Home | History | Annotate | Download | only in src
      1 // Copyright 2008-2009 the V8 project authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style license that can be
      3 // found in the LICENSE file.
      4 
      5 #include "src/v8.h"
      6 
      7 #include "src/ast.h"
      8 #include "src/bytecodes-irregexp.h"
      9 #include "src/regexp-macro-assembler.h"
     10 #include "src/regexp-macro-assembler-irregexp.h"
     11 #include "src/regexp-macro-assembler-irregexp-inl.h"
     12 
     13 
     14 namespace v8 {
     15 namespace internal {
     16 
     17 #ifdef V8_INTERPRETED_REGEXP
     18 
     19 RegExpMacroAssemblerIrregexp::RegExpMacroAssemblerIrregexp(Vector<byte> buffer,
     20                                                            Zone* zone)
     21     : RegExpMacroAssembler(zone),
     22       buffer_(buffer),
     23       pc_(0),
     24       own_buffer_(false),
     25       advance_current_end_(kInvalidPC),
     26       isolate_(zone->isolate()) { }
     27 
     28 
     29 RegExpMacroAssemblerIrregexp::~RegExpMacroAssemblerIrregexp() {
     30   if (backtrack_.is_linked()) backtrack_.Unuse();
     31   if (own_buffer_) buffer_.Dispose();
     32 }
     33 
     34 
     35 RegExpMacroAssemblerIrregexp::IrregexpImplementation
     36 RegExpMacroAssemblerIrregexp::Implementation() {
     37   return kBytecodeImplementation;
     38 }
     39 
     40 
     41 void RegExpMacroAssemblerIrregexp::Bind(Label* l) {
     42   advance_current_end_ = kInvalidPC;
     43   DCHECK(!l->is_bound());
     44   if (l->is_linked()) {
     45     int pos = l->pos();
     46     while (pos != 0) {
     47       int fixup = pos;
     48       pos = *reinterpret_cast<int32_t*>(buffer_.start() + fixup);
     49       *reinterpret_cast<uint32_t*>(buffer_.start() + fixup) = pc_;
     50     }
     51   }
     52   l->bind_to(pc_);
     53 }
     54 
     55 
     56 void RegExpMacroAssemblerIrregexp::EmitOrLink(Label* l) {
     57   if (l == NULL) l = &backtrack_;
     58   if (l->is_bound()) {
     59     Emit32(l->pos());
     60   } else {
     61     int pos = 0;
     62     if (l->is_linked()) {
     63       pos = l->pos();
     64     }
     65     l->link_to(pc_);
