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

Diff of /code/trunk/pcre_compile.c

Parent Directory | Revision Log | View Patch Patch

-revision 528 by ph10,
Sat May 29 16:40:22 2010 UTC
+revision 598 by ph10,
Sat May  7 15:37:31 2011 UTC
 Line 6
  and semantics are as close as possible to those of the Perl 5 language.
                         Written by Philip Hazel
-            Copyright (c) 1997-2010 University of Cambridge
+            Copyright (c) 1997-2011 University of Cambridge
  -----------------------------------------------------------------------------
  Redistribution and use in source and binary forms, with or without
 Line 261 
 static const int posix_class_maps[] = {
    cbit_xdigit,-1,          0              /* xdigit */
  };
  /* Table of substitutes for \d etc when PCRE_UCP is set. The POSIX class
  substitutes must be in the order of the names, defined above, and there are
  both positive and negative cases. NULL means no substitute. */
  #ifdef SUPPORT_UCP
 Line 272 
 static const uschar *substitutes[] = {
    (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */
    (uschar *)"\\p{Xsp}",   /* \s */
    (uschar *)"\\P{Xwd}",   /* \W */
    (uschar *)"\\p{Xwd}"    /* \w */
  };
  static const uschar *posix_substitutes[] = {
    (uschar *)"\\p{L}",     /* alpha */
    (uschar *)"\\p{Ll}",    /* lower */
    (uschar *)"\\p{Lu}",    /* upper */
    (uschar *)"\\p{Xan}",   /* alnum */
    NULL,                   /* ascii */
    (uschar *)"\\h",        /* blank */
    NULL,                   /* cntrl */
 Line 289 
 static const uschar *posix_substitutes[]
    NULL,                   /* punct */
    (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */
    (uschar *)"\\p{Xwd}",   /* word */
    NULL,                   /* xdigit */
    /* Negated cases */
    (uschar *)"\\P{L}",     /* ^alpha */
    (uschar *)"\\P{Ll}",    /* ^lower */
    (uschar *)"\\P{Lu}",    /* ^upper */
    (uschar *)"\\P{Xan}",   /* ^alnum */
    NULL,                   /* ^ascii */
    (uschar *)"\\H",        /* ^blank */
    NULL,                   /* ^cntrl */
 Line 304 
 static const uschar *posix_substitutes[]
    NULL,                   /* ^punct */
    (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */
    (uschar *)"\\P{Xwd}",   /* ^word */
    NULL                    /* ^xdigit */
  };
  #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))
  #endif
  #define STRING(a)  # a
  #define XSTRING(s) STRING(s)
 Line 407 
 static const char error_texts[] =
    /* 65 */
    "different names for subpatterns of the same number are not allowed\0"
    "(*MARK) must have an argument\0"
    "this version of PCRE is not compiled with PCRE_UCP support\0"
+   "\\c must be followed by an ASCII character\0"
    ;
  /* Table to identify digits and hex digits. This is used when compiling
-Line 841 
 else
+Line 842 
 else
      break;
      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.
-     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
+     coding is ASCII-specific, but then the whole concept of \cx is
      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */
      case CHAR_c:
-Line 851 
 else
+Line 853 
 else
        *errorcodeptr = ERR2;
        break;
        }
+ #ifndef EBCDIC    /* ASCII/UTF-8 coding */
- #ifndef EBCDIC  /* ASCII/UTF-8 coding */
+     if (c > 127)  /* Excludes all non-ASCII in either mode */
+       {
+       *errorcodeptr = ERR68;
+       break;
+       }
      if (c >= CHAR_a && c <= CHAR_z) c -= 32;
      c ^= 0x40;
- #else           /* EBCDIC coding */
+ #else             /* EBCDIC coding */
      if (c >= CHAR_a && c <= CHAR_z) c += 64;
      c ^= 0xC0;
  #endif
-Line 1099 
 top-level call starts at the beginning o
+Line 1105 
 top-level call starts at the beginning o
  start at a parenthesis. It scans along a pattern's text looking for capturing
  subpatterns, and counting them. If it finds a named pattern that matches the
  name it is given, it returns its number. Alternatively, if the name is NULL, it
- 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
- encountered, the name will be terminated by '>' because that is checked in the
+ track of subpatterns that reset the capturing group numbers - the (?| feature.
- first pass. Recursion is used to keep track of subpatterns that reset the
- capturing group numbers - the (?| feature.
+ This function was originally called only from the second pass, in which we know
+ that if (?< or (?' or (?P< is encountered, the name will be correctly
+ terminated because that is checked in the first pass. There is now one call to
+ this function in the first pass, to check for a recursive back reference by
+ name (so that we can make the whole group atomic). In this case, we need check
+ only up to the current position in the pattern, and that is still OK because
+ and previous occurrences will have been checked. To make this work, the test
+ for "end of pattern" is a check against cd->end_pattern in the main loop,
+ instead of looking for a binary zero. This means that the special first-pass
+ call can adjust cd->end_pattern temporarily. (Checks for binary zero while
+ processing items within the loop are OK, because afterwards the main loop will
+ terminate.)
