/[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 1114 by chpe, Tue Oct 16 15:57:16 2012 UTC revision 1219 by ph10, Sun Nov 11 18:04:37 2012 UTC
# Line 667  find_error_text(int n) Line 667  find_error_text(int n)
667  const char *s = error_texts;  const char *s = error_texts;
668  for (; n > 0; n--)  for (; n > 0; n--)
669    {    {
670    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
671    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
672    }    }
673  return s;  return s;
674  }  }
# Line 788  ptr--;                            /* Set Line 788  ptr--;                            /* Set
788    
789  /* If backslash is at the end of the pattern, it's an error. */  /* If backslash is at the end of the pattern, it's an error. */
790    
791  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
792    
793  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
794  in a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
# Line 898  else Line 898  else
898      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
899        {        {
900        const pcre_uchar *p;        const pcre_uchar *p;
901        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
902          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
903        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
904          {          {
905          escape = ESC_k;          escape = ESC_k;
906          break;          break;
# Line 1134  else Line 1134  else
1134    
1135      case CHAR_c:      case CHAR_c:
1136      c = *(++ptr);      c = *(++ptr);
1137      if (c == 0)      if (c == CHAR_NULL)
1138        {        {
1139        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1140        break;        break;
# Line 1203  escape sequence. Line 1203  escape sequence.
1203  Argument:  Argument:
1204    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1205    negptr         points to a boolean that is set TRUE for negation else FALSE    negptr         points to a boolean that is set TRUE for negation else FALSE
1206    dptr           points to an int that is set to the detailed property value    ptypeptr       points to an unsigned int that is set to the type value
1207      pdataptr       points to an unsigned int that is set to the detailed property value
1208    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1209    
1210  Returns:         type value from ucp_type_table, or -1 for an invalid type  Returns:         TRUE if the type value was found, or FALSE for an invalid type
1211  */  */
1212    
1213  static int  static BOOL
1214  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1215      unsigned int *pdataptr, int *errorcodeptr)
1216  {  {
1217  pcre_uchar c;  pcre_uchar c;
1218  int i, bot, top;  int i, bot, top;
# Line 1218  const pcre_uchar *ptr = *ptrptr; Line 1220  const pcre_uchar *ptr = *ptrptr;
1220  pcre_uchar name[32];  pcre_uchar name[32];
1221    
1222  c = *(++ptr);  c = *(++ptr);
1223  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1224    
1225  *negptr = FALSE;  *negptr = FALSE;
1226    
# Line 1235  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1237  if (c == CHAR_LEFT_CURLY_BRACKET)
1237    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1238      {      {
1239      c = *(++ptr);      c = *(++ptr);
1240      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1241      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1242      name[i] = c;      name[i] = c;
1243      }      }
# Line 1265  while (bot < top) Line 1267  while (bot < top)
1267    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1268    if (r == 0)    if (r == 0)
1269      {      {
1270      *dptr = PRIV(utt)[i].value;      *ptypeptr = PRIV(utt)[i].type;
1271      return PRIV(utt)[i].type;      *pdataptr = PRIV(utt)[i].value;
1272        return TRUE;
1273      }      }
1274    if (r > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1275    }    }
1276    
1277  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1278  *ptrptr = ptr;  *ptrptr = ptr;
1279  return -1;  return FALSE;
1280    
1281  ERROR_RETURN:  ERROR_RETURN:
1282  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1283  *ptrptr = ptr;  *ptrptr = ptr;
1284  return -1;  return FALSE;
1285  }  }
1286  #endif  #endif
1287    
# Line 1429  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1432  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1432    
1433    else if (ptr[2] == CHAR_NUMBER_SIGN)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1434      {      {
1435      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;      for (ptr += 3; *ptr != CHAR_NULL; ptr++)
1436          if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1437      goto FAIL_EXIT;      goto FAIL_EXIT;
1438      }      }
1439    
# Line 1442  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1446  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1446      ptr += 2;      ptr += 2;
1447      if (ptr[1] != CHAR_QUESTION_MARK)      if (ptr[1] != CHAR_QUESTION_MARK)