     66     Emit32(pos);
     67   }
     68 }
     69 
     70 
     71 void RegExpMacroAssemblerIrregexp::PopRegister(int register_index) {
     72   DCHECK(register_index >= 0);
     73   DCHECK(register_index <= kMaxRegister);
     74   Emit(BC_POP_REGISTER, register_index);
     75 }
     76 
     77 
     78 void RegExpMacroAssemblerIrregexp::PushRegister(
     79     int register_index,
     80     StackCheckFlag check_stack_limit) {
     81   DCHECK(register_index >= 0);
     82   DCHECK(register_index <= kMaxRegister);
     83   Emit(BC_PUSH_REGISTER, register_index);
     84 }
     85 
     86 
     87 void RegExpMacroAssemblerIrregexp::WriteCurrentPositionToRegister(
     88     int register_index, int cp_offset) {
     89   DCHECK(register_index >= 0);
     90   DCHECK(register_index <= kMaxRegister);
     91   Emit(BC_SET_REGISTER_TO_CP, register_index);
     92   Emit32(cp_offset);  // Current position offset.
     93 }
     94 
     95 
     96 void RegExpMacroAssemblerIrregexp::ClearRegisters(int reg_from, int reg_to) {
     97   DCHECK(reg_from <= reg_to);
     98   for (int reg = reg_from; reg <= reg_to; reg++) {
     99     SetRegister(reg, -1);
    100   }
    101 }
    102 
    103 
    104 void RegExpMacroAssemblerIrregexp::ReadCurrentPositionFromRegister(
    105     int register_index) {
    106   DCHECK(register_index >= 0);
    107   DCHECK(register_index <= kMaxRegister);
    108   Emit(BC_SET_CP_TO_REGISTER, register_index);
    109 }
    110 
    111 
    112 void RegExpMacroAssemblerIrregexp::WriteStackPointerToRegister(
    113     int register_index) {
    114   DCHECK(register_index >= 0);
    115   DCHECK(register_index <= kMaxRegister);
    116   Emit(BC_SET_REGISTER_TO_SP, register_index);
    117 }
    118 
    119 
    120 void RegExpMacroAssemblerIrregexp::ReadStackPointerFromRegister(
    121     int register_index) {
    122   DCHECK(register_index >= 0);
    123   DCHECK(register_index <= kMaxRegister);
    124   Emit(BC_SET_SP_TO_REGISTER, register_index);
    125 }
    126 
    127 
    128 void RegExpMacroAssemblerIrregexp::SetCurrentPositionFromEnd(int by) {
    129   DCHECK(is_uint24(by));
    130   Emit(BC_SET_CURRENT_POSITION_FROM_END, by);
    131 }
    132 
    133 
    134 void RegExpMacroAssemblerIrregexp::SetRegister(int register_index, int to) {
    135   DCHECK(register_index >= 0);
    136   DCHECK(register_index <= kMaxRegister);
    137   Emit(BC_SET_REGISTER, register_index);
    138   Emit32(to);
    139 }
    140 
    141 
    142 void RegExpMacroAssemblerIrregexp::AdvanceRegister(int register_index, int by) {
    143   DCHECK(register_index >= 0);
    144   DCHECK(register_index <= kMaxRegister);
    145   Emit(BC_ADVANCE_REGISTER, register_index);
    146   Emit32(by);
    147 }
    148 
    149 
    150 void RegExpMacroAssemblerIrregexp::PopCurrentPosition() {
    151   Emit(BC_POP_CP, 0);
    152 }
    153 
    154 
    155 void RegExpMacroAssemblerIrregexp::PushCurrentPosition() {
    156   Emit(BC_PUSH_CP, 0);
    157 }
    158 
    159 
    160 void RegExpMacroAssemblerIrregexp::Backtrack() {
    161   Emit(BC_POP_BT, 0);
    162 }
    163 
    164 
    165 void RegExpMacroAssemblerIrregexp::GoTo(Label* l) {
    166   if (advance_current_end_ == pc_) {
    167     // Combine advance current and goto.
    168     pc_ = advance_current_start_;
    169     Emit(BC_ADVANCE_CP_AND_GOTO, advance_current_offset_);
    170     EmitOrLink(l);
    171     advance_current_end_ = kInvalidPC;
    172   } else {
    173     // Regular goto.
    174     Emit(BC_GOTO, 0);
    175     EmitOrLink(l);
    176   }
    177 }
    178 
    179 
    180 void RegExpMacroAssemblerIrregexp::PushBacktrack(Label* l) {
    181   Emit(BC_PUSH_BT, 0);
    182   EmitOrLink(l);
    183 }
    184 
    185 
    186 bool RegExpMacroAssemblerIrregexp::Succeed() {
    187   Emit(BC_SUCCEED, 0);
    188   return false;  // Restart matching for global regexp not supported.
    189 }
    190 
    191 
    192 void RegExpMacroAssemblerIrregexp::Fail() {
    193   Emit(BC_FAIL, 0);
    194 }
    195 
    196 
    197 void RegExpMacroAssemblerIrregexp::AdvanceCurrentPosition(int by) {
    198   DCHECK(by >= kMinCPOffset);
    199   DCHECK(by <= kMaxCPOffset);
