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

Diff of /code/trunk/pcre_compile.c

Parent Directory | Revision Log | View Patch Patch

-revision 514 by ph10,
Mon May  3 12:54:22 2010 UTC
+revision 533 by ph10,
Wed Jun  2 19:02:41 2010 UTC
 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
+ static const uschar *substitutes[] = {
+   (uschar *)"\\P{Nd}",    /* \D */
+   (uschar *)"\\p{Nd}",    /* \d */
+   (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 */
+   (uschar *)"\\p{Nd}",    /* digit */
+   NULL,                   /* graph */
+   NULL,                   /* print */
+   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 */
+   (uschar *)"\\P{Nd}",    /* ^digit */
+   NULL,                   /* ^graph */
+   NULL,                   /* ^print */
+   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 360 
 static const char error_texts[] =
+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"
    ;
  /* Table to identify digits and hex digits. This is used when compiling
-Line 829 
 else
+Line 877 
 else
      break;
      }
    }
  /* Perl supports \N{name} for character names, as well as plain \N for "not
  newline". PCRE does not support \N{name}. */
  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET)
    *errorcodeptr = ERR37;
+ /* If PCRE_UCP is set, we change the values for \d etc. */
+ if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)
+   c -= (ESC_DU - ESC_D);
+ /* Set the pointer to the final character before returning. */
  *ptrptr = ptr;
  return c;
-Line 2261 
 auto_callout(uschar *code, const uschar
+Line 2316 
 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 2287 
 Returns:             nothing
+Line 2342 
 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 2337 
 for (++c; c <= d; c++)
+Line 2392 
 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 2350 
 whether the next thing could possibly ma
+Line 2468 
 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 2362 
 Returns:        TRUE if possessifying is
+Line 2478 
 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 2426 
 if (*ptr == CHAR_ASTERISK || *ptr == CHA
+Line 2543 
 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
    high-valued characters. */
    case OP_CHARNC:
  #ifdef SUPPORT_UTF8
-   if (utf8 && item > 127) { GETCHAR(item, utf8_char); }
+   GETCHARTEST(c, previous);
- #endif
+ #else
-   if (item == next) return FALSE;
+   c = *previous;
+ #endif
+   if (c == next) return FALSE;
  #ifdef SUPPORT_UTF8
    if (utf8)
      {
-Line 2463 
 if (next >= 0) switch(op_code)
+Line 2577 
 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 2484 
 if (next >= 0) switch(op_code)
+Line 2598 
 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 2531 
 if (next >= 0) switch(op_code)
+Line 2648 
 if (next >= 0) switch(op_code)
      case 0x202f:
      case 0x205f:
      case 0x3000:
-     return op_code != OP_HSPACE;
+     return op_code == OP_NOT_HSPACE;
      default:
-     return op_code == OP_HSPACE;
+     return op_code != OP_NOT_HSPACE;
      }
+   case OP_ANYNL:
    case OP_VSPACE:
    case OP_NOT_VSPACE:
    switch(next)
-Line 2547 
 if (next >= 0) switch(op_code)
+Line 2665 
 if (next >= 0) switch(op_code)
      case 0x85:
      case 0x2028:
      case 0x2029:
-     return op_code != OP_VSPACE;
+     return op_code == OP_NOT_VSPACE;
      default:
-     return op_code == OP_VSPACE;
+     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. */
+ /* 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. 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);
- #endif
+ #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 2616 
 switch(op_code)
+Line 2748 
 switch(op_code)
      case ESC_v:
      case ESC_V:
-     switch(item)
+     switch(c)
        {
        case 0x0a:
        case 0x0b:
-Line 2629 
 switch(op_code)
+Line 2761 
 switch(op_code)
        default:
        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_h || next == -ESC_v || next == -ESC_R;
    case OP_NOT_DIGIT:
    return next == -ESC_d;
    case OP_WHITESPACE:
-   return next == -ESC_S || next == -ESC_d || next == -ESC_w;
+   return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;
    case OP_NOT_WHITESPACE:
    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;
+   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;
    case OP_NOT_VSPACE:
-   return next == -ESC_v;
+   return next == -ESC_v || next == -ESC_R;
    case OP_WORDCHAR:
-   return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
