/[pcre]/code/tags/pcre-8.33/pcre_compile.c
ViewVC logotype

Diff of /code/tags/pcre-8.33/pcre_compile.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 836 by ph10, Wed Dec 28 17:16:11 2011 UTC revision 1152 by chpe, Sun Oct 21 16:53:51 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. */
60    
61  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# 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, int, int,
88        pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *,
89        compile_data *, int *);
90    
91    
92    
93  /*************************************************  /*************************************************
94  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 110  overrun before it actually does run off Line 122  overrun before it actually does run off
122    
123  /* Private flags added to firstchar and reqchar. */  /* Private flags added to firstchar and reqchar. */
124    
125  #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */  #define REQ_CASELESS    (1 << 0)        /* Indicates caselessness */
126  #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */  #define REQ_VARY        (1 << 1)        /* Reqchar followed non-literal item */
127    /* Negative values for the firstchar and reqchar flags */
128    #define REQ_UNSET       (-2)
129    #define REQ_NONE        (-1)
130    
131  /* Repeated character flags. */  /* Repeated character flags. */
132    
# Line 438  static const char error_texts[] = Line 453  static const char error_texts[] =
453    /* 30 */    /* 30 */
454    "unknown POSIX class name\0"    "unknown POSIX class name\0"
455    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
456    "this version of PCRE is not compiled with PCRE_UTF8 support\0"    "this version of PCRE is compiled without UTF support\0"
457    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
458    "character value in \\x{...} sequence is too large\0"    "character value in \\x{...} sequence is too large\0"
459    /* 35 */    /* 35 */
# Line 461  static const char error_texts[] = Line 476  static const char error_texts[] =
476    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
477    /* 50 */    /* 50 */
478    "repeated subpattern is too long\0"    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
479    "octal value is greater than \\377 (not in UTF-8 mode)\0"    "octal value is greater than \\377 in 8-bit non-UTF-8 mode\0"
480    "internal error: overran compiling workspace\0"    "internal error: overran compiling workspace\0"
481    "internal error: previously-checked referenced subpattern not found\0"    "internal error: previously-checked referenced subpattern not found\0"
482    "DEFINE group contains more than one branch\0"    "DEFINE group contains more than one branch\0"
# Line 480  static const char error_texts[] = Line 495  static const char error_texts[] =
495    /* 65 */    /* 65 */
496    "different names for subpatterns of the same number are not allowed\0"    "different names for subpatterns of the same number are not allowed\0"
497    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
498    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with Unicode property support\0"
499    "\\c must be followed by an ASCII character\0"    "\\c must be followed by an ASCII character\0"
500    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"
501    /* 70 */    /* 70 */
502    "internal error: unknown opcode in find_fixedlength()\0"    "internal error: unknown opcode in find_fixedlength()\0"
503    "\\N is not supported in a class\0"    "\\N is not supported in a class\0"
504    "too many forward references\0"    "too many forward references\0"
505    "disallowed UTF-8/16 code point (>= 0xd800 && <= 0xdfff)\0"    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
506      "invalid UTF-16 string\0"
507      /* 75 */
508      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
509      "character value in \\u.... sequence is too large\0"
510      "invalid UTF-32 string\0"
511    ;    ;
512    
513  /* 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 627  static const pcre_uint8 ebcdic_chartab[] Line 647  static const pcre_uint8 ebcdic_chartab[]
647  #endif  #endif
648    
649    
 /* 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 *);  
   
   
650    
651  /*************************************************  /*************************************************
652  *            Find an error text                  *  *            Find an error text                  *
# Line 738  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 751  return (*p == CHAR_RIGHT_CURLY_BRACKET);
751  *************************************************/  *************************************************/
752    
753  /* 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
754  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
755  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
756  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
757  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
758  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
759  sequence.  escape sequence.
760    
761  Arguments:  Arguments:
762    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
763      chptr          points to the data character
764    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
765    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
766    options        the options bits    options        the options bits
767    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
768    
769  Returns:         zero or positive => a data character  Returns:         zero => a data character
770                   negative => a special escape sequence                   positive => a special escape sequence
771                     negative => a back reference
772                   on error, errorcodeptr is set                   on error, errorcodeptr is set
773  */  */
774    
775  static int  static int
776  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
777    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
778  {  {
779  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
780  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
781  const pcre_uchar *ptr = *ptrptr + 1;  const pcre_uchar *ptr = *ptrptr + 1;
782  pcre_int32 c;  pcre_uint32 c;
783    int escape = 0;
784  int i;  int i;
785    
786  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
# Line 781  Otherwise further processing may be requ Line 797  Otherwise further processing may be requ
797  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
798  /* Not alphanumeric */  /* Not alphanumeric */
799  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
800  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
801    
802  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
803  /* Not alphanumeric */  /* Not alphanumeric */
804  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)) {}
805  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
806  #endif  #endif
807    
808  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
# Line 794  else if ((i = escapes[c - 0x48]) != 0) Line 810  else if ((i = escapes[c - 0x48]) != 0)
810  else  else
811    {    {
812    const pcre_uchar *oldptr;    const pcre_uchar *oldptr;
813    BOOL braced, negated;    BOOL braced, negated, overflow;
814      int s;
815    
816    switch (c)    switch (c)
817      {      {
# Line 819  else Line 836  else
836          c = 0;          c = 0;
837          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
838            {            {
839            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
840  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
841            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
842            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 828  else Line 845  else
845            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
846  #endif  #endif
847            }            }
848    
849    #if defined COMPILE_PCRE8
850            if (c > (utf ? 0x10ffff : 0xff))
851    #elif defined COMPILE_PCRE16
852            if (c > (utf ? 0x10ffff : 0xffff))
853    #elif defined COMPILE_PCRE32
854            if (utf && c > 0x10ffff)
855    #endif
856              {
857              *errorcodeptr = ERR76;
858              }
859            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
860          }          }
861        }        }
862      else      else
# Line 854  else Line 883  else
883      (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
884      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
885      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
886      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
887    
888      case CHAR_g:      case CHAR_g:
889      if (isclass) break;      if (isclass) break;
890      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
891        {        {
892        c = -ESC_g;        escape = ESC_g;
893        break;        break;
894        }        }
895    
# Line 873  else Line 902  else
902          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
903        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
904          {          {
905          c = -ESC_k;          escape = ESC_k;
906          break;          break;
907          }          }
908        braced = TRUE;        braced = TRUE;
# Line 889  else Line 918  else
918      else negated = FALSE;      else negated = FALSE;
919    
920      /* The integer range is limited by the machine's int representation. */      /* The integer range is limited by the machine's int representation. */
921      c = 0;      s = 0;
922        overflow = FALSE;
923      while (IS_DIGIT(ptr[1]))      while (IS_DIGIT(ptr[1]))
924        {        {
925        if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */        if (s > INT_MAX / 10 - 1) /* Integer overflow */
926          {          {
927          c = -1;          overflow = TRUE;
928          break;          break;
929          }          }
930        c = c * 10 + *(++ptr) - CHAR_0;        s = s * 10 + (int)(*(++ptr) - CHAR_0);
931        }        }
932      if (((unsigned int)c) > INT_MAX) /* Integer overflow */      if (overflow) /* Integer overflow */
933        {        {
934        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
935          ptr++;          ptr++;
# Line 913  else Line 943  else
943        break;        break;
944        }        }
945    
946      if (c == 0)      if (s == 0)
947        {        {
948        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
949        break;        break;
# Line 921  else Line 951  else
951    
952      if (negated)      if (negated)
953        {        {
954        if (c > bracount)        if (s > bracount)
955          {          {
956          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
957          break;          break;
958          }          }
959        c = bracount - (c - 1);        s = bracount - (s - 1);
960        }        }
961    
962      c = -(ESC_REF + c);      escape = -s;
963      break;      break;
964    
965      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 951  else Line 981  else
981        {        {
982        oldptr = ptr;        oldptr = ptr;
983        /* The integer range is limited by the machine's int representation. */        /* The integer range is limited by the machine's int representation. */
984        c -= CHAR_0;        s = (int)(c -CHAR_0);
985          overflow = FALSE;
986        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
987          {          {
988          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */          if (s > INT_MAX / 10 - 1) /* Integer overflow */
989            {            {
990            c = -1;            overflow = TRUE;
991            break;            break;
992            }            }
993          c = c * 10 + *(++ptr) - CHAR_0;          s = s * 10 + (int)(*(++ptr) - CHAR_0);
994          }          }
995        if (((unsigned int)c) > INT_MAX) /* Integer overflow */        if (overflow) /* Integer overflow */
996          {          {
997          while (IS_DIGIT(ptr[1]))          while (IS_DIGIT(ptr[1]))
998            ptr++;            ptr++;
999          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1000          break;          break;
1001          }          }
1002        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1003          {          {
1004          c = -(ESC_REF + c);          escape = -s;
1005          break;          break;
1006          }          }
1007        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 990  else Line 1021  else
1021      /* \0 always starts an octal number, but we may drop through to here with a      /* \0 always starts an octal number, but we may drop through to here with a
1022      larger first octal digit. The original code used just to take the least      larger first octal digit. The original code used just to take the least
1023      significant 8 bits of octal numbers (I think this is what early Perls used      significant 8 bits of octal numbers (I think this is what early Perls used
1024      to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more      to do). Nowadays we allow for larger numbers in UTF-8 mode and 16-bit mode,
1025      than 3 octal digits. */      but no more than 3 octal digits. */
1026    
1027      case CHAR_0:      case CHAR_0:
1028      c -= CHAR_0;      c -= CHAR_0;
1029      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1030          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1031    #ifdef COMPILE_PCRE8
1032      if (!utf && c > 0xff) *errorcodeptr = ERR51;      if (!utf && c > 0xff) *errorcodeptr = ERR51;
1033    #endif
1034      break;      break;
1035    
1036      /* \x is complicated. \x{ddd} is a character number which can be greater      /* \x is complicated. \x{ddd} is a character number which can be greater
# Line 1015  else Line 1048  else
1048          c = 0;          c = 0;
1049          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1050            {            {
1051            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1052  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1053            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1054            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 1033  else Line 1066  else
1066        const pcre_uchar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1067    
1068        c = 0;        c = 0;
1069          overflow = FALSE;
1070        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1071          {          {
1072          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1073          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1074    
1075    #ifdef COMPILE_PCRE32
1076            if (c >= 0x10000000l) { overflow = TRUE; break; }
1077    #endif
1078    
1079  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1080          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1081          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 1046  else Line 1084  else
1084          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1085  #endif  #endif
1086    
1087  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1088          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1089  #else  #elif defined COMPILE_PCRE16
1090  #ifdef COMPILE_PCRE16          if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1091          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }  #elif defined COMPILE_PCRE32
1092  #endif          if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1093  #endif  #endif
1094          }          }
1095    
1096        if (c < 0)        if (overflow)
1097          {          {
1098          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1099          *errorcodeptr = ERR34;          *errorcodeptr = ERR34;
# Line 1077  else Line 1115  else
1115      c = 0;      c = 0;
1116      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1117        {        {
1118        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1119        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1120  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1121        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 1136  else Line 1174  else
1174  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1175  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1176    
1177  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1178       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1179    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1180    
1181  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1182    
1183  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1184    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1185    
1186  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1187    
1188  *ptrptr = ptr;  *ptrptr = ptr;
1189  return c;  *chptr = c;
1190    return escape;
1191  }  }
1192    
   
   
1193  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1194  /*************************************************  /*************************************************
1195  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1166  escape sequence. Line 1203  escape sequence.
