pcre_compile.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

     find one.
     
     When we hit a repeat whose minimum is zero, we may have to adjust these values
     to take the zero repeat into account. This is implemented by setting them to
     zeroFirstByte and zeroReqByte when such a repeat is encountered. The individual
     item types that can be repeated set these backoff variables appropriately. */
    
    int firstByte = REQ_UNSET;
    int reqByte = REQ_UNSET;
    int zeroReqByte = REQ_UNSET;
    int zeroFirstByte = REQ_UNSET;
    
    /* The variable reqCaseOpt contains either the REQ_IGNORE_CASE value or zero,
     according to the current setting of the ignores-case flag. REQ_IGNORE_CASE is a bit
     value > 255. It is added into the firstByte or reqByte variables to record the
     case status of the value. This is used only for ASCII characters. */
    
    int reqCaseOpt = (options & IgnoreCaseOption) ? REQ_IGNORE_CASE : 0;
    
    /* Switch on next character until the end of the branch */
    
    for (;; ptr++) {
        bool negateClass;
        bool shouldFlipNegation; /* If a negative special such as \S is used, we should negate the whole class to properly support Unicode. */
        int classCharCount;
        int classLastChar;
        int skipBytes;
        int subReqByte;
        int subFirstByte;
        int mcLength;
        unsigned char mcbuffer[8];
        
        /* Next byte in the pattern */
        
        c = ptr < patternEnd ? *ptr : 0;
        
        /* Fill in length of a previous callout, except when the next thing is
         a quantifier. */
        
        bool isQuantifier = c == '*' || c == '+' || c == '?' || (c == '{' && isCountedRepeat(ptr + 1, patternEnd));
        
        switch (c) {
            /* The branch terminates at end of string, |, or ). */
                
            case 0:
                if (ptr < patternEnd)
                    goto NORMAL_CHAR;
                // End of string; fall through
            case '|':
            case ')':
                *firstbyteptr = firstByte;
                *reqbyteptr = reqByte;
                *codePtr = code;
                *ptrPtr = ptr;
                return true;
                
            /* Handle single-character metacharacters. In multiline mode, ^ disables
             the setting of any following char as a first character. */

            case '^':
                if (options & MatchAcrossMultipleLinesOption) {
                    if (firstByte == REQ_UNSET)
                        firstByte = REQ_NONE;
                    *code++ = OP_BOL;
                } else
                    *code++ = OP_CIRC;
                previous = NULL;
                break;

            case '$':
                previous = NULL;
                if (options & MatchAcrossMultipleLinesOption)
                  *code++ = OP_EOL;
                else
                  *code++ = OP_DOLL;
                break;

            /* There can never be a first char if '.' is first, whatever happens about
             repeats. The value of reqByte doesn't change either. */

            case '.':
                if (firstByte == REQ_UNSET)
                    firstByte = REQ_NONE;
                zeroFirstByte = firstByte;
                zeroReqByte = reqByte;
                previous = code;
                *code++ = OP_NOT_NEWLINE;
                break;
                
            /* Character classes. If the included characters are all < 256, we build a
             32-byte bitmap of the permitted characters, except in the special case
             where there is only one such character. For negated classes, we build the
             map as usual, then invert it at the end. However, we use a different opcode
             so that data characters > 255 can be handled correctly.
             
             If the class contains characters outside the 0-255 range, a different
             opcode is compiled. It may optionally have a bit map for characters < 256,
             but those above are are explicitly listed afterwards. A flag byte tells
             whether the bitmap is present, and whether this is a negated class or not.
             */
                
            case '[': {
                previous = code;
                shouldFlipNegation = false;
                
                /* PCRE supports POSIX class stuff inside a class. Perl gives an error if
                 they are encountered at the top level, so we'll do that too. */
                
                /* If the first character is '^', set the negation flag and skip it. */

                if (ptr + 1 >= patternEnd) {
                    *errorCodePtr = ERR6;
                    return false;
                }

                if (ptr[1] == '^') {
                    negateClass = true;
                    ++ptr;
                } else
                    negateClass = false;
                
                /* Keep a count of chars with values < 256 so that we can optimize the case
                 of just a single character (as long as it's < 256). For higher valued UTF-8
                 characters, we don't yet do any optimization. */
                
                classCharCount = 0;
                classLastChar = -1;
                
                class_utf8 = false;                       /* No chars >= 256 */
                class_utf8data = code + LINK_SIZE + 34;   /* For UTF-8 items */
                
                /* Initialize the 32-char bit map to all zeros. We have to build the
                 map in a temporary bit of store, in case the class contains only 1
                 character (< 256), because in that case the compiled code doesn't use the
                 bit map. */
                
                memset(classbits, 0, 32 * sizeof(unsigned char));
                
                /* Process characters until ] is reached. The first pass
                 through the regex checked the overall syntax, so we don't need to be very
                 strict here. At the start of the loop, c contains the first byte of the
                 character. */

                while ((++ptr < patternEnd) && (c = *ptr) != ']') {
                    /* Backslash may introduce a single character, or it may introduce one
                     of the specials, which just set a flag. Escaped items are checked for
                     validity in the pre-compiling pass. The sequence \b is a special case.
                     Inside a class (and only there) it is treated as backspace. Elsewhere
                     it marks a word boundary. Other escapes have preset maps ready to
                     or into the one we are building. We assume they have more than one
                     character in them, so set classCharCount bigger than one. */
                    
                    if (c == '\\') {
                        c = checkEscape(&ptr, patternEnd, errorCodePtr, cd.numCapturingBrackets, true);
                        if (c < 0) {
                            classCharCount += 2;     /* Greater than 1 is what matters */
                            switch (-c) {
                                case ESC_d:
                                    for (c = 0; c < 32; c++)
                                        classbits[c] |= classBitmapForChar(c + cbit_digit);
                                    continue;
                                    
