 | QuickCheckDetails::Position* pos = |
| 1784 | details->positions(characters_filled_in); |
| 1785 | RegExpCharacterClass* tree = elm.char_class(); |
| 1786 | ZoneGrowableArray<CharacterRange>* ranges = tree->ranges(); |
| 1787 | ASSERT(!ranges->is_empty()); |
| 1788 | if (tree->is_negated()) { |
| 1789 | // A quick check uses multi-character mask and compare. There is no |
| 1790 | // useful way to incorporate a negative char class into this scheme |
| 1791 | // so we just conservatively create a mask and value that will always |
| 1792 | // succeed. |
| 1793 | pos->mask = 0; |
| 1794 | pos->value = 0; |
| 1795 | } else { |
| 1796 | intptr_t first_range = 0; |
| 1797 | while (ranges->At(first_range).from() > char_mask) { |
| 1798 | first_range++; |
| 1799 | if (first_range == ranges->length()) { |
| 1800 | details->set_cannot_match(); |
| 1801 | pos->determines_perfectly = false; |
| 1802 | return; |
| 1803 | } |
| 1804 | } |
| 1805 | CharacterRange range = ranges->At(first_range); |
| 1806 | uint16_t from = range.from(); |
| 1807 | uint16_t to = range.to(); |
| 1808 | if (to > char_mask) { |
| 1809 | to = char_mask; |
| 1810 | } |
| 1811 | uint32_t differing_bits = (from ^ to); |
| 1812 | // A mask and compare is only perfect if the differing bits form a |
| 1813 | // number like 00011111 with one single block of trailing 1s. |
| 1814 | if ((differing_bits & (differing_bits + 1)) == 0 && |
| 1815 | from + differing_bits == to) { |
| 1816 | pos->determines_perfectly = true; |
| 1817 | } |
| 1818 | uint32_t common_bits = ~SmearBitsRight(differing_bits); |
| 1819 | uint32_t bits = (from & common_bits); |
| 1820 | for (intptr_t i = first_range + 1; i < ranges->length(); i++) { |
| 1821 | CharacterRange range = ranges->At(i); |
| 1822 | uint16_t from = range.from(); |
| 1823 | uint16_t to = range.to(); |
| 1824 | if (from > char_mask) continue; |
| 1825 | if (to > char_mask) to = char_mask; |
| 1826 | // Here we are combining more ranges into the mask and compare |
| 1827 | // value. With each new range the mask becomes more sparse and |
| 1828 | // so the chances of a false positive rise. A character class |
| 1829 | // with multiple ranges is assumed never to be equivalent to a |
| 1830 | // mask and compare operation. |
| 1831 | pos->determines_perfectly = false; |
| 1832 | uint32_t new_common_bits = (from ^ to); |
| 1833 | new_common_bits = ~SmearBitsRight(new_common_bits); |
| 1834 | common_bits &= new_common_bits; |
| 1835 | bits &= new_common_bits; |
| 1836 | uint32_t differing_bits = (from & common_bits) ^ bits; |
| 1837 | common_bits ^= differing_bits; |
| 1838 | bits &= common_bits; |
| 1839 | } |
| 1840 | pos->mask = common_bits; |
| 1841 | pos->value = bits; |
| 1842 | } |
| 1843 | characters_filled_in++; |
| 1844 | ASSERT(characters_filled_in <= details->characters()); |
| 1845 | if (characters_filled_in == details->characters()) { |
| 1846 | return; |
| 1847 | } |
| 1848 | } |
| 1849 | } |
| 1850 | ASSERT(characters_filled_in != details->characters()); |
| 1851 | if (!details->cannot_match()) { |
| 1852 | on_success()->GetQuickCheckDetails(details, compiler, characters_filled_in, |
| 1853 | true); |
| 1854 | } |
| 1855 | } |
| 1856 | |
| 1857 | void QuickCheckDetails::Clear() { |
| 1858 | for (int i = 0; i < characters_; i++) { |
| 1859 | positions_[i].mask = 0; |
| 1860 | positions_[i].value = 0; |
| 1861 | positions_[i].determines_perfectly = false; |
| 1862 | } |
| 1863 | characters_ = 0; |
| 1864 | } |
| 1865 | |
| 1866 | void QuickCheckDetails::Advance(intptr_t by, bool one_byte) { |
| 1867 | if (by >= characters_ || by < 0) { |
| 1868 | // check that by < 0 => characters_ == 0 |
| 1869 | ASSERT(by >= 0 || characters_ == 0); |
| 1870 | Clear(); |
| 1871 | return; |
| 1872 | } |
| 1873 | for (intptr_t i = 0; i < characters_ - by; i++) { |
| 1874 | positions_[i] = positions_[by + i]; |
| 1875 | } |
| 1876 | for (intptr_t i = characters_ - by; i < characters_; i++) { |
| 1877 | positions_[i].mask = 0; |
| 1878 | positions_[i].value = 0; |
| 1879 | positions_[i].determines_perfectly = false; |
| 1880 | } |
| 1881 | characters_ -= by; |
| 1882 | // We could change mask_ and value_ here but we would never advance unless |
| 1883 | // they had already been used in a check and they won't be used again because |
| 1884 | // it would gain us nothing. So there's no point. |
| 1885 | } |
| 1886 | |
| 1887 | void QuickCheckDetails::Merge(QuickCheckDetails* other, intptr_t from_index) { |
| 1888 | ASSERT(characters_ == other->characters_); |
| 1889 | if (other->cannot_match_) { |
| 1890 | return; |
| 1891 | } |
| 1892 | if (cannot_match_) { |
| 1893 | *this = *other; |
| 1894 | return; |
| 1895 | } |
| 1896 | for (intptr_t i = from_index; i < characters_; i++) { |
| 1897 | QuickCheckDetails::Position* pos = positions(i); |
| 1898 | QuickCheckDetails::Position* other_pos = other->positions(i); |
| 1899 | if (pos->mask != other_pos->mask || pos->value != other_pos->value || |
| 1900 | !other_pos->determines_perfectly) { |
| 1901 | // Our mask-compare operation will be approximate unless we have the |
| 1902 | // exact same operation on both sides of the alternation. |
| 1903 | pos->determines_perfectly = false; |
| 1904 | } |
| 1905 | pos->mask &= other_pos->mask; |
| 1906 | pos->value &= pos->mask; |
| 1907 | other_pos->value &= pos->mask; |
| 1908 | uint16_t differing_bits = (pos->value ^ other_pos->value); |
| 1909 | pos->mask &= ~differing_bits; |
| 1910 | pos->value &= pos->mask; |
| 1911 | } |
| 1912 | } |
| 1913 | |
| 1914 | class VisitMarker : public ValueObject { |
| 1915 | public: |
| 1916 | explicit VisitMarker(NodeInfo* info) : info_(info) { |
| 1917 | ASSERT(!info->visited); |
| 1918 | info->visited = true; |
| 1919 | } |
| 1920 | ~VisitMarker() { info_->visited = false; } |
| 1921 | |
| 1922 | private: |
| 1923 | NodeInfo* info_; |
| 1924 | }; |
| 1925 | |
| 1926 | RegExpNode* SeqRegExpNode::FilterOneByte(intptr_t depth) { |
| 1927 | if (info()->replacement_calculated) return replacement(); |
| 1928 | if (depth < 0) return this; |
| 1929 | ASSERT(!info()->visited); |
| 1930 | VisitMarker marker(info()); |
| 1931 | return FilterSuccessor(depth - 1); |
| 1932 | } |
| 1933 | |
| 1934 | RegExpNode* SeqRegExpNode::FilterSuccessor(intptr_t depth) { |
| 1935 | RegExpNode* next = on_success_->FilterOneByte(depth - 1); |
| 1936 | if (next == NULL) return set_replacement(NULL); |
| 1937 | on_success_ = next; |
| 1938 | return set_replacement(this); |
| 1939 | } |
| 1940 | |
| 1941 | // We need to check for the following characters: 0x39c 0x3bc 0x178. |
| 1942 | static inline bool RangeContainsLatin1Equivalents(CharacterRange range) { |
| 1943 | // TODO(dcarney): this could be a lot more efficient. |
| 1944 | return range.Contains(0x39c) || range.Contains(0x3bc) || |
| 1945 | range.Contains(0x178); |
| 1946 | } |
| 1947 | |
| 1948 | static bool RangesContainLatin1Equivalents( |
| 1949 | ZoneGrowableArray<CharacterRange>* ranges) { |
| 1950 | for (intptr_t i = 0; i < ranges->length(); i++) { |
| 1951 | // TODO(dcarney): this could be a lot more efficient. |
| 1952 | if (RangeContainsLatin1Equivalents(ranges->At(i))) return true; |
| 1953 | } |
| 1954 | return false; |
| 1955 | } |
| 1956 | |
| 1957 | static uint16_t ConvertNonLatin1ToLatin1(uint16_t c) { |
| 1958 | ASSERT(c > Symbols::kMaxOneCharCodeSymbol); |
| 1959 | switch (c) { |
| 1960 | // This are equivalent characters in unicode. |
| 1961 | case 0x39c: |
| 1962 | case 0x3bc: |
| 1963 | return 0xb5; |
| 1964 | // This is an uppercase of a Latin-1 character |
| 1965 | // outside of Latin-1. |
| 1966 | case 0x178: |
| 1967 | return 0xff; |
| 1968 | } |
| 1969 | return 0; |
| 1970 | } |
| 1971 | |
| 1972 | RegExpNode* TextNode::FilterOneByte(intptr_t depth) { |
| 1973 | if (info()->replacement_calculated) return replacement(); |
| 1974 | if (depth < 0) return this; |
| 1975 | ASSERT(!info()->visited); |
| 1976 | VisitMarker marker(info()); |
| 1977 | intptr_t element_count = elms_->length(); |
| 1978 | for (intptr_t i = 0; i < element_count; i++) { |
| 1979 | TextElement elm = elms_->At(i); |
| 1980 | if (elm.text_type() == TextElement::ATOM) { |
| 1981 | ZoneGrowableArray<uint16_t>* quarks = elm.atom()->data(); |
| 1982 | for (intptr_t j = 0; j < quarks->length(); j++) { |
| 1983 | uint16_t c = quarks->At(j); |
| 1984 | if (c <= Symbols::kMaxOneCharCodeSymbol) continue; |
| 1985 | if (!elm.atom()->ignore_case()) return set_replacement(NULL); |
| 1986 | // Here, we need to check for characters whose upper and lower cases |
| 1987 | // are outside the Latin-1 range. |
| 1988 | uint16_t converted = ConvertNonLatin1ToLatin1(c); |
| 1989 | // Character is outside Latin-1 completely |
| 1990 | if (converted == 0) return set_replacement(NULL); |
| 1991 | // Convert quark to Latin-1 in place. |
| 1992 | (*quarks)[0] = converted; |
| 1993 | } |
| 1994 | } else { |
| 1995 | ASSERT(elm.text_type() == TextElement::CHAR_CLASS); |
| 1996 | RegExpCharacterClass* cc = elm.char_class(); |
| 1997 | ZoneGrowableArray<CharacterRange>* ranges = cc->ranges(); |
| 1998 | if (!CharacterRange::IsCanonical(ranges)) { |
| 1999 | CharacterRange::Canonicalize(ranges); |
| 2000 | } |
| 2001 | // Now they are in order so we only need to look at the first. |
| 2002 | intptr_t range_count = ranges->length(); |
| 2003 | if (cc->is_negated()) { |
| 2004 | if (range_count != 0 && ranges->At(0).from() == 0 && |
| 2005 | ranges->At(0).to() >= Symbols::kMaxOneCharCodeSymbol) { |
| 2006 | // This will be handled in a later filter. |
| 2007 | if (cc->flags().IgnoreCase() && |
| 2008 | RangesContainLatin1Equivalents(ranges)) { |
| 2009 | continue; |
| 2010 | } |
| 2011 | return set_replacement(NULL); |
| 2012 | } |
| 2013 | } else { |
| 2014 | if (range_count == 0 || |
| 2015 | ranges->At(0).from() > Symbols::kMaxOneCharCodeSymbol) { |
| 2016 | // This will be handled in a later filter. |
| 2017 | if (cc->flags().IgnoreCase() && |
| 2018 | RangesContainLatin1Equivalents(ranges)) |
| 2019 | continue; |
| 2020 | return set_replacement(NULL); |
| 2021 | } |
| 2022 | } |
| 2023 | } |
| 2024 | } |
| 2025 | return FilterSuccessor(depth - 1); |
| 2026 | } |
| 2027 | |
| 2028 | RegExpNode* LoopChoiceNode::FilterOneByte(intptr_t depth) { |
| 2029 | if (info()->replacement_calculated) return replacement(); |
| 2030 | if (depth < 0) return this; |
| 2031 | if (info()->visited) return this; |
| 2032 | { |
| 2033 | VisitMarker marker(info()); |
| 2034 | |
| 2035 | RegExpNode* continue_replacement = continue_node_->FilterOneByte(depth - 1); |
| 2036 | // If we can't continue after the loop then there is no sense in doing the |
| 2037 | // loop. |
| 2038 | if (continue_replacement == NULL) return set_replacement(NULL); |
| 2039 | } |
| 2040 | |
| 2041 | return ChoiceNode::FilterOneByte(depth - 1); |
| 2042 | } |
| 2043 | |
| 2044 | RegExpNode* ChoiceNode::FilterOneByte(intptr_t depth) { |
| 2045 | if (info()->replacement_calculated) return replacement(); |
| 2046 | if (depth < 0) return this; |
| 2047 | if (info()->visited) return this; |
| 2048 | VisitMarker marker(info()); |
| 2049 | intptr_t choice_count = alternatives_->length(); |
| 2050 | |
| 2051 | for (intptr_t i = 0; i < choice_count; i++) { |
| 2052 | GuardedAlternative alternative = alternatives_->At(i); |
| 2053 | if (alternative.guards() != NULL && alternative.guards()->length() != 0) { |
| 2054 | set_replacement(this); |
| 2055 | return this; |
| 2056 | } |
| 2057 | } |
| 2058 | |
| 2059 | intptr_t surviving = 0; |
| 2060 | RegExpNode* survivor = NULL; |
| 2061 | for (intptr_t i = 0; i < choice_count; i++) { |
| 2062 | GuardedAlternative alternative = alternatives_->At(i); |
| 2063 | RegExpNode* replacement = alternative.node()->FilterOneByte(depth - 1); |
| 2064 | ASSERT(replacement != this); // No missing EMPTY_MATCH_CHECK. |
| 2065 | if (replacement != NULL) { |
| 2066 | (*alternatives_)[i].set_node(replacement); |
| 2067 | surviving++; |
| 2068 | survivor = replacement; |
| 2069 | } |
| 2070 | } |
| 2071 | if (surviving < 2) return set_replacement(survivor); |
| 2072 | |
| 2073 | set_replacement(this); |
| 2074 | if (surviving == choice_count) { |
| 2075 | return this; |
| 2076 | } |
| 2077 | // Only some of the nodes survived the filtering. We need to rebuild the |
| 2078 | // alternatives list. |
| 2079 | ZoneGrowableArray<GuardedAlternative>* new_alternatives = |
| 2080 | new (Z) ZoneGrowableArray<GuardedAlternative>(surviving); |
| 2081 | for (intptr_t i = 0; i < choice_count; i++) { |
| 2082 | RegExpNode* replacement = |
| 2083 | (*alternatives_)[i].node()->FilterOneByte(depth - 1); |
| 2084 | if (replacement != NULL) { |
| 2085 | (*alternatives_)[i].set_node(replacement); |
| 2086 | new_alternatives->Add((*alternatives_)[i]); |
| 2087 | } |
| 2088 | } |
| 2089 | alternatives_ = new_alternatives; |
| 2090 | return this; |
| 2091 | } |
| 2092 | |
| 2093 | RegExpNode* NegativeLookaroundChoiceNode::FilterOneByte(intptr_t depth) { |
| 2094 | if (info()->replacement_calculated) return replacement(); |
| 2095 | if (depth < 0) return this; |
| 2096 | if (info()->visited) return this; |
| 2097 | VisitMarker marker(info()); |
| 2098 | // Alternative 0 is the negative lookahead, alternative 1 is what comes |
| 2099 | // afterwards. |
| 2100 | RegExpNode* node = (*alternatives_)[1].node(); |
| 2101 | RegExpNode* replacement = node->FilterOneByte(depth - 1); |
| 2102 | if (replacement == NULL) return set_replacement(NULL); |
| 2103 | (*alternatives_)[1].set_node(replacement); |
| 2104 | |
| 2105 | RegExpNode* neg_node = (*alternatives_)[0].node(); |
| 2106 | RegExpNode* neg_replacement = neg_node->FilterOneByte(depth - 1); |
| 2107 | // If the negative lookahead is always going to fail then |
| 2108 | // we don't need to check it. |
| 2109 | if (neg_replacement == NULL) return set_replacement(replacement); |
| 2110 | (*alternatives_)[0].set_node(neg_replacement); |
| 2111 | return set_replacement(this); |
| 2112 | } |
| 2113 | |
| 2114 | void LoopChoiceNode::GetQuickCheckDetails(QuickCheckDetails* details, |
| 2115 | RegExpCompiler* compiler, |
| 2116 | intptr_t characters_filled_in, |
| 2117 | bool not_at_start) { |
| 2118 | if (body_can_be_zero_length_ || info()->visited) return; |
| 2119 | VisitMarker marker(info()); |
| 2120 | return ChoiceNode::GetQuickCheckDetails(details, compiler, |
| 2121 | characters_filled_in, not_at_start); |
| 2122 | } |
| 2123 | |
| 2124 | void LoopChoiceNode::FillInBMInfo(intptr_t offset, |
| 2125 | intptr_t budget, |
| 2126 | BoyerMooreLookahead* bm, |
| 2127 | bool not_at_start) { |
| 2128 | if (body_can_be_zero_length_ || budget <= 0) { |
| 2129 | bm->SetRest(offset); |
| 2130 | SaveBMInfo(bm, not_at_start, offset); |
| 2131 | return; |
| 2132 | } |
| 2133 | ChoiceNode::FillInBMInfo(offset, budget - 1, bm, not_at_start); |
| 2134 | SaveBMInfo(bm, not_at_start, offset); |
| 2135 | } |
| 2136 | |
| 2137 | void ChoiceNode::GetQuickCheckDetails(QuickCheckDetails* details, |
| 2138 | RegExpCompiler* compiler, |
| 2139 | intptr_t characters_filled_in, |
| 2140 | bool not_at_start) { |
| 2141 | not_at_start = (not_at_start || not_at_start_); |
| 2142 | intptr_t choice_count = alternatives_->length(); |
| 2143 | ASSERT(choice_count > 0); |
| 2144 | (*alternatives_)[0].node()->GetQuickCheckDetails( |
| 2145 | details, compiler, characters_filled_in, not_at_start); |
| 2146 | for (intptr_t i = 1; i < choice_count; i++) { |
| 2147 | QuickCheckDetails new_details(details->characters()); |
| 2148 | RegExpNode* node = (*alternatives_)[i].node(); |
| 2149 | node->GetQuickCheckDetails(&new_details, compiler, characters_filled_in, |
| 2150 | not_at_start); |
| 2151 | // Here we merge the quick match details of the two branches. |
| 2152 | details->Merge(&new_details, characters_filled_in); |
| 2153 | } |
| 2154 | } |
| 2155 | |
| 2156 | // Check for [0-9A-Z_a-z]. |
| 2157 | static void EmitWordCheck(RegExpMacroAssembler* assembler, |
| 2158 | BlockLabel* word, |
| 2159 | BlockLabel* non_word, |
| 2160 | bool fall_through_on_word) { |
| 2161 | if (assembler->CheckSpecialCharacterClass( |
| 2162 | fall_through_on_word ? 'w' : 'W', |
| 2163 | fall_through_on_word ? non_word : word)) { |
| 2164 | // Optimized implementation available. |
| 2165 | return; |
| 2166 | } |
| 2167 | assembler->CheckCharacterGT('z', non_word); |
| 2168 | assembler->CheckCharacterLT('0', non_word); |
| 2169 | assembler->CheckCharacterGT('a' - 1, word); |
| 2170 | assembler->CheckCharacterLT('9' + 1, word); |
| 2171 | assembler->CheckCharacterLT('A', non_word); |
| 2172 | assembler->CheckCharacterLT('Z' + 1, word); |
| 2173 | if (fall_through_on_word) { |
| 2174 | assembler->CheckNotCharacter('_', non_word); |
| 2175 | } else { |
| 2176 | assembler->CheckCharacter('_', word); |
| 2177 | } |
| 2178 | } |
| 2179 | |
| 2180 | // Emit the code to check for a ^ in multiline mode (1-character lookbehind |
| 2181 | // that matches newline or the start of input). |
| 2182 | static void EmitHat(RegExpCompiler* compiler, |
| 2183 | RegExpNode* on_success, |
| 2184 | Trace* trace) { |
| 2185 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 2186 | // We will be loading the previous character into the current character |
| 2187 | // register. |
| 2188 | Trace new_trace(*trace); |
| 2189 | new_trace.InvalidateCurrentCharacter(); |
| 2190 | |
| 2191 | BlockLabel ok; |
| 2192 | if (new_trace.cp_offset() == 0) { |
| 2193 | // The start of input counts as a newline in this context, so skip to |
| 2194 | // ok if we are at the start. |
| 2195 | assembler->CheckAtStart(&ok); |
| 2196 | } |
| 2197 | // We already checked that we are not at the start of input so it must be |
| 2198 | // OK to load the previous character. |
| 2199 | assembler->LoadCurrentCharacter(new_trace.cp_offset() - 1, |
| 2200 | new_trace.backtrack(), false); |
| 2201 | if (!assembler->CheckSpecialCharacterClass('n', new_trace.backtrack())) { |
| 2202 | // Newline means \n, \r, 0x2028 or 0x2029. |
| 2203 | if (!compiler->one_byte()) { |
| 2204 | assembler->CheckCharacterAfterAnd(0x2028, 0xfffe, &ok); |
| 2205 | } |
| 2206 | assembler->CheckCharacter('\n', &ok); |
| 2207 | assembler->CheckNotCharacter('\r', new_trace.backtrack()); |
| 2208 | } |
| 2209 | assembler->BindBlock(&ok); |
| 2210 | on_success->Emit(compiler, &new_trace); |
| 2211 | } |
| 2212 | |
| 2213 | // Emit the code to handle \b and \B (word-boundary or non-word-boundary). |
| 2214 | void AssertionNode::EmitBoundaryCheck(RegExpCompiler* compiler, Trace* trace) { |
| 2215 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 2216 | Trace::TriBool next_is_word_character = Trace::UNKNOWN; |
| 2217 | bool not_at_start = (trace->at_start() == Trace::FALSE_VALUE); |
| 2218 | BoyerMooreLookahead* lookahead = bm_info(not_at_start); |
| 2219 | if (lookahead == NULL) { |
| 2220 | intptr_t eats_at_least = |
| 2221 | Utils::Minimum(kMaxLookaheadForBoyerMoore, |
| 2222 | EatsAtLeast(kMaxLookaheadForBoyerMoore, kRecursionBudget, |
| 2223 | not_at_start)); |
| 2224 | if (eats_at_least >= 1) { |
| 2225 | BoyerMooreLookahead* bm = |
| 2226 | new (Z) BoyerMooreLookahead(eats_at_least, compiler, Z); |
| 2227 | FillInBMInfo(0, kRecursionBudget, bm, not_at_start); |
| 2228 | if (bm->at(0)->is_non_word()) next_is_word_character = Trace::FALSE_VALUE; |
| 2229 | if (bm->at(0)->is_word()) next_is_word_character = Trace::TRUE_VALUE; |
| 2230 | } |
| 2231 | } else { |
| 2232 | if (lookahead->at(0)->is_non_word()) |
| 2233 | next_is_word_character = Trace::FALSE_VALUE; |
| 2234 | if (lookahead->at(0)->is_word()) next_is_word_character = Trace::TRUE_VALUE; |
| 2235 | } |
| 2236 | bool at_boundary = (assertion_type_ == AssertionNode::AT_BOUNDARY); |
| 2237 | if (next_is_word_character == Trace::UNKNOWN) { |
| 2238 | BlockLabel before_non_word; |
| 2239 | BlockLabel before_word; |
| 2240 | if (trace->characters_preloaded() != 1) { |
| 2241 | assembler->LoadCurrentCharacter(trace->cp_offset(), &before_non_word); |
| 2242 | } |
| 2243 | // Fall through on non-word. |
| 2244 | EmitWordCheck(assembler, &before_word, &before_non_word, false); |
| 2245 | // Next character is not a word character. |
| 2246 | assembler->BindBlock(&before_non_word); |
| 2247 | BlockLabel ok; |
| 2248 | // Backtrack on \B (non-boundary check) if previous is a word, |
| 2249 | // since we know next *is not* a word and this would be a boundary. |
| 2250 | BacktrackIfPrevious(compiler, trace, at_boundary ? kIsNonWord : kIsWord); |
| 2251 | |
| 2252 | if (!assembler->IsClosed()) { |
| 2253 | assembler->GoTo(&ok); |
| 2254 | } |
| 2255 | |
| 2256 | assembler->BindBlock(&before_word); |
| 2257 | BacktrackIfPrevious(compiler, trace, at_boundary ? kIsWord : kIsNonWord); |
| 2258 | assembler->BindBlock(&ok); |
| 2259 | } else if (next_is_word_character == Trace::TRUE_VALUE) { |
| 2260 | BacktrackIfPrevious(compiler, trace, at_boundary ? kIsWord : kIsNonWord); |
| 2261 | } else { |
| 2262 | ASSERT(next_is_word_character == Trace::FALSE_VALUE); |
| 2263 | BacktrackIfPrevious(compiler, trace, at_boundary ? kIsNonWord : kIsWord); |
| 2264 | } |
| 2265 | } |
| 2266 | |
| 2267 | void AssertionNode::BacktrackIfPrevious( |
