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

Diff of /code/trunk/pcre_compile.c

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

revision 982 by ph10, Wed Jun 20 15:15:27 2012 UTC revision 1348 by ph10, Fri Jul 5 10:38:37 2013 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2012 University of Cambridge             Copyright (c) 1997-2013 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  is also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. We do not need to select pcre16_printint.c specially, because the  library. We do not need to select pcre16_printint.c specially, because the
59  COMPILE_PCREx macro will already be appropriately set. */  COMPILE_PCREx macro will already be appropriately set. */
# Line 68  COMPILE_PCREx macro will already be appr Line 68  COMPILE_PCREx macro will already be appr
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 77  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, 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 472  static const char error_texts[] = Line 487  static const char error_texts[] =
487    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
488    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"
489    /* 60 */    /* 60 */
490    "(*VERB) not recognized\0"    "(*VERB) not recognized or malformed\0"
491    "number is too big\0"    "number is too big\0"
492    "subpattern name expected\0"    "subpattern name expected\0"
493    "digit expected after (?+\0"    "digit expected after (?+\0"
# Line 492  static const char error_texts[] = Line 507  static const char error_texts[] =
507    /* 75 */    /* 75 */
508    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
509    "character value in \\u.... sequence is too large\0"    "character value in \\u.... sequence is too large\0"
510      "invalid UTF-32 string\0"
511      "setting UTF is disabled by the application\0"
512    ;    ;
513    
514  /* 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 631  static const pcre_uint8 ebcdic_chartab[] Line 648  static const pcre_uint8 ebcdic_chartab[]
648  #endif  #endif
649    
650    
 /* 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 *);  
   
   
651    
652  /*************************************************  /*************************************************
653  *            Find an error text                  *  *            Find an error text                  *
# Line 658  find_error_text(int n) Line 668  find_error_text(int n)
668  const char *s = error_texts;  const char *s = error_texts;
669  for (; n > 0; n--)  for (; n > 0; n--)
670    {    {
671    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
672    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
673    }    }
674  return s;  return s;
675  }  }
# Line 742  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 752  return (*p == CHAR_RIGHT_CURLY_BRACKET);
752  *************************************************/  *************************************************/
753    
754  /* 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
755  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
756  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
757  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
758  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
759  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
760  sequence.  escape sequence.
761    
762  Arguments:  Arguments:
763    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
764      chptr          points to the data character
765    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
766    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
767    options        the options bits    options        the options bits
768    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
769    
770  Returns:         zero or positive => a data character  Returns:         zero => a data character
771                   negative => a special escape sequence                   positive => a special escape sequence
772                     negative => a back reference
773                   on error, errorcodeptr is set                   on error, errorcodeptr is set
774  */  */
775    
776  static int  static int
777  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
778    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
779  {  {
780  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
781  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
782  const pcre_uchar *ptr = *ptrptr + 1;  const pcre_uchar *ptr = *ptrptr + 1;
783  pcre_int32 c;  pcre_uint32 c;
784    int escape = 0;
785  int i;  int i;
786    
787  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
# Line 776  ptr--;                            /* Set Line 789  ptr--;                            /* Set
789    
790  /* If backslash is at the end of the pattern, it's an error. */  /* If backslash is at the end of the pattern, it's an error. */
791    
792  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
793    
794  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
795  in a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
# Line 785  Otherwise further processing may be requ Line 798  Otherwise further processing may be requ
798  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
799  /* Not alphanumeric */  /* Not alphanumeric */
800  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
801  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0)
802      { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
803    
804  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
805  /* Not alphanumeric */  /* Not alphanumeric */
806  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)) {}
807  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; }
808  #endif  #endif
809    
810  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
# Line 798  else if ((i = escapes[c - 0x48]) != 0) Line 812  else if ((i = escapes[c - 0x48]) != 0)
812  else  else
813    {    {
814    const pcre_uchar *oldptr;    const pcre_uchar *oldptr;
815    BOOL braced, negated;    BOOL braced, negated, overflow;
816      int s;
817    
818    switch (c)    switch (c)
819      {      {
# Line 823  else Line 838  else
838          c = 0;          c = 0;
839          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
840            {            {
841            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
842  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
843            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
844            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 833  else Line 848  else
848  #endif  #endif
849            }            }
850    
851  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
852          if (c > (utf ? 0x10ffff : 0xff))          if (c > (utf ? 0x10ffffU : 0xffU))
853  #else  #elif defined COMPILE_PCRE16
854  #ifdef COMPILE_PCRE16          if (c > (utf ? 0x10ffffU : 0xffffU))
855          if (c > (utf ? 0x10ffff : 0xffff))  #elif defined COMPILE_PCRE32
856  #endif          if (utf && c > 0x10ffffU)
857  #endif  #endif
858            {            {
859            *errorcodeptr = ERR76;            *errorcodeptr = ERR76;
# Line 870  else Line 885  else
885      (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
886      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
887      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
888      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
889    
890      case CHAR_g:      case CHAR_g:
891      if (isclass) break;      if (isclass) break;
892      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
893        {        {
894        c = -ESC_g;        escape = ESC_g;
895        break;        break;
896        }        }
897    
# Line 885  else Line 900  else
900      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
901        {        {
902        const pcre_uchar *p;        const pcre_uchar *p;
903        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
904          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
905        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
906          {          {
907          c = -ESC_k;          escape = ESC_k;
908          break;          break;
909          }          }
910        braced = TRUE;        braced = TRUE;
# Line 905  else Line 920  else
920      else negated = FALSE;      else negated = FALSE;
921    
922      /* The integer range is limited by the machine's int representation. */      /* The integer range is limited by the machine's int representation. */
923      c = 0;      s = 0;
924        overflow = FALSE;
925      while (IS_DIGIT(ptr[1]))      while (IS_DIGIT(ptr[1]))
926        {        {
927        if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */        if (s > INT_MAX / 10 - 1) /* Integer overflow */
928          {          {
929          c = -1;          overflow = TRUE;
930          break;          break;
931          }          }
932        c = c * 10 + *(++ptr) - CHAR_0;        s = s * 10 + (int)(*(++ptr) - CHAR_0);
933        }        }
934      if (((unsigned int)c) > INT_MAX) /* Integer overflow */      if (overflow) /* Integer overflow */
935        {        {
936        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
937          ptr++;          ptr++;
# Line 929  else Line 945  else
945        break;        break;
946        }        }
947    
948      if (c == 0)      if (s == 0)
949        {        {
950        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
951        break;        break;
# Line 937  else Line 953  else
953    
954      if (negated)      if (negated)
955        {        {
956        if (c > bracount)        if (s > bracount)
957          {          {
958          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
959          break;          break;
960          }          }
961        c = bracount - (c - 1);        s = bracount - (s - 1);
962        }        }
963    
964      c = -(ESC_REF + c);      escape = -s;
965      break;      break;
966    
967      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 967  else Line 983  else
983        {        {
984        oldptr = ptr;        oldptr = ptr;
985        /* The integer range is limited by the machine's int representation. */        /* The integer range is limited by the machine's int representation. */
986        c -= CHAR_0;        s = (int)(c -CHAR_0);
987          overflow = FALSE;
988        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
989          {          {
990          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */          if (s > INT_MAX / 10 - 1) /* Integer overflow */
991            {            {
992            c = -1;            overflow = TRUE;
993            break;            break;
994            }            }
995          c = c * 10 + *(++ptr) - CHAR_0;          s = s * 10 + (int)(*(++ptr) - CHAR_0);
996          }          }
997        if (((unsigned int)c) > INT_MAX) /* Integer overflow */        if (overflow) /* Integer overflow */
998          {          {
999          while (IS_DIGIT(ptr[1]))          while (IS_DIGIT(ptr[1]))
1000            ptr++;            ptr++;
1001          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1002          break;          break;
1003          }          }
1004        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1005          {          {
1006          c = -(ESC_REF + c);          escape = -s;
1007          break;          break;
1008          }          }
1009        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 1033  else Line 1050  else
1050          c = 0;          c = 0;
1051          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1052            {            {
1053            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1054  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1055            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1056            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 1051  else Line 1068  else
1068        const pcre_uchar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1069    
1070        c = 0;        c = 0;
1071          overflow = FALSE;
1072        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1073          {          {
1074          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1075          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1076    
1077    #ifdef COMPILE_PCRE32
1078            if (c >= 0x10000000l) { overflow = TRUE; break; }
1079    #endif
1080    
1081  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1082          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1083          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 1064  else Line 1086  else
1086          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1087  #endif  #endif
1088    
1089  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1090          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1091  #else  #elif defined COMPILE_PCRE16
1092  #ifdef COMPILE_PCRE16          if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1093          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }  #elif defined COMPILE_PCRE32
1094  #endif          if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1095  #endif  #endif
1096          }          }
1097    
1098        if (c < 0)        if (overflow)
1099          {          {
1100          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1101          *errorcodeptr = ERR34;          *errorcodeptr = ERR34;
# Line 1095  else Line 1117  else
1117      c = 0;      c = 0;
1118      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)
1119        {        {
1120        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1121        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1122  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1123        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 1114  else Line 1136  else
1136    
1137      case CHAR_c:      case CHAR_c:
1138      c = *(++ptr);      c = *(++ptr);
1139      if (c == 0)      if (c == CHAR_NULL)
1140        {        {
1141        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1142        break;        break;
# Line 1154  else Line 1176  else
1176  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1177  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1178    
1179  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1180       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1181    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1182    
1183  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1184    
1185  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1186    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1187    
1188  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1189    
1190  *ptrptr = ptr;  *ptrptr = ptr;
1191  return c;  *chptr = c;
1192    return escape;
1193  }  }
1194    
   