1203  Argument:  Argument:
1204    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1205    negptr         points to a boolean that is set TRUE for negation else FALSE    negptr         points to a boolean that is set TRUE for negation else FALSE
1206    dptr           points to an int that is set to the detailed property value    ptypeptr       points to an unsigned int that is set to the type value
1207      pdataptr       points to an unsigned int that is set to the detailed property value
1208    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1209    
1210  Returns:         type value from ucp_type_table, or -1 for an invalid type  Returns:         TRUE if the type value was found, or FALSE for an invalid type
1211  */  */
1212    
1213  static int  static BOOL
1214  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1215      unsigned int *pdataptr, int *errorcodeptr)
1216  {  {
1217  int c, i, bot, top;  pcre_uchar c;
1218    int i, bot, top;
1219  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1220  pcre_uchar name[32];  pcre_uchar name[32];
1221    
# Line 1222  top = PRIV(utt_size); Line 1262  top = PRIV(utt_size);
1262    
1263  while (bot < top)  while (bot < top)
1264    {    {
1265      int r;
1266    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1267    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1268    if (c == 0)    if (r == 0)
1269      {      {
1270      *dptr = PRIV(utt)[i].value;      *ptypeptr = PRIV(utt)[i].type;
1271      return PRIV(utt)[i].type;      *pdataptr = PRIV(utt)[i].value;
1272        return TRUE;
1273      }      }
1274    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1275    }    }
1276    
1277  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1278  *ptrptr = ptr;  *ptrptr = ptr;
1279  return -1;  return FALSE;
1280    
1281  ERROR_RETURN:  ERROR_RETURN:
1282  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1283  *ptrptr = ptr;  *ptrptr = ptr;
1284  return -1;  return FALSE;
1285  }  }
1286  #endif  #endif
1287    
# Line 1274  int max = -1; Line 1316  int max = -1;
1316  /* Read the minimum value and do a paranoid check: a negative value indicates  /* Read the minimum value and do a paranoid check: a negative value indicates
1317  an integer overflow. */  an integer overflow. */
1318    
1319  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1320  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1321    {    {
1322    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1289  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1331  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1331    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1332      {      {
1333      max = 0;      max = 0;
1334      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1335      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1336        {        {
1337        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1344  Arguments: Line 1386  Arguments:
1386    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1387    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1388    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1389    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1390    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1391    
1392  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 1420  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1462  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1462      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1463          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1464        {        {
1465        int term;        pcre_uchar term;
1466        const pcre_uchar *thisname;        const pcre_uchar *thisname;
1467        *count += 1;        *count += 1;
1468        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
# Line 1428  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1470  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1470        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1471        thisname = ptr;        thisname = ptr;
1472        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1473        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1474            STRNCMP_UC_UC(name, thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1475          return *count;          return *count;
1476        term++;        term++;
1477        }        }
# Line 1578  Arguments: Line 1620  Arguments:
1620    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1621    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1622    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1623    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1624    
1625  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1626  */  */
# Line 1681  and doing the check at the end; a flag s Line 1723  and doing the check at the end; a flag s
1723    
1724  Arguments:  Arguments:
1725    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1726    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1727    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1728    cd       the "compile data" structure    cd       the "compile data" structure
1729    
# Line 1707  for (;;) Line 1749  for (;;)
1749    {    {
1750    int d;    int d;
1751    pcre_uchar *ce, *cs;    pcre_uchar *ce, *cs;
1752    register int op = *cc;    register pcre_uchar op = *cc;
1753    
1754    switch (op)    switch (op)
1755      {      {
1756      /* We only need to continue for OP_CBRA (normal capturing bracket) and      /* We only need to continue for OP_CBRA (normal capturing bracket) and
# Line 1769  for (;;) Line 1811  for (;;)
1811      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1812      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1813      cc += PRIV(OP_lengths)[*cc];      cc += PRIV(OP_lengths)[*cc];
1814      break;      break;
1815    
1816      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1817    
# Line 1827  for (;;) Line 1869  for (;;)
1869      case OP_EXACTI:      case OP_EXACTI:
1870      case OP_NOTEXACT:      case OP_NOTEXACT:
1871      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1872      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1873      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1874  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
1875      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
# Line 1836  for (;;) Line 1878  for (;;)
1878    
1879      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1880      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1881      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)
1882          cc += 2;
1883      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1884      break;      break;
1885    
# Line 1871  for (;;) Line 1914  for (;;)
1914    
1915      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1916    
 #if defined SUPPORT_UTF || defined COMPILE_PCRE16  
     case OP_XCLASS:  
     cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];  
     /* Fall through */  
 #endif  
   
1917      case OP_CLASS:      case OP_CLASS:
1918      case OP_NCLASS:      case OP_NCLASS:
1919    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1920        case OP_XCLASS:
1921        /* The original code caused an unsigned overflow in 64 bit systems,
1922        so now we use a conditional statement. */
1923        if (op == OP_XCLASS)
1924          cc += GET(cc, 1);
1925        else
1926          cc += PRIV(OP_lengths)[OP_CLASS];
1927    #else
1928      cc += PRIV(OP_lengths)[OP_CLASS];      cc += PRIV(OP_lengths)[OP_CLASS];
1929    #endif
1930    
1931      switch (*cc)      switch (*cc)
1932        {        {
# Line 1894  for (;;) Line 1941  for (;;)
1941        case OP_CRRANGE:        case OP_CRRANGE:
1942        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1943        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1944        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1945        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
1946        break;        break;
1947    
# Line 2010  length. Line 2057  length.
2057    
2058  Arguments:  Arguments:
2059    code        points to start of expression    code        points to start of expression
2060    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2061    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2062    
2063  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 2021  PRIV(find_bracket)(const pcre_uchar *cod Line 2068  PRIV(find_bracket)(const pcre_uchar *cod
2068  {  {
2069  for (;;)  for (;;)
2070    {    {
2071    register int c = *code;    register pcre_uchar c = *code;
2072    
2073    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2074    
# Line 2044  for (;;) Line 2091  for (;;)
2091    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2092             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2093      {      {
2094      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2095      if (n == number) return (pcre_uchar *)code;      if (n == number) return (pcre_uchar *)code;
2096      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
2097      }      }
# Line 2074  for (;;) Line 2121  for (;;)
2121        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2122        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2123        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2124        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2125          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2126        break;        break;
2127    
2128        case OP_MARK:        case OP_MARK:
# Line 2097  for (;;) Line 2144  for (;;)
2144    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
2145    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2146    
2147  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2148      if (utf) switch(c)      if (utf) switch(c)
2149        {        {
2150        case OP_CHAR:        case OP_CHAR:
# Line 2149  instance of OP_RECURSE. Line 2196  instance of OP_RECURSE.
2196    
2197  Arguments:  Arguments:
2198    code        points to start of expression    code        points to start of expression
2199    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2200    
2201  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
2202  */  */
# Line 2159  find_recurse(const pcre_uchar *code, BOO Line 2206  find_recurse(const pcre_uchar *code, BOO
2206  {  {
2207  for (;;)  for (;;)
2208    {    {
2209    register int c = *code;    register pcre_uchar c = *code;
2210    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2211    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2212    
# Line 2194  for (;;) Line 2241  for (;;)
2241        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2242        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2243        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2244        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2245          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2246        break;        break;
2247    
2248        case OP_MARK:        case OP_MARK:
# Line 2217  for (;;) Line 2264  for (;;)
2264      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
2265      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2266    
2267  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2268      if (utf) switch(c)      if (utf) switch(c)
2269        {        {
2270        case OP_CHAR:        case OP_CHAR:
2271        case OP_CHARI:        case OP_CHARI:
2272          case OP_NOT:
2273          case OP_NOTI:
2274        case OP_EXACT:        case OP_EXACT:
2275        case OP_EXACTI:        case OP_EXACTI:
2276          case OP_NOTEXACT:
2277          case OP_NOTEXACTI:
2278        case OP_UPTO:        case OP_UPTO:
2279        case OP_UPTOI:        case OP_UPTOI:
2280          case OP_NOTUPTO:
2281          case OP_NOTUPTOI:
2282        case OP_MINUPTO:        case OP_MINUPTO:
2283        case OP_MINUPTOI:        case OP_MINUPTOI:
2284          case OP_NOTMINUPTO:
2285          case OP_NOTMINUPTOI:
2286        case OP_POSUPTO:        case OP_POSUPTO:
2287        case OP_POSUPTOI:        case OP_POSUPTOI:
2288          case OP_NOTPOSUPTO:
2289          case OP_NOTPOSUPTOI:
2290        case OP_STAR:        case OP_STAR:
2291        case OP_STARI:        case OP_STARI:
2292          case OP_NOTSTAR:
2293          case OP_NOTSTARI:
2294        case OP_MINSTAR:        case OP_MINSTAR:
2295        case OP_MINSTARI:        case OP_MINSTARI:
2296          case OP_NOTMINSTAR:
2297          case OP_NOTMINSTARI:
2298        case OP_POSSTAR:        case OP_POSSTAR:
2299        case OP_POSSTARI:        case OP_POSSTARI:
2300          case OP_NOTPOSSTAR:
2301          case OP_NOTPOSSTARI:
2302        case OP_PLUS:        case OP_PLUS:
2303        case OP_PLUSI:        case OP_PLUSI:
2304          case OP_NOTPLUS:
2305          case OP_NOTPLUSI:
2306        case OP_MINPLUS:        case OP_MINPLUS:
2307        case OP_MINPLUSI:        case OP_MINPLUSI:
2308          case OP_NOTMINPLUS:
2309          case OP_NOTMINPLUSI:
2310        case OP_POSPLUS:        case OP_POSPLUS:
2311        case OP_POSPLUSI:        case OP_POSPLUSI:
2312          case OP_NOTPOSPLUS:
2313          case OP_NOTPOSPLUSI:
2314        case OP_QUERY:        case OP_QUERY:
2315        case OP_QUERYI:        case OP_QUERYI:
2316          case OP_NOTQUERY:
2317          case OP_NOTQUERYI:
2318        case OP_MINQUERY:        case OP_MINQUERY:
2319        case OP_MINQUERYI:        case OP_MINQUERYI:
2320          case OP_NOTMINQUERY:
2321          case OP_NOTMINQUERYI:
2322        case OP_POSQUERY:        case OP_POSQUERY:
2323        case OP_POSQUERYI:        case OP_POSQUERYI:
2324          case OP_NOTPOSQUERY:
2325          case OP_NOTPOSQUERYI:
2326        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2327        break;        break;
2328        }        }
# Line 2275  bracket whose current branch will alread Line 2350  bracket whose current branch will alread
2350  Arguments:  Arguments:
2351    code        points to start of search    code        points to start of search
2352    endcode     points to where to stop    endcode     points to where to stop
2353    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2354    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2355    
2356  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2285  static BOOL Line 2360  static BOOL
2360  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2361    BOOL utf, compile_data *cd)    BOOL utf, compile_data *cd)
2362  {  {
2363  register int c;  register pcre_uchar c;
2364  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2365       code < endcode;       code < endcode;
2366       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
# Line 2319  for (code = first_significant_code(code Line 2394  for (code = first_significant_code(code
2394      /* Test for forward reference */      /* Test for forward reference */
2395    
2396      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2397        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2398    
2399      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2400    
# Line 2492  for (code = first_significant_code(code Line 2567  for (code = first_significant_code(code
2567      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2568      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2569      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2570      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2571        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2572      break;      break;
2573    
2574      /* End of branch */      /* End of branch */
# Line 2508  for (code = first_significant_code(code Line 2583  for (code = first_significant_code(code
2583      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2584      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2585    
2586  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2587      case OP_STAR:      case OP_STAR:
2588      case OP_STARI:      case OP_STARI:
2589      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2574  Arguments: Line 2649  Arguments:
2649    code        points to start of the recursion    code        points to start of the recursion
2650    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2651    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2652    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2653    cd          pointers to tables etc    cd          pointers to tables etc
2654    
2655  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2640  Returns:   TRUE or FALSE Line 2715  Returns:   TRUE or FALSE
2715  static BOOL  static BOOL
2716  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2717  {  {
2718  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2719  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2720  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2721    {    {
# Line 2689  register int yield = 0; Line 2764  register int yield = 0;
2764  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2765    {    {
2766    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2767      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2768    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2769    yield++;    yield++;
2770    }    }
# Line 2721  value in the reference (which is a group Line 2796  value in the reference (which is a group
2796  Arguments:  Arguments:
2797    group      points to the start of the group    group      points to the start of the group
2798    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2799    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2800    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2801    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
2802    
# Line 2744  while ((ptr = (pcre_uchar *)find_recurse Line 2819  while ((ptr = (pcre_uchar *)find_recurse
2819    
2820    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2821      {      {
2822      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2823      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2824        {        {
2825        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2757  while ((ptr = (pcre_uchar *)find_recurse Line 2832  while ((ptr = (pcre_uchar *)find_recurse
2832    
2833    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2834      {      {
2835      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2836      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2837      }      }
2838    
# Line 2825  PUT(previous_callout, 2 + LINK_SIZE, len Line 2900  PUT(previous_callout, 2 + LINK_SIZE, len
2900  *************************************************/  *************************************************/
2901    
2902  /* 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
2903  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
2904  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
2905  start address.  start address. A character with multiple other cases is returned on its own
2906    with a special return value.