| 2268 | RegExpCompiler* compiler, |
| 2269 | Trace* trace, |
| 2270 | AssertionNode::IfPrevious backtrack_if_previous) { |
| 2271 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 2272 | Trace new_trace(*trace); |
| 2273 | new_trace.InvalidateCurrentCharacter(); |
| 2274 | |
| 2275 | BlockLabel fall_through, dummy; |
| 2276 | |
| 2277 | BlockLabel* non_word = backtrack_if_previous == kIsNonWord |
| 2278 | ? new_trace.backtrack() |
| 2279 | : &fall_through; |
| 2280 | BlockLabel* word = backtrack_if_previous == kIsNonWord |
| 2281 | ? &fall_through |
| 2282 | : new_trace.backtrack(); |
| 2283 | |
| 2284 | if (new_trace.cp_offset() == 0) { |
| 2285 | // The start of input counts as a non-word character, so the question is |
| 2286 | // decided if we are at the start. |
| 2287 | assembler->CheckAtStart(non_word); |
| 2288 | } |
| 2289 | // We already checked that we are not at the start of input so it must be |
| 2290 | // OK to load the previous character. |
| 2291 | assembler->LoadCurrentCharacter(new_trace.cp_offset() - 1, &dummy, false); |
| 2292 | EmitWordCheck(assembler, word, non_word, backtrack_if_previous == kIsNonWord); |
| 2293 | |
| 2294 | assembler->BindBlock(&fall_through); |
| 2295 | on_success()->Emit(compiler, &new_trace); |
| 2296 | } |
| 2297 | |
| 2298 | void AssertionNode::GetQuickCheckDetails(QuickCheckDetails* details, |
| 2299 | RegExpCompiler* compiler, |
| 2300 | intptr_t filled_in, |
| 2301 | bool not_at_start) { |
| 2302 | if (assertion_type_ == AT_START && not_at_start) { |
| 2303 | details->set_cannot_match(); |
| 2304 | return; |
| 2305 | } |
| 2306 | return on_success()->GetQuickCheckDetails(details, compiler, filled_in, |
| 2307 | not_at_start); |
| 2308 | } |
| 2309 | |
| 2310 | void AssertionNode::Emit(RegExpCompiler* compiler, Trace* trace) { |
| 2311 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 2312 | switch (assertion_type_) { |
| 2313 | case AT_END: { |
| 2314 | BlockLabel ok; |
| 2315 | assembler->CheckPosition(trace->cp_offset(), &ok); |
| 2316 | assembler->GoTo(trace->backtrack()); |
| 2317 | assembler->BindBlock(&ok); |
| 2318 | break; |
| 2319 | } |
| 2320 | case AT_START: { |
| 2321 | if (trace->at_start() == Trace::FALSE_VALUE) { |
| 2322 | assembler->GoTo(trace->backtrack()); |
| 2323 | return; |
| 2324 | } |
| 2325 | if (trace->at_start() == Trace::UNKNOWN) { |
| 2326 | assembler->CheckNotAtStart(trace->cp_offset(), trace->backtrack()); |
| 2327 | Trace at_start_trace = *trace; |
| 2328 | at_start_trace.set_at_start(Trace::TRUE_VALUE); |
| 2329 | on_success()->Emit(compiler, &at_start_trace); |
| 2330 | return; |
| 2331 | } |
| 2332 | } break; |
| 2333 | case AFTER_NEWLINE: |
| 2334 | EmitHat(compiler, on_success(), trace); |
| 2335 | return; |
| 2336 | case AT_BOUNDARY: |
| 2337 | case AT_NON_BOUNDARY: { |
| 2338 | EmitBoundaryCheck(compiler, trace); |
| 2339 | return; |
| 2340 | } |
| 2341 | } |
| 2342 | on_success()->Emit(compiler, trace); |
| 2343 | } |
| 2344 | |
| 2345 | static bool DeterminedAlready(QuickCheckDetails* quick_check, intptr_t offset) { |
| 2346 | if (quick_check == NULL) return false; |
| 2347 | if (offset >= quick_check->characters()) return false; |
| 2348 | return quick_check->positions(offset)->determines_perfectly; |
| 2349 | } |
| 2350 | |
| 2351 | static void UpdateBoundsCheck(intptr_t index, intptr_t* checked_up_to) { |
| 2352 | if (index > *checked_up_to) { |
| 2353 | *checked_up_to = index; |
| 2354 | } |
| 2355 | } |
| 2356 | |
| 2357 | // We call this repeatedly to generate code for each pass over the text node. |
| 2358 | // The passes are in increasing order of difficulty because we hope one |
| 2359 | // of the first passes will fail in which case we are saved the work of the |
| 2360 | // later passes. for example for the case independent regexp /%[asdfghjkl]a/ |
| 2361 | // we will check the '%' in the first pass, the case independent 'a' in the |
| 2362 | // second pass and the character class in the last pass. |
| 2363 | // |
| 2364 | // The passes are done from right to left, so for example to test for /bar/ |
| 2365 | // we will first test for an 'r' with offset 2, then an 'a' with offset 1 |
| 2366 | // and then a 'b' with offset 0. This means we can avoid the end-of-input |
| 2367 | // bounds check most of the time. In the example we only need to check for |
| 2368 | // end-of-input when loading the putative 'r'. |
| 2369 | // |
| 2370 | // A slight complication involves the fact that the first character may already |
| 2371 | // be fetched into a register by the previous node. In this case we want to |
| 2372 | // do the test for that character first. We do this in separate passes. The |
| 2373 | // 'preloaded' argument indicates that we are doing such a 'pass'. If such a |
| 2374 | // pass has been performed then subsequent passes will have true in |
| 2375 | // first_element_checked to indicate that that character does not need to be |
| 2376 | // checked again. |
| 2377 | // |
| 2378 | // In addition to all this we are passed a Trace, which can |
| 2379 | // contain an AlternativeGeneration object. In this AlternativeGeneration |
| 2380 | // object we can see details of any quick check that was already passed in |
| 2381 | // order to get to the code we are now generating. The quick check can involve |
| 2382 | // loading characters, which means we do not need to recheck the bounds |
| 2383 | // up to the limit the quick check already checked. In addition the quick |
| 2384 | // check can have involved a mask and compare operation which may simplify |
| 2385 | // or obviate the need for further checks at some character positions. |
| 2386 | void TextNode::TextEmitPass(RegExpCompiler* compiler, |
| 2387 | TextEmitPassType pass, |
| 2388 | bool preloaded, |
| 2389 | Trace* trace, |
| 2390 | bool first_element_checked, |
| 2391 | intptr_t* checked_up_to) { |
| 2392 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 2393 | bool one_byte = compiler->one_byte(); |
| 2394 | BlockLabel* backtrack = trace->backtrack(); |
| 2395 | QuickCheckDetails* quick_check = trace->quick_check_performed(); |
| 2396 | intptr_t element_count = elms_->length(); |
| 2397 | intptr_t backward_offset = read_backward() ? -Length() : 0; |
| 2398 | for (intptr_t i = preloaded ? 0 : element_count - 1; i >= 0; i--) { |
| 2399 | TextElement elm = elms_->At(i); |
| 2400 | intptr_t cp_offset = trace->cp_offset() + elm.cp_offset() + backward_offset; |
| 2401 | if (elm.text_type() == TextElement::ATOM) { |
| 2402 | ZoneGrowableArray<uint16_t>* quarks = elm.atom()->data(); |
| 2403 | for (intptr_t j = preloaded ? 0 : quarks->length() - 1; j >= 0; j--) { |
| 2404 | if (SkipPass(pass, elm.atom()->ignore_case())) continue; |
| 2405 | if (first_element_checked && i == 0 && j == 0) continue; |
| 2406 | if (DeterminedAlready(quick_check, elm.cp_offset() + j)) continue; |
| 2407 | EmitCharacterFunction* emit_function = NULL; |
| 2408 | uint16_t quark = quarks->At(j); |
| 2409 | if (elm.atom()->ignore_case()) { |
| 2410 | // Everywhere else we assume that a non-Latin-1 character cannot match |
| 2411 | // a Latin-1 character. Avoid the cases where this is assumption is |
| 2412 | // invalid by using the Latin1 equivalent instead. |
| 2413 | quark = Latin1::TryConvertToLatin1(quark); |
| 2414 | } |
| 2415 | switch (pass) { |
| 2416 | case NON_LATIN1_MATCH: |
| 2417 | ASSERT(one_byte); |
| 2418 | if (quark > Symbols::kMaxOneCharCodeSymbol) { |
| 2419 | assembler->GoTo(backtrack); |
| 2420 | return; |
| 2421 | } |
| 2422 | break; |
| 2423 | case NON_LETTER_CHARACTER_MATCH: |
| 2424 | emit_function = &EmitAtomNonLetter; |
| 2425 | break; |
| 2426 | case SIMPLE_CHARACTER_MATCH: |
| 2427 | emit_function = &EmitSimpleCharacter; |
| 2428 | break; |
| 2429 | case CASE_CHARACTER_MATCH: |
| 2430 | emit_function = &EmitAtomLetter; |
| 2431 | break; |
| 2432 | default: |
| 2433 | break; |
| 2434 | } |
| 2435 | if (emit_function != NULL) { |
| 2436 | const bool bounds_check = |
| 2437 | *checked_up_to < (cp_offset + j) || read_backward(); |
| 2438 | bool bound_checked = |
| 2439 | emit_function(Z, compiler, quarks->At(j), backtrack, |
| 2440 | cp_offset + j, bounds_check, preloaded); |
| 2441 | if (bound_checked) UpdateBoundsCheck(cp_offset + j, checked_up_to); |
| 2442 | } |
| 2443 | } |
| 2444 | } else { |
| 2445 | ASSERT(elm.text_type() == TextElement::CHAR_CLASS); |
| 2446 | if (pass == CHARACTER_CLASS_MATCH) { |
| 2447 | if (first_element_checked && i == 0) continue; |
| 2448 | if (DeterminedAlready(quick_check, elm.cp_offset())) continue; |
| 2449 | RegExpCharacterClass* cc = elm.char_class(); |
| 2450 | bool bounds_check = *checked_up_to < cp_offset || read_backward(); |
| 2451 | EmitCharClass(assembler, cc, one_byte, backtrack, cp_offset, |
| 2452 | bounds_check, preloaded, Z); |
| 2453 | UpdateBoundsCheck(cp_offset, checked_up_to); |
| 2454 | } |
| 2455 | } |
| 2456 | } |
| 2457 | } |
| 2458 | |
| 2459 | intptr_t TextNode::Length() { |
| 2460 | TextElement elm = elms_->Last(); |
| 2461 | ASSERT(elm.cp_offset() >= 0); |
| 2462 | return elm.cp_offset() + elm.length(); |
| 2463 | } |
| 2464 | |
| 2465 | bool TextNode::SkipPass(intptr_t intptr_t_pass, bool ignore_case) { |
| 2466 | TextEmitPassType pass = static_cast<TextEmitPassType>(intptr_t_pass); |
| 2467 | if (ignore_case) { |
| 2468 | return pass == SIMPLE_CHARACTER_MATCH; |
| 2469 | } else { |
| 2470 | return pass == NON_LETTER_CHARACTER_MATCH || pass == CASE_CHARACTER_MATCH; |
| 2471 | } |
| 2472 | } |
| 2473 | |
| 2474 | TextNode* TextNode::CreateForCharacterRanges( |
| 2475 | ZoneGrowableArray<CharacterRange>* ranges, |
| 2476 | bool read_backward, |
| 2477 | RegExpNode* on_success, |
| 2478 | RegExpFlags flags) { |
| 2479 | ASSERT(ranges != nullptr); |
| 2480 | ZoneGrowableArray<TextElement>* elms = new ZoneGrowableArray<TextElement>(1); |
| 2481 | elms->Add(TextElement::CharClass(new RegExpCharacterClass(ranges, flags))); |
| 2482 | return new TextNode(elms, read_backward, on_success); |
| 2483 | } |
| 2484 | |
| 2485 | TextNode* TextNode::CreateForSurrogatePair(CharacterRange lead, |
| 2486 | CharacterRange trail, |
| 2487 | bool read_backward, |
| 2488 | RegExpNode* on_success, |
| 2489 | RegExpFlags flags) { |
| 2490 | auto lead_ranges = CharacterRange::List(on_success->zone(), lead); |
| 2491 | auto trail_ranges = CharacterRange::List(on_success->zone(), trail); |
| 2492 | auto elms = new ZoneGrowableArray<TextElement>(2); |
| 2493 | |
| 2494 | elms->Add( |
| 2495 | TextElement::CharClass(new RegExpCharacterClass(lead_ranges, flags))); |
| 2496 | elms->Add( |
| 2497 | TextElement::CharClass(new RegExpCharacterClass(trail_ranges, flags))); |
| 2498 | |
| 2499 | return new TextNode(elms, read_backward, on_success); |
| 2500 | } |
| 2501 | |
| 2502 | // This generates the code to match a text node. A text node can contain |
| 2503 | // straight character sequences (possibly to be matched in a case-independent |
| 2504 | // way) and character classes. For efficiency we do not do this in a single |
| 2505 | // pass from left to right. Instead we pass over the text node several times, |
| 2506 | // emitting code for some character positions every time. See the comment on |
| 2507 | // TextEmitPass for details. |
| 2508 | void TextNode::Emit(RegExpCompiler* compiler, Trace* trace) { |
| 2509 | LimitResult limit_result = LimitVersions(compiler, trace); |
| 2510 | if (limit_result == DONE) return; |
| 2511 | ASSERT(limit_result == CONTINUE); |
| 2512 | |
| 2513 | if (trace->cp_offset() + Length() > RegExpMacroAssembler::kMaxCPOffset) { |
| 2514 | compiler->SetRegExpTooBig(); |
| 2515 | return; |
| 2516 | } |
| 2517 | |
| 2518 | if (compiler->one_byte()) { |
| 2519 | intptr_t dummy = 0; |
| 2520 | TextEmitPass(compiler, NON_LATIN1_MATCH, false, trace, false, &dummy); |
| 2521 | } |
| 2522 | |
| 2523 | bool first_elt_done = false; |
| 2524 | intptr_t bound_checked_to = trace->cp_offset() - 1; |
| 2525 | bound_checked_to += trace->bound_checked_up_to(); |
| 2526 | |
| 2527 | // If a character is preloaded into the current character register then |
| 2528 | // check that now. |
| 2529 | if (trace->characters_preloaded() == 1) { |
| 2530 | for (intptr_t pass = kFirstRealPass; pass <= kLastPass; pass++) { |
| 2531 | TextEmitPass(compiler, static_cast<TextEmitPassType>(pass), true, trace, |
| 2532 | false, &bound_checked_to); |
| 2533 | } |
| 2534 | first_elt_done = true; |
| 2535 | } |
| 2536 | |
| 2537 | for (intptr_t pass = kFirstRealPass; pass <= kLastPass; pass++) { |
| 2538 | TextEmitPass(compiler, static_cast<TextEmitPassType>(pass), false, trace, |
| 2539 | first_elt_done, &bound_checked_to); |
| 2540 | } |
| 2541 | |
| 2542 | Trace successor_trace(*trace); |
| 2543 | // If we advance backward, we may end up at the start. |
| 2544 | successor_trace.AdvanceCurrentPositionInTrace( |
| 2545 | read_backward() ? -Length() : Length(), compiler); |
| 2546 | successor_trace.set_at_start(read_backward() ? Trace::UNKNOWN |
| 2547 | : Trace::FALSE_VALUE); |
| 2548 | RecursionCheck rc(compiler); |
| 2549 | on_success()->Emit(compiler, &successor_trace); |
| 2550 | } |
| 2551 | |
| 2552 | void Trace::InvalidateCurrentCharacter() { |
| 2553 | characters_preloaded_ = 0; |
| 2554 | } |
| 2555 | |
| 2556 | void Trace::AdvanceCurrentPositionInTrace(intptr_t by, |
| 2557 | RegExpCompiler* compiler) { |
| 2558 | // We don't have an instruction for shifting the current character register |
| 2559 | // down or for using a shifted value for anything so lets just forget that |
| 2560 | // we preloaded any characters into it. |
| 2561 | characters_preloaded_ = 0; |
| 2562 | // Adjust the offsets of the quick check performed information. This |
| 2563 | // information is used to find out what we already determined about the |
| 2564 | // characters by means of mask and compare. |
| 2565 | quick_check_performed_.Advance(by, compiler->one_byte()); |
| 2566 | cp_offset_ += by; |
| 2567 | if (cp_offset_ > RegExpMacroAssembler::kMaxCPOffset) { |
| 2568 | compiler->SetRegExpTooBig(); |
| 2569 | cp_offset_ = 0; |
| 2570 | } |
| 2571 | bound_checked_up_to_ = |
| 2572 | Utils::Maximum(static_cast<intptr_t>(0), bound_checked_up_to_ - by); |
| 2573 | } |
| 2574 | |
| 2575 | void TextNode::MakeCaseIndependent(bool is_one_byte) { |
| 2576 | intptr_t element_count = elms_->length(); |
| 2577 | for (intptr_t i = 0; i < element_count; i++) { |
| 2578 | TextElement elm = elms_->At(i); |
| 2579 | if (elm.text_type() == TextElement::CHAR_CLASS) { |
| 2580 | RegExpCharacterClass* cc = elm.char_class(); |
| 2581 | bool case_equivalents_already_added = |
| 2582 | cc->flags().NeedsUnicodeCaseEquivalents(); |
| 2583 | if (cc->flags().IgnoreCase() && !case_equivalents_already_added) { |
| 2584 | // None of the standard character classes is different in the case |
| 2585 | // independent case and it slows us down if we don't know that. |
| 2586 | if (cc->is_standard()) continue; |
| 2587 | CharacterRange::AddCaseEquivalents(cc->ranges(), is_one_byte, Z); |
| 2588 | } |
| 2589 | } |
| 2590 | } |
| 2591 | } |
| 2592 | |
| 2593 | intptr_t TextNode::GreedyLoopTextLength() { |
| 2594 | TextElement elm = elms_->At(elms_->length() - 1); |
| 2595 | return elm.cp_offset() + elm.length(); |
| 2596 | } |
| 2597 | |
| 2598 | RegExpNode* TextNode::GetSuccessorOfOmnivorousTextNode( |
| 2599 | RegExpCompiler* compiler) { |
| 2600 | if (read_backward()) return nullptr; |
| 2601 | if (elms_->length() != 1) return NULL; |
| 2602 | TextElement elm = elms_->At(0); |
| 2603 | if (elm.text_type() != TextElement::CHAR_CLASS) return NULL; |
| 2604 | RegExpCharacterClass* node = elm.char_class(); |
| 2605 | ZoneGrowableArray<CharacterRange>* ranges = node->ranges(); |
| 2606 | if (!CharacterRange::IsCanonical(ranges)) { |
| 2607 | CharacterRange::Canonicalize(ranges); |
| 2608 | } |
| 2609 | if (node->is_negated()) { |
| 2610 | return ranges->length() == 0 ? on_success() : NULL; |
| 2611 | } |
| 2612 | if (ranges->length() != 1) return NULL; |
| 2613 | uint32_t max_char; |
| 2614 | if (compiler->one_byte()) { |
| 2615 | max_char = Symbols::kMaxOneCharCodeSymbol; |
| 2616 | } else { |
| 2617 | max_char = Utf16::kMaxCodeUnit; |
| 2618 | } |
| 2619 | return ranges->At(0).IsEverything(max_char) ? on_success() : NULL; |
| 2620 | } |
| 2621 | |
| 2622 | // Finds the fixed match length of a sequence of nodes that goes from |
| 2623 | // this alternative and back to this choice node. If there are variable |
| 2624 | // length nodes or other complications in the way then return a sentinel |
| 2625 | // value indicating that a greedy loop cannot be constructed. |
| 2626 | intptr_t ChoiceNode::GreedyLoopTextLengthForAlternative( |
| 2627 | const GuardedAlternative* alternative) { |
| 2628 | intptr_t length = 0; |
| 2629 | RegExpNode* node = alternative->node(); |
| 2630 | // Later we will generate code for all these text nodes using recursion |
| 2631 | // so we have to limit the max number. |
| 2632 | intptr_t recursion_depth = 0; |
| 2633 | while (node != this) { |
| 2634 | if (recursion_depth++ > RegExpCompiler::kMaxRecursion) { |
| 2635 | return kNodeIsTooComplexForGreedyLoops; |
| 2636 | } |
| 2637 | intptr_t node_length = node->GreedyLoopTextLength(); |
| 2638 | if (node_length == kNodeIsTooComplexForGreedyLoops) { |
| 2639 | return kNodeIsTooComplexForGreedyLoops; |
| 2640 | } |
| 2641 | length += node_length; |
| 2642 | SeqRegExpNode* seq_node = static_cast<SeqRegExpNode*>(node); |
| 2643 | node = seq_node->on_success(); |
| 2644 | } |
| 2645 | return read_backward() ? -length : length; |
| 2646 | } |
| 2647 | |
| 2648 | void LoopChoiceNode::AddLoopAlternative(GuardedAlternative alt) { |
| 2649 | ASSERT(loop_node_ == NULL); |
| 2650 | AddAlternative(alt); |
| 2651 | loop_node_ = alt.node(); |
| 2652 | } |
| 2653 | |
| 2654 | void LoopChoiceNode::AddContinueAlternative(GuardedAlternative alt) { |
| 2655 | ASSERT(continue_node_ == NULL); |
| 2656 | AddAlternative(alt); |
| 2657 | continue_node_ = alt.node(); |
| 2658 | } |
| 2659 | |
| 2660 | void LoopChoiceNode::Emit(RegExpCompiler* compiler, Trace* trace) { |
| 2661 | RegExpMacroAssembler* macro_assembler = compiler->macro_assembler(); |
| 2662 | if (trace->stop_node() == this) { |
| 2663 | // Back edge of greedy optimized loop node graph. |
| 2664 | intptr_t text_length = |
| 2665 | GreedyLoopTextLengthForAlternative(&alternatives_->At(0)); |
| 2666 | ASSERT(text_length != kNodeIsTooComplexForGreedyLoops); |
| 2667 | // Update the counter-based backtracking info on the stack. This is an |
| 2668 | // optimization for greedy loops (see below). |
| 2669 | ASSERT(trace->cp_offset() == text_length); |
| 2670 | macro_assembler->AdvanceCurrentPosition(text_length); |
| 2671 | macro_assembler->GoTo(trace->loop_label()); |
| 2672 | return; |
| 2673 | } |
| 2674 | ASSERT(trace->stop_node() == NULL); |
| 2675 | if (!trace->is_trivial()) { |
| 2676 | trace->Flush(compiler, this); |
| 2677 | return; |
| 2678 | } |
| 2679 | ChoiceNode::Emit(compiler, trace); |
| 2680 | } |
| 2681 | |
| 2682 | intptr_t ChoiceNode::CalculatePreloadCharacters(RegExpCompiler* compiler, |
| 2683 | intptr_t eats_at_least) { |
| 2684 | intptr_t preload_characters = |
| 2685 | Utils::Minimum(static_cast<intptr_t>(4), eats_at_least); |
| 2686 | if (compiler->macro_assembler()->CanReadUnaligned()) { |
| 2687 | bool one_byte = compiler->one_byte(); |
| 2688 | if (one_byte) { |
| 2689 | if (preload_characters > 4) preload_characters = 4; |
| 2690 | // We can't preload 3 characters because there is no machine instruction |
| 2691 | // to do that. We can't just load 4 because we could be reading |
| 2692 | // beyond the end of the string, which could cause a memory fault. |
| 2693 | if (preload_characters == 3) preload_characters = 2; |
| 2694 | } else { |
| 2695 | if (preload_characters > 2) preload_characters = 2; |
| 2696 | } |
| 2697 | } else { |
| 2698 | if (preload_characters > 1) preload_characters = 1; |
| 2699 | } |
| 2700 | return preload_characters; |
| 2701 | } |
| 2702 | |
| 2703 | // This structure is used when generating the alternatives in a choice node. It |
| 2704 | // records the way the alternative is being code generated. |
| 2705 | struct AlternativeGeneration { |
| 2706 | AlternativeGeneration() |
| 2707 | : possible_success(), |
| 2708 | expects_preload(false), |
| 2709 | after(), |
| 2710 | quick_check_details() {} |
| 2711 | BlockLabel possible_success; |
| 2712 | bool expects_preload; |
| 2713 | BlockLabel after; |
| 2714 | QuickCheckDetails quick_check_details; |
| 2715 | }; |
| 2716 | |
| 2717 | // Creates a list of AlternativeGenerations. If the list has a reasonable |
| 2718 | // size then it is on the stack, otherwise the excess is on the heap. |
| 2719 | class AlternativeGenerationList { |
| 2720 | public: |
| 2721 | explicit AlternativeGenerationList(intptr_t count) : count_(count) { |
| 2722 | ASSERT(count >= 0); |
| 2723 | if (count > kAFew) { |
| 2724 | excess_alt_gens_.reset(new AlternativeGeneration[count - kAFew]); |
| 2725 | } |
| 2726 | } |
| 2727 | |
| 2728 | AlternativeGeneration* at(intptr_t i) { |
| 2729 | ASSERT(0 <= i); |
| 2730 | ASSERT(i < count_); |
