pcre_exec.cpp.svn-base

Google浏览器V8内核代码

            /* Match a single ASCII character. */
                
            BEGIN_OPCODE(ASCII_CHAR):
                if (md.endSubject == stack.currentFrame->args.subjectPtr)
                    RRETURN_NO_MATCH;
                if (*stack.currentFrame->args.subjectPtr != stack.currentFrame->args.instructionPtr[1])
                    RRETURN_NO_MATCH;
                ++stack.currentFrame->args.subjectPtr;
                stack.currentFrame->args.instructionPtr += 2;
                NEXT_OPCODE;
                
            /* Match one of two cases of an ASCII letter. */
                
            BEGIN_OPCODE(ASCII_LETTER_IGNORING_CASE):
                if (md.endSubject == stack.currentFrame->args.subjectPtr)
                    RRETURN_NO_MATCH;
                if ((*stack.currentFrame->args.subjectPtr | 0x20) != stack.currentFrame->args.instructionPtr[1])
                    RRETURN_NO_MATCH;
                ++stack.currentFrame->args.subjectPtr;
                stack.currentFrame->args.instructionPtr += 2;
                NEXT_OPCODE;
                
            /* Match a single character repeatedly; different opcodes share code. */
                
            BEGIN_OPCODE(EXACT):
                min = stack.currentFrame->locals.max = get2ByteValue(stack.currentFrame->args.instructionPtr + 1);
                minimize = false;
                stack.currentFrame->args.instructionPtr += 3;
                goto REPEATCHAR;
                
            BEGIN_OPCODE(UPTO):
            BEGIN_OPCODE(MINUPTO):
                min = 0;
                stack.currentFrame->locals.max = get2ByteValue(stack.currentFrame->args.instructionPtr + 1);
                minimize = *stack.currentFrame->args.instructionPtr == OP_MINUPTO;
                stack.currentFrame->args.instructionPtr += 3;
                goto REPEATCHAR;
                
            BEGIN_OPCODE(STAR):
            BEGIN_OPCODE(MINSTAR):
            BEGIN_OPCODE(PLUS):
            BEGIN_OPCODE(MINPLUS):
            BEGIN_OPCODE(QUERY):
            BEGIN_OPCODE(MINQUERY):
                repeatInformationFromInstructionOffset(*stack.currentFrame->args.instructionPtr++ - OP_STAR, minimize, min, stack.currentFrame->locals.max);
                
                /* Common code for all repeated single-character matches. We can give
                 up quickly if there are fewer than the minimum number of characters left in
                 the subject. */
                
            REPEATCHAR:
                
                stack.currentFrame->locals.length = 1;
                getUTF8CharAndIncrementLength(stack.currentFrame->locals.fc, stack.currentFrame->args.instructionPtr, stack.currentFrame->locals.length);
                if (min * (stack.currentFrame->locals.fc > 0xFFFF ? 2 : 1) > md.endSubject - stack.currentFrame->args.subjectPtr)
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr += stack.currentFrame->locals.length;
                
                if (stack.currentFrame->locals.fc <= 0xFFFF) {
                    int othercase = md.ignoreCase ? kjs_pcre_ucp_othercase(stack.currentFrame->locals.fc) : -1;
                    
                    for (int i = 1; i <= min; i++) {
                        if (*stack.currentFrame->args.subjectPtr != stack.currentFrame->locals.fc && *stack.currentFrame->args.subjectPtr != othercase)
                            RRETURN_NO_MATCH;
                        ++stack.currentFrame->args.subjectPtr;
                    }
                    
                    if (min == stack.currentFrame->locals.max)
                        NEXT_OPCODE;
                    
