1 // Copyright 2008 the V8 project authors. All rights reserved. 2 // Redistribution and use in source and binary forms, with or without 3 // modification, are permitted provided that the following conditions are 4 // met: 5 // 6 // * Redistributions of source code must retain the above copyright 7 // notice, this list of conditions and the following disclaimer. 8 // * Redistributions in binary form must reproduce the above 9 // copyright notice, this list of conditions and the following 10 // disclaimer in the documentation and/or other materials provided 11 // with the distribution. 12 // * Neither the name of Google Inc. nor the names of its 13 // contributors may be used to endorse or promote products derived 14 // from this software without specific prior written permission. 15 // 16 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 17 // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 18 // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 19 // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 20 // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 21 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 22 // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 23 // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 24 // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 25 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 26 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 27 28 #ifndef V8_REGEXP_MACRO_ASSEMBLER_H_ 29 #define V8_REGEXP_MACRO_ASSEMBLER_H_ 30 31 namespace v8 { 32 namespace internal { 33 34 struct DisjunctDecisionRow { 35 RegExpCharacterClass cc; 36 Label* on_match; 37 }; 38 39 40 class RegExpMacroAssembler { 41 public: 42 // The implementation must be able to handle at least: 43 static const int kMaxRegister = (1 << 16) - 1; 44 static const int kMaxCPOffset = (1 << 15) - 1; 45 static const int kMinCPOffset = -(1 << 15); 46 enum IrregexpImplementation { 47 kIA32Implementation, 48 kARMImplementation, 49 kX64Implementation, 50 kBytecodeImplementation 51 }; 52 53 enum StackCheckFlag { 54 kNoStackLimitCheck = false, 55 kCheckStackLimit = true 56 }; 57 58 RegExpMacroAssembler(); 59 virtual ~RegExpMacroAssembler(); 60 // The maximal number of pushes between stack checks. Users must supply 61 // kCheckStackLimit flag to push operations (instead of kNoStackLimitCheck) 62 // at least once for every stack_limit() pushes that are executed. 63 virtual int stack_limit_slack() = 0; 64 virtual bool CanReadUnaligned(); 65 virtual void AdvanceCurrentPosition(int by) = 0; // Signed cp change. 66 virtual void AdvanceRegister(int reg, int by) = 0; // r[reg] += by. 67 // Continues execution from the position pushed on the top of the backtrack 68 // stack by an earlier PushBacktrack(Label*). 69 virtual void Backtrack() = 0; 70 virtual void Bind(Label* label) = 0; 71 virtual void CheckAtStart(Label* on_at_start) = 0; 72 // Dispatch after looking the current character up in a 2-bits-per-entry 73 // map. The destinations vector has up to 4 labels. 74 virtual void CheckCharacter(uint32_t c, Label* on_equal) = 0; 75 // Bitwise and the current character with the given constant and then 76 // check for a match with c. 77 virtual void CheckCharacterAfterAnd(uint32_t c, 78 uint32_t and_with, 79 Label* on_equal) = 0; 80 virtual void CheckCharacterGT(uc16 limit, Label* on_greater) = 0; 81 virtual void CheckCharacterLT(uc16 limit, Label* on_less) = 0; 82 // Check the current character for a match with a literal string. If we 83 // fail to match then goto the on_failure label. If check_eos is set then 84 // the end of input always fails. If check_eos is clear then it is the 85 // caller's responsibility to ensure that the end of string is not hit. 86 // If the label is NULL then we should pop a backtrack address off 87 // the stack and go to that. 88 virtual void CheckCharacters( 89 Vector<const uc16> str, 90 int cp_offset, 91 Label* on_failure, 92 bool check_eos) = 0; 93 virtual void CheckGreedyLoop(Label* on_tos_equals_current_position) = 0; 94 virtual void CheckNotAtStart(Label* on_not_at_start) = 0; 95 virtual void CheckNotBackReference(int start_reg, Label* on_no_match) = 0; 96 virtual void CheckNotBackReferenceIgnoreCase(int start_reg, 97 Label* on_no_match) = 0; 98 // Check the current character for a match with a literal character. If we 99 // fail to match then goto the on_failure label. End of input always 100 // matches. If the label is NULL then we should pop a backtrack address off 101 // the stack and go to that. 102 virtual void CheckNotCharacter(uint32_t c, Label* on_not_equal) = 0; 103 virtual void CheckNotCharacterAfterAnd(uint32_t c, 104 uint32_t and_with, 105 Label* on_not_equal) = 0; 106 // Subtract a constant from the current character, then or with the given 107 // constant and then check for a match with c. 108 virtual void CheckNotCharacterAfterMinusAnd(uc16 c, 109 uc16 minus, 110 uc16 and_with, 111 Label* on_not_equal) = 0; 112 virtual void CheckNotRegistersEqual(int reg1, 113 int reg2, 114 Label* on_not_equal) = 0; 115 116 // Checks whether the given offset from the current position is before 117 // the end of the string. May overwrite the current character. CheckPosition(int cp_offset,Label * on_outside_input)118 virtual void CheckPosition(int cp_offset, Label* on_outside_input) { 119 LoadCurrentCharacter(cp_offset, on_outside_input, true); 120 } 121 // Check whether a standard/default character class matches the current 122 // character. Returns false if the type of special character class does 123 // not have custom support. 