1448        {        {
1449        while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;        while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1450        if (*ptr != 0) ptr++;        if (*ptr != CHAR_NULL) ptr++;
1451        }        }
1452      }      }
1453    
# Line 1486  for (; ptr < cd->end_pattern; ptr++) Line 1490  for (; ptr < cd->end_pattern; ptr++)
1490    
1491    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1492      {      {
1493      if (*(++ptr) == 0) goto FAIL_EXIT;      if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1494      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1495        {        {
1496        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1497        if (*ptr == 0) goto FAIL_EXIT;        if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1498        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1499        }        }
1500      continue;      continue;
# Line 1534  for (; ptr < cd->end_pattern; ptr++) Line 1538  for (; ptr < cd->end_pattern; ptr++)
1538    
1539      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1540        {        {
1541        if (*ptr == 0) return -1;        if (*ptr == CHAR_NULL) return -1;
1542        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1543          {          {
1544          if (*(++ptr) == 0) goto FAIL_EXIT;          if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1545          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1546            {            {
1547            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1548            if (*ptr == 0) goto FAIL_EXIT;            if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1549            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1550            }            }
1551          continue;          continue;
# Line 1555  for (; ptr < cd->end_pattern; ptr++) Line 1559  for (; ptr < cd->end_pattern; ptr++)
1559    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1560      {      {
1561      ptr++;      ptr++;
1562      while (*ptr != 0)      while (*ptr != CHAR_NULL)
1563        {        {
1564        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1565        ptr++;        ptr++;
# Line 1563  for (; ptr < cd->end_pattern; ptr++) Line 1567  for (; ptr < cd->end_pattern; ptr++)
1567        if (utf) FORWARDCHAR(ptr);        if (utf) FORWARDCHAR(ptr);
1568  #endif  #endif
1569        }        }
1570      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1571      continue;      continue;
1572      }      }
1573    
# Line 1573  for (; ptr < cd->end_pattern; ptr++) Line 1577  for (; ptr < cd->end_pattern; ptr++)
1577      {      {
1578      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1579      if (rc > 0) return rc;      if (rc > 0) return rc;
1580      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1581      }      }
1582    
1583    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
# Line 1638  matching closing parens. That is why we Line 1642  matching closing parens. That is why we
1642  for (;;)  for (;;)
1643    {    {
1644    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1645    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == CHAR_NULL) break;
1646    }    }
1647    
1648  return rc;  return rc;
# Line 1911  for (;;) Line 1915  for (;;)
1915    
1916      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1917    
 #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  
     case OP_XCLASS:  
     cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];  
     /* Fall through */  
 #endif  
   
1918      case OP_CLASS:      case OP_CLASS:
1919      case OP_NCLASS:      case OP_NCLASS:
1920    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1921        case OP_XCLASS:
1922        /* The original code caused an unsigned overflow in 64 bit systems,
1923        so now we use a conditional statement. */
1924        if (op == OP_XCLASS)
1925          cc += GET(cc, 1);
1926        else
1927          cc += PRIV(OP_lengths)[OP_CLASS];
1928    #else
1929      cc += PRIV(OP_lengths)[OP_CLASS];      cc += PRIV(OP_lengths)[OP_CLASS];
1930    #endif
1931    
1932      switch (*cc)      switch (*cc)
1933        {        {
# Line 2710  check_posix_syntax(const pcre_uchar *ptr Line 2718  check_posix_syntax(const pcre_uchar *ptr
2718  {  {
2719  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2720  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2721  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
2722    {    {
2723    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2724      ptr++;      ptr++;
# Line 2915  get_othercase_range(pcre_uint32 *cptr, p Line 2923  get_othercase_range(pcre_uint32 *cptr, p
2923    pcre_uint32 *odptr)    pcre_uint32 *odptr)
2924  {  {
2925  pcre_uint32 c, othercase, next;  pcre_uint32 c, othercase, next;
2926  int co;  unsigned int co;
2927    
2928  /* Find the first character that has an other case. If it has multiple other  /* Find the first character that has an other case. If it has multiple other
