| OLD | NEW |
| 1 // Copyright (c) 2015, the Dart project authors. Please see the AUTHORS file | 1 // Copyright (c) 2015, the Dart project authors. Please see the AUTHORS file |
| 2 // for details. All rights reserved. Use of this source code is governed by a | 2 // for details. All rights reserved. Use of this source code is governed by a |
| 3 // BSD-style license that can be found in the LICENSE file. | 3 // BSD-style license that can be found in the LICENSE file. |
| 4 | 4 |
| 5 // A simple interpreter for the Irregexp byte code. | 5 // A simple interpreter for the Irregexp byte code. |
| 6 | 6 |
| 7 #include "vm/regexp_interpreter.h" | 7 #include "vm/regexp_interpreter.h" |
| 8 | 8 |
| 9 #include "vm/regexp_bytecodes.h" | 9 #include "vm/regexp_bytecodes.h" |
| 10 #include "vm/regexp_assembler.h" | 10 #include "vm/regexp_assembler.h" |
| 11 #include "vm/object.h" | 11 #include "vm/object.h" |
| 12 #include "vm/unicode.h" | 12 #include "vm/unicode.h" |
| 13 #include "vm/unibrow.h" | 13 #include "vm/unibrow.h" |
| 14 #include "vm/unibrow-inl.h" | 14 #include "vm/unibrow-inl.h" |
| 15 | 15 |
| 16 namespace dart { | 16 namespace dart { |
| 17 | 17 |
| 18 DEFINE_FLAG(bool, trace_regexp_bytecodes, false, "trace_regexp_bytecodes"); | 18 DEFINE_FLAG(bool, trace_regexp_bytecodes, false, "trace_regexp_bytecodes"); |
| 19 | 19 |
| 20 typedef unibrow::Mapping<unibrow::Ecma262Canonicalize> Canonicalize; | 20 typedef unibrow::Mapping<unibrow::Ecma262Canonicalize> Canonicalize; |
| 21 | 21 |
| 22 template<typename Char> | 22 template <typename Char> |
| 23 static bool BackRefMatchesNoCase(Canonicalize* interp_canonicalize, | 23 static bool BackRefMatchesNoCase(Canonicalize* interp_canonicalize, |
| 24 intptr_t from, | 24 intptr_t from, |
| 25 intptr_t current, | 25 intptr_t current, |
| 26 intptr_t len, | 26 intptr_t len, |
| 27 const String& subject); | 27 const String& subject); |
| 28 | 28 |
| 29 template <> | 29 template <> |
| 30 bool BackRefMatchesNoCase<uint16_t>(Canonicalize* interp_canonicalize, | 30 bool BackRefMatchesNoCase<uint16_t>(Canonicalize* interp_canonicalize, |
| 31 intptr_t from, | 31 intptr_t from, |
| 32 intptr_t current, | 32 intptr_t current, |
| 33 intptr_t len, | 33 intptr_t len, |
| 34 const String& subject) { | 34 const String& subject) { |
| 35 for (int i = 0; i < len; i++) { | 35 for (int i = 0; i < len; i++) { |
| 36 int32_t old_char = subject.CharAt(from++); | 36 int32_t old_char = subject.CharAt(from++); |
| 37 int32_t new_char = subject.CharAt(current++); | 37 int32_t new_char = subject.CharAt(current++); |
| 38 if (old_char == new_char) continue; | 38 if (old_char == new_char) continue; |
| 39 int32_t old_string[1] = { old_char }; | 39 int32_t old_string[1] = {old_char}; |
| 40 int32_t new_string[1] = { new_char }; | 40 int32_t new_string[1] = {new_char}; |
| 41 interp_canonicalize->get(old_char, '\0', old_string); | 41 interp_canonicalize->get(old_char, '\0', old_string); |
| 42 interp_canonicalize->get(new_char, '\0', new_string); | 42 interp_canonicalize->get(new_char, '\0', new_string); |
| 43 if (old_string[0] != new_string[0]) { | 43 if (old_string[0] != new_string[0]) { |
| 44 return false; | 44 return false; |
| 45 } | 45 } |
| 46 } | 46 } |
| 47 return true; | 47 return true; |
| 48 } | 48 } |
| 49 | 49 |
| 50 template <> | 50 template <> |
| (...skipping 24 matching lines...) Expand all Loading... |