| 2731 | if (i < kAFew) { |
| 2732 | return &a_few_alt_gens_[i]; |
| 2733 | } |
| 2734 | return &excess_alt_gens_[i - kAFew]; |
| 2735 | } |
| 2736 | |
| 2737 | private: |
| 2738 | static const intptr_t kAFew = 10; |
| 2739 | |
| 2740 | intptr_t count_; |
| 2741 | AlternativeGeneration a_few_alt_gens_[kAFew]; |
| 2742 | std::unique_ptr<AlternativeGeneration[]> excess_alt_gens_; |
| 2743 | |
| 2744 | DISALLOW_ALLOCATION(); |
| 2745 | DISALLOW_COPY_AND_ASSIGN(AlternativeGenerationList); |
| 2746 | }; |
| 2747 | |
| 2748 | static const int32_t kRangeEndMarker = Utf::kMaxCodePoint + 1; |
| 2749 | |
| 2750 | // The '2' variant is inclusive from and exclusive to. |
| 2751 | // This covers \s as defined in ECMA-262 5.1, 15.10.2.12, |
| 2752 | // which include WhiteSpace (7.2) or LineTerminator (7.3) values. |
| 2753 | // 0x180E has been removed from Unicode's Zs category and thus |
| 2754 | // from ECMAScript's WhiteSpace category as of Unicode 6.3. |
| 2755 | static const int32_t kSpaceRanges[] = { |
| 2756 | '\t', '\r' + 1, ' ', ' ' + 1, 0x00A0, 0x00A1, 0x1680, |
| 2757 | 0x1681, 0x2000, 0x200B, 0x2028, 0x202A, 0x202F, 0x2030, |
| 2758 | 0x205F, 0x2060, 0x3000, 0x3001, 0xFEFF, 0xFF00, kRangeEndMarker}; |
| 2759 | static const intptr_t kSpaceRangeCount = ARRAY_SIZE(kSpaceRanges); |
| 2760 | static const int32_t kWordRanges[] = { |
| 2761 | '0', '9' + 1, 'A', 'Z' + 1, '_', '_' + 1, 'a', 'z' + 1, kRangeEndMarker}; |
| 2762 | static const intptr_t kWordRangeCount = ARRAY_SIZE(kWordRanges); |
| 2763 | static const int32_t kDigitRanges[] = {'0', '9' + 1, kRangeEndMarker}; |
| 2764 | static const intptr_t kDigitRangeCount = ARRAY_SIZE(kDigitRanges); |
| 2765 | static const int32_t kSurrogateRanges[] = {0xd800, 0xe000, kRangeEndMarker}; |
| 2766 | static const intptr_t kSurrogateRangeCount = ARRAY_SIZE(kSurrogateRanges); |
| 2767 | static const int32_t kLineTerminatorRanges[] = { |
| 2768 | 0x000A, 0x000B, 0x000D, 0x000E, 0x2028, 0x202A, kRangeEndMarker}; |
| 2769 | static const intptr_t kLineTerminatorRangeCount = |
| 2770 | ARRAY_SIZE(kLineTerminatorRanges); |
| 2771 | |
| 2772 | void BoyerMoorePositionInfo::Set(intptr_t character) { |
| 2773 | SetInterval(Interval(character, character)); |
| 2774 | } |
| 2775 | |
| 2776 | void BoyerMoorePositionInfo::SetInterval(const Interval& interval) { |
| 2777 | s_ = AddRange(s_, kSpaceRanges, kSpaceRangeCount, interval); |
| 2778 | w_ = AddRange(w_, kWordRanges, kWordRangeCount, interval); |
| 2779 | d_ = AddRange(d_, kDigitRanges, kDigitRangeCount, interval); |
| 2780 | surrogate_ = |
| 2781 | AddRange(surrogate_, kSurrogateRanges, kSurrogateRangeCount, interval); |
| 2782 | if (interval.to() - interval.from() >= kMapSize - 1) { |
| 2783 | if (map_count_ != kMapSize) { |
| 2784 | map_count_ = kMapSize; |
| 2785 | for (intptr_t i = 0; i < kMapSize; i++) |
| 2786 | (*map_)[i] = true; |
| 2787 | } |
| 2788 | return; |
| 2789 | } |
| 2790 | for (intptr_t i = interval.from(); i <= interval.to(); i++) { |
| 2791 | intptr_t mod_character = (i & kMask); |
| 2792 | if (!map_->At(mod_character)) { |
| 2793 | map_count_++; |
| 2794 | (*map_)[mod_character] = true; |
| 2795 | } |
| 2796 | if (map_count_ == kMapSize) return; |
| 2797 | } |
| 2798 | } |
| 2799 | |
| 2800 | void BoyerMoorePositionInfo::SetAll() { |
| 2801 | s_ = w_ = d_ = kLatticeUnknown; |
| 2802 | if (map_count_ != kMapSize) { |
| 2803 | map_count_ = kMapSize; |
| 2804 | for (intptr_t i = 0; i < kMapSize; i++) |
| 2805 | (*map_)[i] = true; |
| 2806 | } |
| 2807 | } |
| 2808 | |
| 2809 | BoyerMooreLookahead::BoyerMooreLookahead(intptr_t length, |
| 2810 | RegExpCompiler* compiler, |
| 2811 | Zone* zone) |
| 2812 | : length_(length), compiler_(compiler) { |
| 2813 | if (compiler->one_byte()) { |
| 2814 | max_char_ = Symbols::kMaxOneCharCodeSymbol; |
| 2815 | } else { |
| 2816 | max_char_ = Utf16::kMaxCodeUnit; |
| 2817 | } |
| 2818 | bitmaps_ = new (zone) ZoneGrowableArray<BoyerMoorePositionInfo*>(length); |
| 2819 | for (intptr_t i = 0; i < length; i++) { |
| 2820 | bitmaps_->Add(new (zone) BoyerMoorePositionInfo(zone)); |
| 2821 | } |
| 2822 | } |
| 2823 | |
| 2824 | // Find the longest range of lookahead that has the fewest number of different |
| 2825 | // characters that can occur at a given position. Since we are optimizing two |
| 2826 | // different parameters at once this is a tradeoff. |
| 2827 | bool BoyerMooreLookahead::FindWorthwhileInterval(intptr_t* from, intptr_t* to) { |
| 2828 | intptr_t biggest_points = 0; |
| 2829 | // If more than 32 characters out of 128 can occur it is unlikely that we can |
| 2830 | // be lucky enough to step forwards much of the time. |
| 2831 | const intptr_t kMaxMax = 32; |
| 2832 | for (intptr_t max_number_of_chars = 4; max_number_of_chars < kMaxMax; |
| 2833 | max_number_of_chars *= 2) { |
| 2834 | biggest_points = |
| 2835 | FindBestInterval(max_number_of_chars, biggest_points, from, to); |
| 2836 | } |
| 2837 | if (biggest_points == 0) return false; |
| 2838 | return true; |
| 2839 | } |
| 2840 | |
| 2841 | // Find the highest-points range between 0 and length_ where the character |
| 2842 | // information is not too vague. 'Too vague' means that there are more than |
| 2843 | // max_number_of_chars that can occur at this position. Calculates the number |
| 2844 | // of points as the product of width-of-the-range and |
| 2845 | // probability-of-finding-one-of-the-characters, where the probability is |
| 2846 | // calculated using the frequency distribution of the sample subject string. |
| 2847 | intptr_t BoyerMooreLookahead::FindBestInterval(intptr_t max_number_of_chars, |
| 2848 | intptr_t old_biggest_points, |
| 2849 | intptr_t* from, |
| 2850 | intptr_t* to) { |
| 2851 | intptr_t biggest_points = old_biggest_points; |
| 2852 | static const intptr_t kSize = RegExpMacroAssembler::kTableSize; |
| 2853 | for (intptr_t i = 0; i < length_;) { |
| 2854 | while (i < length_ && Count(i) > max_number_of_chars) |
| 2855 | i++; |
| 2856 | if (i == length_) break; |
| 2857 | intptr_t remembered_from = i; |
| 2858 | bool union_map[kSize]; |
| 2859 | for (intptr_t j = 0; j < kSize; j++) |
| 2860 | union_map[j] = false; |
| 2861 | while (i < length_ && Count(i) <= max_number_of_chars) { |
| 2862 | BoyerMoorePositionInfo* map = bitmaps_->At(i); |
| 2863 | for (intptr_t j = 0; j < kSize; j++) |
| 2864 | union_map[j] |= map->at(j); |
| 2865 | i++; |
| 2866 | } |
| 2867 | intptr_t frequency = 0; |
| 2868 | for (intptr_t j = 0; j < kSize; j++) { |
| 2869 | if (union_map[j]) { |
| 2870 | // Add 1 to the frequency to give a small per-character boost for |
| 2871 | // the cases where our sampling is not good enough and many |
| 2872 | // characters have a frequency of zero. This means the frequency |
| 2873 | // can theoretically be up to 2*kSize though we treat it mostly as |
| 2874 | // a fraction of kSize. |
| 2875 | frequency += compiler_->frequency_collator()->Frequency(j) + 1; |
| 2876 | } |
| 2877 | } |
| 2878 | // We use the probability of skipping times the distance we are skipping to |
| 2879 | // judge the effectiveness of this. Actually we have a cut-off: By |
| 2880 | // dividing by 2 we switch off the skipping if the probability of skipping |
| 2881 | // is less than 50%. This is because the multibyte mask-and-compare |
| 2882 | // skipping in quickcheck is more likely to do well on this case. |
| 2883 | bool in_quickcheck_range = |
| 2884 | ((i - remembered_from < 4) || |
| 2885 | (compiler_->one_byte() ? remembered_from <= 4 : remembered_from <= 2)); |
| 2886 | // Called 'probability' but it is only a rough estimate and can actually |
| 2887 | // be outside the 0-kSize range. |
| 2888 | intptr_t probability = |
| 2889 | (in_quickcheck_range ? kSize / 2 : kSize) - frequency; |
| 2890 | intptr_t points = (i - remembered_from) * probability; |
| 2891 | if (points > biggest_points) { |
| 2892 | *from = remembered_from; |
| 2893 | *to = i - 1; |
| 2894 | biggest_points = points; |
| 2895 | } |
| 2896 | } |
| 2897 | return biggest_points; |
| 2898 | } |
| 2899 | |
| 2900 | // Take all the characters that will not prevent a successful match if they |
| 2901 | // occur in the subject string in the range between min_lookahead and |
| 2902 | // max_lookahead (inclusive) measured from the current position. If the |
| 2903 | // character at max_lookahead offset is not one of these characters, then we |
| 2904 | // can safely skip forwards by the number of characters in the range. |
| 2905 | intptr_t BoyerMooreLookahead::GetSkipTable( |
| 2906 | intptr_t min_lookahead, |
| 2907 | intptr_t max_lookahead, |
| 2908 | const TypedData& boolean_skip_table) { |
| 2909 | const intptr_t kSize = RegExpMacroAssembler::kTableSize; |
| 2910 | |
| 2911 | const intptr_t kSkipArrayEntry = 0; |
| 2912 | const intptr_t kDontSkipArrayEntry = 1; |
| 2913 | |
| 2914 | for (intptr_t i = 0; i < kSize; i++) { |
| 2915 | boolean_skip_table.SetUint8(i, kSkipArrayEntry); |
| 2916 | } |
| 2917 | intptr_t skip = max_lookahead + 1 - min_lookahead; |
| 2918 | |
| 2919 | for (intptr_t i = max_lookahead; i >= min_lookahead; i--) { |
| 2920 | BoyerMoorePositionInfo* map = bitmaps_->At(i); |
| 2921 | for (intptr_t j = 0; j < kSize; j++) { |
| 2922 | if (map->at(j)) { |
| 2923 | boolean_skip_table.SetUint8(j, kDontSkipArrayEntry); |
| 2924 | } |
| 2925 | } |
| 2926 | } |
| 2927 | |
| 2928 | return skip; |
| 2929 | } |
| 2930 | |
| 2931 | // See comment above on the implementation of GetSkipTable. |
| 2932 | void BoyerMooreLookahead::EmitSkipInstructions(RegExpMacroAssembler* masm) { |
| 2933 | const intptr_t kSize = RegExpMacroAssembler::kTableSize; |
| 2934 | |
| 2935 | intptr_t min_lookahead = 0; |
| 2936 | intptr_t max_lookahead = 0; |
| 2937 | |
| 2938 | if (!FindWorthwhileInterval(&min_lookahead, &max_lookahead)) return; |
| 2939 | |
| 2940 | bool found_single_character = false; |
| 2941 | intptr_t single_character = 0; |
| 2942 | for (intptr_t i = max_lookahead; i >= min_lookahead; i--) { |
| 2943 | BoyerMoorePositionInfo* map = bitmaps_->At(i); |
| 2944 | if (map->map_count() > 1 || |
| 2945 | (found_single_character && map->map_count() != 0)) { |
| 2946 | found_single_character = false; |
| 2947 | break; |
| 2948 | } |
| 2949 | for (intptr_t j = 0; j < kSize; j++) { |
| 2950 | if (map->at(j)) { |
| 2951 | found_single_character = true; |
| 2952 | single_character = j; |
| 2953 | break; |
| 2954 | } |
| 2955 | } |
| 2956 | } |
| 2957 | |
| 2958 | intptr_t lookahead_width = max_lookahead + 1 - min_lookahead; |
| 2959 | |
| 2960 | if (found_single_character && lookahead_width == 1 && max_lookahead < 3) { |
| 2961 | // The mask-compare can probably handle this better. |
| 2962 | return; |
| 2963 | } |
| 2964 | |
| 2965 | if (found_single_character) { |
| 2966 | BlockLabel cont, again; |
| 2967 | masm->BindBlock(&again); |
| 2968 | masm->LoadCurrentCharacter(max_lookahead, &cont, true); |
| 2969 | if (max_char_ > kSize) { |
| 2970 | masm->CheckCharacterAfterAnd(single_character, |
| 2971 | RegExpMacroAssembler::kTableMask, &cont); |
| 2972 | } else { |
| 2973 | masm->CheckCharacter(single_character, &cont); |
| 2974 | } |
| 2975 | masm->AdvanceCurrentPosition(lookahead_width); |
| 2976 | masm->GoTo(&again); |
| 2977 | masm->BindBlock(&cont); |
| 2978 | return; |
| 2979 | } |
| 2980 | |
| 2981 | const TypedData& boolean_skip_table = TypedData::ZoneHandle( |
| 2982 | compiler_->zone(), |
| 2983 | TypedData::New(kTypedDataUint8ArrayCid, kSize, Heap::kOld)); |
| 2984 | intptr_t skip_distance = |
| 2985 | GetSkipTable(min_lookahead, max_lookahead, boolean_skip_table); |
| 2986 | ASSERT(skip_distance != 0); |
| 2987 | |
| 2988 | BlockLabel cont, again; |
| 2989 | |
| 2990 | masm->BindBlock(&again); |
| 2991 | masm->CheckPreemption(/*is_backtrack=*/false); |
| 2992 | masm->LoadCurrentCharacter(max_lookahead, &cont, true); |
| 2993 | masm->CheckBitInTable(boolean_skip_table, &cont); |
| 2994 | masm->AdvanceCurrentPosition(skip_distance); |
| 2995 | masm->GoTo(&again); |
| 2996 | masm->BindBlock(&cont); |
| 2997 | |
| 2998 | return; |
| 2999 | } |
| 3000 | |
| 3001 | /* Code generation for choice nodes. |
| 3002 | * |
| 3003 | * We generate quick checks that do a mask and compare to eliminate a |
| 3004 | * choice. If the quick check succeeds then it jumps to the continuation to |
| 3005 | * do slow checks and check subsequent nodes. If it fails (the common case) |
| 3006 | * it falls through to the next choice. |
| 3007 | * |
| 3008 | * Here is the desired flow graph. Nodes directly below each other imply |
| 3009 | * fallthrough. Alternatives 1 and 2 have quick checks. Alternative |
| 3010 | * 3 doesn't have a quick check so we have to call the slow check. |
| 3011 | * Nodes are marked Qn for quick checks and Sn for slow checks. The entire |
| 3012 | * regexp continuation is generated directly after the Sn node, up to the |
| 3013 | * next GoTo if we decide to reuse some already generated code. Some |
| 3014 | * nodes expect preload_characters to be preloaded into the current |
| 3015 | * character register. R nodes do this preloading. Vertices are marked |
| 3016 | * F for failures and S for success (possible success in the case of quick |
| 3017 | * nodes). L, V, < and > are used as arrow heads. |
| 3018 | * |
| 3019 | * ----------> R |
| 3020 | * | |
| 3021 | * V |
| 3022 | * Q1 -----> S1 |
| 3023 | * | S / |
| 3024 | * F| / |
| 3025 | * | F/ |
| 3026 | * | / |
| 3027 | * | R |
| 3028 | * | / |
| 3029 | * V L |
| 3030 | * Q2 -----> S2 |
| 3031 | * | S / |
| 3032 | * F| / |
| 3033 | * | F/ |
| 3034 | * | / |
| 3035 | * | R |
| 3036 | * | / |
| 3037 | * V L |
| 3038 | * S3 |
| 3039 | * | |
| 3040 | * F| |
| 3041 | * | |
| 3042 | * R |
| 3043 | * | |
| 3044 | * backtrack V |
| 3045 | * <----------Q4 |
| 3046 | * \ F | |
| 3047 | * \ |S |
| 3048 | * \ F V |
| 3049 | * \-----S4 |
| 3050 | * |
| 3051 | * For greedy loops we push the current position, then generate the code that |
| 3052 | * eats the input specially in EmitGreedyLoop. The other choice (the |
| 3053 | * continuation) is generated by the normal code in EmitChoices, and steps back |
| 3054 | * in the input to the starting position when it fails to match. The loop code |
| 3055 | * looks like this (U is the unwind code that steps back in the greedy loop). |
| 3056 | * |
| 3057 | * _____ |
| 3058 | * / \ |
| 3059 | * V | |
| 3060 | * ----------> S1 | |
| 3061 | * /| | |
| 3062 | * / |S | |
| 3063 | * F/ \_____/ |
| 3064 | * / |
| 3065 | * |<----- |
| 3066 | * | \ |
| 3067 | * V |S |
| 3068 | * Q2 ---> U----->backtrack |
| 3069 | * | F / |
| 3070 | * S| / |
| 3071 | * V F / |
| 3072 | * S2--/ |
| 3073 | */ |
| 3074 | |
| 3075 | GreedyLoopState::GreedyLoopState(bool not_at_start) { |
| 3076 | counter_backtrack_trace_.set_backtrack(&label_); |
| 3077 | if (not_at_start) counter_backtrack_trace_.set_at_start(Trace::FALSE_VALUE); |
| 3078 | } |
| 3079 | |
| 3080 | void ChoiceNode::AssertGuardsMentionRegisters(Trace* trace) { |
| 3081 | #ifdef DEBUG |
| 3082 | intptr_t choice_count = alternatives_->length(); |
| 3083 | for (intptr_t i = 0; i < choice_count - 1; i++) { |
| 3084 | GuardedAlternative alternative = alternatives_->At(i); |
| 3085 | ZoneGrowableArray<Guard*>* guards = alternative.guards(); |
| 3086 | intptr_t guard_count = (guards == NULL) ? 0 : guards->length(); |
| 3087 | for (intptr_t j = 0; j < guard_count; j++) { |
| 3088 | ASSERT(!trace->mentions_reg(guards->At(j)->reg())); |
| 3089 | } |
| 3090 | } |
| 3091 | #endif |
| 3092 | } |
| 3093 | |
| 3094 | void ChoiceNode::SetUpPreLoad(RegExpCompiler* compiler, |
| 3095 | Trace* current_trace, |
| 3096 | PreloadState* state) { |
| 3097 | if (state->eats_at_least_ == PreloadState::kEatsAtLeastNotYetInitialized) { |
| 3098 | // Save some time by looking at most one machine word ahead. |
| 3099 | state->eats_at_least_ = |
| 3100 | EatsAtLeast(compiler->one_byte() ? 4 : 2, kRecursionBudget, |
| 3101 | current_trace->at_start() == Trace::FALSE_VALUE); |
| 3102 | } |
| 3103 | state->preload_characters_ = |
| 3104 | CalculatePreloadCharacters(compiler, state->eats_at_least_); |
| 3105 | |
| 3106 | state->preload_is_current_ = |
| 3107 | (current_trace->characters_preloaded() == state->preload_characters_); |
| 3108 | state->preload_has_checked_bounds_ = state->preload_is_current_; |
| 3109 | } |
| 3110 | |
| 3111 | void ChoiceNode::Emit(RegExpCompiler* compiler, Trace* trace) { |
| 3112 | intptr_t choice_count = alternatives_->length(); |
| 3113 | |
| 3114 | if (choice_count == 1 && alternatives_->At(0).guards() == nullptr) { |
| 3115 | alternatives_->At(0).node()->Emit(compiler, trace); |
| 3116 | return; |
| 3117 | } |
| 3118 | |
| 3119 | AssertGuardsMentionRegisters(trace); |
| 3120 | |
| 3121 | LimitResult limit_result = LimitVersions(compiler, trace); |
| 3122 | if (limit_result == DONE) return; |
| 3123 | ASSERT(limit_result == CONTINUE); |
| 3124 | |
| 3125 | // For loop nodes we already flushed (see LoopChoiceNode::Emit), but for |
| 3126 | // other choice nodes we only flush if we are out of code size budget. |
| 3127 | if (trace->flush_budget() == 0 && trace->actions() != NULL) { |
| 3128 | trace->Flush(compiler, this); |
| 3129 | return; |
| 3130 | } |
| 3131 | |
| 3132 | RecursionCheck rc(compiler); |
| 3133 | |
| 3134 | PreloadState preload; |
| 3135 | preload.init(); |
| 3136 | GreedyLoopState greedy_loop_state(not_at_start()); |
| 3137 | |
| 3138 | intptr_t text_length = |
| 3139 | GreedyLoopTextLengthForAlternative(&alternatives_->At(0)); |
| 3140 | AlternativeGenerationList alt_gens(choice_count); |
| 3141 | |
| 3142 | if (choice_count > 1 && text_length != kNodeIsTooComplexForGreedyLoops) { |
| 3143 | trace = EmitGreedyLoop(compiler, trace, &alt_gens, &preload, |
| 3144 | &greedy_loop_state, text_length); |
| 3145 | } else { |
| 3146 | // TODO(erikcorry): Delete this. We don't need this label, but it makes us |
| 3147 | // match the traces produced pre-cleanup. |
| 3148 | BlockLabel second_choice; |
| 3149 | compiler->macro_assembler()->BindBlock(&second_choice); |
| 3150 | |
| 3151 | preload.eats_at_least_ = EmitOptimizedUnanchoredSearch(compiler, trace); |
| 3152 | |
| 3153 | EmitChoices(compiler, &alt_gens, 0, trace, &preload); |
| 3154 | } |
| 3155 | |
| 3156 | // At this point we need to generate slow checks for the alternatives where |
| 3157 | // the quick check was inlined. We can recognize these because the associated |
| 3158 | // label was bound. |
| 3159 | intptr_t new_flush_budget = trace->flush_budget() / choice_count; |
| 3160 | for (intptr_t i = 0; i < choice_count; i++) { |
| 3161 | AlternativeGeneration* alt_gen = alt_gens.at(i); |
| 3162 | Trace new_trace(*trace); |
| 3163 | // If there are actions to be flushed we have to limit how many times |
| 3164 | // they are flushed. Take the budget of the parent trace and distribute |
| 3165 | // it fairly amongst the children. |
| 3166 | if (new_trace.actions() != NULL) { |
| 3167 | new_trace.set_flush_budget(new_flush_budget); |
| 3168 | } |
| 3169 | bool next_expects_preload = |
| 3170 | i == choice_count - 1 ? false : alt_gens.at(i + 1)->expects_preload; |
| 3171 | EmitOutOfLineContinuation(compiler, &new_trace, alternatives_->At(i), |
| 3172 | alt_gen, preload.preload_characters_, |
| 3173 | next_expects_preload); |
| 3174 | } |
| 3175 | } |
| 3176 | |
| 3177 | Trace* ChoiceNode::EmitGreedyLoop(RegExpCompiler* compiler, |
| 3178 | Trace* trace, |
| 3179 | AlternativeGenerationList* alt_gens, |
| 3180 | PreloadState* preload, |
| 3181 | GreedyLoopState* greedy_loop_state, |
| 3182 | intptr_t text_length) { |
| 3183 | RegExpMacroAssembler* macro_assembler = compiler->macro_assembler(); |
| 3184 | // Here we have special handling for greedy loops containing only text nodes |
| 3185 | // and other simple nodes. These are handled by pushing the current |
| 3186 | // position on the stack and then incrementing the current position each |
| 3187 | // time around the switch. On backtrack we decrement the current position |
| 3188 | // and check it against the pushed value. This avoids pushing backtrack |
| 3189 | // information for each iteration of the loop, which could take up a lot of |
| 3190 | // space. |
| 3191 | ASSERT(trace->stop_node() == NULL); |
| 3192 | macro_assembler->PushCurrentPosition(); |
| 3193 | BlockLabel greedy_match_failed; |
| 3194 | Trace greedy_match_trace; |
| 3195 | if (not_at_start()) greedy_match_trace.set_at_start(Trace::FALSE_VALUE); |
| 3196 | greedy_match_trace.set_backtrack(&greedy_match_failed); |
| 3197 | BlockLabel loop_label; |
| 3198 | macro_assembler->BindBlock(&loop_label); |
| 3199 | macro_assembler->CheckPreemption(/*is_backtrack=*/false); |
| 3200 | greedy_match_trace.set_stop_node(this); |
| 3201 | greedy_match_trace.set_loop_label(&loop_label); |
| 3202 | (*alternatives_)[0].node()->Emit(compiler, &greedy_match_trace); |
| 3203 | macro_assembler->BindBlock(&greedy_match_failed); |
| 3204 | |
| 3205 | BlockLabel second_choice; // For use in greedy matches. |
| 3206 | macro_assembler->BindBlock(&second_choice); |
| 3207 | |
| 3208 | Trace* new_trace = greedy_loop_state->counter_backtrack_trace(); |
| 3209 | |