  Arguments:
    ptrptr       address of the current character pointer (updated)
-Line 1110 
 Arguments:
+Line 1127 
 Arguments:
    name         name to seek, or NULL if seeking a numbered subpattern
    lorn         name length, or subpattern number if name is NULL
    xmode        TRUE if we are in /x mode
+   utf8         TRUE if we are in UTF-8 mode
    count        pointer to the current capturing subpattern number (updated)
  Returns:       the number of the named subpattern, or -1 if not found
-Line 1117 
 Returns:       the number of the named s
+Line 1135 
 Returns:       the number of the named s
  static int
  find_parens_sub(uschar **ptrptr, compile_data *cd, const uschar *name, int lorn,
-   BOOL xmode, int *count)
+   BOOL xmode, BOOL utf8, int *count)
  {
  uschar *ptr = *ptrptr;
  int start_count = *count;
-Line 1129 
 dealing with. The very first call may no
+Line 1147 
 dealing with. The very first call may no
  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
    {
-   if (ptr[1] == CHAR_QUESTION_MARK &&
+   /* Handle specials such as (*SKIP) or (*UTF8) etc. */
-       ptr[2] == CHAR_VERTICAL_LINE)
+   if (ptr[1] == CHAR_ASTERISK) ptr += 2;
+   /* Handle a normal, unnamed capturing parenthesis. */
+   else if (ptr[1] != CHAR_QUESTION_MARK)
+     {
+     *count += 1;
+     if (name == NULL && *count == lorn) return *count;
+     ptr++;
+     }
+   /* All cases now have (? at the start. Remember when we are in a group
+   where the parenthesis numbers are duplicated. */
+   else if (ptr[2] == CHAR_VERTICAL_LINE)
      {
      ptr += 3;
      dup_parens = TRUE;
      }
-   /* Handle a normal, unnamed capturing parenthesis */
+   /* Handle comments; all characters are allowed until a ket is reached. */
-   else if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)
+   else if (ptr[2] == CHAR_NUMBER_SIGN)
      {
-     *count += 1;
+     for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
-     if (name == NULL && *count == lorn) return *count;
+     goto FAIL_EXIT;
-     ptr++;
      }
    /* Handle a condition. If it is an assertion, just carry on so that it
    is processed as normal. If not, skip to the closing parenthesis of the
-   condition (there can't be any nested parens. */
+   condition (there can't be any nested parens). */
    else if (ptr[2] == CHAR_LEFT_PARENTHESIS)
      {
-Line 1159 
 if (ptr[0] == CHAR_LEFT_PARENTHESIS)
+Line 1191 
 if (ptr[0] == CHAR_LEFT_PARENTHESIS)
        }
      }
-   /* We have either (? or (* and not a condition */
+   /* Start with (? but not a condition. */
    else
      {
-Line 1188 
 if (ptr[0] == CHAR_LEFT_PARENTHESIS)
+Line 1220 
 if (ptr[0] == CHAR_LEFT_PARENTHESIS)
    }
  /* Past any initial parenthesis handling, scan for parentheses or vertical
- bars. */
+ bars. Stop if we get to cd->end_pattern. Note that this is important for the
+ first-pass call when this value is temporarily adjusted to stop at the current
+ position. So DO NOT change this to a test for binary zero. */
- for (; *ptr != 0; ptr++)
+ for (; ptr < cd->end_pattern; ptr++)
    {
    /* Skip over backslashed characters and also entire \Q...\E */
-Line 1264 
 for (; *ptr != 0; ptr++)
+Line 1298 
 for (; *ptr != 0; ptr++)
    if (xmode && *ptr == CHAR_NUMBER_SIGN)
      {
-     while (*(++ptr) != 0 && *ptr != CHAR_NL) {};
+     ptr++;
+     while (*ptr != 0)
+       {
+       if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
+       ptr++;
+ #ifdef SUPPORT_UTF8
+       if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
+ #endif
+       }
      if (*ptr == 0) goto FAIL_EXIT;
      continue;
      }
-Line 1273 
 for (; *ptr != 0; ptr++)
+Line 1315 
 for (; *ptr != 0; ptr++)
    if (*ptr == CHAR_LEFT_PARENTHESIS)
      {
-     int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, count);
+     int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);
      if (rc > 0) return rc;
      if (*ptr == 0) goto FAIL_EXIT;
      }
-Line 1281 
 for (; *ptr != 0; ptr++)
+Line 1323 
 for (; *ptr != 0; ptr++)
    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
      {
      if (dup_parens && *count < hwm_count) *count = hwm_count;
-     *ptrptr = ptr;
+     goto FAIL_EXIT;
-     return -1;
      }
    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)
-Line 1320 
 Arguments:
+Line 1361 
 Arguments:
    name         name to seek, or NULL if seeking a numbered subpattern
    lorn         name length, or subpattern number if name is NULL
    xmode        TRUE if we are in /x mode
+   utf8         TRUE if we are in UTF-8 mode
  Returns:       the number of the found subpattern, or -1 if not found
  */
  static int
- find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode)