| 75 static void TraceInterpreter(const uint8_t* code_base, | 75 static void TraceInterpreter(const uint8_t* code_base, |
| 76 const uint8_t* pc, | 76 const uint8_t* pc, |
| 77 int stack_depth, | 77 int stack_depth, |
| 78 int current_position, | 78 int current_position, |
| 79 uint32_t current_char, | 79 uint32_t current_char, |
| 80 int bytecode_length, | 80 int bytecode_length, |
| 81 const char* bytecode_name) { | 81 const char* bytecode_name) { |
| 82 if (FLAG_trace_regexp_bytecodes) { | 82 if (FLAG_trace_regexp_bytecodes) { |
| 83 bool printable = (current_char < 127 && current_char >= 32); | 83 bool printable = (current_char < 127 && current_char >= 32); |
| 84 const char* format = | 84 const char* format = |
| 85 printable ? | 85 printable |
| 86 "pc = %02x, sp = %d, curpos = %d, curchar = %08x (%c), bc = %s" : | 86 ? "pc = %02x, sp = %d, curpos = %d, curchar = %08x (%c), bc = %s" |
| 87 "pc = %02x, sp = %d, curpos = %d, curchar = %08x .%c., bc = %s"; | 87 : "pc = %02x, sp = %d, curpos = %d, curchar = %08x .%c., bc = %s"; |
| 88 OS::Print(format, | 88 OS::Print(format, pc - code_base, stack_depth, current_position, |
| 89 pc - code_base, | 89 current_char, printable ? current_char : '.', bytecode_name); |
| 90 stack_depth, | |
| 91 current_position, | |
| 92 current_char, | |
| 93 printable ? current_char : '.', | |
| 94 bytecode_name); | |
| 95 for (int i = 0; i < bytecode_length; i++) { | 90 for (int i = 0; i < bytecode_length; i++) { |
| 96 OS::Print(", %02x", pc[i]); | 91 OS::Print(", %02x", pc[i]); |
| 97 } | 92 } |
| 98 OS::Print(" "); | 93 OS::Print(" "); |
| 99 for (int i = 1; i < bytecode_length; i++) { | 94 for (int i = 1; i < bytecode_length; i++) { |
| 100 unsigned char b = pc[i]; | 95 unsigned char b = pc[i]; |
| 101 if (b < 127 && b >= 32) { | 96 if (b < 127 && b >= 32) { |
| 102 OS::Print("%c", b); | 97 OS::Print("%c", b); |
| 103 } else { | 98 } else { |
| 104 OS::Print("."); | 99 OS::Print("."); |
| 105 } | 100 } |
| 106 } | 101 } |
| 107 OS::Print("\n"); | 102 OS::Print("\n"); |
| 108 } | 103 } |
| 109 } | 104 } |
| 110 | 105 |
| 111 | 106 |
| 112 #define BYTECODE(name) \ | 107 #define BYTECODE(name) \ |
| 113 case BC_##name: \ | 108 case BC_##name: \ |
| 114 TraceInterpreter(code_base, \ | 109 TraceInterpreter(code_base, pc, \ |
| 115 pc, \ | 110 static_cast<int>(backtrack_sp - backtrack_stack_base), \ |
| 116 static_cast<int>(backtrack_sp - backtrack_stack_base), \ | 111 current, current_char, BC_##name##_LENGTH, #name); |
| 117 current, \ | |
| 118 current_char, \ | |
| 119 BC_##name##_LENGTH, \ | |
| 120 #name); | |
| 121 #else | 112 #else |
| 122 #define BYTECODE(name) \ | 113 #define BYTECODE(name) case BC_##name: |
| 123 case BC_##name: | |
| 124 #endif | 114 #endif |
| 125 | 115 |
| 126 | 116 |
| 127 static int32_t Load32Aligned(const uint8_t* pc) { | 117 static int32_t Load32Aligned(const uint8_t* pc) { |
| 128 ASSERT((reinterpret_cast<intptr_t>(pc) & 3) == 0); | 118 ASSERT((reinterpret_cast<intptr_t>(pc) & 3) == 0); |
| 129 return *reinterpret_cast<const int32_t *>(pc); | 119 return *reinterpret_cast<const int32_t*>(pc); |
| 130 } | 120 } |
| 131 | 121 |
| 132 | 122 |
| 133 static int32_t Load16Aligned(const uint8_t* pc) { | 123 static int32_t Load16Aligned(const uint8_t* pc) { |
| 134 ASSERT((reinterpret_cast<intptr_t>(pc) & 1) == 0); | 124 ASSERT((reinterpret_cast<intptr_t>(pc) & 1) == 0); |