    200   advance_current_start_ = pc_;
    201   advance_current_offset_ = by;
    202   Emit(BC_ADVANCE_CP, by);
    203   advance_current_end_ = pc_;
    204 }
    205 
    206 
    207 void RegExpMacroAssemblerIrregexp::CheckGreedyLoop(
    208       Label* on_tos_equals_current_position) {
    209   Emit(BC_CHECK_GREEDY, 0);
    210   EmitOrLink(on_tos_equals_current_position);
    211 }
    212 
    213 
    214 void RegExpMacroAssemblerIrregexp::LoadCurrentCharacter(int cp_offset,
    215                                                         Label* on_failure,
    216                                                         bool check_bounds,
    217                                                         int characters) {
    218   DCHECK(cp_offset >= kMinCPOffset);
    219   DCHECK(cp_offset <= kMaxCPOffset);
    220   int bytecode;
    221   if (check_bounds) {
    222     if (characters == 4) {
    223       bytecode = BC_LOAD_4_CURRENT_CHARS;
    224     } else if (characters == 2) {
    225       bytecode = BC_LOAD_2_CURRENT_CHARS;
    226     } else {
    227       DCHECK(characters == 1);
    228       bytecode = BC_LOAD_CURRENT_CHAR;
    229     }
    230   } else {
    231     if (characters == 4) {
    232       bytecode = BC_LOAD_4_CURRENT_CHARS_UNCHECKED;
    233     } else if (characters == 2) {
    234       bytecode = BC_LOAD_2_CURRENT_CHARS_UNCHECKED;
    235     } else {
    236       DCHECK(characters == 1);
    237       bytecode = BC_LOAD_CURRENT_CHAR_UNCHECKED;
    238     }
    239   }
    240   Emit(bytecode, cp_offset);
    241   if (check_bounds) EmitOrLink(on_failure);
    242 }
    243 
    244 
    245 void RegExpMacroAssemblerIrregexp::CheckCharacterLT(uc16 limit,
    246                                                     Label* on_less) {
    247   Emit(BC_CHECK_LT, limit);
    248   EmitOrLink(on_less);
    249 }
    250 
    251 
    252 void RegExpMacroAssemblerIrregexp::CheckCharacterGT(uc16 limit,
    253                                                     Label* on_greater) {
    254   Emit(BC_CHECK_GT, limit);
    255   EmitOrLink(on_greater);
    256 }
    257 
    258 
    259 void RegExpMacroAssemblerIrregexp::CheckCharacter(uint32_t c, Label* on_equal) {
    260   if (c > MAX_FIRST_ARG) {
    261     Emit(BC_CHECK_4_CHARS, 0);
    262     Emit32(c);
    263   } else {
    264     Emit(BC_CHECK_CHAR, c);
    265   }
    266   EmitOrLink(on_equal);
    267 }
    268 
    269 
    270 void RegExpMacroAssemblerIrregexp::CheckAtStart(Label* on_at_start) {
    271   Emit(BC_CHECK_AT_START, 0);
    272   EmitOrLink(on_at_start);
    273 }
    274 
    275 
    276 void RegExpMacroAssemblerIrregexp::CheckNotAtStart(Label* on_not_at_start) {
    277   Emit(BC_CHECK_NOT_AT_START, 0);
    278   EmitOrLink(on_not_at_start);
    279 }
    280 
    281 
    282 void RegExpMacroAssemblerIrregexp::CheckNotCharacter(uint32_t c,
    283                                                      Label* on_not_equal) {
    284   if (c > MAX_FIRST_ARG) {
    285     Emit(BC_CHECK_NOT_4_CHARS, 0);
    286     Emit32(c);
    287   } else {
    288     Emit(BC_CHECK_NOT_CHAR, c);
    289   }
    290   EmitOrLink(on_not_equal);
    291 }
    292 
    293 
    294 void RegExpMacroAssemblerIrregexp::CheckCharacterAfterAnd(
    295     uint32_t c,
    296     uint32_t mask,
    297     Label* on_equal) {
    298   if (c > MAX_FIRST_ARG) {
    299     Emit(BC_AND_CHECK_4_CHARS, 0);
    300     Emit32(c);
    301   } else {
    302     Emit(BC_AND_CHECK_CHAR, c);
    303   }
    304   Emit32(mask);
    305   EmitOrLink(on_equal);
    306 }
    307 
    308 
    309 void RegExpMacroAssemblerIrregexp::CheckNotCharacterAfterAnd(
    310     uint32_t c,
    311     uint32_t mask,
    312     Label* on_not_equal) {
    313   if (c > MAX_FIRST_ARG) {
    314     Emit(BC_AND_CHECK_NOT_4_CHARS, 0);
    315     Emit32(c);
    316   } else {
    317     Emit(BC_AND_CHECK_NOT_CHAR, c);
    318   }
    319   Emit32(mask);
    320   EmitOrLink(on_not_equal);
    321 }
    322 
    323 
    324 void RegExpMacroAssemblerIrregexp::CheckNotCharacterAfterMinusAnd(
    325     uc16 c,
    326     uc16 minus,
    327     uc16 mask,
    328     Label* on_not_equal) {
    329   Emit(BC_MINUS_AND_CHECK_NOT_CHAR, c);
    330   Emit16(minus);