2907    
2908  Arguments:  Arguments:
2909    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2835  Arguments: Line 2911  Arguments:
2911    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2912    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2913    
2914  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2915                   0 when a range is returned
2916                  >0 the CASESET offset for char with multiple other cases
2917                    in this case, ocptr contains the original
2918  */  */
2919    
2920  static BOOL  static int
2921  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2922    unsigned int *odptr)    pcre_uint32 *odptr)
2923  {  {
2924  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2925    unsigned int co;
2926    
2927    /* Find the first character that has an other case. If it has multiple other
2928    cases, return its case offset value. */
2929    
2930  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2931    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2932      if ((co = UCD_CASESET(c)) != 0)
2933        {
2934        *ocptr = c++;   /* Character that has the set */
2935        *cptr = c;      /* Rest of input range */
2936        return (int)co;
2937        }
2938      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2939      }
2940    
2941  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2942    
2943  *ocptr = othercase;  *ocptr = othercase;
2944  next = othercase + 1;  next = othercase + 1;
# Line 2858  for (++c; c <= d; c++) Line 2949  for (++c; c <= d; c++)
2949    next++;    next++;
2950    }    }
2951    
2952  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2953  *cptr = c;  *cptr = c;             /* Rest of input range */
2954    return 0;
 return TRUE;  
2955  }  }
2956    
2957    
# Line 2883  Returns:       TRUE if auto-possessifyin Line 2973  Returns:       TRUE if auto-possessifyin
2973  */  */
2974    
2975  static BOOL  static BOOL
2976  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata, BOOL negated)
2977  {  {
2978    #ifdef SUPPORT_UCP
2979    const pcre_uint32 *p;
2980    #endif
2981    
2982  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2983    
2984  switch(ptype)  switch(ptype)
2985    {    {
2986    case PT_LAMP:    case PT_LAMP:
# Line 2923  switch(ptype) Line 3018  switch(ptype)
3018    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3019            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3020            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3021    
3022    #ifdef SUPPORT_UCP
3023      case PT_CLIST:
3024      p = PRIV(ucd_caseless_sets) + prop->caseset;
3025      for (;;)
3026        {
3027        if ((unsigned int)c < *p) return !negated;
3028        if ((unsigned int)c == *p++) return negated;
3029        }
3030      break;  /* Control never reaches here */
3031    #endif
3032    }    }
3033    
3034  return FALSE;  return FALSE;
3035  }  }
3036  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2940  sense to automatically possessify the re Line 3047  sense to automatically possessify the re
3047    
3048  Arguments:  Arguments:
3049    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3050    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3051    ptr           next character in pattern    ptr           next character in pattern
3052    options       options bits    options       options bits
3053    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2952  static BOOL Line 3059  static BOOL
3059  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3060    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3061  {  {
3062  pcre_int32 c, next;  pcre_uint32 c = NOTACHAR;
3063  int op_code = *previous++;  pcre_uint32 next;
3064    int escape;
3065    pcre_uchar op_code = *previous++;
3066    
3067  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3068    
# Line 2984  value is a character, a negative value i Line 3093  value is a character, a negative value i
3093  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3094    {    {
3095    int temperrorcode = 0;    int temperrorcode = 0;
3096    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3097    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3098    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3099    }    }
3100  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
3101    {    {
3102      escape = 0;
3103  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3104    if (utf) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3105  #endif  #endif
# Line 3026  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3136  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3136    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)
3137      return FALSE;      return FALSE;
3138    
3139  /* 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. */  
3140    
3141  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3142        op_code == OP_NOT || op_code == OP_NOTI)
3143    {    {
   case OP_CHAR:  
3144  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3145    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3146  #else  #else
3147    c = *previous;    c = *previous;
3148  #endif  #endif
3149    return c != next;    }
3150    
3151    /* 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
3152    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3153    
3154    case OP_CHARI:  if (escape == 0)
3155  #ifdef SUPPORT_UTF    {
3156    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3157  #else    case, which maps to the special PT_CLIST property. Check this first. */
3158    c = *previous;  
3159    #ifdef SUPPORT_UCP
3160      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3161        {
3162        unsigned int ocs = UCD_CASESET(next);
3163        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3164        }
3165  #endif  #endif
3166    if (c == next) return FALSE;  
3167  #ifdef SUPPORT_UTF    switch(op_code)
   if (utf)  
3168      {      {
3169      unsigned int othercase;      case OP_CHAR:
3170      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3171    
3172        /* For CHARI (caseless character) we must check the other case. If we have
3173        Unicode property support, we can use it to test the other case of
3174        high-valued characters. We know that next can have only one other case,
3175        because multi-other-case characters are dealt with above. */
3176    
3177        case OP_CHARI:
3178        if (c == next) return FALSE;
3179    #ifdef SUPPORT_UTF
3180        if (utf)
3181          {
3182          pcre_uint32 othercase;
3183          if (next < 128) othercase = cd->fcc[next]; else
3184  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3185      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3186  #else  #else
3187      othercase = NOTACHAR;        othercase = NOTACHAR;
3188  #endif  #endif
3189      return (unsigned int)c != othercase;        return c != othercase;
3190      }        }
3191    else      else
3192  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3193    return (c != TABLE_GET(next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3194    
3195    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3196    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3197    an XCLASS instead. */  
3198        case OP_NOTI:
3199    case OP_NOT:      if (c == next) return TRUE;
   return (c = *previous) == next;  
   
   case OP_NOTI:  
   if ((c = *previous) == next) return TRUE;  
3200  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3201    if (utf)      if (utf)
3202      {        {
3203      unsigned int othercase;        pcre_uint32 othercase;
3204      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3205  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3206      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3207  #else  #else
3208      othercase = NOTACHAR;        othercase = NOTACHAR;
3209  #endif  #endif
3210      return (unsigned int)c == othercase;        return c == othercase;
3211      }        }
3212    else      else
3213  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3214    return (c == TABLE_GET(next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3215    
3216    /* 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.
3217    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. */
3218    
3219    case OP_DIGIT:      case OP_DIGIT:
3220    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3221    
3222    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3223    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3224    
3225    case OP_WHITESPACE:      case OP_WHITESPACE:
3226    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3227    
3228    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3229    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3230    
3231    case OP_WORDCHAR:      case OP_WORDCHAR:
3232    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3233    
3234    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3235    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3236    
3237    case OP_HSPACE:      case OP_HSPACE:
3238    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3239    switch(next)      switch(next)
3240      {        {
3241      case 0x09:        HSPACE_CASES:
3242      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;  
     }  
3243    
3244    case OP_ANYNL:        default:
3245    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3246    case OP_NOT_VSPACE:        }
3247    switch(next)  
3248      {      case OP_ANYNL:
3249      case 0x0a:      case OP_VSPACE:
3250      case 0x0b:      case OP_NOT_VSPACE:
3251      case 0x0c:      switch(next)
3252      case 0x0d:        {
3253      case 0x85:        VSPACE_CASES:
3254      case 0x2028:        return op_code == OP_NOT_VSPACE;
3255      case 0x2029:  
3256      return op_code == OP_NOT_VSPACE;        default:
3257      default:        return op_code != OP_NOT_VSPACE;
3258      return op_code != OP_NOT_VSPACE;        }
     }  
3259    
3260  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3261    case OP_PROP:      case OP_PROP:
3262    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3263    
3264    case OP_NOTPROP:      case OP_NOTPROP:
3265    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3266  #endif  #endif
3267    
3268    default:      default:
3269    return FALSE;      return FALSE;
3270        }
3271    }    }
3272    
   
3273  /* 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
3274  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
3275  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 3180  switch(op_code) Line 3280  switch(op_code)
3280    {    {
3281    case OP_CHAR:    case OP_CHAR:
3282    case OP_CHARI:    case OP_CHARI:
3283  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3284      {      {
3285      case ESC_d:      case ESC_d:
3286      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3287    
3288      case ESC_D:      case ESC_D:
3289      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3290    
3291      case ESC_s:      case ESC_s:
3292      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3293    
3294      case ESC_S:      case ESC_S:
3295      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3296    
3297      case ESC_w:      case ESC_w:
3298      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3299    
3300      case ESC_W:      case ESC_W:
3301      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3302    
3303      case ESC_h:      case ESC_h:
3304      case ESC_H:      case ESC_H:
3305      switch(c)      switch(c)
3306        {        {
3307        case 0x09:        HSPACE_CASES:
3308        case 0x20:        return escape != ESC_h;
3309        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;  
3310        default:        default:
3311        return -next == ESC_h;        return escape == ESC_h;
3312        }        }
3313    
3314      case ESC_v:      case ESC_v:
3315      case ESC_V:      case ESC_V:
3316      switch(c)      switch(c)
3317        {        {
3318        case 0x0a:        VSPACE_CASES:
3319        case 0x0b:        return escape != ESC_v;
3320        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3321        default:        default:
3322        return -next == ESC_v;        return escape == ESC_v;
3323        }        }
3324    
3325      /* 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
3326      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3327      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3328    
3329  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3330      case ESC_du:      case ESC_du:
# Line 3262  switch(op_code) Line 3335  switch(op_code)
3335      case ESC_SU:      case ESC_SU:
3336        {        {
3337        int temperrorcode = 0;        int temperrorcode = 0;
3338        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3339        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3340        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3341        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3342        }        }
# Line 3272  switch(op_code) Line 3345  switch(op_code)
3345      case ESC_p:      case ESC_p:
3346      case ESC_P:      case ESC_P:
3347        {        {
3348        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3349          int errorcodeptr;
3350        BOOL negated;        BOOL negated;
3351    
3352        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3353        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3354        if (ptype < 0) return FALSE;          return FALSE;
3355        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3356    
3357        /* If the property item is optional, we have to give up. (When generated        /* If the property item is optional, we have to give up. (When generated
# Line 3290  switch(op_code) Line 3364  switch(op_code)
3364    
3365        /* Do the property check. */        /* Do the property check. */
3366    
3367        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3368        }        }
3369  #endif  #endif
3370    
# Line 3305  switch(op_code) Line 3379  switch(op_code)
3379    these op-codes are never generated.) */    these op-codes are never generated.) */
3380    
3381    case OP_DIGIT:    case OP_DIGIT:
3382    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3383           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3384    
3385    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3386    return next == -ESC_d;    return escape == ESC_d;
3387    
3388    case OP_WHITESPACE:    case OP_WHITESPACE:
3389    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3390    
3391    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3392    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3393    
3394    case OP_HSPACE:    case OP_HSPACE:
3395    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3396           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3397    
3398    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3399    return next == -ESC_h;    return escape == ESC_h;
3400    
3401    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3402    case OP_ANYNL:    case OP_ANYNL:
3403    case OP_VSPACE:    case OP_VSPACE:
3404    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3405    
3406    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3407    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3408    
3409    case OP_WORDCHAR:    case OP_WORDCHAR:
3410    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3411           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3412    
3413    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3414    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3415    
3416    default:    default:
3417    return FALSE;    return FALSE;
# Line 3349  switch(op_code) Line 3423  switch(op_code)
3423    
3424    
3425  /*************************************************  /*************************************************
3426    *        Add a character or range to a class     *
3427    *************************************************/
3428    
3429    /* This function packages up the logic of adding a character or range of
3430    characters to a class. The character values in the arguments will be within the
3431    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3432    mutually recursive with the function immediately below.
3433    
3434    Arguments:
3435      classbits     the bit map for characters < 256
3436      uchardptr     points to the pointer for extra data
3437      options       the options word
3438      cd            contains pointers to tables etc.