| 135 return *reinterpret_cast<const uint16_t *>(pc); | 125 return *reinterpret_cast<const uint16_t*>(pc); |
| 136 } | 126 } |
| 137 | 127 |
| 138 | 128 |
| 139 // A simple abstraction over the backtracking stack used by the interpreter. | 129 // A simple abstraction over the backtracking stack used by the interpreter. |
| 140 // This backtracking stack does not grow automatically, but it ensures that the | 130 // This backtracking stack does not grow automatically, but it ensures that the |
| 141 // the memory held by the stack is released or remembered in a cache if the | 131 // the memory held by the stack is released or remembered in a cache if the |
| 142 // matching terminates. | 132 // matching terminates. |
| 143 class BacktrackStack { | 133 class BacktrackStack { |
| 144 public: | 134 public: |
| 145 explicit BacktrackStack(Zone* zone) { | 135 explicit BacktrackStack(Zone* zone) { |
| (...skipping 37 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 183 | 173 |
| 184 #ifdef DEBUG | 174 #ifdef DEBUG |
| 185 if (FLAG_trace_regexp_bytecodes) { | 175 if (FLAG_trace_regexp_bytecodes) { |
| 186 OS::Print("Start irregexp bytecode interpreter\n"); | 176 OS::Print("Start irregexp bytecode interpreter\n"); |
| 187 } | 177 } |
| 188 #endif | 178 #endif |
| 189 while (true) { | 179 while (true) { |
| 190 int32_t insn = Load32Aligned(pc); | 180 int32_t insn = Load32Aligned(pc); |
| 191 switch (insn & BYTECODE_MASK) { | 181 switch (insn & BYTECODE_MASK) { |
| 192 BYTECODE(BREAK) | 182 BYTECODE(BREAK) |
| 193 UNREACHABLE(); | 183 UNREACHABLE(); |
| 194 return IrregexpInterpreter::RE_FAILURE; | 184 return IrregexpInterpreter::RE_FAILURE; |
| 195 BYTECODE(PUSH_CP) | 185 BYTECODE(PUSH_CP) |
| 196 if (--backtrack_stack_space < 0) { | 186 if (--backtrack_stack_space < 0) { |
| 197 return IrregexpInterpreter::RE_EXCEPTION; | 187 return IrregexpInterpreter::RE_EXCEPTION; |
| 198 } | 188 } |
| 199 *backtrack_sp++ = current; | 189 *backtrack_sp++ = current; |
| 200 pc += BC_PUSH_CP_LENGTH; | 190 pc += BC_PUSH_CP_LENGTH; |
| 201 break; | 191 break; |
| 202 BYTECODE(PUSH_BT) | 192 BYTECODE(PUSH_BT) |
| 203 if (--backtrack_stack_space < 0) { | 193 if (--backtrack_stack_space < 0) { |
| 204 return IrregexpInterpreter::RE_EXCEPTION; | 194 return IrregexpInterpreter::RE_EXCEPTION; |
| 205 } | 195 } |
| 206 *backtrack_sp++ = Load32Aligned(pc + 4); | 196 *backtrack_sp++ = Load32Aligned(pc + 4); |
| 207 pc += BC_PUSH_BT_LENGTH; | 197 pc += BC_PUSH_BT_LENGTH; |
| 208 break; | 198 break; |
| 209 BYTECODE(PUSH_REGISTER) | 199 BYTECODE(PUSH_REGISTER) |
| 210 if (--backtrack_stack_space < 0) { | 200 if (--backtrack_stack_space < 0) { |
| 211 return IrregexpInterpreter::RE_EXCEPTION; | 201 return IrregexpInterpreter::RE_EXCEPTION; |
| 212 } | 202 } |
| 213 *backtrack_sp++ = registers[insn >> BYTECODE_SHIFT]; | 203 *backtrack_sp++ = registers[insn >> BYTECODE_SHIFT]; |
| 214 pc += BC_PUSH_REGISTER_LENGTH; | 204 pc += BC_PUSH_REGISTER_LENGTH; |
| 215 break; | 205 break; |
| 216 BYTECODE(SET_REGISTER) | 206 BYTECODE(SET_REGISTER) |
| 217 registers[insn >> BYTECODE_SHIFT] = Load32Aligned(pc + 4); | 207 registers[insn >> BYTECODE_SHIFT] = Load32Aligned(pc + 4); |
| 218 pc += BC_SET_REGISTER_LENGTH; | 208 pc += BC_SET_REGISTER_LENGTH; |
| 219 break; | 209 break; |
| 220 BYTECODE(ADVANCE_REGISTER) | 210 BYTECODE(ADVANCE_REGISTER) |