| 3210 | EmitChoices(compiler, alt_gens, 1, new_trace, preload); |
| 3211 | |
| 3212 | macro_assembler->BindBlock(greedy_loop_state->label()); |
| 3213 | // If we have unwound to the bottom then backtrack. |
| 3214 | macro_assembler->CheckGreedyLoop(trace->backtrack()); |
| 3215 | // Otherwise try the second priority at an earlier position. |
| 3216 | macro_assembler->AdvanceCurrentPosition(-text_length); |
| 3217 | macro_assembler->GoTo(&second_choice); |
| 3218 | return new_trace; |
| 3219 | } |
| 3220 | |
| 3221 | intptr_t ChoiceNode::EmitOptimizedUnanchoredSearch(RegExpCompiler* compiler, |
| 3222 | Trace* trace) { |
| 3223 | intptr_t eats_at_least = PreloadState::kEatsAtLeastNotYetInitialized; |
| 3224 | if (alternatives_->length() != 2) return eats_at_least; |
| 3225 | |
| 3226 | GuardedAlternative alt1 = alternatives_->At(1); |
| 3227 | if (alt1.guards() != NULL && alt1.guards()->length() != 0) { |
| 3228 | return eats_at_least; |
| 3229 | } |
| 3230 | RegExpNode* eats_anything_node = alt1.node(); |
| 3231 | if (eats_anything_node->GetSuccessorOfOmnivorousTextNode(compiler) != this) { |
| 3232 | return eats_at_least; |
| 3233 | } |
| 3234 | |
| 3235 | // Really we should be creating a new trace when we execute this function, |
| 3236 | // but there is no need, because the code it generates cannot backtrack, and |
| 3237 | // we always arrive here with a trivial trace (since it's the entry to a |
| 3238 | // loop. That also implies that there are no preloaded characters, which is |
| 3239 | // good, because it means we won't be violating any assumptions by |
| 3240 | // overwriting those characters with new load instructions. |
| 3241 | ASSERT(trace->is_trivial()); |
| 3242 | |
| 3243 | RegExpMacroAssembler* macro_assembler = compiler->macro_assembler(); |
| 3244 | // At this point we know that we are at a non-greedy loop that will eat |
| 3245 | // any character one at a time. Any non-anchored regexp has such a |
| 3246 | // loop prepended to it in order to find where it starts. We look for |
| 3247 | // a pattern of the form ...abc... where we can look 6 characters ahead |
| 3248 | // and step forwards 3 if the character is not one of abc. Abc need |
| 3249 | // not be atoms, they can be any reasonably limited character class or |
| 3250 | // small alternation. |
| 3251 | BoyerMooreLookahead* bm = bm_info(false); |
| 3252 | if (bm == NULL) { |
| 3253 | eats_at_least = Utils::Minimum( |
| 3254 | kMaxLookaheadForBoyerMoore, |
| 3255 | EatsAtLeast(kMaxLookaheadForBoyerMoore, kRecursionBudget, false)); |
| 3256 | if (eats_at_least >= 1) { |
| 3257 | bm = new (Z) BoyerMooreLookahead(eats_at_least, compiler, Z); |
| 3258 | GuardedAlternative alt0 = alternatives_->At(0); |
| 3259 | alt0.node()->FillInBMInfo(0, kRecursionBudget, bm, false); |
| 3260 | } |
| 3261 | } |
| 3262 | if (bm != NULL) { |
| 3263 | bm->EmitSkipInstructions(macro_assembler); |
| 3264 | } |
| 3265 | return eats_at_least; |
| 3266 | } |
| 3267 | |
| 3268 | void ChoiceNode::EmitChoices(RegExpCompiler* compiler, |
| 3269 | AlternativeGenerationList* alt_gens, |
| 3270 | intptr_t first_choice, |
| 3271 | Trace* trace, |
| 3272 | PreloadState* preload) { |
| 3273 | RegExpMacroAssembler* macro_assembler = compiler->macro_assembler(); |
| 3274 | SetUpPreLoad(compiler, trace, preload); |
| 3275 | |
| 3276 | // For now we just call all choices one after the other. The idea ultimately |
| 3277 | // is to use the Dispatch table to try only the relevant ones. |
| 3278 | intptr_t choice_count = alternatives_->length(); |
| 3279 | |
| 3280 | intptr_t new_flush_budget = trace->flush_budget() / choice_count; |
| 3281 | |
| 3282 | for (intptr_t i = first_choice; i < choice_count; i++) { |
| 3283 | bool is_last = i == choice_count - 1; |
| 3284 | bool fall_through_on_failure = !is_last; |
| 3285 | GuardedAlternative alternative = alternatives_->At(i); |
| 3286 | AlternativeGeneration* alt_gen = alt_gens->at(i); |
| 3287 | alt_gen->quick_check_details.set_characters(preload->preload_characters_); |
| 3288 | ZoneGrowableArray<Guard*>* guards = alternative.guards(); |
| 3289 | intptr_t guard_count = (guards == NULL) ? 0 : guards->length(); |
| 3290 | Trace new_trace(*trace); |
| 3291 | new_trace.set_characters_preloaded( |
| 3292 | preload->preload_is_current_ ? preload->preload_characters_ : 0); |
| 3293 | if (preload->preload_has_checked_bounds_) { |
| 3294 | new_trace.set_bound_checked_up_to(preload->preload_characters_); |
| 3295 | } |
| 3296 | new_trace.quick_check_performed()->Clear(); |
| 3297 | if (not_at_start_) new_trace.set_at_start(Trace::FALSE_VALUE); |
| 3298 | if (!is_last) { |
| 3299 | new_trace.set_backtrack(&alt_gen->after); |
| 3300 | } |
| 3301 | alt_gen->expects_preload = preload->preload_is_current_; |
| 3302 | bool generate_full_check_inline = false; |
| 3303 | if (kRegexpOptimization && |
| 3304 | try_to_emit_quick_check_for_alternative(i == 0) && |
| 3305 | alternative.node()->EmitQuickCheck( |
| 3306 | compiler, trace, &new_trace, preload->preload_has_checked_bounds_, |
| 3307 | &alt_gen->possible_success, &alt_gen->quick_check_details, |
| 3308 | fall_through_on_failure)) { |
| 3309 | // Quick check was generated for this choice. |
| 3310 | preload->preload_is_current_ = true; |
| 3311 | preload->preload_has_checked_bounds_ = true; |
| 3312 | // If we generated the quick check to fall through on possible success, |
| 3313 | // we now need to generate the full check inline. |
| 3314 | if (!fall_through_on_failure) { |
| 3315 | macro_assembler->BindBlock(&alt_gen->possible_success); |
| 3316 | new_trace.set_quick_check_performed(&alt_gen->quick_check_details); |
| 3317 | new_trace.set_characters_preloaded(preload->preload_characters_); |
| 3318 | new_trace.set_bound_checked_up_to(preload->preload_characters_); |
| 3319 | generate_full_check_inline = true; |
| 3320 | } |
| 3321 | } else if (alt_gen->quick_check_details.cannot_match()) { |
| 3322 | if (!fall_through_on_failure) { |
| 3323 | macro_assembler->GoTo(trace->backtrack()); |
| 3324 | } |
| 3325 | continue; |
| 3326 | } else { |
| 3327 | // No quick check was generated. Put the full code here. |
| 3328 | // If this is not the first choice then there could be slow checks from |
| 3329 | // previous cases that go here when they fail. There's no reason to |
| 3330 | // insist that they preload characters since the slow check we are about |
| 3331 | // to generate probably can't use it. |
| 3332 | if (i != first_choice) { |
| 3333 | alt_gen->expects_preload = false; |
| 3334 | new_trace.InvalidateCurrentCharacter(); |
| 3335 | } |
| 3336 | generate_full_check_inline = true; |
| 3337 | } |
| 3338 | if (generate_full_check_inline) { |
| 3339 | if (new_trace.actions() != NULL) { |
| 3340 | new_trace.set_flush_budget(new_flush_budget); |
| 3341 | } |
| 3342 | for (intptr_t j = 0; j < guard_count; j++) { |
| 3343 | GenerateGuard(macro_assembler, guards->At(j), &new_trace); |
| 3344 | } |
| 3345 | alternative.node()->Emit(compiler, &new_trace); |
| 3346 | preload->preload_is_current_ = false; |
| 3347 | } |
| 3348 | macro_assembler->BindBlock(&alt_gen->after); |
| 3349 | } |
| 3350 | } |
| 3351 | |
| 3352 | void ChoiceNode::EmitOutOfLineContinuation(RegExpCompiler* compiler, |
| 3353 | Trace* trace, |
| 3354 | GuardedAlternative alternative, |
| 3355 | AlternativeGeneration* alt_gen, |
| 3356 | intptr_t preload_characters, |
| 3357 | bool next_expects_preload) { |
| 3358 | if (!alt_gen->possible_success.is_linked()) return; |
| 3359 | |
| 3360 | RegExpMacroAssembler* macro_assembler = compiler->macro_assembler(); |
| 3361 | macro_assembler->BindBlock(&alt_gen->possible_success); |
| 3362 | Trace out_of_line_trace(*trace); |
| 3363 | out_of_line_trace.set_characters_preloaded(preload_characters); |
| 3364 | out_of_line_trace.set_quick_check_performed(&alt_gen->quick_check_details); |
| 3365 | if (not_at_start_) out_of_line_trace.set_at_start(Trace::FALSE_VALUE); |
| 3366 | ZoneGrowableArray<Guard*>* guards = alternative.guards(); |
| 3367 | intptr_t guard_count = (guards == NULL) ? 0 : guards->length(); |
| 3368 | if (next_expects_preload) { |
| 3369 | BlockLabel reload_current_char; |
| 3370 | out_of_line_trace.set_backtrack(&reload_current_char); |
| 3371 | for (intptr_t j = 0; j < guard_count; j++) { |
| 3372 | GenerateGuard(macro_assembler, guards->At(j), &out_of_line_trace); |
| 3373 | } |
| 3374 | alternative.node()->Emit(compiler, &out_of_line_trace); |
| 3375 | macro_assembler->BindBlock(&reload_current_char); |
| 3376 | // Reload the current character, since the next quick check expects that. |
| 3377 | // We don't need to check bounds here because we only get into this |
| 3378 | // code through a quick check which already did the checked load. |
| 3379 | macro_assembler->LoadCurrentCharacter(trace->cp_offset(), NULL, false, |
| 3380 | preload_characters); |
| 3381 | macro_assembler->GoTo(&(alt_gen->after)); |
| 3382 | } else { |
| 3383 | out_of_line_trace.set_backtrack(&(alt_gen->after)); |
| 3384 | for (intptr_t j = 0; j < guard_count; j++) { |
| 3385 | GenerateGuard(macro_assembler, guards->At(j), &out_of_line_trace); |
| 3386 | } |
| 3387 | alternative.node()->Emit(compiler, &out_of_line_trace); |
| 3388 | } |
| 3389 | } |
| 3390 | |
| 3391 | void ActionNode::Emit(RegExpCompiler* compiler, Trace* trace) { |
| 3392 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 3393 | LimitResult limit_result = LimitVersions(compiler, trace); |
| 3394 | if (limit_result == DONE) return; |
| 3395 | ASSERT(limit_result == CONTINUE); |
| 3396 | |
| 3397 | RecursionCheck rc(compiler); |
| 3398 | |
| 3399 | switch (action_type_) { |
| 3400 | case STORE_POSITION: { |
| 3401 | Trace::DeferredCapture new_capture(data_.u_position_register.reg, |
| 3402 | data_.u_position_register.is_capture, |
| 3403 | trace); |
| 3404 | Trace new_trace = *trace; |
| 3405 | new_trace.add_action(&new_capture); |
| 3406 | on_success()->Emit(compiler, &new_trace); |
| 3407 | break; |
| 3408 | } |
| 3409 | case INCREMENT_REGISTER: { |
| 3410 | Trace::DeferredIncrementRegister new_increment( |
| 3411 | data_.u_increment_register.reg); |
| 3412 | Trace new_trace = *trace; |
| 3413 | new_trace.add_action(&new_increment); |
| 3414 | on_success()->Emit(compiler, &new_trace); |
| 3415 | break; |
| 3416 | } |
| 3417 | case SET_REGISTER: { |
| 3418 | Trace::DeferredSetRegister new_set(data_.u_store_register.reg, |
| 3419 | data_.u_store_register.value); |
| 3420 | Trace new_trace = *trace; |
| 3421 | new_trace.add_action(&new_set); |
| 3422 | on_success()->Emit(compiler, &new_trace); |
| 3423 | break; |
| 3424 | } |
| 3425 | case CLEAR_CAPTURES: { |
| 3426 | Trace::DeferredClearCaptures new_capture(Interval( |
| 3427 | data_.u_clear_captures.range_from, data_.u_clear_captures.range_to)); |
| 3428 | Trace new_trace = *trace; |
| 3429 | new_trace.add_action(&new_capture); |
| 3430 | on_success()->Emit(compiler, &new_trace); |
| 3431 | break; |
| 3432 | } |
| 3433 | case BEGIN_SUBMATCH: |
| 3434 | if (!trace->is_trivial()) { |
| 3435 | trace->Flush(compiler, this); |
| 3436 | } else { |
| 3437 | assembler->WriteCurrentPositionToRegister( |
| 3438 | data_.u_submatch.current_position_register, 0); |
| 3439 | assembler->WriteStackPointerToRegister( |
| 3440 | data_.u_submatch.stack_pointer_register); |
| 3441 | on_success()->Emit(compiler, trace); |
| 3442 | } |
| 3443 | break; |
| 3444 | case EMPTY_MATCH_CHECK: { |
| 3445 | intptr_t start_pos_reg = data_.u_empty_match_check.start_register; |
| 3446 | intptr_t stored_pos = 0; |
| 3447 | intptr_t rep_reg = data_.u_empty_match_check.repetition_register; |
| 3448 | bool has_minimum = (rep_reg != RegExpCompiler::kNoRegister); |
| 3449 | bool know_dist = trace->GetStoredPosition(start_pos_reg, &stored_pos); |
| 3450 | if (know_dist && !has_minimum && stored_pos == trace->cp_offset()) { |
| 3451 | // If we know we haven't advanced and there is no minimum we |
| 3452 | // can just backtrack immediately. |
| 3453 | assembler->GoTo(trace->backtrack()); |
| 3454 | } else if (know_dist && stored_pos < trace->cp_offset()) { |
| 3455 | // If we know we've advanced we can generate the continuation |
| 3456 | // immediately. |
| 3457 | on_success()->Emit(compiler, trace); |
| 3458 | } else if (!trace->is_trivial()) { |
| 3459 | trace->Flush(compiler, this); |
| 3460 | } else { |
| 3461 | BlockLabel skip_empty_check; |
| 3462 | // If we have a minimum number of repetitions we check the current |
| 3463 | // number first and skip the empty check if it's not enough. |
| 3464 | if (has_minimum) { |
| 3465 | intptr_t limit = data_.u_empty_match_check.repetition_limit; |
| 3466 | assembler->IfRegisterLT(rep_reg, limit, &skip_empty_check); |
| 3467 | } |
| 3468 | // If the match is empty we bail out, otherwise we fall through |
| 3469 | // to the on-success continuation. |
| 3470 | assembler->IfRegisterEqPos(data_.u_empty_match_check.start_register, |
| 3471 | trace->backtrack()); |
| 3472 | assembler->BindBlock(&skip_empty_check); |
| 3473 | on_success()->Emit(compiler, trace); |
| 3474 | } |
| 3475 | break; |
| 3476 | } |
| 3477 | case POSITIVE_SUBMATCH_SUCCESS: { |
| 3478 | if (!trace->is_trivial()) { |
| 3479 | trace->Flush(compiler, this); |
| 3480 | return; |
| 3481 | } |
| 3482 | assembler->ReadCurrentPositionFromRegister( |
| 3483 | data_.u_submatch.current_position_register); |
| 3484 | assembler->ReadStackPointerFromRegister( |
| 3485 | data_.u_submatch.stack_pointer_register); |
| 3486 | intptr_t clear_register_count = data_.u_submatch.clear_register_count; |
| 3487 | if (clear_register_count == 0) { |
| 3488 | on_success()->Emit(compiler, trace); |
| 3489 | return; |
| 3490 | } |
| 3491 | intptr_t clear_registers_from = data_.u_submatch.clear_register_from; |
| 3492 | BlockLabel clear_registers_backtrack; |
| 3493 | Trace new_trace = *trace; |
| 3494 | new_trace.set_backtrack(&clear_registers_backtrack); |
| 3495 | on_success()->Emit(compiler, &new_trace); |
| 3496 | |
| 3497 | assembler->BindBlock(&clear_registers_backtrack); |
| 3498 | intptr_t clear_registers_to = |
| 3499 | clear_registers_from + clear_register_count - 1; |
| 3500 | assembler->ClearRegisters(clear_registers_from, clear_registers_to); |
| 3501 | |
| 3502 | ASSERT(trace->backtrack() == NULL); |
| 3503 | assembler->Backtrack(); |
| 3504 | return; |
| 3505 | } |
| 3506 | default: |
| 3507 | UNREACHABLE(); |
| 3508 | } |
| 3509 | } |
| 3510 | |
| 3511 | void BackReferenceNode::Emit(RegExpCompiler* compiler, Trace* trace) { |
| 3512 | RegExpMacroAssembler* assembler = compiler->macro_assembler(); |
| 3513 | if (!trace->is_trivial()) { |
| 3514 | trace->Flush(compiler, this); |
| 3515 | return; |
| 3516 | } |
| 3517 | |
| 3518 | LimitResult limit_result = LimitVersions(compiler, trace); |
| 3519 | if (limit_result == DONE) return; |
| 3520 | ASSERT(limit_result == CONTINUE); |
| 3521 | |
| 3522 | RecursionCheck rc(compiler); |
| 3523 | |
| 3524 | ASSERT(start_reg_ + 1 == end_reg_); |
| 3525 | if (flags_.IgnoreCase()) { |
| 3526 | assembler->CheckNotBackReferenceIgnoreCase( |
| 3527 | start_reg_, read_backward(), flags_.IsUnicode(), trace->backtrack()); |
| 3528 | } else { |
| 3529 | assembler->CheckNotBackReference(start_reg_, read_backward(), |
| 3530 | trace->backtrack()); |
| 3531 | } |
| 3532 | // We are going to advance backward, so we may end up at the start. |
| 3533 | if (read_backward()) trace->set_at_start(Trace::UNKNOWN); |
| 3534 | |
| 3535 | // Check that the back reference does not end inside a surrogate pair. |
| 3536 | if (flags_.IsUnicode() && !compiler->one_byte()) { |
| 3537 | assembler->CheckNotInSurrogatePair(trace->cp_offset(), trace->backtrack()); |
| 3538 | } |
| 3539 | |
| 3540 | on_success()->Emit(compiler, trace); |
| 3541 | } |
| 3542 | |
| 3543 | // ------------------------------------------------------------------- |
| 3544 | // Dot/dotty output |
| 3545 | |
| 3546 | #ifdef DEBUG |
| 3547 | |
| 3548 | class DotPrinter : public NodeVisitor { |
| 3549 | public: |
| 3550 | explicit DotPrinter(bool ignore_case) {} |
| 3551 | void PrintNode(const char* label, RegExpNode* node); |
| 3552 | void Visit(RegExpNode* node); |
| 3553 | void PrintAttributes(RegExpNode* from); |
| 3554 | void PrintOnFailure(RegExpNode* from, RegExpNode* to); |
| 3555 | #define DECLARE_VISIT(Type) virtual void Visit##Type(Type##Node* that); |
| 3556 | FOR_EACH_NODE_TYPE(DECLARE_VISIT) |
| 3557 | #undef DECLARE_VISIT |
| 3558 | }; |
| 3559 | |
| 3560 | void DotPrinter::PrintNode(const char* label, RegExpNode* node) { |
| 3561 | OS::PrintErr("digraph G {\n graph [label=\""); |
| 3562 | for (intptr_t i = 0; label[i] != '\0'; i++) { |
| 3563 | switch (label[i]) { |
| 3564 | case '\\': |
| 3565 | OS::PrintErr("\\\\"); |
| 3566 | break; |
| 3567 | case '"': |
| 3568 | OS::PrintErr("\""); |
| 3569 | break; |
| 3570 | default: |
| 3571 | OS::PrintErr("%c", label[i]); |
| 3572 | break; |
| 3573 | } |
| 3574 | } |
| 3575 | OS::PrintErr("\"];\n"); |
| 3576 | Visit(node); |
| 3577 | OS::PrintErr("}\n"); |
| 3578 | } |
| 3579 | |
| 3580 | void DotPrinter::Visit(RegExpNode* node) { |
| 3581 | if (node->info()->visited) return; |
| 3582 | node->info()->visited = true; |
| 3583 | node->Accept(this); |
| 3584 | } |
| 3585 | |
| 3586 | void DotPrinter::PrintOnFailure(RegExpNode* from, RegExpNode* on_failure) { |
| 3587 | OS::PrintErr(" n%p -> n%p [style=dotted];\n", from, on_failure); |
| 3588 | Visit(on_failure); |
| 3589 | } |
| 3590 | |
| 3591 | class AttributePrinter : public ValueObject { |
| 3592 | public: |
| 3593 | AttributePrinter() : first_(true) {} |
| 3594 | void PrintSeparator() { |
| 3595 | if (first_) { |
| 3596 | first_ = false; |
| 3597 | } else { |
| 3598 | OS::PrintErr("|"); |
| 3599 | } |
| 3600 | } |
| 3601 | void PrintBit(const char* name, bool value) { |
| 3602 | if (!value) return; |
| 3603 | PrintSeparator(); |
| 3604 | OS::PrintErr("{%s}", name); |
| 3605 | } |
| 3606 | void PrintPositive(const char* name, intptr_t value) { |
| 3607 | if (value < 0) return; |
| 3608 | PrintSeparator(); |
| 3609 | OS::PrintErr("{%s|%"Pd "}", name, value); |
| 3610 | } |
| 3611 | |
| 3612 | private: |
| 3613 | bool first_; |
| 3614 | }; |
| 3615 | |
| 3616 | void DotPrinter::PrintAttributes(RegExpNode* that) { |
| 3617 | OS::PrintErr( |
| 3618 | " a%p [shape=Mrecord, color=grey, fontcolor=grey, " |
| 3619 | "margin=0.1, fontsize=10, label=\"{", |
| 3620 | that); |
| 3621 | AttributePrinter printer; |
| 3622 | NodeInfo* info = that->info(); |
| 3623 | printer.PrintBit("NI", info->follows_newline_interest); |
| 3624 | printer.PrintBit("WI", info->follows_word_interest); |
| 3625 | printer.PrintBit("SI", info->follows_start_interest); |
| 3626 | BlockLabel* label = that->label(); |
| 3627 | if (label->is_bound()) printer.PrintPositive("@", label->pos()); |
| 3628 | OS::PrintErr( |
| 3629 | "}\"];\n" |
| 3630 | " a%p -> n%p [style=dashed, color=grey, arrowhead=none];\n", |
| 3631 | that, that); |
| 3632 | } |
| 3633 | |
| 3634 | void DotPrinter::VisitChoice(ChoiceNode* that) { |
| 3635 | OS::PrintErr(" n%p [shape=Mrecord, label=\"?\"];\n", that); |
| 3636 | for (intptr_t i = 0; i < that->alternatives()->length(); i++) { |
| 3637 | GuardedAlternative alt = that->alternatives()->At(i); |
| 3638 | OS::PrintErr(" n%p -> n%p", that, alt.node()); |
| 3639 | } |
| 3640 | for (intptr_t i = 0; i < that->alternatives()->length(); i++) { |
| 3641 | GuardedAlternative alt = that->alternatives()->At(i); |
| 3642 | alt.node()->Accept(this); |
| 3643 | } |
| 3644 | } |
| 3645 | |
| 3646 | void DotPrinter::VisitText(TextNode* that) { |
| 3647 | OS::PrintErr(" n%p [label=\"", that); |
| 3648 | for (intptr_t i = 0; i < that->elements()->length(); i++) { |
| 3649 | if (i > 0) OS::PrintErr(" "); |
| 3650 | TextElement elm = that->elements()->At(i); |
| 3651 | switch (elm.text_type()) { |
| 3652 | case TextElement::ATOM: { |
| 3653 | ZoneGrowableArray<uint16_t>* data = elm.atom()->data(); |
| 3654 | for (intptr_t i = 0; i < data->length(); i++) { |
| 3655 | OS::PrintErr("%c", static_cast<char>(data->At(i))); |
| 3656 | } |
| 3657 | break; |
| 3658 | } |
| 3659 | case TextElement::CHAR_CLASS: { |
| 3660 | RegExpCharacterClass* node = elm.char_class(); |
| 3661 | OS::PrintErr("["); |
| 3662 | if (node->is_negated()) OS::PrintErr("^"); |
| 3663 | for (intptr_t j = 0; j < node->ranges()->length(); j++) { |
| 3664 | CharacterRange range = node->ranges()->At(j); |
| 3665 | PrintUtf16(range.from()); |
| 3666 | OS::PrintErr("-"); |
| 3667 | PrintUtf16(range.to()); |
| 3668 | } |
| 3669 | OS::PrintErr("]"); |
| 3670 | break; |
| 3671 | } |
| 3672 | default: |
| 3673 | UNREACHABLE(); |
| 3674 | } |
| 3675 | } |
| 3676 | OS::PrintErr("\", shape=box, peripheries=2];\n"); |
| 3677 | PrintAttributes(that); |
| 3678 | OS::PrintErr(" n%p -> n%p;\n", that, that->on_success()); |
| 3679 | Visit(that->on_success()); |
| 3680 | } |
| 3681 | |
| 3682 | void DotPrinter::VisitBackReference(BackReferenceNode* that) { |
| 3683 | OS::PrintErr(" n%p [label=\"$%"Pd "..$%"Pd "\", shape=doubleoctagon];\n", |
| 3684 | that, that->start_register(), that->end_register()); |
| 3685 | PrintAttributes(that); |
| 3686 | OS::PrintErr(" n%p -> n%p;\n", that, that->on_success()); |