+ find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode,
+   BOOL utf8)
  {
  uschar *ptr = (uschar *)cd->start_pattern;
  int count = 0;
-Line 1338 
 matching closing parens. That is why we
+Line 1381 
 matching closing parens. That is why we
  for (;;)
    {
-   rc = find_parens_sub(&ptr, cd, name, lorn, xmode, &count);
+   rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);
    if (rc > 0 || *ptr++ == 0) break;
    }
-Line 1711 
 for (;;)
+Line 1754 
 for (;;)
        case OP_MARK:
        case OP_PRUNE_ARG:
        case OP_SKIP_ARG:
-       case OP_THEN_ARG:
        code += code[1];
        break;
+       case OP_THEN_ARG:
+       code += code[1+LINK_SIZE];
+       break;
        }
      /* Add in the fixed length from the table */
-Line 1814 
 for (;;)
+Line 1860 
 for (;;)
        case OP_MARK:
        case OP_PRUNE_ARG:
        case OP_SKIP_ARG:
-       case OP_THEN_ARG:
        code += code[1];
        break;
+       case OP_THEN_ARG:
+       code += code[1+LINK_SIZE];
+       break;
        }
      /* Add in the fixed length from the table */
-Line 2092 
 for (code = first_significant_code(code
+Line 2141 
 for (code = first_significant_code(code
      case OP_MARK:
      case OP_PRUNE_ARG:
      case OP_SKIP_ARG:
-     case OP_THEN_ARG:
      code += code[1];
      break;
+     case OP_THEN_ARG:
+     code += code[1+LINK_SIZE];
+     break;
      /* None of the remaining opcodes are required to match a character. */
      default:
-Line 2316 
 auto_callout(uschar *code, const uschar
+Line 2368 
 auto_callout(uschar *code, const uschar
  {
  *code++ = OP_CALLOUT;
  *code++ = 255;
- PUT(code, 0, ptr - cd->start_pattern);  /* Pattern offset */
+ PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
- PUT(code, LINK_SIZE, 0);                /* Default length */
+ PUT(code, LINK_SIZE, 0);                       /* Default length */
  return code + 2*LINK_SIZE;
  }
-Line 2342 
 Returns:             nothing
+Line 2394 
 Returns:             nothing
  static void
  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)
  {
- int length = ptr - cd->start_pattern - GET(previous_callout, 2);
+ int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
  PUT(previous_callout, 2 + LINK_SIZE, length);
  }
-Line 2392 
 for (++c; c <= d; c++)
+Line 2444 
 for (++c; c <= d; c++)
  return TRUE;
  }
+ /*************************************************
+ *        Check a character and a property        *
+ *************************************************/
+ /* This function is called by check_auto_possessive() when a property item
+ is adjacent to a fixed character.
+ Arguments:
+   c            the character
+   ptype        the property type
+   pdata        the data for the type
+   negated      TRUE if it's a negated property (\P or \p{^)
+ Returns:       TRUE if auto-possessifying is OK
+ */
+ static BOOL
+ check_char_prop(int c, int ptype, int pdata, BOOL negated)
+ {
+ const ucd_record *prop = GET_UCD(c);
+ switch(ptype)
+   {
+   case PT_LAMP:
+   return (prop->chartype == ucp_Lu ||
+           prop->chartype == ucp_Ll ||
+           prop->chartype == ucp_Lt) == negated;
+   case PT_GC:
+   return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;
+   case PT_PC:
+   return (pdata == prop->chartype) == negated;
+   case PT_SC:
+   return (pdata == prop->script) == negated;
+   /* These are specials */
+   case PT_ALNUM:
+   return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
+           _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;
+   case PT_SPACE:    /* Perl space */
+   return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
+           c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
+           == negated;
+   case PT_PXSPACE:  /* POSIX space */
+   return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
+           c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
+           c == CHAR_FF || c == CHAR_CR)
+           == negated;
+   case PT_WORD:
+   return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
+           _pcre_ucp_gentype[prop->chartype] == ucp_N ||
+           c == CHAR_UNDERSCORE) == negated;
+   }
+ return FALSE;
+ }
  #endif  /* SUPPORT_UCP */
-Line 2405 
 whether the next thing could possibly ma
+Line 2520 
 whether the next thing could possibly ma
  sense to automatically possessify the repeated item.
  Arguments:
-   op_code       the repeated op code
+   previous      pointer to the repeated opcode
-   this          data for this item, depends on the opcode
    utf8          TRUE in UTF-8 mode
-   utf8_char     used for utf8 character bytes, NULL if not relevant
    ptr           next character in pattern
    options       options bits
    cd            contains pointers to tables etc.