                                case ESC_D:
                                    shouldFlipNegation = true;
                                    for (c = 0; c < 32; c++)
                                        classbits[c] |= ~classBitmapForChar(c + cbit_digit);
                                    continue;
                                    
                                case ESC_w:
                                    for (c = 0; c < 32; c++)
                                        classbits[c] |= classBitmapForChar(c + cbit_word);
                                    continue;
                                    
                                case ESC_W:
                                    shouldFlipNegation = true;
                                    for (c = 0; c < 32; c++)
                                        classbits[c] |= ~classBitmapForChar(c + cbit_word);
                                    continue;
                                    
                                case ESC_s:
                                    for (c = 0; c < 32; c++)
                                         classbits[c] |= classBitmapForChar(c + cbit_space);
                                    continue;
                                    
                                case ESC_S:
                                    shouldFlipNegation = true;
                                    for (c = 0; c < 32; c++)
                                         classbits[c] |= ~classBitmapForChar(c + cbit_space);
                                    continue;
                                    
                                    /* Unrecognized escapes are faulted if PCRE is running in its
                                     strict mode. By default, for compatibility with Perl, they are
                                     treated as literals. */
                                    
                                default:
                                    c = *ptr;              /* The final character */
                                    classCharCount -= 2;  /* Undo the default count from above */
                            }
                        }
                        
                        /* Fall through if we have a single character (c >= 0). This may be
                         > 256 in UTF-8 mode. */
                        
                    }   /* End of backslash handling */
                    
                    /* A single character may be followed by '-' to form a range. However,
                     Perl does not permit ']' to be the end of the range. A '-' character
                     here is treated as a literal. */
                    
                    if ((ptr + 2 < patternEnd) && ptr[1] == '-' && ptr[2] != ']') {
                        ptr += 2;
                        
                        int d = *ptr;
                        
                        /* The second part of a range can be a single-character escape, but
                         not any of the other escapes. Perl 5.6 treats a hyphen as a literal
                         in such circumstances. */
                        
                        if (d == '\\') {
                            const UChar* oldptr = ptr;
                            d = checkEscape(&ptr, patternEnd, errorCodePtr, cd.numCapturingBrackets, true);
                            
                            /* \X is literal X; any other special means the '-' was literal */
                            if (d < 0) {
                                ptr = oldptr - 2;
                                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
                            }
                        }
                        
                        /* The check that the two values are in the correct order happens in
                         the pre-pass. Optimize one-character ranges */
                        
                        if (d == c)
                            goto LONE_SINGLE_CHARACTER;  /* A few lines below */
                        
                        /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless
                         matching, we have to use an XCLASS with extra data items. Caseless
                         matching for characters > 127 is available only if UCP support is
                         available. */
                        
                        if ((d > 255 || ((options & IgnoreCaseOption) && d > 127))) {
                            class_utf8 = true;
                            
                            /* With UCP support, we can find the other case equivalents of
                             the relevant characters. There may be several ranges. Optimize how
                             they fit with the basic range. */
                            
                            if (options & IgnoreCaseOption) {
                                int occ, ocd;
                                int cc = c;
                                int origd = d;
                                while (getOthercaseRange(&cc, origd, &occ, &ocd)) {
                                    if (occ >= c && ocd <= d)
                                        continue;  /* Skip embedded ranges */
                                    
                                    if (occ < c  && ocd >= c - 1)        /* Extend the basic range */
                                    {                                  /* if there is overlap,   */
                                        c = occ;                           /* noting that if occ < c */
                                        continue;                          /* we can't have ocd > d  */
                                    }                                  /* because a subrange is  */
                                    if (ocd > d && occ <= d + 1)         /* always shorter than    */
                                    {                                  /* the basic range.       */
                                        d = ocd;
                                        continue;
                                    }
                                    
                                    if (occ == ocd)
                                        *class_utf8data++ = XCL_SINGLE;
                                    else {
                                        *class_utf8data++ = XCL_RANGE;
                                        class_utf8data += encodeUTF8(occ, class_utf8data);
                                    }
                                    class_utf8data += encodeUTF8(ocd, class_utf8data);
                                }
                            }
                            
                            /* Now record the original range, possibly modified for UCP caseless
                             overlapping ranges. */
                            
                            *class_utf8data++ = XCL_RANGE;
                            class_utf8data += encodeUTF8(c, class_utf8data);
                            class_utf8data += encodeUTF8(d, class_utf8data);
                            
                            /* With UCP support, we are done. Without UCP support, there is no
                             caseless matching for UTF-8 characters > 127; we can use the bit map
                             for the smaller ones. */
                            
                            continue;    /* With next character in the class */
                        }
                        
                        /* We use the bit map for all cases when not in UTF-8 mode; else
                         ranges that lie entirely within 0-127 when there is UCP support; else
                         for partial ranges without UCP support. */
                        
                        for (; c <= d; c++) {
                            classbits[c/8] |= (1 << (c&7));