3439      start         start of range character
3440      end           end of range character
3441    
3442    Returns:        the number of < 256 characters added
3443                    the pointer to extra data is updated
3444    */
3445    
3446    static int
3447    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3448      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3449    {
3450    pcre_uint32 c;
3451    int n8 = 0;
3452    
3453    /* If caseless matching is required, scan the range and process alternate
3454    cases. In Unicode, there are 8-bit characters that have alternate cases that
3455    are greater than 255 and vice-versa. Sometimes we can just extend the original
3456    range. */
3457    
3458    if ((options & PCRE_CASELESS) != 0)
3459      {
3460    #ifdef SUPPORT_UCP
3461      if ((options & PCRE_UTF8) != 0)
3462        {
3463        int rc;
3464        pcre_uint32 oc, od;
3465    
3466        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3467        c = start;
3468    
3469        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3470          {
3471          /* Handle a single character that has more than one other case. */
3472    
3473          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3474            PRIV(ucd_caseless_sets) + rc, oc);
3475    
3476          /* Do nothing if the other case range is within the original range. */
3477    
3478          else if (oc >= start && od <= end) continue;
3479    
3480          /* Extend the original range if there is overlap, noting that if oc < c, we
3481          can't have od > end because a subrange is always shorter than the basic
3482          range. Otherwise, use a recursive call to add the additional range. */
3483    
3484          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3485          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3486          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3487          }
3488        }
3489      else
3490    #endif  /* SUPPORT_UCP */
3491    
3492      /* Not UTF-mode, or no UCP */
3493    
3494      for (c = start; c <= end && c < 256; c++)
3495        {
3496        SETBIT(classbits, cd->fcc[c]);
3497        n8++;
3498        }
3499      }
3500    
3501    /* Now handle the original range. Adjust the final value according to the bit
3502    length - this means that the same lists of (e.g.) horizontal spaces can be used
3503    in all cases. */
3504    
3505    #if defined COMPILE_PCRE8
3506    #ifdef SUPPORT_UTF
3507      if ((options & PCRE_UTF8) == 0)
3508    #endif
3509      if (end > 0xff) end = 0xff;
3510    
3511    #elif defined COMPILE_PCRE16
3512    #ifdef SUPPORT_UTF
3513      if ((options & PCRE_UTF16) == 0)
3514    #endif
3515      if (end > 0xffff) end = 0xffff;
3516    
3517    #endif /* COMPILE_PCRE[8|16] */
3518    
3519    /* If all characters are less than 256, use the bit map. Otherwise use extra
3520    data. */
3521    
3522    if (end < 0x100)
3523      {
3524      for (c = start; c <= end; c++)
3525        {
3526        n8++;
3527        SETBIT(classbits, c);
3528        }
3529      }
3530    
3531    else
3532      {
3533      pcre_uchar *uchardata = *uchardptr;
3534    
3535    #ifdef SUPPORT_UTF
3536      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3537        {
3538        if (start < end)
3539          {
3540          *uchardata++ = XCL_RANGE;
3541          uchardata += PRIV(ord2utf)(start, uchardata);
3542          uchardata += PRIV(ord2utf)(end, uchardata);
3543          }
3544        else if (start == end)
3545          {
3546          *uchardata++ = XCL_SINGLE;
3547          uchardata += PRIV(ord2utf)(start, uchardata);
3548          }
3549        }
3550      else
3551    #endif  /* SUPPORT_UTF */
3552    
3553      /* Without UTF support, character values are constrained by the bit length,
3554      and can only be > 256 for 16-bit and 32-bit libraries. */
3555    
3556    #ifdef COMPILE_PCRE8
3557        {}
3558    #else
3559      if (start < end)
3560        {
3561        *uchardata++ = XCL_RANGE;
3562        *uchardata++ = start;
3563        *uchardata++ = end;
3564        }
3565      else if (start == end)
3566        {
3567        *uchardata++ = XCL_SINGLE;
3568        *uchardata++ = start;
3569        }
3570    #endif
3571    
3572      *uchardptr = uchardata;   /* Updata extra data pointer */
3573      }
3574    
3575    return n8;    /* Number of 8-bit characters */
3576    }
3577    
3578    
3579    
3580    
3581    /*************************************************
3582    *        Add a list of characters to a class     *
3583    *************************************************/
3584    
3585    /* This function is used for adding a list of case-equivalent characters to a
3586    class, and also for adding a list of horizontal or vertical whitespace. If the
3587    list is in order (which it should be), ranges of characters are detected and
3588    handled appropriately. This function is mutually recursive with the function
3589    above.
3590    
3591    Arguments:
3592      classbits     the bit map for characters < 256
3593      uchardptr     points to the pointer for extra data
3594      options       the options word
3595      cd            contains pointers to tables etc.
3596      p             points to row of 32-bit values, terminated by NOTACHAR
3597      except        character to omit; this is used when adding lists of
3598                      case-equivalent characters to avoid including the one we
3599                      already know about
3600    
3601    Returns:        the number of < 256 characters added
3602                    the pointer to extra data is updated
3603    */
3604    
3605    static int
3606    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3607      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3608    {
3609    int n8 = 0;
3610    while (p[0] < NOTACHAR)
3611      {
3612      int n = 0;
3613      if (p[0] != except)
3614        {
3615        while(p[n+1] == p[0] + n + 1) n++;
3616        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3617        }
3618      p += n + 1;
3619      }
3620    return n8;
3621    }
3622    
3623    
3624    
3625    /*************************************************
3626    *    Add characters not in a list to a class     *
3627    *************************************************/
3628    
3629    /* This function is used for adding the complement of a list of horizontal or
3630    vertical whitespace to a class. The list must be in order.
3631    
3632    Arguments:
3633      classbits     the bit map for characters < 256
3634      uchardptr     points to the pointer for extra data
3635      options       the options word
3636      cd            contains pointers to tables etc.
3637      p             points to row of 32-bit values, terminated by NOTACHAR
3638    
3639    Returns:        the number of < 256 characters added
3640                    the pointer to extra data is updated
3641    */
3642    
3643    static int
3644    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3645      int options, compile_data *cd, const pcre_uint32 *p)
3646    {
3647    BOOL utf = (options & PCRE_UTF8) != 0;
3648    int n8 = 0;
3649    if (p[0] > 0)
3650      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3651    while (p[0] < NOTACHAR)
3652      {
3653      while (p[1] == p[0] + 1) p++;
3654      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3655        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3656      p++;
3657      }
3658    return n8;
3659    }
3660    
3661    
3662    
3663    /*************************************************
3664  *           Compile one branch                   *  *           Compile one branch                   *
3665  *************************************************/  *************************************************/
3666    
# Line 3363  Arguments: Line 3675  Arguments:
3675    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3676    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3677    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3678    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3679    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3680      reqcharptr     place to put the last required character
3681      reqcharflagsptr place to put the last required character flags, or a negative number
3682    bcptr          points to current branch chain    bcptr          points to current branch chain
3683    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3684    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3377  Returns:         TRUE on success Line 3691  Returns:         TRUE on success
3691    
3692  static BOOL  static BOOL
3693  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3694    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3695    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3696      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3697      branch_chain *bcptr, int cond_depth,
3698    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3699  {  {
3700  int repeat_type, op_type;  int repeat_type, op_type;
3701  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3702  int bravalue = 0;  int bravalue = 0;
3703  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3704  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
3705  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
3706    pcre_uint32 zeroreqchar, zerofirstchar;
3707    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3708  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3709  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3710  int after_manual_callout = 0;  int after_manual_callout = 0;
3711  int length_prevgroup = 0;  int length_prevgroup = 0;
3712  register int c;  register pcre_uint32 c;
3713    int escape;
3714  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3715  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3716  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3411  must not do this for other options (e.g. Line 3730  must not do this for other options (e.g.
3730  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3731    
3732  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3733  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3734  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3735    #ifndef COMPILE_PCRE32
3736  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3737    #endif
3738  #else  #else
3739  BOOL utf = FALSE;  BOOL utf = FALSE;
3740  #endif  #endif
3741    
3742  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3743    class_uchardata always so that it can be passed to add_to_class() always,
3744    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3745    alternative calls for the different cases. */
3746    
3747    pcre_uchar *class_uchardata;
3748  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3749  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3750  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3751  #endif  #endif
3752    
# Line 3445  to take the zero repeat into account. Th Line 3769  to take the zero repeat into account. Th
3769  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3770  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3771    
3772  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3773    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3774    
3775  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
3776  or zero, according to the current setting of the caseless flag. The  or zero, according to the current setting of the caseless flag. The
# Line 3466  for (;; ptr++) Line 3791  for (;; ptr++)
3791    BOOL is_recurse;    BOOL is_recurse;
3792    BOOL reset_bracount;    BOOL reset_bracount;
3793    int class_has_8bitchar;    int class_has_8bitchar;
3794    int class_single_char;    int class_one_char;
3795    int newoptions;    int newoptions;
3796    int recno;    int recno;
3797    int refsign;    int refsign;
3798    int skipbytes;    int skipbytes;
3799    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
3800    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
3801    int terminator;    int terminator;
3802    int mclength;    unsigned int mclength;
3803    int tempbracount;    unsigned int tempbracount;
3804      pcre_uint32 ec;
3805    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3806    
3807    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3526  for (;; ptr++) Line 3852  for (;; ptr++)
3852      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3853      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3854        (int)(code - last_code), c, c));        (int)(code - last_code), c, c));
3855    
3856      /* If "previous" is set and it is not at the start of the work space, move      /* If "previous" is set and it is not at the start of the work space, move
3857      it back to there, in order to avoid filling up the work space. Otherwise,      it back to there, in order to avoid filling up the work space. Otherwise,
3858      if "previous" is NULL, reset the current code pointer to the start. */      if "previous" is NULL, reset the current code pointer to the start. */
# Line 3638  for (;; ptr++) Line 3964  for (;; ptr++)
3964      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3965      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3966      *firstcharptr = firstchar;      *firstcharptr = firstchar;
3967        *firstcharflagsptr = firstcharflags;
3968      *reqcharptr = reqchar;      *reqcharptr = reqchar;
3969        *reqcharflagsptr = reqcharflags;
3970      *codeptr = code;      *codeptr = code;
3971      *ptrptr = ptr;      *ptrptr = ptr;
3972      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3662  for (;; ptr++) Line 3990  for (;; ptr++)
3990      previous = NULL;      previous = NULL;
3991      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3992        {        {
3993        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3994        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3995        }        }
3996      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3677  for (;; ptr++) Line 4005  for (;; ptr++)
4005      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4006    
4007      case CHAR_DOT:      case CHAR_DOT:
4008      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4009      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4010        zerofirstcharflags = firstcharflags;
4011      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4012        zeroreqcharflags = reqcharflags;
4013      previous = code;      previous = code;
4014      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4015      break;      break;
# Line 3753  for (;; ptr++) Line 4083  for (;; ptr++)
4083          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4084        {        {
4085        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4086        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4087        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4088          zerofirstcharflags = firstcharflags;
4089        break;        break;
4090        }        }
4091    
# Line 3764  for (;; ptr++) Line 4095  for (;; ptr++)
4095    
4096      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4097    
4098      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4099      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 <
4100      < 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
4101      a single character. */      character. */
4102    
4103      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4104      class_single_char = 0;      class_one_char = 0;
4105    
4106      /* 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
4107      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
4108      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
4109      */      map. */
4110    
4111      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4112    
4113  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4114      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4115      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4116      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4117  #endif  #endif
4118    
4119      /* 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 3804  for (;; ptr++) Line 4135  for (;; ptr++)
4135        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4136        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
4137        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4138        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4139          however. */
4140    
4141        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4142          {          {
4143            xclass = TRUE;
4144          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4145          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4146          }          }
# Line 3866  for (;; ptr++) Line 4199  for (;; ptr++)
4199          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4200          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4201    
4202          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4203            posix_class = 0;            posix_class = 0;
4204    
4205          /* 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 3909  for (;; ptr++) Line 4242  for (;; ptr++)
4242              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4243            }            }
4244    
4245          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4246          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4247    
4248          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3925  for (;; ptr++) Line 4258  for (;; ptr++)
4258            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4259    
4260          ptr = tempptr + 1;          ptr = tempptr + 1;
4261          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4262          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4263          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4264          class_single_char = 2;          class_one_char = 2;
4265          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4266          }          }
4267    
# Line 3936  for (;; ptr++) Line 4269  for (;; ptr++)
4269        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
4270        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
4271        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4272        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4273        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4274        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4275        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4276    
4277        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4278          {          {
4279          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4280    
4281          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4282    
4283          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4284          else if (-c == ESC_N)            /* \N is not supported in a class */            c = ec;
4285            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4286            else if (escape == ESC_N)            /* \N is not supported in a class */
4287            {            {
4288            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4289            goto FAILED;            goto FAILED;
4290            }            }
4291          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4292            {            {
4293            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4294              {              {
# Line 3961  for (;; ptr++) Line 4297  for (;; ptr++)
4297            else inescq = TRUE;            else inescq = TRUE;
4298            continue;            continue;
4299            }            }
4300          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4301    
4302          if (c < 0)          else
4303            {            {
4304            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4305            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4306            class_has_8bitchar++;            class_has_8bitchar++;
4307            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4308            class_single_char += 2;            class_one_char += 2;
4309    
4310            switch (-c)            switch (escape)
4311              {              {
4312  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4313              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3981  for (;; ptr++) Line 4317  for (;; ptr++)
4317              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4318              case ESC_SU:              case ESC_SU:
4319              nestptr = ptr;              nestptr = ptr;
4320              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4321              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4322              continue;              continue;
4323  #endif  #endif
# Line 4005  for (;; ptr++) Line 4341  for (;; ptr++)
4341    
4342              /* 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
4343              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4344              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4345                EBCDIC, so we lazily just adjust the appropriate bit. */
4346    
4347              case ESC_s:              case ESC_s:
4348              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4018  for (;; ptr++) Line 4355  for (;; ptr++)
4355              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4356              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4357              continue;              continue;
4358    
4359                /* The rest apply in both UCP and non-UCP cases. */
4360    
4361              case ESC_h:              case ESC_h:
4362              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4363              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  
4364              continue;              continue;
4365    
4366              case ESC_H:              case ESC_H:
4367              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4368                {                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  
4369              continue;              continue;
4370    
4371              case ESC_v:              case ESC_v:
4372              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4373              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  
4374              continue;              continue;
4375    
4376              case ESC_V:              case ESC_V:
4377              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4378                {                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  
4379              continue;              continue;
4380    
4381  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4200  for (;; ptr++) Line 4383  for (;; ptr++)
4383              case ESC_P:              case ESC_P:
4384                {                {
4385                BOOL negated;                BOOL negated;
4386                int pdata;                unsigned int ptype = 0, pdata = 0;
4387                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4388                if (ptype < 0) goto FAILED;                  goto FAILED;
4389                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4390                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4391                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4392                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4223  for (;; ptr++) Line 4405  for (;; ptr++)
4405                goto FAILED;                goto FAILED;
4406                }                }
4407              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4408              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4409              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4410              break;              break;
4411              }              }
4412            }            }
4413    
4414          /* 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).
