/[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 924 by zherczeg, Wed Feb 22 10:23:56 2012 UTC revision 1059 by chpe, Tue Oct 16 15:53:53 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 739  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 749  return (*p == CHAR_RIGHT_CURLY_BRACKET);
749  *************************************************/  *************************************************/
750    
751  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
752  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
753  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group
754  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  n is returned as ESC_REF + n; ESC_REF is the highest ESC_xxx macro. When
755  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  UTF-8 is enabled, a positive value greater than 255 may be returned in chptr.
756  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
757  sequence.  escape sequence.
758    
759  Arguments:  Arguments:
760    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
761      chptr          points to the data character
762    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
763    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
764    options        the options bits    options        the options bits
765    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
766    
767  Returns:         zero or positive => a data character  Returns:         zero => a data character
768                   negative => a special escape sequence                   positive => a special escape sequence
769                   on error, errorcodeptr is set                   on error, errorcodeptr is set
770  */  */
771    
772  static int  static int
773  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, int *chptr, int *errorcodeptr,
774    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
775  {  {
776  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
777  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
778  const pcre_uchar *ptr = *ptrptr + 1;  const pcre_uchar *ptr = *ptrptr + 1;
779  pcre_int32 c;  pcre_int32 c;
780    int escape = 0;
781  int i;  int i;
782    
783  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
# Line 782  Otherwise further processing may be requ Line 794  Otherwise further processing may be requ
794  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
795  /* Not alphanumeric */  /* Not alphanumeric */
796  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
797  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = i; else escape = -i; }
798    
799  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
800  /* Not alphanumeric */  /* Not alphanumeric */
801  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)) {}
802  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = i; else escape = -i; }
803  #endif  #endif
804    
805  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
# Line 829  else Line 841  else
841            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
842  #endif  #endif
843            }            }
844    
845    #if defined COMPILE_PCRE8
846            if (c > (utf ? 0x10ffff : 0xff))
847    #elif defined COMPILE_PCRE16
848            if (c > (utf ? 0x10ffff : 0xffff))
849    #elif defined COMPILE_PCRE32
850            if (utf && c > 0x10ffff)
851    #endif
852              {
853              *errorcodeptr = ERR76;
854              }
855            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
856          }          }
857        }        }
858      else      else
# Line 855  else Line 879  else
879      (3) For Oniguruma compatibility we also support \g followed by a name or a      (3) For Oniguruma compatibility we also support \g followed by a name or a
880      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
881      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
882      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
883    
884      case CHAR_g:      case CHAR_g:
885      if (isclass) break;      if (isclass) break;
886      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
887        {        {
888        c = -ESC_g;        escape = ESC_g;
889        break;        break;
890        }        }
891    
# Line 874  else Line 898  else
898          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
899        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
900          {          {
901          c = -ESC_k;          escape = ESC_k;
902          break;          break;
903          }          }
904        braced = TRUE;        braced = TRUE;
# Line 930  else Line 954  else
954        c = bracount - (c - 1);        c = bracount - (c - 1);
955        }        }
956    
957      c = -(ESC_REF + c);      escape = ESC_REF + c;
958      break;      break;
959    
960      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 971  else Line 995  else
995          }          }
996        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
997          {          {
998          c = -(ESC_REF + c);          escape = ESC_REF + c;
999          break;          break;
1000          }          }
1001        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 1049  else Line 1073  else
1073          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1074  #endif  #endif
1075    
1076  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1077          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }
1078  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
1079          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }
1080  #endif  #elif defined COMPILE_PCRE32
1081            if (utf && c > 0x10ffff) { c = -1; break; }
1082  #endif  #endif
1083          }          }
1084    
# Line 1139  else Line 1163  else
1163  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1164  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1165    
1166  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1167       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1168    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1169    
1170  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1171    
1172  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1173    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1174    
1175  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1176    
1177  *ptrptr = ptr;  *ptrptr = ptr;
1178  return c;  *chptr = c;
1179    return escape;
1180  }  }
1181    
   
   
1182  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1183  /*************************************************  /*************************************************
1184  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1347  Arguments: Line 1370  Arguments:
1370    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1371    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1372    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1373    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1374    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1375    
1376  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 1604  Arguments:
1604    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1605    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1606    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1607    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1608    
1609  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1610  */  */
# Line 1684  and doing the check at the end; a flag s Line 1707  and doing the check at the end; a flag s
1707    
1708  Arguments:  Arguments:
1709    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1710    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1711    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1712    cd       the "compile data" structure    cd       the "compile data" structure
1713    
# Line 1818  for (;;) Line 1841  for (;;)
1841      case OP_NOTI:      case OP_NOTI:
1842      branchlength++;      branchlength++;
1843      cc += 2;      cc += 2;
1844  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1845      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1846  #endif  #endif
1847      break;      break;
# Line 1832  for (;;) Line 1855  for (;;)
1855      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1856      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1857      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1858  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1859      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1860  #endif  #endif
1861      break;      break;
1862    
1863      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1864      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1865      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)
1866          cc += 2;
1867      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1868      break;      break;
1869    
# Line 1874  for (;;) Line 1898  for (;;)
1898    
1899      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1900    
1901  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1902      case OP_XCLASS:      case OP_XCLASS:
1903      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1904      /* Fall through */      /* Fall through */
# Line 2013  length. Line 2037  length.
2037    
2038  Arguments:  Arguments:
2039    code        points to start of expression    code        points to start of expression
2040    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2041    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2042    
2043  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 2101  for (;;)
2101        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2102        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2103        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2104        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2105          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2106        break;        break;
2107    
2108        case OP_MARK:        case OP_MARK:
# Line 2100  for (;;) Line 2124  for (;;)
2124    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
2125    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2126    
2127  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2128      if (utf) switch(c)      if (utf) switch(c)
2129        {        {
2130        case OP_CHAR:        case OP_CHAR:
# Line 2152  instance of OP_RECURSE. Line 2176  instance of OP_RECURSE.
2176    
2177  Arguments:  Arguments:
2178    code        points to start of expression    code        points to start of expression
2179    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2180    
2181  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
2182  */  */
# Line 2197  for (;;) Line 2221  for (;;)
2221        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2222        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2223        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2224        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2225          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2226        break;        break;
2227    
2228        case OP_MARK:        case OP_MARK:
# Line 2220  for (;;) Line 2244  for (;;)
2244      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
2245      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2246    
2247  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2248      if (utf) switch(c)      if (utf) switch(c)
2249        {        {
2250        case OP_CHAR:        case OP_CHAR:
# Line 2306  bracket whose current branch will alread Line 2330  bracket whose current branch will alread
2330  Arguments:  Arguments:
2331    code        points to start of search    code        points to start of search
2332    endcode     points to where to stop    endcode     points to where to stop
2333    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2334    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2335    
2336  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 2547  for (code = first_significant_code(code
2547      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2548      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2549      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2550      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2551        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2552      break;      break;
2553    
2554      /* End of branch */      /* End of branch */
# Line 2539  for (code = first_significant_code(code Line 2563  for (code = first_significant_code(code
2563      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2564      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2565    
2566  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2567      case OP_STAR:      case OP_STAR:
2568      case OP_STARI:      case OP_STARI:
2569      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2605  Arguments: Line 2629  Arguments:
2629    code        points to start of the recursion    code        points to start of the recursion
2630    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2631    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2632    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2633    cd          pointers to tables etc    cd          pointers to tables etc
2634    
2635  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 2776  value in the reference (which is a group
2776  Arguments:  Arguments:
2777    group      points to the start of the group    group      points to the start of the group
2778    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2779    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2780    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2781    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
2782    
# Line 2856  PUT(previous_callout, 2 + LINK_SIZE, len Line 2880  PUT(previous_callout, 2 + LINK_SIZE, len
2880  *************************************************/  *************************************************/
2881    
2882  /* 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
2883  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
2884  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
2885  start address.  start address. A character with multiple other cases is returned on its own
2886    with a special return value.