    331   Emit16(mask);
    332   EmitOrLink(on_not_equal);
    333 }
    334 
    335 
    336 void RegExpMacroAssemblerIrregexp::CheckCharacterInRange(
    337     uc16 from,
    338     uc16 to,
    339     Label* on_in_range) {
    340   Emit(BC_CHECK_CHAR_IN_RANGE, 0);
    341   Emit16(from);
    342   Emit16(to);
    343   EmitOrLink(on_in_range);
    344 }
    345 
    346 
    347 void RegExpMacroAssemblerIrregexp::CheckCharacterNotInRange(
    348     uc16 from,
    349     uc16 to,
    350     Label* on_not_in_range) {
    351   Emit(BC_CHECK_CHAR_NOT_IN_RANGE, 0);
    352   Emit16(from);
    353   Emit16(to);
    354   EmitOrLink(on_not_in_range);
    355 }
    356 
    357 
    358 void RegExpMacroAssemblerIrregexp::CheckBitInTable(
    359     Handle<ByteArray> table, Label* on_bit_set) {
    360   Emit(BC_CHECK_BIT_IN_TABLE, 0);
    361   EmitOrLink(on_bit_set);
    362   for (int i = 0; i < kTableSize; i += kBitsPerByte) {
    363     int byte = 0;
    364     for (int j = 0; j < kBitsPerByte; j++) {
    365       if (table->get(i + j) != 0) byte |= 1 << j;
    366     }
    367     Emit8(byte);
    368   }
    369 }
    370 
    371 
    372 void RegExpMacroAssemblerIrregexp::CheckNotBackReference(int start_reg,
    373                                                          Label* on_not_equal) {
    374   DCHECK(start_reg >= 0);
    375   DCHECK(start_reg <= kMaxRegister);
    376   Emit(BC_CHECK_NOT_BACK_REF, start_reg);
    377   EmitOrLink(on_not_equal);
    378 }
    379 
    380 
    381 void RegExpMacroAssemblerIrregexp::CheckNotBackReferenceIgnoreCase(
    382     int start_reg,
    383     Label* on_not_equal) {
    384   DCHECK(start_reg >= 0);
    385   DCHECK(start_reg <= kMaxRegister);
    386   Emit(BC_CHECK_NOT_BACK_REF_NO_CASE, start_reg);
    387   EmitOrLink(on_not_equal);
    388 }
    389 
    390 
    391 void RegExpMacroAssemblerIrregexp::IfRegisterLT(int register_index,
    392                                                 int comparand,
    393                                                 Label* on_less_than) {
    394   DCHECK(register_index >= 0);
    395   DCHECK(register_index <= kMaxRegister);
    396   Emit(BC_CHECK_REGISTER_LT, register_index);
    397   Emit32(comparand);
    398   EmitOrLink(on_less_than);
    399 }
    400 
    401 
    402 void RegExpMacroAssemblerIrregexp::IfRegisterGE(int register_index,
    403                                                 int comparand,
    404                                                 Label* on_greater_or_equal) {
    405   DCHECK(register_index >= 0);
    406   DCHECK(register_index <= kMaxRegister);
    407   Emit(BC_CHECK_REGISTER_GE, register_index);
    408   Emit32(comparand);
    409   EmitOrLink(on_greater_or_equal);
    410 }
    411 
    412 
    413 void RegExpMacroAssemblerIrregexp::IfRegisterEqPos(int register_index,
    414                                                    Label* on_eq) {
    415   DCHECK(register_index >= 0);
    416   DCHECK(register_index <= kMaxRegister);
    417   Emit(BC_CHECK_REGISTER_EQ_POS, register_index);
    418   EmitOrLink(on_eq);
    419 }
    420 
    421 
    422 Handle<HeapObject> RegExpMacroAssemblerIrregexp::GetCode(
    423     Handle<String> source) {
    424   Bind(&backtrack_);
    425   Emit(BC_POP_BT, 0);
    426   Handle<ByteArray> array = isolate_->factory()->NewByteArray(length());
    427   Copy(array->GetDataStartAddress());
    428   return array;
    429 }
    430 
    431 
    432 int RegExpMacroAssemblerIrregexp::length() {
    433   return pc_;
    434 }
    435 
    436 
    437 void RegExpMacroAssemblerIrregexp::Copy(Address a) {
    438   MemCopy(a, buffer_.start(), length());
    439 }
    440 
    441 
    442 void RegExpMacroAssemblerIrregexp::Expand() {
    443   bool old_buffer_was_our_own = own_buffer_;
    444   Vector<byte> old_buffer = buffer_;
    445   buffer_ = Vector<byte>::New(old_buffer.length() * 2);
    446   own_buffer_ = true;
    447   MemCopy(buffer_.start(), old_buffer.start(), old_buffer.length());
    448   if (old_buffer_was_our_own) {
    449     old_buffer.Dispose();
    450   }
    451 }
    452 
    453 #endif  // V8_INTERPRETED_REGEXP
    454 
    455 } }  // namespace v8::internal
    456