   
1195  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1196  /*************************************************  /*************************************************
1197  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1184  escape sequence. Line 1205  escape sequence.
1205  Argument:  Argument:
1206    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1207    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
1208    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
1209      pdataptr       points to an unsigned int that is set to the detailed property value
1210    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1211    
1212  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
1213  */  */
1214    
1215  static int  static BOOL
1216  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1217      unsigned int *pdataptr, int *errorcodeptr)
1218  {  {
1219  int c, i, bot, top;  pcre_uchar c;
1220    int i, bot, top;
1221  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1222  pcre_uchar name[32];  pcre_uchar name[32];
1223    
1224  c = *(++ptr);  c = *(++ptr);
1225  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1226    
1227  *negptr = FALSE;  *negptr = FALSE;
1228    
# Line 1215  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1239  if (c == CHAR_LEFT_CURLY_BRACKET)
1239    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1240      {      {
1241      c = *(++ptr);      c = *(++ptr);
1242      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1243      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1244      name[i] = c;      name[i] = c;
1245      }      }
# Line 1240  top = PRIV(utt_size); Line 1264  top = PRIV(utt_size);
1264    
1265  while (bot < top)  while (bot < top)
1266    {    {
1267      int r;
1268    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1269    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);
1270    if (c == 0)    if (r == 0)
1271      {      {
1272      *dptr = PRIV(utt)[i].value;      *ptypeptr = PRIV(utt)[i].type;
1273      return PRIV(utt)[i].type;      *pdataptr = PRIV(utt)[i].value;
1274        return TRUE;
1275      }      }
1276    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1277    }    }
1278    
1279  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1280  *ptrptr = ptr;  *ptrptr = ptr;
1281  return -1;  return FALSE;
1282    
1283  ERROR_RETURN:  ERROR_RETURN:
1284  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1285  *ptrptr = ptr;  *ptrptr = ptr;
1286  return -1;  return FALSE;
1287  }  }
1288  #endif  #endif
1289    
# Line 1292  int max = -1; Line 1318  int max = -1;
1318  /* 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
1319  an integer overflow. */  an integer overflow. */
1320    
1321  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1322  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1323    {    {
1324    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1307  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1333  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1333    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1334      {      {
1335      max = 0;      max = 0;
1336      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1337      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1338        {        {
1339        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1362  Arguments: Line 1388  Arguments:
1388    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1389    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1390    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1391    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1392    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1393    
1394  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 1384  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1410  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1410    {    {
1411    /* Handle specials such as (*SKIP) or (*UTF8) etc. */    /* Handle specials such as (*SKIP) or (*UTF8) etc. */
1412    
1413    if (ptr[1] == CHAR_ASTERISK) ptr += 2;    if (ptr[1] == CHAR_ASTERISK)
1414        {
1415        ptr += 2;
1416        while (ptr < cd->end_pattern && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1417        }
1418    
1419    /* Handle a normal, unnamed capturing parenthesis. */    /* Handle a normal, unnamed capturing parenthesis. */
1420    
# Line 1408  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1438  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1438    
1439    else if (ptr[2] == CHAR_NUMBER_SIGN)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1440      {      {
1441      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;      for (ptr += 3; *ptr != CHAR_NULL; ptr++)
1442          if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1443      goto FAIL_EXIT;      goto FAIL_EXIT;
1444      }      }
1445    
# Line 1421  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1452  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1452      ptr += 2;      ptr += 2;
1453      if (ptr[1] != CHAR_QUESTION_MARK)      if (ptr[1] != CHAR_QUESTION_MARK)
1454        {        {
1455        while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;        while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1456        if (*ptr != 0) ptr++;        if (*ptr != CHAR_NULL) ptr++;
1457        }        }
1458      }      }
1459    
# Line 1438  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1469  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1469      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1470          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1471        {        {
1472        int term;        pcre_uchar term;
1473        const pcre_uchar *thisname;        const pcre_uchar *thisname;
1474        *count += 1;        *count += 1;
1475        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
# Line 1446  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1477  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1477        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1478        thisname = ptr;        thisname = ptr;
1479        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1480        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1481            STRNCMP_UC_UC(name, thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1482          return *count;          return *count;
1483        term++;        term++;
1484        }        }
# Line 1465  for (; ptr < cd->end_pattern; ptr++) Line 1496  for (; ptr < cd->end_pattern; ptr++)
1496    
1497    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1498      {      {
1499      if (*(++ptr) == 0) goto FAIL_EXIT;      if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1500      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1501        {        {
1502        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1503        if (*ptr == 0) goto FAIL_EXIT;        if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1504        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1505        }        }
1506      continue;      continue;
# Line 1513  for (; ptr < cd->end_pattern; ptr++) Line 1544  for (; ptr < cd->end_pattern; ptr++)
1544    
1545      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1546        {        {
1547        if (*ptr == 0) return -1;        if (*ptr == CHAR_NULL) return -1;
1548        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1549          {          {
1550          if (*(++ptr) == 0) goto FAIL_EXIT;          if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1551          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1552            {            {
1553            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1554            if (*ptr == 0) goto FAIL_EXIT;            if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1555            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1556            }            }
1557          continue;          continue;
# Line 1534  for (; ptr < cd->end_pattern; ptr++) Line 1565  for (; ptr < cd->end_pattern; ptr++)
1565    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1566      {      {
1567      ptr++;      ptr++;
1568      while (*ptr != 0)      while (*ptr != CHAR_NULL)
1569        {        {
1570        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1571        ptr++;        ptr++;
# Line 1542  for (; ptr < cd->end_pattern; ptr++) Line 1573  for (; ptr < cd->end_pattern; ptr++)
1573        if (utf) FORWARDCHAR(ptr);        if (utf) FORWARDCHAR(ptr);
1574  #endif  #endif
1575        }        }
1576      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1577      continue;      continue;
1578      }      }
1579    
# Line 1552  for (; ptr < cd->end_pattern; ptr++) Line 1583  for (; ptr < cd->end_pattern; ptr++)
1583      {      {
1584      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1585      if (rc > 0) return rc;      if (rc > 0) return rc;
1586      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1587      }      }
1588    
1589    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
# Line 1596  Arguments: Line 1627  Arguments:
1627    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1628    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1629    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1630    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1631    
1632  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1633  */  */
# Line 1617  matching closing parens. That is why we Line 1648  matching closing parens. That is why we
1648  for (;;)  for (;;)
1649    {    {
1650    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1651    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == CHAR_NULL) break;
1652    }    }
1653    
1654  return rc;  return rc;
# Line 1699  and doing the check at the end; a flag s Line 1730  and doing the check at the end; a flag s
1730    
1731  Arguments:  Arguments:
1732    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1733    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1734    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1735    cd       the "compile data" structure    cd       the "compile data" structure
1736    
# Line 1725  for (;;) Line 1756  for (;;)
1756    {    {
1757    int d;    int d;
1758    pcre_uchar *ce, *cs;    pcre_uchar *ce, *cs;
1759    register int op = *cc;    register pcre_uchar op = *cc;
1760    
1761    switch (op)    switch (op)
1762      {      {
# Line 1845  for (;;) Line 1876  for (;;)
1876      case OP_EXACTI:      case OP_EXACTI:
1877      case OP_NOTEXACT:      case OP_NOTEXACT:
1878      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1879      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1880      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1881  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
1882      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
# Line 1854  for (;;) Line 1885  for (;;)
1885    
1886      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1887      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1888      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)
1889          cc += 2;
1890      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1891      break;      break;
1892    
# Line 1889  for (;;) Line 1921  for (;;)
1921    
1922      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1923    
 #if defined SUPPORT_UTF || defined COMPILE_PCRE16  
     case OP_XCLASS:  
     cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];  
     /* Fall through */  
 #endif  
   
1924      case OP_CLASS:      case OP_CLASS:
1925      case OP_NCLASS:      case OP_NCLASS:
1926    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1927        case OP_XCLASS:
1928        /* The original code caused an unsigned overflow in 64 bit systems,
1929        so now we use a conditional statement. */
1930        if (op == OP_XCLASS)
1931          cc += GET(cc, 1);
1932        else
1933          cc += PRIV(OP_lengths)[OP_CLASS];
1934    #else
1935      cc += PRIV(OP_lengths)[OP_CLASS];      cc += PRIV(OP_lengths)[OP_CLASS];
1936    #endif
1937    
1938      switch (*cc)      switch (*cc)
1939        {        {
# Line 1912  for (;;) Line 1948  for (;;)
1948        case OP_CRRANGE:        case OP_CRRANGE:
1949        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1950        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1951        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1952        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
1953        break;        break;
1954    
# Line 2028  length. Line 2064  length.
2064    
2065  Arguments:  Arguments:
2066    code        points to start of expression    code        points to start of expression
2067    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2068    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2069    
2070  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 2039  PRIV(find_bracket)(const pcre_uchar *cod Line 2075  PRIV(find_bracket)(const pcre_uchar *cod
2075  {  {
2076  for (;;)  for (;;)
2077    {    {
2078    register int c = *code;    register pcre_uchar c = *code;
2079    
2080    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2081    
# Line 2062  for (;;) Line 2098  for (;;)
2098    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2099             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2100      {      {
2101      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2102      if (n == number) return (pcre_uchar *)code;      if (n == number) return (pcre_uchar *)code;
2103      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
2104      }      }
# Line 2092  for (;;) Line 2128  for (;;)
2128        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2129        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2130        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2131        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2132          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2133        break;        break;
2134    
2135        case OP_MARK:        case OP_MARK:
2136        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2137        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2138        case OP_THEN_ARG:        case OP_THEN_ARG:
2139        code += code[1];        code += code[1];
2140        break;        break;
# Line 2115  for (;;) Line 2148  for (;;)
2148    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
2149    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2150    
2151  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2152      if (utf) switch(c)      if (utf) switch(c)
2153        {        {
2154        case OP_CHAR:        case OP_CHAR:
# Line 2167  instance of OP_RECURSE. Line 2200  instance of OP_RECURSE.
2200    
2201  Arguments:  Arguments:
2202    code        points to start of expression    code        points to start of expression
2203    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2204    
2205  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
2206  */  */
# Line 2177  find_recurse(const pcre_uchar *code, BOO Line 2210  find_recurse(const pcre_uchar *code, BOO
2210  {  {
2211  for (;;)  for (;;)
2212    {    {
2213    register int c = *code;    register pcre_uchar c = *code;
2214    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2215    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2216    
# Line 2212  for (;;) Line 2245  for (;;)
2245        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2246        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2247        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2248        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2249          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2250        break;        break;
2251    
2252        case OP_MARK:        case OP_MARK:
2253        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2254        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2255        case OP_THEN_ARG:        case OP_THEN_ARG:
2256        code += code[1];        code += code[1];
2257        break;        break;
# Line 2235  for (;;) Line 2265  for (;;)
2265      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
2266      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2267    
2268  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2269      if (utf) switch(c)      if (utf) switch(c)
2270        {        {
2271        case OP_CHAR:        case OP_CHAR:
# Line 2321  bracket whose current branch will alread Line 2351  bracket whose current branch will alread
2351  Arguments:  Arguments:
2352    code        points to start of search    code        points to start of search
2353    endcode     points to where to stop    endcode     points to where to stop
2354    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2355    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2356      recurses    chain of recurse_check to catch mutual recursion
2357    
2358  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2359  */  */
2360    
2361    typedef struct recurse_check {
2362      struct recurse_check *prev;
2363      const pcre_uchar *group;
2364    } recurse_check;
2365    
2366  static BOOL  static BOOL
2367  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2368    BOOL utf, compile_data *cd)    BOOL utf, compile_data *cd, recurse_check *recurses)
2369  {  {
2370  register int c;  register pcre_uchar c;
2371    recurse_check this_recurse;
2372    
2373  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2374       code < endcode;       code < endcode;
2375       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
# Line 2339  for (code = first_significant_code(code Line 2377  for (code = first_significant_code(code
2377    const pcre_uchar *ccode;    const pcre_uchar *ccode;
2378    
2379    c = *code;    c = *code;
2380    
2381    /* Skip over forward assertions; the other assertions are skipped by    /* Skip over forward assertions; the other assertions are skipped by
2382    first_significant_code() with a TRUE final argument. */    first_significant_code() with a TRUE final argument. */
2383    
# Line 2359  for (code = first_significant_code(code Line 2397  for (code = first_significant_code(code
2397    
2398    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2399      {      {
2400      const pcre_uchar *scode;      const pcre_uchar *scode = cd->start_code + GET(code, 1);
2401      BOOL empty_branch;      BOOL empty_branch;
2402    
2403      /* Test for forward reference */      /* Test for forward reference or uncompleted reference. This is disabled
2404        when called to scan a completed pattern by setting cd->start_workspace to
2405      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      NULL. */
2406        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;  
2407        if (cd->start_workspace != NULL)
2408          {
2409          const pcre_uchar *tcode;
2410          for (tcode = cd->start_workspace; tcode < cd->hwm; tcode += LINK_SIZE)
2411            if ((int)GET(tcode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2412          if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2413          }
2414    
2415        /* If we are scanning a completed pattern, there are no forward references
2416        and all groups are complete. We need to detect whether this is a recursive
2417        call, as otherwise there will be an infinite loop. If it is a recursion,
2418        just skip over it. Simple recursions are easily detected. For mutual
2419        recursions we keep a chain on the stack. */
2420    
2421        else
2422          {
2423          recurse_check *r = recurses;
2424          const pcre_uchar *endgroup = scode;
2425    
2426          do endgroup += GET(endgroup, 1); while (*endgroup == OP_ALT);
2427          if (code >= scode && code <= endgroup) continue;  /* Simple recursion */
2428    
2429          for (r = recurses; r != NULL; r = r->prev)
2430            if (r->group == scode) break;
2431          if (r != NULL) continue;   /* Mutual recursion */
2432          }
2433    
2434      /* Not a forward reference, test for completed backward reference */      /* Completed reference; scan the referenced group, remembering it on the
2435        stack chain to detect mutual recursions. */
2436    
2437      empty_branch = FALSE;      empty_branch = FALSE;
2438      scode = cd->start_code + GET(code, 1);      this_recurse.prev = recurses;
2439      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      this_recurse.group = scode;
2440    
     /* Completed backwards reference */  
   