-Line 2417 
 Returns:        TRUE if possessifying is
+Line 2530 
 Returns:        TRUE if possessifying is
  */
  static BOOL
- check_auto_possessive(int op_code, int item, BOOL utf8, uschar *utf8_char,
+ check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,
-   const uschar *ptr, int options, compile_data *cd)
+   int options, compile_data *cd)
  {
- int next;
+ int c, next;
+ int op_code = *previous++;
  /* Skip whitespace and comments in extended mode */
-Line 2431 
 if ((options & PCRE_EXTENDED) != 0)
+Line 2545 
 if ((options & PCRE_EXTENDED) != 0)
      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
      if (*ptr == CHAR_NUMBER_SIGN)
        {
-       while (*(++ptr) != 0)
+       ptr++;
+       while (*ptr != 0)
+         {
          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
+         ptr++;
+ #ifdef SUPPORT_UTF8
+         if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
+ #endif
+         }
        }
      else break;
      }
-Line 2468 
 if ((options & PCRE_EXTENDED) != 0)
+Line 2589 
 if ((options & PCRE_EXTENDED) != 0)
      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
      if (*ptr == CHAR_NUMBER_SIGN)
        {
-       while (*(++ptr) != 0)
+       ptr++;
+       while (*ptr != 0)
+         {
          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
+         ptr++;
+ #ifdef SUPPORT_UTF8
+         if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
+ #endif
+         }
        }
      else break;
      }
-Line 2481 
 if (*ptr == CHAR_ASTERISK || *ptr == CHA
+Line 2609 
 if (*ptr == CHAR_ASTERISK || *ptr == CHA
    strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
      return FALSE;
- /* Now compare the next item with the previous opcode. If the previous is a
+ /* Now compare the next item with the previous opcode. First, handle cases when
- positive single character match, "item" either contains the character or, if
+ the next item is a character. */
- "item" is greater than 127 in utf8 mode, the character's bytes are in
- utf8_char. */
- /* Handle cases when the next item is a character. */
  if (next >= 0) switch(op_code)
    {
    case OP_CHAR:
  #ifdef SUPPORT_UTF8
-   if (utf8 && item > 127) { GETCHAR(item, utf8_char); }
+   GETCHARTEST(c, previous);
  #else
-   (void)(utf8_char);  /* Keep compiler happy by referencing function argument */
+   c = *previous;
  #endif
-   return item != next;
+   return c != next;
    /* For CHARNC (caseless character) we must check the other case. If we have
    Unicode property support, we can use it to test the other case of
-Line 2505 
 if (next >= 0) switch(op_code)
+Line 2628 
 if (next >= 0) switch(op_code)
    case OP_CHARNC:
  #ifdef SUPPORT_UTF8
-   if (utf8 && item > 127) { GETCHAR(item, utf8_char); }
+   GETCHARTEST(c, previous);
+ #else
+   c = *previous;
  #endif
-   if (item == next) return FALSE;
+   if (c == next) return FALSE;
  #ifdef SUPPORT_UTF8
    if (utf8)
      {
-Line 2518 
 if (next >= 0) switch(op_code)
+Line 2643 
 if (next >= 0) switch(op_code)
  #else
      othercase = NOTACHAR;
  #endif
-     return (unsigned int)item != othercase;
+     return (unsigned int)c != othercase;
      }
    else
  #endif  /* SUPPORT_UTF8 */
-   return (item != cd->fcc[next]);  /* Non-UTF-8 mode */
+   return (c != cd->fcc[next]);  /* Non-UTF-8 mode */
-   /* For OP_NOT, "item" must be a single-byte character. */
+   /* For OP_NOT, its data is always a single-byte character. */
    case OP_NOT:
-   if (item == next) return TRUE;
+   if ((c = *previous) == next) return TRUE;
    if ((options & PCRE_CASELESS) == 0) return FALSE;
  #ifdef SUPPORT_UTF8
    if (utf8)
-Line 2539 
 if (next >= 0) switch(op_code)
+Line 2664 
 if (next >= 0) switch(op_code)
  #else
      othercase = NOTACHAR;
  #endif
-     return (unsigned int)item == othercase;
+     return (unsigned int)c == othercase;
      }
    else
  #endif  /* SUPPORT_UTF8 */
-   return (item == cd->fcc[next]);  /* Non-UTF-8 mode */
+   return (c == cd->fcc[next]);  /* Non-UTF-8 mode */
    /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.
    When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
    case OP_DIGIT:
    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;
-Line 2611 
 if (next >= 0) switch(op_code)
+Line 2736 
 if (next >= 0) switch(op_code)
      return op_code != OP_NOT_VSPACE;
      }
+ #ifdef SUPPORT_UCP
+   case OP_PROP:
+   return check_char_prop(next, previous[0], previous[1], FALSE);
+   case OP_NOTPROP:
+   return check_char_prop(next, previous[0], previous[1], TRUE);
+ #endif
    default:
    return FALSE;
    }
  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP
  is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
  generated only when PCRE_UCP is *not* set, that is, when only ASCII
- characteristics are recognized. */
+ characteristics are recognized. Similarly, the opcodes OP_DIGIT etc. are
+ replaced by OP_PROP codes when PCRE_UCP is set. */
  switch(op_code)
    {
    case OP_CHAR:
    case OP_CHARNC:
  #ifdef SUPPORT_UTF8
-   if (utf8 && item > 127) { GETCHAR(item, utf8_char); }
+   GETCHARTEST(c, previous);
+ #else
+   c = *previous;
  #endif
    switch(-next)
      {
      case ESC_d:
-     return item > 127 || (cd->ctypes[item] & ctype_digit) == 0;
+     return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;
      case ESC_D:
-     return item <= 127 && (cd->ctypes[item] & ctype_digit) != 0;
+     return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;
      case ESC_s:
-     return item > 127 || (cd->ctypes[item] & ctype_space) == 0;
+     return c > 127 || (cd->ctypes[c] & ctype_space) == 0;
      case ESC_S:
-     return item <= 127 && (cd->ctypes[item] & ctype_space) != 0;
+     return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;
      case ESC_w:
-     return item > 127 || (cd->ctypes[item] & ctype_word) == 0;
+     return c > 127 || (cd->ctypes[c] & ctype_word) == 0;
      case ESC_W:
-     return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
+     return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;
      case ESC_h:
      case ESC_H:
-     switch(item)
+     switch(c)
        {
        case 0x09:
        case 0x20:
-Line 2678 
 switch(op_code)
+Line 2814 
 switch(op_code)
      case ESC_v:
      case ESC_V:
-     switch(item)
+     switch(c)
        {
        case 0x0a:
        case 0x0b:
-Line 2692 
 switch(op_code)
+Line 2828 
 switch(op_code)
        return -next == ESC_v;
        }
+     /* When PCRE_UCP is set, these values get generated for \d etc. Find
+     their substitutions and process them. The result will always be either
+     -ESC_p or -ESC_P. Then fall through to process those values. */
+ #ifdef SUPPORT_UCP
+     case ESC_du:
+     case ESC_DU:
+     case ESC_wu:
+     case ESC_WU:
+     case ESC_su:
+     case ESC_SU:
+       {
+       int temperrorcode = 0;
+       ptr = substitutes[-next - ESC_DU];
+       next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);
+       if (temperrorcode != 0) return FALSE;
+       ptr++;    /* For compatibility */
+       }
+     /* Fall through */
+     case ESC_p:
+     case ESC_P:
+       {
+       int ptype, pdata, errorcodeptr;
+       BOOL negated;
+       ptr--;      /* Make ptr point at the p or P */
+       ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);
+       if (ptype < 0) return FALSE;
+       ptr++;      /* Point past the final curly ket */
+       /* If the property item is optional, we have to give up. (When generated
+       from \d etc by PCRE_UCP, this test will have been applied much earlier,
+       to the original \d etc. At this point, ptr will point to a zero byte. */
+       if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
+         strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
+           return FALSE;
+       /* Do the property check. */
+       return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);
+       }
+ #endif
      default:
      return FALSE;
      }
+   /* In principle, support for Unicode properties should be integrated here as
+   well. It means re-organizing the above code so as to get hold of the property
+   values before switching on the op-code. However, I wonder how many patterns
+   combine ASCII \d etc with Unicode properties? (Note that if PCRE_UCP is set,
+   these op-codes are never generated.) */
    case OP_DIGIT:
    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
           next == -ESC_h || next == -ESC_v || next == -ESC_R;
-Line 2710 
 switch(op_code)
+Line 2897 
 switch(op_code)
    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
    case OP_HSPACE:
    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||
           next == -ESC_w || next == -ESC_v || next == -ESC_R;
    case OP_NOT_HSPACE:
    return next == -ESC_h;
    /* Can't have \S in here because VT matches \S (Perl anomaly) */
    case OP_ANYNL:
    case OP_VSPACE:
    return next == -ESC_V || next == -ESC_d || next == -ESC_w;
-Line 2725 
 switch(op_code)
+Line 2912 
 switch(op_code)
    return next == -ESC_v || next == -ESC_R;
    case OP_WORDCHAR:
    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||
           next == -ESC_v || next == -ESC_R;
    case OP_NOT_WORDCHAR:
-Line 2861 
 for (;; ptr++)
+Line 3048 
 for (;; ptr++)
    c = *ptr;
    /* If we are at the end of a nested substitution, revert to the outer level
    string. Nesting only happens one level deep. */
    if (c == 0 && nestptr != NULL)
-Line 2901 
 for (;; ptr++)
+Line 3088 
 for (;; ptr++)
        goto FAILED;
        }
-     *lengthptr += code - last_code;
+     *lengthptr += (int)(code - last_code);
      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));
      /* If "previous" is set and it is not at the start of the work space, move
-Line 2983 
 for (;; ptr++)
+Line 3170 
 for (;; ptr++)
      if ((cd->ctypes[c] & ctype_space) != 0) continue;
      if (c == CHAR_NUMBER_SIGN)
        {
-       while (*(++ptr) != 0)
+       ptr++;
+       while (*ptr != 0)
          {
          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
+         ptr++;
+ #ifdef SUPPORT_UTF8
+         if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
+ #endif
          }
        if (*ptr != 0) continue;
-Line 3019 
 for (;; ptr++)
+Line 3211 
 for (;; ptr++)
          *errorcodeptr = ERR20;
          goto FAILED;
          }
-       *lengthptr += code - last_code;   /* To include callout length */
+       *lengthptr += (int)(code - last_code);   /* To include callout length */
        DPRINTF((">> end branch\n"));
        }
      return TRUE;
-Line 3168 
 for (;; ptr++)
+Line 3360 
 for (;; ptr++)
          {                           /* Braces are required because the */
          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
          }
        /* In the pre-compile phase, accumulate the length of any UTF-8 extra
        data and reset the pointer. This is so that very large classes that
        contain a zillion UTF-8 characters no longer overwrite the work space
-Line 3224 
 for (;; ptr++)
+Line 3416 
 for (;; ptr++)
            ptr++;
            }
-         posix_class = check_posix_name(ptr, tempptr - ptr);
+         posix_class = check_posix_name(ptr, (int)(tempptr - ptr));
          if (posix_class < 0)
            {
            *errorcodeptr = ERR30;
-Line 3237 
 for (;; ptr++)
+Line 3429 
 for (;; ptr++)
          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
            posix_class = 0;
          /* When PCRE_UCP is set, some of the POSIX classes are converted to
          different escape sequences that use Unicode properties. */
  #ifdef SUPPORT_UCP
          if ((options & PCRE_UCP) != 0)
            {
            int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
            if (posix_substitutes[pc] != NULL)
              {
              nestptr = tempptr + 1;
              ptr = posix_substitutes[pc] - 1;
              continue;
              }
            }
  #endif
          /* In the non-UCP case, we build the bit map for the POSIX class in a
          chunk of local store because we may be adding and subtracting from it,
          and we don't want to subtract bits that may be in the main map already.