4415          greater than 256. */          This may be greater than 256. */
4416    
4417            escape = 0;
4418    
4419          }   /* End of backslash handling */          }   /* End of backslash handling */
4420    
4421        /* 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
4422        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
4423        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4424        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4425    
4426        CHECK_RANGE:        CHECK_RANGE:
4427        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4245  for (;; ptr++) Line 4429  for (;; ptr++)
4429          inescq = FALSE;          inescq = FALSE;
4430          ptr += 2;          ptr += 2;
4431          }          }
   
4432        oldptr = ptr;        oldptr = ptr;
4433    
4434        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4435    
4436        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4437    
# Line 4256  for (;; ptr++) Line 4439  for (;; ptr++)
4439    
4440        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4441          {          {
4442          int d;          pcre_uint32 d;
4443          ptr += 2;          ptr += 2;
4444          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4445    
# Line 4271  for (;; ptr++) Line 4454  for (;; ptr++)
4454            inescq = TRUE;            inescq = TRUE;
4455            break;            break;
4456            }            }
4457    
4458            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4459            back the pointer and jump to handle the character that preceded it. */
4460    
4461          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4462            {            {
4463            ptr = oldptr;            ptr = oldptr;
4464            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4465            }            }
4466    
4467            /* Otherwise, we have a potential range; pick up the next character */
4468    
4469  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4470          if (utf)          if (utf)
# Line 4293  for (;; ptr++) Line 4481  for (;; ptr++)
4481    
4482          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4483            {            {
4484            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4485              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4486            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4487    
4488            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4489    
4490            if (d < 0)            if (descape != 0)
4491              {              {
4492              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4493                {                {
4494                ptr = oldptr;                ptr = oldptr;
4495                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4496                }                }
4497              }              }
4498            }            }
4499    
4500          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4501          one-character ranges */          one-character ranges. */
4502    
4503          if (d < c)          if (d < c)
4504            {            {
4505            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4506            goto FAILED;            goto FAILED;
4507            }            }
4508            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4509    
4510          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4511            cannot be done anymore. Any value greater than 1 indicates that there
4512            is more than one character. */
4513    
4514            class_one_char = 2;
4515    
4516          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4517    
4518          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4519    
4520          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4521          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4522          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));  
             }  
           }  
   
4523          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4524          }          }
4525    
4526        /* 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
4527        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
4528        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4529          increase it if it is already 2 or more ... just in case there's a class
4530        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4531    
4532        /* Only the value of 1 matters for class_single_char. */        CLASS_SINGLE_CHARACTER:
4533        if (class_single_char < 2) class_single_char++;        if (class_one_char < 2) class_one_char++;
4534    
4535        /* 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
4536        there were no negated characters >= 128 and there was no use of \p or \P,        class, and there have been no prior ranges, or XCLASS items generated by
4537        in other words, no use of any XCLASS features, we can optimize.        escapes. If this is the final character in the class, we can optimize by
4538          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4539        In UTF-8 mode, we can optimize the negative case only if there were no        OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4540        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR        to be set. Otherwise, there can be no first char if this item is first,
4541        operate on single-bytes characters only. This is an historical hangover.        whatever repeat count may follow. In the case of reqchar, save the
4542        Maybe one day we can tidy these opcodes to handle multi-byte characters.        previous value for reinstating. */
   
       The optimization throws away the bit map. We turn the item into a  
       1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
       Note that OP_NOT[I] does not support multibyte characters. In the positive  
       case, it can cause firstchar to be set. Otherwise, there can be no first  
       char if this item is first, whatever repeat count may follow. In the case  
       of reqchar, save the previous value for reinstating. */  
4543    
4544  #ifdef SUPPORT_UTF        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET  
         && (!utf || !negate_class || c < (MAX_VALUE_FOR_SINGLE_CHAR + 1)))  
 #else  
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)  
 #endif  
4545          {          {
4546          ptr++;          ptr++;
4547          zeroreqchar = reqchar;          zeroreqchar = reqchar;
4548            zeroreqcharflags = reqcharflags;
         /* The OP_NOT[I] opcodes work on single characters only. */  
4549    
4550          if (negate_class)          if (negate_class)
4551            {            {
4552            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;  #ifdef SUPPORT_UCP
4553              int d;
4554    #endif
4555              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4556            zerofirstchar = firstchar;            zerofirstchar = firstchar;
4557            *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;            zerofirstcharflags = firstcharflags;
4558            *code++ = c;  
4559            goto NOT_CHAR;            /* For caseless UTF-8 mode when UCP support is available, check
4560              whether this character has more than one other case. If so, generate
4561              a special OP_NOTPROP item instead of OP_NOTI. */
4562    
4563    #ifdef SUPPORT_UCP
4564              if (utf && (options & PCRE_CASELESS) != 0 &&
4565                  (d = UCD_CASESET(c)) != 0)
4566                {
4567                *code++ = OP_NOTPROP;
4568                *code++ = PT_CLIST;
4569                *code++ = d;
4570                }
4571              else
4572    #endif
4573              /* Char has only one other case, or UCP not available */
4574    
4575                {
4576                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4577    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4578                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4579                  code += PRIV(ord2utf)(c, code);
4580                else
4581    #endif
4582                  *code++ = c;
4583                }
4584    
4585              /* We are finished with this character class */
4586    
4587              goto END_CLASS;
4588            }            }
4589    
4590          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4591          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4592    
4593  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4594          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4595            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4596          else          else
# Line 4532  for (;; ptr++) Line 4601  for (;; ptr++)
4601            }            }
4602          goto ONE_CHAR;          goto ONE_CHAR;
4603          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4604    
4605        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4606          has been generated. Add this character to the class. */
4607  #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
4608        if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))        class_has_8bitchar +=
4609  #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 ((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));  
           }  
         }  
4610        }        }
4611    
4612      /* 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 4609  for (;; ptr++) Line 4626  for (;; ptr++)
4626        goto FAILED;        goto FAILED;
4627        }        }
4628    
4629        /* We will need an XCLASS if data has been placed in class_uchardata. In
4630        the second phase this is a sufficient test. However, in the pre-compile
4631        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4632        only if the very last character in the class needs XCLASS will it contain
4633        anything at this point. For this reason, xclass gets set TRUE above when
4634        uchar_classdata is emptied, and that's why this code is the way it is here
4635        instead of just doing a test on class_uchardata below. */
4636    
4637    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4638        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4639    #endif
4640    
4641      /* 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
4642      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4643      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
4644    
4645      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4646      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4647        zerofirstcharflags = firstcharflags;
4648      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4649        zeroreqcharflags = reqcharflags;
4650    
4651      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4652      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
# Line 4671  for (;; ptr++) Line 4702  for (;; ptr++)
4702        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4703        }        }
4704      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4705      NOT_CHAR:  
4706        END_CLASS:
4707      break;      break;
4708    
4709    
# Line 4709  for (;; ptr++) Line 4741  for (;; ptr++)
4741      if (repeat_min == 0)      if (repeat_min == 0)
4742        {        {
4743        firstchar = zerofirstchar;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4744          firstcharflags = zerofirstcharflags;
4745        reqchar = zeroreqchar;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4746          reqcharflags = zeroreqcharflags;
4747        }        }
4748    
4749      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4772  for (;; ptr++) Line 4806  for (;; ptr++)
4806    
4807      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4808    
4809      /* If previous was a character match, abolish the item and generate a      /* If previous was a character or negated character match, abolish the item
4810      repeat item instead. If a char item has a minumum of more than one, ensure      and generate a repeat item instead. If a char item has a minimum of more
4811      that it is set in reqchar - it might not be if a sequence such as x{3} is      than one, ensure that it is set in reqchar - it might not be if a sequence
4812      the first thing in a branch because the x will have gone into firstchar      such as x{3} is the first thing in a branch because the x will have gone
4813      instead.  */      into firstchar instead.  */
4814    
4815      if (*previous == OP_CHAR || *previous == OP_CHARI)      if (*previous == OP_CHAR || *previous == OP_CHARI
4816        {          || *previous == OP_NOT || *previous == OP_NOTI)
4817        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;        {
4818          switch (*previous)
4819            {
4820            default: /* Make compiler happy. */
4821            case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
4822            case OP_CHARI: op_type = OP_STARI - OP_STAR; break;
4823            case OP_NOT:   op_type = OP_NOTSTAR - OP_STAR; break;
4824            case OP_NOTI:  op_type = OP_NOTSTARI - OP_STAR; break;
4825            }
4826    
4827        /* 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
4828        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
4829        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
4830        it's a length rather than a small character. */        it's a length rather than a small character. */
4831    
4832  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4833        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4834          {          {
4835          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4803  for (;; ptr++) Line 4845  for (;; ptr++)
4845        with UTF disabled, or for a single character UTF character. */        with UTF disabled, or for a single character UTF character. */
4846          {          {
4847          c = code[-1];          c = code[-1];
4848          if (repeat_min > 1) reqchar = c | req_caseopt | cd->req_varyopt;          if (*previous <= OP_CHARI && repeat_min > 1)
4849              {
4850              reqchar = c;
4851              reqcharflags = req_caseopt | cd->req_varyopt;
4852              }
4853          }          }
4854    
4855        /* If the repetition is unlimited, it pays to see if the next thing on        /* If the repetition is unlimited, it pays to see if the next thing on
# Line 4822  for (;; ptr++) Line 4868  for (;; ptr++)
4868        goto OUTPUT_SINGLE_REPEAT;   /* Code shared with single character types */        goto OUTPUT_SINGLE_REPEAT;   /* Code shared with single character types */
4869        }        }
4870    
     /* If previous was a single negated character ([^a] or similar), we use  
     one of the special opcodes, replacing it. The code is shared with single-  
     character repeats by setting opt_type to add a suitable offset into  
     repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI  
     are currently used only for single-byte chars. */  
   
     else if (*previous == OP_NOT || *previous == OP_NOTI)  
       {  
       op_type = ((*previous == OP_NOT)? OP_NOTSTAR : OP_NOTSTARI) - OP_STAR;  
       c = previous[1];  
       if (!possessive_quantifier &&  
           repeat_max < 0 &&  
           check_auto_possessive(previous, utf, ptr + 1, options, cd))  
         {  
         repeat_type = 0;    /* Force greedy */  
         possessive_quantifier = TRUE;  
         }  
       goto OUTPUT_SINGLE_REPEAT;  
       }  
   
4871      /* If previous was a character type match (\d or similar), abolish it and      /* If previous was a character type match (\d or similar), abolish it and
4872      create a suitable repeat item. The code is shared with single-character      create a suitable repeat item. The code is shared with single-character
4873      repeats by setting op_type to add a suitable offset into repeat_type. Note      repeats by setting op_type to add a suitable offset into repeat_type. Note
# Line 4942  for (;; ptr++) Line 4968  for (;; ptr++)
4968    
4969          if (repeat_max < 0)          if (repeat_max < 0)
4970            {            {