2441      do      do
2442        {        {
2443        if (could_be_empty_branch(scode, endcode, utf, cd))        if (could_be_empty_branch(scode, endcode, utf, cd, &this_recurse))
2444          {          {
2445          empty_branch = TRUE;          empty_branch = TRUE;
2446          break;          break;
# Line 2433  for (code = first_significant_code(code Line 2496  for (code = first_significant_code(code
2496        empty_branch = FALSE;        empty_branch = FALSE;
2497        do        do
2498          {          {
2499          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd, NULL))
2500            empty_branch = TRUE;            empty_branch = TRUE;
2501          code += GET(code, 1);          code += GET(code, 1);
2502          }          }
# Line 2491  for (code = first_significant_code(code Line 2554  for (code = first_significant_code(code
2554    
2555      /* Opcodes that must match a character */      /* Opcodes that must match a character */
2556    
2557        case OP_ANY:
2558        case OP_ALLANY:
2559        case OP_ANYBYTE:
2560    
2561      case OP_PROP:      case OP_PROP:
2562      case OP_NOTPROP:      case OP_NOTPROP:
2563        case OP_ANYNL:
2564    
2565        case OP_NOT_HSPACE:
2566        case OP_HSPACE:
2567        case OP_NOT_VSPACE:
2568        case OP_VSPACE:
2569      case OP_EXTUNI:      case OP_EXTUNI:
2570    
2571      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2572      case OP_DIGIT:      case OP_DIGIT:
2573      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2574      case OP_WHITESPACE:      case OP_WHITESPACE:
2575      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2576      case OP_WORDCHAR:      case OP_WORDCHAR:
2577      case OP_ANY:  
     case OP_ALLANY:  
     case OP_ANYBYTE:  
2578      case OP_CHAR:      case OP_CHAR:
2579      case OP_CHARI:      case OP_CHARI:
2580      case OP_NOT:      case OP_NOT:
2581      case OP_NOTI:      case OP_NOTI:
2582    
2583      case OP_PLUS:      case OP_PLUS:
2584        case OP_PLUSI:
2585      case OP_MINPLUS:      case OP_MINPLUS:
2586      case OP_POSPLUS:      case OP_MINPLUSI:
2587      case OP_EXACT:  
2588      case OP_NOTPLUS:      case OP_NOTPLUS:
2589        case OP_NOTPLUSI:
2590      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
2591        case OP_NOTMINPLUSI:
2592    
2593        case OP_POSPLUS:
2594        case OP_POSPLUSI:
2595      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
2596        case OP_NOTPOSPLUSI:
2597    
2598        case OP_EXACT:
2599        case OP_EXACTI:
2600      case OP_NOTEXACT:      case OP_NOTEXACT:
2601        case OP_NOTEXACTI:
2602    
2603      case OP_TYPEPLUS:      case OP_TYPEPLUS:
2604      case OP_TYPEMINPLUS:      case OP_TYPEMINPLUS:
2605      case OP_TYPEPOSPLUS:      case OP_TYPEPOSPLUS:
2606      case OP_TYPEEXACT:      case OP_TYPEEXACT:
2607    
2608      return FALSE;      return FALSE;
2609    
2610      /* These are going to continue, as they may be empty, but we have to      /* These are going to continue, as they may be empty, but we have to
# Line 2538  for (code = first_significant_code(code Line 2624  for (code = first_significant_code(code
2624      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2625      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2626      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2627      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2628        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2629      break;      break;
2630    
2631      /* End of branch */      /* End of branch */
# Line 2552  for (code = first_significant_code(code Line 2638  for (code = first_significant_code(code
2638      return TRUE;      return TRUE;
2639    
2640      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2641      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO and their caseless and negative versions may be
2642        followed by a multibyte character. */
2643    
2644  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2645      case OP_STAR:      case OP_STAR:
2646      case OP_STARI:      case OP_STARI:
2647        case OP_NOTSTAR:
2648        case OP_NOTSTARI:
2649    
2650      case OP_MINSTAR:      case OP_MINSTAR:
2651      case OP_MINSTARI:      case OP_MINSTARI:
2652        case OP_NOTMINSTAR:
2653        case OP_NOTMINSTARI:
2654    
2655      case OP_POSSTAR:      case OP_POSSTAR:
2656      case OP_POSSTARI:      case OP_POSSTARI:
2657        case OP_NOTPOSSTAR:
2658        case OP_NOTPOSSTARI:
2659    
2660      case OP_QUERY:      case OP_QUERY:
2661      case OP_QUERYI:      case OP_QUERYI:
2662        case OP_NOTQUERY:
2663        case OP_NOTQUERYI:
2664    
2665      case OP_MINQUERY:      case OP_MINQUERY:
2666      case OP_MINQUERYI:      case OP_MINQUERYI:
2667        case OP_NOTMINQUERY:
2668        case OP_NOTMINQUERYI:
2669    
2670      case OP_POSQUERY:      case OP_POSQUERY:
2671      case OP_POSQUERYI:      case OP_POSQUERYI:
2672        case OP_NOTPOSQUERY:
2673        case OP_NOTPOSQUERYI:
2674    
2675      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2676      break;      break;
2677    
2678      case OP_UPTO:      case OP_UPTO:
2679      case OP_UPTOI:      case OP_UPTOI:
2680        case OP_NOTUPTO:
2681        case OP_NOTUPTOI:
2682    
2683      case OP_MINUPTO:      case OP_MINUPTO:
2684      case OP_MINUPTOI:      case OP_MINUPTOI:
2685        case OP_NOTMINUPTO:
2686        case OP_NOTMINUPTOI:
2687    
2688      case OP_POSUPTO:      case OP_POSUPTO:
2689      case OP_POSUPTOI:      case OP_POSUPTOI:
2690        case OP_NOTPOSUPTO:
2691        case OP_NOTPOSUPTOI:
2692    
2693      if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);      if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2694      break;      break;
2695  #endif  #endif
# Line 2586  for (code = first_significant_code(code Line 2700  for (code = first_significant_code(code
2700      case OP_MARK:      case OP_MARK:
2701      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
2702      case OP_SKIP_ARG:      case OP_SKIP_ARG:
     code += code[1];  
     break;  
   
2703      case OP_THEN_ARG:      case OP_THEN_ARG:
2704      code += code[1];      code += code[1];
2705      break;      break;
# Line 2620  Arguments: Line 2731  Arguments:
2731    code        points to start of the recursion    code        points to start of the recursion
2732    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2733    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2734    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2735    cd          pointers to tables etc    cd          pointers to tables etc
2736    
2737  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2632  could_be_empty(const pcre_uchar *code, c Line 2743  could_be_empty(const pcre_uchar *code, c
2743  {  {
2744  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2745    {    {
2746    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd, NULL))
2747      return FALSE;      return FALSE;
2748    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2749    }    }
# Line 2686  Returns:   TRUE or FALSE Line 2797  Returns:   TRUE or FALSE
2797  static BOOL  static BOOL
2798  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2799  {  {
2800  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2801  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2802  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
2803    {    {
2804    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2805      ptr++;      ptr++;
# Line 2735  register int yield = 0; Line 2846  register int yield = 0;
2846  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2847    {    {
2848    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2849      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2850    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2851    yield++;    yield++;
2852    }    }
# Line 2767  value in the reference (which is a group Line 2878  value in the reference (which is a group
2878  Arguments:  Arguments:
2879    group      points to the start of the group    group      points to the start of the group
2880    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2881    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2882    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2883    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
2884    
# Line 2790  while ((ptr = (pcre_uchar *)find_recurse Line 2901  while ((ptr = (pcre_uchar *)find_recurse
2901    
2902    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2903      {      {
2904      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2905      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2906        {        {
2907        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2803  while ((ptr = (pcre_uchar *)find_recurse Line 2914  while ((ptr = (pcre_uchar *)find_recurse
2914    
2915    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2916      {      {
2917      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2918      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2919      }      }
2920    
# Line 2871  PUT(previous_callout, 2 + LINK_SIZE, len Line 2982  PUT(previous_callout, 2 + LINK_SIZE, len
2982  *************************************************/  *************************************************/
2983    
2984  /* 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
2985  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
2986  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
2987  start address.  start address. A character with multiple other cases is returned on its own
2988    with a special return value.
2989    
2990  Arguments:  Arguments:
2991    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2881  Arguments: Line 2993  Arguments:
2993    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2994    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2995    
2996  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2997                   0 when a range is returned
2998                  >0 the CASESET offset for char with multiple other cases
2999                    in this case, ocptr contains the original
3000  */  */
3001    
3002  static BOOL  static int
3003  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
3004    unsigned int *odptr)    pcre_uint32 *odptr)
3005  {  {
3006  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
3007    unsigned int co;
3008    
3009    /* Find the first character that has an other case. If it has multiple other
3010    cases, return its case offset value. */
3011    
3012  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
3013    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
3014      if ((co = UCD_CASESET(c)) != 0)
3015        {
3016        *ocptr = c++;   /* Character that has the set */
3017        *cptr = c;      /* Rest of input range */
3018        return (int)co;
3019        }
3020      if ((othercase = UCD_OTHERCASE(c)) != c) break;
3021      }
3022    
3023  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
3024    
3025  *ocptr = othercase;  *ocptr = othercase;
3026  next = othercase + 1;  next = othercase + 1;
# Line 2904  for (++c; c <= d; c++) Line 3031  for (++c; c <= d; c++)
3031    next++;    next++;
3032    }    }
3033    
3034  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
3035  *cptr = c;  *cptr = c;             /* Rest of input range */
3036    return 0;
 return TRUE;  
3037  }  }
3038    
3039    
# Line 2929  Returns:       TRUE if auto-possessifyin Line 3055  Returns:       TRUE if auto-possessifyin
3055  */  */
3056    
3057  static BOOL  static BOOL
3058  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)
3059  {  {
3060    #ifdef SUPPORT_UCP
3061    const pcre_uint32 *p;
3062    #endif
3063    
3064  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
3065    
3066  switch(ptype)  switch(ptype)
3067    {    {
3068    case PT_LAMP:    case PT_LAMP:
# Line 2969  switch(ptype) Line 3100  switch(ptype)
3100    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3101            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3102            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3103    
3104    #ifdef SUPPORT_UCP
3105      case PT_CLIST:
3106      p = PRIV(ucd_caseless_sets) + prop->caseset;
3107      for (;;)
3108        {
3109        if (c < *p) return !negated;
3110        if (c == *p++) return negated;
3111        }
3112      break;  /* Control never reaches here */
3113    #endif
3114    }    }
3115    
3116  return FALSE;  return FALSE;
3117  }  }
3118  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2986  sense to automatically possessify the re Line 3129  sense to automatically possessify the re
3129    
3130  Arguments:  Arguments:
3131    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3132    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3133    ptr           next character in pattern    ptr           next character in pattern
3134    options       options bits    options       options bits
3135    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2998  static BOOL Line 3141  static BOOL
3141  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3142    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3143  {  {
3144  pcre_int32 c, next;  pcre_uint32 c = NOTACHAR;
3145  int op_code = *previous++;  pcre_uint32 next;
3146    int escape;
3147    pcre_uchar op_code = *previous++;
3148    
3149  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3150    
# Line 3011  if ((options & PCRE_EXTENDED) != 0) Line 3156  if ((options & PCRE_EXTENDED) != 0)
3156      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3157        {        {
3158        ptr++;        ptr++;
3159        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3160          {          {
3161          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3162          ptr++;          ptr++;
# Line 3030  value is a character, a negative value i Line 3175  value is a character, a negative value i
3175  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3176    {    {
3177    int temperrorcode = 0;    int temperrorcode = 0;
3178    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options,
3179        FALSE);
3180    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3181    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3182    }    }
3183  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
3184    {    {
3185      escape = 0;
3186  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3187    if (utf) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3188  #endif  #endif
# Line 3053  if ((options & PCRE_EXTENDED) != 0) Line 3200  if ((options & PCRE_EXTENDED) != 0)
3200      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3201        {        {
3202        ptr++;        ptr++;
3203        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3204          {          {
3205          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3206          ptr++;          ptr++;
# Line 3072  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3219  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3219    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)
3220      return FALSE;      return FALSE;
3221    
3222  /* 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. */  
3223    
3224  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3225        op_code == OP_NOT || op_code == OP_NOTI)
3226    {    {
   case OP_CHAR:  
3227  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3228    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3229  #else  #else
3230    c = *previous;    c = *previous;
3231  #endif  #endif
3232    return c != next;    }
3233    
3234    /* Now compare the next item with the previous opcode. First, handle cases when
3235    the next item is a character. */
3236    
3237    /* For CHARI (caseless character) we must check the other case. If we have  if (escape == 0)
3238    Unicode property support, we can use it to test the other case of    {
3239    high-valued characters. */    /* For a caseless UTF match, the next character may have more than one other
3240      case, which maps to the special PT_CLIST property. Check this first. */
3241    
3242    case OP_CHARI:  #ifdef SUPPORT_UCP
3243  #ifdef SUPPORT_UTF    if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3244    GETCHARTEST(c, previous);      {
3245  #else      unsigned int ocs = UCD_CASESET(next);
3246    c = *previous;      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3247        }
3248  #endif  #endif
3249    if (c == next) return FALSE;  
3250  #ifdef SUPPORT_UTF    switch(op_code)
   if (utf)  
3251      {      {
3252      unsigned int othercase;      case OP_CHAR:
3253      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3254    
3255        /* For CHARI (caseless character) we must check the other case. If we have
3256        Unicode property support, we can use it to test the other case of
3257        high-valued characters. We know that next can have only one other case,
3258        because multi-other-case characters are dealt with above. */
3259    
3260        case OP_CHARI:
3261        if (c == next) return FALSE;
3262    #ifdef SUPPORT_UTF
3263        if (utf)
3264          {
3265          pcre_uint32 othercase;
3266          if (next < 128) othercase = cd->fcc[next]; else
3267  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3268      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3269  #else  #else
3270      othercase = NOTACHAR;        othercase = NOTACHAR;
3271  #endif  #endif
3272      return (unsigned int)c != othercase;        return c != othercase;
3273      }        }
3274    else      else
3275  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3276    return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3277    
3278    case OP_NOT:      case OP_NOT:
3279  #ifdef SUPPORT_UTF      return c == next;
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   return c == next;  
3280    
3281    case OP_NOTI:      case OP_NOTI:
3282  #ifdef SUPPORT_UTF      if (c == next) return TRUE;
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   if (c == next) return TRUE;  
3283  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3284    if (utf)      if (utf)
3285      {        {
3286      unsigned int othercase;        pcre_uint32 othercase;
3287      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3288  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3289      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3290  #else  #else
3291      othercase = NOTACHAR;        othercase = NOTACHAR;
3292  #endif  #endif
3293      return (unsigned int)c == othercase;        return c == othercase;
3294      }        }
3295    else      else
3296  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3297    return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3298    
3299    /* 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.
3300    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. */
3301    
3302    case OP_DIGIT:      case OP_DIGIT:
3303    return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3304    
3305    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3306    return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3307    
3308    case OP_WHITESPACE:      case OP_WHITESPACE:
3309    return next > 255 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3310    
3311    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3312    return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3313    
3314    case OP_WORDCHAR:      case OP_WORDCHAR:
3315    return next > 255 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3316    
3317    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3318    return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3319    
3320    case OP_HSPACE:      case OP_HSPACE:
3321    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3322    switch(next)      switch(next)
3323      {        {
3324      case 0x09:        HSPACE_CASES:
3325      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;  
     }  
3326    
3327    case OP_ANYNL:        default:
3328    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3329    case OP_NOT_VSPACE:        }
3330    switch(next)  
3331      {      case OP_ANYNL:
3332      case 0x0a:      case OP_VSPACE:
3333      case 0x0b:      case OP_NOT_VSPACE:
3334      case 0x0c:      switch(next)
3335      case 0x0d:        {
3336      case 0x85:        VSPACE_CASES:
3337      case 0x2028:        return op_code == OP_NOT_VSPACE;
3338      case 0x2029:  
3339      return op_code == OP_NOT_VSPACE;        default:
3340      default:        return op_code != OP_NOT_VSPACE;
3341      return op_code != OP_NOT_VSPACE;        }
     }  
3342    
3343  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3344    case OP_PROP:      case OP_PROP:
3345    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3346    
3347    case OP_NOTPROP:      case OP_NOTPROP:
3348    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3349  #endif  #endif
3350    
3351    default:      default:
3352    return FALSE;      return FALSE;
3353        }
3354    }    }
3355    
   