-Line 3339 
 for (;; ptr++)
+Line 3531 
 for (;; ptr++)
              case ESC_SU:
              nestptr = ptr;
              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
              class_charcount -= 2;                /* Undo! */
              continue;
  #endif
              case ESC_d:
-Line 3360 
 for (;; ptr++)
+Line 3552 
 for (;; ptr++)
              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
              continue;
+             /* Perl 5.004 onwards omits VT from \s, but we must preserve it
+             if it was previously set by something earlier in the character
+             class. */
              case ESC_s:
-             for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
+             classbits[0] |= cbits[cbit_space];
-             classbits[1] &= ~0x08;   /* Perl 5.004 onwards omits VT from \s */
+             classbits[1] |= cbits[cbit_space+1] & ~0x08;
+             for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
              continue;
              case ESC_S:
-Line 3790 
 for (;; ptr++)
+Line 3987 
 for (;; ptr++)
      can cause firstbyte to be set. Otherwise, there can be no first char if
      this item is first, whatever repeat count may follow. In the case of
      reqbyte, save the previous value for reinstating. */
  #ifdef SUPPORT_UTF8
      if (class_charcount == 1 && !class_utf8 &&
        (!utf8 || !negate_class || class_lastchar < 128))
-Line 3870 
 for (;; ptr++)
+Line 4067 
 for (;; ptr++)
        }
  #endif
      /* If there are no characters > 255, or they are all to be included or
      excluded, set the opcode to OP_CLASS or OP_NCLASS, depending on whether the
      whole class was negated and whether there were negative specials such as \S
      (non-UCP) in the class. Then copy the 32-byte map into the code vector,
-Line 3998 
 for (;; ptr++)
+Line 4195 
 for (;; ptr++)
        if (!possessive_quantifier &&
            repeat_max < 0 &&
-           check_auto_possessive(*previous, c, utf8, utf8_char, ptr + 1,
+           check_auto_possessive(previous, utf8, ptr + 1, options, cd))
-             options, cd))
          {
          repeat_type = 0;    /* Force greedy */
          possessive_quantifier = TRUE;
-Line 4020 
 for (;; ptr++)
+Line 4216 
 for (;; ptr++)
        c = previous[1];
        if (!possessive_quantifier &&
            repeat_max < 0 &&
-           check_auto_possessive(OP_NOT, c, utf8, NULL, ptr + 1, options, cd))
+           check_auto_possessive(previous, utf8, ptr + 1, options, cd))
          {
          repeat_type = 0;    /* Force greedy */
          possessive_quantifier = TRUE;
-Line 4044 
 for (;; ptr++)
+Line 4240 
 for (;; ptr++)
        if (!possessive_quantifier &&
            repeat_max < 0 &&
-           check_auto_possessive(c, 0, utf8, NULL, ptr + 1, options, cd))
+           check_auto_possessive(previous, utf8, ptr + 1, options, cd))
          {
          repeat_type = 0;    /* Force greedy */
          possessive_quantifier = TRUE;
-Line 4254 
 for (;; ptr++)
+Line 4450 
 for (;; ptr++)
        {
        register int i;
        int ketoffset = 0;
-       int len = code - previous;
+       int len = (int)(code - previous);
        uschar *bralink = NULL;
        /* Repeating a DEFINE group is pointless */
-Line 4275 
 for (;; ptr++)
+Line 4471 
 for (;; ptr++)
          {
          register uschar *ket = previous;
          do ket += GET(ket, 1); while (*ket != OP_KET);
-         ketoffset = code - ket;
+         ketoffset = (int)(code - ket);
          }
        /* The case of a zero minimum is special because of the need to stick
-Line 4343 
 for (;; ptr++)
+Line 4539 
 for (;; ptr++)
            /* We chain together the bracket offset fields that have to be
            filled in later when the ends of the brackets are reached. */
-           offset = (bralink == NULL)? 0 : previous - bralink;
+           offset = (bralink == NULL)? 0 : (int)(previous - bralink);
            bralink = previous;
            PUTINC(previous, 0, offset);
            }
-Line 4452 
 for (;; ptr++)
+Line 4648 
 for (;; ptr++)
              {
              int offset;
              *code++ = OP_BRA;
-             offset = (bralink == NULL)? 0 : code - bralink;
+             offset = (bralink == NULL)? 0 : (int)(code - bralink);
              bralink = code;
              PUTINC(code, 0, offset);
              }
-Line 4473 
 for (;; ptr++)
+Line 4669 
 for (;; ptr++)
          while (bralink != NULL)
            {
            int oldlinkoffset;
-           int offset = code - bralink + 1;
+           int offset = (int)(code - bralink + 1);
            uschar *bra = code - offset;
            oldlinkoffset = GET(bra, 1);
            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;
-Line 4561 
 for (;; ptr++)
+Line 4757 
 for (;; ptr++)
  #endif
          }
-       len = code - tempcode;
+       len = (int)(code - tempcode);
        if (len > 0) switch (*tempcode)
          {
          case OP_STAR:  *tempcode = OP_POSSTAR; break;
-Line 4630 
 for (;; ptr++)
+Line 4826 
 for (;; ptr++)
        const uschar *arg = NULL;
        previous = NULL;
        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
-       namelen = ptr - name;
+       namelen = (int)(ptr - name);
        if (*ptr == CHAR_COLON)
          {
          arg = ++ptr;
          while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0
            || *ptr == '_') ptr++;
-         arglen = ptr - arg;
+         arglen = (int)(ptr - arg);