| 221 registers[insn >> BYTECODE_SHIFT] += Load32Aligned(pc + 4); | 211 registers[insn >> BYTECODE_SHIFT] += Load32Aligned(pc + 4); |
| 222 pc += BC_ADVANCE_REGISTER_LENGTH; | 212 pc += BC_ADVANCE_REGISTER_LENGTH; |
| 223 break; | 213 break; |
| 224 BYTECODE(SET_REGISTER_TO_CP) | 214 BYTECODE(SET_REGISTER_TO_CP) |
| 225 registers[insn >> BYTECODE_SHIFT] = current + Load32Aligned(pc + 4); | 215 registers[insn >> BYTECODE_SHIFT] = current + Load32Aligned(pc + 4); |
| 226 pc += BC_SET_REGISTER_TO_CP_LENGTH; | 216 pc += BC_SET_REGISTER_TO_CP_LENGTH; |
| 227 break; | 217 break; |
| 228 BYTECODE(SET_CP_TO_REGISTER) | 218 BYTECODE(SET_CP_TO_REGISTER) |
| 229 current = registers[insn >> BYTECODE_SHIFT]; | 219 current = registers[insn >> BYTECODE_SHIFT]; |
| 230 pc += BC_SET_CP_TO_REGISTER_LENGTH; | 220 pc += BC_SET_CP_TO_REGISTER_LENGTH; |
| 231 break; | 221 break; |
| 232 BYTECODE(SET_REGISTER_TO_SP) | 222 BYTECODE(SET_REGISTER_TO_SP) |
| 233 registers[insn >> BYTECODE_SHIFT] = | 223 registers[insn >> BYTECODE_SHIFT] = |
| 234 static_cast<int>(backtrack_sp - backtrack_stack_base); | 224 static_cast<int>(backtrack_sp - backtrack_stack_base); |
| 235 pc += BC_SET_REGISTER_TO_SP_LENGTH; | 225 pc += BC_SET_REGISTER_TO_SP_LENGTH; |
| 236 break; | 226 break; |
| 237 BYTECODE(SET_SP_TO_REGISTER) | 227 BYTECODE(SET_SP_TO_REGISTER) |
| 238 backtrack_sp = backtrack_stack_base + registers[insn >> BYTECODE_SHIFT]; | 228 backtrack_sp = backtrack_stack_base + registers[insn >> BYTECODE_SHIFT]; |
| 239 backtrack_stack_space = backtrack_stack.max_size() - | 229 backtrack_stack_space = |
| 240 static_cast<int>(backtrack_sp - backtrack_stack_base); | 230 backtrack_stack.max_size() - |
| 241 pc += BC_SET_SP_TO_REGISTER_LENGTH; | 231 static_cast<int>(backtrack_sp - backtrack_stack_base); |
| 242 break; | 232 pc += BC_SET_SP_TO_REGISTER_LENGTH; |
| 233 break; |
| 243 BYTECODE(POP_CP) | 234 BYTECODE(POP_CP) |
| 235 backtrack_stack_space++; |
| 236 --backtrack_sp; |
| 237 current = *backtrack_sp; |
| 238 pc += BC_POP_CP_LENGTH; |
| 239 break; |
| 240 BYTECODE(POP_BT) |
| 241 backtrack_stack_space++; |
| 242 --backtrack_sp; |
| 243 pc = code_base + *backtrack_sp; |
| 244 break; |
| 245 BYTECODE(POP_REGISTER) |
| 246 backtrack_stack_space++; |
| 247 --backtrack_sp; |
| 248 registers[insn >> BYTECODE_SHIFT] = *backtrack_sp; |
| 249 pc += BC_POP_REGISTER_LENGTH; |
| 250 break; |
| 251 BYTECODE(FAIL) |
| 252 return IrregexpInterpreter::RE_FAILURE; |
| 253 BYTECODE(SUCCEED) |
| 254 return IrregexpInterpreter::RE_SUCCESS; |
| 255 BYTECODE(ADVANCE_CP) |
| 256 current += insn >> BYTECODE_SHIFT; |
| 257 pc += BC_ADVANCE_CP_LENGTH; |
| 258 break; |
| 259 BYTECODE(GOTO) |
| 260 pc = code_base + Load32Aligned(pc + 4); |
| 261 break; |
| 262 BYTECODE(ADVANCE_CP_AND_GOTO) |
| 263 current += insn >> BYTECODE_SHIFT; |
| 264 pc = code_base + Load32Aligned(pc + 4); |
| 265 break; |
| 266 BYTECODE(CHECK_GREEDY) |
| 267 if (current == backtrack_sp[-1]) { |
| 268 backtrack_sp--; |
| 244 backtrack_stack_space++; | 269 backtrack_stack_space++; |
| 245 --backtrack_sp; | |
| 246 current = *backtrack_sp; | |
| 247 pc += BC_POP_CP_LENGTH; | |
| 248 break; | |
| 249 BYTECODE(POP_BT) | |
| 250 backtrack_stack_space++; | |
| 251 --backtrack_sp; | |
| 252 pc = code_base + *backtrack_sp; | |
| 253 break; | |
| 254 BYTECODE(POP_REGISTER) | |
| 255 backtrack_stack_space++; | |
| 256 --backtrack_sp; | |
| 257 registers[insn >> BYTECODE_SHIFT] = *backtrack_sp; | |