2887    
2888  Arguments:  Arguments:
2889    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2866  Arguments: Line 2891  Arguments:
2891    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2892    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2893    
2894  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2895                   0 when a range is returned
2896                  >0 the CASESET offset for char with multiple other cases
2897                    in this case, ocptr contains the original
2898  */  */
2899    
2900  static BOOL  static int
2901  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,
2902    unsigned int *odptr)    unsigned int *odptr)
2903  {  {
2904  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2905    int co;
2906    
2907    /* Find the first character that has an other case. If it has multiple other
2908    cases, return its case offset value. */
2909    
2910  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2911    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2912      if ((co = UCD_CASESET(c)) != 0)
2913        {
2914        *ocptr = c++;   /* Character that has the set */
2915        *cptr = c;      /* Rest of input range */
2916        return co;
2917        }
2918      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2919      }
2920    
2921  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2922    
2923  *ocptr = othercase;  *ocptr = othercase;
2924  next = othercase + 1;  next = othercase + 1;
# Line 2889  for (++c; c <= d; c++) Line 2929  for (++c; c <= d; c++)
2929    next++;    next++;
2930    }    }
2931    
2932  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2933  *cptr = c;  *cptr = c;             /* Rest of input range */
2934    return 0;
 return TRUE;  
2935  }  }
2936    
2937    
# Line 2916  Returns:       TRUE if auto-possessifyin Line 2955  Returns:       TRUE if auto-possessifyin
2955  static BOOL  static BOOL
2956  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(int c, int ptype, int pdata, BOOL negated)
2957  {  {
2958    #ifdef SUPPORT_UCP
2959    const pcre_uint32 *p;
2960    #endif
2961    
2962  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2963    
2964  switch(ptype)  switch(ptype)
2965    {    {
2966    case PT_LAMP:    case PT_LAMP:
# Line 2954  switch(ptype) Line 2998  switch(ptype)
2998    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2999            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3000            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3001    
3002    #ifdef SUPPORT_UCP
3003      case PT_CLIST:
3004      p = PRIV(ucd_caseless_sets) + prop->caseset;
3005      for (;;)
3006        {
3007        if ((unsigned int)c < *p) return !negated;
3008        if ((unsigned int)c == *p++) return negated;
3009        }
3010      break;  /* Control never reaches here */
3011    #endif
3012    }    }
3013    
3014  return FALSE;  return FALSE;
3015  }  }
3016  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2971  sense to automatically possessify the re Line 3027  sense to automatically possessify the re
3027    
3028  Arguments:  Arguments:
3029    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3030    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3031    ptr           next character in pattern    ptr           next character in pattern
3032    options       options bits    options       options bits
3033    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2983  static BOOL Line 3039  static BOOL
3039  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3040    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3041  {  {
3042  pcre_int32 c, next;  pcre_int32 c = NOTACHAR; // FIXMEchpe pcre_uint32
3043    pcre_int32 next;
3044    int escape;
3045  int op_code = *previous++;  int op_code = *previous++;
3046    
3047  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 3015  value is a character, a negative value i Line 3073  value is a character, a negative value i
3073  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3074    {    {
3075    int temperrorcode = 0;    int temperrorcode = 0;
3076    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3077    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3078    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3079    }    }
3080  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
3081    {    {
3082      escape = 0;
3083  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3084    if (utf) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3085  #endif  #endif
# Line 3057  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3116  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3116    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)
3117      return FALSE;      return FALSE;
3118    
3119  /* 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. */  
3120    
3121  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3122        op_code == OP_NOT || op_code == OP_NOTI)
3123      //if (escape == 0) switch(op_code)
3124    {    {
   case OP_CHAR:  
3125  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3126    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3127  #else  #else
3128    c = *previous;    c = *previous;
3129  #endif  #endif
3130    return c != next;    }
3131    
3132    /* 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
3133    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3134    
3135    case OP_CHARI:  if (escape == 0)
3136  #ifdef SUPPORT_UTF    {
3137    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3138  #else    case, which maps to the special PT_CLIST property. Check this first. */
3139    c = *previous;  
 #endif  
   if (c == next) return FALSE;  
 #ifdef SUPPORT_UTF  
   if (utf)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3140  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3141      othercase = UCD_OTHERCASE((unsigned int)next);    if (utf && (unsigned int)c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3142  #else      {
3143      othercase = NOTACHAR;      int ocs = UCD_CASESET(next);
3144  #endif      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
     return (unsigned int)c != othercase;  
3145      }      }
   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;  
3146  #endif  #endif
   return c == next;  
3147    
3148    case OP_NOTI:    switch(op_code)
3149        {
3150        case OP_CHAR:
3151        return c != next;
3152    
3153        /* For CHARI (caseless character) we must check the other case. If we have
3154        Unicode property support, we can use it to test the other case of
3155        high-valued characters. We know that next can have only one other case,
3156        because multi-other-case characters are dealt with above. */
3157    
3158        case OP_CHARI:
3159        if (c == next) return FALSE;
3160  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3161    GETCHARTEST(c, previous);      if (utf)
3162          {
3163          unsigned int othercase;
3164          if (next < 128) othercase = cd->fcc[next]; else
3165    #ifdef SUPPORT_UCP
3166          othercase = UCD_OTHERCASE((unsigned int)next);
3167  #else  #else
3168    c = *previous;        othercase = NOTACHAR;
3169  #endif  #endif
3170    if (c == next) return TRUE;        return (unsigned int)c != othercase;
3171          }
3172        else
3173    #endif  /* SUPPORT_UTF */
3174        return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3175    
3176        case OP_NOT:
3177        return c == next;
3178    
3179        case OP_NOTI:
3180        if (c == next) return TRUE;
3181  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3182    if (utf)      if (utf)
3183      {        {
3184      unsigned int othercase;        unsigned int othercase;
3185      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3186  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3187      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE((unsigned int)next);
3188  #else  #else
3189      othercase = NOTACHAR;        othercase = NOTACHAR;
3190  #endif  #endif
3191      return (unsigned int)c == othercase;        return (unsigned int)c == othercase;
3192      }        }
3193    else      else
3194  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3195    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 */
3196    
3197    /* 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.
