/[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 962 by ph10, Fri Apr 20 17:28:23 2012 UTC revision 1055 by chpe, Tue Oct 16 15:53:30 2012 UTC
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  is also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. We do not need to select pcre16_printint.c specially, because the  library. We do not need to select pcre16_printint.c specially, because the
59  COMPILE_PCREx macro will already be appropriately set. */  COMPILE_PCREx macro will already be appropriately set. */
# Line 68  COMPILE_PCREx macro will already be appr Line 68  COMPILE_PCREx macro will already be appr
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 77  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL,
88        int, int, int *, int *, branch_chain *, compile_data *, int *);
89    
90    
91    
92  /*************************************************  /*************************************************
93  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 112  overrun before it actually does run off Line 123  overrun before it actually does run off
123    
124  #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */  #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */
125  #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */  #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */
126    #define REQ_MASK       (REQ_CASELESS | REQ_VARY)
127    
128  /* Repeated character flags. */  /* Repeated character flags. */
129    
# Line 489  static const char error_texts[] = Line 501  static const char error_texts[] =
501    "too many forward references\0"    "too many forward references\0"
502    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
503    "invalid UTF-16 string\0"    "invalid UTF-16 string\0"
504      /* 75 */
505      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
506      "character value in \\u.... sequence is too large\0"
507      "invalid UTF-32 string\0"
508    ;    ;
509    
510  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 628  static const pcre_uint8 ebcdic_chartab[] Line 644  static const pcre_uint8 ebcdic_chartab[]
644  #endif  #endif
645    
646    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
647    
648    
649  /*************************************************  /*************************************************
# Line 786  else if ((i = escapes[c - CHAR_0]) != 0) Line 796  else if ((i = escapes[c - CHAR_0]) != 0)
796    
797  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
798  /* Not alphanumeric */  /* Not alphanumeric */
799  else if (c < 'a' || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
800  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
801  #endif  #endif
802    
# Line 829  else Line 839  else
839            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
840  #endif  #endif
841            }            }
842    
843    #if defined COMPILE_PCRE8
844            if (c > (utf ? 0x10ffff : 0xff))
845    #elif defined COMPILE_PCRE16
846            if (c > (utf ? 0x10ffff : 0xffff))
847    #elif defined COMPILE_PCRE32
848            if (utf && c > 0x10ffff)
849    #endif
850              {
851              *errorcodeptr = ERR76;
852              }
853            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
854          }          }
855        }        }
856      else      else
# Line 1049  else Line 1071  else
1071          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1072  #endif  #endif
1073    
1074  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1075          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }
1076  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
1077          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }
1078  #endif  #elif defined COMPILE_PCRE32
1079            if (utf && c > 0x10ffff) { c = -1; break; }
1080  #endif  #endif
1081          }          }
1082    
# Line 1347  Arguments: Line 1369  Arguments:
1369    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1370    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1371    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1372    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1373    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1374    
1375  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
# Line 1581  Arguments: Line 1603  Arguments:
1603    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1604    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1605    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1606    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1607    
1608  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1609  */  */
# Line 1684  and doing the check at the end; a flag s Line 1706  and doing the check at the end; a flag s
1706    
1707  Arguments:  Arguments:
1708    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1709    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1710    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1711    cd       the "compile data" structure    cd       the "compile data" structure
1712    
# Line 1818  for (;;) Line 1840  for (;;)
1840      case OP_NOTI:      case OP_NOTI:
1841      branchlength++;      branchlength++;
1842      cc += 2;      cc += 2;
1843  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1844      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1845  #endif  #endif
1846      break;      break;
# Line 1832  for (;;) Line 1854  for (;;)
1854      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1855      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1856      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1857  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1858      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1859  #endif  #endif
1860      break;      break;
1861    
1862      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1863      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1864      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1865          cc += 2;
1866      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1867      break;      break;
1868    
# Line 1874  for (;;) Line 1897  for (;;)
1897    
1898      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1899    
1900  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1901      case OP_XCLASS:      case OP_XCLASS:
1902      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1903      /* Fall through */      /* Fall through */
# Line 2013  length. Line 2036  length.
2036    
2037  Arguments:  Arguments:
2038    code        points to start of expression    code        points to start of expression
2039    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2040    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2041    
2042  Returns:      pointer to the opcode for the bracket, or NULL if not found  Returns:      pointer to the opcode for the bracket, or NULL if not found
# Line 2077  for (;;) Line 2100  for (;;)
2100        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2101        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2102        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2103        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2104          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2105        break;        break;
2106    
2107        case OP_MARK:        case OP_MARK:
# Line 2100  for (;;) Line 2123  for (;;)
2123    a multi-byte character. The length in the table is a minimum, so we have to    a multi-byte character. The length in the table is a minimum, so we have to
2124    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2125    
2126  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2127      if (utf) switch(c)      if (utf) switch(c)
2128        {        {
2129        case OP_CHAR:        case OP_CHAR:
# Line 2152  instance of OP_RECURSE. Line 2175  instance of OP_RECURSE.
2175    
2176  Arguments:  Arguments:
2177    code        points to start of expression    code        points to start of expression
2178    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2179    
2180  Returns:      pointer to the opcode for OP_RECURSE, or NULL if not found  Returns:      pointer to the opcode for OP_RECURSE, or NULL if not found
2181  */  */
# Line 2197  for (;;) Line 2220  for (;;)
2220        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2221        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2222        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2223        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2224          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2225        break;        break;
2226    
2227        case OP_MARK:        case OP_MARK:
# Line 2220  for (;;) Line 2243  for (;;)
2243      by a multi-byte character. The length in the table is a minimum, so we have      by a multi-byte character. The length in the table is a minimum, so we have
2244      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2245    
2246  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2247      if (utf) switch(c)      if (utf) switch(c)
2248        {        {
2249        case OP_CHAR:        case OP_CHAR:
# Line 2306  bracket whose current branch will alread Line 2329  bracket whose current branch will alread
2329  Arguments:  Arguments:
2330    code        points to start of search    code        points to start of search
2331    endcode     points to where to stop    endcode     points to where to stop
2332    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2333    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2334    
2335  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2523  for (code = first_significant_code(code Line 2546  for (code = first_significant_code(code
2546      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2547      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2548      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2549      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2550        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2551      break;      break;
2552    
2553      /* End of branch */      /* End of branch */
# Line 2539  for (code = first_significant_code(code Line 2562  for (code = first_significant_code(code
2562      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2563      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2564    
2565  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2566      case OP_STAR:      case OP_STAR:
2567      case OP_STARI:      case OP_STARI:
2568      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2605  Arguments: Line 2628  Arguments:
2628    code        points to start of the recursion    code        points to start of the recursion
2629    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2630    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2631    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2632    cd          pointers to tables etc    cd          pointers to tables etc
2633    
2634  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2752  value in the reference (which is a group Line 2775  value in the reference (which is a group
2775  Arguments:  Arguments:
2776    group      points to the start of the group    group      points to the start of the group
2777    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2778    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2779    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2780    save_hwm   the hwm forward reference pointer at the start of the group    save_hwm   the hwm forward reference pointer at the start of the group
2781    
# Line 2856  PUT(previous_callout, 2 + LINK_SIZE, len Line 2879  PUT(previous_callout, 2 + LINK_SIZE, len
2879  *************************************************/  *************************************************/
2880    
2881  /* This function is passed the start and end of a class range, in UTF-8 mode  /* This function is passed the start and end of a class range, in UTF-8 mode
2882  with UCP support. It searches up the characters, looking for internal ranges of  with UCP support. It searches up the characters, looking for ranges of
2883  characters in the "other" case. Each call returns the next one, updating the  characters in the "other" case. Each call returns the next one, updating the
2884  start address.  start address. A character with multiple other cases is returned on its own
2885    with a special return value.
2886    
2887  Arguments:  Arguments:
2888    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2866  Arguments: Line 2890  Arguments:
2890    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2891    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2892    
2893  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2894                   0 when a range is returned
2895                  >0 the CASESET offset for char with multiple other cases
2896                    in this case, ocptr contains the original
2897  */  */
2898    
2899  static BOOL  static int
2900  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,
2901    unsigned int *odptr)    unsigned int *odptr)
2902  {  {
2903  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2904    int co;
2905    
2906    /* Find the first character that has an other case. If it has multiple other
2907    cases, return its case offset value. */
2908    
2909  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2910    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2911      if ((co = UCD_CASESET(c)) != 0)
2912        {
2913        *ocptr = c++;   /* Character that has the set */
2914        *cptr = c;      /* Rest of input range */
2915        return co;
2916        }
2917      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2918      }
2919    
2920  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2921    
2922  *ocptr = othercase;  *ocptr = othercase;
2923  next = othercase + 1;  next = othercase + 1;
# Line 2889  for (++c; c <= d; c++) Line 2928  for (++c; c <= d; c++)
2928    next++;    next++;
2929    }    }
2930    
2931  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2932  *cptr = c;  *cptr = c;             /* Rest of input range */
2933    return 0;
 return TRUE;  
2934  }  }
2935    
2936    
# Line 2916  Returns:       TRUE if auto-possessifyin Line 2954  Returns:       TRUE if auto-possessifyin
2954  static BOOL  static BOOL
2955  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(int c, int ptype, int pdata, BOOL negated)
2956  {  {
2957    #ifdef SUPPORT_UCP
2958    const pcre_uint32 *p;
2959    #endif
2960    
2961  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2962    
2963  switch(ptype)  switch(ptype)
2964    {    {
2965    case PT_LAMP:    case PT_LAMP:
# Line 2954  switch(ptype) Line 2997  switch(ptype)
2997    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2998            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2999            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3000    
3001    #ifdef SUPPORT_UCP
3002      case PT_CLIST:
3003      p = PRIV(ucd_caseless_sets) + prop->caseset;
3004      for (;;)
3005        {
3006        if ((unsigned int)c < *p) return !negated;
3007        if ((unsigned int)c == *p++) return negated;
3008        }
3009      break;  /* Control never reaches here */
3010    #endif
3011    }    }
3012    
3013  return FALSE;  return FALSE;
3014  }  }
3015  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2971  sense to automatically possessify the re Line 3026  sense to automatically possessify the re
3026    
3027  Arguments:  Arguments:
3028    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3029    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3030    ptr           next character in pattern    ptr           next character in pattern
3031    options       options bits    options       options bits
3032    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2983  static BOOL Line 3038  static BOOL
3038  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3039    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3040  {  {
3041  pcre_int32 c, next;  pcre_int32 c = NOTACHAR;
3042    pcre_int32 next;
3043  int op_code = *previous++;  int op_code = *previous++;
3044    
3045  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 3057  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3113  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3113    STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)    STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
3114      return FALSE;      return FALSE;
3115    
3116  /* Now compare the next item with the previous opcode. First, handle cases when  /* If the previous item is a character, get its value. */
 the next item is a character. */  
3117    
3118  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3119        op_code == OP_NOT || op_code == OP_NOTI)
3120    {    {
   case OP_CHAR:  
3121  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3122    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3123  #else  #else
3124    c = *previous;    c = *previous;
3125  #endif  #endif
3126    return c != next;    }
3127    
3128    /* For CHARI (caseless character) we must check the other case. If we have  /* Now compare the next item with the previous opcode. First, handle cases when
3129    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3130    
3131    case OP_CHARI:  if (next >= 0)
3132  #ifdef SUPPORT_UTF    {
3133    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3134  #else    case, which maps to the special PT_CLIST property. Check this first. */
3135    c = *previous;  
 #endif  
   if (c == next) return FALSE;  
 #ifdef SUPPORT_UTF  
   if (utf)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3136  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3137      othercase = UCD_OTHERCASE((unsigned int)next);    if (utf && (unsigned int)c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3138  #else      {
3139      othercase = NOTACHAR;      int ocs = UCD_CASESET(next);
3140  #endif      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
     return (unsigned int)c != othercase;  
3141      }      }
   else  
 #endif  /* SUPPORT_UTF */  
   return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */  
   
