/[pcre]/code/trunk/pcre_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_compile.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 556 by ph10, Tue Oct 26 11:06:44 2010 UTC revision 602 by ph10, Wed May 25 08:29:03 2011 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2010 University of Cambridge             Copyright (c) 1997-2011 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 408  static const char error_texts[] = Line 408  static const char error_texts[] =
408    "different names for subpatterns of the same number are not allowed\0"    "different names for subpatterns of the same number are not allowed\0"
409    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
410    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with PCRE_UCP support\0"
411      "\\c must be followed by an ASCII character\0"
412    ;    ;
413    
414  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 841  else Line 842  else
842      break;      break;
843    
844      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.
845      This coding is ASCII-specific, but then the whole concept of \cx is      An error is given if the byte following \c is not an ASCII character. This
846        coding is ASCII-specific, but then the whole concept of \cx is
847      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */
848    
849      case CHAR_c:      case CHAR_c:
# Line 851  else Line 853  else
853        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
854        break;        break;
855        }        }
856    #ifndef EBCDIC    /* ASCII/UTF-8 coding */
857  #ifndef EBCDIC  /* ASCII/UTF-8 coding */      if (c > 127)  /* Excludes all non-ASCII in either mode */
858          {
859          *errorcodeptr = ERR68;
860          break;
861          }
862      if (c >= CHAR_a && c <= CHAR_z) c -= 32;      if (c >= CHAR_a && c <= CHAR_z) c -= 32;
863      c ^= 0x40;      c ^= 0x40;
864  #else           /* EBCDIC coding */  #else             /* EBCDIC coding */
865      if (c >= CHAR_a && c <= CHAR_z) c += 64;      if (c >= CHAR_a && c <= CHAR_z) c += 64;
866      c ^= 0xC0;      c ^= 0xC0;
867  #endif  #endif
# Line 1099  top-level call starts at the beginning o Line 1105  top-level call starts at the beginning o
1105  start at a parenthesis. It scans along a pattern's text looking for capturing  start at a parenthesis. It scans along a pattern's text looking for capturing
1106  subpatterns, and counting them. If it finds a named pattern that matches the  subpatterns, and counting them. If it finds a named pattern that matches the
1107  name it is given, it returns its number. Alternatively, if the name is NULL, it  name it is given, it returns its number. Alternatively, if the name is NULL, it
1108  returns when it reaches a given numbered subpattern. We know that if (?P< is  returns when it reaches a given numbered subpattern. Recursion is used to keep
1109  encountered, the name will be terminated by '>' because that is checked in the  track of subpatterns that reset the capturing group numbers - the (?| feature.
1110  first pass. Recursion is used to keep track of subpatterns that reset the  
1111  capturing group numbers - the (?| feature.  This function was originally called only from the second pass, in which we know
1112    that if (?< or (?' or (?P< is encountered, the name will be correctly
1113    terminated because that is checked in the first pass. There is now one call to
1114    this function in the first pass, to check for a recursive back reference by
1115    name (so that we can make the whole group atomic). In this case, we need check
1116    only up to the current position in the pattern, and that is still OK because
1117    and previous occurrences will have been checked. To make this work, the test
1118    for "end of pattern" is a check against cd->end_pattern in the main loop,
1119    instead of looking for a binary zero. This means that the special first-pass
1120    call can adjust cd->end_pattern temporarily. (Checks for binary zero while
1121    processing items within the loop are OK, because afterwards the main loop will
1122    terminate.)