3356  /* 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
3357  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
3358  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 3232  switch(op_code) Line 3363  switch(op_code)
3363    {    {
3364    case OP_CHAR:    case OP_CHAR:
3365    case OP_CHARI:    case OP_CHARI:
3366  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3367      {      {
3368      case ESC_d:      case ESC_d:
3369      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
# Line 3261  switch(op_code) Line 3387  switch(op_code)
3387      case ESC_H:      case ESC_H:
3388      switch(c)      switch(c)
3389        {        {
3390        case 0x09:        HSPACE_CASES:
3391        case 0x20:        return escape != ESC_h;
3392        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;  
3393        default:        default:
3394        return -next == ESC_h;        return escape == ESC_h;
3395        }        }
3396    
3397      case ESC_v:      case ESC_v:
3398      case ESC_V:      case ESC_V:
3399      switch(c)      switch(c)
3400        {        {
3401        case 0x0a:        VSPACE_CASES:
3402        case 0x0b:        return escape != ESC_v;
3403        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3404        default:        default:
3405        return -next == ESC_v;        return escape == ESC_v;
3406        }        }
3407    
3408      /* 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
3409      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3410      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3411    
3412  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3413      case ESC_du:      case ESC_du:
# Line 3314  switch(op_code) Line 3418  switch(op_code)
3418      case ESC_SU:      case ESC_SU:
3419        {        {
3420        int temperrorcode = 0;        int temperrorcode = 0;
3421        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3422        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3423        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3424        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3425        }        }
# Line 3324  switch(op_code) Line 3428  switch(op_code)
3428      case ESC_p:      case ESC_p:
3429      case ESC_P:      case ESC_P:
3430        {        {
3431        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3432          int errorcodeptr;
3433        BOOL negated;        BOOL negated;
3434    
3435        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3436        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3437        if (ptype < 0) return FALSE;          return FALSE;
3438        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3439    
3440        /* 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 3342  switch(op_code) Line 3447  switch(op_code)
3447    
3448        /* Do the property check. */        /* Do the property check. */
3449    
3450        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3451        }        }
3452  #endif  #endif
3453    
# Line 3357  switch(op_code) Line 3462  switch(op_code)
3462    these op-codes are never generated.) */    these op-codes are never generated.) */
3463    
3464    case OP_DIGIT:    case OP_DIGIT:
3465    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3466           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3467    
3468    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3469    return next == -ESC_d;    return escape == ESC_d;
3470    
3471    case OP_WHITESPACE:    case OP_WHITESPACE:
3472    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3473    
3474    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3475    return next == -ESC_s || next == -ESC_h || next == -ESC_v || next == -ESC_R;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3476    
3477    case OP_HSPACE:    case OP_HSPACE:
3478    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3479           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3480    
3481    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3482    return next == -ESC_h;    return escape == ESC_h;
3483    
3484    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3485    case OP_ANYNL:    case OP_ANYNL:
3486    case OP_VSPACE:    case OP_VSPACE:
3487    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3488    
3489    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3490    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3491    
3492    case OP_WORDCHAR:    case OP_WORDCHAR:
3493    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3494           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3495    
3496    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3497    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3498    
3499    default:    default:
3500    return FALSE;    return FALSE;
# Line 3401  switch(op_code) Line 3506  switch(op_code)
3506    
3507    
3508  /*************************************************  /*************************************************
3509    *        Add a character or range to a class     *
3510    *************************************************/
3511    
3512    /* This function packages up the logic of adding a character or range of
3513    characters to a class. The character values in the arguments will be within the
3514    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3515    mutually recursive with the function immediately below.
3516    
3517    Arguments:
3518      classbits     the bit map for characters < 256
3519      uchardptr     points to the pointer for extra data
3520      options       the options word
3521      cd            contains pointers to tables etc.
3522      start         start of range character
3523      end           end of range character
3524    
3525    Returns:        the number of < 256 characters added
3526                    the pointer to extra data is updated
3527    */
3528    
3529    static int
3530    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3531      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3532    {
3533    pcre_uint32 c;
3534    int n8 = 0;
3535    
3536    /* If caseless matching is required, scan the range and process alternate
3537    cases. In Unicode, there are 8-bit characters that have alternate cases that
3538    are greater than 255 and vice-versa. Sometimes we can just extend the original
3539    range. */
3540    
3541    if ((options & PCRE_CASELESS) != 0)
3542      {
3543    #ifdef SUPPORT_UCP
3544      if ((options & PCRE_UTF8) != 0)
3545        {
3546        int rc;
3547        pcre_uint32 oc, od;
3548    
3549        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3550        c = start;
3551    
3552        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3553          {
3554          /* Handle a single character that has more than one other case. */
3555    
3556          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3557            PRIV(ucd_caseless_sets) + rc, oc);
3558    
3559          /* Do nothing if the other case range is within the original range. */
3560    
3561          else if (oc >= start && od <= end) continue;
3562    
3563          /* Extend the original range if there is overlap, noting that if oc < c, we
3564          can't have od > end because a subrange is always shorter than the basic
3565          range. Otherwise, use a recursive call to add the additional range. */
3566    
3567          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3568          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3569          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3570          }
3571        }
3572      else
3573    #endif  /* SUPPORT_UCP */
3574    
3575      /* Not UTF-mode, or no UCP */
3576    
3577      for (c = start; c <= end && c < 256; c++)
3578        {
3579        SETBIT(classbits, cd->fcc[c]);
3580        n8++;
3581        }
3582      }
3583    
3584    /* Now handle the original range. Adjust the final value according to the bit
3585    length - this means that the same lists of (e.g.) horizontal spaces can be used
3586    in all cases. */
3587    
3588    #if defined COMPILE_PCRE8
3589    #ifdef SUPPORT_UTF
3590      if ((options & PCRE_UTF8) == 0)
3591    #endif
3592      if (end > 0xff) end = 0xff;
3593    
3594    #elif defined COMPILE_PCRE16
3595    #ifdef SUPPORT_UTF
3596      if ((options & PCRE_UTF16) == 0)
3597    #endif
3598      if (end > 0xffff) end = 0xffff;
3599    
3600    #endif /* COMPILE_PCRE[8|16] */
3601    
3602    /* If all characters are less than 256, use the bit map. Otherwise use extra
3603    data. */
3604    
3605    if (end < 0x100)
3606      {
3607      for (c = start; c <= end; c++)
3608        {
3609        n8++;
3610        SETBIT(classbits, c);
3611        }
3612      }
3613    
3614    else
3615      {
3616      pcre_uchar *uchardata = *uchardptr;
3617    
3618    #ifdef SUPPORT_UTF
3619      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3620        {
3621        if (start < end)
3622          {
3623          *uchardata++ = XCL_RANGE;
3624          uchardata += PRIV(ord2utf)(start, uchardata);
3625          uchardata += PRIV(ord2utf)(end, uchardata);
3626          }
3627        else if (start == end)
3628          {
3629          *uchardata++ = XCL_SINGLE;
3630          uchardata += PRIV(ord2utf)(start, uchardata);
3631          }
3632        }
3633      else
3634    #endif  /* SUPPORT_UTF */
3635    
3636      /* Without UTF support, character values are constrained by the bit length,
3637      and can only be > 256 for 16-bit and 32-bit libraries. */
3638    
3639    #ifdef COMPILE_PCRE8
3640        {}
3641    #else
3642      if (start < end)
3643        {
3644        *uchardata++ = XCL_RANGE;
3645        *uchardata++ = start;
3646        *uchardata++ = end;
3647        }
3648      else if (start == end)
3649        {
3650        *uchardata++ = XCL_SINGLE;
3651        *uchardata++ = start;
3652        }
3653    #endif
3654    
3655      *uchardptr = uchardata;   /* Updata extra data pointer */
3656      }
3657    
3658    return n8;    /* Number of 8-bit characters */
3659    }
3660    
3661    
3662    
3663    
3664    /*************************************************
3665    *        Add a list of characters to a class     *
3666    *************************************************/
3667    
3668    /* This function is used for adding a list of case-equivalent characters to a
3669    class, and also for adding a list of horizontal or vertical whitespace. If the
3670    list is in order (which it should be), ranges of characters are detected and
3671    handled appropriately. This function is mutually recursive with the function
3672    above.
3673    
3674    Arguments:
3675      classbits     the bit map for characters < 256
3676      uchardptr     points to the pointer for extra data
3677      options       the options word
3678      cd            contains pointers to tables etc.
3679      p             points to row of 32-bit values, terminated by NOTACHAR
3680      except        character to omit; this is used when adding lists of
3681                      case-equivalent characters to avoid including the one we
3682                      already know about
3683    
3684    Returns:        the number of < 256 characters added
3685                    the pointer to extra data is updated
3686    */
3687    
3688    static int
3689    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3690      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3691    {
3692    int n8 = 0;
3693    while (p[0] < NOTACHAR)
3694      {
3695      int n = 0;
3696      if (p[0] != except)
3697        {
3698        while(p[n+1] == p[0] + n + 1) n++;
3699        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3700        }
3701      p += n + 1;
3702      }
3703    return n8;
3704    }
3705    
3706    
3707    
3708    /*************************************************
3709    *    Add characters not in a list to a class     *
3710    *************************************************/
3711    
3712    /* This function is used for adding the complement of a list of horizontal or
3713    vertical whitespace to a class. The list must be in order.
3714    
3715    Arguments:
3716      classbits     the bit map for characters < 256
3717      uchardptr     points to the pointer for extra data
3718      options       the options word
3719      cd            contains pointers to tables etc.
3720      p             points to row of 32-bit values, terminated by NOTACHAR
3721    
3722    Returns:        the number of < 256 characters added
3723                    the pointer to extra data is updated
3724    */
3725    
3726    static int
3727    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3728      int options, compile_data *cd, const pcre_uint32 *p)
3729    {
3730    BOOL utf = (options & PCRE_UTF8) != 0;
3731    int n8 = 0;
3732    if (p[0] > 0)
3733      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3734    while (p[0] < NOTACHAR)
3735      {
3736      while (p[1] == p[0] + 1) p++;
3737      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3738        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3739      p++;
3740      }
3741    return n8;
3742    }
3743    
3744    
3745    
3746    /*************************************************
3747  *           Compile one branch                   *  *           Compile one branch                   *
3748  *************************************************/  *************************************************/
3749    
# Line 3415  Arguments: Line 3758  Arguments:
3758    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3759    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3760    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3761    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3762    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3763      reqcharptr     place to put the last required character
3764      reqcharflagsptr place to put the last required character flags, or a negative number
3765    bcptr          points to current branch chain    bcptr          points to current branch chain
3766    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3767    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3429  Returns:         TRUE on success Line 3774  Returns:         TRUE on success
3774    
3775  static BOOL  static BOOL
3776  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3777    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3778    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3779      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3780      branch_chain *bcptr, int cond_depth,
3781    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3782  {  {
3783  int repeat_type, op_type;  int repeat_type, op_type;
3784  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3785  int bravalue = 0;  int bravalue = 0;
3786  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3787  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
3788  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
3789    pcre_uint32 zeroreqchar, zerofirstchar;
3790    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3791  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3792  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3793  int after_manual_callout = 0;  int after_manual_callout = 0;
3794  int length_prevgroup = 0;  int length_prevgroup = 0;
3795  register int c;  register pcre_uint32 c;
3796    int escape;
3797  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3798  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3799  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3463  must not do this for other options (e.g. Line 3813  must not do this for other options (e.g.
3813  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3814    
3815  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3816  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3817  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3818    #ifndef COMPILE_PCRE32
3819  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3820    #endif
3821  #else  #else
3822  BOOL utf = FALSE;  BOOL utf = FALSE;
3823  #endif  #endif
3824    
3825  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3826    class_uchardata always so that it can be passed to add_to_class() always,
3827    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3828    alternative calls for the different cases. */
3829    
3830    pcre_uchar *class_uchardata;
3831  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3832  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3833  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3834  #endif  #endif
3835    
# Line 3497  to take the zero repeat into account. Th Line 3852  to take the zero repeat into account. Th
3852  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3853  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3854    
3855  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3856    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3857    
3858  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
3859  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 3518  for (;; ptr++) Line 3874  for (;; ptr++)
3874    BOOL is_recurse;    BOOL is_recurse;
3875    BOOL reset_bracount;    BOOL reset_bracount;
3876    int class_has_8bitchar;    int class_has_8bitchar;
3877    int class_single_char;    int class_one_char;
3878    int newoptions;    int newoptions;
3879    int recno;    int recno;
3880    int refsign;    int refsign;
3881    int skipbytes;    int skipbytes;
3882    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
3883    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
3884    int terminator;    int terminator;
3885    int mclength;    unsigned int mclength;
3886    int tempbracount;    unsigned int tempbracount;
3887      pcre_uint32 ec;
3888    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3889    
3890    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3537  for (;; ptr++) Line 3894  for (;; ptr++)
3894    /* If we are at the end of a nested substitution, revert to the outer level    /* If we are at the end of a nested substitution, revert to the outer level
3895    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
3896    
3897    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
3898      {      {
3899      ptr = nestptr;      ptr = nestptr;
3900      nestptr = NULL;      nestptr = NULL;
# Line 3612  for (;; ptr++) Line 3969  for (;; ptr++)
3969    
3970    /* If in \Q...\E, check for the end; if not, we have a literal */    /* If in \Q...\E, check for the end; if not, we have a literal */
3971    
3972    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
3973      {      {
3974      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3975        {        {
# Line 3660  for (;; ptr++) Line 4017  for (;; ptr++)
4017      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
4018        {        {
4019        ptr++;        ptr++;
4020        while (*ptr != 0)        while (*ptr != CHAR_NULL)
4021          {          {
4022          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
4023          ptr++;          ptr++;
# Line 3668  for (;; ptr++) Line 4025  for (;; ptr++)
4025          if (utf) FORWARDCHAR(ptr);          if (utf) FORWARDCHAR(ptr);
4026  #endif  #endif
4027          }          }
4028        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
4029    
4030        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
4031        c = 0;        c = 0;
# Line 3690  for (;; ptr++) Line 4047  for (;; ptr++)
4047      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
4048      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
4049      *firstcharptr = firstchar;      *firstcharptr = firstchar;
4050        *firstcharflagsptr = firstcharflags;
4051      *reqcharptr = reqchar;      *reqcharptr = reqchar;
4052        *reqcharflagsptr = reqcharflags;
4053      *codeptr = code;      *codeptr = code;
4054      *ptrptr = ptr;      *ptrptr = ptr;
4055      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3714  for (;; ptr++) Line 4073  for (;; ptr++)
4073      previous = NULL;      previous = NULL;
4074      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
4075        {        {
4076        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4077        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
4078        }        }
4079      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3729  for (;; ptr++) Line 4088  for (;; ptr++)
4088      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4089    
4090      case CHAR_DOT:      case CHAR_DOT:
4091      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4092      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4093        zerofirstcharflags = firstcharflags;
4094      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4095        zeroreqcharflags = reqcharflags;
4096      previous = code;      previous = code;
4097      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4098      break;      break;
# Line 3805  for (;; ptr++) Line 4166  for (;; ptr++)
4166          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4167        {        {
4168        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4169        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4170        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4171          zerofirstcharflags = firstcharflags;
4172        break;        break;
4173        }        }
4174    
# Line 3816  for (;; ptr++) Line 4178  for (;; ptr++)
4178    
4179      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4180    
4181      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4182      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 <
4183      < 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
4184      a single character. */      character. */
4185    
4186      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4187      class_single_char = 0;      class_one_char = 0;
4188    
4189      /* 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
4190      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
4191      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
4192      */      map. */
4193    
4194      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4195    
4196  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4197      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4198      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4199      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4200  #endif  #endif
4201    
4202      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4203      means that an initial ] is taken as a data character. At the start of the      means that an initial ] is taken as a data character. At the start of the
4204      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4205    
4206      if (c != 0) do      if (c != CHAR_NULL) do
4207        {        {
4208        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4209    
# Line 3856  for (;; ptr++) Line 4218  for (;; ptr++)
4218        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4219        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
4220        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4221        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4222          however. */
4223    
4224        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4225          {          {
4226            xclass = TRUE;
4227          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4228          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4229          }          }
# Line 3961  for (;; ptr++) Line 4325  for (;; ptr++)
4325              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4326            }            }
4327    
4328          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4329          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4330    
4331          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3977  for (;; ptr++) Line 4341  for (;; ptr++)
4341            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4342    
4343          ptr = tempptr + 1;          ptr = tempptr + 1;
4344          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4345          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4346          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4347          class_single_char = 2;          class_one_char = 2;
4348          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4349          }          }
4350    
# Line 3988  for (;; ptr++) Line 4352  for (;; ptr++)
4352        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
4353        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
4354        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4355        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4356        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4357        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4358        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4359    
4360        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4361          {          {
4362          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options,
4363              TRUE);
4364          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4365            if (escape == 0) c = ec;
4366          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4367          else if (-c == ESC_N)            /* \N is not supported in a class */          else if (escape == ESC_N)          /* \N is not supported in a class */
4368            {            {
4369            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4370            goto FAILED;            goto FAILED;
4371            }            }
4372          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4373            {            {
4374            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4375              {              {
# Line 4013  for (;; ptr++) Line 4378  for (;; ptr++)
4378            else inescq = TRUE;            else inescq = TRUE;
4379            continue;            continue;
4380            }            }
4381          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4382    
4383          if (c < 0)          else
4384            {            {
4385            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4386            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4387            class_has_8bitchar++;            class_has_8bitchar++;
4388            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4389            class_single_char += 2;            class_one_char += 2;
4390    
4391            switch (-c)            switch (escape)
4392              {              {
4393  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4394              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 4033  for (;; ptr++) Line 4398  for (;; ptr++)
4398              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4399              case ESC_SU:              case ESC_SU:
4400              nestptr = ptr;              nestptr = ptr;
4401              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4402              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4403              continue;              continue;
4404  #endif  #endif
# Line 4057  for (;; ptr++) Line 4422  for (;; ptr++)
4422    
4423              /* 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
4424              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4425              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4426                EBCDIC, so we lazily just adjust the appropriate bit. */
4427    
4428              case ESC_s:              case ESC_s:
4429              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4071  for (;; ptr++) Line 4437  for (;; ptr++)
4437              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4438              continue;              continue;
4439    
4440              case ESC_h:              /* The rest apply in both UCP and non-UCP cases. */
4441              SETBIT(classbits, 0x09); /* VT */  
4442              SETBIT(classbits, 0x20); /* SPACE */              case ESC_h:
4443              SETBIT(classbits, 0xa0); /* NSBP */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4444  #ifndef COMPILE_PCRE8                PRIV(hspace_list), NOTACHAR);
4445              xclass = TRUE;              continue;
4446              *class_uchardata++ = XCL_SINGLE;  
4447              *class_uchardata++ = 0x1680;              case ESC_H:
4448              *class_uchardata++ = XCL_SINGLE;              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4449              *class_uchardata++ = 0x180e;                cd, PRIV(hspace_list));
             *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  
             continue;  
   