          }
        if (*ptr != CHAR_RIGHT_PARENTHESIS)
-Line 4675 
 for (;; ptr++)
+Line 4871 
 for (;; ptr++)
                *errorcodeptr = ERR66;
                goto FAILED;
                }
-             *code++ = verbs[i].op;
+             *code = verbs[i].op;
+             if (*code++ == OP_THEN)
+               {
+               PUT(code, 0, code - bcptr->current_branch - 1);
+               code += LINK_SIZE;
+               }
              }
            else
-Line 4685 
 for (;; ptr++)
+Line 4886 
 for (;; ptr++)
                *errorcodeptr = ERR59;
                goto FAILED;
                }
-             *code++ = verbs[i].op_arg;
+             *code = verbs[i].op_arg;
+             if (*code++ == OP_THEN_ARG)
+               {
+               PUT(code, 0, code - bcptr->current_branch - 1);
+               code += LINK_SIZE;
+               }
              *code++ = arglen;
              memcpy(code, arg, arglen);
              code += arglen;
-Line 4818 
 for (;; ptr++)
+Line 5024 
 for (;; ptr++)
                recno * 10 + *ptr - CHAR_0 : -1;
            ptr++;
            }
-         namelen = ptr - name;
+         namelen = (int)(ptr - name);
          if ((terminator > 0 && *ptr++ != terminator) ||
              *ptr++ != CHAR_RIGHT_PARENTHESIS)
-Line 4879 
 for (;; ptr++)
+Line 5085 
 for (;; ptr++)
          /* Search the pattern for a forward reference */
          else if ((i = find_parens(cd, name, namelen,
-                         (options & PCRE_EXTENDED) != 0)) > 0)
+                         (options & PCRE_EXTENDED) != 0, utf8)) > 0)
            {
            PUT2(code, 2+LINK_SIZE, i);
            code[1+LINK_SIZE]++;
-Line 5014 
 for (;; ptr++)
+Line 5220 
 for (;; ptr++)
              goto FAILED;
              }
            *code++ = n;
-           PUT(code, 0, ptr - cd->start_pattern + 1);  /* Pattern offset */
+           PUT(code, 0, (int)(ptr - cd->start_pattern + 1)); /* Pattern offset */
-           PUT(code, LINK_SIZE, 0);                    /* Default length */
+           PUT(code, LINK_SIZE, 0);                          /* Default length */
            code += 2 * LINK_SIZE;
            }
          previous = NULL;
-Line 5048 
 for (;; ptr++)
+Line 5254 
 for (;; ptr++)
            name = ++ptr;
            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
-           namelen = ptr - name;
+           namelen = (int)(ptr - name);
            /* In the pre-compile phase, just do a syntax check. */
-Line 5178 
 for (;; ptr++)
+Line 5384 
 for (;; ptr++)
          NAMED_REF_OR_RECURSE:
          name = ++ptr;
          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
-         namelen = ptr - name;
+         namelen = (int)(ptr - name);
-         /* 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
-         reference number. */
+         a dummy reference number, because it was not used in the first pass.
+         However, with the change of recursive back references to be atomic,
+         we have to look for the number so that this state can be identified, as
+         otherwise the incorrect length is computed. If it's not a backwards
+         reference, the dummy number will do. */
          if (lengthptr != NULL)
            {
+           const uschar *temp;
            if (namelen == 0)
              {
              *errorcodeptr = ERR62;
-Line 5200 
 for (;; ptr++)
+Line 5412 
 for (;; ptr++)
              *errorcodeptr = ERR48;
              goto FAILED;
              }
-           recno = 0;
+           /* The name table does not exist in the first pass, so we cannot
+           do a simple search as in the code below. Instead, we have to scan the
+           pattern to find the number. It is important that we scan it only as
+           far as we have got because the syntax of named subpatterns has not
+           been checked for the rest of the pattern, and find_parens() assumes
+           correct syntax. In any case, it's a waste of resources to scan
+           further. We stop the scan at the current point by temporarily
+           adjusting the value of cd->endpattern. */
+           temp = cd->end_pattern;
+           cd->end_pattern = ptr;
+           recno = find_parens(cd, name, namelen,
+             (options & PCRE_EXTENDED) != 0, utf8);
+           cd->end_pattern = temp;
+           if (recno < 0) recno = 0;    /* Forward ref; set dummy number */
            }
          /* In the real compile, seek the name in the table. We check the name
-Line 5225 
 for (;; ptr++)
+Line 5452 
 for (;; ptr++)
              }
            else if ((recno =                /* Forward back reference */
                      find_parens(cd, name, namelen,
-                       (options & PCRE_EXTENDED) != 0)) <= 0)
+                       (options & PCRE_EXTENDED) != 0, utf8)) <= 0)
              {
              *errorcodeptr = ERR15;
              goto FAILED;
-Line 5336 
 for (;; ptr++)
+Line 5563 
 for (;; ptr++)
              if (called == NULL)
                {
                if (find_parens(cd, NULL, recno,
-                     (options & PCRE_EXTENDED) != 0) < 0)
+                     (options & PCRE_EXTENDED) != 0, utf8) < 0)
                  {
                  *errorcodeptr = ERR15;
                  goto FAILED;
-Line 5347 
 for (;; ptr++)
+Line 5574 
 for (;; ptr++)
                of the group. */
                called = cd->start_code + recno;
-               PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);