   case OP_NOT:  
 #ifdef SUPPORT_UTF  
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
3142  #endif  #endif
   return c == next;  
3143    
3144    case OP_NOTI:    switch(op_code)
3145        {
3146        case OP_CHAR:
3147        return c != next;
3148    
3149        /* For CHARI (caseless character) we must check the other case. If we have
3150        Unicode property support, we can use it to test the other case of
3151        high-valued characters. We know that next can have only one other case,
3152        because multi-other-case characters are dealt with above. */
3153    
3154        case OP_CHARI:
3155        if (c == next) return FALSE;
3156  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3157    GETCHARTEST(c, previous);      if (utf)
3158          {
3159          unsigned int othercase;
3160          if (next < 128) othercase = cd->fcc[next]; else
3161    #ifdef SUPPORT_UCP
3162          othercase = UCD_OTHERCASE((unsigned int)next);
3163  #else  #else
3164    c = *previous;        othercase = NOTACHAR;
3165  #endif  #endif
3166    if (c == next) return TRUE;        return (unsigned int)c != othercase;
3167          }
3168        else
3169    #endif  /* SUPPORT_UTF */
3170        return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3171    
3172        case OP_NOT:
3173        return c == next;
3174    
3175        case OP_NOTI:
3176        if (c == next) return TRUE;
3177  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3178    if (utf)      if (utf)
3179      {        {
3180      unsigned int othercase;        unsigned int othercase;
3181      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3182  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3183      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE((unsigned int)next);
3184  #else  #else
3185      othercase = NOTACHAR;        othercase = NOTACHAR;
3186  #endif  #endif
3187      return (unsigned int)c == othercase;        return (unsigned int)c == othercase;
3188      }        }
3189    else      else
3190  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3191    return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3192    
3193    /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.      /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.
3194    When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */      When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
3195    
3196    case OP_DIGIT:      case OP_DIGIT:
3197    return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3198    
3199    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3200    return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3201    
3202    case OP_WHITESPACE:      case OP_WHITESPACE:
3203    return next > 255 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3204    
3205    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3206    return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3207    
3208    case OP_WORDCHAR:      case OP_WORDCHAR:
3209    return next > 255 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3210    
3211    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3212    return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3213    
3214    case OP_HSPACE:      case OP_HSPACE:
3215    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3216    switch(next)      switch(next)
3217      {        {
3218      case 0x09:        HSPACE_CASES:
3219      case 0x20:        return op_code == OP_NOT_HSPACE;
     case 0xa0:  
     case 0x1680:  
     case 0x180e:  
     case 0x2000:  
     case 0x2001:  
     case 0x2002:  
     case 0x2003:  
     case 0x2004:  
     case 0x2005:  
     case 0x2006:  
     case 0x2007:  
     case 0x2008:  
     case 0x2009:  
     case 0x200A:  
     case 0x202f:  
     case 0x205f:  
     case 0x3000:  
     return op_code == OP_NOT_HSPACE;  
     default:  
     return op_code != OP_NOT_HSPACE;  
     }  
3220    
3221    case OP_ANYNL:        default:
3222    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3223    case OP_NOT_VSPACE:        }
3224    switch(next)  
3225      {      case OP_ANYNL:
3226      case 0x0a:      case OP_VSPACE:
3227      case 0x0b:      case OP_NOT_VSPACE:
3228      case 0x0c:      switch(next)
3229      case 0x0d:        {
3230      case 0x85:        VSPACE_CASES:
3231      case 0x2028:        return op_code == OP_NOT_VSPACE;
3232      case 0x2029:  
3233      return op_code == OP_NOT_VSPACE;        default:
3234      default:        return op_code != OP_NOT_VSPACE;
3235      return op_code != OP_NOT_VSPACE;        }
     }  
3236    
3237  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3238    case OP_PROP:      case OP_PROP:
3239    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3240    
3241    case OP_NOTPROP:      case OP_NOTPROP:
3242    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3243  #endif  #endif
3244    
3245    default:      default:
3246    return FALSE;      return FALSE;
3247        }
3248    }    }
3249    
   