| 3687 | Visit(that->on_success()); |
| 3688 | } |
| 3689 | |
| 3690 | void DotPrinter::VisitEnd(EndNode* that) { |
| 3691 | OS::PrintErr(" n%p [style=bold, shape=point];\n", that); |
| 3692 | PrintAttributes(that); |
| 3693 | } |
| 3694 | |
| 3695 | void DotPrinter::VisitAssertion(AssertionNode* that) { |
| 3696 | OS::PrintErr(" n%p [", that); |
| 3697 | switch (that->assertion_type()) { |
| 3698 | case AssertionNode::AT_END: |
| 3699 | OS::PrintErr("label=\"$\", shape=septagon"); |
| 3700 | break; |
| 3701 | case AssertionNode::AT_START: |
| 3702 | OS::PrintErr("label=\"^\", shape=septagon"); |
| 3703 | break; |
| 3704 | case AssertionNode::AT_BOUNDARY: |
| 3705 | OS::PrintErr("label=\"\\b\", shape=septagon"); |
| 3706 | break; |
| 3707 | case AssertionNode::AT_NON_BOUNDARY: |
| 3708 | OS::PrintErr("label=\"\\B\", shape=septagon"); |
| 3709 | break; |
| 3710 | case AssertionNode::AFTER_NEWLINE: |
| 3711 | OS::PrintErr("label=\"(?<=\\n)\", shape=septagon"); |
| 3712 | break; |
| 3713 | } |
| 3714 | OS::PrintErr("];\n"); |
| 3715 | PrintAttributes(that); |
| 3716 | RegExpNode* successor = that->on_success(); |
| 3717 | OS::PrintErr(" n%p -> n%p;\n", that, successor); |
| 3718 | Visit(successor); |
| 3719 | } |
| 3720 | |
| 3721 | void DotPrinter::VisitAction(ActionNode* that) { |
| 3722 | OS::PrintErr(" n%p [", that); |
| 3723 | switch (that->action_type_) { |
| 3724 | case ActionNode::SET_REGISTER: |
| 3725 | OS::PrintErr("label=\"$%"Pd ":=%"Pd "\", shape=octagon", |
| 3726 | that->data_.u_store_register.reg, |
| 3727 | that->data_.u_store_register.value); |
| 3728 | break; |
| 3729 | case ActionNode::INCREMENT_REGISTER: |
| 3730 | OS::PrintErr("label=\"$%"Pd "++\", shape=octagon", |
| 3731 | that->data_.u_increment_register.reg); |
| 3732 | break; |
| 3733 | case ActionNode::STORE_POSITION: |
| 3734 | OS::PrintErr("label=\"$%"Pd ":=$pos\", shape=octagon", |
| 3735 | that->data_.u_position_register.reg); |
| 3736 | break; |
| 3737 | case ActionNode::BEGIN_SUBMATCH: |
| 3738 | OS::PrintErr("label=\"$%"Pd ":=$pos,begin\", shape=septagon", |
| 3739 | that->data_.u_submatch.current_position_register); |
| 3740 | break; |
| 3741 | case ActionNode::POSITIVE_SUBMATCH_SUCCESS: |
| 3742 | OS::PrintErr("label=\"escape\", shape=septagon"); |
| 3743 | break; |
| 3744 | case ActionNode::EMPTY_MATCH_CHECK: |
| 3745 | OS::PrintErr("label=\"$%"Pd "=$pos?,$%"Pd "<%"Pd "?\", shape=septagon", |
| 3746 | that->data_.u_empty_match_check.start_register, |
| 3747 | that->data_.u_empty_match_check.repetition_register, |
| 3748 | that->data_.u_empty_match_check.repetition_limit); |
| 3749 | break; |
| 3750 | case ActionNode::CLEAR_CAPTURES: { |
| 3751 | OS::PrintErr("label=\"clear $%"Pd " to $%"Pd "\", shape=septagon", |
| 3752 | that->data_.u_clear_captures.range_from, |
| 3753 | that->data_.u_clear_captures.range_to); |
| 3754 | break; |
| 3755 | } |
| 3756 | } |
| 3757 | OS::PrintErr("];\n"); |
| 3758 | PrintAttributes(that); |
| 3759 | RegExpNode* successor = that->on_success(); |
| 3760 | OS::PrintErr(" n%p -> n%p;\n", that, successor); |
| 3761 | Visit(successor); |
| 3762 | } |
| 3763 | |
| 3764 | void RegExpEngine::DotPrint(const char* label, |
| 3765 | RegExpNode* node, |
| 3766 | bool ignore_case) { |
| 3767 | DotPrinter printer(ignore_case); |
| 3768 | printer.PrintNode(label, node); |
| 3769 | } |
| 3770 | |
| 3771 | #endif // DEBUG |
| 3772 | |
| 3773 | // ------------------------------------------------------------------- |
| 3774 | // Tree to graph conversion |
| 3775 | |
| 3776 | // The zone in which we allocate graph nodes. |
| 3777 | #define OZ (on_success->zone()) |
| 3778 | |
| 3779 | RegExpNode* RegExpAtom::ToNode(RegExpCompiler* compiler, |
| 3780 | RegExpNode* on_success) { |
| 3781 | ZoneGrowableArray<TextElement>* elms = |
| 3782 | new (OZ) ZoneGrowableArray<TextElement>(1); |
| 3783 | elms->Add(TextElement::Atom(this)); |
| 3784 | return new (OZ) TextNode(elms, compiler->read_backward(), on_success); |
| 3785 | } |
| 3786 | |
| 3787 | RegExpNode* RegExpText::ToNode(RegExpCompiler* compiler, |
| 3788 | RegExpNode* on_success) { |
| 3789 | ZoneGrowableArray<TextElement>* elms = |
| 3790 | new (OZ) ZoneGrowableArray<TextElement>(1); |
| 3791 | for (intptr_t i = 0; i < elements()->length(); i++) { |
| 3792 | elms->Add(elements()->At(i)); |
| 3793 | } |
| 3794 | return new (OZ) TextNode(elms, compiler->read_backward(), on_success); |
| 3795 | } |
| 3796 | |
| 3797 | static bool CompareInverseRanges(ZoneGrowableArray<CharacterRange>* ranges, |
| 3798 | const int32_t* special_class, |
| 3799 | intptr_t length) { |
| 3800 | length--; // Remove final kRangeEndMarker. |
| 3801 | ASSERT(special_class[length] == kRangeEndMarker); |
| 3802 | ASSERT(ranges->length() != 0); |
| 3803 | ASSERT(length != 0); |
| 3804 | ASSERT(special_class[0] != 0); |
| 3805 | if (ranges->length() != (length >> 1) + 1) { |
| 3806 | return false; |
| 3807 | } |
| 3808 | CharacterRange range = ranges->At(0); |
| 3809 | if (range.from() != 0) { |
| 3810 | return false; |
| 3811 | } |
| 3812 | for (intptr_t i = 0; i < length; i += 2) { |
| 3813 | if (special_class[i] != (range.to() + 1)) { |
| 3814 | return false; |
| 3815 | } |
| 3816 | range = ranges->At((i >> 1) + 1); |
| 3817 | if (special_class[i + 1] != range.from()) { |
| 3818 | return false; |
| 3819 | } |
| 3820 | } |
| 3821 | if (range.to() != Utf::kMaxCodePoint) { |
| 3822 | return false; |
| 3823 | } |
| 3824 | return true; |
| 3825 | } |
| 3826 | |
| 3827 | static bool CompareRanges(ZoneGrowableArray<CharacterRange>* ranges, |
| 3828 | const int32_t* special_class, |
| 3829 | intptr_t length) { |
| 3830 | length--; // Remove final kRangeEndMarker. |
| 3831 | ASSERT(special_class[length] == kRangeEndMarker); |
| 3832 | if (ranges->length() * 2 != length) { |
| 3833 | return false; |
| 3834 | } |
| 3835 | for (intptr_t i = 0; i < length; i += 2) { |
| 3836 | CharacterRange range = ranges->At(i >> 1); |
| 3837 | if (range.from() != special_class[i] || |
| 3838 | range.to() != special_class[i + 1] - 1) { |
| 3839 | return false; |
| 3840 | } |
| 3841 | } |
| 3842 | return true; |
| 3843 | } |
| 3844 | |
| 3845 | bool RegExpCharacterClass::is_standard() { |
| 3846 | // TODO(lrn): Remove need for this function, by not throwing away information |
| 3847 | // along the way. |
| 3848 | if (is_negated()) { |
| 3849 | return false; |
| 3850 | } |
| 3851 | if (set_.is_standard()) { |
| 3852 | return true; |
| 3853 | } |
| 3854 | if (CompareRanges(set_.ranges(), kSpaceRanges, kSpaceRangeCount)) { |
| 3855 | set_.set_standard_set_type('s'); |
| 3856 | return true; |
| 3857 | } |
| 3858 | if (CompareInverseRanges(set_.ranges(), kSpaceRanges, kSpaceRangeCount)) { |
| 3859 | set_.set_standard_set_type('S'); |
| 3860 | return true; |
| 3861 | } |
| 3862 | if (CompareInverseRanges(set_.ranges(), kLineTerminatorRanges, |
| 3863 | kLineTerminatorRangeCount)) { |
| 3864 | set_.set_standard_set_type('.'); |
| 3865 | return true; |
| 3866 | } |
| 3867 | if (CompareRanges(set_.ranges(), kLineTerminatorRanges, |
| 3868 | kLineTerminatorRangeCount)) { |
| 3869 | set_.set_standard_set_type('n'); |
| 3870 | return true; |
| 3871 | } |
| 3872 | if (CompareRanges(set_.ranges(), kWordRanges, kWordRangeCount)) { |
| 3873 | set_.set_standard_set_type('w'); |
| 3874 | return true; |
| 3875 | } |
| 3876 | if (CompareInverseRanges(set_.ranges(), kWordRanges, kWordRangeCount)) { |
| 3877 | set_.set_standard_set_type('W'); |
| 3878 | return true; |
| 3879 | } |
| 3880 | return false; |
| 3881 | } |
| 3882 | |
| 3883 | UnicodeRangeSplitter::UnicodeRangeSplitter( |
| 3884 | Zone* zone, |
| 3885 | ZoneGrowableArray<CharacterRange>* base) |
| 3886 | : zone_(zone), |
| 3887 | table_(zone), |
| 3888 | bmp_(nullptr), |
| 3889 | lead_surrogates_(nullptr), |
| 3890 | trail_surrogates_(nullptr), |
| 3891 | non_bmp_(nullptr) { |
| 3892 | // The unicode range splitter categorizes given character ranges into: |
| 3893 | // - Code points from the BMP representable by one code unit. |
| 3894 | // - Code points outside the BMP that need to be split into surrogate pairs. |
| 3895 | // - Lone lead surrogates. |
| 3896 | // - Lone trail surrogates. |
| 3897 | // Lone surrogates are valid code points, even though no actual characters. |
| 3898 | // They require special matching to make sure we do not split surrogate pairs. |
| 3899 | // We use the dispatch table to accomplish this. The base range is split up |
| 3900 | // by the table by the overlay ranges, and the Call callback is used to |
| 3901 | // filter and collect ranges for each category. |
| 3902 | for (intptr_t i = 0; i < base->length(); i++) { |
| 3903 | table_.AddRange(base->At(i), kBase, zone_); |
| 3904 | } |
| 3905 | // Add overlay ranges. |
| 3906 | table_.AddRange(CharacterRange::Range(0, Utf16::kLeadSurrogateStart - 1), |
| 3907 | kBmpCodePoints, zone_); |
| 3908 | table_.AddRange(CharacterRange::Range(Utf16::kLeadSurrogateStart, |
| 3909 | Utf16::kLeadSurrogateEnd), |
| 3910 | kLeadSurrogates, zone_); |
| 3911 | table_.AddRange(CharacterRange::Range(Utf16::kTrailSurrogateStart, |
| 3912 | Utf16::kTrailSurrogateEnd), |
| 3913 | kTrailSurrogates, zone_); |
| 3914 | table_.AddRange( |
| 3915 | CharacterRange::Range(Utf16::kTrailSurrogateEnd + 1, Utf16::kMaxCodeUnit), |
| 3916 | kBmpCodePoints, zone_); |
| 3917 | table_.AddRange( |
| 3918 | CharacterRange::Range(Utf16::kMaxCodeUnit + 1, Utf::kMaxCodePoint), |
| 3919 | kNonBmpCodePoints, zone_); |
| 3920 | table_.ForEach(this); |
| 3921 | } |
| 3922 | |
| 3923 | void UnicodeRangeSplitter::Call(uint32_t from, ChoiceTable::Entry entry) { |
| 3924 | OutSet* outset = entry.out_set(); |
| 3925 | if (!outset->Get(kBase)) return; |
| 3926 | ZoneGrowableArray<CharacterRange>** target = nullptr; |
| 3927 | if (outset->Get(kBmpCodePoints)) { |
| 3928 | target = &bmp_; |
| 3929 | } else if (outset->Get(kLeadSurrogates)) { |
| 3930 | target = &lead_surrogates_; |
| 3931 | } else if (outset->Get(kTrailSurrogates)) { |
| 3932 | target = &trail_surrogates_; |
| 3933 | } else { |
| 3934 | ASSERT(outset->Get(kNonBmpCodePoints)); |
| 3935 | target = &non_bmp_; |
| 3936 | } |
| 3937 | if (*target == nullptr) { |
| 3938 | *target = new (zone_) ZoneGrowableArray<CharacterRange>(2); |
| 3939 | } |
| 3940 | (*target)->Add(CharacterRange::Range(entry.from(), entry.to())); |
| 3941 | } |
| 3942 | |
| 3943 | void AddBmpCharacters(RegExpCompiler* compiler, |
| 3944 | ChoiceNode* result, |
| 3945 | RegExpNode* on_success, |
| 3946 | UnicodeRangeSplitter* splitter) { |
| 3947 | ZoneGrowableArray<CharacterRange>* bmp = splitter->bmp(); |
| 3948 | if (bmp == nullptr) return; |
| 3949 | result->AddAlternative(GuardedAlternative(TextNode::CreateForCharacterRanges( |
| 3950 | bmp, compiler->read_backward(), on_success, RegExpFlags()))); |
| 3951 | } |
| 3952 | |
| 3953 | void AddNonBmpSurrogatePairs(RegExpCompiler* compiler, |
| 3954 | ChoiceNode* result, |
| 3955 | RegExpNode* on_success, |
| 3956 | UnicodeRangeSplitter* splitter) { |
| 3957 | ZoneGrowableArray<CharacterRange>* non_bmp = splitter->non_bmp(); |
| 3958 | if (non_bmp == nullptr) return; |
| 3959 | ASSERT(!compiler->one_byte()); |
| 3960 | CharacterRange::Canonicalize(non_bmp); |
| 3961 | for (int i = 0; i < non_bmp->length(); i++) { |
| 3962 | // Match surrogate pair. |
| 3963 | // E.g. [\u10005-\u11005] becomes |
| 3964 | // \ud800[\udc05-\udfff]| |
| 3965 | // [\ud801-\ud803][\udc00-\udfff]| |
| 3966 | // \ud804[\udc00-\udc05] |
| 3967 | uint32_t from = non_bmp->At(i).from(); |
| 3968 | uint32_t to = non_bmp->At(i).to(); |
| 3969 | uint16_t from_points[2]; |
| 3970 | Utf16::Encode(from, from_points); |
| 3971 | uint16_t to_points[2]; |
| 3972 | Utf16::Encode(to, to_points); |
| 3973 | if (from_points[0] == to_points[0]) { |
| 3974 | // The lead surrogate is the same. |
| 3975 | result->AddAlternative( |
| 3976 | GuardedAlternative(TextNode::CreateForSurrogatePair( |
| 3977 | CharacterRange::Singleton(from_points[0]), |
| 3978 | CharacterRange::Range(from_points[1], to_points[1]), |
| 3979 | compiler->read_backward(), on_success, RegExpFlags()))); |
| 3980 | } else { |
| 3981 | if (from_points[1] != Utf16::kTrailSurrogateStart) { |
| 3982 | // Add [from_l][from_t-\udfff] |
| 3983 | result->AddAlternative( |
| 3984 | GuardedAlternative(TextNode::CreateForSurrogatePair( |
| 3985 | CharacterRange::Singleton(from_points[0]), |
| 3986 | CharacterRange::Range(from_points[1], |
| 3987 | Utf16::kTrailSurrogateEnd), |
| 3988 | compiler->read_backward(), on_success, RegExpFlags()))); |
| 3989 | from_points[0]++; |
| 3990 | } |
| 3991 | if (to_points[1] != Utf16::kTrailSurrogateEnd) { |
| 3992 | // Add [to_l][\udc00-to_t] |
| 3993 | result->AddAlternative( |
| 3994 | GuardedAlternative(TextNode::CreateForSurrogatePair( |
| 3995 | CharacterRange::Singleton(to_points[0]), |
| 3996 | CharacterRange::Range(Utf16::kTrailSurrogateStart, |
| 3997 | to_points[1]), |
| 3998 | compiler->read_backward(), on_success, RegExpFlags()))); |
| 3999 | to_points[0]--; |
| 4000 | } |
| 4001 | if (from_points[0] <= to_points[0]) { |
| 4002 | // Add [from_l-to_l][\udc00-\udfff] |
| 4003 | result->AddAlternative( |
| 4004 | GuardedAlternative(TextNode::CreateForSurrogatePair( |
| 4005 | CharacterRange::Range(from_points[0], to_points[0]), |
| 4006 | CharacterRange::Range(Utf16::kTrailSurrogateStart, |
| 4007 | Utf16::kTrailSurrogateEnd), |
| 4008 | compiler->read_backward(), on_success, RegExpFlags()))); |
| 4009 | } |
| 4010 | } |
| 4011 | } |
| 4012 | } |
| 4013 | |
| 4014 | RegExpNode* NegativeLookaroundAgainstReadDirectionAndMatch( |
| 4015 | RegExpCompiler* compiler, |
| 4016 | ZoneGrowableArray<CharacterRange>* lookbehind, |
| 4017 | ZoneGrowableArray<CharacterRange>* match, |
| 4018 | RegExpNode* on_success, |
| 4019 | bool read_backward, |
| 4020 | RegExpFlags flags) { |
| 4021 | RegExpNode* match_node = TextNode::CreateForCharacterRanges( |
| 4022 | match, read_backward, on_success, flags); |
| 4023 | int stack_register = compiler->UnicodeLookaroundStackRegister(); |
| 4024 | int position_register = compiler->UnicodeLookaroundPositionRegister(); |
| 4025 | RegExpLookaround::Builder lookaround(false, match_node, stack_register, |
| 4026 | position_register); |
| 4027 | RegExpNode* negative_match = TextNode::CreateForCharacterRanges( |
| 4028 | lookbehind, !read_backward, lookaround.on_match_success(), flags); |
| 4029 | return lookaround.ForMatch(negative_match); |
| 4030 | } |
| 4031 | |
| 4032 | RegExpNode* MatchAndNegativeLookaroundInReadDirection( |
| 4033 | RegExpCompiler* compiler, |
| 4034 | ZoneGrowableArray<CharacterRange>* match, |
| 4035 | ZoneGrowableArray<CharacterRange>* lookahead, |
| 4036 | RegExpNode* on_success, |
| 4037 | bool read_backward, |
| 4038 | RegExpFlags flags) { |
| 4039 | int stack_register = compiler->UnicodeLookaroundStackRegister(); |
| 4040 | int position_register = compiler->UnicodeLookaroundPositionRegister(); |
| 4041 | RegExpLookaround::Builder lookaround(false, on_success, stack_register, |
| 4042 | position_register); |
| 4043 | RegExpNode* negative_match = TextNode::CreateForCharacterRanges( |
| 4044 | lookahead, read_backward, lookaround.on_match_success(), flags); |
| 4045 | return TextNode::CreateForCharacterRanges( |
| 4046 | match, read_backward, lookaround.ForMatch(negative_match), flags); |
| 4047 | } |
| 4048 | |
| 4049 | void AddLoneLeadSurrogates(RegExpCompiler* compiler, |
| 4050 | ChoiceNode* result, |
| 4051 | RegExpNode* on_success, |
| 4052 | UnicodeRangeSplitter* splitter) { |
| 4053 | auto lead_surrogates = splitter->lead_surrogates(); |
| 4054 | if (lead_surrogates == nullptr) return; |
| 4055 | // E.g. \ud801 becomes \ud801(?![\udc00-\udfff]). |
| 4056 | auto trail_surrogates = CharacterRange::List( |
| 4057 | on_success->zone(), CharacterRange::Range(Utf16::kTrailSurrogateStart, |
| 4058 | Utf16::kTrailSurrogateEnd)); |
| 4059 | |
| 4060 | RegExpNode* match; |
| 4061 | if (compiler->read_backward()) { |
| 4062 | // Reading backward. Assert that reading forward, there is no trail |
| 4063 | // surrogate, and then backward match the lead surrogate. |
| 4064 | match = NegativeLookaroundAgainstReadDirectionAndMatch( |
| 4065 | compiler, trail_surrogates, lead_surrogates, on_success, true, |
| 4066 | RegExpFlags()); |
| 4067 | } else { |
| 4068 | // Reading forward. Forward match the lead surrogate and assert that |
| 4069 | // no trail surrogate follows. |
| 4070 | match = MatchAndNegativeLookaroundInReadDirection( |
| 4071 | compiler, lead_surrogates, trail_surrogates, on_success, false, |
| 4072 | RegExpFlags()); |
| 4073 | } |
| 4074 | result->AddAlternative(GuardedAlternative(match)); |
| 4075 | } |
| 4076 | |
| 4077 | void AddLoneTrailSurrogates(RegExpCompiler* compiler, |
| 4078 | ChoiceNode* result, |
| 4079 | RegExpNode* on_success, |
| 4080 | UnicodeRangeSplitter* splitter) { |
| 4081 | auto trail_surrogates = splitter->trail_surrogates(); |
| 4082 | if (trail_surrogates == nullptr) return; |
| 4083 | // E.g. \udc01 becomes (?<![\ud800-\udbff])\udc01 |
| 4084 | auto lead_surrogates = CharacterRange::List( |
| 4085 | on_success->zone(), CharacterRange::Range(Utf16::kLeadSurrogateStart, |
| 4086 | Utf16::kLeadSurrogateEnd)); |
| 4087 | |
| 4088 | RegExpNode* match; |
| 4089 | if (compiler->read_backward()) { |
| 4090 | // Reading backward. Backward match the trail surrogate and assert that no |
| 4091 | // lead surrogate precedes it. |
| 4092 | match = MatchAndNegativeLookaroundInReadDirection( |
| 4093 | compiler, trail_surrogates, lead_surrogates, on_success, true, |
| 4094 | RegExpFlags()); |
| 4095 | } else { |
| 4096 | // Reading forward. Assert that reading backward, there is no lead |
| 4097 | // surrogate, and then forward match the trail surrogate. |
| 4098 | match = NegativeLookaroundAgainstReadDirectionAndMatch( |
| 4099 | compiler, lead_surrogates, trail_surrogates, on_success, false, |
| 4100 | RegExpFlags()); |
| 4101 | } |
| 4102 | result->AddAlternative(GuardedAlternative(match)); |
| 4103 | } |
| 4104 | |
| 4105 | RegExpNode* UnanchoredAdvance(RegExpCompiler* compiler, |
| 4106 | RegExpNode* on_success) { |
| 4107 | // This implements ES2015 21.2.5.2.3, AdvanceStringIndex. |
| 4108 | ASSERT(!compiler->read_backward()); |
| 4109 | // Advance any character. If the character happens to be a lead surrogate and |
| 4110 | // we advanced into the middle of a surrogate pair, it will work out, as |
| 4111 | // nothing will match from there. We will have to advance again, consuming |
| 4112 | // the associated trail surrogate. |
| 4113 | auto range = CharacterRange::List( |
| 4114 | on_success->zone(), CharacterRange::Range(0, Utf16::kMaxCodeUnit)); |
| 4115 | return TextNode::CreateForCharacterRanges(range, false, on_success, |
| 4116 | RegExpFlags()); |
| 4117 | } |
| 4118 | |
| 4119 | void AddUnicodeCaseEquivalents(ZoneGrowableArray<CharacterRange>* ranges) { |
| 4120 | ASSERT(CharacterRange::IsCanonical(ranges)); |
| 4121 | |
| 4122 | // Micro-optimization to avoid passing large ranges to UnicodeSet::closeOver. |
| 4123 | // See also https://crbug.com/v8/6727. |
| 4124 | // TODO(sstrickl): This only covers the special case of the {0,0x10FFFF} |
| 4125 | // range, which we use frequently internally. But large ranges can also easily |
| 4126 | // be created by the user. We might want to have a more general caching |
| 4127 | // mechanism for such ranges. |
| 4128 | if (ranges->length() == 1 && ranges->At(0).IsEverything(Utf::kMaxCodePoint)) { |
| 4129 | return; |
| 4130 | } |
| 4131 | |
| 4132 | icu::UnicodeSet set; |
| 4133 | for (int i = 0; i < ranges->length(); i++) { |
| 4134 | set.add(ranges->At(i).from(), ranges->At(i).to()); |
| 4135 | } |
| 4136 | ranges->Clear(); |
| 4137 | set.closeOver(USET_CASE_INSENSITIVE); |
| 4138 | // Full case mapping map single characters to multiple characters. |
| 4139 | // Those are represented as strings in the set. Remove them so that |
| 4140 | // we end up with only simple and common case mappings. |
| 4141 | set.removeAllStrings(); |
| 4142 | for (int i = 0; i < set.getRangeCount(); i++) { |
| 4143 | ranges->Add( |
| 4144 | CharacterRange::Range(set.getRangeStart(i), set.getRangeEnd(i))); |
| 4145 | } |
| 4146 | // No errors and everything we collected have been ranges. |
| 4147 | CharacterRange::Canonicalize(ranges); |
| 4148 | } |
| 4149 | |
| 4150 | RegExpNode* RegExpCharacterClass::ToNode(RegExpCompiler* compiler, |
| 4151 | RegExpNode* on_success) { |
| 4152 | set_.Canonicalize(); |
| 4153 | ZoneGrowableArray<CharacterRange>* ranges = this->ranges(); |
| 4154 | if (flags_.NeedsUnicodeCaseEquivalents()) { |
| 4155 | AddUnicodeCaseEquivalents(ranges); |
| 4156 | } |
| 4157 | if (flags_.IsUnicode() && !compiler->one_byte() && |
| 4158 | !contains_split_surrogate()) { |
| 4159 | if (is_negated()) { |
| 4160 | ZoneGrowableArray<CharacterRange>* negated = |
| 4161 | new ZoneGrowableArray<CharacterRange>(2); |
| 4162 | CharacterRange::Negate(ranges, negated); |
| 4163 | ranges = negated; |
| 4164 | } |
| 4165 | if (ranges->length() == 0) { |
| 4166 | RegExpCharacterClass* fail = |