+               PUTINC(cd->hwm, 0, (int)(code + 2 + LINK_SIZE - cd->start_code));
                }
              /* If not a forward reference, and the subpattern is still open,
-Line 5371 
 for (;; ptr++)
+Line 5598 
 for (;; ptr++)
            code += 1 + LINK_SIZE;
            *code = OP_RECURSE;
-           PUT(code, 1, called - cd->start_code);
+           PUT(code, 1, (int)(called - cd->start_code));
            code += 1 + LINK_SIZE;
            *code = OP_KET;
-Line 5675 
 for (;; ptr++)
+Line 5902 
 for (;; ptr++)
      /* ===================================================================*/
      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
      are arranged to be the negation of the corresponding OP_values in the
      default case when PCRE_UCP is not set. For the back references, the values
      are ESC_REF plus the reference number. Only back references and those types
      that consume a character may be repeated. We can test for values between
-Line 5853 
 for (;; ptr++)
+Line 6080 
 for (;; ptr++)
            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
            }
          else
  #endif
            {
            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
            *code++ = -c;
            }
          }
        continue;
        }
-Line 6189 
 for (;;)
+Line 6416 
 for (;;)
      {
      if (lengthptr == NULL)
        {
-       int branch_length = code - last_branch;
+       int branch_length = (int)(code - last_branch);
        do
          {
          int prev_length = GET(last_branch, 1);
-Line 6203 
 for (;;)
+Line 6430 
 for (;;)
      /* Fill in the ket */
      *code = OP_KET;
-     PUT(code, 1, code - start_bracket);
+     PUT(code, 1, (int)(code - start_bracket));
      code += 1 + LINK_SIZE;
      /* If it was a capturing subpattern, check to see if it contained any
-Line 6218 
 for (;;)
+Line 6445 
 for (;;)
            code - start_bracket);
          *start_bracket = OP_ONCE;
          code += 1 + LINK_SIZE;
-         PUT(start_bracket, 1, code - start_bracket);
+         PUT(start_bracket, 1, (int)(code - start_bracket));
          *code = OP_KET;
-         PUT(code, 1, code - start_bracket);
+         PUT(code, 1, (int)(code - start_bracket));
          code += 1 + LINK_SIZE;
          length += 2 + 2*LINK_SIZE;
          }
-Line 6275 
 for (;;)
+Line 6502 
 for (;;)
    else
      {
      *code = OP_ALT;
-     PUT(code, 1, code - last_branch);
+     PUT(code, 1, (int)(code - last_branch));
      bc.current_branch = last_branch = code;
      code += 1 + LINK_SIZE;
      }
-Line 6666 
 while (ptr[skipatstart] == CHAR_LEFT_PAR
+Line 6893 
 while (ptr[skipatstart] == CHAR_LEFT_PAR
      { skipatstart += 7; options |= PCRE_UTF8; continue; }
    else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)
      { skipatstart += 6; options |= PCRE_UCP; continue; }
+   else if (strncmp((char *)(ptr+skipatstart+2), STRING_NO_START_OPT_RIGHTPAR, 13) == 0)
+     { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; }
    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
-Line 6689 
 while (ptr[skipatstart] == CHAR_LEFT_PAR
+Line 6918 
 while (ptr[skipatstart] == CHAR_LEFT_PAR
      options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
    else break;
    }
  utf8 = (options & PCRE_UTF8) != 0;
- /* 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
+ return of an error code from _pcre_valid_utf8() is a new feature, introduced in
+ release 8.13. The only use we make of it here is to adjust the offset value to
+ the end of the string for a short string error, for compatibility with previous
+ versions. */
  #ifdef SUPPORT_UTF8
  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 &&
-      (*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1)) >= 0)
+      (*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1, &errorcode)) >= 0)
    {
    errorcode = ERR44;
    goto PCRE_EARLY_ERROR_RETURN2;
-Line 6715 
 if (utf8)
+Line 6948 
 if (utf8)
  if ((options & PCRE_UCP) != 0)
    {
    errorcode = ERR67;
    goto PCRE_EARLY_ERROR_RETURN;
    }
  #endif
  /* Check validity of \R options. */
-Line 6847 
 regex compiled on a system with 4-byte p
+Line 7080 
 regex compiled on a system with 4-byte p
  pointers. */
  re->magic_number = MAGIC_NUMBER;
- re->size = size;
+ re->size = (int)size;
  re->options = cd->external_options;
  re->flags = cd->external_flags;
  re->dummy1 = 0;
-Line 6918 
 while (errorcode == 0 && cd->hwm > cwork
+Line 7151 
 while (errorcode == 0 && cd->hwm > cwork
    recno = GET(codestart, offset);
    groupptr = _pcre_find_bracket(codestart, utf8, recno);
    if (groupptr == NULL) errorcode = ERR53;
-     else PUT(((uschar *)codestart), offset, groupptr - codestart);
+     else PUT(((uschar *)codestart), offset, (int)(groupptr - codestart));
    }
  /* Give an error if there's back reference to a non-existent capturing
-Line 6973 
 if (errorcode != 0)
+Line 7206 
 if (errorcode != 0)
    {
    (pcre_free)(re);
    PCRE_EARLY_ERROR_RETURN:
-   *erroroffset = ptr - (const uschar *)pattern;
+   *erroroffset = (int)(ptr - (const uschar *)pattern);
    PCRE_EARLY_ERROR_RETURN2:
    *errorptr = find_error_text(errorcode);
    if (errorcodeptr != NULL) *errorcodeptr = errorcode;

 Legend:



Removed from v.528
 


changed lines


 
Added in v.598
 Legend:



Removed from v.528
 


changed lines


 
Added in v.598
-Removed from v.528
+Added in v.598

webmaster@exim.org	ViewVC Help
Powered by ViewVC 1.1.12