1123    
1124  Arguments:  Arguments:
1125    ptrptr       address of the current character pointer (updated)    ptrptr       address of the current character pointer (updated)
# Line 1110  Arguments: Line 1127  Arguments:
1127    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1128    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1129    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1130    utf8         TRUE if we are in UTF-8 mode    utf8         TRUE if we are in UTF-8 mode
1131    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1132    
1133  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
# Line 1203  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1220  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1220    }    }
1221    
1222  /* Past any initial parenthesis handling, scan for parentheses or vertical  /* Past any initial parenthesis handling, scan for parentheses or vertical
1223  bars. */  bars. Stop if we get to cd->end_pattern. Note that this is important for the
1224    first-pass call when this value is temporarily adjusted to stop at the current
1225    position. So DO NOT change this to a test for binary zero. */
1226    
1227  for (; *ptr != 0; ptr++)  for (; ptr < cd->end_pattern; ptr++)
1228    {    {
1229    /* Skip over backslashed characters and also entire \Q...\E */    /* Skip over backslashed characters and also entire \Q...\E */
1230    
# Line 1279  for (; *ptr != 0; ptr++) Line 1298  for (; *ptr != 0; ptr++)
1298    
1299    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1300      {      {
1301      ptr++;      ptr++;
1302      while (*ptr != 0)      while (*ptr != 0)
1303        {        {
1304        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1305        ptr++;        ptr++;
1306  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1307        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
1308  #endif  #endif
1309        }        }
# Line 1342  Arguments: Line 1361  Arguments:
1361    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1362    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1363    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1364    utf8         TRUE if we are in UTF-8 mode    utf8         TRUE if we are in UTF-8 mode
1365    
1366  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1367  */  */
# Line 1378  return rc; Line 1397  return rc;
1397    
1398  /* This is called by several functions that scan a compiled expression looking  /* This is called by several functions that scan a compiled expression looking
1399  for a fixed first character, or an anchoring op code etc. It skips over things  for a fixed first character, or an anchoring op code etc. It skips over things
1400  that do not influence this. For some calls, a change of option is important.  that do not influence this. For some calls, it makes sense to skip negative
1401  For some calls, it makes sense to skip negative forward and all backward  forward and all backward assertions, and also the \b assertion; for others it
1402  assertions, and also the \b assertion; for others it does not.  does not.
1403    
1404  Arguments:  Arguments:
1405    code         pointer to the start of the group    code         pointer to the start of the group
# Line 1400  for (;;) Line 1419  for (;;)
1419    {    {
1420    switch ((int)*code)    switch ((int)*code)
1421      {      {
     case OP_OPT:  
     if (optbit > 0 && ((int)code[1] & optbit) != (*options & optbit))  
       *options = (int)code[1];  
     code += 2;  
     break;  
   
1422      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1423      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1424      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
# Line 1542  for (;;) Line 1555  for (;;)
1555      case OP_RREF:      case OP_RREF:
1556      case OP_NRREF:      case OP_NRREF:
1557      case OP_DEF:      case OP_DEF:
     case OP_OPT:  
1558      case OP_CALLOUT:      case OP_CALLOUT:
1559      case OP_SOD:      case OP_SOD:
1560      case OP_SOM:      case OP_SOM:
# Line 1550  for (;;) Line 1562  for (;;)
1562      case OP_EOD:      case OP_EOD:
1563      case OP_EODN:      case OP_EODN:
1564      case OP_CIRC:      case OP_CIRC:
1565        case OP_CIRCM:
1566      case OP_DOLL:      case OP_DOLL:
1567        case OP_DOLLM:
1568      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1569      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1570      cc += _pcre_OP_lengths[*cc];      cc += _pcre_OP_lengths[*cc];
# Line 1559  for (;;) Line 1573  for (;;)
1573      /* Handle literal characters */      /* Handle literal characters */
1574    
1575      case OP_CHAR:      case OP_CHAR:
1576      case OP_CHARNC:      case OP_CHARI:
1577      case OP_NOT:      case OP_NOT:
1578        case OP_NOTI:
1579      branchlength++;      branchlength++;
1580      cc += 2;      cc += 2;
1581  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1755  for (;;) Line 1770  for (;;)
1770      if (utf8) switch(c)      if (utf8) switch(c)
1771        {        {
1772        case OP_CHAR:        case OP_CHAR:
1773        case OP_CHARNC:        case OP_CHARI:
1774        case OP_EXACT:        case OP_EXACT:
1775          case OP_EXACTI:
1776        case OP_UPTO:        case OP_UPTO:
1777          case OP_UPTOI:
1778        case OP_MINUPTO:        case OP_MINUPTO:
1779          case OP_MINUPTOI:
1780        case OP_POSUPTO:        case OP_POSUPTO:
1781          case OP_POSUPTOI:
1782        case OP_STAR:        case OP_STAR:
1783          case OP_STARI:
1784        case OP_MINSTAR:        case OP_MINSTAR:
1785          case OP_MINSTARI:
1786        case OP_POSSTAR:        case OP_POSSTAR:
1787          case OP_POSSTARI:
1788        case OP_PLUS:        case OP_PLUS:
1789          case OP_PLUSI:
1790        case OP_MINPLUS:        case OP_MINPLUS:
1791          case OP_MINPLUSI:
1792        case OP_POSPLUS:        case OP_POSPLUS:
1793          case OP_POSPLUSI:
1794        case OP_QUERY:        case OP_QUERY:
1795          case OP_QUERYI:
1796        case OP_MINQUERY:        case OP_MINQUERY:
1797          case OP_MINQUERYI:
1798        case OP_POSQUERY:        case OP_POSQUERY:
1799          case OP_POSQUERYI:
1800        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1801        break;        break;
1802        }        }
# Line 1861  for (;;) Line 1889  for (;;)
1889      if (utf8) switch(c)      if (utf8) switch(c)
1890        {        {
1891        case OP_CHAR:        case OP_CHAR:
1892        case OP_CHARNC:        case OP_CHARI:
1893        case OP_EXACT:        case OP_EXACT:
1894          case OP_EXACTI:
1895        case OP_UPTO:        case OP_UPTO:
1896          case OP_UPTOI:
1897        case OP_MINUPTO:        case OP_MINUPTO:
1898          case OP_MINUPTOI:
1899        case OP_POSUPTO:        case OP_POSUPTO:
1900          case OP_POSUPTOI:
1901        case OP_STAR:        case OP_STAR:
1902          case OP_STARI:
1903        case OP_MINSTAR:        case OP_MINSTAR:
1904          case OP_MINSTARI:
1905        case OP_POSSTAR:        case OP_POSSTAR:
1906          case OP_POSSTARI:
1907        case OP_PLUS:        case OP_PLUS:
1908          case OP_PLUSI:
1909        case OP_MINPLUS:        case OP_MINPLUS:
1910          case OP_MINPLUSI:
1911        case OP_POSPLUS:        case OP_POSPLUS:
1912          case OP_POSPLUSI:
1913        case OP_QUERY:        case OP_QUERY:
1914          case OP_QUERYI:
1915        case OP_MINQUERY:        case OP_MINQUERY:
1916          case OP_MINQUERYI:
1917        case OP_POSQUERY:        case OP_POSQUERY:
1918          case OP_POSQUERYI:
1919        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1920        break;        break;
1921        }        }
# Line 2052  for (code = first_significant_code(code Line 2093  for (code = first_significant_code(code
2093      case OP_ALLANY:      case OP_ALLANY:
2094      case OP_ANYBYTE:      case OP_ANYBYTE:
2095      case OP_CHAR:      case OP_CHAR:
2096      case OP_CHARNC:      case OP_CHARI:
2097      case OP_NOT:      case OP_NOT:
2098        case OP_NOTI:
2099      case OP_PLUS:      case OP_PLUS:
2100      case OP_MINPLUS:      case OP_MINPLUS:
2101      case OP_POSPLUS:      case OP_POSPLUS:
# Line 2101  for (code = first_significant_code(code Line 2143  for (code = first_significant_code(code
2143    
2144  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2145      case OP_STAR:      case OP_STAR:
2146        case OP_STARI:
2147      case OP_MINSTAR:      case OP_MINSTAR:
2148        case OP_MINSTARI:
2149      case OP_POSSTAR:      case OP_POSSTAR:
2150        case OP_POSSTARI:
2151      case OP_QUERY:      case OP_QUERY:
2152        case OP_QUERYI:
2153      case OP_MINQUERY:      case OP_MINQUERY:
2154        case OP_MINQUERYI:
2155      case OP_POSQUERY:      case OP_POSQUERY:
2156        case OP_POSQUERYI:
2157      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];
2158      break;      break;
2159    
2160      case OP_UPTO:      case OP_UPTO:
2161        case OP_UPTOI:
2162      case OP_MINUPTO:      case OP_MINUPTO:
2163        case OP_MINUPTOI:
2164      case OP_POSUPTO:      case OP_POSUPTO:
2165        case OP_POSUPTOI:
2166      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];
2167      break;      break;
2168  #endif  #endif
# Line 2526  if ((options & PCRE_EXTENDED) != 0) Line 2577  if ((options & PCRE_EXTENDED) != 0)
2577      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2578      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
2579        {        {
2580        ptr++;        ptr++;
2581        while (*ptr != 0)        while (*ptr != 0)
2582          {          {
2583          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
2584          ptr++;          ptr++;
2585  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2586          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
2587  #endif  #endif
2588          }          }
# Line 2570  if ((options & PCRE_EXTENDED) != 0) Line 2621  if ((options & PCRE_EXTENDED) != 0)
2621      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2622      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
2623        {        {
2624        ptr++;        ptr++;
2625        while (*ptr != 0)        while (*ptr != 0)
2626          {          {