3250  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP
3251  is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are  is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
3252  generated only when PCRE_UCP is *not* set, that is, when only ASCII  generated only when PCRE_UCP is *not* set, that is, when only ASCII
# Line 3217  switch(op_code) Line 3257  switch(op_code)
3257    {    {
3258    case OP_CHAR:    case OP_CHAR:
3259    case OP_CHARI:    case OP_CHARI:
 #ifdef SUPPORT_UTF  
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
3260    switch(-next)    switch(-next)
3261      {      {
3262      case ESC_d:      case ESC_d:
# Line 3246  switch(op_code) Line 3281  switch(op_code)
3281      case ESC_H:      case ESC_H:
3282      switch(c)      switch(c)
3283        {        {
3284        case 0x09:        HSPACE_CASES:
       case 0x20:  
       case 0xa0:  
       case 0x1680:  
       case 0x180e:  
       case 0x2000:  
       case 0x2001:  
       case 0x2002:  
       case 0x2003:  
       case 0x2004:  
       case 0x2005:  
       case 0x2006:  
       case 0x2007:  
       case 0x2008:  
       case 0x2009:  
       case 0x200A:  
       case 0x202f:  
       case 0x205f:  
       case 0x3000:  
3285        return -next != ESC_h;        return -next != ESC_h;
3286    
3287        default:        default:
3288        return -next == ESC_h;        return -next == ESC_h;
3289        }        }
# Line 3274  switch(op_code) Line 3292  switch(op_code)
3292      case ESC_V:      case ESC_V:
3293      switch(c)      switch(c)
3294        {        {
3295        case 0x0a:        VSPACE_CASES:
       case 0x0b:  
       case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
3296        return -next != ESC_v;        return -next != ESC_v;
3297    
3298        default:        default:
3299        return -next == ESC_v;        return -next == ESC_v;
3300        }        }
# Line 3386  switch(op_code) Line 3399  switch(op_code)
3399    
3400    
3401  /*************************************************  /*************************************************
3402    *        Add a character or range to a class     *
3403    *************************************************/
3404    
3405    /* This function packages up the logic of adding a character or range of
3406    characters to a class. The character values in the arguments will be within the
3407    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3408    mutually recursive with the function immediately below.
3409    
3410    Arguments:
3411      classbits     the bit map for characters < 256
3412      uchardptr     points to the pointer for extra data
3413      options       the options word
3414      cd            contains pointers to tables etc.
3415      start         start of range character
3416      end           end of range character
3417    
3418    Returns:        the number of < 256 characters added
3419                    the pointer to extra data is updated
3420    */
3421    
3422    static int
3423    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3424      compile_data *cd, unsigned int start, unsigned int end)
3425    {
3426    unsigned int c;
3427    int n8 = 0;
3428    
3429    /* If caseless matching is required, scan the range and process alternate
3430    cases. In Unicode, there are 8-bit characters that have alternate cases that
3431    are greater than 255 and vice-versa. Sometimes we can just extend the original
3432    range. */
3433    
3434    if ((options & PCRE_CASELESS) != 0)
3435      {
3436    #ifdef SUPPORT_UCP
3437      if ((options & PCRE_UTF8) != 0)
3438        {
3439        int rc;
3440        unsigned int oc, od;
3441    
3442        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3443        c = start;
3444    
3445        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3446          {
3447          /* Handle a single character that has more than one other case. */
3448    
3449          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3450            PRIV(ucd_caseless_sets) + rc, oc);
3451    
3452          /* Do nothing if the other case range is within the original range. */
3453    
3454          else if (oc >= start && od <= end) continue;
3455    
3456          /* Extend the original range if there is overlap, noting that if oc < c, we
3457          can't have od > end because a subrange is always shorter than the basic
3458          range. Otherwise, use a recursive call to add the additional range. */
3459    
3460          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3461          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3462          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3463          }
3464        }
3465      else
3466    #endif  /* SUPPORT_UCP */
3467    
3468      /* Not UTF-mode, or no UCP */
3469    
3470      for (c = start; c <= end && c < 256; c++)
3471        {
3472        SETBIT(classbits, cd->fcc[c]);
3473        n8++;
3474        }
3475      }
3476    
3477    /* Now handle the original range. Adjust the final value according to the bit
3478    length - this means that the same lists of (e.g.) horizontal spaces can be used
3479    in all cases. */
3480    
3481    #if defined COMPILE_PCRE8
3482    #ifdef SUPPORT_UTF
3483      if ((options & PCRE_UTF8) == 0)
3484    #endif
3485      if (end > 0xff) end = 0xff;
3486    
3487    #elif defined COMPILE_PCRE16
3488    #ifdef SUPPORT_UTF
3489      if ((options & PCRE_UTF16) == 0)
3490    #endif
3491      if (end > 0xffff) end = 0xffff;
3492    
3493    #elif defined COMPILE_PCRE32
3494    #ifdef SUPPORT_UTF
3495      if ((options & PCRE_UTF32) == 0)
3496        if (end > 0xffffu) end = 0xffffu; // FIXMEchpe rebase fix this
3497    #endif
3498    #endif /* COMPILE_PCRE[8|16|32] */
3499    
3500    /* If all characters are less than 256, use the bit map. Otherwise use extra
3501    data. */
3502    
3503    if (end < 0x100)
3504      {
3505      for (c = start; c <= end; c++)
3506        {
3507        n8++;
3508        SETBIT(classbits, c);
3509        }
3510      }
3511    
3512    else
3513      {
3514      pcre_uchar *uchardata = *uchardptr;
3515    
3516    #ifdef SUPPORT_UTF
3517      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3518        {
3519        if (start < end)
3520          {
3521          *uchardata++ = XCL_RANGE;
3522          uchardata += PRIV(ord2utf)(start, uchardata);
3523          uchardata += PRIV(ord2utf)(end, uchardata);
3524          }
3525        else if (start == end)
3526          {
3527          *uchardata++ = XCL_SINGLE;
3528          uchardata += PRIV(ord2utf)(start, uchardata);
3529          }
3530        }
3531      else
3532    #endif  /* SUPPORT_UTF */
3533    
3534      /* Without UTF support, character values are constrained by the bit length,
3535      and can only be > 256 for 16-bit and 32-bit libraries. */
3536    
3537    #ifdef COMPILE_PCRE8
3538        {}
3539    #else
3540      if (start < end)
3541        {
3542        *uchardata++ = XCL_RANGE;
3543        *uchardata++ = start;
3544        *uchardata++ = end;
3545        }
3546      else if (start == end)
3547        {
3548        *uchardata++ = XCL_SINGLE;
3549        *uchardata++ = start;
3550        }
3551    #endif
3552    
3553      *uchardptr = uchardata;   /* Updata extra data pointer */
3554      }
3555    
3556    return n8;    /* Number of 8-bit characters */
3557    }
3558    
3559    
3560    
3561    
3562    /*************************************************
3563    *        Add a list of characters to a class     *
3564    *************************************************/
3565    
3566    /* This function is used for adding a list of case-equivalent characters to a
3567    class, and also for adding a list of horizontal or vertical whitespace. If the
3568    list is in order (which it should be), ranges of characters are detected and
3569    handled appropriately. This function is mutually recursive with the function
3570    above.
3571    
3572    Arguments:
3573      classbits     the bit map for characters < 256
3574      uchardptr     points to the pointer for extra data
3575      options       the options word
3576      cd            contains pointers to tables etc.
3577      p             points to row of 32-bit values, terminated by NOTACHAR
3578      except        character to omit; this is used when adding lists of
3579                      case-equivalent characters to avoid including the one we
3580                      already know about
3581    
3582    Returns:        the number of < 256 characters added
3583                    the pointer to extra data is updated
3584    */
3585    
3586    static int
3587    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3588      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3589    {
3590    int n8 = 0;
3591    while (p[0] < NOTACHAR)
3592      {
3593      int n = 0;
3594      if (p[0] != except)
3595        {
3596        while(p[n+1] == p[0] + n + 1) n++;
3597        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3598        }
3599      p += n + 1;
3600      }
3601    return n8;
3602    }
3603    
3604    
3605    
3606    /*************************************************
3607    *    Add characters not in a list to a class     *
3608    *************************************************/
3609    
3610    /* This function is used for adding the complement of a list of horizontal or
3611    vertical whitespace to a class. The list must be in order.
3612    
3613    Arguments:
3614      classbits     the bit map for characters < 256
3615      uchardptr     points to the pointer for extra data
3616      options       the options word
3617      cd            contains pointers to tables etc.
3618      p             points to row of 32-bit values, terminated by NOTACHAR
3619    
3620    Returns:        the number of < 256 characters added
3621                    the pointer to extra data is updated
3622    */
3623    
3624    static int
3625    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3626      int options, compile_data *cd, const pcre_uint32 *p)
3627    {
3628    int n8 = 0;
3629    if (p[0] > 0)
3630      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3631    while (p[0] < NOTACHAR)
3632      {
3633      while (p[1] == p[0] + 1) p++;
3634      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3635        (p[1] == NOTACHAR)? 0x10ffff : p[1] - 1);
3636      p++;
3637      }
3638    return n8;
3639    }
3640    
3641    
3642    
3643    /*************************************************
3644  *           Compile one branch                   *  *           Compile one branch                   *
3645  *************************************************/  *************************************************/
3646    
# Line 3448  must not do this for other options (e.g. Line 3703  must not do this for other options (e.g.
3703  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3704    
3705  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3706  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */
3707  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3708  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3709  #else  #else
3710  BOOL utf = FALSE;  BOOL utf = FALSE;
3711  #endif  #endif
3712    
3713  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3714    class_uchardata always so that it can be passed to add_to_class() always,