             case ESC_H:  
             for (c = 0; c < 32; c++)  
               {  
               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  
4450              continue;              continue;
4451    
4452              case ESC_v:              case ESC_v:
4453              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4454              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  
4455              continue;              continue;
4456    
4457              case ESC_V:              case ESC_V:
4458              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4459                {                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  
4460              continue;              continue;
4461    
4462  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4252  for (;; ptr++) Line 4464  for (;; ptr++)
4464              case ESC_P:              case ESC_P:
4465                {                {
4466                BOOL negated;                BOOL negated;
4467                int pdata;                unsigned int ptype = 0, pdata = 0;
4468                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4469                if (ptype < 0) goto FAILED;                  goto FAILED;
4470                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4471                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4472                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4473                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4275  for (;; ptr++) Line 4486  for (;; ptr++)
4486                goto FAILED;                goto FAILED;
4487                }                }
4488              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4489              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4490              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4491              break;              break;
4492              }              }
4493            }            }
4494    
4495          /* 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).
4496          greater than 256. */          This may be greater than 256. */
4497    
4498            escape = 0;
4499    
4500          }   /* End of backslash handling */          }   /* End of backslash handling */
4501    
4502        /* 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
4503        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
4504        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4505        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4506    
4507        CHECK_RANGE:        CHECK_RANGE:
4508        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4297  for (;; ptr++) Line 4510  for (;; ptr++)
4510          inescq = FALSE;          inescq = FALSE;
4511          ptr += 2;          ptr += 2;
4512          }          }
   