2627          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
2628          ptr++;          ptr++;
2629  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2630          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
2631  #endif  #endif
2632          }          }
# Line 2603  if (next >= 0) switch(op_code) Line 2654  if (next >= 0) switch(op_code)
2654  #endif  #endif
2655    return c != next;    return c != next;
2656    
2657    /* For CHARNC (caseless character) we must check the other case. If we have    /* For CHARI (caseless character) we must check the other case. If we have
2658    Unicode property support, we can use it to test the other case of    Unicode property support, we can use it to test the other case of
2659    high-valued characters. */    high-valued characters. */
2660    
2661    case OP_CHARNC:    case OP_CHARI:
2662  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2663    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
2664  #else  #else
# Line 2630  if (next >= 0) switch(op_code) Line 2681  if (next >= 0) switch(op_code)
2681  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2682    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */
2683    
2684    /* For OP_NOT, its data is always a single-byte character. */    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These
2685      opcodes are not used for multi-byte characters, because they are coded using
2686      an XCLASS instead. */
2687    
2688    case OP_NOT:    case OP_NOT:
2689      return (c = *previous) == next;
2690    
2691      case OP_NOTI:
2692    if ((c = *previous) == next) return TRUE;    if ((c = *previous) == next) return TRUE;
   if ((options & PCRE_CASELESS) == 0) return FALSE;  
2693  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2694    if (utf8)    if (utf8)
2695      {      {
# Line 2739  replaced by OP_PROP codes when PCRE_UCP Line 2794  replaced by OP_PROP codes when PCRE_UCP
2794  switch(op_code)  switch(op_code)
2795    {    {
2796    case OP_CHAR:    case OP_CHAR:
2797    case OP_CHARNC:    case OP_CHARI:
2798  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2799    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
2800  #else  #else
# Line 3151  for (;; ptr++) Line 3206  for (;; ptr++)
3206      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if ((cd->ctypes[c] & ctype_space) != 0) continue;
3207      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3208        {        {
3209        ptr++;        ptr++;
3210        while (*ptr != 0)        while (*ptr != 0)
3211          {          {
3212          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3213          ptr++;          ptr++;
3214  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3215          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
3216  #endif  #endif
3217          }          }
# Line 3203  for (;; ptr++) Line 3258  for (;; ptr++)
3258      the setting of any following char as a first character. */      the setting of any following char as a first character. */
3259    
3260      case CHAR_CIRCUMFLEX_ACCENT:      case CHAR_CIRCUMFLEX_ACCENT:
3261        previous = NULL;
3262      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3263        {        {
3264        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
3265          *code++ = OP_CIRCM;
3266        }        }
3267      previous = NULL;      else *code++ = OP_CIRC;
     *code++ = OP_CIRC;  
3268      break;      break;
3269    
3270      case CHAR_DOLLAR_SIGN:      case CHAR_DOLLAR_SIGN:
3271      previous = NULL;      previous = NULL;
3272      *code++ = OP_DOLL;      *code++ = ((options & PCRE_MULTILINE) != 0)? OP_DOLLM : OP_DOLL;
3273      break;      break;
3274    
3275      /* There can never be a first char if '.' is first, whatever happens about      /* There can never be a first char if '.' is first, whatever happens about
# Line 3534  for (;; ptr++) Line 3590  for (;; ptr++)
3590              continue;              continue;
3591    
3592              /* Perl 5.004 onwards omits VT from \s, but we must preserve it              /* Perl 5.004 onwards omits VT from \s, but we must preserve it
3593              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
3594              class. */              class. */
3595    
3596              case ESC_s:              case ESC_s:
3597              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
3598              classbits[1] |= cbits[cbit_space+1] & ~0x08;              classbits[1] |= cbits[cbit_space+1] & ~0x08;
3599              for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];              for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
3600              continue;              continue;
3601    
# Line 3959  for (;; ptr++) Line 4015  for (;; ptr++)
4015    
4016      In UTF-8 mode, we can optimize the negative case only if there were no      In UTF-8 mode, we can optimize the negative case only if there were no
4017      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
4018      operate on single-bytes only. This is an historical hangover. Maybe one day      operate on single-bytes characters only. This is an historical hangover.