                    if (minimize) {
                        stack.currentFrame->locals.repeatOthercase = othercase;
                        for (stack.currentFrame->locals.fi = min;; stack.currentFrame->locals.fi++) {
                            RECURSIVE_MATCH(28, stack.currentFrame->args.instructionPtr, stack.currentFrame->args.bracketChain);
                            if (isMatch)
                                RRETURN;
                            if (stack.currentFrame->locals.fi >= stack.currentFrame->locals.max || stack.currentFrame->args.subjectPtr >= md.endSubject)
                                RRETURN;
                            if (*stack.currentFrame->args.subjectPtr != stack.currentFrame->locals.fc && *stack.currentFrame->args.subjectPtr != stack.currentFrame->locals.repeatOthercase)
                                RRETURN;
                            ++stack.currentFrame->args.subjectPtr;
                        }
                        /* Control never reaches here */
                    } else {
                        stack.currentFrame->locals.subjectPtrAtStartOfInstruction = stack.currentFrame->args.subjectPtr;
                        for (int i = min; i < stack.currentFrame->locals.max; i++) {
                            if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                                break;
                            if (*stack.currentFrame->args.subjectPtr != stack.currentFrame->locals.fc && *stack.currentFrame->args.subjectPtr != othercase)
                                break;
                            ++stack.currentFrame->args.subjectPtr;
                        }
                        while (stack.currentFrame->args.subjectPtr >= stack.currentFrame->locals.subjectPtrAtStartOfInstruction) {
                            RECURSIVE_MATCH(29, stack.currentFrame->args.instructionPtr, stack.currentFrame->args.bracketChain);
                            if (isMatch)
                                RRETURN;
                            --stack.currentFrame->args.subjectPtr;
                        }
                        RRETURN_NO_MATCH;
                    }
                    /* Control never reaches here */
                } else {
                    /* No case on surrogate pairs, so no need to bother with "othercase". */
                    
                    for (int i = 1; i <= min; i++) {
                        if (*stack.currentFrame->args.subjectPtr != stack.currentFrame->locals.fc)
                            RRETURN_NO_MATCH;
                        stack.currentFrame->args.subjectPtr += 2;
                    }
                    
                    if (min == stack.currentFrame->locals.max)
                        NEXT_OPCODE;
                    
                    if (minimize) {
                        for (stack.currentFrame->locals.fi = min;; stack.currentFrame->locals.fi++) {
                            RECURSIVE_MATCH(30, stack.currentFrame->args.instructionPtr, stack.currentFrame->args.bracketChain);
                            if (isMatch)
                                RRETURN;
                            if (stack.currentFrame->locals.fi >= stack.currentFrame->locals.max || stack.currentFrame->args.subjectPtr >= md.endSubject)
                                RRETURN;
                            if (*stack.currentFrame->args.subjectPtr != stack.currentFrame->locals.fc)
                                RRETURN;
                            stack.currentFrame->args.subjectPtr += 2;
                        }
                        /* Control never reaches here */
                    } else {
                        stack.currentFrame->locals.subjectPtrAtStartOfInstruction = stack.currentFrame->args.subjectPtr;
                        for (int i = min; i < stack.currentFrame->locals.max; i++) {
                            if (stack.currentFrame->args.subjectPtr > md.endSubject - 2)
                                break;
                            if (*stack.currentFrame->args.subjectPtr != stack.currentFrame->locals.fc)
                                break;
                            stack.currentFrame->args.subjectPtr += 2;
                        }
                        while (stack.currentFrame->args.subjectPtr >= stack.currentFrame->locals.subjectPtrAtStartOfInstruction) {
                            RECURSIVE_MATCH(31, stack.currentFrame->args.instructionPtr, stack.currentFrame->args.bracketChain);
                            if (isMatch)
                                RRETURN;
                            stack.currentFrame->args.subjectPtr -= 2;
                        }
                        RRETURN_NO_MATCH;
                    }
                    /* Control never reaches here */
                }
                /* Control never reaches here */
                
            /* Match a negated single one-byte character. */
                
            BEGIN_OPCODE(NOT): {
                if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                int c = *stack.currentFrame->args.subjectPtr++;
                if (md.ignoreCase) {
                    if (c < 128)
                        c = toLowerCase(c);
                    if (toLowerCase(*stack.currentFrame->args.instructionPtr++) == c)
                        RRETURN_NO_MATCH;
                } else {
                    if (*stack.currentFrame->args.instructionPtr++ == c)
                        RRETURN_NO_MATCH;
                }
                NEXT_OPCODE;
            }
                
            /* Match a negated single one-byte character repeatedly. This is almost a
             repeat of the code for a repeated single character, but I haven't found a
             nice way of commoning these up that doesn't require a test of the
             positive/negative option for each character match. Maybe that wouldn't add
             very much to the time taken, but character matching *is* what this is all
             about... */
                
            BEGIN_OPCODE(NOTEXACT):
                min = stack.currentFrame->locals.max = get2ByteValue(stack.currentFrame->args.instructionPtr + 1);
                minimize = false;
                stack.currentFrame->args.instructionPtr += 3;
                goto REPEATNOTCHAR;
                