4513        oldptr = ptr;        oldptr = ptr;
4514    
4515        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4516    
4517        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4518    
# Line 4308  for (;; ptr++) Line 4520  for (;; ptr++)
4520    
4521        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4522          {          {
4523          int d;          pcre_uint32 d;
4524          ptr += 2;          ptr += 2;
4525          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4526    
# Line 4324  for (;; ptr++) Line 4536  for (;; ptr++)
4536            break;            break;
4537            }            }
4538    
4539          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          /* Minus (hyphen) at the end of a class is treated as a literal, so put
4540            back the pointer and jump to handle the character that preceded it. */
4541    
4542            if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4543            {            {
4544            ptr = oldptr;            ptr = oldptr;
4545            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4546            }            }
4547    
4548            /* Otherwise, we have a potential range; pick up the next character */
4549    
4550  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4551          if (utf)          if (utf)
4552            {                           /* Braces are required because the */            {                           /* Braces are required because the */
# Line 4345  for (;; ptr++) Line 4562  for (;; ptr++)
4562    
4563          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4564            {            {
4565            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4566              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4567            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4568    
4569            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4570    
4571            if (d < 0)            if (descape != 0)
4572              {              {
4573              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4574                {                {
4575                ptr = oldptr;                ptr = oldptr;
4576                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4577                }                }
4578              }              }
4579            }            }
4580    
4581          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4582          one-character ranges */          one-character ranges. */
4583    
4584          if (d < c)          if (d < c)
4585            {            {
4586            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4587            goto FAILED;            goto FAILED;
4588            }            }
4589            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4590    
4591          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4592            cannot be done anymore. Any value greater than 1 indicates that there
4593            is more than one character. */
4594    
4595          /* Remember \r or \n */          class_one_char = 2;
4596    
4597            /* Remember an explicit \r or \n, and add the range to the class. */
4598    
4599          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4600    
4601          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4602          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
         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. */  
4603    
4604  #ifdef SUPPORT_UCP          continue;   /* Go get the next char in the class */
4605  #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;  
                 }  
4606    
4607                if (occ == ocd)        /* Handle a single character - we can get here for a normal non-escape
4608                  {        char, or after \ that introduces a single character or for an apparent
4609                  *class_uchardata++ = XCL_SINGLE;        range that isn't. Only the value 1 matters for class_one_char, so don't
4610                  }        increase it if it is already 2 or more ... just in case there's a class
4611                else        with a zillion characters in it. */
4612                  {  
4613                  *class_uchardata++ = XCL_RANGE;        CLASS_SINGLE_CHARACTER:
4614                  class_uchardata += PRIV(ord2utf)(occ, class_uchardata);        if (class_one_char < 2) class_one_char++;
4615                  }  
4616                class_uchardata += PRIV(ord2utf)(ocd, class_uchardata);        /* If class_one_char is 1, we have the first single character in the
4617                }        class, and there have been no prior ranges, or XCLASS items generated by
4618              }        escapes. If this is the final character in the class, we can optimize by
4619  #endif  /* SUPPORT_UCP */        turning the item into a 1-character OP_CHAR[I] if it's positive, or
4620          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4621          to be set. Otherwise, there can be no first char if this item is first,
4622          whatever repeat count may follow. In the case of reqchar, save the
4623          previous value for reinstating. */
4624    
4625            /* Now record the original range, possibly modified for UCP caseless        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
4626            overlapping ranges. */          {
4627            ptr++;
4628            zeroreqchar = reqchar;
4629            zeroreqcharflags = reqcharflags;
4630    
4631            *class_uchardata++ = XCL_RANGE;          if (negate_class)
4632  #ifdef SUPPORT_UTF            {
4633  #ifndef COMPILE_PCRE8  #ifdef SUPPORT_UCP
4634            if (utf)            int d;
             {  
             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);  
4635  #endif  #endif
4636  #else /* SUPPORT_UTF */            if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4637            *class_uchardata++ = c;            zerofirstchar = firstchar;
4638            *class_uchardata++ = d;            zerofirstcharflags = firstcharflags;
 #endif /* SUPPORT_UTF */  
4639    
4640            /* With UCP support, we are done. Without UCP support, there is no            /* For caseless UTF-8 mode when UCP support is available, check
4641            caseless matching for UTF characters > 127; we can use the bit map            whether this character has more than one other case. If so, generate
4642            for the smaller ones. As for 16 bit characters without UTF, we            a special OP_NOTPROP item instead of OP_NOTI. */
           can still use  */  
4643    
4644  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4645  #ifndef COMPILE_PCRE8            if (utf && (options & PCRE_CASELESS) != 0 &&
4646            if (utf)                (d = UCD_CASESET(c)) != 0)
 #endif  
             continue;    /* With next character in the class */  
 #endif  /* SUPPORT_UCP */  
   
 #if defined SUPPORT_UTF && !defined(SUPPORT_UCP) && !(defined COMPILE_PCRE8)  
           if (utf)  
4647              {              {
4648              if ((options & PCRE_CASELESS) == 0 || c > 127) continue;              *code++ = OP_NOTPROP;
4649              /* Adjust upper limit and fall through to set up the map */              *code++ = PT_CLIST;
4650              d = 127;              *code++ = d;
4651              }              }
4652            else            else
4653              {  #endif
4654              if (c > 255) continue;            /* Char has only one other case, or UCP not available */
             /* 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. */  
4655    
         if (lengthptr == NULL) for (; c <= d; c++)  
           {  
           classbits[c/8] |= (1 << (c&7));  
           if ((options & PCRE_CASELESS) != 0)  
4656              {              {
4657              int uc = cd->fcc[c]; /* flip case */              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4658              classbits[uc/8] |= (1 << (uc&7));  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4659                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4660                  code += PRIV(ord2utf)(c, code);
4661                else
4662    #endif
4663                  *code++ = c;
4664              }              }
           }  
   
         continue;   /* Go get the next char in the class */  
         }  
   
       /* Handle a lone single character - we can get here for a normal  
       non-escape char, or after \ that introduces a single character or for an  
       apparent range that isn't. */  
   
       LONE_SINGLE_CHARACTER:  
   
       /* Only the value of 1 matters for class_single_char. */  
4665    
4666        if (class_single_char < 2) class_single_char++;            /* We are finished with this character class */
   
       /* If class_charcount is 1, we saw precisely one character. As long as  
       there was no use of \p or \P, in other words, no use of any XCLASS  
       features, we can optimize.  
   
       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.  
       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. */  
   
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)  
         {  
         ptr++;  
         zeroreqchar = reqchar;  
4667    
4668          if (negate_class)            goto END_CLASS;
           {  
           if (firstchar == REQ_UNSET) firstchar = REQ_NONE;  
           zerofirstchar = firstchar;  
           *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
 #ifdef SUPPORT_UTF  
           if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)  
             code += PRIV(ord2utf)(c, code);  
           else  
 #endif  
             *code++ = c;  
           goto NOT_CHAR;  
4669            }            }
4670    
4671          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4672          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4673    
4674  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4675          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4676            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4677          else          else
# Line 4578  for (;; ptr++) Line 4683  for (;; ptr++)
4683          goto ONE_CHAR;          goto ONE_CHAR;
4684          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4685    
4686        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4687          has been generated. Add this character to the class. */
 #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
       if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))  
 #elif defined SUPPORT_UTF  
       if (utf && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
 #elif !(defined COMPILE_PCRE8)  
       if (c > 255)  
 #endif  
   
 #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)  
         {  
         xclass = TRUE;  
         *class_uchardata++ = XCL_SINGLE;  
 #ifdef SUPPORT_UTF  
 #ifndef COMPILE_PCRE8  
         /* In non 8 bit mode, we can get here even if we are not in UTF mode. */  
         if (!utf)  
           *class_uchardata++ = c;  
         else  
 #endif  
           class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
 #else /* SUPPORT_UTF */  
         *class_uchardata++ = c;  
 #endif /* SUPPORT_UTF */  
   
 #ifdef SUPPORT_UCP  
 #ifdef COMPILE_PCRE8  
         if ((options & PCRE_CASELESS) != 0)  
 #else  
         /* In non 8 bit mode, we can get here even if we are not in UTF mode. */  
         if (utf && (options & PCRE_CASELESS) != 0)  
 #endif  
           {  
           unsigned int othercase;  
           if ((int)(othercase = UCD_OTHERCASE(c)) != c)  
             {  
             *class_uchardata++ = XCL_SINGLE;  
             class_uchardata += PRIV(ord2utf)(othercase, class_uchardata);  
             }  
           }  
 #endif  /* SUPPORT_UCP */  
   
         }  
       else  
 #endif  /* SUPPORT_UTF || COMPILE_PCRE16 */  
4688    
4689        /* Handle a single-byte character */        class_has_8bitchar +=
4690          {          add_to_class(classbits, &class_uchardata, options, cd, c, c);
         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));  
           }  
         }  
4691        }        }
4692    
4693      /* 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.
4694      If we are at the end of an internal nested string, revert to the outer      If we are at the end of an internal nested string, revert to the outer
4695      string. */      string. */
4696    
4697      while (((c = *(++ptr)) != 0 ||      while (((c = *(++ptr)) != CHAR_NULL ||
4698             (nestptr != NULL &&             (nestptr != NULL &&
4699               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != CHAR_NULL)) &&
4700             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
4701    
4702      /* Check for missing terminating ']' */      /* Check for missing terminating ']' */
4703    
4704      if (c == 0)      if (c == CHAR_NULL)
4705        {        {
4706        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
4707        goto FAILED;        goto FAILED;
4708        }        }
4709    
4710        /* We will need an XCLASS if data has been placed in class_uchardata. In
4711        the second phase this is a sufficient test. However, in the pre-compile
4712        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4713        only if the very last character in the class needs XCLASS will it contain
4714        anything at this point. For this reason, xclass gets set TRUE above when
4715        uchar_classdata is emptied, and that's why this code is the way it is here
4716        instead of just doing a test on class_uchardata below. */
4717    
4718    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4719        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4720    #endif
4721    
4722      /* 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
4723      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4724      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
4725    
4726      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4727      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4728        zerofirstcharflags = firstcharflags;
4729      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4730        zeroreqcharflags = reqcharflags;
4731    
4732      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4733      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 4716  for (;; ptr++) Line 4783  for (;; ptr++)
4783        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4784        }        }
4785      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4786      NOT_CHAR:  
4787        END_CLASS:
4788      break;      break;
4789    
4790    
# Line 4754  for (;; ptr++) Line 4822  for (;; ptr++)
4822      if (repeat_min == 0)      if (repeat_min == 0)
4823        {        {
4824        firstchar = zerofirstchar;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4825          firstcharflags = zerofirstcharflags;
4826        reqchar = zeroreqchar;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4827          reqcharflags = zeroreqcharflags;
4828        }        }
4829    
4830      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4840  for (;; ptr++) Line 4910  for (;; ptr++)
4910        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
4911        it's a length rather than a small character. */        it's a length rather than a small character. */
4912    
4913  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4914        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4915          {          {
4916          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4857  for (;; ptr++) Line 4927  for (;; ptr++)
4927          {          {
4928          c = code[-1];          c = code[-1];
4929          if (*previous <= OP_CHARI && repeat_min > 1)          if (*previous <= OP_CHARI && repeat_min > 1)
4930            reqchar = c | req_caseopt | cd->req_varyopt;            {
4931              reqchar = c;
4932              reqcharflags = req_caseopt | cd->req_varyopt;
4933              }
4934          }          }
4935    
4936        /* 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 4914  for (;; ptr++) Line 4987  for (;; ptr++)
4987    
4988        if (repeat_max == 0) goto END_REPEAT;        if (repeat_max == 0) goto END_REPEAT;
4989    
       /*--------------------------------------------------------------------*/  
       /* This code is obsolete from release 8.00; the restriction was finally  
       removed: */  
   
       /* All real repeats make it impossible to handle partial matching (maybe  
       one day we will be able to remove this restriction). */  
   
       /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */  
       /*--------------------------------------------------------------------*/  
   
4990        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
4991    
4992        repeat_type += op_type;        repeat_type += op_type;
# Line 4976  for (;; ptr++) Line 5039  for (;; ptr++)
5039    
5040          if (repeat_max < 0)          if (repeat_max < 0)
5041            {            {
5042  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5043            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
5044              {              {
5045              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5001  for (;; ptr++) Line 5064  for (;; ptr++)
5064    
5065          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
5066            {            {
5067  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5068            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
5069              {              {
5070              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5031  for (;; ptr++) Line 5094  for (;; ptr++)
5094    
5095        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
5096    
5097  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5098        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
5099          {          {
5100          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5070  for (;; ptr++) Line 5133  for (;; ptr++)
5133          goto END_REPEAT;          goto END_REPEAT;
5134          }          }
5135    
       /*--------------------------------------------------------------------*/  
       /* This code is obsolete from release 8.00; the restriction was finally  
       removed: */  
   
       /* All real repeats make it impossible to handle partial matching (maybe  
       one day we will be able to remove this restriction). */  
   
       /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */  
       /*--------------------------------------------------------------------*/  
   