4019      we can tidy these opcodes to handle multi-byte characters.      Maybe one day we can tidy these opcodes to handle multi-byte characters.
4020    
4021      The optimization throws away the bit map. We turn the item into a      The optimization throws away the bit map. We turn the item into a
4022      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note      1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.
4023      that OP_NOT does not support multibyte characters. In the positive case, it      Note that OP_NOT[I] does not support multibyte characters. In the positive
4024      can cause firstbyte to be set. Otherwise, there can be no first char if      case, it can cause firstbyte to be set. Otherwise, there can be no first
4025      this item is first, whatever repeat count may follow. In the case of      char if this item is first, whatever repeat count may follow. In the case
4026      reqbyte, save the previous value for reinstating. */      of reqbyte, save the previous value for reinstating. */
4027    
4028  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
4029      if (class_charcount == 1 && !class_utf8 &&      if (class_charcount == 1 && !class_utf8 &&
# Line 3978  for (;; ptr++) Line 4034  for (;; ptr++)
4034        {        {
4035        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
4036    
4037        /* The OP_NOT opcode works on one-byte characters only. */        /* The OP_NOT[I] opcodes work on one-byte characters only. */
4038    
4039        if (negate_class)        if (negate_class)
4040          {          {
4041          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
4042          zerofirstbyte = firstbyte;          zerofirstbyte = firstbyte;
4043          *code++ = OP_NOT;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4044          *code++ = class_lastchar;          *code++ = class_lastchar;
4045          break;          break;
4046          }          }
# Line 4142  for (;; ptr++) Line 4198  for (;; ptr++)
4198      the first thing in a branch because the x will have gone into firstbyte      the first thing in a branch because the x will have gone into firstbyte
4199      instead.  */      instead.  */
4200    
4201      if (*previous == OP_CHAR || *previous == OP_CHARNC)      if (*previous == OP_CHAR || *previous == OP_CHARI)
4202        {        {
4203          op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;
4204    
4205        /* Deal with UTF-8 characters that take up more than one byte. It's        /* Deal with UTF-8 characters that take up more than one byte. It's
4206        easier to write this out separately than try to macrify it. Use c to        easier to write this out separately than try to macrify it. Use c to
4207        hold the length of the character in bytes, plus 0x80 to flag that it's a        hold the length of the character in bytes, plus 0x80 to flag that it's a
# Line 4188  for (;; ptr++) Line 4246  for (;; ptr++)
4246      /* If previous was a single negated character ([^a] or similar), we use      /* If previous was a single negated character ([^a] or similar), we use
4247      one of the special opcodes, replacing it. The code is shared with single-      one of the special opcodes, replacing it. The code is shared with single-
4248      character repeats by setting opt_type to add a suitable offset into      character repeats by setting opt_type to add a suitable offset into
4249      repeat_type. We can also test for auto-possessification. OP_NOT is      repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI
4250      currently used only for single-byte chars. */      are currently used only for single-byte chars. */
4251    
4252      else if (*previous == OP_NOT)      else if (*previous == OP_NOT || *previous == OP_NOTI)
4253        {        {
4254        op_type = OP_NOTSTAR - OP_STAR;  /* Use "not" opcodes */        op_type = ((*previous == OP_NOT)? OP_NOTSTAR : OP_NOTSTARI) - OP_STAR;
4255        c = previous[1];        c = previous[1];
4256        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4257            repeat_max < 0 &&            repeat_max < 0 &&
# Line 4390  for (;; ptr++) Line 4448  for (;; ptr++)
4448  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
4449               *previous == OP_XCLASS ||               *previous == OP_XCLASS ||
4450  #endif  #endif
4451               *previous == OP_REF)               *previous == OP_REF ||
4452                 *previous == OP_REFI)