| 4167 | new RegExpCharacterClass(ranges, RegExpFlags()); |
| 4168 | return new TextNode(fail, compiler->read_backward(), on_success); |
| 4169 | } |
| 4170 | if (standard_type() == '*') { |
| 4171 | return UnanchoredAdvance(compiler, on_success); |
| 4172 | } else { |
| 4173 | ChoiceNode* result = new (OZ) ChoiceNode(2, OZ); |
| 4174 | UnicodeRangeSplitter splitter(OZ, ranges); |
| 4175 | AddBmpCharacters(compiler, result, on_success, &splitter); |
| 4176 | AddNonBmpSurrogatePairs(compiler, result, on_success, &splitter); |
| 4177 | AddLoneLeadSurrogates(compiler, result, on_success, &splitter); |
| 4178 | AddLoneTrailSurrogates(compiler, result, on_success, &splitter); |
| 4179 | return result; |
| 4180 | } |
| 4181 | } else { |
| 4182 | return new TextNode(this, compiler->read_backward(), on_success); |
| 4183 | } |
| 4184 | return new (OZ) TextNode(this, compiler->read_backward(), on_success); |
| 4185 | } |
| 4186 | |
| 4187 | RegExpNode* RegExpDisjunction::ToNode(RegExpCompiler* compiler, |
| 4188 | RegExpNode* on_success) { |
| 4189 | ZoneGrowableArray<RegExpTree*>* alternatives = this->alternatives(); |
| 4190 | intptr_t length = alternatives->length(); |
| 4191 | ChoiceNode* result = new (OZ) ChoiceNode(length, OZ); |
| 4192 | for (intptr_t i = 0; i < length; i++) { |
| 4193 | GuardedAlternative alternative( |
| 4194 | alternatives->At(i)->ToNode(compiler, on_success)); |
| 4195 | result->AddAlternative(alternative); |
| 4196 | } |
| 4197 | return result; |
| 4198 | } |
| 4199 | |
| 4200 | RegExpNode* RegExpQuantifier::ToNode(RegExpCompiler* compiler, |
| 4201 | RegExpNode* on_success) { |
| 4202 | return ToNode(min(), max(), is_greedy(), body(), compiler, on_success); |
| 4203 | } |
| 4204 | |
| 4205 | // Scoped object to keep track of how much we unroll quantifier loops in the |
| 4206 | // regexp graph generator. |
| 4207 | class RegExpExpansionLimiter : public ValueObject { |
| 4208 | public: |
| 4209 | static const intptr_t kMaxExpansionFactor = 6; |
| 4210 | RegExpExpansionLimiter(RegExpCompiler* compiler, intptr_t factor) |
| 4211 | : compiler_(compiler), |
| 4212 | saved_expansion_factor_(compiler->current_expansion_factor()), |
| 4213 | ok_to_expand_(saved_expansion_factor_ <= kMaxExpansionFactor) { |
| 4214 | ASSERT(factor > 0); |
| 4215 | if (ok_to_expand_) { |
| 4216 | if (factor > kMaxExpansionFactor) { |
| 4217 | // Avoid integer overflow of the current expansion factor. |
| 4218 | ok_to_expand_ = false; |
| 4219 | compiler->set_current_expansion_factor(kMaxExpansionFactor + 1); |
| 4220 | } else { |
| 4221 | intptr_t new_factor = saved_expansion_factor_ * factor; |
| 4222 | ok_to_expand_ = (new_factor <= kMaxExpansionFactor); |
| 4223 | compiler->set_current_expansion_factor(new_factor); |
| 4224 | } |
| 4225 | } |
| 4226 | } |
| 4227 | |
| 4228 | ~RegExpExpansionLimiter() { |
| 4229 | compiler_->set_current_expansion_factor(saved_expansion_factor_); |
| 4230 | } |
| 4231 | |
| 4232 | bool ok_to_expand() { return ok_to_expand_; } |
| 4233 | |
| 4234 | private: |
| 4235 | RegExpCompiler* compiler_; |
| 4236 | intptr_t saved_expansion_factor_; |
| 4237 | bool ok_to_expand_; |
| 4238 | |
| 4239 | DISALLOW_IMPLICIT_CONSTRUCTORS(RegExpExpansionLimiter); |
| 4240 | }; |
| 4241 | |
| 4242 | RegExpNode* RegExpQuantifier::ToNode(intptr_t min, |
| 4243 | intptr_t max, |
| 4244 | bool is_greedy, |
| 4245 | RegExpTree* body, |
| 4246 | RegExpCompiler* compiler, |
| 4247 | RegExpNode* on_success, |
| 4248 | bool not_at_start) { |
| 4249 | // x{f, t} becomes this: |
| 4250 | // |
| 4251 | // (r++)<-. |
| 4252 | // | ` |
| 4253 | // | (x) |
| 4254 | // v ^ |
| 4255 | // (r=0)-->(?)---/ [if r < t] |
| 4256 | // | |
| 4257 | // [if r >= f] \----> ... |
| 4258 | // |
| 4259 | |
| 4260 | // 15.10.2.5 RepeatMatcher algorithm. |
| 4261 | // The parser has already eliminated the case where max is 0. In the case |
| 4262 | // where max_match is zero the parser has removed the quantifier if min was |
| 4263 | // > 0 and removed the atom if min was 0. See AddQuantifierToAtom. |
| 4264 | |
| 4265 | // If we know that we cannot match zero length then things are a little |
| 4266 | // simpler since we don't need to make the special zero length match check |
| 4267 | // from step 2.1. If the min and max are small we can unroll a little in |
| 4268 | // this case. |
| 4269 | // Unroll (foo)+ and (foo){3,} |
| 4270 | static const intptr_t kMaxUnrolledMinMatches = 3; |
| 4271 | // Unroll (foo)? and (foo){x,3} |
| 4272 | static const intptr_t kMaxUnrolledMaxMatches = 3; |
| 4273 | if (max == 0) return on_success; // This can happen due to recursion. |
| 4274 | bool body_can_be_empty = (body->min_match() == 0); |
| 4275 | intptr_t body_start_reg = RegExpCompiler::kNoRegister; |
| 4276 | Interval capture_registers = body->CaptureRegisters(); |
| 4277 | bool needs_capture_clearing = !capture_registers.is_empty(); |
| 4278 | Zone* zone = compiler->zone(); |
| 4279 | |
| 4280 | if (body_can_be_empty) { |
| 4281 | body_start_reg = compiler->AllocateRegister(); |
| 4282 | } else if (kRegexpOptimization && !needs_capture_clearing) { |
| 4283 | // Only unroll if there are no captures and the body can't be |
| 4284 | // empty. |
| 4285 | { |
| 4286 | RegExpExpansionLimiter limiter(compiler, min + ((max != min) ? 1 : 0)); |
| 4287 | if (min > 0 && min <= kMaxUnrolledMinMatches && limiter.ok_to_expand()) { |
| 4288 | intptr_t new_max = (max == kInfinity) ? max : max - min; |
| 4289 | // Recurse once to get the loop or optional matches after the fixed |
| 4290 | // ones. |
| 4291 | RegExpNode* answer = |
| 4292 | ToNode(0, new_max, is_greedy, body, compiler, on_success, true); |
| 4293 | // Unroll the forced matches from 0 to min. This can cause chains of |
| 4294 | // TextNodes (which the parser does not generate). These should be |
| 4295 | // combined if it turns out they hinder good code generation. |
| 4296 | for (intptr_t i = 0; i < min; i++) { |
| 4297 | answer = body->ToNode(compiler, answer); |
| 4298 | } |
| 4299 | return answer; |
| 4300 | } |
| 4301 | } |
| 4302 | if (max <= kMaxUnrolledMaxMatches && min == 0) { |
| 4303 | ASSERT(max > 0); // Due to the 'if' above. |
| 4304 | RegExpExpansionLimiter limiter(compiler, max); |
| 4305 | if (limiter.ok_to_expand()) { |
| 4306 | // Unroll the optional matches up to max. |
| 4307 | RegExpNode* answer = on_success; |
| 4308 | for (intptr_t i = 0; i < max; i++) { |
| 4309 | ChoiceNode* alternation = new (zone) ChoiceNode(2, zone); |
| 4310 | if (is_greedy) { |
| 4311 | alternation->AddAlternative( |
| 4312 | GuardedAlternative(body->ToNode(compiler, answer))); |
| 4313 | alternation->AddAlternative(GuardedAlternative(on_success)); |
| 4314 | } else { |
| 4315 | alternation->AddAlternative(GuardedAlternative(on_success)); |
| 4316 | alternation->AddAlternative( |
| 4317 | GuardedAlternative(body->ToNode(compiler, answer))); |
| 4318 | } |
| 4319 | answer = alternation; |
| 4320 | if (not_at_start && !compiler->read_backward()) { |
| 4321 | alternation->set_not_at_start(); |
| 4322 | } |
| 4323 | } |
| 4324 | return answer; |
| 4325 | } |
| 4326 | } |
| 4327 | } |
| 4328 | bool has_min = min > 0; |
| 4329 | bool has_max = max < RegExpTree::kInfinity; |
| 4330 | bool needs_counter = has_min || has_max; |
| 4331 | intptr_t reg_ctr = needs_counter ? compiler->AllocateRegister() |
| 4332 | : RegExpCompiler::kNoRegister; |
| 4333 | LoopChoiceNode* center = new (zone) |
| 4334 | LoopChoiceNode(body->min_match() == 0, compiler->read_backward(), zone); |
| 4335 | if (not_at_start && !compiler->read_backward()) center->set_not_at_start(); |
| 4336 | RegExpNode* loop_return = |
| 4337 | needs_counter ? static_cast<RegExpNode*>( |
| 4338 | ActionNode::IncrementRegister(reg_ctr, center)) |
| 4339 | : static_cast<RegExpNode*>(center); |
| 4340 | if (body_can_be_empty) { |
| 4341 | // If the body can be empty we need to check if it was and then |
| 4342 | // backtrack. |
| 4343 | loop_return = |
| 4344 | ActionNode::EmptyMatchCheck(body_start_reg, reg_ctr, min, loop_return); |
| 4345 | } |
| 4346 | RegExpNode* body_node = body->ToNode(compiler, loop_return); |
| 4347 | if (body_can_be_empty) { |
| 4348 | // If the body can be empty we need to store the start position |
| 4349 | // so we can bail out if it was empty. |
| 4350 | body_node = ActionNode::StorePosition(body_start_reg, false, body_node); |
| 4351 | } |
| 4352 | if (needs_capture_clearing) { |
| 4353 | // Before entering the body of this loop we need to clear captures. |
| 4354 | body_node = ActionNode::ClearCaptures(capture_registers, body_node); |
| 4355 | } |
| 4356 | GuardedAlternative body_alt(body_node); |
| 4357 | if (has_max) { |
| 4358 | Guard* body_guard = new (zone) Guard(reg_ctr, Guard::LT, max); |
| 4359 | body_alt.AddGuard(body_guard, zone); |
| 4360 | } |
| 4361 | GuardedAlternative rest_alt(on_success); |
| 4362 | if (has_min) { |
| 4363 | Guard* rest_guard = new (zone) Guard(reg_ctr, Guard::GEQ, min); |
| 4364 | rest_alt.AddGuard(rest_guard, zone); |
| 4365 | } |
| 4366 | if (is_greedy) { |
| 4367 | center->AddLoopAlternative(body_alt); |
| 4368 | center->AddContinueAlternative(rest_alt); |
| 4369 | } else { |
| 4370 | center->AddContinueAlternative(rest_alt); |
| 4371 | center->AddLoopAlternative(body_alt); |
| 4372 | } |
| 4373 | if (needs_counter) { |
| 4374 | return ActionNode::SetRegister(reg_ctr, 0, center); |
| 4375 | } else { |
| 4376 | return center; |
| 4377 | } |
| 4378 | } |
| 4379 | |
| 4380 | namespace { |
| 4381 | // Desugar \b to (?<=\w)(?=\W)|(?<=\W)(?=\w) and |
| 4382 | // \B to (?<=\w)(?=\w)|(?<=\W)(?=\W) |
| 4383 | RegExpNode* BoundaryAssertionAsLookaround(RegExpCompiler* compiler, |
| 4384 | RegExpNode* on_success, |
| 4385 | RegExpAssertion::AssertionType type, |
| 4386 | RegExpFlags flags) { |
| 4387 | ASSERT(flags.NeedsUnicodeCaseEquivalents()); |
| 4388 | ZoneGrowableArray<CharacterRange>* word_range = |
| 4389 | new ZoneGrowableArray<CharacterRange>(2); |
| 4390 | CharacterRange::AddClassEscape('w', word_range, true); |
| 4391 | int stack_register = compiler->UnicodeLookaroundStackRegister(); |
| 4392 | int position_register = compiler->UnicodeLookaroundPositionRegister(); |
| 4393 | ChoiceNode* result = new (OZ) ChoiceNode(2, OZ); |
| 4394 | // Add two choices. The (non-)boundary could start with a word or |
| 4395 | // a non-word-character. |
| 4396 | for (int i = 0; i < 2; i++) { |
| 4397 | bool lookbehind_for_word = i == 0; |
| 4398 | bool lookahead_for_word = |
| 4399 | (type == RegExpAssertion::BOUNDARY) ^ lookbehind_for_word; |
| 4400 | // Look to the left. |
| 4401 | RegExpLookaround::Builder lookbehind(lookbehind_for_word, on_success, |
| 4402 | stack_register, position_register); |
| 4403 | RegExpNode* backward = TextNode::CreateForCharacterRanges( |
| 4404 | word_range, true, lookbehind.on_match_success(), flags); |
| 4405 | // Look to the right. |
| 4406 | RegExpLookaround::Builder lookahead(lookahead_for_word, |
| 4407 | lookbehind.ForMatch(backward), |
| 4408 | stack_register, position_register); |
| 4409 | RegExpNode* forward = TextNode::CreateForCharacterRanges( |
| 4410 | word_range, false, lookahead.on_match_success(), flags); |
| 4411 | result->AddAlternative(GuardedAlternative(lookahead.ForMatch(forward))); |
| 4412 | } |
| 4413 | return result; |
| 4414 | } |
| 4415 | } // anonymous namespace |
| 4416 | |
| 4417 | RegExpNode* RegExpAssertion::ToNode(RegExpCompiler* compiler, |
| 4418 | RegExpNode* on_success) { |
| 4419 | switch (assertion_type()) { |
| 4420 | case START_OF_LINE: |
| 4421 | return AssertionNode::AfterNewline(on_success); |
| 4422 | case START_OF_INPUT: |
| 4423 | return AssertionNode::AtStart(on_success); |
| 4424 | case BOUNDARY: |
| 4425 | return flags_.NeedsUnicodeCaseEquivalents() |
| 4426 | ? BoundaryAssertionAsLookaround(compiler, on_success, BOUNDARY, |
| 4427 | flags_) |
| 4428 | : AssertionNode::AtBoundary(on_success); |
| 4429 | case NON_BOUNDARY: |
| 4430 | return flags_.NeedsUnicodeCaseEquivalents() |
| 4431 | ? BoundaryAssertionAsLookaround(compiler, on_success, |
| 4432 | NON_BOUNDARY, flags_) |
| 4433 | : AssertionNode::AtNonBoundary(on_success); |
| 4434 | case END_OF_INPUT: |
| 4435 | return AssertionNode::AtEnd(on_success); |
| 4436 | case END_OF_LINE: { |
| 4437 | // Compile $ in multiline regexps as an alternation with a positive |
| 4438 | // lookahead in one side and an end-of-input on the other side. |
| 4439 | // We need two registers for the lookahead. |
| 4440 | intptr_t stack_pointer_register = compiler->AllocateRegister(); |
| 4441 | intptr_t position_register = compiler->AllocateRegister(); |
| 4442 | // The ChoiceNode to distinguish between a newline and end-of-input. |
| 4443 | ChoiceNode* result = new ChoiceNode(2, on_success->zone()); |
| 4444 | // Create a newline atom. |
| 4445 | ZoneGrowableArray<CharacterRange>* newline_ranges = |
| 4446 | new ZoneGrowableArray<CharacterRange>(3); |
| 4447 | CharacterRange::AddClassEscape('n', newline_ranges); |
| 4448 | RegExpCharacterClass* newline_atom = |
| 4449 | new RegExpCharacterClass('n', RegExpFlags()); |
| 4450 | TextNode* newline_matcher = |
| 4451 | new TextNode(newline_atom, /*read_backwards=*/false, |
| 4452 | ActionNode::PositiveSubmatchSuccess( |
| 4453 | stack_pointer_register, position_register, |
| 4454 | 0, // No captures inside. |
| 4455 | -1, // Ignored if no captures. |
| 4456 | on_success)); |
| 4457 | // Create an end-of-input matcher. |
| 4458 | RegExpNode* end_of_line = ActionNode::BeginSubmatch( |
| 4459 | stack_pointer_register, position_register, newline_matcher); |
| 4460 | // Add the two alternatives to the ChoiceNode. |
| 4461 | GuardedAlternative eol_alternative(end_of_line); |
| 4462 | result->AddAlternative(eol_alternative); |
| 4463 | GuardedAlternative end_alternative(AssertionNode::AtEnd(on_success)); |
| 4464 | result->AddAlternative(end_alternative); |
| 4465 | return result; |
| 4466 | } |
| 4467 | default: |
| 4468 | UNREACHABLE(); |
| 4469 | } |
| 4470 | return on_success; |
| 4471 | } |
| 4472 | |
| 4473 | RegExpNode* RegExpBackReference::ToNode(RegExpCompiler* compiler, |
| 4474 | RegExpNode* on_success) { |
| 4475 | return new (OZ) BackReferenceNode(RegExpCapture::StartRegister(index()), |
| 4476 | RegExpCapture::EndRegister(index()), flags_, |
| 4477 | compiler->read_backward(), on_success); |
| 4478 | } |
| 4479 | |
| 4480 | RegExpNode* RegExpEmpty::ToNode(RegExpCompiler* compiler, |
| 4481 | RegExpNode* on_success) { |
| 4482 | return on_success; |
| 4483 | } |
| 4484 | |
| 4485 | RegExpLookaround::Builder::Builder(bool is_positive, |
| 4486 | RegExpNode* on_success, |
| 4487 | intptr_t stack_pointer_register, |
| 4488 | intptr_t position_register, |
| 4489 | intptr_t capture_register_count, |
| 4490 | intptr_t capture_register_start) |
| 4491 | : is_positive_(is_positive), |
| 4492 | on_success_(on_success), |
| 4493 | stack_pointer_register_(stack_pointer_register), |
| 4494 | position_register_(position_register) { |
| 4495 | if (is_positive_) { |
| 4496 | on_match_success_ = ActionNode::PositiveSubmatchSuccess( |
| 4497 | stack_pointer_register, position_register, capture_register_count, |
| 4498 | capture_register_start, on_success); |
| 4499 | } else { |
| 4500 | on_match_success_ = new (OZ) NegativeSubmatchSuccess( |
| 4501 | stack_pointer_register, position_register, capture_register_count, |
| 4502 | capture_register_start, OZ); |
| 4503 | } |
| 4504 | } |
| 4505 | |
| 4506 | RegExpNode* RegExpLookaround::Builder::ForMatch(RegExpNode* match) { |
| 4507 | if (is_positive_) { |
| 4508 | return ActionNode::BeginSubmatch(stack_pointer_register_, |
| 4509 | position_register_, match); |
| 4510 | } else { |
| 4511 | Zone* zone = on_success_->zone(); |
| 4512 | // We use a ChoiceNode to represent the negative lookaround. The first |
| 4513 | // alternative is the negative match. On success, the end node backtracks. |
| 4514 | // On failure, the second alternative is tried and leads to success. |
| 4515 | // NegativeLookaroundChoiceNode is a special ChoiceNode that ignores the |
| 4516 | // first exit when calculating quick checks. |
| 4517 | ChoiceNode* choice_node = new (zone) NegativeLookaroundChoiceNode( |
| 4518 | GuardedAlternative(match), GuardedAlternative(on_success_), zone); |
| 4519 | return ActionNode::BeginSubmatch(stack_pointer_register_, |
| 4520 | position_register_, choice_node); |
| 4521 | } |
| 4522 | } |
| 4523 | |
| 4524 | RegExpNode* RegExpLookaround::ToNode(RegExpCompiler* compiler, |
| 4525 | RegExpNode* on_success) { |
| 4526 | intptr_t stack_pointer_register = compiler->AllocateRegister(); |
| 4527 | intptr_t position_register = compiler->AllocateRegister(); |
| 4528 | |
| 4529 | const intptr_t registers_per_capture = 2; |
| 4530 | const intptr_t register_of_first_capture = 2; |
| 4531 | intptr_t register_count = capture_count_ * registers_per_capture; |
| 4532 | intptr_t register_start = |
| 4533 | register_of_first_capture + capture_from_ * registers_per_capture; |
| 4534 | |
| 4535 | RegExpNode* result; |
| 4536 | bool was_reading_backward = compiler->read_backward(); |
| 4537 | compiler->set_read_backward(type() == LOOKBEHIND); |
| 4538 | Builder builder(is_positive(), on_success, stack_pointer_register, |
| 4539 | position_register, register_count, register_start); |
| 4540 | RegExpNode* match = body_->ToNode(compiler, builder.on_match_success()); |
| 4541 | result = builder.ForMatch(match); |
| 4542 | compiler->set_read_backward(was_reading_backward); |
| 4543 | return result; |
| 4544 | } |
| 4545 | |
| 4546 | RegExpNode* RegExpCapture::ToNode(RegExpCompiler* compiler, |
| 4547 | RegExpNode* on_success) { |
| 4548 | return ToNode(body(), index(), compiler, on_success); |
| 4549 | } |
| 4550 | |
| 4551 | RegExpNode* RegExpCapture::ToNode(RegExpTree* body, |
| 4552 | intptr_t index, |
| 4553 | RegExpCompiler* compiler, |
| 4554 | RegExpNode* on_success) { |
| 4555 | ASSERT(body != nullptr); |
| 4556 | intptr_t start_reg = RegExpCapture::StartRegister(index); |
| 4557 | intptr_t end_reg = RegExpCapture::EndRegister(index); |
| 4558 | if (compiler->read_backward()) { |
| 4559 | intptr_t tmp = end_reg; |
| 4560 | end_reg = start_reg; |
| 4561 | start_reg = tmp; |
| 4562 | } |
| 4563 | RegExpNode* store_end = ActionNode::StorePosition(end_reg, true, on_success); |
| 4564 | RegExpNode* body_node = body->ToNode(compiler, store_end); |
| 4565 | return ActionNode::StorePosition(start_reg, true, body_node); |
| 4566 | } |
| 4567 | |
| 4568 | RegExpNode* RegExpAlternative::ToNode(RegExpCompiler* compiler, |
| 4569 | RegExpNode* on_success) { |
| 4570 | ZoneGrowableArray<RegExpTree*>* children = nodes(); |
| 4571 | RegExpNode* current = on_success; |
| 4572 | if (compiler->read_backward()) { |
| 4573 | for (intptr_t i = 0; i < children->length(); i++) { |
| 4574 | current = children->At(i)->ToNode(compiler, current); |
| 4575 | } |
| 4576 | } else { |
| 4577 | for (intptr_t i = children->length() - 1; i >= 0; i--) { |
| 4578 | current = children->At(i)->ToNode(compiler, current); |
| 4579 | } |
| 4580 | } |
| 4581 | return current; |
| 4582 | } |
| 4583 | |
| 4584 | static void AddClass(const int32_t* elmv, |
| 4585 | intptr_t elmc, |
| 4586 | ZoneGrowableArray<CharacterRange>* ranges) { |
| 4587 | elmc--; |
| 4588 | ASSERT(elmv[elmc] == kRangeEndMarker); |
| 4589 | for (intptr_t i = 0; i < elmc; i += 2) { |
| 4590 | ASSERT(elmv[i] < elmv[i + 1]); |
| 4591 | ranges->Add(CharacterRange(elmv[i], elmv[i + 1] - 1)); |
| 4592 | } |
| 4593 | } |
| 4594 | |
| 4595 | static void AddClassNegated(const int32_t* elmv, |
| 4596 | intptr_t elmc, |
| 4597 | ZoneGrowableArray<CharacterRange>* ranges) { |
| 4598 | elmc--; |
| 4599 | ASSERT(elmv[elmc] == kRangeEndMarker); |
| 4600 | ASSERT(elmv[0] != 0x0000); |
| 4601 | ASSERT(elmv[elmc - 1] != Utf::kMaxCodePoint); |
| 4602 | uint16_t last = 0x0000; |
| 4603 | for (intptr_t i = 0; i < elmc; i += 2) { |
| 4604 | ASSERT(last <= elmv[i] - 1); |
| 4605 | ASSERT(elmv[i] < elmv[i + 1]); |
| 4606 | ranges->Add(CharacterRange(last, elmv[i] - 1)); |
| 4607 | last = elmv[i + 1]; |
| 4608 | } |
| 4609 | ranges->Add(CharacterRange(last, Utf::kMaxCodePoint)); |
| 4610 | } |
| 4611 | |
| 4612 | void CharacterRange::AddClassEscape(uint16_t type, |
| 4613 | ZoneGrowableArray<CharacterRange>* ranges, |
| 4614 | bool add_unicode_case_equivalents) { |
| 4615 | if (add_unicode_case_equivalents && (type == 'w' || type == 'W')) { |
| 4616 | // See #sec-runtime-semantics-wordcharacters-abstract-operation |
| 4617 | // In case of unicode and ignore_case, we need to create the closure over |
| 4618 | // case equivalent characters before negating. |
| 4619 | ZoneGrowableArray<CharacterRange>* new_ranges = |
| 4620 | new ZoneGrowableArray<CharacterRange>(2); |
| 4621 | AddClass(kWordRanges, kWordRangeCount, new_ranges); |
| 4622 | AddUnicodeCaseEquivalents(new_ranges); |
| 4623 | if (type == 'W') { |
| 4624 | ZoneGrowableArray<CharacterRange>* negated = |
| 4625 | new ZoneGrowableArray<CharacterRange>(2); |
| 4626 | CharacterRange::Negate(new_ranges, negated); |
| 4627 | new_ranges = negated; |
| 4628 | } |
| 4629 | ranges->AddArray(*new_ranges); |