+   return next == -ESC_W || next == -ESC_s || next == -ESC_h ||
+          next == -ESC_v || next == -ESC_R;
    case OP_NOT_WORDCHAR:
    return next == -ESC_w || next == -ESC_d;
-Line 2725 
 BOOL inescq = FALSE;
+Line 2911 
 BOOL inescq = FALSE;
  BOOL groupsetfirstbyte = FALSE;
  const uschar *ptr = *ptrptr;
  const uschar *tempptr;
+ const uschar *nestptr = NULL;
  uschar *previous = NULL;
  uschar *previous_callout = NULL;
  uschar *save_hwm = NULL;
-Line 2795 
 for (;; ptr++)
+Line 2982 
 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)
+     {
+     ptr = nestptr;
+     nestptr = NULL;
+     c = *ptr;
+     }
    /* If we are in the pre-compile phase, accumulate the length used for the
    previous cycle of this loop. */
-Line 2825 
 for (;; ptr++)
+Line 3022 
 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 2943 
 for (;; ptr++)
+Line 3140 
 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 3092 
 for (;; ptr++)
+Line 3289 
 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 3148 
 for (;; ptr++)
+Line 3345 
 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 3161 
 for (;; ptr++)
+Line 3358 
 for (;; ptr++)
          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
            posix_class = 0;
-         /* We build the bit map for the POSIX class in a chunk of local store
+         /* When PCRE_UCP is set, some of the POSIX classes are converted to
-         because we may be adding and subtracting from it, and we don't want to
+         different escape sequences that use Unicode properties. */
-         subtract bits that may be in the main map already. At the end we or the
-         result into the bit map that is being built. */
+ #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.
+         At the end we or the result into the bit map that is being built. */
          posix_class *= 3;
-Line 3237 
 for (;; ptr++)
+Line 3449 
 for (;; ptr++)
            register const uschar *cbits = cd->cbits;
            class_charcount += 2;     /* Greater than 1 is what matters */
-           /* Save time by not doing this in the pre-compile phase. */
+           switch (-c)
-           if (lengthptr == NULL) switch (-c)
              {
+ #ifdef SUPPORT_UCP
+             case ESC_du:     /* These are the values given for \d etc */
+             case ESC_DU:     /* when PCRE_UCP is set. We replace the */
+             case ESC_wu:     /* escape sequence with an appropriate \p */
+             case ESC_WU:     /* or \P to test Unicode properties instead */
+             case ESC_su:     /* of the default ASCII testing. */
+             case ESC_SU:
+             nestptr = ptr;
+             ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
+             class_charcount -= 2;                /* Undo! */
+             continue;
+ #endif
              case ESC_d:
              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];
              continue;
-Line 3270 
 for (;; ptr++)
+Line 3492 
 for (;; ptr++)
              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
              continue;
-             default:    /* Not recognized; fall through */
+             case ESC_h:
-             break;      /* Need "default" setting to stop compiler warning. */
-             }
-           /* In the pre-compile phase, just do the recognition. */
-           else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
-                    c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
-           /* We need to deal with \H, \h, \V, and \v in both phases because
-           they use extra memory. */
-           if (-c == ESC_h)
-             {
              SETBIT(classbits, 0x09); /* VT */
              SETBIT(classbits, 0x20); /* SPACE */
              SETBIT(classbits, 0xa0); /* NSBP */
-Line 3307 
 for (;; ptr++)
+Line 3516 
 for (;; ptr++)
                }
  #endif
              continue;
-             }
-           if (-c == ESC_H)
+             case ESC_H:
-             {
              for (c = 0; c < 32; c++)
                {
                int x = 0xff;
-Line 3352 
 for (;; ptr++)
+Line 3559 
 for (;; ptr++)
                }
  #endif
              continue;
-             }
-           if (-c == ESC_v)
+             case ESC_v:
-             {
              SETBIT(classbits, 0x0a); /* LF */
              SETBIT(classbits, 0x0b); /* VT */
              SETBIT(classbits, 0x0c); /* FF */
-Line 3371 
 for (;; ptr++)
+Line 3576 
 for (;; ptr++)
                }
  #endif
              continue;
-             }