4453        {        {
4454        if (repeat_max == 0)        if (repeat_max == 0)
4455          {          {
# Line 4444  for (;; ptr++) Line 4503  for (;; ptr++)
4503    
4504        /* If the maximum repeat count is unlimited, find the end of the bracket        /* If the maximum repeat count is unlimited, find the end of the bracket
4505        by scanning through from the start, and compute the offset back to it        by scanning through from the start, and compute the offset back to it
4506        from the current code pointer. There may be an OP_OPT setting following        from the current code pointer. */
       the final KET, so we can't find the end just by going back from the code  
       pointer. */  
4507    
4508        if (repeat_max == -1)        if (repeat_max == -1)
4509          {          {
# Line 4746  for (;; ptr++) Line 4803  for (;; ptr++)
4803          case OP_QUERY: *tempcode = OP_POSQUERY; break;          case OP_QUERY: *tempcode = OP_POSQUERY; break;
4804          case OP_UPTO:  *tempcode = OP_POSUPTO; break;          case OP_UPTO:  *tempcode = OP_POSUPTO; break;
4805    
4806          case OP_TYPESTAR:  *tempcode = OP_TYPEPOSSTAR; break;          case OP_STARI:  *tempcode = OP_POSSTARI; break;
4807          case OP_TYPEPLUS:  *tempcode = OP_TYPEPOSPLUS; break;          case OP_PLUSI:  *tempcode = OP_POSPLUSI; break;
4808          case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break;          case OP_QUERYI: *tempcode = OP_POSQUERYI; break;
4809          case OP_TYPEUPTO:  *tempcode = OP_TYPEPOSUPTO; break;          case OP_UPTOI:  *tempcode = OP_POSUPTOI; break;
4810    
4811          case OP_NOTSTAR:  *tempcode = OP_NOTPOSSTAR; break;          case OP_NOTSTAR:  *tempcode = OP_NOTPOSSTAR; break;
4812          case OP_NOTPLUS:  *tempcode = OP_NOTPOSPLUS; break;          case OP_NOTPLUS:  *tempcode = OP_NOTPOSPLUS; break;
4813          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;
4814          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;
4815    
4816            case OP_NOTSTARI:  *tempcode = OP_NOTPOSSTARI; break;
4817            case OP_NOTPLUSI:  *tempcode = OP_NOTPOSPLUSI; break;
4818            case OP_NOTQUERYI: *tempcode = OP_NOTPOSQUERYI; break;
4819            case OP_NOTUPTOI:  *tempcode = OP_NOTPOSUPTOI; break;
4820    
4821            case OP_TYPESTAR:  *tempcode = OP_TYPEPOSSTAR; break;
4822            case OP_TYPEPLUS:  *tempcode = OP_TYPEPOSPLUS; break;
4823            case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break;
4824            case OP_TYPEUPTO:  *tempcode = OP_TYPEPOSUPTO; break;
4825    
4826          /* Because we are moving code along, we must ensure that any          /* Because we are moving code along, we must ensure that any
4827          pending recursive references are updated. */          pending recursive references are updated. */
4828    
# Line 4856  for (;; ptr++) Line 4923  for (;; ptr++)
4923              if (*code++ == OP_THEN)              if (*code++ == OP_THEN)
4924                {                {
4925                PUT(code, 0, code - bcptr->current_branch - 1);                PUT(code, 0, code - bcptr->current_branch - 1);
4926                code += LINK_SIZE;                code += LINK_SIZE;
4927                }                }
4928              }              }
4929    
4930            else            else
# Line 4871  for (;; ptr++) Line 4938  for (;; ptr++)
4938              if (*code++ == OP_THEN_ARG)              if (*code++ == OP_THEN_ARG)
4939                {                {
4940                PUT(code, 0, code - bcptr->current_branch - 1);                PUT(code, 0, code - bcptr->current_branch - 1);
4941                code += LINK_SIZE;                code += LINK_SIZE;
4942                }                }
4943              *code++ = arglen;              *code++ = arglen;
4944              memcpy(code, arg, arglen);              memcpy(code, arg, arglen);
4945              code += arglen;              code += arglen;
# Line 5367  for (;; ptr++) Line 5434  for (;; ptr++)
5434          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5435          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
5436    
5437          /* In the pre-compile phase, do a syntax check and set a dummy          /* In the pre-compile phase, do a syntax check. We used to just set
5438          reference number. */          a dummy reference number, because it was not used in the first pass.
