pcre_exec.cpp

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

                NEXT_OPCODE;
                
            /* Negative assertion: all branches must fail to match */
                
            BEGIN_OPCODE(ASSERT_NOT):
                do {
                    RECURSIVE_MATCH_NEW_GROUP(7, stack.currentFrame->args.instructionPtr + 1 + LINK_SIZE, NULL);
                    if (isMatch)
                        RRETURN_NO_MATCH;
                    stack.currentFrame->args.instructionPtr += getLinkValue(stack.currentFrame->args.instructionPtr + 1);
                } while (*stack.currentFrame->args.instructionPtr == OP_ALT);
                
                stack.currentFrame->args.instructionPtr += 1 + LINK_SIZE;
                NEXT_OPCODE;
                
            /* An alternation is the end of a branch; scan along to find the end of the
             bracketed group and go to there. */
                
            BEGIN_OPCODE(ALT):
                advanceToEndOfBracket(stack.currentFrame->args.instructionPtr);
                NEXT_OPCODE;
                
            /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating
             that it may occur zero times. It may repeat infinitely, or not at all -
             i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper
             repeat limits are compiled as a number of copies, with the optional ones
             preceded by BRAZERO or BRAMINZERO. */
                
            BEGIN_OPCODE(BRAZERO): {
                stack.currentFrame->locals.startOfRepeatingBracket = stack.currentFrame->args.instructionPtr + 1;
                RECURSIVE_MATCH_NEW_GROUP(14, stack.currentFrame->locals.startOfRepeatingBracket, stack.currentFrame->args.bracketChain);
                if (isMatch)
                    RRETURN;
                advanceToEndOfBracket(stack.currentFrame->locals.startOfRepeatingBracket);
                stack.currentFrame->args.instructionPtr = stack.currentFrame->locals.startOfRepeatingBracket + 1 + LINK_SIZE;
                NEXT_OPCODE;
            }
                
            BEGIN_OPCODE(BRAMINZERO): {
                stack.currentFrame->locals.startOfRepeatingBracket = stack.currentFrame->args.instructionPtr + 1;
                advanceToEndOfBracket(stack.currentFrame->locals.startOfRepeatingBracket);
                RECURSIVE_MATCH_NEW_GROUP(15, stack.currentFrame->locals.startOfRepeatingBracket + 1 + LINK_SIZE, stack.currentFrame->args.bracketChain);
                if (isMatch)
                    RRETURN;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;
            }
                
            /* End of a group, repeated or non-repeating. If we are at the end of
             an assertion "group", stop matching and return 1, but record the
             current high water mark for use by positive assertions. Do this also
             for the "once" (not-backup up) groups. */
                
            BEGIN_OPCODE(KET):
            BEGIN_OPCODE(KETRMIN):
            BEGIN_OPCODE(KETRMAX):
                stack.currentFrame->locals.instructionPtrAtStartOfOnce = stack.currentFrame->args.instructionPtr - getLinkValue(stack.currentFrame->args.instructionPtr + 1);
                stack.currentFrame->locals.subjectPtrAtStartOfInstruction = stack.currentFrame->args.bracketChain->bracketStart;

                /* Back up the stack of bracket start pointers. */

                stack.currentFrame->args.bracketChain = stack.currentFrame->args.bracketChain->previousBracket;

                if (*stack.currentFrame->locals.instructionPtrAtStartOfOnce == OP_ASSERT || *stack.currentFrame->locals.instructionPtrAtStartOfOnce == OP_ASSERT_NOT) {
                    md.endOffsetTop = stack.currentFrame->args.offsetTop;
                    isMatch = true;
                    RRETURN;
                }
                
                /* In all other cases except a conditional group we have to check the
                 group number back at the start and if necessary complete handling an
                 extraction by setting the offsets and bumping the high water mark. */
                
                stack.currentFrame->locals.number = *stack.currentFrame->locals.instructionPtrAtStartOfOnce - OP_BRA;
                
