/[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 905 by zherczeg, Mon Jan 23 19:26:03 2012 UTC revision 1266 by ph10, Sun Mar 3 11:14:26 2013 UTC
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  is also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. We do not need to select pcre16_printint.c specially, because the  library. We do not need to select pcre16_printint.c specially, because the
59  COMPILE_PCREx macro will already be appropriately set. */  COMPILE_PCREx macro will already be appropriately set. */
# Line 68  COMPILE_PCREx macro will already be appr Line 68  COMPILE_PCREx macro will already be appr
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 77  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, 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 489  static const char error_texts[] = Line 504  static const char error_texts[] =
504    "too many forward references\0"    "too many forward references\0"
505    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
506    "invalid UTF-16 string\0"    "invalid UTF-16 string\0"
507      /* 75 */
508      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
509      "character value in \\u.... sequence is too large\0"
510      "invalid UTF-32 string\0"
511    ;    ;
512    
513  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 628  static const pcre_uint8 ebcdic_chartab[] Line 647  static const pcre_uint8 ebcdic_chartab[]
647  #endif  #endif
648    
649    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
   
650    
651  /*************************************************  /*************************************************
652  *            Find an error text                  *  *            Find an error text                  *
# Line 655  find_error_text(int n) Line 667  find_error_text(int n)
667  const char *s = error_texts;  const char *s = error_texts;
668  for (; n > 0; n--)  for (; n > 0; n--)
669    {    {
670    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
671    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
672    }    }
673  return s;  return s;
674  }  }
# Line 739  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 751  return (*p == CHAR_RIGHT_CURLY_BRACKET);
751  *************************************************/  *************************************************/
752    
753  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
754  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
755  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
756  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
757  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
758  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
759  sequence.  escape sequence.
760    
761  Arguments:  Arguments:
762    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
763      chptr          points to the data character
764    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
765    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
766    options        the options bits    options        the options bits
767    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
768    
769  Returns:         zero or positive => a data character  Returns:         zero => a data character
770                   negative => a special escape sequence                   positive => a special escape sequence
771                     negative => a back reference
772                   on error, errorcodeptr is set                   on error, errorcodeptr is set
773  */  */
774    
775  static int  static int
776  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
777    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
778  {  {
779  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
780  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
781  const pcre_uchar *ptr = *ptrptr + 1;  const pcre_uchar *ptr = *ptrptr + 1;
782  pcre_int32 c;  pcre_uint32 c;
783    int escape = 0;
784  int i;  int i;
785    
786  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
# Line 773  ptr--;                            /* Set Line 788  ptr--;                            /* Set
788    
789  /* 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. */
790    
791  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
792    
793  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
794  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 782  Otherwise further processing may be requ Line 797  Otherwise further processing may be requ
797  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
798  /* Not alphanumeric */  /* Not alphanumeric */
799  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
800  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0)
801      { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
802    
803  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
804  /* Not alphanumeric */  /* Not alphanumeric */
805  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)) {}
806  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; }
807  #endif  #endif
808    
809  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
# Line 795  else if ((i = escapes[c - 0x48]) != 0) Line 811  else if ((i = escapes[c - 0x48]) != 0)
811  else  else
812    {    {
813    const pcre_uchar *oldptr;    const pcre_uchar *oldptr;
814    BOOL braced, negated;    BOOL braced, negated, overflow;
815      int s;
816    
817    switch (c)    switch (c)
818      {      {
# Line 820  else Line 837  else
837          c = 0;          c = 0;
838          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
839            {            {
840            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
841  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
842            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
843            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 829  else Line 846  else
846            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
847  #endif  #endif
848            }            }
849    
850    #if defined COMPILE_PCRE8
851            if (c > (utf ? 0x10ffffU : 0xffU))
852    #elif defined COMPILE_PCRE16
853            if (c > (utf ? 0x10ffffU : 0xffffU))
854    #elif defined COMPILE_PCRE32
855            if (utf && c > 0x10ffffU)
856    #endif
857              {
858              *errorcodeptr = ERR76;
859              }
860            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
861          }          }
862        }        }
863      else      else
# Line 855  else Line 884  else
884      (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
885      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
886      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
887      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
888    
889      case CHAR_g:      case CHAR_g:
890      if (isclass) break;      if (isclass) break;
891      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
892        {        {
893        c = -ESC_g;        escape = ESC_g;
894        break;        break;
895        }        }
896    
# Line 870  else Line 899  else
899      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
900        {        {
901        const pcre_uchar *p;        const pcre_uchar *p;
902        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++)
903          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
904        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
905          {          {
906          c = -ESC_k;          escape = ESC_k;
907          break;          break;
908          }          }
909        braced = TRUE;        braced = TRUE;
# Line 890  else Line 919  else
919      else negated = FALSE;      else negated = FALSE;
920    
921      /* The integer range is limited by the machine's int representation. */      /* The integer range is limited by the machine's int representation. */
922      c = 0;      s = 0;
923        overflow = FALSE;
924      while (IS_DIGIT(ptr[1]))      while (IS_DIGIT(ptr[1]))
925        {        {
926        if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */        if (s > INT_MAX / 10 - 1) /* Integer overflow */
927          {          {
928          c = -1;          overflow = TRUE;
929          break;          break;
930          }          }
931        c = c * 10 + *(++ptr) - CHAR_0;        s = s * 10 + (int)(*(++ptr) - CHAR_0);
932        }        }
933      if (((unsigned int)c) > INT_MAX) /* Integer overflow */      if (overflow) /* Integer overflow */
934        {        {
935        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
936          ptr++;          ptr++;
# Line 914  else Line 944  else
944        break;        break;
945        }        }
946    
947      if (c == 0)      if (s == 0)
948        {        {
949        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
950        break;        break;
# Line 922  else Line 952  else
952    
953      if (negated)      if (negated)
954        {        {
955        if (c > bracount)        if (s > bracount)
956          {          {
957          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
958          break;          break;
959          }          }
960        c = bracount - (c - 1);        s = bracount - (s - 1);
961        }        }
962    
963      c = -(ESC_REF + c);      escape = -s;
964      break;      break;
965    
966      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 952  else Line 982  else