3198    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. */
3199    
3200    case OP_DIGIT:      case OP_DIGIT:
3201    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3202    
3203    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3204    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3205    
3206    case OP_WHITESPACE:      case OP_WHITESPACE:
3207    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3208    
3209    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3210    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3211    
3212    case OP_WORDCHAR:      case OP_WORDCHAR:
3213    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3214    
3215    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3216    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3217    
3218    case OP_HSPACE:      case OP_HSPACE:
3219    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3220    switch(next)      switch(next)
3221      {        {
3222      case 0x09:        HSPACE_CASES:
3223      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;  
     }  
3224    
3225    case OP_ANYNL:        default:
3226    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3227    case OP_NOT_VSPACE:        }
3228    switch(next)  
3229      {      case OP_ANYNL:
3230      case 0x0a:      case OP_VSPACE:
3231      case 0x0b:      case OP_NOT_VSPACE:
3232      case 0x0c:      switch(next)
3233      case 0x0d:        {
3234      case 0x85:        VSPACE_CASES:
3235      case 0x2028:        return op_code == OP_NOT_VSPACE;
3236      case 0x2029:  
3237      return op_code == OP_NOT_VSPACE;        default:
3238      default:        return op_code != OP_NOT_VSPACE;
3239      return op_code != OP_NOT_VSPACE;        }
     }  
3240    
3241  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3242    case OP_PROP:      case OP_PROP:
3243    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3244    
3245    case OP_NOTPROP:      case OP_NOTPROP:
3246    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3247  #endif  #endif
3248    
3249    default:      default:
3250    return FALSE;      return FALSE;
3251        }
3252    }    }
3253    
   
3254  /* 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
3255  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
3256  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 3261  switch(op_code)
3261    {    {
3262    case OP_CHAR:    case OP_CHAR:
3263    case OP_CHARI:    case OP_CHARI:
3264  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3265      {      {
3266      case ESC_d:      case ESC_d:
3267      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3268    
3269      case ESC_D:      case ESC_D:
3270      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3271    
3272      case ESC_s:      case ESC_s:
3273      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3274    
3275      case ESC_S:      case ESC_S:
3276      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3277    
3278      case ESC_w:      case ESC_w:
3279      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3280    
3281      case ESC_W:      case ESC_W:
3282      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3283    
3284      case ESC_h:      case ESC_h:
3285      case ESC_H:      case ESC_H:
3286      switch(c)      switch(c)
3287        {        {
3288        case 0x09:        HSPACE_CASES:
3289        case 0x20:        return escape != ESC_h;
3290        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 -next != ESC_h;  
3291        default:        default:
3292        return -next == ESC_h;        return escape == ESC_h;
3293        }        }
3294    
3295      case ESC_v:      case ESC_v:
3296      case ESC_V:      case ESC_V:
3297      switch(c)      switch(c)
3298        {        {
3299        case 0x0a:        VSPACE_CASES:
3300        case 0x0b:        return escape != ESC_v;
3301        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3302        default:        default:
3303        return -next == ESC_v;        return escape == ESC_v;
3304        }        }
3305    
3306      /* When PCRE_UCP is set, these values get generated for \d etc. Find      /* When PCRE_UCP is set, these values get generated for \d etc. Find
3307      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3308      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3309    
3310  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3311      case ESC_du:      case ESC_du:
# Line 3299  switch(op_code) Line 3316  switch(op_code)
3316      case ESC_SU:      case ESC_SU:
3317        {        {
3318        int temperrorcode = 0;        int temperrorcode = 0;
3319        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3320        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3321        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3322        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3323        }        }
# Line 3327  switch(op_code) Line 3344  switch(op_code)
3344    
3345        /* Do the property check. */        /* Do the property check. */
3346    
3347        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3348        }        }
3349  #endif  #endif
3350    
# Line 3342  switch(op_code) Line 3359  switch(op_code)
3359    these op-codes are never generated.) */    these op-codes are never generated.) */
3360    
3361    case OP_DIGIT:    case OP_DIGIT:
3362    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3363           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3364    
3365    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3366    return next == -ESC_d;    return escape == ESC_d;
3367    
3368    case OP_WHITESPACE:    case OP_WHITESPACE:
3369    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3370    
3371    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3372    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3373    
3374    case OP_HSPACE:    case OP_HSPACE:
3375    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3376           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3377    
3378    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3379    return next == -ESC_h;    return escape == ESC_h;
3380    
3381    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3382    case OP_ANYNL:    case OP_ANYNL:
3383    case OP_VSPACE:    case OP_VSPACE:
3384    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3385    
3386    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3387    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3388    
3389    case OP_WORDCHAR:    case OP_WORDCHAR:
3390    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3391           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3392    
3393    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3394    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3395    
3396    default:    default:
3397    return FALSE;    return FALSE;
# Line 3386  switch(op_code) Line 3403  switch(op_code)
3403    
3404    
3405  /*************************************************  /*************************************************
3406    *        Add a character or range to a class     *
3407    *************************************************/
3408    
3409    /* This function packages up the logic of adding a character or range of
3410    characters to a class. The character values in the arguments will be within the
3411    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3412    mutually recursive with the function immediately below.
3413    
3414    Arguments:
3415      classbits     the bit map for characters < 256
3416      uchardptr     points to the pointer for extra data
3417      options       the options word
3418      cd            contains pointers to tables etc.
3419      start         start of range character
3420      end           end of range character
3421    
3422    Returns:        the number of < 256 characters added
3423                    the pointer to extra data is updated
3424    */
3425    
3426    static int
3427    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3428      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3429    {
3430    pcre_uint32 c;
3431    int n8 = 0;
3432    
3433    /* If caseless matching is required, scan the range and process alternate
3434    cases. In Unicode, there are 8-bit characters that have alternate cases that
3435    are greater than 255 and vice-versa. Sometimes we can just extend the original
3436    range. */
3437    
3438    if ((options & PCRE_CASELESS) != 0)
3439      {
3440    #ifdef SUPPORT_UCP
3441      if ((options & PCRE_UTF8) != 0)
3442        {
3443        int rc;
3444        pcre_uint32 oc, od;
3445    
3446        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3447        c = start;
3448    
3449        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3450          {
3451          /* Handle a single character that has more than one other case. */
3452    
3453          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3454            PRIV(ucd_caseless_sets) + rc, oc);
3455    
3456          /* Do nothing if the other case range is within the original range. */
3457    
3458          else if (oc >= start && od <= end) continue;
3459    
3460          /* Extend the original range if there is overlap, noting that if oc < c, we
3461          can't have od > end because a subrange is always shorter than the basic
3462          range. Otherwise, use a recursive call to add the additional range. */
3463    
3464          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3465          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3466          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3467          }
3468        }
3469      else
3470    #endif  /* SUPPORT_UCP */
3471    
3472      /* Not UTF-mode, or no UCP */
3473    
3474      for (c = start; c <= end && c < 256; c++)
3475        {
3476        SETBIT(classbits, cd->fcc[c]);
3477        n8++;
3478        }
3479      }
3480    
3481    /* Now handle the original range. Adjust the final value according to the bit
3482    length - this means that the same lists of (e.g.) horizontal spaces can be used
3483    in all cases. */
3484    
3485    #if defined COMPILE_PCRE8
3486    #ifdef SUPPORT_UTF
3487      if ((options & PCRE_UTF8) == 0)
3488    #endif
3489      if (end > 0xff) end = 0xff;
3490    
3491    #elif defined COMPILE_PCRE16
3492    #ifdef SUPPORT_UTF
3493      if ((options & PCRE_UTF16) == 0)
3494    #endif
3495      if (end > 0xffff) end = 0xffff;
3496    
3497    #endif /* COMPILE_PCRE[8|16] */
3498    
3499    /* If all characters are less than 256, use the bit map. Otherwise use extra
3500    data. */
3501    
3502    if (end < 0x100)
3503      {
3504      for (c = start; c <= end; c++)
3505        {
3506        n8++;
3507        SETBIT(classbits, c);
3508        }
3509      }
3510    
3511    else
3512      {
3513      pcre_uchar *uchardata = *uchardptr;
3514    
3515    #ifdef SUPPORT_UTF
3516      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3517        {
3518        if (start < end)
3519          {
3520          *uchardata++ = XCL_RANGE;
3521          uchardata += PRIV(ord2utf)(start, uchardata);
3522          uchardata += PRIV(ord2utf)(end, uchardata);
3523          }
3524        else if (start == end)
3525          {
3526          *uchardata++ = XCL_SINGLE;
3527          uchardata += PRIV(ord2utf)(start, uchardata);
3528          }
3529        }
3530      else
3531    #endif  /* SUPPORT_UTF */
3532    
3533      /* Without UTF support, character values are constrained by the bit length,
3534      and can only be > 256 for 16-bit and 32-bit libraries. */
3535    
3536    #ifdef COMPILE_PCRE8
3537        {}
3538    #else
3539      if (start < end)
3540        {
3541        *uchardata++ = XCL_RANGE;
3542        *uchardata++ = start;
3543        *uchardata++ = end;
3544        }
3545      else if (start == end)
3546        {
3547        *uchardata++ = XCL_SINGLE;
3548        *uchardata++ = start;
3549        }
3550    #endif
3551    
3552      *uchardptr = uchardata;   /* Updata extra data pointer */
3553      }
3554    
3555    return n8;    /* Number of 8-bit characters */
3556    }
3557    
3558    
3559    
3560    
3561    /*************************************************
3562    *        Add a list of characters to a class     *
3563    *************************************************/
3564    
3565    /* This function is used for adding a list of case-equivalent characters to a
3566    class, and also for adding a list of horizontal or vertical whitespace. If the
3567    list is in order (which it should be), ranges of characters are detected and
3568    handled appropriately. This function is mutually recursive with the function
3569    above.