2929  cases, return its case offset value. */  cases, return its case offset value. */
# Line 2926  for (c = *cptr; c <= d; c++) Line 2934  for (c = *cptr; c <= d; c++)
2934      {      {
2935      *ocptr = c++;   /* Character that has the set */      *ocptr = c++;   /* Character that has the set */
2936      *cptr = c;      /* Rest of input range */      *cptr = c;      /* Rest of input range */
2937      return co;      return (int)co;
2938      }      }
2939    if ((othercase = UCD_OTHERCASE(c)) != c) break;    if ((othercase = UCD_OTHERCASE(c)) != c) break;
2940    }    }
# Line 2966  Returns:       TRUE if auto-possessifyin Line 2974  Returns:       TRUE if auto-possessifyin
2974  */  */
2975    
2976  static BOOL  static BOOL
2977  check_char_prop(pcre_uint32 c, int ptype, int pdata, BOOL negated)  check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata, BOOL negated)
2978  {  {
2979  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2980  const pcre_uint32 *p;  const pcre_uint32 *p;
# Line 3017  switch(ptype) Line 3025  switch(ptype)
3025    p = PRIV(ucd_caseless_sets) + prop->caseset;    p = PRIV(ucd_caseless_sets) + prop->caseset;
3026    for (;;)    for (;;)
3027      {      {
3028      if ((unsigned int)c < *p) return !negated;      if (c < *p) return !negated;
3029      if ((unsigned int)c == *p++) return negated;      if (c == *p++) return negated;
3030      }      }
3031    break;  /* Control never reaches here */    break;  /* Control never reaches here */
3032  #endif  #endif
# Line 3067  if ((options & PCRE_EXTENDED) != 0) Line 3075  if ((options & PCRE_EXTENDED) != 0)
3075      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3076        {        {
3077        ptr++;        ptr++;
3078        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3079          {          {
3080          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3081          ptr++;          ptr++;
# Line 3110  if ((options & PCRE_EXTENDED) != 0) Line 3118  if ((options & PCRE_EXTENDED) != 0)
3118      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3119        {        {
3120        ptr++;        ptr++;
3121        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3122          {          {
3123          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3124          ptr++;          ptr++;
# Line 3133  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3141  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3141    
3142  if (op_code == OP_CHAR || op_code == OP_CHARI ||  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3143      op_code == OP_NOT || op_code == OP_NOTI)      op_code == OP_NOT || op_code == OP_NOTI)
   //if (escape == 0) switch(op_code)  
3144    {    {
3145  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3146    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
# Line 3153  if (escape == 0) Line 3160  if (escape == 0)
3160  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3161    if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)    if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3162      {      {
3163      int ocs = UCD_CASESET(next);      unsigned int ocs = UCD_CASESET(next);
3164      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3165      }      }
3166  #endif  #endif
# Line 3253  if (escape == 0) Line 3260  if (escape == 0)
3260    
3261  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3262      case OP_PROP:      case OP_PROP:
3263      return check_char_prop(next, (int)previous[0], (int)previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3264    
3265      case OP_NOTPROP:      case OP_NOTPROP:
3266      return check_char_prop(next, (int)previous[0], (int)previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3267  #endif  #endif
3268    
3269      default:      default:
# Line 3339  switch(op_code) Line 3346  switch(op_code)
3346      case ESC_p:      case ESC_p:
3347      case ESC_P:      case ESC_P:
3348        {        {
3349        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3350          int errorcodeptr;
3351        BOOL negated;        BOOL negated;
3352    
3353        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3354        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3355        if (ptype < 0) return FALSE;          return FALSE;
3356        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3357    
3358        /* If the property item is optional, we have to give up. (When generated        /* If the property item is optional, we have to give up. (When generated
# Line 3725  dynamically as we process the pattern. * Line 3733  dynamically as we process the pattern. *
3733  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3734  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3735  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3736    #ifndef COMPILE_PCRE32
3737  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3738    #endif
3739  #else  #else
3740  BOOL utf = FALSE;  BOOL utf = FALSE;