982        {        {
983        oldptr = ptr;        oldptr = ptr;
984        /* The integer range is limited by the machine's int representation. */        /* The integer range is limited by the machine's int representation. */
985        c -= CHAR_0;        s = (int)(c -CHAR_0);
986          overflow = FALSE;
987        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
988          {          {
989          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */          if (s > INT_MAX / 10 - 1) /* Integer overflow */
990            {            {
991            c = -1;            overflow = TRUE;
992            break;            break;
993            }            }
994          c = c * 10 + *(++ptr) - CHAR_0;          s = s * 10 + (int)(*(++ptr) - CHAR_0);
995          }          }
996        if (((unsigned int)c) > INT_MAX) /* Integer overflow */        if (overflow) /* Integer overflow */
997          {          {
998          while (IS_DIGIT(ptr[1]))          while (IS_DIGIT(ptr[1]))
999            ptr++;            ptr++;
1000          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1001          break;          break;
1002          }          }
1003        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1004          {          {
1005          c = -(ESC_REF + c);          escape = -s;
1006          break;          break;
1007          }          }
1008        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 1018  else Line 1049  else
1049          c = 0;          c = 0;
1050          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1051            {            {
1052            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1053  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1054            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1055            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 1036  else Line 1067  else
1067        const pcre_uchar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1068    
1069        c = 0;        c = 0;
1070          overflow = FALSE;
1071        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1072          {          {
1073          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1074          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1075    
1076    #ifdef COMPILE_PCRE32
1077            if (c >= 0x10000000l) { overflow = TRUE; break; }
1078    #endif
1079    
1080  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1081          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1082          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 1049  else Line 1085  else
1085          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1086  #endif  #endif
1087    
1088  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1089          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1090  #else  #elif defined COMPILE_PCRE16
1091  #ifdef COMPILE_PCRE16          if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1092          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }  #elif defined COMPILE_PCRE32
1093  #endif          if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1094  #endif  #endif
1095          }          }
1096    
1097        if (c < 0)        if (overflow)
1098          {          {
1099          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1100          *errorcodeptr = ERR34;          *errorcodeptr = ERR34;
# Line 1080  else Line 1116  else
1116      c = 0;      c = 0;
1117      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)
1118        {        {
1119        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1120        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1121  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1122        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 1099  else Line 1135  else
1135    
1136      case CHAR_c:      case CHAR_c:
1137      c = *(++ptr);      c = *(++ptr);
1138      if (c == 0)      if (c == CHAR_NULL)
1139        {        {
1140        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1141        break;        break;
# Line 1139  else Line 1175  else
1175  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1176  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1177    
1178  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1179       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1180    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1181    
1182  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1183    
1184  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1185    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1186    
1187  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1188    
1189  *ptrptr = ptr;  *ptrptr = ptr;
1190  return c;  *chptr = c;
1191    return escape;
1192  }  }
1193    
   
   
1194  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1195  /*************************************************  /*************************************************
1196  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1169  escape sequence. Line 1204  escape sequence.
1204  Argument:  Argument:
1205    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1206    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
1207    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
1208      pdataptr       points to an unsigned int that is set to the detailed property value
1209    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1210    
1211  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
1212  */  */
1213    
1214  static int  static BOOL
1215  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1216      unsigned int *pdataptr, int *errorcodeptr)
1217  {  {
1218  int c, i, bot, top;  pcre_uchar c;
1219    int i, bot, top;
1220  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1221  pcre_uchar name[32];  pcre_uchar name[32];
1222    
1223  c = *(++ptr);  c = *(++ptr);
1224  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1225    
1226  *negptr = FALSE;  *negptr = FALSE;
1227    
# Line 1200  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1238  if (c == CHAR_LEFT_CURLY_BRACKET)
1238    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1239      {      {
1240      c = *(++ptr);      c = *(++ptr);
1241      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1242      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1243      name[i] = c;      name[i] = c;
1244      }      }
# Line 1225  top = PRIV(utt_size); Line 1263  top = PRIV(utt_size);
1263    
1264  while (bot < top)  while (bot < top)
1265    {    {
1266      int r;
1267    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1268    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);
1269    if (c == 0)    if (r == 0)
1270      {      {
1271      *dptr = PRIV(utt)[i].value;      *ptypeptr = PRIV(utt)[i].type;
1272      return PRIV(utt)[i].type;      *pdataptr = PRIV(utt)[i].value;
1273        return TRUE;
1274      }      }
1275    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1276    }    }
1277    
1278  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1279  *ptrptr = ptr;  *ptrptr = ptr;
1280  return -1;  return FALSE;
1281    
1282  ERROR_RETURN:  ERROR_RETURN:
1283  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1284  *ptrptr = ptr;  *ptrptr = ptr;
1285  return -1;  return FALSE;
1286  }  }
1287  #endif  #endif
1288    
# Line 1277  int max = -1; Line 1317  int max = -1;
1317  /* 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
1318  an integer overflow. */  an integer overflow. */
1319    
1320  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1321  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1322    {    {
1323    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1292  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1332  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1332    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1333      {      {
1334      max = 0;      max = 0;
1335      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1336      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1337        {        {
1338        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1347  Arguments: Line 1387  Arguments:
1387    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1388    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1389    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1390    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1391    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1392    
1393  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 1393  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1433  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1433    
1434    else if (ptr[2] == CHAR_NUMBER_SIGN)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1435      {      {
1436      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;      for (ptr += 3; *ptr != CHAR_NULL; ptr++)
1437          if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1438      goto FAIL_EXIT;      goto FAIL_EXIT;
1439      }      }
1440    
# Line 1406  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1447  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1447      ptr += 2;      ptr += 2;
1448      if (ptr[1] != CHAR_QUESTION_MARK)      if (ptr[1] != CHAR_QUESTION_MARK)
1449        {        {