3570    
3571    Arguments:
3572      classbits     the bit map for characters < 256
3573      uchardptr     points to the pointer for extra data
3574      options       the options word
3575      cd            contains pointers to tables etc.
3576      p             points to row of 32-bit values, terminated by NOTACHAR
3577      except        character to omit; this is used when adding lists of
3578                      case-equivalent characters to avoid including the one we
3579                      already know about
3580    
3581    Returns:        the number of < 256 characters added
3582                    the pointer to extra data is updated
3583    */
3584    
3585    static int
3586    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3587      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3588    {
3589    int n8 = 0;
3590    while (p[0] < NOTACHAR)
3591      {
3592      int n = 0;
3593      if (p[0] != except)
3594        {
3595        while(p[n+1] == p[0] + n + 1) n++;
3596        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3597        }
3598      p += n + 1;
3599      }
3600    return n8;
3601    }
3602    
3603    
3604    
3605    /*************************************************
3606    *    Add characters not in a list to a class     *
3607    *************************************************/
3608    
3609    /* This function is used for adding the complement of a list of horizontal or
3610    vertical whitespace to a class. The list must be in order.
3611    
3612    Arguments:
3613      classbits     the bit map for characters < 256
3614      uchardptr     points to the pointer for extra data
3615      options       the options word
3616      cd            contains pointers to tables etc.
3617      p             points to row of 32-bit values, terminated by NOTACHAR
3618    
3619    Returns:        the number of < 256 characters added
3620                    the pointer to extra data is updated
3621    */
3622    
3623    static int
3624    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3625      int options, compile_data *cd, const pcre_uint32 *p)
3626    {
3627    BOOL utf = (options & PCRE_UTF8) != 0;
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) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3636      p++;
3637      }
3638    return n8;
3639    }
3640    
3641    
3642    
3643    /*************************************************
3644  *           Compile one branch                   *  *           Compile one branch                   *
3645  *************************************************/  *************************************************/
3646    
# Line 3429  int options = *optionsptr; Line 3684  int options = *optionsptr;
3684  int after_manual_callout = 0;  int after_manual_callout = 0;
3685  int length_prevgroup = 0;  int length_prevgroup = 0;
3686  register int c;  register int c;
3687    int escape;
3688  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3689  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3690  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3448  must not do this for other options (e.g. Line 3704  must not do this for other options (e.g.
3704  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3705    
3706  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3707  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3708  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3709  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3710  #else  #else
3711  BOOL utf = FALSE;  BOOL utf = FALSE;
3712  #endif  #endif
3713    
3714  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3715    class_uchardata always so that it can be passed to add_to_class() always,
3716    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3717    alternative calls for the different cases. */
3718    
3719    pcre_uchar *class_uchardata;
3720  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3721  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3722  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3723  #endif  #endif
3724    
# Line 3503  for (;; ptr++) Line 3762  for (;; ptr++)
3762    BOOL is_recurse;    BOOL is_recurse;
3763    BOOL reset_bracount;    BOOL reset_bracount;
3764    int class_has_8bitchar;    int class_has_8bitchar;
3765    int class_single_char;    int class_one_char;
3766    int newoptions;    int newoptions;
3767    int recno;    int recno;
3768    int refsign;    int refsign;
# Line 3513  for (;; ptr++) Line 3772  for (;; ptr++)
3772    int terminator;    int terminator;
3773    int mclength;    int mclength;
3774    int tempbracount;    int tempbracount;
3775      int ec; // FIXMEchpe pcre_uint32
3776    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3777    
3778    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3801  for (;; ptr++) Line 4061  for (;; ptr++)
4061    
4062      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4063    
4064      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4065      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 <
4066      < 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
4067      a single character. */      character. */
4068    
4069      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4070      class_single_char = 0;      class_one_char = 0;
4071    
4072      /* 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
4073      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
4074      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
4075      */      map. */
4076    
4077      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4078    
4079  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4080      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4081      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4082      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4083  #endif  #endif
4084    
4085      /* 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 4090  for (;; ptr++)
4090        {        {
4091        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4092    
4093  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4094        if (utf && HAS_EXTRALEN(c))        if (utf && HAS_EXTRALEN(c))
4095          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4096          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
# Line 3841  for (;; ptr++) Line 4101  for (;; ptr++)
4101        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4102        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
4103        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4104        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4105          however. */
4106    
4107        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4108          {          {
4109            xclass = TRUE;
4110          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4111          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4112          }          }
# Line 3903  for (;; ptr++) Line 4165  for (;; ptr++)
4165          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4166          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4167    
4168          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4169            posix_class = 0;            posix_class = 0;
4170    
4171          /* 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 4208  for (;; ptr++)
4208              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4209            }            }
4210    
4211          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4212          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4213    
4214          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3962  for (;; ptr++) Line 4224  for (;; ptr++)
4224            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4225    
4226          ptr = tempptr + 1;          ptr = tempptr + 1;
4227          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4228          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4229          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4230          class_single_char = 2;          class_one_char = 2;
4231          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4232          }          }
4233    
# Line 3973  for (;; ptr++) Line 4235  for (;; ptr++)
4235        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
4236        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
4237        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4238        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4239        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4240        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4241        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4242    
4243        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4244          {          {
4245          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4246    
4247          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4248    
4249          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4250          else if (-c == ESC_N)            /* \N is not supported in a class */            c = ec;
4251            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4252            else if (escape == ESC_N)            /* \N is not supported in a class */
4253            {            {
4254            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4255            goto FAILED;            goto FAILED;
4256            }            }
4257          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4258            {            {
4259            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4260              {              {
# Line 3998  for (;; ptr++) Line 4263  for (;; ptr++)
4263            else inescq = TRUE;            else inescq = TRUE;
4264            continue;            continue;
4265            }            }
4266          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4267    
4268          if (c < 0)          else
4269            {            {
4270            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4271            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4272            class_has_8bitchar++;            class_has_8bitchar++;
4273            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4274            class_single_char += 2;            class_one_char += 2;
4275    
4276            switch (-c)            switch (escape)
4277              {              {
4278  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4279              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 4018  for (;; ptr++) Line 4283  for (;; ptr++)
4283              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4284              case ESC_SU:              case ESC_SU:
4285              nestptr = ptr;              nestptr = ptr;
4286              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4287              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4288              continue;              continue;
4289  #endif  #endif
# Line 4042  for (;; ptr++) Line 4307  for (;; ptr++)
4307    
4308              /* 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
4309              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4310              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4311                EBCDIC, so we lazily just adjust the appropriate bit. */
4312    
4313              case ESC_s:              case ESC_s:
4314              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4055  for (;; ptr++) Line 4321  for (;; ptr++)
4321              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4322              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4323              continue;              continue;
4324    
4325                /* The rest apply in both UCP and non-UCP cases. */
4326    
4327              case ESC_h:              case ESC_h:
4328              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4329              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  
4330              continue;              continue;
4331    
4332              case ESC_H:              case ESC_H:
4333              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4334                {                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  
4335              continue;              continue;
4336    
4337              case ESC_v:              case ESC_v:
4338              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4339              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  
4340              continue;              continue;
4341    
4342              case ESC_V:              case ESC_V:
4343              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4344                {                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  
4345              continue;              continue;
4346    
4347  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4240  for (;; ptr++) Line 4352  for (;; ptr++)
4352                int pdata;                int pdata;
4353                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4354                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4355                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4356                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4357                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4358                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4260  for (;; ptr++) Line 4371  for (;; ptr++)
4371                goto FAILED;                goto FAILED;
4372                }                }
4373              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4374              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4375              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4376              break;              break;
4377              }              }
4378            }            }
4379    
4380          /* 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).