3741  #endif  #endif
# Line 3790  for (;; ptr++) Line 3800  for (;; ptr++)
3800    pcre_uint32 subreqchar, subfirstchar;    pcre_uint32 subreqchar, subfirstchar;
3801    pcre_int32 subreqcharflags, subfirstcharflags;    pcre_int32 subreqcharflags, subfirstcharflags;
3802    int terminator;    int terminator;
3803    int mclength;    unsigned int mclength;
3804    int tempbracount;    unsigned int tempbracount;
3805    pcre_uint32 ec;    pcre_uint32 ec;
3806    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3807    
# Line 3802  for (;; ptr++) Line 3812  for (;; ptr++)
3812    /* If we are at the end of a nested substitution, revert to the outer level    /* If we are at the end of a nested substitution, revert to the outer level
3813    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
3814    
3815    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
3816      {      {
3817      ptr = nestptr;      ptr = nestptr;
3818      nestptr = NULL;      nestptr = NULL;
# Line 3877  for (;; ptr++) Line 3887  for (;; ptr++)
3887    
3888    /* If in \Q...\E, check for the end; if not, we have a literal */    /* If in \Q...\E, check for the end; if not, we have a literal */
3889    
3890    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
3891      {      {
3892      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3893        {        {
# Line 3925  for (;; ptr++) Line 3935  for (;; ptr++)
3935      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3936        {        {
3937        ptr++;        ptr++;
3938        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3939          {          {
3940          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3941          ptr++;          ptr++;
# Line 3933  for (;; ptr++) Line 3943  for (;; ptr++)
3943          if (utf) FORWARDCHAR(ptr);          if (utf) FORWARDCHAR(ptr);
3944  #endif  #endif
3945          }          }
3946        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
3947    
3948        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
3949        c = 0;        c = 0;
# Line 4111  for (;; ptr++) Line 4121  for (;; ptr++)
4121      means that an initial ] is taken as a data character. At the start of the      means that an initial ] is taken as a data character. At the start of the
4122      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4123    
4124      if (c != 0) do      if (c != CHAR_NULL) do
4125        {        {
4126        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4127    
# Line 4374  for (;; ptr++) Line 4384  for (;; ptr++)
4384              case ESC_P:              case ESC_P:
4385                {                {
4386                BOOL negated;                BOOL negated;
4387                int pdata;                unsigned int ptype = 0, pdata = 0;
4388                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4389                if (ptype < 0) goto FAILED;                  goto FAILED;
4390                *class_uchardata++ = ((escape == ESC_p) != negated)?                *class_uchardata++ = ((escape == ESC_p) != negated)?
4391                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4392                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
# Line 4449  for (;; ptr++) Line 4459  for (;; ptr++)
4459          /* Minus (hyphen) at the end of a class is treated as a literal, so put          /* Minus (hyphen) at the end of a class is treated as a literal, so put
4460          back the pointer and jump to handle the character that preceded it. */          back the pointer and jump to handle the character that preceded it. */
4461    
4462          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4463            {            {
4464            ptr = oldptr;            ptr = oldptr;
4465            goto CLASS_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
# Line 4604  for (;; ptr++) Line 4614  for (;; ptr++)
4614      If we are at the end of an internal nested string, revert to the outer      If we are at the end of an internal nested string, revert to the outer
4615      string. */      string. */
4616    
4617      while (((c = *(++ptr)) != 0 ||      while (((c = *(++ptr)) != CHAR_NULL ||
4618             (nestptr != NULL &&             (nestptr != NULL &&
4619               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != CHAR_NULL)) &&
4620             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
4621    
4622      /* Check for missing terminating ']' */      /* Check for missing terminating ']' */
4623    
4624      if (c == 0)      if (c == CHAR_NULL)
4625        {        {
4626        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
4627        goto FAILED;        goto FAILED;
# Line 5601  for (;; ptr++) Line 5611  for (;; ptr++)
5611        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
5612          {          {
5613          arg = ++ptr;          arg = ++ptr;