5439            However, with the change of recursive back references to be atomic,
5440            we have to look for the number so that this state can be identified, as
5441            otherwise the incorrect length is computed. If it's not a backwards
5442            reference, the dummy number will do. */
5443    
5444          if (lengthptr != NULL)          if (lengthptr != NULL)
5445            {            {
5446              const uschar *temp;
5447    
5448            if (namelen == 0)            if (namelen == 0)
5449              {              {
5450              *errorcodeptr = ERR62;              *errorcodeptr = ERR62;
# Line 5387  for (;; ptr++) Line 5460  for (;; ptr++)
5460              *errorcodeptr = ERR48;              *errorcodeptr = ERR48;
5461              goto FAILED;              goto FAILED;
5462              }              }
5463            recno = 0;  
5464              /* The name table does not exist in the first pass, so we cannot
5465              do a simple search as in the code below. Instead, we have to scan the
5466              pattern to find the number. It is important that we scan it only as
5467              far as we have got because the syntax of named subpatterns has not
5468              been checked for the rest of the pattern, and find_parens() assumes
5469              correct syntax. In any case, it's a waste of resources to scan
5470              further. We stop the scan at the current point by temporarily
5471              adjusting the value of cd->endpattern. */
5472    
5473              temp = cd->end_pattern;
5474              cd->end_pattern = ptr;
5475              recno = find_parens(cd, name, namelen,
5476                (options & PCRE_EXTENDED) != 0, utf8);
5477              cd->end_pattern = temp;
5478              if (recno < 0) recno = 0;    /* Forward ref; set dummy number */
5479            }            }
5480    
5481          /* In the real compile, seek the name in the table. We check the name          /* In the real compile, seek the name in the table. We check the name
# Line 5639  for (;; ptr++) Line 5727  for (;; ptr++)
5727              }              }
5728            else            else
5729              {              {
             if ((options & PCRE_IMS) != (newoptions & PCRE_IMS))  
               {  
               *code++ = OP_OPT;  
               *code++ = newoptions & PCRE_IMS;  
               }  
5730              greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);              greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);
5731              greedy_non_default = greedy_default ^ 1;              greedy_non_default = greedy_default ^ 1;
5732              req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;              req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;
# Line 5981  for (;; ptr++) Line 6064  for (;; ptr++)
6064          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6065          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
6066          previous = code;          previous = code;
6067          *code++ = OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6068          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
6069          cd->backref_map |= (recno < 32)? (1 << recno) : 1;          cd->backref_map |= (recno < 32)? (1 << recno) : 1;
6070          if (recno > cd->top_backref) cd->top_backref = recno;          if (recno > cd->top_backref) cd->top_backref = recno;
# Line 6089  for (;; ptr++) Line 6172  for (;; ptr++)
6172    
6173      ONE_CHAR:      ONE_CHAR:
6174      previous = code;      previous = code;
6175      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6176      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6177    
6178      /* Remember if \r or \n were seen */      /* Remember if \r or \n were seen */
# Line 6153  return FALSE; Line 6236  return FALSE;
6236  /* On entry, ptr is pointing past the bracket character, but on return it  /* On entry, ptr is pointing past the bracket character, but on return it
6237  points to the closing bracket, or vertical bar, or end of string. The code  points to the closing bracket, or vertical bar, or end of string. The code
6238  variable is pointing at the byte into which the BRA operator has been stored.  variable is pointing at the byte into which the BRA operator has been stored.
 If the ims options are changed at the start (for a (?ims: group) or during any  
 branch, we need to insert an OP_OPT item at the start of every following branch  
 to ensure they get set correctly at run time, and also pass the new options  
 into every subsequent branch compile.  
   
6239  This function is used during the pre-compile phase when we are trying to find  This function is used during the pre-compile phase when we are trying to find
6240  out the amount of memory needed, as well as during the real compile phase. The  out the amount of memory needed, as well as during the real compile phase. The
6241  value of lengthptr distinguishes the two phases.  value of lengthptr distinguishes the two phases.