                /* For extended extraction brackets (large number), we have to fish out
                 the number from a dummy opcode at the start. */
                
                if (stack.currentFrame->locals.number > EXTRACT_BASIC_MAX)
                    stack.currentFrame->locals.number = get2ByteValue(stack.currentFrame->locals.instructionPtrAtStartOfOnce + 2 + LINK_SIZE);
                stack.currentFrame->locals.offset = stack.currentFrame->locals.number << 1;
                
#ifdef DEBUG
                printf("end bracket %d", stack.currentFrame->locals.number);
                printf("\n");
#endif
                
                /* Test for a numbered group. This includes groups called as a result
                 of recursion. Note that whole-pattern recursion is coded as a recurse
                 into group 0, so it won't be picked up here. Instead, we catch it when
                 the OP_END is reached. */
                
                if (stack.currentFrame->locals.number > 0) {
                    if (stack.currentFrame->locals.offset >= md.offsetMax)
                        md.offsetOverflow = true;
                    else {
                        md.offsetVector[stack.currentFrame->locals.offset] =
                        md.offsetVector[md.offsetEnd - stack.currentFrame->locals.number];
                        md.offsetVector[stack.currentFrame->locals.offset+1] = stack.currentFrame->args.subjectPtr - md.startSubject;
                        if (stack.currentFrame->args.offsetTop <= stack.currentFrame->locals.offset)
                            stack.currentFrame->args.offsetTop = stack.currentFrame->locals.offset + 2;
                    }
                }
                
                /* For a non-repeating ket, just continue at this level. This also
                 happens for a repeating ket if no characters were matched in the group.
                 This is the forcible breaking of infinite loops as implemented in Perl
                 5.005. If there is an options reset, it will get obeyed in the normal
                 course of events. */
                
                if (*stack.currentFrame->args.instructionPtr == OP_KET || stack.currentFrame->args.subjectPtr == stack.currentFrame->locals.subjectPtrAtStartOfInstruction) {
                    stack.currentFrame->args.instructionPtr += 1 + LINK_SIZE;
                    NEXT_OPCODE;
                }
                
                /* The repeating kets try the rest of the pattern or restart from the
                 preceding bracket, in the appropriate order. */
                
                if (*stack.currentFrame->args.instructionPtr == OP_KETRMIN) {
                    RECURSIVE_MATCH(16, stack.currentFrame->args.instructionPtr + 1 + LINK_SIZE, stack.currentFrame->args.bracketChain);
                    if (isMatch)
                        RRETURN;
                    RECURSIVE_MATCH_NEW_GROUP(17, stack.currentFrame->locals.instructionPtrAtStartOfOnce, stack.currentFrame->args.bracketChain);
                    if (isMatch)
                        RRETURN;
                } else { /* OP_KETRMAX */
                    RECURSIVE_MATCH_NEW_GROUP(18, stack.currentFrame->locals.instructionPtrAtStartOfOnce, stack.currentFrame->args.bracketChain);
                    if (isMatch)
                        RRETURN;
                    RECURSIVE_MATCH(19, stack.currentFrame->args.instructionPtr + 1 + LINK_SIZE, stack.currentFrame->args.bracketChain);
                    if (isMatch)
                        RRETURN;
                }
                RRETURN;
                
            /* Start of subject. */

            BEGIN_OPCODE(CIRC):
                if (stack.currentFrame->args.subjectPtr != md.startSubject)
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;

            /* After internal newline if multiline. */

            BEGIN_OPCODE(BOL):
                if (stack.currentFrame->args.subjectPtr != md.startSubject && !isNewline(stack.currentFrame->args.subjectPtr[-1]))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;

            /* End of subject. */

            BEGIN_OPCODE(DOLL):
                if (stack.currentFrame->args.subjectPtr < md.endSubject)
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;

            /* Before internal newline if multiline. */

            BEGIN_OPCODE(EOL):
                if (stack.currentFrame->args.subjectPtr < md.endSubject && !isNewline(*stack.currentFrame->args.subjectPtr))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;
                