3715    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3716    alternative calls for the different cases. */
3717    
3718    pcre_uchar *class_uchardata;
3719  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3720  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3721  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3722  #endif  #endif
3723    
# Line 3503  for (;; ptr++) Line 3761  for (;; ptr++)
3761    BOOL is_recurse;    BOOL is_recurse;
3762    BOOL reset_bracount;    BOOL reset_bracount;
3763    int class_has_8bitchar;    int class_has_8bitchar;
3764    int class_single_char;    int class_one_char;
3765    int newoptions;    int newoptions;
3766    int recno;    int recno;
3767    int refsign;    int refsign;
# Line 3801  for (;; ptr++) Line 4059  for (;; ptr++)
4059    
4060      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4061    
4062      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4063      class_has_8bitchar will be non-zero, if the class contains at least one      class_has_8bitchar will be non-zero if the class contains at least one <
4064      < 256 character. class_single_char will be 1 if the class contains only      256 character; class_one_char will be 1 if the class contains just one
4065      a single character. */      character. */
4066    
4067      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4068      class_single_char = 0;      class_one_char = 0;
4069    
4070      /* Initialize the 32-char bit map to all zeros. We build the map in a      /* Initialize the 32-char bit map to all zeros. We build the map in a
4071      temporary bit of memory, in case the class contains only 1 character (less      temporary bit of memory, in case the class contains fewer than two
4072      than 256), because in that case the compiled code doesn't use the bit map.      8-bit characters because in that case the compiled code doesn't use the bit
4073      */      map. */
4074    
4075      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4076    
4077  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4078      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4079      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4080      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4081  #endif  #endif
4082    
4083      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
# Line 3830  for (;; ptr++) Line 4088  for (;; ptr++)
4088        {        {
4089        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4090    
4091  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4092        if (utf && HAS_EXTRALEN(c))        if (utf && HAS_EXTRALEN(c))
4093          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4094          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
# Line 3841  for (;; ptr++) Line 4099  for (;; ptr++)
4099        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4100        data and reset the pointer. This is so that very large classes that        data and reset the pointer. This is so that very large classes that
4101        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4102        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4103          however. */
4104    
4105        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4106          {          {
4107            xclass = TRUE;
4108          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4109          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4110          }          }
# Line 3903  for (;; ptr++) Line 4163  for (;; ptr++)
4163          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4164          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4165    
4166          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4167            posix_class = 0;            posix_class = 0;
4168    
4169          /* When PCRE_UCP is set, some of the POSIX classes are converted to          /* When PCRE_UCP is set, some of the POSIX classes are converted to
# Line 3946  for (;; ptr++) Line 4206  for (;; ptr++)
4206              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4207            }            }
4208    
4209          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4210          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4211    
4212          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3962  for (;; ptr++) Line 4222  for (;; ptr++)
4222            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4223    
4224          ptr = tempptr + 1;          ptr = tempptr + 1;
4225          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4226          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4227          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4228          class_single_char = 2;          class_one_char = 2;
4229          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4230          }          }
4231    
# Line 3973  for (;; ptr++) Line 4233  for (;; ptr++)
4233        of the specials, which just set a flag. The sequence \b is a special        of the specials, which just set a flag. The sequence \b is a special
4234        case. Inside a class (and only there) it is treated as backspace. We        case. Inside a class (and only there) it is treated as backspace. We
4235        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4236        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4237        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4238        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4239        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
# Line 4006  for (;; ptr++) Line 4266  for (;; ptr++)
4266            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4267            class_has_8bitchar++;            class_has_8bitchar++;
4268            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4269            class_single_char += 2;            class_one_char += 2;
4270    
4271            switch (-c)            switch (-c)
4272              {              {
# Line 4042  for (;; ptr++) Line 4302  for (;; ptr++)
4302    
4303              /* Perl 5.004 onwards omits VT from \s, but we must preserve it              /* Perl 5.004 onwards omits VT from \s, but we must preserve it
4304              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4305              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4306                EBCDIC, so we lazily just adjust the appropriate bit. */
4307    
4308              case ESC_s:              case ESC_s:
4309              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4055  for (;; ptr++) Line 4316  for (;; ptr++)
4316              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4317              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4318              continue;              continue;
4319    
4320                /* The rest apply in both UCP and non-UCP cases. */
4321    
4322              case ESC_h:              case ESC_h:
4323              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4324              SETBIT(classbits, 0x20); /* SPACE */                PRIV(hspace_list), NOTACHAR);
             SETBIT(classbits, 0xa0); /* NSBP */  
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x1680;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x180e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2000;  
             *class_uchardata++ = 0x200a;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x202f;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x205f;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x3000;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x1680, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x180e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2000, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x200a, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x202f, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x205f, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x3000, class_uchardata);  
               }  
 #endif  
4325              continue;              continue;
4326    
4327              case ESC_H:              case ESC_H:
4328              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4329                {                cd, PRIV(hspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x09/8: x ^= 1 << (0x09%8); break;  
                 case 0x20/8: x ^= 1 << (0x20%8); break;  
                 case 0xa0/8: x ^= 1 << (0xa0%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x0100;  
             *class_uchardata++ = 0x167f;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x1681;  
             *class_uchardata++ = 0x180d;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x180f;  
             *class_uchardata++ = 0x1fff;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x200b;  
             *class_uchardata++ = 0x202e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2030;  
             *class_uchardata++ = 0x205e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2060;  
             *class_uchardata++ = 0x2fff;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x3001;  
 #ifdef SUPPORT_UTF  
             if (utf)  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
             else  
 #endif  
               *class_uchardata++ = 0xffff;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x167f, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x1681, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x180d, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x180f, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x1fff, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x200b, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x202e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2030, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x205e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2060, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x2fff, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x3001, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
               }  
 #endif  
4330              continue;              continue;
4331    
4332              case ESC_v:              case ESC_v:
4333              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4334              SETBIT(classbits, 0x0b); /* VT */                PRIV(vspace_list), NOTACHAR);
             SETBIT(classbits, 0x0c); /* FF */  
             SETBIT(classbits, 0x0d); /* CR */  
             SETBIT(classbits, 0x85); /* NEL */  
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2028;  
             *class_uchardata++ = 0x2029;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2028, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x2029, class_uchardata);  
               }  
 #endif  
4335              continue;              continue;
4336    
4337              case ESC_V:              case ESC_V:
4338              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4339                {                cd, PRIV(vspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x0a/8: x ^= 1 << (0x0a%8);  
                              x ^= 1 << (0x0b%8);  
                              x ^= 1 << (0x0c%8);  
                              x ^= 1 << (0x0d%8);  
                              break;  
                 case 0x85/8: x ^= 1 << (0x85%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
   