-           if (-c == ESC_V)
+             case ESC_V:
-             {
              for (c = 0; c < 32; c++)
                {
                int x = 0xff;
-Line 3404 
 for (;; ptr++)
+Line 3607 
 for (;; ptr++)
                }
  #endif
              continue;
-             }
-           /* We need to deal with \P and \p in both phases. */
  #ifdef SUPPORT_UCP
-           if (-c == ESC_p || -c == ESC_P)
+             case ESC_p:
-             {
+             case ESC_P:
-             BOOL negated;
+               {
-             int pdata;
+               BOOL negated;
-             int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
+               int pdata;
-             if (ptype < 0) goto FAILED;
+               int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
-             class_utf8 = TRUE;
+               if (ptype < 0) goto FAILED;
-             *class_utf8data++ = ((-c == ESC_p) != negated)?
+               class_utf8 = TRUE;
-               XCL_PROP : XCL_NOTPROP;
+               *class_utf8data++ = ((-c == ESC_p) != negated)?
-             *class_utf8data++ = ptype;
+                 XCL_PROP : XCL_NOTPROP;
-             *class_utf8data++ = pdata;
+               *class_utf8data++ = ptype;
-             class_charcount -= 2;   /* Not a < 256 character */
+               *class_utf8data++ = pdata;
-             continue;
+               class_charcount -= 2;   /* Not a < 256 character */
-             }
+               continue;
+               }
  #endif
-           /* Unrecognized escapes are faulted if PCRE is running in its
+             /* Unrecognized escapes are faulted if PCRE is running in its
-           strict mode. By default, for compatibility with Perl, they are
+             strict mode. By default, for compatibility with Perl, they are
-           treated as literals. */
+             treated as literals. */
-           if ((options & PCRE_EXTRA) != 0)
+             default:
-             {
+             if ((options & PCRE_EXTRA) != 0)
-             *errorcodeptr = ERR7;
+               {
-             goto FAILED;
+               *errorcodeptr = ERR7;
+               goto FAILED;
+               }
+             class_charcount -= 2;  /* Undo the default count from above */
+             c = *ptr;              /* Get the final character and fall through */
+             break;
              }
-           class_charcount -= 2;  /* Undo the default count from above */
-           c = *ptr;              /* Get the final character and fall through */
            }
          /* Fall through if we have a single character (c >= 0). This may be
-Line 3675 
 for (;; ptr++)
+Line 3878 
 for (;; ptr++)
          }
        }
-     /* Loop until ']' reached. This "while" is the end of the "do" above. */
+     /* Loop until ']' reached. This "while" is the end of the "do" far above.
+     If we are at the end of an internal nested string, revert to the outer
+     string. */
+     while (((c = *(++ptr)) != 0 ||
+            (nestptr != NULL &&
+              (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&
+            (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
-     while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
+     /* Check for missing terminating ']' */
-     if (c == 0)                          /* Missing terminating ']' */
+     if (c == 0)
        {
        *errorcodeptr = ERR6;
        goto FAILED;
        }
- /* This code has been disabled because it would mean that \s counts as
- an explicit \r or \n reference, and that's not really what is wanted. Now
- we set the flag only if there is a literal "\r" or "\n" in the class. */
- #if 0
-     /* Remember whether \r or \n are in this class */
-     if (negate_class)
-       {
-       if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
-       }
-     else
-       {
-       if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
-       }
- #endif
      /* If class_charcount is 1, we saw precisely one character whose value is
      less than 256. As long as there were no characters >= 128 and there was no
      use of \p or \P, in other words, no use of any XCLASS features, we can
-Line 3720 
 we set the flag only if there is a liter
+Line 3911 
 we set the flag only if there is a liter
      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 3767 
 we set the flag only if there is a liter
+Line 3958 
 we set the flag only if there is a liter
      /* If there are characters with values > 255, we have to compile an