5614          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5615          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5616          if ((unsigned int)arglen > MAX_MARK)          if ((unsigned int)arglen > MAX_MARK)
5617            {            {
# Line 5715  for (;; ptr++) Line 5725  for (;; ptr++)
5725          {          {
5726          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */
5727          ptr++;          ptr++;
5728          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5729          if (*ptr == 0)          if (*ptr == CHAR_NULL)
5730            {            {
5731            *errorcodeptr = ERR18;            *errorcodeptr = ERR18;
5732            goto FAILED;            goto FAILED;
# Line 5792  for (;; ptr++) Line 5802  for (;; ptr++)
5802            }            }
5803          else          else
5804            {            {
5805            terminator = 0;            terminator = CHAR_NULL;
5806            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);
5807            }            }
5808    
# Line 5817  for (;; ptr++) Line 5827  for (;; ptr++)
5827            }            }
5828          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
5829    
5830          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||
5831              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
5832            {            {
5833            ptr--;      /* Error offset */            ptr--;      /* Error offset */
# Line 5882  for (;; ptr++) Line 5892  for (;; ptr++)
5892            code[1+LINK_SIZE]++;            code[1+LINK_SIZE]++;
5893            }            }
5894    
5895          /* If terminator == 0 it means that the name followed directly after          /* If terminator == CHAR_NULL it means that the name followed directly
5896          the opening parenthesis [e.g. (?(abc)...] and in this case there are          after the opening parenthesis [e.g. (?(abc)...] and in this case there
5897          some further alternatives to try. For the cases where terminator != 0          are some further alternatives to try. For the cases where terminator !=
5898          [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have          0 [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have
5899          now checked all the possibilities, so give an error. */          now checked all the possibilities, so give an error. */
5900    
5901          else if (terminator != 0)          else if (terminator != CHAR_NULL)
5902            {            {
5903            *errorcodeptr = ERR15;            *errorcodeptr = ERR15;
5904            goto FAILED;            goto FAILED;
# Line 6057  for (;; ptr++) Line 6067  for (;; ptr++)
6067    
6068            if (lengthptr != NULL)            if (lengthptr != NULL)
6069              {              {
6070              if (*ptr != terminator)              if (*ptr != (pcre_uchar)terminator)
6071                {                {
6072                *errorcodeptr = ERR42;                *errorcodeptr = ERR42;
6073                goto FAILED;                goto FAILED;
# Line 6199  for (;; ptr++) Line 6209  for (;; ptr++)
6209              *errorcodeptr = ERR62;              *errorcodeptr = ERR62;
6210              goto FAILED;              goto FAILED;
6211              }              }
6212            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6213              {              {
6214              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
6215              goto FAILED;              goto FAILED;
# Line 6305  for (;; ptr++) Line 6315  for (;; ptr++)
6315            while(IS_DIGIT(*ptr))            while(IS_DIGIT(*ptr))
6316              recno = recno * 10 + *ptr++ - CHAR_0;              recno = recno * 10 + *ptr++ - CHAR_0;
6317    
6318            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6319              {              {
6320              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
6321              goto FAILED;              goto FAILED;
# Line 6785  for (;; ptr++) Line 6795  for (;; ptr++)
6795          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
6796            {            {
6797            BOOL is_a_number = TRUE;            BOOL is_a_number = TRUE;
6798            for (p = ptr + 1; *p != 0 && *p != terminator; p++)            for (p = ptr + 1; *p != CHAR_NULL && *p != (pcre_uchar)terminator; p++)
6799              {              {
6800              if (!MAX_255(*p)) { is_a_number = FALSE; break; }              if (!MAX_255(*p)) { is_a_number = FALSE; break; }
6801              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;
6802              if ((cd->ctypes[*p] & ctype_word) == 0) break;              if ((cd->ctypes[*p] & ctype_word) == 0) break;
6803              }              }
6804            if (*p != terminator)            if (*p != (pcre_uchar)terminator)
6805              {              {
6806              *errorcodeptr = ERR57;              *errorcodeptr = ERR57;
6807              break;              break;
# Line 6809  for (;; ptr++) Line 6819  for (;; ptr++)
6819    
6820          p = ptr + 2;          p = ptr + 2;