| 258 pc += BC_POP_REGISTER_LENGTH; | |
| 259 break; | |
| 260 BYTECODE(FAIL) | |
| 261 return IrregexpInterpreter::RE_FAILURE; | |
| 262 BYTECODE(SUCCEED) | |
| 263 return IrregexpInterpreter::RE_SUCCESS; | |
| 264 BYTECODE(ADVANCE_CP) | |
| 265 current += insn >> BYTECODE_SHIFT; | |
| 266 pc += BC_ADVANCE_CP_LENGTH; | |
| 267 break; | |
| 268 BYTECODE(GOTO) | |
| 269 pc = code_base + Load32Aligned(pc + 4); | 270 pc = code_base + Load32Aligned(pc + 4); |
| 270 break; | 271 } else { |
| 271 BYTECODE(ADVANCE_CP_AND_GOTO) | 272 pc += BC_CHECK_GREEDY_LENGTH; |
| 272 current += insn >> BYTECODE_SHIFT; | 273 } |
| 273 pc = code_base + Load32Aligned(pc + 4); | 274 break; |
| 274 break; | |
| 275 BYTECODE(CHECK_GREEDY) | |
| 276 if (current == backtrack_sp[-1]) { | |
| 277 backtrack_sp--; | |
| 278 backtrack_stack_space++; | |
| 279 pc = code_base + Load32Aligned(pc + 4); | |
| 280 } else { | |
| 281 pc += BC_CHECK_GREEDY_LENGTH; | |
| 282 } | |
| 283 break; | |
| 284 BYTECODE(LOAD_CURRENT_CHAR) { | 275 BYTECODE(LOAD_CURRENT_CHAR) { |
| 285 int pos = current + (insn >> BYTECODE_SHIFT); | 276 int pos = current + (insn >> BYTECODE_SHIFT); |
| 286 if (pos >= subject_length) { | 277 if (pos >= subject_length) { |
| 287 pc = code_base + Load32Aligned(pc + 4); | 278 pc = code_base + Load32Aligned(pc + 4); |
| 288 } else { | 279 } else { |
| 289 current_char = subject.CharAt(pos); | 280 current_char = subject.CharAt(pos); |
| 290 pc += BC_LOAD_CURRENT_CHAR_LENGTH; | 281 pc += BC_LOAD_CURRENT_CHAR_LENGTH; |
| 291 } | 282 } |
| 292 break; | 283 break; |
| 293 } | 284 } |
| 294 BYTECODE(LOAD_CURRENT_CHAR_UNCHECKED) { | 285 BYTECODE(LOAD_CURRENT_CHAR_UNCHECKED) { |
| 295 int pos = current + (insn >> BYTECODE_SHIFT); | 286 int pos = current + (insn >> BYTECODE_SHIFT); |
| 296 current_char = subject.CharAt(pos); | 287 current_char = subject.CharAt(pos); |
| 297 pc += BC_LOAD_CURRENT_CHAR_UNCHECKED_LENGTH; | 288 pc += BC_LOAD_CURRENT_CHAR_UNCHECKED_LENGTH; |
| 298 break; | 289 break; |
| 299 } | 290 } |
| 300 BYTECODE(LOAD_2_CURRENT_CHARS) { | 291 BYTECODE(LOAD_2_CURRENT_CHARS) { |
| 301 int pos = current + (insn >> BYTECODE_SHIFT); | 292 int pos = current + (insn >> BYTECODE_SHIFT); |
| 302 if (pos + 2 > subject_length) { | 293 if (pos + 2 > subject_length) { |
| 303 pc = code_base + Load32Aligned(pc + 4); | 294 pc = code_base + Load32Aligned(pc + 4); |
| 304 } else { | 295 } else { |
| 305 Char next = subject.CharAt(pos + 1); | 296 Char next = subject.CharAt(pos + 1); |
| 306 current_char = subject.CharAt(pos) | | 297 current_char = |
| 307 (next << (kBitsPerByte * sizeof(Char))); | 298 subject.CharAt(pos) | (next << (kBitsPerByte * sizeof(Char))); |
| 308 pc += BC_LOAD_2_CURRENT_CHARS_LENGTH; | 299 pc += BC_LOAD_2_CURRENT_CHARS_LENGTH; |
| 309 } | 300 } |
| 310 break; | 301 break; |
| 311 } | 302 } |
| 312 BYTECODE(LOAD_2_CURRENT_CHARS_UNCHECKED) { | 303 BYTECODE(LOAD_2_CURRENT_CHARS_UNCHECKED) { |
| 313 int pos = current + (insn >> BYTECODE_SHIFT); | 304 int pos = current + (insn >> BYTECODE_SHIFT); |
| 314 Char next = subject.CharAt(pos + 1); | 305 Char next = subject.CharAt(pos + 1); |
| 315 current_char = subject.CharAt(pos) | | 306 current_char = |
| 316 (next << (kBitsPerByte * sizeof(Char))); | 307 subject.CharAt(pos) | (next << (kBitsPerByte * sizeof(Char))); |
| 317 pc += BC_LOAD_2_CURRENT_CHARS_UNCHECKED_LENGTH; | 308 pc += BC_LOAD_2_CURRENT_CHARS_UNCHECKED_LENGTH; |