1450        while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;        while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1451        if (*ptr != 0) ptr++;        if (*ptr != CHAR_NULL) ptr++;
1452        }        }
1453      }      }
1454    
# Line 1423  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1464  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1464      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1465          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1466        {        {
1467        int term;        pcre_uchar term;
1468        const pcre_uchar *thisname;        const pcre_uchar *thisname;
1469        *count += 1;        *count += 1;
1470        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
# Line 1431  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1472  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1472        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1473        thisname = ptr;        thisname = ptr;
1474        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1475        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1476            STRNCMP_UC_UC(name, thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1477          return *count;          return *count;
1478        term++;        term++;
1479        }        }
# Line 1450  for (; ptr < cd->end_pattern; ptr++) Line 1491  for (; ptr < cd->end_pattern; ptr++)
1491    
1492    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1493      {      {
1494      if (*(++ptr) == 0) goto FAIL_EXIT;      if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1495      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1496        {        {
1497        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1498        if (*ptr == 0) goto FAIL_EXIT;        if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1499        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1500        }        }
1501      continue;      continue;
# Line 1498  for (; ptr < cd->end_pattern; ptr++) Line 1539  for (; ptr < cd->end_pattern; ptr++)
1539    
1540      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1541        {        {
1542        if (*ptr == 0) return -1;        if (*ptr == CHAR_NULL) return -1;
1543        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1544          {          {
1545          if (*(++ptr) == 0) goto FAIL_EXIT;          if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1546          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1547            {            {
1548            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1549            if (*ptr == 0) goto FAIL_EXIT;            if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1550            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1551            }            }
1552          continue;          continue;
# Line 1519  for (; ptr < cd->end_pattern; ptr++) Line 1560  for (; ptr < cd->end_pattern; ptr++)
1560    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1561      {      {
1562      ptr++;      ptr++;
1563      while (*ptr != 0)      while (*ptr != CHAR_NULL)
1564        {        {
1565        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1566        ptr++;        ptr++;
# Line 1527  for (; ptr < cd->end_pattern; ptr++) Line 1568  for (; ptr < cd->end_pattern; ptr++)
1568        if (utf) FORWARDCHAR(ptr);        if (utf) FORWARDCHAR(ptr);
1569  #endif  #endif
1570        }        }
1571      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1572      continue;      continue;
1573      }      }
1574    
# Line 1537  for (; ptr < cd->end_pattern; ptr++) Line 1578  for (; ptr < cd->end_pattern; ptr++)
1578      {      {
1579      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1580      if (rc > 0) return rc;      if (rc > 0) return rc;
1581      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1582      }      }
1583    
1584    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
# Line 1581  Arguments: Line 1622  Arguments:
1622    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1623    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1624    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1625    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1626    
1627  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1628  */  */
# Line 1602  matching closing parens. That is why we Line 1643  matching closing parens. That is why we
1643  for (;;)  for (;;)
1644    {    {
1645    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1646    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == CHAR_NULL) break;
1647    }    }
1648    
1649  return rc;  return rc;
# Line 1684  and doing the check at the end; a flag s Line 1725  and doing the check at the end; a flag s
1725    
1726  Arguments:  Arguments:
1727    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1728    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1729    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1730    cd       the "compile data" structure    cd       the "compile data" structure
1731    
# Line 1710  for (;;) Line 1751  for (;;)
1751    {    {
1752    int d;    int d;
1753    pcre_uchar *ce, *cs;    pcre_uchar *ce, *cs;
1754    register int op = *cc;    register pcre_uchar op = *cc;
1755    
1756    switch (op)    switch (op)
1757      {      {
# Line 1830  for (;;) Line 1871  for (;;)
1871      case OP_EXACTI:      case OP_EXACTI:
1872      case OP_NOTEXACT:      case OP_NOTEXACT:
1873      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1874      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1875      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1876  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
1877      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
# Line 1839  for (;;) Line 1880  for (;;)
1880    
1881      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1882      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1883      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)
1884          cc += 2;
1885      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1886      break;      break;
1887    
# Line 1874  for (;;) Line 1916  for (;;)
1916    
1917      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1918    
 #if defined SUPPORT_UTF || defined COMPILE_PCRE16  
     case OP_XCLASS:  
     cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];  
     /* Fall through */  
 #endif  
   
1919      case OP_CLASS:      case OP_CLASS:
1920      case OP_NCLASS:      case OP_NCLASS:
1921    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1922        case OP_XCLASS:
1923        /* The original code caused an unsigned overflow in 64 bit systems,
1924        so now we use a conditional statement. */
1925        if (op == OP_XCLASS)
1926          cc += GET(cc, 1);
1927        else
1928          cc += PRIV(OP_lengths)[OP_CLASS];
1929    #else
1930      cc += PRIV(OP_lengths)[OP_CLASS];      cc += PRIV(OP_lengths)[OP_CLASS];
1931    #endif
1932    
1933      switch (*cc)      switch (*cc)
1934        {        {
# Line 1897  for (;;) Line 1943  for (;;)
1943        case OP_CRRANGE:        case OP_CRRANGE:
1944        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1945        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1946        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1947        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
1948        break;        break;
1949    
# Line 2013  length. Line 2059  length.
2059    
2060  Arguments:  Arguments:
2061    code        points to start of expression    code        points to start of expression
2062    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2063    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2064    
2065  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 2024  PRIV(find_bracket)(const pcre_uchar *cod Line 2070  PRIV(find_bracket)(const pcre_uchar *cod
2070  {  {
2071  for (;;)  for (;;)
2072    {    {
2073    register int c = *code;    register pcre_uchar c = *code;
2074    
2075    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2076    
# Line 2047  for (;;) Line 2093  for (;;)
2093    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2094             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2095      {      {
2096      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2097      if (n == number) return (pcre_uchar *)code;      if (n == number) return (pcre_uchar *)code;
2098      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
2099      }      }
# Line 2077  for (;;) Line 2123  for (;;)
2123        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2124        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2125        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2126        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2127          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2128        break;        break;
2129    
2130        case OP_MARK:        case OP_MARK:
# Line 2100  for (;;) Line 2146  for (;;)
2146    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
2147    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2148    
2149  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2150      if (utf) switch(c)      if (utf) switch(c)
2151        {        {
2152        case OP_CHAR:        case OP_CHAR:
# Line 2152  instance of OP_RECURSE. Line 2198  instance of OP_RECURSE.
2198    
2199  Arguments:  Arguments:
2200    code        points to start of expression    code        points to start of expression
2201    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2202    
2203  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
2204  */  */
# Line 2162  find_recurse(const pcre_uchar *code, BOO Line 2208  find_recurse(const pcre_uchar *code, BOO
2208  {  {
2209  for (;;)  for (;;)
2210    {    {
2211    register int c = *code;    register pcre_uchar c = *code;
2212    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2213    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2214    
# Line 2197  for (;;) Line 2243  for (;;)
2243        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2244        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2245        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2246        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2247          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2248        break;        break;
2249    
2250        case OP_MARK:        case OP_MARK:
# Line 2220  for (;;) Line 2266  for (;;)
2266      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
2267      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2268    
2269  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2270      if (utf) switch(c)      if (utf) switch(c)
2271        {        {
2272        case OP_CHAR:        case OP_CHAR:
2273        case OP_CHARI:        case OP_CHARI:
2274          case OP_NOT:
2275          case OP_NOTI:
2276        case OP_EXACT:        case OP_EXACT:
2277        case OP_EXACTI:        case OP_EXACTI:
2278          case OP_NOTEXACT:
2279          case OP_NOTEXACTI:
2280        case OP_UPTO:        case OP_UPTO:
2281        case OP_UPTOI:        case OP_UPTOI:
2282          case OP_NOTUPTO:
2283          case OP_NOTUPTOI:
2284        case OP_MINUPTO:        case OP_MINUPTO:
2285        case OP_MINUPTOI:        case OP_MINUPTOI:
2286          case OP_NOTMINUPTO:
2287          case OP_NOTMINUPTOI:
2288        case OP_POSUPTO:        case OP_POSUPTO:
2289        case OP_POSUPTOI:        case OP_POSUPTOI:
2290          case OP_NOTPOSUPTO:
2291          case OP_NOTPOSUPTOI:
2292        case OP_STAR:        case OP_STAR:
2293        case OP_STARI:        case OP_STARI:
2294          case OP_NOTSTAR:
2295          case OP_NOTSTARI:
2296        case OP_MINSTAR:        case OP_MINSTAR:
2297        case OP_MINSTARI:        case OP_MINSTARI:
2298          case OP_NOTMINSTAR:
2299          case OP_NOTMINSTARI:
2300        case OP_POSSTAR:        case OP_POSSTAR:
2301        case OP_POSSTARI:        case OP_POSSTARI:
2302          case OP_NOTPOSSTAR:
2303          case OP_NOTPOSSTARI:
2304        case OP_PLUS:        case OP_PLUS:
2305        case OP_PLUSI:        case OP_PLUSI:
2306          case OP_NOTPLUS:
2307          case OP_NOTPLUSI:
2308        case OP_MINPLUS:        case OP_MINPLUS:
2309        case OP_MINPLUSI:        case OP_MINPLUSI:
2310          case OP_NOTMINPLUS:
2311          case OP_NOTMINPLUSI:
2312        case OP_POSPLUS:        case OP_POSPLUS:
2313        case OP_POSPLUSI:        case OP_POSPLUSI:
2314          case OP_NOTPOSPLUS:
2315          case OP_NOTPOSPLUSI:
2316        case OP_QUERY:        case OP_QUERY:
2317        case OP_QUERYI:        case OP_QUERYI:
2318          case OP_NOTQUERY:
2319          case OP_NOTQUERYI:
2320        case OP_MINQUERY:        case OP_MINQUERY:
2321        case OP_MINQUERYI:        case OP_MINQUERYI:
2322          case OP_NOTMINQUERY:
2323          case OP_NOTMINQUERYI:
2324        case OP_POSQUERY:        case OP_POSQUERY:
2325        case OP_POSQUERYI:        case OP_POSQUERYI:
2326          case OP_NOTPOSQUERY:
2327          case OP_NOTPOSQUERYI:
2328        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2329        break;        break;
2330        }        }
# Line 2278  bracket whose current branch will alread Line 2352  bracket whose current branch will alread
2352  Arguments:  Arguments:
2353    code        points to start of search    code        points to start of search
2354    endcode     points to where to stop    endcode     points to where to stop
2355    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2356    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2357    
2358  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2288  static BOOL Line 2362  static BOOL
2362  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2363    BOOL utf, compile_data *cd)    BOOL utf, compile_data *cd)
2364  {  {
2365  register int c;  register pcre_uchar c;
2366  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2367       code < endcode;       code < endcode;
2368       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
# Line 2322  for (code = first_significant_code(code Line 2396  for (code = first_significant_code(code
2396      /* Test for forward reference */      /* Test for forward reference */
2397    
2398      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2399        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2400    
2401      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2402    
# Line 2495  for (code = first_significant_code(code Line 2569  for (code = first_significant_code(code
2569      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2570      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2571      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2572      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2573        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2574      break;      break;
2575    
2576      /* End of branch */      /* End of branch */
# Line 2511  for (code = first_significant_code(code Line 2585  for (code = first_significant_code(code
2585      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2586      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2587    
2588  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2589      case OP_STAR:      case OP_STAR:
2590      case OP_STARI:      case OP_STARI:
2591      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2577  Arguments: Line 2651  Arguments:
2651    code        points to start of the recursion    code        points to start of the recursion
2652    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2653    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2654    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2655    cd          pointers to tables etc    cd          pointers to tables etc
2656    
2657  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2643  Returns:   TRUE or FALSE Line 2717  Returns:   TRUE or FALSE
2717  static BOOL  static BOOL
2718  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2719  {  {
2720  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2721  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2722  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
2723    {    {
2724    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2725      ptr++;      ptr++;
# Line 2692  register int yield = 0; Line 2766  register int yield = 0;
2766  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2767    {    {
2768    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2769      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2770    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2771    yield++;    yield++;
2772    }    }
# Line 2724  value in the reference (which is a group Line 2798  value in the reference (which is a group
2798  Arguments:  Arguments:
2799    group      points to the start of the group    group      points to the start of the group
2800    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2801    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2802    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2803    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
2804    
# Line 2747  while ((ptr = (pcre_uchar *)find_recurse Line 2821  while ((ptr = (pcre_uchar *)find_recurse
2821    
2822    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2823      {      {
2824      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2825      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2826        {        {
2827        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2760  while ((ptr = (pcre_uchar *)find_recurse Line 2834  while ((ptr = (pcre_uchar *)find_recurse
2834    
2835    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2836      {      {
2837      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2838      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2839      }      }
2840    
# Line 2828  PUT(previous_callout, 2 + LINK_SIZE, len Line 2902  PUT(previous_callout, 2 + LINK_SIZE, len
2902  *************************************************/  *************************************************/
2903    
2904  /* 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
2905  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
2906  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
2907  start address.  start address. A character with multiple other cases is returned on its own
2908    with a special return value.