4381          greater than 256. */          This may be greater than 256. */
4382    
4383            escape = 0;
4384    
4385          }   /* End of backslash handling */          }   /* End of backslash handling */
4386    
4387        /* 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
4388        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
4389        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4390        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4391    
4392        CHECK_RANGE:        CHECK_RANGE:
4393        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4282  for (;; ptr++) Line 4395  for (;; ptr++)
4395          inescq = FALSE;          inescq = FALSE;
4396          ptr += 2;          ptr += 2;
4397          }          }
   
4398        oldptr = ptr;        oldptr = ptr;
4399    
4400        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4401    
4402        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4403    
# Line 4308  for (;; ptr++) Line 4420  for (;; ptr++)
4420            inescq = TRUE;            inescq = TRUE;
4421            break;            break;
4422            }            }
4423    
4424            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4425            back the pointer and jump to handle the character that preceded it. */
4426    
4427          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4428            {            {
4429            ptr = oldptr;            ptr = oldptr;
4430            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4431            }            }
4432    
4433            /* Otherwise, we have a potential range; pick up the next character */
4434    
4435  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4436          if (utf)          if (utf)
# Line 4330  for (;; ptr++) Line 4447  for (;; ptr++)
4447    
4448          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4449            {            {
4450            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4451              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4452            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4453    
4454            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4455    
4456            if (d < 0)            if (descape > 0)
4457              {              {
4458              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4459                {                {
4460                ptr = oldptr;                ptr = oldptr;
4461                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4462                }                }
4463              }              }
4464            }            }
4465    
4466          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4467          one-character ranges */          one-character ranges. */
4468    
4469          if (d < c)          if (d < c)
4470            {            {
4471            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4472            goto FAILED;            goto FAILED;
4473            }            }
4474            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4475    
4476          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4477            cannot be done anymore. Any value greater than 1 indicates that there
4478            is more than one character. */
4479    
4480            class_one_char = 2;
4481    
4482          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4483    
4484          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4485    
4486          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4487          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4488          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));  
             }  
           }  
   
4489          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4490          }          }
4491    
4492        /* 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
4493        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
4494        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4495          increase it if it is already 2 or more ... just in case there's a class
4496        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4497    
4498        /* Only the value of 1 matters for class_single_char. */        CLASS_SINGLE_CHARACTER:
4499        if (class_single_char < 2) class_single_char++;        if (class_one_char < 2) class_one_char++;
4500    
4501        /* If class_charcount is 1, we saw precisely one character. As long as        /* If class_one_char is 1, we have the first single character in the
4502        there was no use of \p or \P, in other words, no use of any XCLASS features,        class, and there have been no prior ranges, or XCLASS items generated by
4503        we can optimize.        escapes. If this is the final character in the class, we can optimize by
4504          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4505        The optimization throws away the bit map. We turn the item into a        OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4506        1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.        to be set. Otherwise, there can be no first char if this item is first,
4507        In the positive case, it can cause firstchar to be set. Otherwise, there        whatever repeat count may follow. In the case of reqchar, save the
4508        can be no first char if this item is first, whatever repeat count may        previous value for reinstating. */
       follow. In the case of reqchar, save the previous value for reinstating. */  
4509    
4510        if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
4511          {          {
4512          ptr++;          ptr++;
4513          zeroreqchar = reqchar;          zeroreqchar = reqchar;
4514    
         /* The OP_NOT[I] opcodes work on single characters only. */  
   
4515          if (negate_class)          if (negate_class)
4516            {            {
4517    #ifdef SUPPORT_UCP
4518                // FIXMEchpe pcreuint32?
4519              int d;
4520    #endif
4521            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4522            zerofirstchar = firstchar;            zerofirstchar = firstchar;
4523            *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
4524  #ifdef SUPPORT_UTF            /* For caseless UTF-8 mode when UCP support is available, check
4525            if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)            whether this character has more than one other case. If so, generate
4526              code += PRIV(ord2utf)(c, code);            a special OP_NOTPROP item instead of OP_NOTI. */
4527            else  
4528    #ifdef SUPPORT_UCP
4529              if (utf && (options & PCRE_CASELESS) != 0 &&
4530                  (d = UCD_CASESET(c)) != 0)
4531                {
4532                *code++ = OP_NOTPROP;
4533                *code++ = PT_CLIST;
4534                *code++ = d;
4535                }
4536              else
4537  #endif  #endif
4538              *code++ = c;            /* Char has only one other case, or UCP not available */
4539            goto NOT_CHAR;  
4540                {
4541                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4542    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4543                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4544                  code += PRIV(ord2utf)(c, code);
4545                else
4546    #endif
4547                  *code++ = c;
4548                }
4549    
4550              /* We are finished with this character class */
4551    
4552              goto END_CLASS;
4553            }            }
4554    
4555          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4556          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4557    
4558  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4559          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4560            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4561          else          else
# Line 4563  for (;; ptr++) Line 4566  for (;; ptr++)
4566            }            }
4567          goto ONE_CHAR;          goto ONE_CHAR;
4568          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4569    
4570        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4571          has been generated. Add this character to the class. */
4572  #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
4573        if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))        class_has_8bitchar +=
4574  #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));  
           }  
         }  
4575        }        }
4576    
4577      /* 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 4640  for (;; ptr++) Line 4591  for (;; ptr++)
4591        goto FAILED;        goto FAILED;
4592        }        }
4593    
4594        /* We will need an XCLASS if data has been placed in class_uchardata. In
4595        the second phase this is a sufficient test. However, in the pre-compile
4596        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4597        only if the very last character in the class needs XCLASS will it contain
4598        anything at this point. For this reason, xclass gets set TRUE above when
4599        uchar_classdata is emptied, and that's why this code is the way it is here
4600        instead of just doing a test on class_uchardata below. */
4601    
4602    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4603        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4604    #endif
4605    
4606      /* 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
4607      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4608      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
# Line 4702  for (;; ptr++) Line 4665  for (;; ptr++)
4665        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4666        }        }
4667      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4668      NOT_CHAR:  
4669        END_CLASS:
4670      break;      break;
4671    
4672    
# Line 4804  for (;; ptr++) Line 4768  for (;; ptr++)
4768      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4769    
4770      /* If previous was a character or negated character match, abolish the item      /* If previous was a character or negated character match, abolish the item
4771      and generate a repeat item instead. If a char item has a minumum of more      and generate a repeat item instead. If a char item has a minimum of more
4772      than one, ensure  that it is set in reqchar - it might not be if a sequence      than one, ensure that it is set in reqchar - it might not be if a sequence
4773      such as x{3} is  the first thing in a branch because the x will have gone      such as x{3} is the first thing in a branch because the x will have gone
4774      into firstchar instead.  */      into firstchar instead.  */
4775    
4776      if (*previous == OP_CHAR || *previous == OP_CHARI      if (*previous == OP_CHAR || *previous == OP_CHARI
4777          || *previous == OP_NOT || *previous == OP_NOTI)          || *previous == OP_NOT || *previous == OP_NOTI)
4778        {        {
4779        switch (*previous) {        switch (*previous)
4780        default: /* Make compiler happy. */          {
4781        case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;          default: /* Make compiler happy. */
4782        case OP_CHARI: op_type = OP_STARI - OP_STAR; break;          case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
4783        case OP_NOT:   op_type = OP_NOTSTAR - OP_STAR; break;          case OP_CHARI: op_type = OP_STARI - OP_STAR; break;
4784        case OP_NOTI:  op_type = OP_NOTSTARI - OP_STAR; break;          case OP_NOT:   op_type = OP_NOTSTAR - OP_STAR; break;
4785        }          case OP_NOTI:  op_type = OP_NOTSTARI - OP_STAR; break;
4786            }
4787    
4788        /* Deal with UTF characters that take up more than one character. It's        /* Deal with UTF characters that take up more than one character. It's
4789        easier to write this out separately than try to macrify it. Use c to        easier to write this out separately than try to macrify it. Use c to
4790        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
4791        it's a length rather than a small character. */        it's a length rather than a small character. */
4792    
4793  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4794        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4795          {          {
4796          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4961  for (;; ptr++) Line 4926  for (;; ptr++)
4926    
4927          if (repeat_max < 0)          if (repeat_max < 0)
4928            {            {
4929  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4930            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4931              {              {