| 318 break; | 309 break; |
| 319 } | 310 } |
| 320 BYTECODE(LOAD_4_CURRENT_CHARS) { | 311 BYTECODE(LOAD_4_CURRENT_CHARS) { |
| 321 ASSERT(sizeof(Char) == 1); | 312 ASSERT(sizeof(Char) == 1); |
| 322 int pos = current + (insn >> BYTECODE_SHIFT); | 313 int pos = current + (insn >> BYTECODE_SHIFT); |
| 323 if (pos + 4 > subject_length) { | 314 if (pos + 4 > subject_length) { |
| 324 pc = code_base + Load32Aligned(pc + 4); | 315 pc = code_base + Load32Aligned(pc + 4); |
| 325 } else { | 316 } else { |
| 326 Char next1 = subject.CharAt(pos + 1); | 317 Char next1 = subject.CharAt(pos + 1); |
| 327 Char next2 = subject.CharAt(pos + 2); | 318 Char next2 = subject.CharAt(pos + 2); |
| 328 Char next3 = subject.CharAt(pos + 3); | 319 Char next3 = subject.CharAt(pos + 3); |
| 329 current_char = (subject.CharAt(pos) | | 320 current_char = (subject.CharAt(pos) | (next1 << 8) | (next2 << 16) | |
| 330 (next1 << 8) | | |
| 331 (next2 << 16) | | |
| 332 (next3 << 24)); | 321 (next3 << 24)); |
| 333 pc += BC_LOAD_4_CURRENT_CHARS_LENGTH; | 322 pc += BC_LOAD_4_CURRENT_CHARS_LENGTH; |
| 334 } | 323 } |
| 335 break; | 324 break; |
| 336 } | 325 } |
| 337 BYTECODE(LOAD_4_CURRENT_CHARS_UNCHECKED) { | 326 BYTECODE(LOAD_4_CURRENT_CHARS_UNCHECKED) { |
| 338 ASSERT(sizeof(Char) == 1); | 327 ASSERT(sizeof(Char) == 1); |
| 339 int pos = current + (insn >> BYTECODE_SHIFT); | 328 int pos = current + (insn >> BYTECODE_SHIFT); |
| 340 Char next1 = subject.CharAt(pos + 1); | 329 Char next1 = subject.CharAt(pos + 1); |
| 341 Char next2 = subject.CharAt(pos + 2); | 330 Char next2 = subject.CharAt(pos + 2); |
| 342 Char next3 = subject.CharAt(pos + 3); | 331 Char next3 = subject.CharAt(pos + 3); |
| 343 current_char = (subject.CharAt(pos) | | 332 current_char = (subject.CharAt(pos) | (next1 << 8) | (next2 << 16) | |
| 344 (next1 << 8) | | |
| 345 (next2 << 16) | | |
| 346 (next3 << 24)); | 333 (next3 << 24)); |
| 347 pc += BC_LOAD_4_CURRENT_CHARS_UNCHECKED_LENGTH; | 334 pc += BC_LOAD_4_CURRENT_CHARS_UNCHECKED_LENGTH; |
| 348 break; | 335 break; |
| 349 } | 336 } |
| 350 BYTECODE(CHECK_4_CHARS) { | 337 BYTECODE(CHECK_4_CHARS) { |
| 351 uint32_t c = Load32Aligned(pc + 4); | 338 uint32_t c = Load32Aligned(pc + 4); |
| 352 if (c == current_char) { | 339 if (c == current_char) { |
| 353 pc = code_base + Load32Aligned(pc + 8); | 340 pc = code_base + Load32Aligned(pc + 8); |
| 354 } else { | 341 } else { |
| 355 pc += BC_CHECK_4_CHARS_LENGTH; | 342 pc += BC_CHECK_4_CHARS_LENGTH; |
| (...skipping 117 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 473 BYTECODE(CHECK_GT) { | 460 BYTECODE(CHECK_GT) { |
| 474 uint32_t limit = (insn >> BYTECODE_SHIFT); | 461 uint32_t limit = (insn >> BYTECODE_SHIFT); |
| 475 if (current_char > limit) { | 462 if (current_char > limit) { |
| 476 pc = code_base + Load32Aligned(pc + 4); | 463 pc = code_base + Load32Aligned(pc + 4); |
| 477 } else { | 464 } else { |
| 478 pc += BC_CHECK_GT_LENGTH; | 465 pc += BC_CHECK_GT_LENGTH; |
| 479 } | 466 } |
| 480 break; | 467 break; |
| 481 } | 468 } |
| 482 BYTECODE(CHECK_REGISTER_LT) | 469 BYTECODE(CHECK_REGISTER_LT) |
| 483 if (registers[insn >> BYTECODE_SHIFT] < Load32Aligned(pc + 4)) { | 470 if (registers[insn >> BYTECODE_SHIFT] < Load32Aligned(pc + 4)) { |
| 484 pc = code_base + Load32Aligned(pc + 8); | 471 pc = code_base + Load32Aligned(pc + 8); |
| 485 } else { | 472 } else { |