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x0100;  
             *class_uchardata++ = 0x2027;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x202a;  
 #ifdef SUPPORT_UTF  
             if (utf)  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
             else  
 #endif  
               *class_uchardata++ = 0xffff;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x2027, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x202a, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
               }  
 #endif  
4340              continue;              continue;
4341    
4342  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4240  for (;; ptr++) Line 4347  for (;; ptr++)
4347                int pdata;                int pdata;
4348                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4349                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
               xclass = TRUE;  
4350                *class_uchardata++ = ((-c == ESC_p) != negated)?                *class_uchardata++ = ((-c == ESC_p) != negated)?
4351                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4352                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
# Line 4260  for (;; ptr++) Line 4366  for (;; ptr++)
4366                goto FAILED;                goto FAILED;
4367                }                }
4368              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4369              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4370              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4371              break;              break;
4372              }              }
4373            }            }
4374    
4375          /* Fall through if we have a single character (c >= 0). This may be          /* Fall through if the escape just defined a single character (c >= 0).
4376          greater than 256. */          This may be greater than 256. */
4377    
4378          }   /* End of backslash handling */          }   /* End of backslash handling */
4379    
4380        /* A single character may be followed by '-' to form a range. However,        /* A character may be followed by '-' to form a range. However, Perl does
4381        Perl does not permit ']' to be the end of the range. A '-' character        not permit ']' to be the end of the range. A '-' character at the end is
4382        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4383        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4384    
4385        CHECK_RANGE:        CHECK_RANGE:
4386        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4282  for (;; ptr++) Line 4388  for (;; ptr++)
4388          inescq = FALSE;          inescq = FALSE;
4389          ptr += 2;          ptr += 2;
4390          }          }
   