4971  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4972            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4973              {              {
4974              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4967  for (;; ptr++) Line 4993  for (;; ptr++)
4993    
4994          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4995            {            {
4996  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4997            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4998              {              {
4999              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4997  for (;; ptr++) Line 5023  for (;; ptr++)
5023    
5024        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
5025    
5026  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5027        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
5028          {          {
5029          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5205  for (;; ptr++) Line 5231  for (;; ptr++)
5231    
5232            else            else
5233              {              {
5234              if (groupsetfirstchar && reqchar < 0) reqchar = firstchar;              if (groupsetfirstchar && reqcharflags < 0)
5235                  {
5236                  reqchar = firstchar;
5237                  reqcharflags = firstcharflags;
5238                  }
5239    
5240              for (i = 1; i < repeat_min; i++)              for (i = 1; i < repeat_min; i++)
5241                {                {
# Line 5582  for (;; ptr++) Line 5612  for (;; ptr++)
5612          arg = ++ptr;          arg = ++ptr;
5613          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5614          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5615            if ((unsigned int)arglen > MAX_MARK)
5616              {
5617              *errorcodeptr = ERR75;
5618              goto FAILED;
5619              }
5620          }          }
5621    
5622        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5597  for (;; ptr++) Line 5632  for (;; ptr++)
5632          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5633              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5634            {            {
5635              int setverb;
5636    
5637            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5638            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5639    
# Line 5614  for (;; ptr++) Line 5651  for (;; ptr++)
5651                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5652                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5653                }                }
5654              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5655                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5656    
5657              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5658              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
5659              }              }
5660    
5661            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5629  for (;; ptr++) Line 5667  for (;; ptr++)
5667                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5668                goto FAILED;                goto FAILED;
5669                }                }
5670              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5671              }              }
5672    
5673            else            else
# Line 5640  for (;; ptr++) Line 5677  for (;; ptr++)
5677                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5678                goto FAILED;                goto FAILED;
5679                }                }
5680              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5681              *code++ = arglen;              *code++ = arglen;
5682              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5683              code += arglen;              code += arglen;
5684              *code++ = 0;              *code++ = 0;
5685              }              }
5686    
5687              switch (setverb)
5688                {
5689                case OP_THEN:
5690                case OP_THEN_ARG:
5691                cd->external_flags |= PCRE_HASTHEN;
5692                break;
5693    
5694                case OP_PRUNE:
5695                case OP_PRUNE_ARG:
5696                case OP_SKIP:
5697                case OP_SKIP_ARG:
5698                cd->had_pruneorskip = TRUE;
5699                break;
5700                }
5701    
5702            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5703            }            }
5704    
# Line 5770  for (;; ptr++) Line 5821  for (;; ptr++)
5821          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
5822            {            {
5823            if (recno >= 0)            if (recno >= 0)
5824              recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1;              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;
5825            ptr++;            ptr++;
5826            }            }
5827          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
5828    
5829          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||
5830              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
5831            {            {
5832            ptr--;      /* Error offset */            ptr--;      /* Error offset */
# Line 6015  for (;; ptr++) Line 6066  for (;; ptr++)
6066    
6067            if (lengthptr != NULL)            if (lengthptr != NULL)
6068              {              {
6069              if (*ptr != terminator)              if (*ptr != (pcre_uchar)terminator)
6070                {                {
6071                *errorcodeptr = ERR42;                *errorcodeptr = ERR42;
6072                goto FAILED;                goto FAILED;
# Line 6157  for (;; ptr++) Line 6208  for (;; ptr++)
6208              *errorcodeptr = ERR62;              *errorcodeptr = ERR62;
6209              goto FAILED;              goto FAILED;
6210              }              }
6211            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6212              {              {
6213              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
6214              goto FAILED;              goto FAILED;
# Line 6263  for (;; ptr++) Line 6314  for (;; ptr++)
6314            while(IS_DIGIT(*ptr))            while(IS_DIGIT(*ptr))
6315              recno = recno * 10 + *ptr++ - CHAR_0;              recno = recno * 10 + *ptr++ - CHAR_0;
6316    
6317            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6318              {              {
6319              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
6320              goto FAILED;              goto FAILED;
# Line 6367  for (;; ptr++) Line 6418  for (;; ptr++)
6418    
6419          /* Can't determine a first byte now */          /* Can't determine a first byte now */
6420    
6421          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6422          continue;          continue;
6423    
6424    
# Line 6501  for (;; ptr++) Line 6552  for (;; ptr++)
6552           cond_depth +           cond_depth +
6553             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
6554           &subfirstchar,                   /* For possible first char */           &subfirstchar,                   /* For possible first char */
6555             &subfirstcharflags,
6556           &subreqchar,                     /* For possible last char */           &subreqchar,                     /* For possible last char */
6557             &subreqcharflags,
6558           bcptr,                           /* Current branch chain */           bcptr,                           /* Current branch chain */
6559           cd,                              /* Tables block */           cd,                              /* Tables block */
6560           (lengthptr == NULL)? NULL :      /* Actual compile phase */           (lengthptr == NULL)? NULL :      /* Actual compile phase */
# Line 6562  for (;; ptr++) Line 6615  for (;; ptr++)
6615            *errorcodeptr = ERR27;            *errorcodeptr = ERR27;
6616            goto FAILED;            goto FAILED;
6617            }            }
6618          if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;          if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE;
6619          }          }
6620        }        }
6621    
# Line 6611  for (;; ptr++) Line 6664  for (;; ptr++)
6664      back off. */      back off. */
6665    
6666      zeroreqchar = reqchar;      zeroreqchar = reqchar;
6667        zeroreqcharflags = reqcharflags;
6668      zerofirstchar = firstchar;      zerofirstchar = firstchar;
6669        zerofirstcharflags = firstcharflags;
6670      groupsetfirstchar = FALSE;      groupsetfirstchar = FALSE;
6671    
6672      if (bravalue >= OP_ONCE)      if (bravalue >= OP_ONCE)
# Line 6622  for (;; ptr++) Line 6677  for (;; ptr++)
6677        no firstchar, set "none" for the whole branch. In both cases, a zero        no firstchar, set "none" for the whole branch. In both cases, a zero
6678        repeat forces firstchar to "none". */        repeat forces firstchar to "none". */
6679    
6680        if (firstchar == REQ_UNSET)        if (firstcharflags == REQ_UNSET)
6681          {          {
6682          if (subfirstchar >= 0)          if (subfirstcharflags >= 0)
6683            {            {
6684            firstchar = subfirstchar;            firstchar = subfirstchar;
6685              firstcharflags = subfirstcharflags;
6686            groupsetfirstchar = TRUE;            groupsetfirstchar = TRUE;
6687            }            }
6688          else firstchar = REQ_NONE;          else firstcharflags = REQ_NONE;
6689          zerofirstchar = REQ_NONE;          zerofirstcharflags = REQ_NONE;
6690          }          }
6691    
6692        /* If firstchar was previously set, convert the subpattern's firstchar        /* If firstchar was previously set, convert the subpattern's firstchar
6693        into reqchar if there wasn't one, using the vary flag that was in        into reqchar if there wasn't one, using the vary flag that was in
6694        existence beforehand. */        existence beforehand. */
6695    
6696        else if (subfirstchar >= 0 && subreqchar < 0)        else if (subfirstcharflags >= 0 && subreqcharflags < 0)
6697          subreqchar = subfirstchar | tempreqvary;          {
6698            subreqchar = subfirstchar;
6699            subreqcharflags = subfirstcharflags | tempreqvary;
6700            }
6701    
6702        /* If the subpattern set a required byte (or set a first byte that isn't        /* If the subpattern set a required byte (or set a first byte that isn't
6703        really the first byte - see above), set it. */        really the first byte - see above), set it. */
6704    
6705        if (subreqchar >= 0) reqchar = subreqchar;        if (subreqcharflags >= 0)
6706            {
6707            reqchar = subreqchar;
6708            reqcharflags = subreqcharflags;
6709            }
6710        }        }
6711    
6712      /* For a forward assertion, we take the reqchar, if set. This can be      /* For a forward assertion, we take the reqchar, if set. This can be
# Line 6654  for (;; ptr++) Line 6717  for (;; ptr++)
6717      of a firstchar. This is overcome by a scan at the end if there's no      of a firstchar. This is overcome by a scan at the end if there's no
6718      firstchar, looking for an asserted first char. */      firstchar, looking for an asserted first char. */
6719    
6720      else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;      else if (bravalue == OP_ASSERT && subreqcharflags >= 0)
6721          {
6722          reqchar = subreqchar;
6723          reqcharflags = subreqcharflags;
6724          }
6725      break;     /* End of processing '(' */      break;     /* End of processing '(' */
6726    
6727    
# Line 6662  for (;; ptr++) Line 6729  for (;; ptr++)
6729      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
6730      are arranged to be the negation of the corresponding OP_values in the      are arranged to be the negation of the corresponding OP_values in the
6731      default case when PCRE_UCP is not set. For the back references, the values      default case when PCRE_UCP is not set. For the back references, the values
6732      are ESC_REF plus the reference number. Only back references and those types      are negative the reference number. Only back references and those types
6733      that consume a character may be repeated. We can test for values between      that consume a character may be repeated. We can test for values between
6734      ESC_b and ESC_Z for the latter; this may have to change if any new ones are      ESC_b and ESC_Z for the latter; this may have to change if any new ones are
6735      ever created. */      ever created. */
6736    
6737      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6738      tempptr = ptr;      tempptr = ptr;
6739      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6740    
6741      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6742    
6743      if (c < 0)      if (escape == 0)
6744          c = ec;
6745        else
6746        {        {
6747        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6748          {          {
6749          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6750            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6682  for (;; ptr++) Line 6752  for (;; ptr++)
6752          continue;          continue;
6753          }          }
6754    
6755        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6756    
6757        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6758        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6759    
6760        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6761          firstchar = REQ_NONE;          firstcharflags = REQ_NONE;
6762    
6763        /* 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. */
6764    
6765        zerofirstchar = firstchar;        zerofirstchar = firstchar;
6766          zerofirstcharflags = firstcharflags;
6767        zeroreqchar = reqchar;        zeroreqchar = reqchar;
6768          zeroreqcharflags = reqcharflags;
6769    
6770        /* \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'
6771        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6772        -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 <
6773        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
6774        -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as        -n, and for the Perl syntax \g{name} the result is ESC_k (as
6775        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6776    
6777        if (-c == ESC_g)        if (escape == ESC_g)
6778          {          {
6779          const pcre_uchar *p;          const pcre_uchar *p;
6780          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
# Line 6722  for (;; ptr++) Line 6794  for (;; ptr++)
6794          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
6795            {            {
6796            BOOL is_a_number = TRUE;            BOOL is_a_number = TRUE;
6797            for (p = ptr + 1; *p != 0 && *p != terminator; p++)            for (p = ptr + 1; *p != 0 && *p != (pcre_uchar)terminator; p++)
6798              {              {
6799              if (!MAX_255(*p)) { is_a_number = FALSE; break; }              if (!MAX_255(*p)) { is_a_number = FALSE; break; }
6800              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;
6801              if ((cd->ctypes[*p] & ctype_word) == 0) break;              if ((cd->ctypes[*p] & ctype_word) == 0) break;
6802              }              }
6803            if (*p != terminator)            if (*p != (pcre_uchar)terminator)
6804              {              {
6805              *errorcodeptr = ERR57;              *errorcodeptr = ERR57;
6806              break;              break;
# Line 6746  for (;; ptr++) Line 6818  for (;; ptr++)
6818    
6819          p = ptr + 2;          p = ptr + 2;
6820          while (IS_DIGIT(*p)) p++;          while (IS_DIGIT(*p)) p++;
6821          if (*p != terminator)          if (*p != (pcre_uchar)terminator)
6822            {            {
6823            *errorcodeptr = ERR57;            *errorcodeptr = ERR57;
6824            break;            break;
# Line 6758  for (;; ptr++) Line 6830  for (;; ptr++)
6830        /* \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).