| 486 pc += BC_CHECK_REGISTER_LT_LENGTH; | 473 pc += BC_CHECK_REGISTER_LT_LENGTH; |
| 487 } | 474 } |
| 488 break; | 475 break; |
| 489 BYTECODE(CHECK_REGISTER_GE) | 476 BYTECODE(CHECK_REGISTER_GE) |
| 490 if (registers[insn >> BYTECODE_SHIFT] >= Load32Aligned(pc + 4)) { | 477 if (registers[insn >> BYTECODE_SHIFT] >= Load32Aligned(pc + 4)) { |
| 491 pc = code_base + Load32Aligned(pc + 8); | 478 pc = code_base + Load32Aligned(pc + 8); |
| 492 } else { | 479 } else { |
| 493 pc += BC_CHECK_REGISTER_GE_LENGTH; | 480 pc += BC_CHECK_REGISTER_GE_LENGTH; |
| 494 } | 481 } |
| 495 break; | 482 break; |
| 496 BYTECODE(CHECK_REGISTER_EQ_POS) | 483 BYTECODE(CHECK_REGISTER_EQ_POS) |
| 497 if (registers[insn >> BYTECODE_SHIFT] == current) { | 484 if (registers[insn >> BYTECODE_SHIFT] == current) { |
| 498 pc = code_base + Load32Aligned(pc + 4); | 485 pc = code_base + Load32Aligned(pc + 4); |
| 499 } else { | 486 } else { |
| 500 pc += BC_CHECK_REGISTER_EQ_POS_LENGTH; | 487 pc += BC_CHECK_REGISTER_EQ_POS_LENGTH; |
| 501 } | 488 } |
| 502 break; | 489 break; |
| 503 BYTECODE(CHECK_NOT_REGS_EQUAL) | 490 BYTECODE(CHECK_NOT_REGS_EQUAL) |
| 504 if (registers[insn >> BYTECODE_SHIFT] == | 491 if (registers[insn >> BYTECODE_SHIFT] == |
| 505 registers[Load32Aligned(pc + 4)]) { | 492 registers[Load32Aligned(pc + 4)]) { |
| 506 pc += BC_CHECK_NOT_REGS_EQUAL_LENGTH; | 493 pc += BC_CHECK_NOT_REGS_EQUAL_LENGTH; |
| 507 } else { | 494 } else { |
| 508 pc = code_base + Load32Aligned(pc + 8); | 495 pc = code_base + Load32Aligned(pc + 8); |
| 509 } | 496 } |
| 510 break; | 497 break; |
| 511 BYTECODE(CHECK_NOT_BACK_REF) { | 498 BYTECODE(CHECK_NOT_BACK_REF) { |
| 512 int from = registers[insn >> BYTECODE_SHIFT]; | 499 int from = registers[insn >> BYTECODE_SHIFT]; |
| 513 int len = registers[(insn >> BYTECODE_SHIFT) + 1] - from; | 500 int len = registers[(insn >> BYTECODE_SHIFT) + 1] - from; |
| 514 if (from < 0 || len <= 0) { | 501 if (from < 0 || len <= 0) { |
| 515 pc += BC_CHECK_NOT_BACK_REF_LENGTH; | 502 pc += BC_CHECK_NOT_BACK_REF_LENGTH; |
| 516 break; | 503 break; |
| 517 } | 504 } |
| 518 if (current + len > subject_length) { | 505 if (current + len > subject_length) { |
| 519 pc = code_base + Load32Aligned(pc + 4); | 506 pc = code_base + Load32Aligned(pc + 4); |
| 520 break; | 507 break; |
| (...skipping 15 matching lines...) Expand all Loading... |
| 536 int from = registers[insn >> BYTECODE_SHIFT]; | 523 int from = registers[insn >> BYTECODE_SHIFT]; |
| 537 int len = registers[(insn >> BYTECODE_SHIFT) + 1] - from; | 524 int len = registers[(insn >> BYTECODE_SHIFT) + 1] - from; |
| 538 if (from < 0 || len <= 0) { | 525 if (from < 0 || len <= 0) { |
| 539 pc += BC_CHECK_NOT_BACK_REF_NO_CASE_LENGTH; | 526 pc += BC_CHECK_NOT_BACK_REF_NO_CASE_LENGTH; |
| 540 break; | 527 break; |
| 541 } | 528 } |
| 542 if (current + len > subject_length) { | 529 if (current + len > subject_length) { |
| 543 pc = code_base + Load32Aligned(pc + 4); | 530 pc = code_base + Load32Aligned(pc + 4); |
| 544 break; | 531 break; |
| 545 } else { | 532 } else { |
| 546 if (BackRefMatchesNoCase<Char>(&canonicalize, | 533 if (BackRefMatchesNoCase<Char>(&canonicalize, from, current, len, |
| 547 from, current, len, subject)) { | 534 subject)) { |
| 548 current += len; | 535 current += len; |
| 549 pc += BC_CHECK_NOT_BACK_REF_NO_CASE_LENGTH; | 536 pc += BC_CHECK_NOT_BACK_REF_NO_CASE_LENGTH; |
| 550 } else { | 537 } else { |