4391        oldptr = ptr;        oldptr = ptr;
4392    
4393        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4394    
4395        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4396    
# Line 4308  for (;; ptr++) Line 4413  for (;; ptr++)
4413            inescq = TRUE;            inescq = TRUE;
4414            break;            break;
4415            }            }
4416    
4417            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4418            back the pointer and jump to handle the character that preceded it. */
4419    
4420          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4421            {            {
4422            ptr = oldptr;            ptr = oldptr;
4423            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4424            }            }
4425    
4426            /* Otherwise, we have a potential range; pick up the next character */
4427    
4428  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4429          if (utf)          if (utf)
# Line 4333  for (;; ptr++) Line 4443  for (;; ptr++)
4443            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
4444            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4445    
4446            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4447    
4448            if (d < 0)            if (d < 0)
4449              {              {
4450              if (d == -ESC_b) d = CHAR_BS; else              if (d == -ESC_b) d = CHAR_BS; else
4451                {                {
4452                ptr = oldptr;                ptr = oldptr;
4453                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4454                }                }
4455              }              }
4456            }            }
4457    
4458          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4459          one-character ranges */          one-character ranges. */
4460    
4461          if (d < c)          if (d < c)
4462            {            {
4463            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4464            goto FAILED;            goto FAILED;
4465            }            }
4466            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4467    
4468          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4469            cannot be done anymore. Any value greater than 1 indicates that there
4470            is more than one character. */
4471    
4472            class_one_char = 2;
4473    
4474          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4475    
4476          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4477    
4478          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4479          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4480          class_single_char = 2;  
   
         /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless  
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
         if ((d > 255) || (utf && ((options & PCRE_CASELESS) != 0 && d > 127)))  
 #elif defined  SUPPORT_UTF  
         if (utf && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
 #elif !(defined COMPILE_PCRE8)  
         if (d > 255)  
 #endif  
 #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)  
           {  
           xclass = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
   
 #ifdef SUPPORT_UCP  
 #ifndef COMPILE_PCRE8  
           if (utf && (options & PCRE_CASELESS) != 0)  
 #else  
           if ((options & PCRE_CASELESS) != 0)  
 #endif  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
   
               if (occ == ocd)  
                 {  
                 *class_uchardata++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_uchardata++ = XCL_RANGE;  
                 class_uchardata += PRIV(ord2utf)(occ, class_uchardata);  
                 }  
               class_uchardata += PRIV(ord2utf)(ocd, class_uchardata);  
               }  
             }  
 #endif  /* SUPPORT_UCP */  
   
           /* Now record the original range, possibly modified for UCP caseless  
           overlapping ranges. */  
   
           *class_uchardata++ = XCL_RANGE;  
 #ifdef SUPPORT_UTF  
 #ifndef COMPILE_PCRE8  
           if (utf)  
             {  
             class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
             class_uchardata += PRIV(ord2utf)(d, class_uchardata);  
             }  
           else  
             {  
             *class_uchardata++ = c;  
             *class_uchardata++ = d;  
             }  
 #else  
           class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
           class_uchardata += PRIV(ord2utf)(d, class_uchardata);  
 #endif  
 #else /* SUPPORT_UTF */  
           *class_uchardata++ = c;  
           *class_uchardata++ = d;  
 #endif /* SUPPORT_UTF */  
   
           /* With UCP support, we are done. Without UCP support, there is no  
           caseless matching for UTF characters > 127; we can use the bit map  
           for the smaller ones. As for 16 bit characters without UTF, we  
           can still use  */  
   
 #ifdef SUPPORT_UCP  
 #ifndef COMPILE_PCRE8  
           if (utf)  
 #endif  
             continue;    /* With next character in the class */  
 #endif  /* SUPPORT_UCP */  
   
 #if defined SUPPORT_UTF && !defined(SUPPORT_UCP) && !(defined COMPILE_PCRE8)  
           if (utf)  
             {  
             if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
             /* Adjust upper limit and fall through to set up the map */  
             d = 127;  
             }  
           else  
             {  
             if (c > 255) continue;  
             /* Adjust upper limit and fall through to set up the map */  
             d = 255;  
             }  
 #elif defined SUPPORT_UTF && !defined(SUPPORT_UCP)  
           if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
           /* Adjust upper limit and fall through to set up the map */  
           d = 127;  
 #else  
           if (c > 255) continue;  
           /* Adjust upper limit and fall through to set up the map */  
           d = 255;  
 #endif  /* SUPPORT_UTF && !SUPPORT_UCP && !COMPILE_PCRE8 */  
           }  
 #endif  /* SUPPORT_UTF || !COMPILE_PCRE8 */  
   
         /* We use the bit map for 8 bit mode, or when the characters fall  
         partially or entirely to [0-255] ([0-127] for UCP) ranges. */  
   
         class_has_8bitchar = 1;  
   
         /* We can save a bit of time by skipping this in the pre-compile. */  
   
         if (lengthptr == NULL) for (; c <= d; c++)  
           {  
           classbits[c/8] |= (1 << (c&7));  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             int uc = cd->fcc[c]; /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
           }  
   
4481          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4482          }          }
4483    
4484        /* Handle a lone single character - we can get here for a normal        /* Handle a single character - we can get here for a normal non-escape
4485        non-escape char, or after \ that introduces a single character or for an        char, or after \ that introduces a single character or for an apparent
4486        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4487          increase it if it is already 2 or more ... just in case there's a class
4488        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4489    
4490        /* Only the value of 1 matters for class_single_char. */        CLASS_SINGLE_CHARACTER:
4491          if (class_one_char < 2) class_one_char++;
4492        if (class_single_char < 2) class_single_char++;  
4493          /* If class_one_char is 1, we have the first single character in the
4494        /* If class_charcount is 1, we saw precisely one character. As long as        class, and there have been no prior ranges, or XCLASS items generated by
4495        there was no use of \p or \P, in other words, no use of any XCLASS        escapes. If this is the final character in the class, we can optimize by
4496        features, we can optimize.        turning the item into a 1-character OP_CHAR[I] if it's positive, or
4497          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4498        The optimization throws away the bit map. We turn the item into a        to be set. Otherwise, there can be no first char if this item is first,
4499        1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.        whatever repeat count may follow. In the case of reqchar, save the
4500        In the positive case, it can cause firstchar to be set. Otherwise, there        previous value for reinstating. */
       can be no first char if this item is first, whatever repeat count may  
       follow. In the case of reqchar, save the previous value for reinstating. */  
4501    
4502        if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
4503          {          {
4504          ptr++;          ptr++;
4505          zeroreqchar = reqchar;          zeroreqchar = reqchar;
4506    
4507          if (negate_class)          if (negate_class)
4508            {            {
4509    #ifdef SUPPORT_UCP
4510                // FIXMEchpe pcreuint32?
4511              int d;
4512    #endif
4513            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4514            zerofirstchar = firstchar;            zerofirstchar = firstchar;
4515            *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
4516  #ifdef SUPPORT_UTF            /* For caseless UTF-8 mode when UCP support is available, check
4517            if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)            whether this character has more than one other case. If so, generate
4518              code += PRIV(ord2utf)(c, code);            a special OP_NOTPROP item instead of OP_NOTI. */
4519            else  
4520    #ifdef SUPPORT_UCP
4521              if (utf && (options & PCRE_CASELESS) != 0 &&
4522                  (d = UCD_CASESET(c)) != 0)
4523                {
4524                *code++ = OP_NOTPROP;
4525                *code++ = PT_CLIST;
4526                *code++ = d;
4527                }
4528              else
4529  #endif  #endif
4530              *code++ = c;            /* Char has only one other case, or UCP not available */
4531            goto NOT_CHAR;  
4532                {
4533                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4534    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4535                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4536                  code += PRIV(ord2utf)(c, code);
4537                else
4538    #endif
4539                  *code++ = c;
4540                }
4541    
4542              /* We are finished with this character class */
4543    
4544              goto END_CLASS;
4545            }            }
4546    
4547          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4548          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4549    
4550  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4551          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4552            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4553          else          else
# Line 4562  for (;; ptr++) Line 4558  for (;; ptr++)
4558            }            }
4559          goto ONE_CHAR;          goto ONE_CHAR;
4560          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4561    
4562        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4563          has been generated. Add this character to the class. */
4564  #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
4565        if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))        class_has_8bitchar +=
4566  #elif defined SUPPORT_UTF          add_to_class(classbits, &class_uchardata, options, cd, c, c);
       if (utf && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
 #elif !(defined COMPILE_PCRE8)  
       if (c > 255)  
 #endif  
   