6831        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6832    
6833        if (-c == ESC_k)        if (escape == ESC_k)
6834          {          {
6835          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6836            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))
# Line 6777  for (;; ptr++) Line 6849  for (;; ptr++)
6849        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
6850        ':' later. */        ':' later. */
6851    
6852        if (-c >= ESC_REF)        if (escape < 0)
6853          {          {
6854          open_capitem *oc;          open_capitem *oc;
6855          recno = -c - ESC_REF;          recno = -escape;
6856    
6857          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6858          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6859          previous = code;          previous = code;
6860          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6861          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
# Line 6807  for (;; ptr++) Line 6879  for (;; ptr++)
6879        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6880    
6881  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6882        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6883          {          {
6884          BOOL negated;          BOOL negated;
6885          int pdata;          unsigned int ptype = 0, pdata = 0;
6886          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
6887          if (ptype < 0) goto FAILED;            goto FAILED;
6888          previous = code;          previous = code;
6889          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6890          *code++ = ptype;          *code++ = ptype;
6891          *code++ = pdata;          *code++ = pdata;
6892          }          }
# Line 6823  for (;; ptr++) Line 6895  for (;; ptr++)
6895        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6896        allowed. */        allowed. */
6897    
6898        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6899          {          {
6900          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6901          goto FAILED;          goto FAILED;
# Line 6833  for (;; ptr++) Line 6905  for (;; ptr++)
6905        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6906        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
6907        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
6908        Unicode property tests. */        Unicode property tests. Note that \b and \B do a one-character
6909          lookbehind. */
6910    
6911        else        else
6912          {          {
6913            if ((escape == ESC_b || escape == ESC_B) && cd->max_lookbehind == 0)
6914              cd->max_lookbehind = 1;
6915  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6916          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6917            {            {
6918            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6919            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6920            }            }
6921          else          else
6922  #endif  #endif
# Line 6849  for (;; ptr++) Line 6924  for (;; ptr++)
6924          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6925    
6926            {            {
6927            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
6928            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
6929            }            }
6930          }          }
6931        continue;        continue;
# Line 6860  for (;; ptr++) Line 6935  for (;; ptr++)
6935      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
6936      handle it as a data character. */      handle it as a data character. */
6937    
6938  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6939      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6940        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6941      else      else
# Line 6893  for (;; ptr++) Line 6968  for (;; ptr++)
6968    
6969      ONE_CHAR:      ONE_CHAR:
6970      previous = code;      previous = code;
6971    
6972        /* For caseless UTF-8 mode when UCP support is available, check whether
6973        this character has more than one other case. If so, generate a special
6974        OP_PROP item instead of OP_CHARI. */
6975    
6976    #ifdef SUPPORT_UCP
6977        if (utf && (options & PCRE_CASELESS) != 0)
6978          {
6979          GETCHAR(c, mcbuffer);
6980          if ((c = UCD_CASESET(c)) != 0)
6981            {
6982            *code++ = OP_PROP;
6983            *code++ = PT_CLIST;
6984            *code++ = c;
6985            if (firstcharflags == REQ_UNSET) firstcharflags = zerofirstcharflags = REQ_NONE;
6986            break;
6987            }
6988          }
6989    #endif
6990    
6991        /* Caseful matches, or not one of the multicase characters. */
6992    
6993      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6994      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6995    
# Line 6906  for (;; ptr++) Line 7003  for (;; ptr++)
7003      Otherwise, leave the firstchar value alone, and don't change it on a zero      Otherwise, leave the firstchar value alone, and don't change it on a zero
7004      repeat. */      repeat. */
7005    
7006      if (firstchar == REQ_UNSET)      if (firstcharflags == REQ_UNSET)
7007        {        {
7008        zerofirstchar = REQ_NONE;        zerofirstcharflags = REQ_NONE;
7009        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7010          zeroreqcharflags = reqcharflags;
7011    
7012        /* If the character is more than one byte long, we can set firstchar        /* If the character is more than one byte long, we can set firstchar
7013        only if it is not to be matched caselessly. */        only if it is not to be matched caselessly. */
# Line 6917  for (;; ptr++) Line 7015  for (;; ptr++)
7015        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7016          {          {
7017          firstchar = mcbuffer[0] | req_caseopt;          firstchar = mcbuffer[0] | req_caseopt;
7018          if (mclength != 1) reqchar = code[-1] | cd->req_varyopt;          firstchar = mcbuffer[0];
7019            firstcharflags = req_caseopt;
7020    
7021            if (mclength != 1)
7022              {
7023              reqchar = code[-1];
7024              reqcharflags = cd->req_varyopt;
7025              }
7026          }          }
7027        else firstchar = reqchar = REQ_NONE;        else firstcharflags = reqcharflags = REQ_NONE;
7028        }        }
7029    
7030      /* firstchar was previously set; we can set reqchar only if the length is      /* firstchar was previously set; we can set reqchar only if the length is
# Line 6928  for (;; ptr++) Line 7033  for (;; ptr++)
7033      else      else
7034        {        {
7035        zerofirstchar = firstchar;        zerofirstchar = firstchar;
7036          zerofirstcharflags = firstcharflags;
7037        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7038          zeroreqcharflags = reqcharflags;
7039        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7040          reqchar = code[-1] | req_caseopt | cd->req_varyopt;          {
7041            reqchar = code[-1];
7042            reqcharflags = req_caseopt | cd->req_varyopt;
7043            }
7044        }        }
7045    
7046      break;            /* End of literal character handling */      break;            /* End of literal character handling */
# Line 6949  return FALSE; Line 7059  return FALSE;
7059    
7060    
7061    
   
7062  /*************************************************  /*************************************************
7063  *     Compile sequence of alternatives           *  *     Compile sequence of alternatives           *
7064  *************************************************/  *************************************************/
# Line 6970  Arguments: Line 7079  Arguments:
7079    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
7080    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
7081    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
7082    firstcharptr   place to put the first required character, or a negative number    firstcharptr    place to put the first required character
7083    reqcharptr     place to put the last required character, or a negative number    firstcharflagsptr place to put the first character flags, or a negative number
7084      reqcharptr     place to put the last required character
7085      reqcharflagsptr place to put the last required character flags, or a negative number
7086    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
7087    cd             points to the data block with tables pointers etc.    cd             points to the data block with tables pointers etc.
7088    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
# Line 6983  Returns:         TRUE on success Line 7094  Returns:         TRUE on success
7094  static BOOL  static BOOL
7095  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,
7096    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
7097    int cond_depth, pcre_int32 *firstcharptr, pcre_int32 *reqcharptr,    int cond_depth,
7098      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
7099      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
7100    branch_chain *bcptr, compile_data *cd, int *lengthptr)    branch_chain *bcptr, compile_data *cd, int *lengthptr)
7101  {  {
7102  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
# Line 6993  pcre_uchar *start_bracket = code; Line 7106  pcre_uchar *start_bracket = code;
7106  pcre_uchar *reverse_count = NULL;  pcre_uchar *reverse_count = NULL;
7107  open_capitem capitem;  open_capitem capitem;
7108  int capnumber = 0;  int capnumber = 0;
7109  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7110  pcre_int32 branchfirstchar, branchreqchar;  pcre_int32 firstcharflags, reqcharflags;
7111    pcre_uint32 branchfirstchar, branchreqchar;
7112    pcre_int32 branchfirstcharflags, branchreqcharflags;
7113  int length;  int length;
7114  int orig_bracount;  unsigned int orig_bracount;
7115  int max_bracount;  unsigned int max_bracount;
7116  branch_chain bc;  branch_chain bc;
7117    
7118  bc.outer = bcptr;  bc.outer = bcptr;
7119  bc.current_branch = code;  bc.current_branch = code;
7120    
7121  firstchar = reqchar = REQ_UNSET;  firstchar = reqchar = 0;
7122    firstcharflags = reqcharflags = REQ_UNSET;
7123    
7124  /* Accumulate the length for use in the pre-compile phase. Start with the  /* Accumulate the length for use in the pre-compile phase. Start with the
7125  length of the BRA and KET and any extra bytes that are required at the  length of the BRA and KET and any extra bytes that are required at the
# Line 7063  for (;;) Line 7179  for (;;)
7179    into the length. */    into the length. */
7180    
7181    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,
7182          &branchreqchar, &bc, cond_depth, cd,          &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc,
7183          (lengthptr == NULL)? NULL : &length))          cond_depth, cd, (lengthptr == NULL)? NULL : &length))
7184      {      {
7185      *ptrptr = ptr;      *ptrptr = ptr;
7186      return FALSE;      return FALSE;
# Line 7085  for (;;) Line 7201  for (;;)
7201      if (*last_branch != OP_ALT)      if (*last_branch != OP_ALT)
7202        {        {
7203        firstchar = branchfirstchar;        firstchar = branchfirstchar;
7204          firstcharflags = branchfirstcharflags;
7205        reqchar = branchreqchar;        reqchar = branchreqchar;
7206          reqcharflags = branchreqcharflags;
7207        }        }
7208    
7209      /* If this is not the first branch, the first char and reqchar have to      /* If this is not the first branch, the first char and reqchar have to
# Line 7099  for (;;) Line 7217  for (;;)
7217        we have to abandon the firstchar for the regex, but if there was        we have to abandon the firstchar for the regex, but if there was
7218        previously no reqchar, it takes on the value of the old firstchar. */        previously no reqchar, it takes on the value of the old firstchar. */
7219    
7220        if (firstchar >= 0 && firstchar != branchfirstchar)        if (firstcharflags >= 0 &&
7221              (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar))
7222          {          {
7223          if (reqchar < 0) reqchar = firstchar;          if (reqcharflags < 0)
7224          firstchar = REQ_NONE;            {
7225              reqchar = firstchar;
7226              reqcharflags = firstcharflags;
7227              }
7228            firstcharflags = REQ_NONE;
7229          }          }
7230    
7231        /* If we (now or from before) have no firstchar, a firstchar from the        /* If we (now or from before) have no firstchar, a firstchar from the
7232        branch becomes a reqchar if there isn't a branch reqchar. */        branch becomes a reqchar if there isn't a branch reqchar. */
7233    
7234        if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0)        if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0)
7235            branchreqchar = branchfirstchar;          {
7236            branchreqchar = branchfirstchar;
7237            branchreqcharflags = branchfirstcharflags;
7238            }
7239    
7240        /* Now ensure that the reqchars match */        /* Now ensure that the reqchars match */
7241    
7242        if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY))        if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) ||
7243          reqchar = REQ_NONE;            reqchar != branchreqchar)
7244        else reqchar |= branchreqchar;   /* To "or" REQ_VARY */          reqcharflags = REQ_NONE;
7245          else
7246            {
7247            reqchar = branchreqchar;
7248            reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */
7249            }
7250        }        }
7251    
7252      /* If lookbehind, check that this branch matches a fixed-length string, and      /* If lookbehind, check that this branch matches a fixed-length string, and
# Line 7144  for (;;) Line 7275  for (;;)
7275          *ptrptr = ptr;          *ptrptr = ptr;
7276          return FALSE;          return FALSE;
7277          }          }
7278        else { PUT(reverse_count, 0, fixed_length); }        else
7279            {
7280            if (fixed_length > cd->max_lookbehind)
7281              cd->max_lookbehind = fixed_length;
7282            PUT(reverse_count, 0, fixed_length);
7283            }
7284        }        }
7285      }      }
7286    
# Line 7206  for (;;) Line 7342  for (;;)
7342      *codeptr = code;      *codeptr = code;
7343      *ptrptr = ptr;      *ptrptr = ptr;
7344      *firstcharptr = firstchar;      *firstcharptr = firstchar;
7345        *firstcharflagsptr = firstcharflags;
7346      *reqcharptr = reqchar;      *reqcharptr = reqchar;
7347        *reqcharflagsptr = reqcharflags;
7348      if (lengthptr != NULL)      if (lengthptr != NULL)
7349        {        {
7350        if (OFLOW_MAX - *lengthptr < length)        if (OFLOW_MAX - *lengthptr < length)
# Line 7276  and the highest back reference was great Line 7414  and the highest back reference was great
7414  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
7415  of the more common cases more precisely.  of the more common cases more precisely.