5136        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
5137          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
5138        else if (repeat_min == 1 && repeat_max == -1)        else if (repeat_min == 1 && repeat_max == -1)
# Line 5239  for (;; ptr++) Line 5292  for (;; ptr++)
5292    
5293            else            else
5294              {              {
5295              if (groupsetfirstchar && reqchar < 0) reqchar = firstchar;              if (groupsetfirstchar && reqcharflags < 0)
5296                  {
5297                  reqchar = firstchar;
5298                  reqcharflags = firstcharflags;
5299                  }
5300    
5301              for (i = 1; i < repeat_min; i++)              for (i = 1; i < repeat_min; i++)
5302                {                {
# Line 5418  for (;; ptr++) Line 5475  for (;; ptr++)
5475              pcre_uchar *scode = bracode;              pcre_uchar *scode = bracode;
5476              do              do
5477                {                {
5478                if (could_be_empty_branch(scode, ketcode, utf, cd))                if (could_be_empty_branch(scode, ketcode, utf, cd, NULL))
5479                  {                  {
5480                  *bracode += OP_SBRA - OP_BRA;                  *bracode += OP_SBRA - OP_BRA;
5481                  break;                  break;
# Line 5614  for (;; ptr++) Line 5671  for (;; ptr++)
5671        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
5672          {          {
5673          arg = ++ptr;          arg = ++ptr;
5674          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5675          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5676          if (arglen > (int)MAX_MARK)          if ((unsigned int)arglen > MAX_MARK)
5677            {            {
5678            *errorcodeptr = ERR75;            *errorcodeptr = ERR75;
5679            goto FAILED;            goto FAILED;
# Line 5636  for (;; ptr++) Line 5693  for (;; ptr++)
5693          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5694              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5695            {            {
5696              int setverb;
5697    
5698            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5699            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5700    
# Line 5653  for (;; ptr++) Line 5712  for (;; ptr++)
5712                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5713                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5714                }                }
5715              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5716                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5717    
5718              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5719              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
5720              }              }
5721    
5722            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5668  for (;; ptr++) Line 5728  for (;; ptr++)
5728                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5729                goto FAILED;                goto FAILED;
5730                }                }
5731              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5732              }              }
5733    
5734            else            else
# Line 5679  for (;; ptr++) Line 5738  for (;; ptr++)
5738                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5739                goto FAILED;                goto FAILED;
5740                }                }
5741              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5742              *code++ = arglen;              *code++ = arglen;
5743              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5744              code += arglen;              code += arglen;
5745              *code++ = 0;              *code++ = 0;
5746              }              }
5747    
5748              switch (setverb)
5749                {
5750                case OP_THEN:
5751                case OP_THEN_ARG:
5752                cd->external_flags |= PCRE_HASTHEN;
5753                break;
5754    
5755                case OP_PRUNE:
5756                case OP_PRUNE_ARG:
5757                case OP_SKIP:
5758                case OP_SKIP_ARG:
5759                cd->had_pruneorskip = TRUE;
5760                break;
5761                }
5762    
5763            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5764            }            }
5765    
# Line 5712  for (;; ptr++) Line 5785  for (;; ptr++)
5785          {          {
5786          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */
5787          ptr++;          ptr++;
5788          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5789          if (*ptr == 0)          if (*ptr == CHAR_NULL)
5790            {            {
5791            *errorcodeptr = ERR18;            *errorcodeptr = ERR18;
5792            goto FAILED;            goto FAILED;
# Line 5736  for (;; ptr++) Line 5809  for (;; ptr++)
5809          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5810          case CHAR_LEFT_PARENTHESIS:          case CHAR_LEFT_PARENTHESIS:
5811          bravalue = OP_COND;       /* Conditional group */          bravalue = OP_COND;       /* Conditional group */
5812            tempptr = ptr;
5813    
5814          /* A condition can be an assertion, a number (referring to a numbered          /* A condition can be an assertion, a number (referring to a numbered
5815          group), a name (referring to a named group), or 'R', referring to          group), a name (referring to a named group), or 'R', referring to
# Line 5748  for (;; ptr++) Line 5822  for (;; ptr++)
5822          be the recursive thing or the name 'R' (and similarly for 'R' followed          be the recursive thing or the name 'R' (and similarly for 'R' followed
5823          by digits), and (b) a number could be a name that consists of digits.          by digits), and (b) a number could be a name that consists of digits.
5824          In both cases, we look for a name first; if not found, we try the other          In both cases, we look for a name first; if not found, we try the other
5825          cases. */          cases.
5826    
5827            For compatibility with auto-callouts, we allow a callout to be
5828            specified before a condition that is an assertion. First, check for the
5829            syntax of a callout; if found, adjust the temporary pointer that is
5830            used to check for an assertion condition. That's all that is needed! */
5831    
5832            if (ptr[1] == CHAR_QUESTION_MARK && ptr[2] == CHAR_C)
5833              {
5834              for (i = 3;; i++) if (!IS_DIGIT(ptr[i])) break;
5835              if (ptr[i] == CHAR_RIGHT_PARENTHESIS)
5836                tempptr += i + 1;
5837              }
5838    
5839          /* For conditions that are assertions, check the syntax, and then exit          /* For conditions that are assertions, check the syntax, and then exit
5840          the switch. This will take control down to where bracketed groups,          the switch. This will take control down to where bracketed groups,
5841          including assertions, are processed. */          including assertions, are processed. */
5842    
5843          if (ptr[1] == CHAR_QUESTION_MARK && (ptr[2] == CHAR_EQUALS_SIGN ||          if (tempptr[1] == CHAR_QUESTION_MARK &&
5844              ptr[2] == CHAR_EXCLAMATION_MARK || ptr[2] == CHAR_LESS_THAN_SIGN))                (tempptr[2] == CHAR_EQUALS_SIGN ||
5845                   tempptr[2] == CHAR_EXCLAMATION_MARK ||
5846                   tempptr[2] == CHAR_LESS_THAN_SIGN))
5847            break;            break;
5848    
5849          /* Most other conditions use OP_CREF (a couple change to OP_RREF          /* Most other conditions use OP_CREF (a couple change to OP_RREF
# Line 5789  for (;; ptr++) Line 5877  for (;; ptr++)
5877            }            }
5878          else          else
5879            {            {
5880            terminator = 0;            terminator = CHAR_NULL;
5881            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);
5882            }            }
5883    
# Line 5809  for (;; ptr++) Line 5897  for (;; ptr++)
5897          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
5898            {            {
5899            if (recno >= 0)            if (recno >= 0)
5900              recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1;              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;
5901            ptr++;            ptr++;
5902            }            }
5903          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
5904    
5905          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||
5906              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
5907            {            {
5908            ptr--;      /* Error offset */            ptr--;      /* Error offset */
# Line 5879  for (;; ptr++) Line 5967  for (;; ptr++)
5967            code[1+LINK_SIZE]++;            code[1+LINK_SIZE]++;
5968            }            }
5969    
5970          /* If terminator == 0 it means that the name followed directly after          /* If terminator == CHAR_NULL it means that the name followed directly
5971          the opening parenthesis [e.g. (?(abc)...] and in this case there are          after the opening parenthesis [e.g. (?(abc)...] and in this case there
5972          some further alternatives to try. For the cases where terminator != 0          are some further alternatives to try. For the cases where terminator !=
5973          [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have          0 [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have
5974          now checked all the possibilities, so give an error. */          now checked all the possibilities, so give an error. */
5975    
5976          else if (terminator != 0)          else if (terminator != CHAR_NULL)
5977            {            {
5978            *errorcodeptr = ERR15;            *errorcodeptr = ERR15;
5979            goto FAILED;            goto FAILED;
# Line 6054  for (;; ptr++) Line 6142  for (;; ptr++)
6142    
6143            if (lengthptr != NULL)            if (lengthptr != NULL)
6144              {              {
6145              if (*ptr != terminator)              if (*ptr != (pcre_uchar)terminator)
6146                {                {
6147                *errorcodeptr = ERR42;                *errorcodeptr = ERR42;
6148                goto FAILED;                goto FAILED;
# Line 6196  for (;; ptr++) Line 6284  for (;; ptr++)
6284              *errorcodeptr = ERR62;              *errorcodeptr = ERR62;
6285              goto FAILED;              goto FAILED;
6286              }              }
6287            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6288              {              {
6289              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
6290              goto FAILED;              goto FAILED;
# Line 6302  for (;; ptr++) Line 6390  for (;; ptr++)
6390            while(IS_DIGIT(*ptr))            while(IS_DIGIT(*ptr))
6391              recno = recno * 10 + *ptr++ - CHAR_0;              recno = recno * 10 + *ptr++ - CHAR_0;
6392    
6393            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6394              {              {
6395              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
6396              goto FAILED;              goto FAILED;
# Line 6406  for (;; ptr++) Line 6494  for (;; ptr++)
6494    
6495          /* Can't determine a first byte now */          /* Can't determine a first byte now */
6496    
6497          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6498          continue;          continue;
6499    
6500    
# Line 6540  for (;; ptr++) Line 6628  for (;; ptr++)
6628           cond_depth +           cond_depth +
6629             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
6630           &subfirstchar,                   /* For possible first char */           &subfirstchar,                   /* For possible first char */
6631             &subfirstcharflags,
6632           &subreqchar,                     /* For possible last char */           &subreqchar,                     /* For possible last char */
6633             &subreqcharflags,
6634           bcptr,                           /* Current branch chain */           bcptr,                           /* Current branch chain */
6635           cd,                              /* Tables block */           cd,                              /* Tables block */
6636           (lengthptr == NULL)? NULL :      /* Actual compile phase */           (lengthptr == NULL)? NULL :      /* Actual compile phase */
# Line 6601  for (;; ptr++) Line 6691  for (;; ptr++)
6691            *errorcodeptr = ERR27;            *errorcodeptr = ERR27;
6692            goto FAILED;            goto FAILED;
6693            }            }
6694          if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;          if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE;
6695          }          }
6696        }        }
6697    
# Line 6650  for (;; ptr++) Line 6740  for (;; ptr++)
6740      back off. */      back off. */
6741    
6742      zeroreqchar = reqchar;      zeroreqchar = reqchar;
6743        zeroreqcharflags = reqcharflags;
6744      zerofirstchar = firstchar;      zerofirstchar = firstchar;
6745        zerofirstcharflags = firstcharflags;
6746      groupsetfirstchar = FALSE;      groupsetfirstchar = FALSE;
6747    
6748      if (bravalue >= OP_ONCE)      if (bravalue >= OP_ONCE)
# Line 6661  for (;; ptr++) Line 6753  for (;; ptr++)
6753        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
6754        repeat forces firstchar to "none". */        repeat forces firstchar to "none". */
6755    
6756        if (firstchar == REQ_UNSET)        if (firstcharflags == REQ_UNSET)
6757          {          {
6758          if (subfirstchar >= 0)          if (subfirstcharflags >= 0)
6759            {            {
6760            firstchar = subfirstchar;            firstchar = subfirstchar;
6761              firstcharflags = subfirstcharflags;
6762            groupsetfirstchar = TRUE;            groupsetfirstchar = TRUE;
6763            }            }
6764          else firstchar = REQ_NONE;          else firstcharflags = REQ_NONE;
6765          zerofirstchar = REQ_NONE;          zerofirstcharflags = REQ_NONE;
6766          }          }
6767    
6768        /* If firstchar was previously set, convert the subpattern's firstchar        /* If firstchar was previously set, convert the subpattern's firstchar
6769        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
6770        existence beforehand. */        existence beforehand. */
6771    
6772        else if (subfirstchar >= 0 && subreqchar < 0)        else if (subfirstcharflags >= 0 && subreqcharflags < 0)
6773          subreqchar = subfirstchar | tempreqvary;          {
6774            subreqchar = subfirstchar;
6775            subreqcharflags = subfirstcharflags | tempreqvary;
6776            }
6777    
6778        /* 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
6779        really the first byte - see above), set it. */        really the first byte - see above), set it. */
6780    
6781        if (subreqchar >= 0) reqchar = subreqchar;        if (subreqcharflags >= 0)
6782            {
6783            reqchar = subreqchar;
6784            reqcharflags = subreqcharflags;
6785            }
6786        }        }
6787    
6788      /* 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 6693  for (;; ptr++) Line 6793  for (;; ptr++)
6793      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
6794      firstchar, looking for an asserted first char. */      firstchar, looking for an asserted first char. */
6795    
6796      else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;      else if (bravalue == OP_ASSERT && subreqcharflags >= 0)
6797          {
6798          reqchar = subreqchar;
6799          reqcharflags = subreqcharflags;
6800          }
6801      break;     /* End of processing '(' */      break;     /* End of processing '(' */
6802    
6803    
# Line 6701  for (;; ptr++) Line 6805  for (;; ptr++)
6805      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
6806      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
6807      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
6808      are ESC_REF plus the reference number. Only back references and those types      are negative the reference number. Only back references and those types
6809      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
6810      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
6811      ever created. */      ever created. */
6812    
6813      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6814      tempptr = ptr;      tempptr = ptr;
6815      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6816      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6817    
6818      if (c < 0)      if (escape == 0)                  /* The escape coded a single character */
6819          c = ec;
6820        else
6821        {        {
6822        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6823          {          {
6824          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6825            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6721  for (;; ptr++) Line 6827  for (;; ptr++)
6827          continue;          continue;
6828          }          }
6829    
6830        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6831    
6832        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6833        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6834    
6835        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6836          firstchar = REQ_NONE;          firstcharflags = REQ_NONE;
6837    
6838        /* 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. */
6839    
6840        zerofirstchar = firstchar;        zerofirstchar = firstchar;
6841          zerofirstcharflags = firstcharflags;
6842        zeroreqchar = reqchar;        zeroreqchar = reqchar;
6843          zeroreqcharflags = reqcharflags;
6844    
6845        /* \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'
6846        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6847        -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 <
6848        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
6849        -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
6850        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6851    
6852        if (-c == ESC_g)        if (escape == ESC_g)
6853          {          {
6854          const pcre_uchar *p;          const pcre_uchar *p;
6855          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
# Line 6761  for (;; ptr++) Line 6869  for (;; ptr++)
6869          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
6870            {            {
6871            BOOL is_a_number = TRUE;            BOOL is_a_number = TRUE;
6872            for (p = ptr + 1; *p != 0 && *p != terminator; p++)            for (p = ptr + 1; *p != CHAR_NULL && *p != (pcre_uchar)terminator; p++)
6873              {              {
6874              if (!MAX_255(*p)) { is_a_number = FALSE; break; }              if (!MAX_255(*p)) { is_a_number = FALSE; break; }
6875              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;
6876              if ((cd->ctypes[*p] & ctype_word) == 0) break;              if ((cd->ctypes[*p] & ctype_word) == 0) break;
6877              }              }
6878            if (*p != terminator)            if (*p != (pcre_uchar)terminator)
6879              {              {
6880              *errorcodeptr = ERR57;              *errorcodeptr = ERR57;
6881              break;              break;
# Line 6785  for (;; ptr++) Line 6893  for (;; ptr++)
6893    
6894          p = ptr + 2;          p = ptr + 2;
6895          while (IS_DIGIT(*p)) p++;          while (IS_DIGIT(*p)) p++;
6896          if (*p != terminator)          if (*p != (pcre_uchar)terminator)
6897            {            {
6898            *errorcodeptr = ERR57;            *errorcodeptr = ERR57;
6899            break;            break;
# Line 6797  for (;; ptr++) Line 6905  for (;; ptr++)
6905        /* \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).
6906        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6907    
6908        if (-c == ESC_k)        if (escape == ESC_k)
6909          {          {
6910          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6911            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))
# Line 6816  for (;; ptr++) Line 6924  for (;; ptr++)
6924        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
6925        ':' later. */        ':' later. */
6926    
6927        if (-c >= ESC_REF)        if (escape < 0)
6928          {          {
6929          open_capitem *oc;          open_capitem *oc;
6930          recno = -c - ESC_REF;          recno = -escape;
6931    
6932          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6933          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6934          previous = code;          previous = code;
6935          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6936          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
# Line 6846  for (;; ptr++) Line 6954  for (;; ptr++)
6954        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6955    
6956  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6957        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6958          {          {
6959          BOOL negated;          BOOL negated;
6960          int pdata;          unsigned int ptype = 0, pdata = 0;
6961          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
6962          if (ptype < 0) goto FAILED;            goto FAILED;
6963          previous = code;          previous = code;
6964          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6965          *code++ = ptype;          *code++ = ptype;
6966          *code++ = pdata;          *code++ = pdata;
6967          }          }
# Line 6862  for (;; ptr++) Line 6970  for (;; ptr++)
6970        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6971        allowed. */        allowed. */
6972    
6973        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6974          {          {
6975          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6976          goto FAILED;          goto FAILED;
# Line 6873  for (;; ptr++) Line 6981  for (;; ptr++)
6981        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
6982        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
6983        Unicode property tests. Note that \b and \B do a one-character        Unicode property tests. Note that \b and \B do a one-character
6984        lookbehind. */        lookbehind, and \A also behaves as if it does. */
6985    
6986        else        else
6987          {          {
6988          if ((-c == ESC_b || -c == ESC_B) && cd->max_lookbehind == 0)          if ((escape == ESC_b || escape == ESC_B || escape == ESC_A) &&
6989                 cd->max_lookbehind == 0)
6990            cd->max_lookbehind = 1;            cd->max_lookbehind = 1;
6991  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6992          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6993            {            {
6994            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6995            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6996            }            }
6997          else          else
6998  #endif  #endif
# Line 6891  for (;; ptr++) Line 7000  for (;; ptr++)
7000          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
7001    
7002            {            {
7003            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
7004            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
7005            }            }
7006          }          }
7007        continue;        continue;
# Line 6902  for (;; ptr++) Line 7011  for (;; ptr++)
7011      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
7012      handle it as a data character. */      handle it as a data character. */
7013    
7014  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
7015      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
7016        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
7017      else      else
# Line 6935  for (;; ptr++) Line 7044  for (;; ptr++)
7044    
7045      ONE_CHAR:      ONE_CHAR:
7046      previous = code;      previous = code;
7047    
7048        /* For caseless UTF-8 mode when UCP support is available, check whether
7049        this character has more than one other case. If so, generate a special
7050        OP_PROP item instead of OP_CHARI. */
7051    
7052    #ifdef SUPPORT_UCP
7053        if (utf && (options & PCRE_CASELESS) != 0)
7054          {
7055          GETCHAR(c, mcbuffer);
7056          if ((c = UCD_CASESET(c)) != 0)
7057            {
7058            *code++ = OP_PROP;
7059            *code++ = PT_CLIST;
7060            *code++ = c;
7061            if (firstcharflags == REQ_UNSET) firstcharflags = zerofirstcharflags = REQ_NONE;
7062            break;
7063            }
7064          }
7065    #endif
7066    
7067        /* Caseful matches, or not one of the multicase characters. */
7068    
7069      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
7070      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
7071    
# Line 6948  for (;; ptr++) Line 7079  for (;; ptr++)
7079      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
7080      repeat. */      repeat. */
7081    
7082      if (firstchar == REQ_UNSET)      if (firstcharflags == REQ_UNSET)
7083        {        {
7084        zerofirstchar = REQ_NONE;        zerofirstcharflags = REQ_NONE;
7085        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7086          zeroreqcharflags = reqcharflags;
7087    
7088        /* 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
7089        only if it is not to be matched caselessly. */        only if it is not to be matched caselessly. */
# Line 6959  for (;; ptr++) Line 7091  for (;; ptr++)
7091        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7092          {          {
7093          firstchar = mcbuffer[0] | req_caseopt;          firstchar = mcbuffer[0] | req_caseopt;
7094          if (mclength != 1) reqchar = code[-1] | cd->req_varyopt;          firstchar = mcbuffer[0];
7095            firstcharflags = req_caseopt;
7096    
7097            if (mclength != 1)
7098              {
7099              reqchar = code[-1];
7100              reqcharflags = cd->req_varyopt;
7101              }
7102          }          }
7103        else firstchar = reqchar = REQ_NONE;        else firstcharflags = reqcharflags = REQ_NONE;
7104        }        }
7105    
7106      /* 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 6970  for (;; ptr++) Line 7109  for (;; ptr++)
7109      else      else
7110        {        {
7111        zerofirstchar = firstchar;        zerofirstchar = firstchar;
7112          zerofirstcharflags = firstcharflags;
7113        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7114          zeroreqcharflags = reqcharflags;
7115        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7116          reqchar = code[-1] | req_caseopt | cd->req_varyopt;          {
7117            reqchar = code[-1];
7118            reqcharflags = req_caseopt | cd->req_varyopt;
7119            }
7120        }        }
7121    
7122      break;            /* End of literal character handling */      break;            /* End of literal character handling */
# Line 6991  return FALSE; Line 7135  return FALSE;
7135    
7136    
7137    
   