 #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)  
         {  
         xclass = TRUE;  
         *class_uchardata++ = XCL_SINGLE;  
 #ifdef SUPPORT_UTF  
 #ifndef COMPILE_PCRE8  
         /* In non 8 bit mode, we can get here even if we are not in UTF mode. */  
         if (!utf)  
           *class_uchardata++ = c;  
         else  
 #endif  
           class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
 #else /* SUPPORT_UTF */  
         *class_uchardata++ = c;  
 #endif /* SUPPORT_UTF */  
   
 #ifdef SUPPORT_UCP  
 #ifdef COMPILE_PCRE8  
         if ((options & PCRE_CASELESS) != 0)  
 #else  
         /* In non 8 bit mode, we can get here even if we are not in UTF mode. */  
         if (utf && (options & PCRE_CASELESS) != 0)  
 #endif  
           {  
           unsigned int othercase;  
           if ((int)(othercase = UCD_OTHERCASE(c)) != c)  
             {  
             *class_uchardata++ = XCL_SINGLE;  
             class_uchardata += PRIV(ord2utf)(othercase, class_uchardata);  
             }  
           }  
 #endif  /* SUPPORT_UCP */  
   
         }  
       else  
 #endif  /* SUPPORT_UTF || COMPILE_PCRE16 */  
   
       /* Handle a single-byte character */  
         {  
         class_has_8bitchar = 1;  
         classbits[c/8] |= (1 << (c&7));  
         if ((options & PCRE_CASELESS) != 0)  
           {  
           c = cd->fcc[c]; /* flip case */  
           classbits[c/8] |= (1 << (c&7));  
           }  
         }  
4567        }        }
4568    
4569      /* Loop until ']' reached. This "while" is the end of the "do" far above.      /* Loop until ']' reached. This "while" is the end of the "do" far above.
# Line 4639  for (;; ptr++) Line 4583  for (;; ptr++)
4583        goto FAILED;        goto FAILED;
4584        }        }
4585    
4586        /* We will need an XCLASS if data has been placed in class_uchardata. In
4587        the second phase this is a sufficient test. However, in the pre-compile
4588        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4589        only if the very last character in the class needs XCLASS will it contain
4590        anything at this point. For this reason, xclass gets set TRUE above when
4591        uchar_classdata is emptied, and that's why this code is the way it is here
4592        instead of just doing a test on class_uchardata below. */
4593    
4594    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4595        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4596    #endif
4597    
4598      /* If this is the first thing in the branch, there can be no first char      /* If this is the first thing in the branch, there can be no first char
4599      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4600      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
# Line 4701  for (;; ptr++) Line 4657  for (;; ptr++)
4657        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4658        }        }
4659      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4660      NOT_CHAR:  
4661        END_CLASS:
4662      break;      break;
4663    
4664    
# Line 4811  for (;; ptr++) Line 4768  for (;; ptr++)
4768      if (*previous == OP_CHAR || *previous == OP_CHARI      if (*previous == OP_CHAR || *previous == OP_CHARI
4769          || *previous == OP_NOT || *previous == OP_NOTI)          || *previous == OP_NOT || *previous == OP_NOTI)
4770        {        {
4771        switch (*previous)        switch (*previous)
4772          {          {
4773          default: /* Make compiler happy. */          default: /* Make compiler happy. */
4774          case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;          case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
# Line 4825  for (;; ptr++) Line 4782  for (;; ptr++)
4782        hold the length of the character in bytes, plus UTF_LENGTH to flag that        hold the length of the character in bytes, plus UTF_LENGTH to flag that
4783        it's a length rather than a small character. */        it's a length rather than a small character. */
4784    
4785  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4786        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4787          {          {
4788          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4961  for (;; ptr++) Line 4918  for (;; ptr++)
4918    
4919          if (repeat_max < 0)          if (repeat_max < 0)
4920            {            {
4921  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4922            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4923              {              {
4924              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4986  for (;; ptr++) Line 4943  for (;; ptr++)
4943    
4944          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4945            {            {
4946  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4947            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4948              {              {
4949              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5016  for (;; ptr++) Line 4973  for (;; ptr++)
4973    
4974        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
4975    
4976  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4977        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
4978          {          {
4979          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5503  for (;; ptr++) Line 5460  for (;; ptr++)
5460        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)
5461          {          {
5462          tempcode += PRIV(OP_lengths)[*tempcode];          tempcode += PRIV(OP_lengths)[*tempcode];
5463  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5464          if (utf && HAS_EXTRALEN(tempcode[-1]))          if (utf && HAS_EXTRALEN(tempcode[-1]))
5465            tempcode += GET_EXTRALEN(tempcode[-1]);            tempcode += GET_EXTRALEN(tempcode[-1]);
5466  #endif  #endif
# Line 5601  for (;; ptr++) Line 5558  for (;; ptr++)
5558          arg = ++ptr;          arg = ++ptr;
5559          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5560          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5561            if ((unsigned int)arglen > MAX_MARK)
5562              {
5563              *errorcodeptr = ERR75;
5564              goto FAILED;
5565              }
5566          }          }
5567    
5568        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5616  for (;; ptr++) Line 5578  for (;; ptr++)
5578          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5579              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5580            {            {
5581              int setverb;
5582    
5583            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5584            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5585    
# Line 5633  for (;; ptr++) Line 5597  for (;; ptr++)
5597                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5598                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5599                }                }
5600              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5601                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5602    
5603              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5604              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
# Line 5648  for (;; ptr++) Line 5613  for (;; ptr++)
5613                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5614                goto FAILED;                goto FAILED;
5615                }                }
5616              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5617              }              }
5618    
5619            else            else
# Line 5659  for (;; ptr++) Line 5623  for (;; ptr++)
5623                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5624                goto FAILED;                goto FAILED;
5625                }                }
5626              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5627              *code++ = arglen;              *code++ = arglen;
5628              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5629              code += arglen;              code += arglen;
5630              *code++ = 0;              *code++ = 0;
5631              }              }
5632    
5633              switch (setverb)
5634                {
5635                case OP_THEN:
5636                case OP_THEN_ARG:
5637                cd->external_flags |= PCRE_HASTHEN;
5638                break;
5639    
5640                case OP_PRUNE:
5641                case OP_PRUNE_ARG:
5642                case OP_SKIP:
5643                case OP_SKIP_ARG:
5644                cd->had_pruneorskip = TRUE;
5645                break;
5646                }
5647    
5648            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5649            }            }
5650    
# Line 6852  for (;; ptr++) Line 6830  for (;; ptr++)
6830        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6831        can obtain the OP value by negating the escape value in the default        can obtain the OP value by negating the escape value in the default
6832        situation when PCRE_UCP is not set. When it *is* set, we substitute        situation when PCRE_UCP is not set. When it *is* set, we substitute
6833        Unicode property tests. Note that \b and \B do a one-character        Unicode property tests. Note that \b and \B do a one-character
6834        lookbehind. */        lookbehind. */
6835    
6836        else        else
6837          {          {
6838          if ((-c == ESC_b || -c == ESC_B) && cd->max_lookbehind == 0)          if ((-c == ESC_b || -c == ESC_B) && cd->max_lookbehind == 0)
6839            cd->max_lookbehind = 1;            cd->max_lookbehind = 1;
6840  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6841          if (-c >= ESC_DU && -c <= ESC_wu)          if (-c >= ESC_DU && -c <= ESC_wu)
6842            {            {
# Line 6882  for (;; ptr++) Line 6860  for (;; ptr++)
6860      a value > 127. We set its representation in the length/buffer, and then      a value > 127. We set its representation in the length/buffer, and then
6861      handle it as a data character. */      handle it as a data character. */
6862    
6863  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6864      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6865        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6866      else      else
# Line 6905  for (;; ptr++) Line 6883  for (;; ptr++)
6883      mclength = 1;      mclength = 1;
6884      mcbuffer[0] = c;      mcbuffer[0] = c;
6885    
6886  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6887      if (utf && HAS_EXTRALEN(c))      if (utf && HAS_EXTRALEN(c))
6888        ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr));        ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr));
6889  #endif  #endif
# Line 6915  for (;; ptr++) Line 6893  for (;; ptr++)
6893    
6894      ONE_CHAR:      ONE_CHAR:
6895      previous = code;      previous = code;
6896    
6897        /* For caseless UTF-8 mode when UCP support is available, check whether
6898        this character has more than one other case. If so, generate a special
6899        OP_PROP item instead of OP_CHARI. */
6900    
6901    #ifdef SUPPORT_UCP
6902        if (utf && (options & PCRE_CASELESS) != 0)
6903          {
6904          GETCHAR(c, mcbuffer);
6905          if ((c = UCD_CASESET(c)) != 0)
6906            {
6907            *code++ = OP_PROP;
6908            *code++ = PT_CLIST;
6909            *code++ = c;
6910            if (firstchar == REQ_UNSET) firstchar = zerofirstchar = REQ_NONE;
6911            break;
6912            }
6913          }
6914    #endif
6915    
6916        /* Caseful matches, or not one of the multicase characters. */
6917    
6918      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6919      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6920    
# Line 7166  for (;;) Line 7166  for (;;)
7166          *ptrptr = ptr;          *ptrptr = ptr;
7167          return FALSE;          return FALSE;
7168          }          }
7169        else        else
7170          {          {
7171          if (fixed_length > cd->max_lookbehind)          if (fixed_length > cd->max_lookbehind)
7172            cd->max_lookbehind = fixed_length;            cd->max_lookbehind = fixed_length;
7173          PUT(reverse_count, 0, fixed_length);          PUT(reverse_count, 0, fixed_length);
7174          }          }
7175        }        }
7176      }      }
# Line 7303  and the highest back reference was great Line 7303  and the highest back reference was great
7303  However, by keeping a bitmap of the first 31 back references, we can catch some  However, by keeping a bitmap of the first 31 back references, we can catch some
7304  of the more common cases more precisely.  of the more common cases more precisely.
7305    
7306    ... A second exception is when the .* appears inside an atomic group, because
7307    this prevents the number of characters it matches from being adjusted.
7308    
7309  Arguments:  Arguments:
7310    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7311    bracket_map    a bitmap of which brackets we are inside while testing; this    bracket_map    a bitmap of which brackets we are inside while testing; this
7312                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7313                    the less precise approach                    the less precise approach
7314    backref_map    the back reference bitmap    cd             points to the compile data block
7315      atomcount      atomic group level
7316    
7317  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7318  */  */
7319    
7320  static BOOL  static BOOL
7321  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7322    unsigned int backref_map)    compile_data *cd, int atomcount)
7323  {  {
7324  do {  do {
7325     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7327  do { Line 7331  do {
7331     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7332         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7333       {       {
7334       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7335       }       }
7336    
7337     /* Capturing brackets */     /* Capturing brackets */
# Line 7337  do { Line 7341  do {
7341       {       {
7342       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7343       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7344       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7345       }       }
7346    
7347     /* Other brackets */     /* Positive forward assertions and conditions */
7348    
7349     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ASSERT || op == OP_COND)
             op == OP_COND)  
7350       {       {
7351       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7352         }
7353    
7354       /* Atomic groups */
7355    