            BEGIN_OPCODE(NOTUPTO):
            BEGIN_OPCODE(NOTMINUPTO):
                min = 0;
                stack.currentFrame->locals.max = get2ByteValue(stack.currentFrame->args.instructionPtr + 1);
                minimize = *stack.currentFrame->args.instructionPtr == OP_NOTMINUPTO;
                stack.currentFrame->args.instructionPtr += 3;
                goto REPEATNOTCHAR;
                
            BEGIN_OPCODE(NOTSTAR):
            BEGIN_OPCODE(NOTMINSTAR):
            BEGIN_OPCODE(NOTPLUS):
            BEGIN_OPCODE(NOTMINPLUS):
            BEGIN_OPCODE(NOTQUERY):
            BEGIN_OPCODE(NOTMINQUERY):
                repeatInformationFromInstructionOffset(*stack.currentFrame->args.instructionPtr++ - OP_NOTSTAR, minimize, min, stack.currentFrame->locals.max);
                
            /* Common code for all repeated single-byte matches. We can give up quickly
             if there are fewer than the minimum number of bytes left in the
             subject. */
                
            REPEATNOTCHAR:
                if (min > md.endSubject - stack.currentFrame->args.subjectPtr)
                    RRETURN_NO_MATCH;
                stack.currentFrame->locals.fc = *stack.currentFrame->args.instructionPtr++;
                
                /* The code is duplicated for the caseless and caseful cases, for speed,
                 since matching characters is likely to be quite common. First, ensure the
                 minimum number of matches are present. If min = max, continue at the same
                 level without recursing. Otherwise, if minimizing, keep trying the rest of
                 the expression and advancing one matching character if failing, up to the
                 maximum. Alternatively, if maximizing, find the maximum number of
                 characters and work backwards. */
                
                DPRINTF(("negative matching %c{%d,%d}\n", stack.currentFrame->locals.fc, min, stack.currentFrame->locals.max));
                
                if (md.ignoreCase) {
                    if (stack.currentFrame->locals.fc < 128)
                        stack.currentFrame->locals.fc = toLowerCase(stack.currentFrame->locals.fc);
                    
                    for (int i = 1; i <= min; i++) {
                        int d = *stack.currentFrame->args.subjectPtr++;
                        if (d < 128)
                            d = toLowerCase(d);
                        if (stack.currentFrame->locals.fc == d)
                            RRETURN_NO_MATCH;
                    }
                    
                    if (min == stack.currentFrame->locals.max)
                        NEXT_OPCODE;      
                    
                    if (minimize) {
                        for (stack.currentFrame->locals.fi = min;; stack.currentFrame->locals.fi++) {
                            RECURSIVE_MATCH(38, stack.currentFrame->args.instructionPtr, stack.currentFrame->args.bracketChain);
                            if (isMatch)
                                RRETURN;
                            int d = *stack.currentFrame->args.subjectPtr++;
                            if (d < 128)
                                d = toLowerCase(d);
                            if (stack.currentFrame->locals.fi >= stack.currentFrame->locals.max || stack.currentFrame->args.subjectPtr >= md.endSubject || stack.currentFrame->locals.fc == d)
                                RRETURN;
                        }
                        /* Control never reaches here */
                    }
                    
                    /* Maximize case */
                    
                    else {
                        stack.currentFrame->locals.subjectPtrAtStartOfInstruction = stack.currentFrame->args.subjectPtr;
                        
                        for (int i = min; i < stack.currentFrame->locals.max; i++) {
                            if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                                break;
                            int d = *stack.currentFrame->args.subjectPtr;
                            if (d < 128)
                                d = toLowerCase(d);
                            if (stack.currentFrame->locals.fc == d)
                                break;
                            ++stack.currentFrame->args.subjectPtr;
                        }
                        for (;;) {
                            RECURSIVE_MATCH(40, stack.currentFrame->args.instructionPtr, stack.currentFrame->args.bracketChain);
                            if (isMatch)
                                RRETURN;
                            if (stack.currentFrame->args.subjectPtr-- == stack.currentFrame->locals.subjectPtrAtStartOfInstruction)
                                break;        /* Stop if tried at original pos */
                        }
                        
                        RRETURN;
                    }
                    /* Control never reaches here */
                }