7138  /*************************************************  /*************************************************
7139  *     Compile sequence of alternatives           *  *     Compile sequence of alternatives           *
7140  *************************************************/  *************************************************/
# Line 7012  Arguments: Line 7155  Arguments:
7155    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
7156    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
7157    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
7158    firstcharptr   place to put the first required character, or a negative number    firstcharptr    place to put the first required character
7159    reqcharptr     place to put the last required character, or a negative number    firstcharflagsptr place to put the first character flags, or a negative number
7160      reqcharptr     place to put the last required character
7161      reqcharflagsptr place to put the last required character flags, or a negative number
7162    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
7163    cd             points to the data block with tables pointers etc.    cd             points to the data block with tables pointers etc.
7164    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
# Line 7025  Returns:         TRUE on success Line 7170  Returns:         TRUE on success
7170  static BOOL  static BOOL
7171  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,
7172    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
7173    int cond_depth, pcre_int32 *firstcharptr, pcre_int32 *reqcharptr,    int cond_depth,
7174      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
7175      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
7176    branch_chain *bcptr, compile_data *cd, int *lengthptr)    branch_chain *bcptr, compile_data *cd, int *lengthptr)
7177  {  {
7178  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
# Line 7035  pcre_uchar *start_bracket = code; Line 7182  pcre_uchar *start_bracket = code;
7182  pcre_uchar *reverse_count = NULL;  pcre_uchar *reverse_count = NULL;
7183  open_capitem capitem;  open_capitem capitem;
7184  int capnumber = 0;  int capnumber = 0;
7185  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7186  pcre_int32 branchfirstchar, branchreqchar;  pcre_int32 firstcharflags, reqcharflags;
7187    pcre_uint32 branchfirstchar, branchreqchar;
7188    pcre_int32 branchfirstcharflags, branchreqcharflags;
7189  int length;  int length;
7190  int orig_bracount;  unsigned int orig_bracount;
7191  int max_bracount;  unsigned int max_bracount;
7192  branch_chain bc;  branch_chain bc;
7193    
7194  bc.outer = bcptr;  bc.outer = bcptr;
7195  bc.current_branch = code;  bc.current_branch = code;
7196    
7197  firstchar = reqchar = REQ_UNSET;  firstchar = reqchar = 0;
7198    firstcharflags = reqcharflags = REQ_UNSET;
7199    
7200  /* 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
7201  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 7105  for (;;) Line 7255  for (;;)
7255    into the length. */    into the length. */
7256    
7257    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,
7258          &branchreqchar, &bc, cond_depth, cd,          &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc,
7259          (lengthptr == NULL)? NULL : &length))          cond_depth, cd, (lengthptr == NULL)? NULL : &length))
7260      {      {
7261      *ptrptr = ptr;      *ptrptr = ptr;
7262      return FALSE;      return FALSE;
# Line 7127  for (;;) Line 7277  for (;;)
7277      if (*last_branch != OP_ALT)      if (*last_branch != OP_ALT)
7278        {        {
7279        firstchar = branchfirstchar;        firstchar = branchfirstchar;
7280          firstcharflags = branchfirstcharflags;
7281        reqchar = branchreqchar;        reqchar = branchreqchar;
7282          reqcharflags = branchreqcharflags;
7283        }        }
7284    
7285      /* 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 7141  for (;;) Line 7293  for (;;)
7293        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
7294        previously no reqchar, it takes on the value of the old firstchar. */        previously no reqchar, it takes on the value of the old firstchar. */
7295    
7296        if (firstchar >= 0 && firstchar != branchfirstchar)        if (firstcharflags >= 0 &&
7297              (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar))
7298          {          {
7299          if (reqchar < 0) reqchar = firstchar;          if (reqcharflags < 0)
7300          firstchar = REQ_NONE;            {
7301              reqchar = firstchar;
7302              reqcharflags = firstcharflags;
7303              }
7304            firstcharflags = REQ_NONE;
7305          }          }
7306    
7307        /* 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
7308        branch becomes a reqchar if there isn't a branch reqchar. */        branch becomes a reqchar if there isn't a branch reqchar. */
7309    
7310        if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0)        if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0)
7311            branchreqchar = branchfirstchar;          {
7312            branchreqchar = branchfirstchar;
7313            branchreqcharflags = branchfirstcharflags;
7314            }
7315    
7316        /* Now ensure that the reqchars match */        /* Now ensure that the reqchars match */
7317    
7318        if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY))        if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) ||
7319          reqchar = REQ_NONE;            reqchar != branchreqchar)
7320        else reqchar |= branchreqchar;   /* To "or" REQ_VARY */          reqcharflags = REQ_NONE;
7321          else
7322            {
7323            reqchar = branchreqchar;
7324            reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */
7325            }
7326        }        }
7327    
7328      /* 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 7253  for (;;) Line 7418  for (;;)
7418      *codeptr = code;      *codeptr = code;
7419      *ptrptr = ptr;      *ptrptr = ptr;
7420      *firstcharptr = firstchar;      *firstcharptr = firstchar;
7421        *firstcharflagsptr = firstcharflags;
7422      *reqcharptr = reqchar;      *reqcharptr = reqchar;
7423        *reqcharflagsptr = reqcharflags;
7424      if (lengthptr != NULL)      if (lengthptr != NULL)
7425        {        {
7426        if (OFLOW_MAX - *lengthptr < length)        if (OFLOW_MAX - *lengthptr < length)
# Line 7323  and the highest back reference was great Line 7490  and the highest back reference was great
7490  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
7491  of the more common cases more precisely.  of the more common cases more precisely.
7492    
7493    ... A second exception is when the .* appears inside an atomic group, because
7494    this prevents the number of characters it matches from being adjusted.
7495    
7496  Arguments:  Arguments:
7497    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7498    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
7499                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7500                    the less precise approach                    the less precise approach
7501    backref_map    the back reference bitmap    cd             points to the compile data block
7502      atomcount      atomic group level
7503    
7504  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7505  */  */
7506    
7507  static BOOL  static BOOL
7508  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7509    unsigned int backref_map)    compile_data *cd, int atomcount)
7510  {  {
7511  do {  do {
7512     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7347  do { Line 7518  do {
7518     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7519         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7520       {       {
7521       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7522       }       }
7523    
7524     /* Capturing brackets */     /* Capturing brackets */
# Line 7357  do { Line 7528  do {