124 // May clobber the current loaded character. CheckSpecialCharacterClass(uc16 type,Label * on_no_match)125 virtual bool CheckSpecialCharacterClass(uc16 type, 126 Label* on_no_match) { 127 return false; 128 } 129 virtual void Fail() = 0; 130 virtual Handle<Object> GetCode(Handle<String> source) = 0; 131 virtual void GoTo(Label* label) = 0; 132 // Check whether a register is >= a given constant and go to a label if it 133 // is. Backtracks instead if the label is NULL. 134 virtual void IfRegisterGE(int reg, int comparand, Label* if_ge) = 0; 135 // Check whether a register is < a given constant and go to a label if it is. 136 // Backtracks instead if the label is NULL. 137 virtual void IfRegisterLT(int reg, int comparand, Label* if_lt) = 0; 138 // Check whether a register is == to the current position and go to a 139 // label if it is. 140 virtual void IfRegisterEqPos(int reg, Label* if_eq) = 0; 141 virtual IrregexpImplementation Implementation() = 0; 142 virtual void LoadCurrentCharacter(int cp_offset, 143 Label* on_end_of_input, 144 bool check_bounds = true, 145 int characters = 1) = 0; 146 virtual void PopCurrentPosition() = 0; 147 virtual void PopRegister(int register_index) = 0; 148 // Pushes the label on the backtrack stack, so that a following Backtrack 149 // will go to this label. Always checks the backtrack stack limit. 150 virtual void PushBacktrack(Label* label) = 0; 151 virtual void PushCurrentPosition() = 0; 152 virtual void PushRegister(int register_index, 153 StackCheckFlag check_stack_limit) = 0; 154 virtual void ReadCurrentPositionFromRegister(int reg) = 0; 155 virtual void ReadStackPointerFromRegister(int reg) = 0; 156 virtual void SetRegister(int register_index, int to) = 0; 157 virtual void Succeed() = 0; 158 virtual void WriteCurrentPositionToRegister(int reg, int cp_offset) = 0; 159 virtual void ClearRegisters(int reg_from, int reg_to) = 0; 160 virtual void WriteStackPointerToRegister(int reg) = 0; 161 }; 162 163 164 #ifdef V8_NATIVE_REGEXP // Avoid compiling unused code. 165 166 class NativeRegExpMacroAssembler: public RegExpMacroAssembler { 167 public: 168 // Type of input string to generate code for. 169 enum Mode { ASCII = 1, UC16 = 2 }; 170 171 // Result of calling generated native RegExp code. 172 // RETRY: Something significant changed during execution, and the matching 173 // should be retried from scratch. 174 // EXCEPTION: Something failed during execution. If no exception has been 175 // thrown, it's an internal out-of-memory, and the caller should 176 // throw the exception. 177 // FAILURE: Matching failed. 178 // SUCCESS: Matching succeeded, and the output array has been filled with 179 // capture positions. 180 enum Result { RETRY = -2, EXCEPTION = -1, FAILURE = 0, SUCCESS = 1 }; 181 182 NativeRegExpMacroAssembler(); 183 virtual ~NativeRegExpMacroAssembler(); 184 virtual bool CanReadUnaligned(); 185 186 static Result Match(Handle<Code> regexp, 187 Handle<String> subject, 188 int* offsets_vector, 189 int offsets_vector_length, 190 int previous_index); 191 192 // Compares two-byte strings case insensitively. 193 // Called from generated RegExp code. 194 static int CaseInsensitiveCompareUC16(Address byte_offset1, 195 Address byte_offset2, 196 size_t byte_length); 197 198 // Called from RegExp if the backtrack stack limit is hit. 199 // Tries to expand the stack. Returns the new stack-pointer if 200 // successful, and updates the stack_top address, or returns 0 if unable 201 // to grow the stack. 202 // This function must not trigger a garbage collection. 203 static Address GrowStack(Address stack_pointer, Address* stack_top); 204 205 static const byte* StringCharacterPosition(String* subject, int start_index); 206 207 // Byte map of ASCII characters with a 0xff if the character is a word 208 // character (digit, letter or underscore) and 0x00 otherwise. 209 // Used by generated RegExp code. 210 static byte word_character_map[128]; 211 word_character_map_address()212 static Address word_character_map_address() { 213 return &word_character_map[0]; 214 } 215 216 static Result Execute(Code* code, 217 String* input, 218 int start_offset, 219 const byte* input_start, 220 const byte* input_end, 221 int* output); 222 }; 223 224 #endif // V8_NATIVE_REGEXP 225 226 } } // namespace v8::internal 227 228 #endif // V8_REGEXP_MACRO_ASSEMBLER_H_ 229