7356       else if (op == OP_ONCE || op == OP_ONCE_NC)
7357         {
7358         if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7359           return FALSE;
7360       }       }
7361    
7362     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7363     it isn't in brackets that are or may be referenced. */     it isn't in brackets that are or may be referenced or inside an atomic
7364       group. */
7365    
7366     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7367               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7368       {       {
7369       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7370             atomcount > 0 || cd->had_pruneorskip)
7371         return FALSE;         return FALSE;
7372       }       }
7373    
7374     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7375    
7376     else if (op != OP_SOD && op != OP_SOM && op != OP_CIRC) return FALSE;     else if (op != OP_SOD && op != OP_SOM && op != OP_CIRC) return FALSE;
7377    
7378     code += GET(code, 1);     code += GET(code, 1);
7379     }     }
7380  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7378  return TRUE; Line 7392  return TRUE;
7392  matching and for non-DOTALL patterns that start with .* (which must start at  matching and for non-DOTALL patterns that start with .* (which must start at
7393  the beginning or after \n). As in the case of is_anchored() (see above), we  the beginning or after \n). As in the case of is_anchored() (see above), we
7394  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7395  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7396    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7397    count, because once again the assumption no longer holds.
7398    
7399  Arguments:  Arguments:
7400    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7401    bracket_map    a bitmap of which brackets we are inside while testing; this    bracket_map    a bitmap of which brackets we are inside while testing; this
7402                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7403                    the less precise approach                    the less precise approach
7404    backref_map    the back reference bitmap    cd             points to the compile data
7405      atomcount      atomic group level
7406    
7407  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7408  */  */
7409    
7410  static BOOL  static BOOL
7411  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7412    unsigned int backref_map)    compile_data *cd, int atomcount)
7413  {  {
7414  do {  do {
7415     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7418  do { Line 7435  do {
7435         return FALSE;         return FALSE;
7436    
7437         default:     /* Assertion */         default:     /* Assertion */
7438         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7439         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7440         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7441         break;         break;
# Line 7432  do { Line 7449  do {
7449     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7450         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7451       {       {
7452       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7453       }       }
7454    
7455     /* Capturing brackets */     /* Capturing brackets */
# Line 7442  do { Line 7459  do {
7459       {       {
7460       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7461       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7462       if (!is_startline(scode, new_map, backref_map)) return FALSE;       if (!is_startline(scode, new_map, cd, atomcount)) return FALSE;
7463       }       }
7464    
7465     /* Other brackets */     /* Positive forward assertions */
7466    
7467     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)     else if (op == OP_ASSERT)
7468       {       {
7469       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7470       }       }
7471    
7472       /* Atomic brackets */
7473    
7474     /* .* means "start at start or after \n" if it isn't in brackets that     else if (op == OP_ONCE || op == OP_ONCE_NC)
7475     may be referenced. */       {
7476         if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE;
7477         }
7478    
7479       /* .* means "start at start or after \n" if it isn't in atomic brackets or
7480       brackets that may be referenced, as long as the pattern does not contain
7481       *PRUNE or *SKIP, because these break the feature. Consider, for example,
7482       /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the
7483       start of a line. */
7484    
7485     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)
7486       {       {
7487       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 ||
7488             atomcount > 0 || cd->had_pruneorskip)
7489           return FALSE;
7490       }       }
7491    
7492     /* Check for explicit circumflex */     /* Check for explicit circumflex; anything else gives a FALSE result. Note
7493       in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC
7494       because the number of characters matched by .* cannot be adjusted inside
7495       them. */
7496    
7497     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;
7498    
# Line 7582  Returns:        pointer to compiled data Line 7614  Returns:        pointer to compiled data
7614                  with errorptr and erroroffset set                  with errorptr and erroroffset set
7615  */  */
7616    
7617  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7618  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7619  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
7620    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7621  #else  #elif defined COMPILE_PCRE16
7622  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7623  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,
7624    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7625    #elif defined COMPILE_PCRE32
7626    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7627    pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr,
7628      int *erroroffset, const unsigned char *tables)
7629  #endif  #endif
7630  {  {
7631  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7632  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7633  #else  #elif defined COMPILE_PCRE16
7634  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7635    #elif defined COMPILE_PCRE32
7636    return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7637  #endif  #endif
7638  }  }
7639    
7640    
7641  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7642  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7643  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
7644    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7645  #else  #elif defined COMPILE_PCRE16
7646  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7647  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,
7648    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7649    #elif defined COMPILE_PCRE32
7650    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7651    pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr,
7652      const char **errorptr, int *erroroffset, const unsigned char *tables)
7653  #endif  #endif
7654  {  {
7655  REAL_PCRE *re;  REAL_PCRE *re;
# Line 7693  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7735  while (ptr[skipatstart] == CHAR_LEFT_PAR
7735    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7736      { skipatstart += 8; options |= PCRE_UTF16; continue; }      { skipatstart += 8; options |= PCRE_UTF16; continue; }
7737  #endif  #endif
7738    #ifdef COMPILE_PCRE32
7739      if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7740        { skipatstart += 8; options |= PCRE_UTF32; continue; }
7741    #endif
7742    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)
7743      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
7744    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 7721  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7767  while (ptr[skipatstart] == CHAR_LEFT_PAR
7767    else break;    else break;
7768    }    }
7769    
7770  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */
7771  utf = (options & PCRE_UTF8) != 0;  utf = (options & PCRE_UTF8) != 0;
7772    
7773  /* Can't support UTF unless PCRE has been compiled to include the code. The  /* Can't support UTF unless PCRE has been compiled to include the code. The
# Line 7733  not used here. */ Line 7779  not used here. */
7779  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&
7780       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)
7781    {    {
7782  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7783    errorcode = ERR44;    errorcode = ERR44;
7784  #else  #elif defined COMPILE_PCRE16
7785    errorcode = ERR74;    errorcode = ERR74;
7786    #elif defined COMPILE_PCRE32
7787      errorcode = ERR77;
7788  #endif  #endif
7789    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;
7790    }    }
# Line 7900  re->name_count = cd->names_found; Line 7948  re->name_count = cd->names_found;
7948  re->ref_count = 0;  re->ref_count = 0;
7949  re->tables = (tables == PRIV(default_tables))? NULL : tables;  re->tables = (tables == PRIV(default_tables))? NULL : tables;
7950  re->nullpad = NULL;  re->nullpad = NULL;
7951    #ifdef COMPILE_PCRE32
7952    re->dummy1 = re->dummy2 = 0;
7953    #endif
7954    
7955  /* The starting points of the name/number translation table and of the code are  /* The starting points of the name/number translation table and of the code are
7956  passed around in the compile data block. The start/end pattern and initial  passed around in the compile data block. The start/end pattern and initial
# Line 7919  cd->start_code = codestart; Line 7970  cd->start_code = codestart;
7970  cd->hwm = (pcre_uchar *)(cd->start_workspace);  cd->hwm = (pcre_uchar *)(cd->start_workspace);
7971  cd->req_varyopt = 0;  cd->req_varyopt = 0;
7972  cd->had_accept = FALSE;  cd->had_accept = FALSE;
7973    cd->had_pruneorskip = FALSE;
7974  cd->check_lookbehind = FALSE;  cd->check_lookbehind = FALSE;
7975  cd->open_caps = NULL;  cd->open_caps = NULL;
7976    
# Line 8042  if (errorcode != 0) Line 8094  if (errorcode != 0)
8094    }    }
8095    
8096  /* If the anchored option was not passed, set the flag if we can determine that  /* If the anchored option was not passed, set the flag if we can determine that
8097  the pattern is anchored by virtue of ^ characters or \A or anything else (such  the pattern is anchored by virtue of ^ characters or \A or anything else, such
8098  as starting with .* when DOTALL is set).  as starting with non-atomic .* when DOTALL is set and there are no occurrences
8099    of *PRUNE or *SKIP.
8100    
8101  Otherwise, if we know what the first byte has to be, save it, because that  Otherwise, if we know what the first byte has to be, save it, because that
8102  speeds up unanchored matches no end. If not, see if we can set the  speeds up unanchored matches no end. If not, see if we can set the
8103  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches
8104  start with ^. and also when all branches start with .* for non-DOTALL matches.  start with ^. and also when all branches start with non-atomic .* for
8105  */  non-DOTALL matches when *PRUNE and SKIP are not present. */
8106    
8107  if ((re->options & PCRE_ANCHORED) == 0)  if ((re->options & PCRE_ANCHORED) == 0)
8108    {    {
8109    if (is_anchored(codestart, 0, cd->backref_map))    if (is_anchored(codestart, 0, cd, 0)) re->options |= PCRE_ANCHORED;
     re->options |= PCRE_ANCHORED;  
8110    else    else
8111      {      {
8112      if (firstchar < 0)      if (firstchar < 0)
8113        firstchar = find_firstassertedchar(codestart, FALSE);        firstchar = find_firstassertedchar(codestart, FALSE);
8114      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */
8115        {        {
8116  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8117        re->first_char = firstchar & 0xff;        re->first_char = firstchar & 0xff;
8118  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8119        re->first_char = firstchar & 0xffff;        re->first_char = firstchar & 0xffff;
8120  #endif  #elif defined COMPILE_PCRE32
8121          re->first_char = firstchar & ~REQ_MASK;
8122  #endif  #endif
8123        if ((firstchar & REQ_CASELESS) != 0)        if ((firstchar & REQ_CASELESS) != 0)
8124          {          {
# Line 8091  if ((re->options & PCRE_ANCHORED) == 0) Line 8143  if ((re->options & PCRE_ANCHORED) == 0)
8143    
8144        re->flags |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
8145        }        }
8146      else if (is_startline(codestart, 0, cd->backref_map))  
8147        re->flags |= PCRE_STARTLINE;      else if (is_startline(codestart, 0, cd, 0)) re->flags |= PCRE_STARTLINE;
8148      }      }
8149    }    }
8150    
# Line 8103  bytes. */ Line 8155  bytes. */
8155  if (reqchar >= 0 &&  if (reqchar >= 0 &&
8156       ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0))       ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0))
8157    {    {
8158  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8159    re->req_char = reqchar & 0xff;    re->req_char = reqchar & 0xff;
8160  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8161    re->req_char = reqchar & 0xffff;    re->req_char = reqchar & 0xffff;
8162  #endif  #elif defined COMPILE_PCRE32
8163      re->req_char = reqchar & ~REQ_MASK;
8164  #endif  #endif
8165    if ((reqchar & REQ_CASELESS) != 0)    if ((reqchar & REQ_CASELESS) != 0)
8166      {      {
# Line 8160  if ((re->flags & PCRE_REQCHSET) != 0) Line 8212  if ((re->flags & PCRE_REQCHSET) != 0)
8212      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
8213    }    }
8214    
8215  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8216  pcre_printint((pcre *)re, stdout, TRUE);  pcre_printint((pcre *)re, stdout, TRUE);
8217  #else  #elif defined COMPILE_PCRE16
8218  pcre16_printint((pcre *)re, stdout, TRUE);  pcre16_printint((pcre *)re, stdout, TRUE);
8219    #elif defined COMPILE_PCRE32
8220    pcre32_printint((pcre *)re, stdout, TRUE);
8221  #endif  #endif
8222    
8223  /* This check is done here in the debugging case so that the code that  /* This check is done here in the debugging case so that the code that
# Line 8179  if (code - codestart > length) Line 8233  if (code - codestart > length)
8233    }    }
8234  #endif   /* PCRE_DEBUG */  #endif   /* PCRE_DEBUG */
8235    
8236  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8237  return (pcre *)re;  return (pcre *)re;
8238  #else  #elif defined COMPILE_PCRE16
8239  return (pcre16 *)re;  return (pcre16 *)re;
8240    #elif defined COMPILE_PCRE32
8241    return (pcre32 *)re;
8242  #endif  #endif
8243  }  }
8244    

Legend:
Removed from v.962  
changed lines
  Added in v.1055

  ViewVC Help
Powered by ViewVC 1.1.5