7416    
7417    ... A second exception is when the .* appears inside an atomic group, because
7418    this prevents the number of characters it matches from being adjusted.
7419    
7420  Arguments:  Arguments:
7421    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7422    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
7423                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7424                    the less precise approach                    the less precise approach
7425    backref_map    the back reference bitmap    cd             points to the compile data block
7426      atomcount      atomic group level
7427    
7428  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7429  */  */
7430    
7431  static BOOL  static BOOL
7432  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7433    unsigned int backref_map)    compile_data *cd, int atomcount)
7434  {  {
7435  do {  do {
7436     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7300  do { Line 7442  do {
7442     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7443         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7444       {       {
7445       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7446       }       }
7447    
7448     /* Capturing brackets */     /* Capturing brackets */
# Line 7310  do { Line 7452  do {
7452       {       {
7453       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7454       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7455       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7456         }
7457    
7458       /* Positive forward assertions and conditions */
7459    
7460       else if (op == OP_ASSERT || op == OP_COND)
7461         {
7462         if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7463       }       }
7464    
7465     /* Other brackets */     /* Atomic groups */
7466    
7467     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ONCE || op == OP_ONCE_NC)
             op == OP_COND)  
7468       {       {
7469       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7470           return FALSE;
7471       }       }
7472    
7473     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7474     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
7475       group. */
7476    
7477     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7478               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7479       {       {
7480       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7481             atomcount > 0 || cd->had_pruneorskip)
7482         return FALSE;         return FALSE;
7483       }       }
7484    
7485     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7486    
7487     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;
7488    
7489     code += GET(code, 1);     code += GET(code, 1);
7490     }     }
7491  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7351  return TRUE; Line 7503  return TRUE;
7503  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
7504  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
7505  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7506  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7507    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7508    count, because once again the assumption no longer holds.
7509    
7510  Arguments:  Arguments:
7511    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7512    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
7513                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7514                    the less precise approach                    the less precise approach
7515    backref_map    the back reference bitmap    cd             points to the compile data
7516      atomcount      atomic group level
7517    
7518  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7519  */  */
7520    
7521  static BOOL  static BOOL
7522  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7523    unsigned int backref_map)    compile_data *cd, int atomcount)
7524  {  {
7525  do {  do {
7526     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7391  do { Line 7546  do {
7546         return FALSE;         return FALSE;
7547    
7548         default:     /* Assertion */         default:     /* Assertion */
7549         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7550         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7551         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7552         break;         break;
# Line 7405  do { Line 7560  do {
7560     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7561         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7562       {       {
7563       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7564       }       }
7565    
7566     /* Capturing brackets */     /* Capturing brackets */
# Line 7415  do { Line 7570  do {
7570       {       {
7571       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7572       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7573       if (!is_startline(scode, new_map, backref_map)) return FALSE;       if (!is_startline(scode, new_map, cd, atomcount)) return FALSE;
7574       }       }
7575    
7576     /* Other brackets */     /* Positive forward assertions */
7577    
7578       else if (op == OP_ASSERT)
7579         {
7580         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7581         }
7582    
7583       /* Atomic brackets */
7584    
7585     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)     else if (op == OP_ONCE || op == OP_ONCE_NC)
7586       {       {
7587       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE;
7588       }       }
7589    
7590     /* .* means "start at start or after \n" if it isn't in brackets that     /* .* means "start at start or after \n" if it isn't in atomic brackets or
7591     may be referenced. */     brackets that may be referenced, as long as the pattern does not contain
7592       *PRUNE or *SKIP, because these break the feature. Consider, for example,
7593       /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the
7594       start of a line. */
7595    
7596     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)
7597       {       {
7598       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 ||
7599             atomcount > 0 || cd->had_pruneorskip)
7600           return FALSE;
7601       }       }
7602    
7603     /* Check for explicit circumflex */     /* Check for explicit circumflex; anything else gives a FALSE result. Note
7604       in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC
7605       because the number of characters matched by .* cannot be adjusted inside
7606       them. */
7607    
7608     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;
7609    
# Line 7461  we return that char, otherwise -1. Line 7631  we return that char, otherwise -1.
7631    
7632  Arguments:  Arguments:
7633    code       points to start of expression (the bracket)    code       points to start of expression (the bracket)
7634      flags       points to the first char flags, or to REQ_NONE
7635    inassert   TRUE if in an assertion    inassert   TRUE if in an assertion
7636    
7637  Returns:     -1 or the fixed first char  Returns:     the fixed first char, or 0 with REQ_NONE in flags
7638  */  */
7639    
7640  static int  static pcre_uint32
7641  find_firstassertedchar(const pcre_uchar *code, BOOL inassert)  find_firstassertedchar(const pcre_uchar *code, pcre_int32 *flags,
7642      BOOL inassert)
7643  {  {
7644  register int c = -1;  register pcre_uint32 c = 0;
7645    int cflags = REQ_NONE;
7646    
7647    *flags = REQ_NONE;
7648  do {  do {
7649     int d;     pcre_uint32 d;
7650       int dflags;
7651     int xl = (*code == OP_CBRA || *code == OP_SCBRA ||     int xl = (*code == OP_CBRA || *code == OP_SCBRA ||
7652               *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0;               *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0;
7653     const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,     const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,
7654       TRUE);       TRUE);
7655     register int op = *scode;     register pcre_uchar op = *scode;
7656    
7657     switch(op)     switch(op)
7658       {       {
7659       default:       default:
7660       return -1;       return 0;
7661    
7662       case OP_BRA:       case OP_BRA:
7663       case OP_BRAPOS:       case OP_BRAPOS:
# Line 7493  do { Line 7669  do {
7669       case OP_ONCE:       case OP_ONCE:
7670       case OP_ONCE_NC:       case OP_ONCE_NC:
7671       case OP_COND:       case OP_COND:
7672       if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0)       d = find_firstassertedchar(scode, &dflags, op == OP_ASSERT);
7673         return -1;       if (dflags < 0)
7674       if (c < 0) c = d; else if (c != d) return -1;         return 0;
7675         if (cflags < 0) { c = d; cflags = dflags; } else if (c != d || cflags != dflags) return 0;
7676       break;       break;
7677    
7678       case OP_EXACT:       case OP_EXACT:
# Line 7506  do { Line 7683  do {
7683       case OP_PLUS:       case OP_PLUS:
7684       case OP_MINPLUS:       case OP_MINPLUS:
7685       case OP_POSPLUS:       case OP_POSPLUS:
7686       if (!inassert) return -1;       if (!inassert) return 0;
7687       if (c < 0) c = scode[1];       if (cflags < 0) { c = scode[1]; cflags = 0; }
7688         else if (c != scode[1]) return -1;         else if (c != scode[1]) return 0;
7689       break;       break;
7690    
7691       case OP_EXACTI:       case OP_EXACTI:
# Line 7519  do { Line 7696  do {
7696       case OP_PLUSI:       case OP_PLUSI:
7697       case OP_MINPLUSI:       case OP_MINPLUSI:
7698       case OP_POSPLUSI:       case OP_POSPLUSI:
7699       if (!inassert) return -1;       if (!inassert) return 0;
7700       if (c < 0) c = scode[1] | REQ_CASELESS;       if (cflags < 0) { c = scode[1]; cflags = REQ_CASELESS; }
7701         else if (c != scode[1]) return -1;         else if (c != scode[1]) return 0;
7702       break;       break;
7703       }       }
7704    
7705     code += GET(code, 1);     code += GET(code, 1);
7706     }     }
7707  while (*code == OP_ALT);  while (*code == OP_ALT);
7708    
7709    *flags = cflags;
7710  return c;  return c;
7711  }  }
7712    
# Line 7555  Returns:        pointer to compiled data Line 7734  Returns:        pointer to compiled data
7734                  with errorptr and erroroffset set                  with errorptr and erroroffset set
7735  */  */
7736    
7737  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7738  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7739  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
7740    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7741  #else  #elif defined COMPILE_PCRE16
7742  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7743  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,
7744    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7745    #elif defined COMPILE_PCRE32
7746    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7747    pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr,
7748      int *erroroffset, const unsigned char *tables)
7749  #endif  #endif
7750  {  {
7751  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7752  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7753  #else  #elif defined COMPILE_PCRE16
7754  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7755    #elif defined COMPILE_PCRE32
7756    return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7757  #endif  #endif
7758  }  }
7759    
7760    
7761  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7762  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7763  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
7764    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7765  #else  #elif defined COMPILE_PCRE16
7766  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7767  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,
7768    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7769    #elif defined COMPILE_PCRE32
7770    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7771    pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr,
7772      const char **errorptr, int *erroroffset, const unsigned char *tables)
7773  #endif  #endif
7774  {  {
7775  real_pcre *re;  REAL_PCRE *re;
7776  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
7777  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7778    pcre_int32 firstcharflags, reqcharflags;
7779  int newline;  int newline;
7780  int errorcode = 0;  int errorcode = 0;
7781  int skipatstart = 0;  int skipatstart = 0;
# Line 7666  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7856  while (ptr[skipatstart] == CHAR_LEFT_PAR
7856    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7857      { skipatstart += 8; options |= PCRE_UTF16; continue; }      { skipatstart += 8; options |= PCRE_UTF16; continue; }
7858  #endif  #endif
7859    #ifdef COMPILE_PCRE32
7860      if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7861        { skipatstart += 8; options |= PCRE_UTF32; continue; }
7862    #endif
7863    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)
7864      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
7865    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 7694  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7888  while (ptr[skipatstart] == CHAR_LEFT_PAR
7888    else break;    else break;
7889    }    }
7890    
7891  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */
7892  utf = (options & PCRE_UTF8) != 0;  utf = (options & PCRE_UTF8) != 0;
7893    
7894  /* 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 7706  not used here. */ Line 7900  not used here. */
7900  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&
7901       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)
7902    {    {
7903    #if defined COMPILE_PCRE8
7904    errorcode = ERR44;    errorcode = ERR44;
7905    #elif defined COMPILE_PCRE16
7906      errorcode = ERR74;
7907    #elif defined COMPILE_PCRE32
7908      errorcode = ERR77;
7909    #endif
7910    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;
7911    }    }
7912  #else  #else
# Line 7810  cd->start_pattern = (const pcre_uchar *) Line 8010  cd->start_pattern = (const pcre_uchar *)
8010  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));
8011  cd->req_varyopt = 0;  cd->req_varyopt = 0;
8012  cd->assert_depth = 0;  cd->assert_depth = 0;
8013    cd->max_lookbehind = 0;
8014  cd->external_options = options;  cd->external_options = options;
8015  cd->external_flags = 0;  cd->external_flags = 0;
8016  cd->open_caps = NULL;  cd->open_caps = NULL;
# Line 7824  ptr += skipatstart; Line 8025  ptr += skipatstart;
8025  code = cworkspace;  code = cworkspace;
8026  *code = OP_BRA;  *code = OP_BRA;
8027  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
8028    FALSE, 0, 0, &firstchar, &reqchar, NULL, cd, &length);    FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL,
8029      cd, &length);
8030  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
8031    
8032  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 7841  externally provided function. Integer ov Line 8043  externally provided function. Integer ov
8043  because nowadays we limit the maximum value of cd->names_found and  because nowadays we limit the maximum value of cd->names_found and
8044  cd->name_entry_size. */  cd->name_entry_size. */
8045    
8046  size = sizeof(real_pcre) + (length + cd->names_found * cd->name_entry_size) * sizeof(pcre_uchar);  size = sizeof(REAL_PCRE) + (length + cd->names_found * cd->name_entry_size) * sizeof(pcre_uchar);
8047  re = (real_pcre *)(PUBL(malloc))(size);  re = (REAL_PCRE *)(PUBL(malloc))(size);
8048    
8049  if (re == NULL)  if (re == NULL)
8050    {    {