2909    
2910  Arguments:  Arguments:
2911    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2838  Arguments: Line 2913  Arguments:
2913    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2914    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2915    
2916  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2917                   0 when a range is returned
2918                  >0 the CASESET offset for char with multiple other cases
2919                    in this case, ocptr contains the original
2920  */  */
2921    
2922  static BOOL  static int
2923  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2924    unsigned int *odptr)    pcre_uint32 *odptr)
2925  {  {
2926  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2927    unsigned int co;
2928    
2929    /* Find the first character that has an other case. If it has multiple other
2930    cases, return its case offset value. */
2931    
2932  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2933    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2934      if ((co = UCD_CASESET(c)) != 0)
2935        {
2936        *ocptr = c++;   /* Character that has the set */
2937        *cptr = c;      /* Rest of input range */
2938        return (int)co;
2939        }
2940      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2941      }
2942    
2943  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2944    
2945  *ocptr = othercase;  *ocptr = othercase;
2946  next = othercase + 1;  next = othercase + 1;
# Line 2861  for (++c; c <= d; c++) Line 2951  for (++c; c <= d; c++)
2951    next++;    next++;
2952    }    }
2953    
2954  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2955  *cptr = c;  *cptr = c;             /* Rest of input range */
2956    return 0;
 return TRUE;  
2957  }  }
2958    
2959    
# Line 2886  Returns:       TRUE if auto-possessifyin Line 2975  Returns:       TRUE if auto-possessifyin
2975  */  */
2976    
2977  static BOOL  static BOOL
2978  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)
2979  {  {
2980    #ifdef SUPPORT_UCP
2981    const pcre_uint32 *p;
2982    #endif
2983    
2984  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2985    
2986  switch(ptype)  switch(ptype)
2987    {    {
2988    case PT_LAMP:    case PT_LAMP:
# Line 2926  switch(ptype) Line 3020  switch(ptype)
3020    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3021            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3022            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3023    
3024    #ifdef SUPPORT_UCP
3025      case PT_CLIST:
3026      p = PRIV(ucd_caseless_sets) + prop->caseset;
3027      for (;;)
3028        {
3029        if (c < *p) return !negated;
3030        if (c == *p++) return negated;
3031        }
3032      break;  /* Control never reaches here */
3033    #endif
3034    }    }
3035    
3036  return FALSE;  return FALSE;
3037  }  }
3038  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2943  sense to automatically possessify the re Line 3049  sense to automatically possessify the re
3049    
3050  Arguments:  Arguments:
3051    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3052    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3053    ptr           next character in pattern    ptr           next character in pattern
3054    options       options bits    options       options bits
3055    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2955  static BOOL Line 3061  static BOOL
3061  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3062    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3063  {  {
3064  pcre_int32 c, next;  pcre_uint32 c = NOTACHAR;
3065  int op_code = *previous++;  pcre_uint32 next;
3066    int escape;
3067    pcre_uchar op_code = *previous++;
3068    
3069  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3070    
# Line 2968  if ((options & PCRE_EXTENDED) != 0) Line 3076  if ((options & PCRE_EXTENDED) != 0)
3076      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3077        {        {
3078        ptr++;        ptr++;
3079        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3080          {          {
3081          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3082          ptr++;          ptr++;
# Line 2987  value is a character, a negative value i Line 3095  value is a character, a negative value i
3095  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3096    {    {
3097    int temperrorcode = 0;    int temperrorcode = 0;
3098    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options,
3099        FALSE);
3100    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3101    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3102    }    }
3103  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
3104    {    {
3105      escape = 0;
3106  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3107    if (utf) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3108  #endif  #endif
# Line 3010  if ((options & PCRE_EXTENDED) != 0) Line 3120  if ((options & PCRE_EXTENDED) != 0)
3120      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3121        {        {
3122        ptr++;        ptr++;
3123        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3124          {          {
3125          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3126          ptr++;          ptr++;
# Line 3029  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3139  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3139    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)
3140      return FALSE;      return FALSE;
3141    
3142  /* 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. */  
3143    
3144  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3145        op_code == OP_NOT || op_code == OP_NOTI)
3146    {    {
   case OP_CHAR:  
3147  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3148    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3149  #else  #else
3150    c = *previous;    c = *previous;
3151  #endif  #endif
3152    return c != next;    }
3153    
3154    /* For CHARI (caseless character) we must check the other case. If we have  /* Now compare the next item with the previous opcode. First, handle cases when
3155    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3156    
3157    case OP_CHARI:  if (escape == 0)
3158  #ifdef SUPPORT_UTF    {
3159    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3160  #else    case, which maps to the special PT_CLIST property. Check this first. */
3161    c = *previous;  
3162    #ifdef SUPPORT_UCP
3163      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3164        {
3165        unsigned int ocs = UCD_CASESET(next);
3166        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3167        }
3168  #endif  #endif
3169    if (c == next) return FALSE;  
3170  #ifdef SUPPORT_UTF    switch(op_code)
   if (utf)  
3171      {      {
3172      unsigned int othercase;      case OP_CHAR:
3173      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3174    
3175        /* For CHARI (caseless character) we must check the other case. If we have
3176        Unicode property support, we can use it to test the other case of
3177        high-valued characters. We know that next can have only one other case,
3178        because multi-other-case characters are dealt with above. */
3179    
3180        case OP_CHARI:
3181        if (c == next) return FALSE;
3182    #ifdef SUPPORT_UTF
3183        if (utf)
3184          {
3185          pcre_uint32 othercase;
3186          if (next < 128) othercase = cd->fcc[next]; else
3187  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3188      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3189  #else  #else
3190      othercase = NOTACHAR;        othercase = NOTACHAR;
3191  #endif  #endif
3192      return (unsigned int)c != othercase;        return c != othercase;
3193      }        }
3194    else      else
3195  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3196    return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
   
   /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These  
   opcodes are not used for multi-byte characters, because they are coded using  
   an XCLASS instead. */  
3197    
3198    case OP_NOT:      case OP_NOT:
3199    return (c = *previous) == next;      return c == next;
3200    
3201    case OP_NOTI:      case OP_NOTI:
3202    if ((c = *previous) == next) return TRUE;      if (c == next) return TRUE;
3203  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3204    if (utf)      if (utf)
3205      {        {
3206      unsigned int othercase;        pcre_uint32 othercase;
3207      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3208  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3209      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3210  #else  #else
3211      othercase = NOTACHAR;        othercase = NOTACHAR;
3212  #endif  #endif
3213      return (unsigned int)c == othercase;        return c == othercase;
3214      }        }
3215    else      else
3216  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3217    return (c == (int)(TABLE_GET((unsigned int)next, cd->fcc, next)));  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3218    
3219    /* 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.