# Line 6249  for (;;) Line 6327  for (;;)
6327    
6328    if (reset_bracount) cd->bracount = orig_bracount;    if (reset_bracount) cd->bracount = orig_bracount;
6329    
   /* Handle a change of ims options at the start of the branch */  
   
   if ((options & PCRE_IMS) != oldims)  
     {  
     *code++ = OP_OPT;  
     *code++ = options & PCRE_IMS;  
     length += 2;  
     }  
   
6330    /* Set up dummy OP_REVERSE if lookbehind assertion */    /* Set up dummy OP_REVERSE if lookbehind assertion */
6331    
6332    if (lookbehind)    if (lookbehind)
# Line 6414  for (;;) Line 6483  for (;;)
6483        cd->open_caps = cd->open_caps->next;        cd->open_caps = cd->open_caps->next;
6484        }        }
6485    
     /* Reset options if needed. */  
   
     if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)  
       {  
       *code++ = OP_OPT;  
       *code++ = oldims;  
       length += 2;  
       }  
   
6486      /* Retain the highest bracket number, in case resetting was used. */      /* Retain the highest bracket number, in case resetting was used. */
6487    
6488      cd->bracount = max_bracount;      cd->bracount = max_bracount;
# Line 6482  for (;;) Line 6542  for (;;)
6542  /* Try to find out if this is an anchored regular expression. Consider each  /* Try to find out if this is an anchored regular expression. Consider each
6543  alternative branch. If they all start with OP_SOD or OP_CIRC, or with a bracket  alternative branch. If they all start with OP_SOD or OP_CIRC, or with a bracket
6544  all of whose alternatives start with OP_SOD or OP_CIRC (recurse ad lib), then  all of whose alternatives start with OP_SOD or OP_CIRC (recurse ad lib), then
6545  it's anchored. However, if this is a multiline pattern, then only OP_SOD  it's anchored. However, if this is a multiline pattern, then only OP_SOD will
6546  counts, since OP_CIRC can match in the middle.  be found, because ^ generates OP_CIRCM in that mode.
6547    
6548  We can also consider a regex to be anchored if OP_SOM starts all its branches.  We can also consider a regex to be anchored if OP_SOM starts all its branches.
6549  This is the code for \G, which means "match at start of match position, taking  This is the code for \G, which means "match at start of match position, taking
# Line 6557  do { Line 6617  do {
6617    
6618     /* Check for explicit anchoring */     /* Check for explicit anchoring */
6619    
6620     else if (op != OP_SOD && op != OP_SOM &&     else if (op != OP_SOD && op != OP_SOM && op != OP_CIRC) return FALSE;
            ((*options & PCRE_MULTILINE) != 0 || op != OP_CIRC))  
      return FALSE;  
6621     code += GET(code, 1);     code += GET(code, 1);
6622     }     }
6623  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 6659  do { Line 6717  do {
6717    
6718     /* Check for explicit circumflex */     /* Check for explicit circumflex */
6719    
6720     else if (op != OP_CIRC) return FALSE;     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;
6721    
6722     /* Move on to the next alternative */     /* Move on to the next alternative */
6723    
# Line 6720  do { Line 6778  do {
6778       scode += 2;       scode += 2;
6779    
6780       case OP_CHAR:       case OP_CHAR:
6781       case OP_CHARNC:       case OP_CHARI:
6782       case OP_PLUS:       case OP_PLUS:
6783       case OP_MINPLUS:       case OP_MINPLUS:
6784       case OP_POSPLUS:       case OP_POSPLUS:
# Line 6853  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6911  while (ptr[skipatstart] == CHAR_LEFT_PAR
6911      { skipatstart += 7; options |= PCRE_UTF8; continue; }      { skipatstart += 7; options |= PCRE_UTF8; continue; }
6912    else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)    else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)
6913      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
6914      else if (strncmp((char *)(ptr+skipatstart+2), STRING_NO_START_OPT_RIGHTPAR, 13) == 0)
6915        { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; }
6916    
6917    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
6918      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
# Line 6879  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6939  while (ptr[skipatstart] == CHAR_LEFT_PAR
6939    
6940  utf8 = (options & PCRE_UTF8) != 0;  utf8 = (options & PCRE_UTF8) != 0;
6941    
6942  /* Can't support UTF8 unless PCRE has been compiled to include the code. */  /* Can't support UTF8 unless PCRE has been compiled to include the code. The
6943    return of an error code from _pcre_valid_utf8() is a new feature, introduced in
6944    release 8.13. The only use we make of it here is to adjust the offset value to
6945    the end of the string for a short string error, for compatibility with previous
6946    versions. */
6947    
6948  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
6949  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 &&
6950       (*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1)) >= 0)       (*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1, &errorcode)) >= 0)
6951    {    {
6952    errorcode = ERR44;    errorcode = ERR44;
6953    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;

Legend:
Removed from v.556  
changed lines
  Added in v.602

  ViewVC Help
Powered by ViewVC 1.1.5