            /* Word boundary assertions */
                
            BEGIN_OPCODE(NOT_WORD_BOUNDARY):
            BEGIN_OPCODE(WORD_BOUNDARY): {
                bool currentCharIsWordChar = false;
                bool previousCharIsWordChar = false;
                
                if (stack.currentFrame->args.subjectPtr > md.startSubject)
                    previousCharIsWordChar = isWordChar(stack.currentFrame->args.subjectPtr[-1]);
                if (stack.currentFrame->args.subjectPtr < md.endSubject)
                    currentCharIsWordChar = isWordChar(*stack.currentFrame->args.subjectPtr);
                
                /* Now see if the situation is what we want */
                bool wordBoundaryDesired = (*stack.currentFrame->args.instructionPtr++ == OP_WORD_BOUNDARY);
                if (wordBoundaryDesired ? currentCharIsWordChar == previousCharIsWordChar : currentCharIsWordChar != previousCharIsWordChar)
                    RRETURN_NO_MATCH;
                NEXT_OPCODE;
            }
                
            /* Match a single character type; inline for speed */
                
            BEGIN_OPCODE(NOT_NEWLINE):
                if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                    RRETURN_NO_MATCH;
                if (isNewline(*stack.currentFrame->args.subjectPtr++))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;

            BEGIN_OPCODE(NOT_DIGIT):
                if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                    RRETURN_NO_MATCH;
                if (isASCIIDigit(*stack.currentFrame->args.subjectPtr++))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;

            BEGIN_OPCODE(DIGIT):
                if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                    RRETURN_NO_MATCH;
                if (!isASCIIDigit(*stack.currentFrame->args.subjectPtr++))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;

            BEGIN_OPCODE(NOT_WHITESPACE):
                if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                    RRETURN_NO_MATCH;
                if (isSpaceChar(*stack.currentFrame->args.subjectPtr++))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;

            BEGIN_OPCODE(WHITESPACE):
                if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                    RRETURN_NO_MATCH;
                if (!isSpaceChar(*stack.currentFrame->args.subjectPtr++))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;
                
            BEGIN_OPCODE(NOT_WORDCHAR):
                if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                    RRETURN_NO_MATCH;
                if (isWordChar(*stack.currentFrame->args.subjectPtr++))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;
                
            BEGIN_OPCODE(WORDCHAR):
                if (stack.currentFrame->args.subjectPtr >= md.endSubject)
                    RRETURN_NO_MATCH;
                if (!isWordChar(*stack.currentFrame->args.subjectPtr++))
                    RRETURN_NO_MATCH;
                stack.currentFrame->args.instructionPtr++;
                NEXT_OPCODE;
                
            /* Match a back reference, possibly repeatedly. Look past the end of the
             item to see if there is repeat information following. The code is similar
             to that for character classes, but repeated for efficiency. Then obey
             similar code to character type repeats - written out again for speed.
             However, if the referenced string is the empty string, always treat
             it as matched, any number of times (otherwise there could be infinite
             loops). */
                
            BEGIN_OPCODE(REF):
                stack.currentFrame->locals.offset = get2ByteValue(stack.currentFrame->args.instructionPtr + 1) << 1;               /* Doubled ref number */
                stack.currentFrame->args.instructionPtr += 3;                                 /* Advance past item */
                
                /* If the reference is unset, set the length to be longer than the amount
                 of subject left; this ensures that every attempt at a match fails. We
                 can't just fail here, because of the possibility of quantifiers with zero
                 minima. */
                
                if (stack.currentFrame->locals.offset >= stack.currentFrame->args.offsetTop || md.offsetVector[stack.currentFrame->locals.offset] < 0)
                    stack.currentFrame->locals.length = 0;
                else
                    stack.currentFrame->locals.length = md.offsetVector[stack.currentFrame->locals.offset+1] - md.offsetVector[stack.currentFrame->locals.offset];
                
                /* Set up for repetition, or handle the non-repeated case */
                
                switch (*stack.currentFrame->args.instructionPtr) {
                    case OP_CRSTAR:
                    case OP_CRMINSTAR:
                    case OP_CRPLUS:
                    case OP_CRMINPLUS:
                    case OP_CRQUERY:
                    case OP_CRMINQUERY:
                        repeatInformationFromInstructionOffset(*stack.currentFrame->args.instructionPtr++ - OP_CRSTAR, minimize, min, stack.currentFrame->locals.max);
                        break;