3220    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. */
3221    
3222    case OP_DIGIT:      case OP_DIGIT:
3223    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3224    
3225    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3226    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3227    
3228    case OP_WHITESPACE:      case OP_WHITESPACE:
3229    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3230    
3231    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3232    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3233    
3234    case OP_WORDCHAR:      case OP_WORDCHAR:
3235    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3236    
3237    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3238    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3239    
3240    case OP_HSPACE:      case OP_HSPACE:
3241    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3242    switch(next)      switch(next)
3243      {        {
3244      case 0x09:        HSPACE_CASES:
3245      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;  
     }  
3246    
3247    case OP_ANYNL:        default:
3248    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3249    case OP_NOT_VSPACE:        }
3250    switch(next)  
3251      {      case OP_ANYNL:
3252      case 0x0a:      case OP_VSPACE:
3253      case 0x0b:      case OP_NOT_VSPACE:
3254      case 0x0c:      switch(next)
3255      case 0x0d:        {
3256      case 0x85:        VSPACE_CASES:
3257      case 0x2028:        return op_code == OP_NOT_VSPACE;
3258      case 0x2029:  
3259      return op_code == OP_NOT_VSPACE;        default:
3260      default:        return op_code != OP_NOT_VSPACE;
3261      return op_code != OP_NOT_VSPACE;        }
     }  
3262    
3263  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3264    case OP_PROP:      case OP_PROP:
3265    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3266    
3267    case OP_NOTPROP:      case OP_NOTPROP:
3268    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3269  #endif  #endif
3270    
3271    default:      default:
3272    return FALSE;      return FALSE;
3273        }
3274    }    }
3275    
   
3276  /* 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
3277  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
3278  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 3183  switch(op_code) Line 3283  switch(op_code)
3283    {    {
3284    case OP_CHAR:    case OP_CHAR:
3285    case OP_CHARI:    case OP_CHARI:
3286  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3287      {      {
3288      case ESC_d:      case ESC_d:
3289      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3290    
3291      case ESC_D:      case ESC_D:
3292      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3293    
3294      case ESC_s:      case ESC_s:
3295      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3296    
3297      case ESC_S:      case ESC_S:
3298      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3299    
3300      case ESC_w:      case ESC_w:
3301      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3302    
3303      case ESC_W:      case ESC_W:
3304      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3305    
3306      case ESC_h:      case ESC_h:
3307      case ESC_H:      case ESC_H:
3308      switch(c)      switch(c)
3309        {        {
3310        case 0x09:        HSPACE_CASES:
3311        case 0x20:        return escape != ESC_h;
3312        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;  
3313        default:        default:
3314        return -next == ESC_h;        return escape == ESC_h;
3315        }        }
3316    
3317      case ESC_v:      case ESC_v:
3318      case ESC_V:      case ESC_V:
3319      switch(c)      switch(c)
3320        {        {
3321        case 0x0a:        VSPACE_CASES:
3322        case 0x0b:        return escape != ESC_v;
3323        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3324        default:        default:
3325        return -next == ESC_v;        return escape == ESC_v;
3326        }        }
3327    
3328      /* 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
3329      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3330      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3331    
3332  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3333      case ESC_du:      case ESC_du:
# Line 3265  switch(op_code) Line 3338  switch(op_code)
3338      case ESC_SU:      case ESC_SU:
3339        {        {
3340        int temperrorcode = 0;        int temperrorcode = 0;
3341        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3342        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3343        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3344        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3345        }        }
# Line 3275  switch(op_code) Line 3348  switch(op_code)
3348      case ESC_p:      case ESC_p:
3349      case ESC_P:      case ESC_P:
3350        {        {
3351        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3352          int errorcodeptr;
3353        BOOL negated;        BOOL negated;
3354    
3355        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3356        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3357        if (ptype < 0) return FALSE;          return FALSE;
3358        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3359    
3360        /* 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 3293  switch(op_code) Line 3367  switch(op_code)
3367    
3368        /* Do the property check. */        /* Do the property check. */
3369    
3370        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3371        }        }
3372  #endif  #endif
3373    
# Line 3308  switch(op_code) Line 3382  switch(op_code)
3382    these op-codes are never generated.) */    these op-codes are never generated.) */
3383    
3384    case OP_DIGIT:    case OP_DIGIT:
3385    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3386           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3387    
3388    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3389    return next == -ESC_d;    return escape == ESC_d;
3390    
3391    case OP_WHITESPACE:    case OP_WHITESPACE:
3392    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3393    
3394    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3395    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3396    
3397    case OP_HSPACE:    case OP_HSPACE:
3398    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3399           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3400    
3401    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3402    return next == -ESC_h;    return escape == ESC_h;
3403    
3404    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3405    case OP_ANYNL:    case OP_ANYNL:
3406    case OP_VSPACE:    case OP_VSPACE:
3407    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3408    
3409    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3410    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3411    
3412    case OP_WORDCHAR:    case OP_WORDCHAR:
3413    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3414           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3415    
3416    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3417    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3418    
3419    default:    default:
3420    return FALSE;    return FALSE;
# Line 3352  switch(op_code) Line 3426  switch(op_code)
3426    
3427    
3428  /*************************************************  /*************************************************
3429    *        Add a character or range to a class     *
3430    *************************************************/
3431    
3432    /* This function packages up the logic of adding a character or range of
3433    characters to a class. The character values in the arguments will be within the
3434    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3435    mutually recursive with the function immediately below.
3436    
3437    Arguments:
3438      classbits     the bit map for characters < 256
3439      uchardptr     points to the pointer for extra data
3440      options       the options word
3441      cd            contains pointers to tables etc.
3442      start         start of range character
3443      end           end of range character
3444    
3445    Returns:        the number of < 256 characters added
3446                    the pointer to extra data is updated
3447    */
3448    
3449    static int
3450    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3451      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3452    {
3453    pcre_uint32 c;
3454    int n8 = 0;
3455    
3456    /* If caseless matching is required, scan the range and process alternate
3457    cases. In Unicode, there are 8-bit characters that have alternate cases that
3458    are greater than 255 and vice-versa. Sometimes we can just extend the original
3459    range. */
3460    
3461    if ((options & PCRE_CASELESS) != 0)
3462      {
3463    #ifdef SUPPORT_UCP
3464      if ((options & PCRE_UTF8) != 0)
3465        {
3466        int rc;
3467        pcre_uint32 oc, od;
3468    
3469        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3470        c = start;
3471    
3472        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3473          {
3474          /* Handle a single character that has more than one other case. */
3475    
3476          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3477            PRIV(ucd_caseless_sets) + rc, oc);
3478    
3479          /* Do nothing if the other case range is within the original range. */
3480    
3481          else if (oc >= start && od <= end) continue;
3482    
3483          /* Extend the original range if there is overlap, noting that if oc < c, we
3484          can't have od > end because a subrange is always shorter than the basic
3485          range. Otherwise, use a recursive call to add the additional range. */
3486    
3487          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3488          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3489          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3490          }
3491        }
3492      else
3493    #endif  /* SUPPORT_UCP */
3494    
3495      /* Not UTF-mode, or no UCP */
3496    
3497      for (c = start; c <= end && c < 256; c++)
3498        {
3499        SETBIT(classbits, cd->fcc[c]);
3500        n8++;
3501        }
3502      }
3503    
3504    /* Now handle the original range. Adjust the final value according to the bit
3505    length - this means that the same lists of (e.g.) horizontal spaces can be used
3506    in all cases. */
3507    
3508    #if defined COMPILE_PCRE8
3509    #ifdef SUPPORT_UTF
3510      if ((options & PCRE_UTF8) == 0)
3511    #endif
3512      if (end > 0xff) end = 0xff;
3513    
3514    #elif defined COMPILE_PCRE16
3515    #ifdef SUPPORT_UTF
3516      if ((options & PCRE_UTF16) == 0)
3517    #endif
3518      if (end > 0xffff) end = 0xffff;
3519    
3520    #endif /* COMPILE_PCRE[8|16] */
3521    
3522    /* If all characters are less than 256, use the bit map. Otherwise use extra
3523    data. */
3524    
3525    if (end < 0x100)
3526      {
3527      for (c = start; c <= end; c++)
3528        {
3529        n8++;
3530        SETBIT(classbits, c);
3531        }
3532      }
3533    
3534    else
3535      {
3536      pcre_uchar *uchardata = *uchardptr;
3537    
3538    #ifdef SUPPORT_UTF
3539      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3540        {
3541        if (start < end)
3542          {
3543          *uchardata++ = XCL_RANGE;
3544          uchardata += PRIV(ord2utf)(start, uchardata);
3545          uchardata += PRIV(ord2utf)(end, uchardata);
3546          }
3547        else if (start == end)
3548          {
3549          *uchardata++ = XCL_SINGLE;
3550          uchardata += PRIV(ord2utf)(start, uchardata);
3551          }
3552        }
3553      else
3554    #endif  /* SUPPORT_UTF */
3555    
3556      /* Without UTF support, character values are constrained by the bit length,
3557      and can only be > 256 for 16-bit and 32-bit libraries. */
3558    
3559    #ifdef COMPILE_PCRE8
3560        {}
3561    #else
3562      if (start < end)
3563        {
3564        *uchardata++ = XCL_RANGE;
3565        *uchardata++ = start;
3566        *uchardata++ = end;
3567        }
3568      else if (start == end)
3569        {
3570        *uchardata++ = XCL_SINGLE;
3571        *uchardata++ = start;
3572        }
3573    #endif
3574    
3575      *uchardptr = uchardata;   /* Updata extra data pointer */
3576      }
3577    
3578    return n8;    /* Number of 8-bit characters */
3579    }
3580    
3581    
3582    
3583    
3584    /*************************************************
3585    *        Add a list of characters to a class     *
3586    *************************************************/
3587    
3588    /* This function is used for adding a list of case-equivalent characters to a
3589    class, and also for adding a list of horizontal or vertical whitespace. If the
3590    list is in order (which it should be), ranges of characters are detected and
3591    handled appropriately. This function is mutually recursive with the function
3592    above.