| 551 pc = code_base + Load32Aligned(pc + 4); | 538 pc = code_base + Load32Aligned(pc + 4); |
| 552 } | 539 } |
| 553 } | 540 } |
| 554 break; | 541 break; |
| 555 } | 542 } |
| 556 BYTECODE(CHECK_AT_START) | 543 BYTECODE(CHECK_AT_START) |
| 557 if (current == 0) { | 544 if (current == 0) { |
| 558 pc = code_base + Load32Aligned(pc + 4); | 545 pc = code_base + Load32Aligned(pc + 4); |
| 559 } else { | 546 } else { |
| 560 pc += BC_CHECK_AT_START_LENGTH; | 547 pc += BC_CHECK_AT_START_LENGTH; |
| 561 } | 548 } |
| 562 break; | 549 break; |
| 563 BYTECODE(CHECK_NOT_AT_START) | 550 BYTECODE(CHECK_NOT_AT_START) |
| 564 if (current == 0) { | 551 if (current == 0) { |
| 565 pc += BC_CHECK_NOT_AT_START_LENGTH; | 552 pc += BC_CHECK_NOT_AT_START_LENGTH; |
| 566 } else { | 553 } else { |
| 567 pc = code_base + Load32Aligned(pc + 4); | 554 pc = code_base + Load32Aligned(pc + 4); |
| 568 } | 555 } |
| 569 break; | 556 break; |
| 570 BYTECODE(SET_CURRENT_POSITION_FROM_END) { | 557 BYTECODE(SET_CURRENT_POSITION_FROM_END) { |
| 571 int by = static_cast<uint32_t>(insn) >> BYTECODE_SHIFT; | 558 int by = static_cast<uint32_t>(insn) >> BYTECODE_SHIFT; |
| 572 if (subject_length - current > by) { | 559 if (subject_length - current > by) { |
| 573 current = subject_length - by; | 560 current = subject_length - by; |
| 574 current_char = subject.CharAt(current - 1); | 561 current_char = subject.CharAt(current - 1); |
| 575 } | 562 } |
| 576 pc += BC_SET_CURRENT_POSITION_FROM_END_LENGTH; | 563 pc += BC_SET_CURRENT_POSITION_FROM_END_LENGTH; |
| 577 break; | 564 break; |
| 578 } | 565 } |
| 579 default: | 566 default: |
| (...skipping 12 matching lines...) Expand all Loading... |
| 592 Zone* zone) { | 579 Zone* zone) { |
| 593 NoSafepointScope no_safepoint; | 580 NoSafepointScope no_safepoint; |
| 594 const uint8_t* code_base = reinterpret_cast<uint8_t*>(bytecode.DataAddr(0)); | 581 const uint8_t* code_base = reinterpret_cast<uint8_t*>(bytecode.DataAddr(0)); |
| 595 | 582 |
| 596 uint16_t previous_char = '\n'; | 583 uint16_t previous_char = '\n'; |
| 597 if (start_position != 0) { | 584 if (start_position != 0) { |
| 598 previous_char = subject.CharAt(start_position - 1); | 585 previous_char = subject.CharAt(start_position - 1); |
| 599 } | 586 } |
| 600 | 587 |
| 601 if (subject.IsOneByteString() || subject.IsExternalOneByteString()) { | 588 if (subject.IsOneByteString() || subject.IsExternalOneByteString()) { |
| 602 return RawMatch<uint8_t>(code_base, | 589 return RawMatch<uint8_t>(code_base, subject, registers, start_position, |
| 603 subject, | 590 previous_char, zone); |
| 604 registers, | |
| 605 start_position, | |
| 606 previous_char, | |
| 607 zone); | |
| 608 } else if (subject.IsTwoByteString() || subject.IsExternalTwoByteString()) { | 591 } else if (subject.IsTwoByteString() || subject.IsExternalTwoByteString()) { |
| 609 return RawMatch<uint16_t>(code_base, | 592 return RawMatch<uint16_t>(code_base, subject, registers, start_position, |
| 610 subject, | 593 previous_char, zone); |
| 611 registers, | |
| 612 start_position, | |
| 613 previous_char, | |
| 614 zone); | |
| 615 } else { | 594 } else { |
| 616 UNREACHABLE(); | 595 UNREACHABLE(); |
| 617 return IrregexpInterpreter::RE_FAILURE; | 596 return IrregexpInterpreter::RE_FAILURE; |
| 618 } | 597 } |
| 619 } | 598 } |
| 620 | 599 |
| 621 } // namespace dart | 600 } // namespace dart |
| OLD | NEW |