      extended class, with its own opcode, unless there was a negated special
-     such as \S in the class, because in that case all characters > 255 are in
+     such as \S in the class, and PCRE_UCP is not set, because in that case all
-     the class, so any that were explicitly given as well can be ignored. If
+     characters > 255 are in the class, so any that were explicitly given as
-     (when there are explicit characters > 255 that must be listed) there are no
+     well can be ignored. If (when there are explicit characters > 255 that must
-     characters < 256, we can omit the bitmap in the actual compiled code. */
+     be listed) there are no characters < 256, we can omit the bitmap in the
+     actual compiled code. */
  #ifdef SUPPORT_UTF8
-     if (class_utf8 && !should_flip_negation)
+     if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))
        {
        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
        *code++ = OP_XCLASS;
-Line 3799 
 we set the flag only if there is a liter
+Line 3991 
 we set the flag only if there is a liter
        }
  #endif
-     /* If there are no characters > 255, set the opcode to OP_CLASS or
+     /* If there are no characters > 255, or they are all to be included or
-     OP_NCLASS, depending on whether the whole class was negated and whether
+     excluded, set the opcode to OP_CLASS or OP_NCLASS, depending on whether the
-     there were negative specials such as \S in the class. Then copy the 32-byte
+     whole class was negated and whether there were negative specials such as \S
-     map into the code vector, negating it if necessary. */
+     (non-UCP) in the class. Then copy the 32-byte map into the code vector,
+     negating it if necessary. */
      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
      if (negate_class)
-Line 3926 
 we set the flag only if there is a liter
+Line 4119 
 we set the flag only if there is a liter
        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 3948 
 we set the flag only if there is a liter
+Line 4140 
 we set the flag only if there is a liter
        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 3972 
 we set the flag only if there is a liter
+Line 4164 
 we set the flag only if there is a liter
        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 4182 
 we set the flag only if there is a liter
+Line 4374 
 we set the flag only if there is a liter
        {
        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 4203 
 we set the flag only if there is a liter
+Line 4395 
 we set the flag only if there is a liter
          {
          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 4271 
 we set the flag only if there is a liter
+Line 4463 
 we set the flag only if there is a liter
            /* 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 4380 
 we set the flag only if there is a liter
+Line 4572 
 we set the flag only if there is a liter
              {
              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 4401 
 we set the flag only if there is a liter
+Line 4593 
 we set the flag only if there is a liter
          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 4489 
 we set the flag only if there is a liter
+Line 4681 
 we set the flag only if there is a liter
  #endif
          }
-       len = code - tempcode;
+       len = (int)(code - tempcode);
        if (len > 0) switch (*tempcode)
          {
          case OP_STAR:  *tempcode = OP_POSSTAR; break;
-Line 4558 
 we set the flag only if there is a liter
+Line 4750 
 we set the flag only if there is a liter
        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 4746 
 we set the flag only if there is a liter
+Line 4938 
 we set the flag only if there is a liter
                recno * 10 + *ptr - CHAR_0 : -1;
            ptr++;
            }
-         namelen = ptr - name;
+         namelen = (int)(ptr - name);
          if ((terminator > 0 && *ptr++ != terminator) ||
              *ptr++ != CHAR_RIGHT_PARENTHESIS)
-Line 4942 
 we set the flag only if there is a liter
+Line 5134 
 we set the flag only if there is a liter
              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 4976 
 we set the flag only if there is a liter
+Line 5168 
 we set the flag only if there is a liter
            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 5106 
 we set the flag only if there is a liter
+Line 5298 
 we set the flag only if there is a liter
          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