3593    
3594    Arguments:
3595      classbits     the bit map for characters < 256
3596      uchardptr     points to the pointer for extra data
3597      options       the options word
3598      cd            contains pointers to tables etc.
3599      p             points to row of 32-bit values, terminated by NOTACHAR
3600      except        character to omit; this is used when adding lists of
3601                      case-equivalent characters to avoid including the one we
3602                      already know about
3603    
3604    Returns:        the number of < 256 characters added
3605                    the pointer to extra data is updated
3606    */
3607    
3608    static int
3609    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3610      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3611    {
3612    int n8 = 0;
3613    while (p[0] < NOTACHAR)
3614      {
3615      int n = 0;
3616      if (p[0] != except)
3617        {
3618        while(p[n+1] == p[0] + n + 1) n++;
3619        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3620        }
3621      p += n + 1;
3622      }
3623    return n8;
3624    }
3625    
3626    
3627    
3628    /*************************************************
3629    *    Add characters not in a list to a class     *
3630    *************************************************/
3631    
3632    /* This function is used for adding the complement of a list of horizontal or
3633    vertical whitespace to a class. The list must be in order.
3634    
3635    Arguments:
3636      classbits     the bit map for characters < 256
3637      uchardptr     points to the pointer for extra data
3638      options       the options word
3639      cd            contains pointers to tables etc.
3640      p             points to row of 32-bit values, terminated by NOTACHAR
3641    
3642    Returns:        the number of < 256 characters added
3643                    the pointer to extra data is updated
3644    */
3645    
3646    static int
3647    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3648      int options, compile_data *cd, const pcre_uint32 *p)
3649    {
3650    BOOL utf = (options & PCRE_UTF8) != 0;
3651    int n8 = 0;
3652    if (p[0] > 0)
3653      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3654    while (p[0] < NOTACHAR)
3655      {
3656      while (p[1] == p[0] + 1) p++;
3657      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3658        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3659      p++;
3660      }
3661    return n8;
3662    }
3663    
3664    
3665    
3666    /*************************************************
3667  *           Compile one branch                   *  *           Compile one branch                   *
3668  *************************************************/  *************************************************/
3669    
# Line 3366  Arguments: Line 3678  Arguments:
3678    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3679    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3680    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3681    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3682    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3683      reqcharptr     place to put the last required character
3684      reqcharflagsptr place to put the last required character flags, or a negative number
3685    bcptr          points to current branch chain    bcptr          points to current branch chain
3686    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3687    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3380  Returns:         TRUE on success Line 3694  Returns:         TRUE on success
3694    
3695  static BOOL  static BOOL
3696  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3697    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3698    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3699      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3700      branch_chain *bcptr, int cond_depth,
3701    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3702  {  {
3703  int repeat_type, op_type;  int repeat_type, op_type;
3704  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3705  int bravalue = 0;  int bravalue = 0;
3706  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3707  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
3708  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
3709    pcre_uint32 zeroreqchar, zerofirstchar;
3710    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3711  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3712  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3713  int after_manual_callout = 0;  int after_manual_callout = 0;
3714  int length_prevgroup = 0;  int length_prevgroup = 0;
3715  register int c;  register pcre_uint32 c;
3716    int escape;
3717  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3718  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3719  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3414  must not do this for other options (e.g. Line 3733  must not do this for other options (e.g.