| 4630 | return; |
| 4631 | } |
| 4632 | AddClassEscape(type, ranges); |
| 4633 | } |
| 4634 | |
| 4635 | void CharacterRange::AddClassEscape(uint16_t type, |
| 4636 | ZoneGrowableArray<CharacterRange>* ranges) { |
| 4637 | switch (type) { |
| 4638 | case 's': |
| 4639 | AddClass(kSpaceRanges, kSpaceRangeCount, ranges); |
| 4640 | break; |
| 4641 | case 'S': |
| 4642 | AddClassNegated(kSpaceRanges, kSpaceRangeCount, ranges); |
| 4643 | break; |
| 4644 | case 'w': |
| 4645 | AddClass(kWordRanges, kWordRangeCount, ranges); |
| 4646 | break; |
| 4647 | case 'W': |
| 4648 | AddClassNegated(kWordRanges, kWordRangeCount, ranges); |
| 4649 | break; |
| 4650 | case 'd': |
| 4651 | AddClass(kDigitRanges, kDigitRangeCount, ranges); |
| 4652 | break; |
| 4653 | case 'D': |
| 4654 | AddClassNegated(kDigitRanges, kDigitRangeCount, ranges); |
| 4655 | break; |
| 4656 | case '.': |
| 4657 | AddClassNegated(kLineTerminatorRanges, kLineTerminatorRangeCount, ranges); |
| 4658 | break; |
| 4659 | // This is not a character range as defined by the spec but a |
| 4660 | // convenient shorthand for a character class that matches any |
| 4661 | // character. |
| 4662 | case '*': |
| 4663 | ranges->Add(CharacterRange::Everything()); |
| 4664 | break; |
| 4665 | // This is the set of characters matched by the $ and ^ symbols |
| 4666 | // in multiline mode. |
| 4667 | case 'n': |
| 4668 | AddClass(kLineTerminatorRanges, kLineTerminatorRangeCount, ranges); |
| 4669 | break; |
| 4670 | default: |
| 4671 | UNREACHABLE(); |
| 4672 | } |
| 4673 | } |
| 4674 | |
| 4675 | void CharacterRange::AddCaseEquivalents( |
| 4676 | ZoneGrowableArray<CharacterRange>* ranges, |
| 4677 | bool is_one_byte, |
| 4678 | Zone* zone) { |
| 4679 | CharacterRange::Canonicalize(ranges); |
| 4680 | int range_count = ranges->length(); |
| 4681 | for (intptr_t i = 0; i < range_count; i++) { |
| 4682 | CharacterRange range = ranges->At(i); |
| 4683 | int32_t bottom = range.from(); |
| 4684 | if (bottom > Utf16::kMaxCodeUnit) continue; |
| 4685 | int32_t top = Utils::Minimum(range.to(), Utf16::kMaxCodeUnit); |
| 4686 | // Nothing to be done for surrogates |
| 4687 | if (bottom >= Utf16::kLeadSurrogateStart && |
| 4688 | top <= Utf16::kTrailSurrogateEnd) { |
| 4689 | continue; |
| 4690 | } |
| 4691 | if (is_one_byte && !RangeContainsLatin1Equivalents(range)) { |
| 4692 | if (bottom > Symbols::kMaxOneCharCodeSymbol) continue; |
| 4693 | if (top > Symbols::kMaxOneCharCodeSymbol) { |
| 4694 | top = Symbols::kMaxOneCharCodeSymbol; |
| 4695 | } |
| 4696 | } |
| 4697 | |
| 4698 | unibrow::Mapping<unibrow::Ecma262UnCanonicalize> jsregexp_uncanonicalize; |
| 4699 | unibrow::Mapping<unibrow::CanonicalizationRange> jsregexp_canonrange; |
| 4700 | int32_t chars[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 4701 | if (top == bottom) { |
| 4702 | // If this is a singleton we just expand the one character. |
| 4703 | intptr_t length = jsregexp_uncanonicalize.get(bottom, '\0', chars); |
| 4704 | for (intptr_t i = 0; i < length; i++) { |
| 4705 | int32_t chr = chars[i]; |
| 4706 | if (chr != bottom) { |
| 4707 | ranges->Add(CharacterRange::Singleton(chars[i])); |
| 4708 | } |
| 4709 | } |
| 4710 | } else { |
| 4711 | // If this is a range we expand the characters block by block, |
| 4712 | // expanding contiguous subranges (blocks) one at a time. |
| 4713 | // The approach is as follows. For a given start character we |
| 4714 | // look up the remainder of the block that contains it (represented |
| 4715 | // by the end point), for instance we find 'z' if the character |
| 4716 | // is 'c'. A block is characterized by the property |
| 4717 | // that all characters uncanonicalize in the same way, except that |
| 4718 | // each entry in the result is incremented by the distance from the first |
| 4719 | // element. So a-z is a block because 'a' uncanonicalizes to ['a', 'A'] |
| 4720 | // and the k'th letter uncanonicalizes to ['a' + k, 'A' + k]. |
| 4721 | // Once we've found the end point we look up its uncanonicalization |
| 4722 | // and produce a range for each element. For instance for [c-f] |
| 4723 | // we look up ['z', 'Z'] and produce [c-f] and [C-F]. We then only |
| 4724 | // add a range if it is not already contained in the input, so [c-f] |
| 4725 | // will be skipped but [C-F] will be added. If this range is not |
| 4726 | // completely contained in a block we do this for all the blocks |
| 4727 | // covered by the range (handling characters that is not in a block |
| 4728 | // as a "singleton block"). |
| 4729 | int32_t range[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 4730 | intptr_t pos = bottom; |
| 4731 | while (pos <= top) { |
| 4732 | intptr_t length = jsregexp_canonrange.get(pos, '\0', range); |
| 4733 | int32_t block_end; |
| 4734 | if (length == 0) { |
| 4735 | block_end = pos; |
| 4736 | } else { |
| 4737 | ASSERT(length == 1); |
| 4738 | block_end = range[0]; |
| 4739 | } |
| 4740 | intptr_t end = (block_end > top) ? top : block_end; |
| 4741 | length = jsregexp_uncanonicalize.get(block_end, '\0', range); |
| 4742 | for (intptr_t i = 0; i < length; i++) { |
| 4743 | int32_t c = range[i]; |
| 4744 | int32_t range_from = c - (block_end - pos); |
| 4745 | int32_t range_to = c - (block_end - end); |
| 4746 | if (!(bottom <= range_from && range_to <= top)) { |
| 4747 | ranges->Add(CharacterRange(range_from, range_to)); |
| 4748 | } |
| 4749 | } |
| 4750 | pos = end + 1; |
| 4751 | } |
| 4752 | } |
| 4753 | } |
| 4754 | } |
| 4755 | |
| 4756 | bool CharacterRange::IsCanonical(ZoneGrowableArray<CharacterRange>* ranges) { |
| 4757 | ASSERT(ranges != NULL); |
| 4758 | intptr_t n = ranges->length(); |
| 4759 | if (n <= 1) return true; |
| 4760 | intptr_t max = ranges->At(0).to(); |
| 4761 | for (intptr_t i = 1; i < n; i++) { |
| 4762 | CharacterRange next_range = ranges->At(i); |
| 4763 | if (next_range.from() <= max + 1) return false; |
| 4764 | max = next_range.to(); |
| 4765 | } |
| 4766 | return true; |
| 4767 | } |
| 4768 | |
| 4769 | ZoneGrowableArray<CharacterRange>* CharacterSet::ranges() { |
| 4770 | if (ranges_ == NULL) { |
| 4771 | ranges_ = new ZoneGrowableArray<CharacterRange>(2); |
| 4772 | CharacterRange::AddClassEscape(standard_set_type_, ranges_); |
| 4773 | } |
| 4774 | return ranges_; |
| 4775 | } |
| 4776 | |
| 4777 | // Move a number of elements in a zone array to another position |
| 4778 | // in the same array. Handles overlapping source and target areas. |
| 4779 | static void MoveRanges(ZoneGrowableArray<CharacterRange>* list, |
| 4780 | intptr_t from, |
| 4781 | intptr_t to, |
| 4782 | intptr_t count) { |
| 4783 | // Ranges are potentially overlapping. |
| 4784 | if (from < to) { |
| 4785 | for (intptr_t i = count - 1; i >= 0; i--) { |
| 4786 | (*list)[to + i] = list->At(from + i); |
| 4787 | } |
| 4788 | } else { |
| 4789 | for (intptr_t i = 0; i < count; i++) { |
| 4790 | (*list)[to + i] = list->At(from + i); |
| 4791 | } |
| 4792 | } |
| 4793 | } |
| 4794 | |
| 4795 | static intptr_t InsertRangeInCanonicalList( |
| 4796 | ZoneGrowableArray<CharacterRange>* list, |
| 4797 | intptr_t count, |
| 4798 | CharacterRange insert) { |
| 4799 | // Inserts a range into list[0..count[, which must be sorted |
| 4800 | // by from value and non-overlapping and non-adjacent, using at most |
| 4801 | // list[0..count] for the result. Returns the number of resulting |
| 4802 | // canonicalized ranges. Inserting a range may collapse existing ranges into |
| 4803 | // fewer ranges, so the return value can be anything in the range 1..count+1. |
| 4804 | int32_t from = insert.from(); |
| 4805 | int32_t to = insert.to(); |
| 4806 | intptr_t start_pos = 0; |
| 4807 | intptr_t end_pos = count; |
| 4808 | for (intptr_t i = count - 1; i >= 0; i--) { |
| 4809 | CharacterRange current = list->At(i); |
| 4810 | if (current.from() > to + 1) { |
| 4811 | end_pos = i; |
| 4812 | } else if (current.to() + 1 < from) { |
| 4813 | start_pos = i + 1; |
| 4814 | break; |
| 4815 | } |
| 4816 | } |
| 4817 | |
| 4818 | // Inserted range overlaps, or is adjacent to, ranges at positions |
| 4819 | // [start_pos..end_pos[. Ranges before start_pos or at or after end_pos are |
| 4820 | // not affected by the insertion. |
| 4821 | // If start_pos == end_pos, the range must be inserted before start_pos. |
| 4822 | // if start_pos < end_pos, the entire range from start_pos to end_pos |
| 4823 | // must be merged with the insert range. |
| 4824 | |
| 4825 | if (start_pos == end_pos) { |
| 4826 | // Insert between existing ranges at position start_pos. |
| 4827 | if (start_pos < count) { |
| 4828 | MoveRanges(list, start_pos, start_pos + 1, count - start_pos); |
| 4829 | } |
| 4830 | (*list)[start_pos] = insert; |
| 4831 | return count + 1; |
| 4832 | } |
| 4833 | if (start_pos + 1 == end_pos) { |
| 4834 | // Replace single existing range at position start_pos. |
| 4835 | CharacterRange to_replace = list->At(start_pos); |
| 4836 | intptr_t new_from = Utils::Minimum(to_replace.from(), from); |
| 4837 | intptr_t new_to = Utils::Maximum(to_replace.to(), to); |
| 4838 | (*list)[start_pos] = CharacterRange(new_from, new_to); |
| 4839 | return count; |
| 4840 | } |
| 4841 | // Replace a number of existing ranges from start_pos to end_pos - 1. |
| 4842 | // Move the remaining ranges down. |
| 4843 | |
| 4844 | intptr_t new_from = Utils::Minimum(list->At(start_pos).from(), from); |
| 4845 | intptr_t new_to = Utils::Maximum(list->At(end_pos - 1).to(), to); |
| 4846 | if (end_pos < count) { |
| 4847 | MoveRanges(list, end_pos, start_pos + 1, count - end_pos); |
| 4848 | } |
| 4849 | (*list)[start_pos] = CharacterRange(new_from, new_to); |
| 4850 | return count - (end_pos - start_pos) + 1; |
| 4851 | } |
| 4852 | |
| 4853 | void CharacterSet::Canonicalize() { |
| 4854 | // Special/default classes are always considered canonical. The result |
| 4855 | // of calling ranges() will be sorted. |
| 4856 | if (ranges_ == NULL) return; |
| 4857 | CharacterRange::Canonicalize(ranges_); |
| 4858 | } |
| 4859 | |
| 4860 | void CharacterRange::Canonicalize( |
| 4861 | ZoneGrowableArray<CharacterRange>* character_ranges) { |
| 4862 | if (character_ranges->length() <= 1) return; |
| 4863 | // Check whether ranges are already canonical (increasing, non-overlapping, |
| 4864 | // non-adjacent). |
| 4865 | intptr_t n = character_ranges->length(); |
| 4866 | intptr_t max = character_ranges->At(0).to(); |
| 4867 | intptr_t i = 1; |
| 4868 | while (i < n) { |
| 4869 | CharacterRange current = character_ranges->At(i); |
| 4870 | if (current.from() <= max + 1) { |
| 4871 | break; |
| 4872 | } |
| 4873 | max = current.to(); |
| 4874 | i++; |
| 4875 | } |
| 4876 | // Canonical until the i'th range. If that's all of them, we are done. |
| 4877 | if (i == n) return; |
| 4878 | |
| 4879 | // The ranges at index i and forward are not canonicalized. Make them so by |
| 4880 | // doing the equivalent of insertion sort (inserting each into the previous |
| 4881 | // list, in order). |
| 4882 | // Notice that inserting a range can reduce the number of ranges in the |
| 4883 | // result due to combining of adjacent and overlapping ranges. |
| 4884 | intptr_t read = i; // Range to insert. |
| 4885 | intptr_t num_canonical = i; // Length of canonicalized part of list. |
| 4886 | do { |
| 4887 | num_canonical = InsertRangeInCanonicalList(character_ranges, num_canonical, |
| 4888 | character_ranges->At(read)); |
| 4889 | read++; |
| 4890 | } while (read < n); |
| 4891 | character_ranges->TruncateTo(num_canonical); |
| 4892 | |
| 4893 | ASSERT(CharacterRange::IsCanonical(character_ranges)); |
| 4894 | } |
| 4895 | |
| 4896 | void CharacterRange::Negate(ZoneGrowableArray<CharacterRange>* ranges, |
| 4897 | ZoneGrowableArray<CharacterRange>* negated_ranges) { |
| 4898 | ASSERT(CharacterRange::IsCanonical(ranges)); |
| 4899 | ASSERT(negated_ranges->length() == 0); |
| 4900 | intptr_t range_count = ranges->length(); |
| 4901 | uint32_t from = 0; |
| 4902 | intptr_t i = 0; |
| 4903 | if (range_count > 0 && ranges->At(0).from() == 0) { |
| 4904 | from = ranges->At(0).to(); |
| 4905 | i = 1; |
| 4906 | } |
| 4907 | while (i < range_count) { |
| 4908 | CharacterRange range = ranges->At(i); |
| 4909 | negated_ranges->Add(CharacterRange(from + 1, range.from() - 1)); |
| 4910 | from = range.to(); |
| 4911 | i++; |
| 4912 | } |
| 4913 | if (from < Utf::kMaxCodePoint) { |
| 4914 | negated_ranges->Add(CharacterRange(from + 1, Utf::kMaxCodePoint)); |
| 4915 | } |
| 4916 | } |
| 4917 | |
| 4918 | // ------------------------------------------------------------------- |
| 4919 | // Splay tree |
| 4920 | |
| 4921 | // Workaround for the fact that ZoneGrowableArray does not have contains(). |
| 4922 | static bool ArrayContains(ZoneGrowableArray<unsigned>* array, unsigned value) { |
| 4923 | for (intptr_t i = 0; i < array->length(); i++) { |
| 4924 | if (array->At(i) == value) { |
| 4925 | return true; |
| 4926 | } |
| 4927 | } |
| 4928 | return false; |
| 4929 | } |
| 4930 | |
| 4931 | OutSet* OutSet::Extend(unsigned value, Zone* zone) { |
| 4932 | if (Get(value)) return this; |
| 4933 | if (successors() != nullptr) { |
| 4934 | for (int i = 0; i < successors()->length(); i++) { |
| 4935 | OutSet* successor = successors()->At(i); |
| 4936 | if (successor->Get(value)) return successor; |
| 4937 | } |
| 4938 | } else { |
| 4939 | successors_ = new (zone) ZoneGrowableArray<OutSet*>(2); |
| 4940 | } |
| 4941 | OutSet* result = new (zone) OutSet(first_, remaining_); |
| 4942 | result->Set(value, zone); |
| 4943 | successors()->Add(result); |
| 4944 | return result; |
| 4945 | } |
| 4946 | |
| 4947 | void OutSet::Set(unsigned value, Zone* zone) { |
| 4948 | if (value < kFirstLimit) { |
| 4949 | first_ |= (1 << value); |
| 4950 | } else { |
| 4951 | if (remaining_ == NULL) |
| 4952 | remaining_ = new (zone) ZoneGrowableArray<unsigned>(1); |
| 4953 | |
| 4954 | bool remaining_contains_value = ArrayContains(remaining_, value); |
| 4955 | if (remaining_->is_empty() || !remaining_contains_value) { |
| 4956 | remaining_->Add(value); |
| 4957 | } |
| 4958 | } |
| 4959 | } |
| 4960 | |
| 4961 | bool OutSet::Get(unsigned value) const { |
| 4962 | if (value < kFirstLimit) { |
| 4963 | return (first_ & (1 << value)) != 0; |
| 4964 | } else if (remaining_ == NULL) { |
| 4965 | return false; |
| 4966 | } else { |
| 4967 | return ArrayContains(remaining_, value); |
| 4968 | } |
| 4969 | } |
| 4970 | |
| 4971 | const int32_t ChoiceTable::Config::kNoKey = Utf::kInvalidChar; |
| 4972 | |
| 4973 | void ChoiceTable::AddRange(CharacterRange full_range, |
| 4974 | int32_t value, |
| 4975 | Zone* zone) { |
| 4976 | CharacterRange current = full_range; |
| 4977 | if (tree()->is_empty()) { |
| 4978 | // If this is the first range we just insert into the table. |
| 4979 | ZoneSplayTree<Config>::Locator loc; |
| 4980 | bool inserted = tree()->Insert(current.from(), &loc); |
| 4981 | ASSERT(inserted); |
| 4982 | USE(inserted); |
| 4983 | loc.set_value( |
| 4984 | Entry(current.from(), current.to(), empty()->Extend(value, zone))); |
| 4985 | return; |
| 4986 | } |
| 4987 | // First see if there is a range to the left of this one that |
| 4988 | // overlaps. |
| 4989 | ZoneSplayTree<Config>::Locator loc; |
| 4990 | if (tree()->FindGreatestLessThan(current.from(), &loc)) { |
| 4991 | Entry* entry = &loc.value(); |
| 4992 | // If we've found a range that overlaps with this one, and it |
| 4993 | // starts strictly to the left of this one, we have to fix it |
| 4994 | // because the following code only handles ranges that start on |
| 4995 | // or after the start point of the range we're adding. |
| 4996 | if (entry->from() < current.from() && entry->to() >= current.from()) { |
| 4997 | // Snap the overlapping range in half around the start point of |
| 4998 | // the range we're adding. |
| 4999 | CharacterRange left = |
| 5000 | CharacterRange::Range(entry->from(), current.from() - 1); |
| 5001 | CharacterRange right = CharacterRange::Range(current.from(), entry->to()); |
| 5002 | // The left part of the overlapping range doesn't overlap. |
| 5003 | // Truncate the whole entry to be just the left part. |
| 5004 | entry->set_to(left.to()); |
| 5005 | // The right part is the one that overlaps. We add this part |
| 5006 | // to the map and let the next step deal with merging it with |
| 5007 | // the range we're adding. |
| 5008 | ZoneSplayTree<Config>::Locator loc; |
| 5009 | bool inserted = tree()->Insert(right.from(), &loc); |
| 5010 | ASSERT(inserted); |
| 5011 | USE(inserted); |
| 5012 | loc.set_value(Entry(right.from(), right.to(), entry->out_set())); |
| 5013 | } |
| 5014 | } |
| 5015 | while (current.is_valid()) { |
| 5016 | if (tree()->FindLeastGreaterThan(current.from(), &loc) && |
| 5017 | (loc.value().from() <= current.to()) && |
| 5018 | (loc.value().to() >= current.from())) { |
| 5019 | Entry* entry = &loc.value(); |
| 5020 | // We have overlap. If there is space between the start point of |
| 5021 | // the range we're adding and where the overlapping range starts |
| 5022 | // then we have to add a range covering just that space. |
| 5023 | if (current.from() < entry->from()) { |
| 5024 | ZoneSplayTree<Config>::Locator ins; |
| 5025 | bool inserted = tree()->Insert(current.from(), &ins); |
| 5026 | ASSERT(inserted); |
| 5027 | USE(inserted); |
| 5028 | ins.set_value(Entry(current.from(), entry->from() - 1, |
| 5029 | empty()->Extend(value, zone))); |
| 5030 | current.set_from(entry->from()); |
| 5031 | } |
| 5032 | ASSERT(current.from() == entry->from()); |
| 5033 | // If the overlapping range extends beyond the one we want to add |
| 5034 | // we have to snap the right part off and add it separately. |
| 5035 | if (entry->to() > current.to()) { |
| 5036 | ZoneSplayTree<Config>::Locator ins; |
| 5037 | bool inserted = tree()->Insert(current.to() + 1, &ins); |
| 5038 | ASSERT(inserted); |
| 5039 | USE(inserted); |
| 5040 | ins.set_value(Entry(current.to() + 1, entry->to(), entry->out_set())); |
| 5041 | entry->set_to(current.to()); |
| 5042 | } |
| 5043 | ASSERT(entry->to() <= current.to()); |
| 5044 | // The overlapping range is now completely contained by the range |
| 5045 | // we're adding so we can just update it and move the start point |
| 5046 | // of the range we're adding just past it. |
| 5047 | entry->AddValue(value, zone); |
| 5048 | ASSERT(entry->to() + 1 > current.from()); |
| 5049 | current.set_from(entry->to() + 1); |
| 5050 | } else { |
| 5051 | // There is no overlap so we can just add the range |
| 5052 | ZoneSplayTree<Config>::Locator ins; |
| 5053 | bool inserted = tree()->Insert(current.from(), &ins); |
| 5054 | ASSERT(inserted); |
| 5055 | USE(inserted); |
| 5056 | ins.set_value( |
| 5057 | Entry(current.from(), current.to(), empty()->Extend(value, zone))); |
| 5058 | break; |
| 5059 | } |
| 5060 | } |
| 5061 | } |
| 5062 | |
| 5063 | OutSet* ChoiceTable::Get(int32_t value) { |
| 5064 | ZoneSplayTree<Config>::Locator loc; |
| 5065 | if (!tree()->FindGreatestLessThan(value, &loc)) return empty(); |
| 5066 | Entry* entry = &loc.value(); |
| 5067 | if (value <= entry->to()) |
| 5068 | return entry->out_set(); |
| 5069 | else |
| 5070 | return empty(); |
| 5071 | } |
| 5072 | |
| 5073 | // ------------------------------------------------------------------- |
| 5074 | // Analysis |
| 5075 | |
| 5076 | void Analysis::EnsureAnalyzed(RegExpNode* that) { |
| 5077 | if (that->info()->been_analyzed || that->info()->being_analyzed) return; |
| 5078 | that->info()->being_analyzed = true; |
| 5079 | that->Accept(this); |
| 5080 | that->info()->being_analyzed = false; |
| 5081 | that->info()->been_analyzed = true; |
| 5082 | } |
| 5083 | |
| 5084 | void Analysis::VisitEnd(EndNode* that) { |
| 5085 | // nothing to do |
| 5086 | } |
| 5087 | |
| 5088 | void TextNode::CalculateOffsets() { |
| 5089 | intptr_t element_count = elements()->length(); |
| 5090 | // Set up the offsets of the elements relative to the start. This is a fixed |
| 5091 | // quantity since a TextNode can only contain fixed-width things. |
| 5092 | intptr_t cp_offset = 0; |
| 5093 | for (intptr_t i = 0; i < element_count; i++) { |
| 5094 | TextElement& elm = (*elements())[i]; |
| 5095 | elm.set_cp_offset(cp_offset); |
| 5096 | cp_offset += elm.length(); |
| 5097 | } |
| 5098 | } |
| 5099 | |
| 5100 | void Analysis::VisitText(TextNode* that) { |
| 5101 | that->MakeCaseIndependent(is_one_byte_); |
| 5102 | EnsureAnalyzed(that->on_success()); |
| 5103 | if (!has_failed()) { |
| 5104 | that->CalculateOffsets(); |
| 5105 | } |
| 5106 | } |
| 5107 | |
| 5108 | void Analysis::VisitAction(ActionNode* that) { |
| 5109 | RegExpNode* target = that->on_success(); |
| 5110 | EnsureAnalyzed(target); |
| 5111 | if (!has_failed()) { |
| 5112 | // If the next node is interested in what it follows then this node |
| 5113 | // has to be interested too so it can pass the information on. |
| 5114 | that->info()->AddFromFollowing(target->info()); |
| 5115 | } |
| 5116 | } |
| 5117 | |
| 5118 | void Analysis::VisitChoice(ChoiceNode* that) { |
| 5119 | NodeInfo* info = that->info(); |
| 5120 | for (intptr_t i = 0; i < that->alternatives()->length(); i++) { |