4932              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4986  for (;; ptr++) Line 4951  for (;; ptr++)
4951    
4952          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4953            {            {
4954  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4955            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4956              {              {
4957              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5016  for (;; ptr++) Line 4981  for (;; ptr++)
4981    
4982        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
4983    
4984  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4985        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
4986          {          {
4987          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5503  for (;; ptr++) Line 5468  for (;; ptr++)
5468        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)
5469          {          {
5470          tempcode += PRIV(OP_lengths)[*tempcode];          tempcode += PRIV(OP_lengths)[*tempcode];
5471  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5472          if (utf && HAS_EXTRALEN(tempcode[-1]))          if (utf && HAS_EXTRALEN(tempcode[-1]))
5473            tempcode += GET_EXTRALEN(tempcode[-1]);            tempcode += GET_EXTRALEN(tempcode[-1]);
5474  #endif  #endif
# Line 5601  for (;; ptr++) Line 5566  for (;; ptr++)
5566          arg = ++ptr;          arg = ++ptr;
5567          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5568          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5569            if ((unsigned int)arglen > MAX_MARK)
5570              {
5571              *errorcodeptr = ERR75;
5572              goto FAILED;
5573              }
5574          }          }
5575    
5576        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5616  for (;; ptr++) Line 5586  for (;; ptr++)
5586          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5587              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5588            {            {
5589              int setverb;
5590    
5591            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5592            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5593    
# Line 5633  for (;; ptr++) Line 5605  for (;; ptr++)
5605                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5606                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5607                }                }
5608              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5609                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5610    
5611              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5612              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
# Line 5648  for (;; ptr++) Line 5621  for (;; ptr++)
5621                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5622                goto FAILED;                goto FAILED;
5623                }                }
5624              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5625              }              }
5626    
5627            else            else
# Line 5659  for (;; ptr++) Line 5631  for (;; ptr++)
5631                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5632                goto FAILED;                goto FAILED;
5633                }                }
5634              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5635              *code++ = arglen;              *code++ = arglen;
5636              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5637              code += arglen;              code += arglen;
5638              *code++ = 0;              *code++ = 0;
5639              }              }
5640    
5641              switch (setverb)
5642                {
5643                case OP_THEN:
5644                case OP_THEN_ARG:
5645                cd->external_flags |= PCRE_HASTHEN;
5646                break;
5647    
5648                case OP_PRUNE:
5649                case OP_PRUNE_ARG:
5650                case OP_SKIP:
5651                case OP_SKIP_ARG:
5652                cd->had_pruneorskip = TRUE;
5653                break;
5654                }
5655    
5656            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5657            }            }
5658    
# Line 6688  for (;; ptr++) Line 6674  for (;; ptr++)
6674    
6675      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6676      tempptr = ptr;      tempptr = ptr;
6677      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6678    
6679      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6680    
6681      if (c < 0)      if (escape == 0)
6682          c = ec;
6683        else
6684        {        {
6685        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6686          {          {
6687          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6688            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6701  for (;; ptr++) Line 6690  for (;; ptr++)
6690          continue;          continue;
6691          }          }
6692    
6693        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6694    
6695        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6696        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6697    
6698        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstchar == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6699          firstchar = REQ_NONE;          firstchar = REQ_NONE;
6700    
6701        /* Set values to reset to if this is followed by a zero repeat. */        /* Set values to reset to if this is followed by a zero repeat. */
# Line 6716  for (;; ptr++) Line 6705  for (;; ptr++)
6705    
6706        /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'        /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'
6707        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6708        -ESC_g is returned only for these cases. So we don't need to check for <        ESC_g is returned only for these cases. So we don't need to check for <
6709        or ' if the value is -ESC_g. For the Perl syntax \g{n} the value is        or ' if the value is ESC_g. For the Perl syntax \g{n} the value is
6710        -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as        ESC_REF+n, and for the Perl syntax \g{name} the result is ESC_k (as
6711        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6712    
6713        if (-c == ESC_g)        if (escape == ESC_g)
6714          {          {
6715          const pcre_uchar *p;          const pcre_uchar *p;
6716          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
# Line 6777  for (;; ptr++) Line 6766  for (;; ptr++)
6766        /* \k<name> or \k'name' is a back reference by name (Perl syntax).        /* \k<name> or \k'name' is a back reference by name (Perl syntax).
6767        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6768    
6769        if (-c == ESC_k)        if (escape == ESC_k)
6770          {          {
6771          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6772            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))
# Line 6796  for (;; ptr++) Line 6785  for (;; ptr++)
6785        not set to cope with cases like (?=(\w+))\1: which would otherwise set        not set to cope with cases like (?=(\w+))\1: which would otherwise set
6786        ':' later. */        ':' later. */
6787    
6788        if (-c >= ESC_REF)        if (escape >= ESC_REF)
6789          {          {
6790          open_capitem *oc;          open_capitem *oc;
6791          recno = -c - ESC_REF;          recno = escape - ESC_REF;
6792    
6793          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6794          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
# Line 6826  for (;; ptr++) Line 6815  for (;; ptr++)
6815        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6816    
6817  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6818        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6819          {          {
6820          BOOL negated;          BOOL negated;
6821          int pdata;          int pdata;
6822          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
6823          if (ptype < 0) goto FAILED;          if (ptype < 0) goto FAILED;
6824          previous = code;          previous = code;
6825          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6826          *code++ = ptype;          *code++ = ptype;
6827          *code++ = pdata;          *code++ = pdata;
6828          }          }
# Line 6842  for (;; ptr++) Line 6831  for (;; ptr++)
6831        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6832        allowed. */        allowed. */
6833    
6834        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6835          {          {
6836          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6837          goto FAILED;          goto FAILED;
# Line 6852  for (;; ptr++) Line 6841  for (;; ptr++)
6841        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6842        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
6843        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
6844        Unicode property tests. */        Unicode property tests. Note that \b and \B do a one-character
6845          lookbehind. */
6846    
6847        else        else
6848          {          {
6849            if ((escape == ESC_b || escape == ESC_B) && cd->max_lookbehind == 0)
6850              cd->max_lookbehind = 1;
6851  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6852          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6853            {            {
6854            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6855            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6856            }            }
6857          else          else
6858  #endif  #endif
# Line 6868  for (;; ptr++) Line 6860  for (;; ptr++)
6860          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6861    
6862            {            {
6863            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
6864            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
6865            }            }
6866          }          }
6867        continue;        continue;
# Line 6879  for (;; ptr++) Line 6871  for (;; ptr++)
6871      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
6872      handle it as a data character. */      handle it as a data character. */
6873    
6874  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6875      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6876        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6877      else      else
# Line 6902  for (;; ptr++) Line 6894  for (;; ptr++)
6894      mclength = 1;      mclength = 1;
6895      mcbuffer[0] = c;      mcbuffer[0] = c;
6896    
6897  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6898      if (utf && HAS_EXTRALEN(c))      if (utf && HAS_EXTRALEN(c))
6899        ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr));        ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr));
6900  #endif  #endif
# Line 6912  for (;; ptr++) Line 6904  for (;; ptr++)
6904    
6905      ONE_CHAR:      ONE_CHAR:
6906      previous = code;      previous = code;
6907    
6908        /* For caseless UTF-8 mode when UCP support is available, check whether
6909        this character has more than one other case. If so, generate a special
6910        OP_PROP item instead of OP_CHARI. */
6911    
6912    #ifdef SUPPORT_UCP
6913        if (utf && (options & PCRE_CASELESS) != 0)
6914          {
6915          GETCHAR(c, mcbuffer);
6916          if ((c = UCD_CASESET(c)) != 0)
6917            {
6918            *code++ = OP_PROP;
6919            *code++ = PT_CLIST;
6920            *code++ = c;
6921            if (firstchar == REQ_UNSET) firstchar = zerofirstchar = REQ_NONE;
6922            break;
6923            }
6924          }
6925    #endif
6926    
6927        /* Caseful matches, or not one of the multicase characters. */
6928    
6929      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6930      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6931    
# Line 7163  for (;;) Line 7177  for (;;)
7177          *ptrptr = ptr;          *ptrptr = ptr;
7178          return FALSE;          return FALSE;
7179          }          }
7180        else { PUT(reverse_count, 0, fixed_length); }        else
7181            {
7182            if (fixed_length > cd->max_lookbehind)
7183              cd->max_lookbehind = fixed_length;
7184            PUT(reverse_count, 0, fixed_length);
7185            }
7186        }        }
7187      }      }
7188    
# Line 7295  and the highest back reference was great Line 7314  and the highest back reference was great
7314  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
7315  of the more common cases more precisely.  of the more common cases more precisely.