6821          while (IS_DIGIT(*p)) p++;          while (IS_DIGIT(*p)) p++;
6822          if (*p != terminator)          if (*p != (pcre_uchar)terminator)
6823            {            {
6824            *errorcodeptr = ERR57;            *errorcodeptr = ERR57;
6825            break;            break;
# Line 6873  for (;; ptr++) Line 6883  for (;; ptr++)
6883        else if (escape == ESC_P || escape == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6884          {          {
6885          BOOL negated;          BOOL negated;
6886          int pdata;          unsigned int ptype = 0, pdata = 0;
6887          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
6888          if (ptype < 0) goto FAILED;            goto FAILED;
6889          previous = code;          previous = code;
6890          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6891          *code++ = ptype;          *code++ = ptype;
# Line 7102  pcre_int32 firstcharflags, reqcharflags; Line 7112  pcre_int32 firstcharflags, reqcharflags;
7112  pcre_uint32 branchfirstchar, branchreqchar;  pcre_uint32 branchfirstchar, branchreqchar;
7113  pcre_int32 branchfirstcharflags, branchreqcharflags;  pcre_int32 branchfirstcharflags, branchreqcharflags;
7114  int length;  int length;
7115  int orig_bracount;  unsigned int orig_bracount;
7116  int max_bracount;  unsigned int max_bracount;
7117  branch_chain bc;  branch_chain bc;
7118    
7119  bc.outer = bcptr;  bc.outer = bcptr;
# Line 7838  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7848  while (ptr[skipatstart] == CHAR_LEFT_PAR
7848    {    {
7849    int newnl = 0;    int newnl = 0;
7850    int newbsr = 0;    int newbsr = 0;
7851    
7852    /* For completeness and backward compatibility, (*UTFn) is supported in the
7853    relevant libraries, but (*UTF) is generic and always supported. Note that
7854    PCRE_UTF8 == PCRE_UTF16 == PCRE_UTF32. */
7855    
7856  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
7857    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 5) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF8_RIGHTPAR, 5) == 0)
7858      { skipatstart += 7; options |= PCRE_UTF8; continue; }      { skipatstart += 7; options |= PCRE_UTF8; continue; }
7859  #endif  #endif
7860  #ifdef COMPILE_PCRE16  #ifdef COMPILE_PCRE16
7861    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF16_RIGHTPAR, 6) == 0)
7862      { skipatstart += 8; options |= PCRE_UTF16; continue; }      { skipatstart += 8; options |= PCRE_UTF16; continue; }
7863  #endif  #endif
7864  #ifdef COMPILE_PCRE32  #ifdef COMPILE_PCRE32
7865    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF32_RIGHTPAR, 6) == 0)
7866      { skipatstart += 8; options |= PCRE_UTF32; continue; }      { skipatstart += 8; options |= PCRE_UTF32; continue; }
7867  #endif  #endif
7868    
7869      else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 4) == 0)
7870        { skipatstart += 6; options |= PCRE_UTF8; continue; }
7871    else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UCP_RIGHTPAR, 4) == 0)    else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UCP_RIGHTPAR, 4) == 0)
7872      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
7873    else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_START_OPT_RIGHTPAR, 13) == 0)    else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_START_OPT_RIGHTPAR, 13) == 0)
# Line 8109  if (cd->had_accept) Line 8126  if (cd->had_accept)
8126    
8127  /* If not reached end of pattern on success, there's an excess bracket. */  /* If not reached end of pattern on success, there's an excess bracket. */
8128    
8129  if (errorcode == 0 && *ptr != 0) errorcode = ERR22;  if (errorcode == 0 && *ptr != CHAR_NULL) errorcode = ERR22;
8130    
8131  /* Fill in the terminating state and check for disastrous overflow, but  /* Fill in the terminating state and check for disastrous overflow, but
8132  if debugging, leave the test till after things are printed out. */  if debugging, leave the test till after things are printed out. */
# Line 8120  if debugging, leave the test till after Line 8137  if debugging, leave the test till after
8137  if (code - codestart > length) errorcode = ERR23;  if (code - codestart > length) errorcode = ERR23;
8138  #endif  #endif
8139    
8140    #ifdef SUPPORT_VALGRIND
8141    /* If the estimated length exceeds the really used length, mark the extra
8142    allocated memory as unadressable, so that any out-of-bound reads can be
8143    detected. */
8144    VALGRIND_MAKE_MEM_NOACCESS(code, (length - (code - codestart)) * sizeof(pcre_uchar));
8145    #endif
8146    
8147  /* Fill in any forward references that are required. There may be repeated  /* Fill in any forward references that are required. There may be repeated
8148  references; optimize for them, as searching a large regex takes time. */  references; optimize for them, as searching a large regex takes time. */
8149    

Legend:
Removed from v.1114  
changed lines
  Added in v.1219

  ViewVC Help
Powered by ViewVC 1.1.5