| 5121 | RegExpNode* node = (*that->alternatives())[i].node(); |
| 5122 | EnsureAnalyzed(node); |
| 5123 | if (has_failed()) return; |
| 5124 | // Anything the following nodes need to know has to be known by |
| 5125 | // this node also, so it can pass it on. |
| 5126 | info->AddFromFollowing(node->info()); |
| 5127 | } |
| 5128 | } |
| 5129 | |
| 5130 | void Analysis::VisitLoopChoice(LoopChoiceNode* that) { |
| 5131 | NodeInfo* info = that->info(); |
| 5132 | for (intptr_t i = 0; i < that->alternatives()->length(); i++) { |
| 5133 | RegExpNode* node = (*that->alternatives())[i].node(); |
| 5134 | if (node != that->loop_node()) { |
| 5135 | EnsureAnalyzed(node); |
| 5136 | if (has_failed()) return; |
| 5137 | info->AddFromFollowing(node->info()); |
| 5138 | } |
| 5139 | } |
| 5140 | // Check the loop last since it may need the value of this node |
| 5141 | // to get a correct result. |
| 5142 | EnsureAnalyzed(that->loop_node()); |
| 5143 | if (!has_failed()) { |
| 5144 | info->AddFromFollowing(that->loop_node()->info()); |
| 5145 | } |
| 5146 | } |
| 5147 | |
| 5148 | void Analysis::VisitBackReference(BackReferenceNode* that) { |
| 5149 | EnsureAnalyzed(that->on_success()); |
| 5150 | } |
| 5151 | |
| 5152 | void Analysis::VisitAssertion(AssertionNode* that) { |
| 5153 | EnsureAnalyzed(that->on_success()); |
| 5154 | } |
| 5155 | |
| 5156 | void BackReferenceNode::FillInBMInfo(intptr_t offset, |
| 5157 | intptr_t budget, |
| 5158 | BoyerMooreLookahead* bm, |
| 5159 | bool not_at_start) { |
| 5160 | // Working out the set of characters that a backreference can match is too |
| 5161 | // hard, so we just say that any character can match. |
| 5162 | bm->SetRest(offset); |
| 5163 | SaveBMInfo(bm, not_at_start, offset); |
| 5164 | } |
| 5165 | |
| 5166 | COMPILE_ASSERT(BoyerMoorePositionInfo::kMapSize == |
| 5167 | RegExpMacroAssembler::kTableSize); |
| 5168 | |
| 5169 | void ChoiceNode::FillInBMInfo(intptr_t offset, |
| 5170 | intptr_t budget, |
| 5171 | BoyerMooreLookahead* bm, |
| 5172 | bool not_at_start) { |
| 5173 | ZoneGrowableArray<GuardedAlternative>* alts = alternatives(); |
| 5174 | budget = (budget - 1) / alts->length(); |
| 5175 | for (intptr_t i = 0; i < alts->length(); i++) { |
| 5176 | GuardedAlternative& alt = (*alts)[i]; |
| 5177 | if (alt.guards() != NULL && alt.guards()->length() != 0) { |
| 5178 | bm->SetRest(offset); // Give up trying to fill in info. |
| 5179 | SaveBMInfo(bm, not_at_start, offset); |
| 5180 | return; |
| 5181 | } |
| 5182 | alt.node()->FillInBMInfo(offset, budget, bm, not_at_start); |
| 5183 | } |
| 5184 | SaveBMInfo(bm, not_at_start, offset); |
| 5185 | } |
| 5186 | |
| 5187 | void TextNode::FillInBMInfo(intptr_t initial_offset, |
| 5188 | intptr_t budget, |
| 5189 | BoyerMooreLookahead* bm, |
| 5190 | bool not_at_start) { |
| 5191 | if (initial_offset >= bm->length()) return; |
| 5192 | intptr_t offset = initial_offset; |
| 5193 | intptr_t max_char = bm->max_char(); |
| 5194 | for (intptr_t i = 0; i < elements()->length(); i++) { |
| 5195 | if (offset >= bm->length()) { |
| 5196 | if (initial_offset == 0) set_bm_info(not_at_start, bm); |
| 5197 | return; |
| 5198 | } |
| 5199 | TextElement text = elements()->At(i); |
| 5200 | if (text.text_type() == TextElement::ATOM) { |
| 5201 | RegExpAtom* atom = text.atom(); |
| 5202 | for (intptr_t j = 0; j < atom->length(); j++, offset++) { |
| 5203 | if (offset >= bm->length()) { |
| 5204 | if (initial_offset == 0) set_bm_info(not_at_start, bm); |
| 5205 | return; |
| 5206 | } |
| 5207 | uint16_t character = atom->data()->At(j); |
| 5208 | if (atom->flags().IgnoreCase()) { |
| 5209 | int32_t chars[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 5210 | intptr_t length = GetCaseIndependentLetters( |
| 5211 | character, bm->max_char() == Symbols::kMaxOneCharCodeSymbol, |
| 5212 | chars); |
| 5213 | for (intptr_t j = 0; j < length; j++) { |
| 5214 | bm->Set(offset, chars[j]); |
| 5215 | } |
| 5216 | } else { |
| 5217 | if (character <= max_char) bm->Set(offset, character); |
| 5218 | } |
| 5219 | } |
| 5220 | } else { |
| 5221 | ASSERT(text.text_type() == TextElement::CHAR_CLASS); |
| 5222 | RegExpCharacterClass* char_class = text.char_class(); |
| 5223 | ZoneGrowableArray<CharacterRange>* ranges = char_class->ranges(); |
| 5224 | if (char_class->is_negated()) { |
| 5225 | bm->SetAll(offset); |
| 5226 | } else { |
| 5227 | for (intptr_t k = 0; k < ranges->length(); k++) { |
| 5228 | const CharacterRange& range = ranges->At(k); |
| 5229 | if (range.from() > max_char) continue; |
| 5230 | intptr_t to = |
| 5231 | Utils::Minimum(max_char, static_cast<intptr_t>(range.to())); |
| 5232 | bm->SetInterval(offset, Interval(range.from(), to)); |
| 5233 | } |
| 5234 | } |
| 5235 | offset++; |
| 5236 | } |
| 5237 | } |
| 5238 | if (offset >= bm->length()) { |
| 5239 | if (initial_offset == 0) set_bm_info(not_at_start, bm); |
| 5240 | return; |
| 5241 | } |
| 5242 | on_success()->FillInBMInfo(offset, budget - 1, bm, |
| 5243 | true); // Not at start after a text node. |
| 5244 | if (initial_offset == 0) set_bm_info(not_at_start, bm); |
| 5245 | } |
| 5246 | |
| 5247 | RegExpNode* OptionallyStepBackToLeadSurrogate(RegExpCompiler* compiler, |
| 5248 | RegExpNode* on_success, |
| 5249 | RegExpFlags flags) { |
| 5250 | // If the regexp matching starts within a surrogate pair, step back |
| 5251 | // to the lead surrogate and start matching from there. |
| 5252 | ASSERT(!compiler->read_backward()); |
| 5253 | Zone* zone = compiler->zone(); |
| 5254 | |
| 5255 | auto lead_surrogates = CharacterRange::List( |
| 5256 | on_success->zone(), CharacterRange::Range(Utf16::kLeadSurrogateStart, |
| 5257 | Utf16::kLeadSurrogateEnd)); |
| 5258 | auto trail_surrogates = CharacterRange::List( |
| 5259 | on_success->zone(), CharacterRange::Range(Utf16::kTrailSurrogateStart, |
| 5260 | Utf16::kTrailSurrogateEnd)); |
| 5261 | |
| 5262 | ChoiceNode* optional_step_back = new (zone) ChoiceNode(2, zone); |
| 5263 | |
| 5264 | int stack_register = compiler->UnicodeLookaroundStackRegister(); |
| 5265 | int position_register = compiler->UnicodeLookaroundPositionRegister(); |
| 5266 | RegExpNode* step_back = TextNode::CreateForCharacterRanges( |
| 5267 | lead_surrogates, /*read_backward=*/true, on_success, flags); |
| 5268 | RegExpLookaround::Builder builder(/*is_positive=*/true, step_back, |
| 5269 | stack_register, position_register); |
| 5270 | RegExpNode* match_trail = TextNode::CreateForCharacterRanges( |
| 5271 | trail_surrogates, /*read_backward=*/false, builder.on_match_success(), |
| 5272 | flags); |
| 5273 | |
| 5274 | optional_step_back->AddAlternative( |
| 5275 | GuardedAlternative(builder.ForMatch(match_trail))); |
| 5276 | optional_step_back->AddAlternative(GuardedAlternative(on_success)); |
| 5277 | |
| 5278 | return optional_step_back; |
| 5279 | } |
| 5280 | |
| 5281 | #if !defined(DART_PRECOMPILED_RUNTIME) |
| 5282 | RegExpEngine::CompilationResult RegExpEngine::CompileIR( |
| 5283 | RegExpCompileData* data, |
| 5284 | const ParsedFunction* parsed_function, |
| 5285 | const ZoneGrowableArray<const ICData*>& ic_data_array, |
| 5286 | intptr_t osr_id) { |
| 5287 | ASSERT(!FLAG_interpret_irregexp); |
| 5288 | Zone* zone = Thread::Current()->zone(); |
| 5289 | |
| 5290 | const Function& function = parsed_function->function(); |
| 5291 | const intptr_t specialization_cid = function.string_specialization_cid(); |
| 5292 | const bool is_sticky = function.is_sticky_specialization(); |
| 5293 | const bool is_one_byte = (specialization_cid == kOneByteStringCid || |
| 5294 | specialization_cid == kExternalOneByteStringCid); |
| 5295 | RegExp& regexp = RegExp::Handle(zone, function.regexp()); |
| 5296 | const String& pattern = String::Handle(zone, regexp.pattern()); |
| 5297 | |
| 5298 | ASSERT(!regexp.IsNull()); |
| 5299 | ASSERT(!pattern.IsNull()); |
| 5300 | |
| 5301 | const bool is_global = regexp.flags().IsGlobal(); |
| 5302 | const bool is_unicode = regexp.flags().IsUnicode(); |
| 5303 | |
| 5304 | RegExpCompiler compiler(data->capture_count, is_one_byte); |
| 5305 | |
| 5306 | // TODO(zerny): Frequency sampling is currently disabled because of several |
| 5307 | // issues. We do not want to store subject strings in the regexp object since |
| 5308 | // they might be long and we should not prevent their garbage collection. |
| 5309 | // Passing them to this function explicitly does not help, since we must |
| 5310 | // generate exactly the same IR for both the unoptimizing and optimizing |
| 5311 | // pipelines (otherwise it gets confused when i.e. deopt id's differ). |
| 5312 | // An option would be to store sampling results in the regexp object, but |
| 5313 | // I'm not sure the performance gains are relevant enough. |
| 5314 | |
| 5315 | // Wrap the body of the regexp in capture #0. |
| 5316 | RegExpNode* captured_body = |
| 5317 | RegExpCapture::ToNode(data->tree, 0, &compiler, compiler.accept()); |
| 5318 | |
| 5319 | RegExpNode* node = captured_body; |
| 5320 | const bool is_end_anchored = data->tree->IsAnchoredAtEnd(); |
| 5321 | const bool is_start_anchored = data->tree->IsAnchoredAtStart(); |
| 5322 | intptr_t max_length = data->tree->max_match(); |
| 5323 | if (!is_start_anchored && !is_sticky) { |
| 5324 | // Add a .*? at the beginning, outside the body capture, unless |
| 5325 | // this expression is anchored at the beginning or is sticky. |
| 5326 | RegExpNode* loop_node = RegExpQuantifier::ToNode( |
| 5327 | 0, RegExpTree::kInfinity, false, |
| 5328 | new (zone) RegExpCharacterClass('*', RegExpFlags()), &compiler, |
| 5329 | captured_body, data->contains_anchor); |
| 5330 | |
| 5331 | if (data->contains_anchor) { |
| 5332 | // Unroll loop once, to take care of the case that might start |
| 5333 | // at the start of input. |
| 5334 | ChoiceNode* first_step_node = new (zone) ChoiceNode(2, zone); |
| 5335 | first_step_node->AddAlternative(GuardedAlternative(captured_body)); |
| 5336 | first_step_node->AddAlternative(GuardedAlternative(new (zone) TextNode( |
| 5337 | new (zone) RegExpCharacterClass('*', RegExpFlags()), |
| 5338 | /*read_backwards=*/false, loop_node))); |
| 5339 | node = first_step_node; |
| 5340 | } else { |
| 5341 | node = loop_node; |
| 5342 | } |
| 5343 | } |
| 5344 | if (is_one_byte) { |
| 5345 | node = node->FilterOneByte(RegExpCompiler::kMaxRecursion); |
| 5346 | // Do it again to propagate the new nodes to places where they were not |
| 5347 | // put because they had not been calculated yet. |
| 5348 | if (node != NULL) { |
| 5349 | node = node->FilterOneByte(RegExpCompiler::kMaxRecursion); |
| 5350 | } |
| 5351 | } else if (is_unicode && (is_global || is_sticky)) { |
| 5352 | node = OptionallyStepBackToLeadSurrogate(&compiler, node, regexp.flags()); |
| 5353 | } |
| 5354 | |
| 5355 | if (node == NULL) node = new (zone) EndNode(EndNode::BACKTRACK, zone); |
| 5356 | data->node = node; |
| 5357 | Analysis analysis(is_one_byte); |
| 5358 | analysis.EnsureAnalyzed(node); |
| 5359 | if (analysis.has_failed()) { |
| 5360 | const char* error_message = analysis.error_message(); |
| 5361 | return CompilationResult(error_message); |
| 5362 | } |
| 5363 | |
| 5364 | // Native regexp implementation. |
| 5365 | |
| 5366 | IRRegExpMacroAssembler* macro_assembler = new (zone) |
| 5367 | IRRegExpMacroAssembler(specialization_cid, data->capture_count, |
| 5368 | parsed_function, ic_data_array, osr_id, zone); |
| 5369 | |
| 5370 | // Inserted here, instead of in Assembler, because it depends on information |
| 5371 | // in the AST that isn't replicated in the Node structure. |
| 5372 | static const intptr_t kMaxBacksearchLimit = 1024; |
| 5373 | if (is_end_anchored && !is_start_anchored && !is_sticky && |
| 5374 | max_length < kMaxBacksearchLimit) { |
| 5375 | macro_assembler->SetCurrentPositionFromEnd(max_length); |
| 5376 | } |
| 5377 | |
| 5378 | if (is_global) { |
| 5379 | RegExpMacroAssembler::GlobalMode mode = RegExpMacroAssembler::GLOBAL; |
| 5380 | if (data->tree->min_match() > 0) { |
| 5381 | mode = RegExpMacroAssembler::GLOBAL_NO_ZERO_LENGTH_CHECK; |
| 5382 | } else if (is_unicode) { |
| 5383 | mode = RegExpMacroAssembler::GLOBAL_UNICODE; |
| 5384 | } |
| 5385 | macro_assembler->set_global_mode(mode); |
| 5386 | } |
| 5387 | |
| 5388 | RegExpEngine::CompilationResult result = |
| 5389 | compiler.Assemble(macro_assembler, node, data->capture_count, pattern); |
| 5390 | |
| 5391 | if (FLAG_trace_irregexp) { |
| 5392 | macro_assembler->PrintBlocks(); |
| 5393 | } |
| 5394 | |
| 5395 | return result; |
| 5396 | } |
| 5397 | #endif // !defined(DART_PRECOMPILED_RUNTIME) |
| 5398 | |
| 5399 | RegExpEngine::CompilationResult RegExpEngine::CompileBytecode( |
| 5400 | RegExpCompileData* data, |
| 5401 | const RegExp& regexp, |
| 5402 | bool is_one_byte, |
| 5403 | bool is_sticky, |
| 5404 | Zone* zone) { |
| 5405 | ASSERT(FLAG_interpret_irregexp); |
| 5406 | const String& pattern = String::Handle(zone, regexp.pattern()); |
| 5407 | |
| 5408 | ASSERT(!regexp.IsNull()); |
| 5409 | ASSERT(!pattern.IsNull()); |
| 5410 | |
| 5411 | const bool is_global = regexp.flags().IsGlobal(); |
| 5412 | const bool is_unicode = regexp.flags().IsUnicode(); |
| 5413 | |
| 5414 | RegExpCompiler compiler(data->capture_count, is_one_byte); |
| 5415 | |
| 5416 | // TODO(zerny): Frequency sampling is currently disabled because of several |
| 5417 | // issues. We do not want to store subject strings in the regexp object since |
| 5418 | // they might be long and we should not prevent their garbage collection. |
| 5419 | // Passing them to this function explicitly does not help, since we must |
| 5420 | // generate exactly the same IR for both the unoptimizing and optimizing |
| 5421 | // pipelines (otherwise it gets confused when i.e. deopt id's differ). |
| 5422 | // An option would be to store sampling results in the regexp object, but |
| 5423 | // I'm not sure the performance gains are relevant enough. |
| 5424 | |
| 5425 | // Wrap the body of the regexp in capture #0. |
| 5426 | RegExpNode* captured_body = |
| 5427 | RegExpCapture::ToNode(data->tree, 0, &compiler, compiler.accept()); |
| 5428 | |
| 5429 | RegExpNode* node = captured_body; |
| 5430 | bool is_end_anchored = data->tree->IsAnchoredAtEnd(); |
| 5431 | bool is_start_anchored = data->tree->IsAnchoredAtStart(); |
| 5432 | intptr_t max_length = data->tree->max_match(); |
| 5433 | if (!is_start_anchored && !is_sticky) { |
| 5434 | // Add a .*? at the beginning, outside the body capture, unless |
| 5435 | // this expression is anchored at the beginning. |
| 5436 | RegExpNode* loop_node = RegExpQuantifier::ToNode( |
| 5437 | 0, RegExpTree::kInfinity, false, |
| 5438 | new (zone) RegExpCharacterClass('*', RegExpFlags()), &compiler, |
| 5439 | captured_body, data->contains_anchor); |
| 5440 | |
| 5441 | if (data->contains_anchor) { |
| 5442 | // Unroll loop once, to take care of the case that might start |
| 5443 | // at the start of input. |
| 5444 | ChoiceNode* first_step_node = new (zone) ChoiceNode(2, zone); |
| 5445 | first_step_node->AddAlternative(GuardedAlternative(captured_body)); |
| 5446 | first_step_node->AddAlternative(GuardedAlternative(new (zone) TextNode( |
| 5447 | new (zone) RegExpCharacterClass('*', RegExpFlags()), |
| 5448 | /*read_backwards=*/false, loop_node))); |
| 5449 | node = first_step_node; |
| 5450 | } else { |
| 5451 | node = loop_node; |
| 5452 | } |
| 5453 | } |
| 5454 | if (is_one_byte) { |
| 5455 | node = node->FilterOneByte(RegExpCompiler::kMaxRecursion); |
| 5456 | // Do it again to propagate the new nodes to places where they were not |
| 5457 | // put because they had not been calculated yet. |
| 5458 | if (node != NULL) { |
| 5459 | node = node->FilterOneByte(RegExpCompiler::kMaxRecursion); |
| 5460 | } |
| 5461 | } else if (is_unicode && (is_global || is_sticky)) { |
| 5462 | node = OptionallyStepBackToLeadSurrogate(&compiler, node, regexp.flags()); |
| 5463 | } |
| 5464 | |
| 5465 | if (node == NULL) node = new (zone) EndNode(EndNode::BACKTRACK, zone); |
| 5466 | data->node = node; |
| 5467 | Analysis analysis(is_one_byte); |
| 5468 | analysis.EnsureAnalyzed(node); |
| 5469 | if (analysis.has_failed()) { |
| 5470 | const char* error_message = analysis.error_message(); |
| 5471 | return CompilationResult(error_message); |
| 5472 | } |
| 5473 | |
| 5474 | // Bytecode regexp implementation. |
| 5475 | |
| 5476 | ZoneGrowableArray<uint8_t> buffer(zone, 1024); |
| 5477 | BytecodeRegExpMacroAssembler* macro_assembler = |
| 5478 | new (zone) BytecodeRegExpMacroAssembler(&buffer, zone); |
| 5479 | |
| 5480 | // Inserted here, instead of in Assembler, because it depends on information |
| 5481 | // in the AST that isn't replicated in the Node structure. |
| 5482 | static const intptr_t kMaxBacksearchLimit = 1024; |
| 5483 | if (is_end_anchored && !is_start_anchored && !is_sticky && |
| 5484 | max_length < kMaxBacksearchLimit) { |
| 5485 | macro_assembler->SetCurrentPositionFromEnd(max_length); |
| 5486 | } |
| 5487 | |
| 5488 | if (is_global) { |
| 5489 | RegExpMacroAssembler::GlobalMode mode = RegExpMacroAssembler::GLOBAL; |
| 5490 | if (data->tree->min_match() > 0) { |
| 5491 | mode = RegExpMacroAssembler::GLOBAL_NO_ZERO_LENGTH_CHECK; |
| 5492 | } else if (is_unicode) { |
| 5493 | mode = RegExpMacroAssembler::GLOBAL_UNICODE; |
| 5494 | } |
| 5495 | macro_assembler->set_global_mode(mode); |
| 5496 | } |
| 5497 | |
| 5498 | RegExpEngine::CompilationResult result = |
| 5499 | compiler.Assemble(macro_assembler, node, data->capture_count, pattern); |
| 5500 | |
| 5501 | if (FLAG_trace_irregexp) { |
| 5502 | macro_assembler->PrintBlocks(); |
| 5503 | } |
| 5504 | |
| 5505 | return result; |
| 5506 | } |
| 5507 | |
| 5508 | static void CreateSpecializedFunction(Thread* thread, |
| 5509 | Zone* zone, |
| 5510 | const RegExp& regexp, |
| 5511 | intptr_t specialization_cid, |
| 5512 | bool sticky, |
| 5513 | const Object& owner) { |
| 5514 | const intptr_t kParamCount = RegExpMacroAssembler::kParamCount; |
| 5515 | |
| 5516 | Function& fn = |
| 5517 | Function::Handle(zone, Function::New(Symbols::ColonMatcher(), |
| 5518 | FunctionLayout::kIrregexpFunction, |
| 5519 | true, // Static. |
| 5520 | false, // Not const. |
| 5521 | false, // Not abstract. |
| 5522 | false, // Not external. |
| 5523 | false, // Not native. |
| 5524 | owner, TokenPosition::kMinSource)); |
| 5525 | |
| 5526 | // TODO(zerny): Share these arrays between all irregexp functions. |
| 5527 | fn.set_num_fixed_parameters(kParamCount); |
| 5528 | fn.set_parameter_types( |
| 5529 | Array::Handle(zone, Array::New(kParamCount, Heap::kOld))); |
| 5530 | fn.set_parameter_names( |
| 5531 | Array::Handle(zone, Array::New(kParamCount, Heap::kOld))); |
| 5532 | fn.SetParameterTypeAt(RegExpMacroAssembler::kParamRegExpIndex, |
| 5533 | Object::dynamic_type()); |
| 5534 | fn.SetParameterNameAt(RegExpMacroAssembler::kParamRegExpIndex, |
| 5535 | Symbols::This()); |
| 5536 | fn.SetParameterTypeAt(RegExpMacroAssembler::kParamStringIndex, |
| 5537 | Object::dynamic_type()); |
| 5538 | fn.SetParameterNameAt(RegExpMacroAssembler::kParamStringIndex, |
| 5539 | Symbols::string_param()); |
| 5540 | fn.SetParameterTypeAt(RegExpMacroAssembler::kParamStartOffsetIndex, |
| 5541 | Object::dynamic_type()); |
| 5542 | fn.SetParameterNameAt(RegExpMacroAssembler::kParamStartOffsetIndex, |
| 5543 | Symbols::start_index_param()); |
| 5544 | fn.set_result_type(Type::Handle(zone, Type::ArrayType())); |
| 5545 | |
| 5546 | // Cache the result. |
| 5547 | regexp.set_function(specialization_cid, sticky, fn); |
| 5548 | |
| 5549 | fn.SetRegExpData(regexp, specialization_cid, sticky); |
| 5550 | fn.set_is_debuggable(false); |
| 5551 | |
| 5552 | // The function is compiled lazily during the first call. |
| 5553 | } |
| 5554 | |
| 5555 | RegExpPtr RegExpEngine::CreateRegExp(Thread* thread, |
| 5556 | const String& pattern, |
| 5557 | RegExpFlags flags) { |
| 5558 | Zone* zone = thread->zone(); |
| 5559 | const RegExp& regexp = RegExp::Handle(RegExp::New()); |
| 5560 | |
| 5561 | regexp.set_pattern(pattern); |
| 5562 | regexp.set_flags(flags); |
| 5563 | |
| 5564 | // TODO(zerny): We might want to use normal string searching algorithms |
| 5565 | // for simple patterns. |
| 5566 | regexp.set_is_complex(); |
| 5567 | regexp.set_is_global(); // All dart regexps are global. |
| 5568 | |
| 5569 | if (!FLAG_interpret_irregexp) { |
| 5570 | const Library& lib = Library::Handle(zone, Library::CoreLibrary()); |
| 5571 | const Class& owner = |
| 5572 | Class::Handle(zone, lib.LookupClass(Symbols::RegExp())); |
| 5573 | |
| 5574 | for (intptr_t cid = kOneByteStringCid; cid <= kExternalTwoByteStringCid; |
| 5575 | cid++) { |
| 5576 | CreateSpecializedFunction(thread, zone, regexp, cid, /*sticky=*/false, |
| 5577 | owner); |
| 5578 | CreateSpecializedFunction(thread, zone, regexp, cid, /*sticky=*/true, |
| 5579 | owner); |
| 5580 | } |
| 5581 | } |
| 5582 | |
| 5583 | return regexp.raw(); |
| 5584 | } |
| 5585 | |
| 5586 | } // namespace dart |
| 5587 |
Generated on 2020-Aug-16 from project engine revision 1.21
Powered by 
Code Browser 2.1
Generator usage only permitted with license.