7316    
7317    ... A second exception is when the .* appears inside an atomic group, because
7318    this prevents the number of characters it matches from being adjusted.
7319    
7320  Arguments:  Arguments:
7321    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7322    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
7323                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7324                    the less precise approach                    the less precise approach
7325    backref_map    the back reference bitmap    cd             points to the compile data block
7326      atomcount      atomic group level
7327    
7328  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7329  */  */
7330    
7331  static BOOL  static BOOL
7332  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7333    unsigned int backref_map)    compile_data *cd, int atomcount)
7334  {  {
7335  do {  do {
7336     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7319  do { Line 7342  do {
7342     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7343         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7344       {       {
7345       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7346       }       }
7347    
7348     /* Capturing brackets */     /* Capturing brackets */
# Line 7329  do { Line 7352  do {
7352       {       {
7353       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7354       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7355       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7356       }       }
7357    
7358     /* Other brackets */     /* Positive forward assertions and conditions */
7359    
7360     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ASSERT || op == OP_COND)
             op == OP_COND)  
7361       {       {
7362       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7363         }
7364    
7365       /* Atomic groups */
7366    
7367       else if (op == OP_ONCE || op == OP_ONCE_NC)
7368         {
7369         if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7370           return FALSE;
7371       }       }
7372    
7373     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7374     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
7375       group. */
7376    
7377     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7378               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7379       {       {
7380       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7381             atomcount > 0 || cd->had_pruneorskip)
7382         return FALSE;         return FALSE;
7383       }       }
7384    
7385     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7386    
7387     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;
7388    
7389     code += GET(code, 1);     code += GET(code, 1);
7390     }     }
7391  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7370  return TRUE; Line 7403  return TRUE;
7403  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
7404  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
7405  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7406  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7407    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7408    count, because once again the assumption no longer holds.
7409    
7410  Arguments:  Arguments:
7411    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7412    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
7413                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7414                    the less precise approach                    the less precise approach
7415    backref_map    the back reference bitmap    cd             points to the compile data
7416      atomcount      atomic group level
7417    
7418  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7419  */  */
7420    
7421  static BOOL  static BOOL
7422  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7423    unsigned int backref_map)    compile_data *cd, int atomcount)
7424  {  {
7425  do {  do {
7426     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7410  do { Line 7446  do {
7446         return FALSE;         return FALSE;
7447    
7448         default:     /* Assertion */         default:     /* Assertion */
7449         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7450         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7451         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7452         break;         break;
# Line 7424  do { Line 7460  do {
7460     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7461         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7462       {       {
7463       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7464       }       }
7465    
7466     /* Capturing brackets */     /* Capturing brackets */
# Line 7434  do { Line 7470  do {
7470       {       {
7471       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7472       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7473       if (!is_startline(scode, new_map, backref_map)) return FALSE;       if (!is_startline(scode, new_map, cd, atomcount)) return FALSE;
7474       }       }
7475    
7476     /* Other brackets */     /* Positive forward assertions */
7477    
7478     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)     else if (op == OP_ASSERT)
7479       {       {
7480       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7481       }       }
7482    
7483       /* Atomic brackets */
7484    
7485     /* .* means "start at start or after \n" if it isn't in brackets that     else if (op == OP_ONCE || op == OP_ONCE_NC)
7486     may be referenced. */       {
7487         if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE;
7488         }
7489    
7490       /* .* means "start at start or after \n" if it isn't in atomic brackets or
7491       brackets that may be referenced, as long as the pattern does not contain
7492       *PRUNE or *SKIP, because these break the feature. Consider, for example,
7493       /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the
7494       start of a line. */
7495    
7496     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)
7497       {       {
7498       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 ||
7499             atomcount > 0 || cd->had_pruneorskip)
7500           return FALSE;
7501       }       }
7502    
7503     /* Check for explicit circumflex */     /* Check for explicit circumflex; anything else gives a FALSE result. Note
7504       in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC
7505       because the number of characters matched by .* cannot be adjusted inside
7506       them. */
7507    
7508     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;
7509    
# Line 7574  Returns:        pointer to compiled data Line 7625  Returns:        pointer to compiled data
7625                  with errorptr and erroroffset set                  with errorptr and erroroffset set
7626  */  */
7627    
7628  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7629  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7630  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
7631    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7632  #else  #elif defined COMPILE_PCRE16
7633  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7634  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,
7635    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7636    #elif defined COMPILE_PCRE32
7637    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7638    pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr,
7639      int *erroroffset, const unsigned char *tables)
7640  #endif  #endif
7641  {  {
7642  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7643  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7644  #else  #elif defined COMPILE_PCRE16
7645  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7646    #elif defined COMPILE_PCRE32
7647    return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7648  #endif  #endif
7649  }  }
7650    
7651    
7652  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7653  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7654  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
7655    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7656  #else  #elif defined COMPILE_PCRE16
7657  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7658  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,
7659    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7660    #elif defined COMPILE_PCRE32
7661    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7662    pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr,
7663      const char **errorptr, int *erroroffset, const unsigned char *tables)
7664  #endif  #endif
7665  {  {
7666  REAL_PCRE *re;  REAL_PCRE *re;
# Line 7685  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7746  while (ptr[skipatstart] == CHAR_LEFT_PAR
7746    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7747      { skipatstart += 8; options |= PCRE_UTF16; continue; }      { skipatstart += 8; options |= PCRE_UTF16; continue; }
7748  #endif  #endif
7749    #ifdef COMPILE_PCRE32