          reference number. */
-Line 5275 
 we set the flag only if there is a liter
+Line 5467 
 we set the flag only if there is a liter
                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 5299 
 we set the flag only if there is a liter
+Line 5491 
 we set the flag only if there is a liter
            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 5603 
 we set the flag only if there is a liter
+Line 5795 
 we set the flag only if there is a liter
      /* ===================================================================*/
      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
-     are arranged to be the negation of the corresponding OP_values. For the
+     are arranged to be the negation of the corresponding OP_values in the
-     back references, the values are ESC_REF plus the reference number. Only
+     default case when PCRE_UCP is not set. For the back references, the values
-     back references and those types that consume a character may be repeated.
+     are ESC_REF plus the reference number. Only back references and those types
-     We can test for values between ESC_b and ESC_Z for the latter; this may
+     that consume a character may be repeated. We can test for values between
-     have to change if any new ones are ever created. */
+     ESC_b and ESC_Z for the latter; this may have to change if any new ones are
+     ever created. */
      case CHAR_BACKSLASH:
      tempptr = ptr;
-Line 5767 
 we set the flag only if there is a liter
+Line 5960 
 we set the flag only if there is a liter
  #endif
        /* For the rest (including \X when Unicode properties are supported), we
-       can obtain the OP value by negating the escape value. */
+       can obtain the OP value by negating the escape value in the default
+       situation when PCRE_UCP is not set. When it *is* set, we substitute
+       Unicode property tests. */
        else
          {
-         previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
+ #ifdef SUPPORT_UCP
-         *code++ = -c;
+         if (-c >= ESC_DU && -c <= ESC_wu)
+           {
+           nestptr = ptr + 1;                   /* Where to resume */
+           ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
+           }
+         else
+ #endif
+           {
+           previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
+           *code++ = -c;
+           }
          }
        continue;
        }
-Line 6104 
 for (;;)
+Line 6309 
 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 6118 
 for (;;)
+Line 6323 
 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 6133 
 for (;;)
+Line 6338 
 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 6190 
 for (;;)
+Line 6395 
 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 6509 
 int length = 1;  /* For final END opcode
+Line 6714 
 int length = 1;  /* For final END opcode
  int firstbyte, reqbyte, newline;
  int errorcode = 0;
  int skipatstart = 0;
- BOOL utf8 = (options & PCRE_UTF8) != 0;
+ BOOL utf8;
  size_t size;
  uschar *code;
  const uschar *codestart;
-Line 6579 
 while (ptr[skipatstart] == CHAR_LEFT_PAR
+Line 6784 
 while (ptr[skipatstart] == CHAR_LEFT_PAR
    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)
      { skipatstart += 7; options |= PCRE_UTF8; continue; }
+   else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)
+     { skipatstart += 6; options |= PCRE_UCP; continue; }
    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
-Line 6602 
 while (ptr[skipatstart] == CHAR_LEFT_PAR
+Line 6809 
 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. */
-Line 6620 
 if (utf8)
+Line 6829 
 if (utf8)
    }
  #endif
+ /* Can't support UCP unless PCRE has been compiled to include the code. */
+ #ifndef SUPPORT_UCP
+ if ((options & PCRE_UCP) != 0)
+   {
+   errorcode = ERR67;
+   goto PCRE_EARLY_ERROR_RETURN;
+   }
+ #endif
  /* Check validity of \R options. */
  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
-Line 6748 
 regex compiled on a system with 4-byte p
+Line 6967 
 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 6819 
 while (errorcode == 0 && cd->hwm > cwork
+Line 7038 
 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 6874 
 if (errorcode != 0)
+Line 7093 
 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.514
 


changed lines


 
Added in v.533
 Legend:



Removed from v.514
 


changed lines


 
Added in v.533
-Removed from v.514
+Added in v.533

webmaster@exim.org	ViewVC Help
Powered by ViewVC 1.1.12