7750      if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7751        { skipatstart += 8; options |= PCRE_UTF32; continue; }
7752    #endif
7753    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)
7754      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
7755    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 7713  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7778  while (ptr[skipatstart] == CHAR_LEFT_PAR
7778    else break;    else break;
7779    }    }
7780    
7781  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */
7782  utf = (options & PCRE_UTF8) != 0;  utf = (options & PCRE_UTF8) != 0;
7783    
7784  /* 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 7725  not used here. */ Line 7790  not used here. */
7790  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&
7791       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)
7792    {    {
7793  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7794    errorcode = ERR44;    errorcode = ERR44;
7795  #else  #elif defined COMPILE_PCRE16
7796    errorcode = ERR74;    errorcode = ERR74;
7797    #elif defined COMPILE_PCRE32
7798      errorcode = ERR77;
7799  #endif  #endif
7800    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;
7801    }    }
# Line 7833  cd->start_pattern = (const pcre_uchar *) Line 7900  cd->start_pattern = (const pcre_uchar *)
7900  cd->end_pattern = (const pcre_uchar *)(pattern + STRLEN_UC((const pcre_uchar *)pattern));  cd->end_pattern = (const pcre_uchar *)(pattern + STRLEN_UC((const pcre_uchar *)pattern));
7901  cd->req_varyopt = 0;  cd->req_varyopt = 0;
7902  cd->assert_depth = 0;  cd->assert_depth = 0;
7903    cd->max_lookbehind = 0;
7904  cd->external_options = options;  cd->external_options = options;
7905  cd->external_flags = 0;  cd->external_flags = 0;
7906  cd->open_caps = NULL;  cd->open_caps = NULL;
# Line 7883  re->magic_number = MAGIC_NUMBER; Line 7951  re->magic_number = MAGIC_NUMBER;
7951  re->size = (int)size;  re->size = (int)size;
7952  re->options = cd->external_options;  re->options = cd->external_options;
7953  re->flags = cd->external_flags;  re->flags = cd->external_flags;
 re->dummy1 = 0;  
7954  re->first_char = 0;  re->first_char = 0;
7955  re->req_char = 0;  re->req_char = 0;
7956  re->name_table_offset = sizeof(REAL_PCRE) / sizeof(pcre_uchar);  re->name_table_offset = sizeof(REAL_PCRE) / sizeof(pcre_uchar);
# Line 7892  re->name_count = cd->names_found; Line 7959  re->name_count = cd->names_found;
7959  re->ref_count = 0;  re->ref_count = 0;
7960  re->tables = (tables == PRIV(default_tables))? NULL : tables;  re->tables = (tables == PRIV(default_tables))? NULL : tables;
7961  re->nullpad = NULL;  re->nullpad = NULL;
7962    #ifdef COMPILE_PCRE32
7963    re->dummy1 = re->dummy2 = 0;
7964    #endif
7965    
7966  /* 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
7967  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 7903  field; this time it's used for rememberi Line 7973  field; this time it's used for rememberi
7973  cd->final_bracount = cd->bracount;  /* Save for checking forward references */  cd->final_bracount = cd->bracount;  /* Save for checking forward references */
7974  cd->assert_depth = 0;  cd->assert_depth = 0;
7975  cd->bracount = 0;  cd->bracount = 0;
7976    cd->max_lookbehind = 0;
7977  cd->names_found = 0;  cd->names_found = 0;
7978  cd->name_table = (pcre_uchar *)re + re->name_table_offset;  cd->name_table = (pcre_uchar *)re + re->name_table_offset;
7979  codestart = cd->name_table + re->name_entry_size * re->name_count;  codestart = cd->name_table + re->name_entry_size * re->name_count;
# Line 7910  cd->start_code = codestart; Line 7981  cd->start_code = codestart;
7981  cd->hwm = (pcre_uchar *)(cd->start_workspace);  cd->hwm = (pcre_uchar *)(cd->start_workspace);
7982  cd->req_varyopt = 0;  cd->req_varyopt = 0;
7983  cd->had_accept = FALSE;  cd->had_accept = FALSE;
7984    cd->had_pruneorskip = FALSE;
7985  cd->check_lookbehind = FALSE;  cd->check_lookbehind = FALSE;
7986  cd->open_caps = NULL;  cd->open_caps = NULL;
7987    
# Line 7924  code = (pcre_uchar *)codestart; Line 7996  code = (pcre_uchar *)codestart;
7996    &firstchar, &reqchar, NULL, cd, NULL);    &firstchar, &reqchar, NULL, cd, NULL);
7997  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
7998  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
7999    re->max_lookbehind = cd->max_lookbehind;
8000  re->flags = cd->external_flags | PCRE_MODE;  re->flags = cd->external_flags | PCRE_MODE;
8001    
8002  if (cd->had_accept) reqchar = REQ_NONE;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqchar = REQ_NONE;   /* Must disable after (*ACCEPT) */
# Line 8011  if (cd->check_lookbehind) Line 8084  if (cd->check_lookbehind)
8084                      (fixed_length == -4)? ERR70 : ERR25;                      (fixed_length == -4)? ERR70 : ERR25;
8085          break;          break;
8086          }          }
8087          if (fixed_length > cd->max_lookbehind) cd->max_lookbehind = fixed_length;
8088        PUT(cc, 1, fixed_length);        PUT(cc, 1, fixed_length);
8089        }        }
8090      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 8031  if (errorcode != 0) Line 8105  if (errorcode != 0)
8105    }    }
8106    
8107  /* 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
8108  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
8109  as starting with .* when DOTALL is set).  as starting with non-atomic .* when DOTALL is set and there are no occurrences
8110    of *PRUNE or *SKIP.
8111    
8112  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
8113  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
8114  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches
8115  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
8116  */  non-DOTALL matches when *PRUNE and SKIP are not present. */
8117    
8118  if ((re->options & PCRE_ANCHORED) == 0)  if ((re->options & PCRE_ANCHORED) == 0)
8119    {    {
8120    if (is_anchored(codestart, 0, cd->backref_map))    if (is_anchored(codestart, 0, cd, 0)) re->options |= PCRE_ANCHORED;
     re->options |= PCRE_ANCHORED;  
8121    else    else
8122      {      {
8123      if (firstchar < 0)      if (firstchar < 0)
8124        firstchar = find_firstassertedchar(codestart, FALSE);        firstchar = find_firstassertedchar(codestart, FALSE);
8125      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */
8126        {        {
8127  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8128        re->first_char = firstchar & 0xff;        re->first_char = firstchar & 0xff;
8129  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8130        re->first_char = firstchar & 0xffff;        re->first_char = firstchar & 0xffff;
8131  #endif  #elif defined COMPILE_PCRE32
8132          re->first_char = firstchar & ~REQ_MASK;
8133  #endif  #endif
8134        if ((firstchar & REQ_CASELESS) != 0)        if ((firstchar & REQ_CASELESS) != 0)
8135          {          {
# Line 8080  if ((re->options & PCRE_ANCHORED) == 0) Line 8154  if ((re->options & PCRE_ANCHORED) == 0)
8154    
8155        re->flags |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
8156        }        }
8157      else if (is_startline(codestart, 0, cd->backref_map))  
8158        re->flags |= PCRE_STARTLINE;      else if (is_startline(codestart, 0, cd, 0)) re->flags |= PCRE_STARTLINE;
8159      }      }
8160    }    }
8161    
# Line 8092  bytes. */ Line 8166  bytes. */
8166  if (reqchar >= 0 &&  if (reqchar >= 0 &&
8167       ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0))       ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0))
8168    {    {
8169  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8170    re->req_char = reqchar & 0xff;    re->req_char = reqchar & 0xff;
8171  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8172    re->req_char = reqchar & 0xffff;    re->req_char = reqchar & 0xffff;
8173  #endif  #elif defined COMPILE_PCRE32
8174      re->req_char = reqchar & ~REQ_MASK;
8175  #endif  #endif
8176    if ((reqchar & REQ_CASELESS) != 0)    if ((reqchar & REQ_CASELESS) != 0)
8177      {      {
# Line 8149  if ((re->flags & PCRE_REQCHSET) != 0) Line 8223  if ((re->flags & PCRE_REQCHSET) != 0)
8223      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
8224    }    }
8225    
8226  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8227  pcre_printint((pcre *)re, stdout, TRUE);  pcre_printint((pcre *)re, stdout, TRUE);
8228  #else  #elif defined COMPILE_PCRE16
8229  pcre16_printint((pcre *)re, stdout, TRUE);  pcre16_printint((pcre *)re, stdout, TRUE);
8230    #elif defined COMPILE_PCRE32
8231    pcre32_printint((pcre *)re, stdout, TRUE);
8232  #endif  #endif
8233    
8234  /* 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 8168  if (code - codestart > length) Line 8244  if (code - codestart > length)
8244    }    }
8245  #endif   /* PCRE_DEBUG */  #endif   /* PCRE_DEBUG */
8246    
8247  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8248  return (pcre *)re;  return (pcre *)re;
8249  #else  #elif defined COMPILE_PCRE16
8250  return (pcre16 *)re;  return (pcre16 *)re;
8251    #elif defined COMPILE_PCRE32
8252    return (pcre32 *)re;
8253  #endif  #endif
8254  }  }
8255    

Legend:
Removed from v.924  
changed lines
  Added in v.1059

  ViewVC Help
Powered by ViewVC 1.1.5