3733  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3734    
3735  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3736  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3737  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3738    #ifndef COMPILE_PCRE32
3739  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3740    #endif
3741  #else  #else
3742  BOOL utf = FALSE;  BOOL utf = FALSE;
3743  #endif  #endif
3744    
3745  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3746    class_uchardata always so that it can be passed to add_to_class() always,
3747    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3748    alternative calls for the different cases. */
3749    
3750    pcre_uchar *class_uchardata;
3751  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3752  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3753  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3754  #endif  #endif
3755    
# Line 3448  to take the zero repeat into account. Th Line 3772  to take the zero repeat into account. Th
3772  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3773  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3774    
3775  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3776    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3777    
3778  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
3779  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 3469  for (;; ptr++) Line 3794  for (;; ptr++)
3794    BOOL is_recurse;    BOOL is_recurse;
3795    BOOL reset_bracount;    BOOL reset_bracount;
3796    int class_has_8bitchar;    int class_has_8bitchar;
3797    int class_single_char;    int class_one_char;
3798    int newoptions;    int newoptions;
3799    int recno;    int recno;
3800    int refsign;    int refsign;
3801    int skipbytes;    int skipbytes;
3802    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
3803    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
3804    int terminator;    int terminator;
3805    int mclength;    unsigned int mclength;
3806    int tempbracount;    unsigned int tempbracount;
3807      pcre_uint32 ec;
3808    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3809    
3810    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3488  for (;; ptr++) Line 3814  for (;; ptr++)
3814    /* 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
3815    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
3816    
3817    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
3818      {      {
3819      ptr = nestptr;      ptr = nestptr;
3820      nestptr = NULL;      nestptr = NULL;
# Line 3563  for (;; ptr++) Line 3889  for (;; ptr++)
3889    
3890    /* 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 */
3891    
3892    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
3893      {      {
3894      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3895        {        {
# Line 3611  for (;; ptr++) Line 3937  for (;; ptr++)
3937      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3938        {        {
3939        ptr++;        ptr++;
3940        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3941          {          {
3942          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3943          ptr++;          ptr++;
# Line 3619  for (;; ptr++) Line 3945  for (;; ptr++)
3945          if (utf) FORWARDCHAR(ptr);          if (utf) FORWARDCHAR(ptr);
3946  #endif  #endif
3947          }          }
3948        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
3949    
3950        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
3951        c = 0;        c = 0;
# Line 3641  for (;; ptr++) Line 3967  for (;; ptr++)
3967      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3968      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3969      *firstcharptr = firstchar;      *firstcharptr = firstchar;
3970        *firstcharflagsptr = firstcharflags;
3971      *reqcharptr = reqchar;      *reqcharptr = reqchar;
3972        *reqcharflagsptr = reqcharflags;
3973      *codeptr = code;      *codeptr = code;
3974      *ptrptr = ptr;      *ptrptr = ptr;
3975      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3665  for (;; ptr++) Line 3993  for (;; ptr++)
3993      previous = NULL;      previous = NULL;
3994      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3995        {        {
3996        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3997        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3998        }        }
3999      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3680  for (;; ptr++) Line 4008  for (;; ptr++)
4008      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4009    
4010      case CHAR_DOT:      case CHAR_DOT:
4011      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4012      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4013        zerofirstcharflags = firstcharflags;
4014      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4015        zeroreqcharflags = reqcharflags;
4016      previous = code;      previous = code;
4017      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4018      break;      break;
# Line 3756  for (;; ptr++) Line 4086  for (;; ptr++)
4086          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4087        {        {
4088        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4089        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4090        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4091          zerofirstcharflags = firstcharflags;
4092        break;        break;
4093        }        }
4094    
# Line 3767  for (;; ptr++) Line 4098  for (;; ptr++)
4098    
4099      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4100    
4101      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4102      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 <
4103      < 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
4104      a single character. */      character. */
4105    
4106      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4107      class_single_char = 0;      class_one_char = 0;
4108    
4109      /* 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
4110      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
4111      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
4112      */      map. */
4113    
4114      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4115    
4116  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4117      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4118      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4119      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4120  #endif  #endif
4121    
4122      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4123      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
4124      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4125    
4126      if (c != 0) do      if (c != CHAR_NULL) do
4127        {        {
4128        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4129    
# Line 3807  for (;; ptr++) Line 4138  for (;; ptr++)
4138        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4139        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
4140        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4141        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4142          however. */
4143    
4144        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4145          {          {
4146            xclass = TRUE;
4147          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4148          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4149          }          }
# Line 3912  for (;; ptr++) Line 4245  for (;; ptr++)
4245              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4246            }            }
4247    
4248          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4249          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4250    
4251          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3928  for (;; ptr++) Line 4261  for (;; ptr++)
4261            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4262    
4263          ptr = tempptr + 1;          ptr = tempptr + 1;
4264          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4265          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4266          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4267          class_single_char = 2;          class_one_char = 2;
4268          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4269          }          }
4270    
# Line 3939  for (;; ptr++) Line 4272  for (;; ptr++)
4272        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
4273        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
4274        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4275        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4276        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4277        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4278        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4279    
4280        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4281          {          {
4282          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options,
4283              TRUE);
4284          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4285            if (escape == 0) c = ec;
4286          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 */
4287          else if (-c == ESC_N)            /* \N is not supported in a class */          else if (escape == ESC_N)          /* \N is not supported in a class */
4288            {            {
4289            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4290            goto FAILED;            goto FAILED;
4291            }            }
4292          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4293            {            {
4294            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4295              {              {
# Line 3964  for (;; ptr++) Line 4298  for (;; ptr++)
4298            else inescq = TRUE;            else inescq = TRUE;
4299            continue;            continue;
4300            }            }
4301          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4302    
4303          if (c < 0)          else
4304            {            {
4305            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4306            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4307            class_has_8bitchar++;            class_has_8bitchar++;
4308            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4309            class_single_char += 2;            class_one_char += 2;
4310    
4311            switch (-c)            switch (escape)
4312              {              {
4313  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4314              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3984  for (;; ptr++) Line 4318  for (;; ptr++)
4318              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4319              case ESC_SU:              case ESC_SU:
4320              nestptr = ptr;              nestptr = ptr;
4321              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4322              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4323              continue;              continue;
4324  #endif  #endif
# Line 4008  for (;; ptr++) Line 4342  for (;; ptr++)
4342    
4343              /* 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
4344              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4345              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4346                EBCDIC, so we lazily just adjust the appropriate bit. */
4347    
4348              case ESC_s:              case ESC_s:
4349              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4022  for (;; ptr++) Line 4357  for (;; ptr++)
4357              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4358              continue;              continue;
4359    
4360                /* The rest apply in both UCP and non-UCP cases. */
4361    
4362              case ESC_h:              case ESC_h:
4363              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4364              SETBIT(classbits, 0x20); /* SPACE */                PRIV(hspace_list), NOTACHAR);
             SETBIT(classbits, 0xa0); /* NSBP */  
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x1680;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x180e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2000;  
             *class_uchardata++ = 0x200a;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x202f;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x205f;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x3000;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x1680, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x180e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2000, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x200a, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x202f, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x205f, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x3000, class_uchardata);  
               }  
 #endif  
4365              continue;              continue;
4366    
4367              case ESC_H:              case ESC_H:
4368              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4369                {                cd, PRIV(hspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x09/8: x ^= 1 << (0x09%8); break;  
                 case 0x20/8: x ^= 1 << (0x20%8); break;  
                 case 0xa0/8: x ^= 1 << (0xa0%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x0100;  
             *class_uchardata++ = 0x167f;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x1681;  
             *class_uchardata++ = 0x180d;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x180f;  
             *class_uchardata++ = 0x1fff;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x200b;  
             *class_uchardata++ = 0x202e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2030;  
             *class_uchardata++ = 0x205e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2060;  
             *class_uchardata++ = 0x2fff;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x3001;  
 #ifdef SUPPORT_UTF  
             if (utf)  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
             else  
 #endif  
               *class_uchardata++ = 0xffff;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x167f, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x1681, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x180d, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x180f, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x1fff, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x200b, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x202e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2030, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x205e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2060, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x2fff, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x3001, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
               }  
 #endif  
4370              continue;              continue;
4371    
4372              case ESC_v:              case ESC_v:
4373              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4374              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  
4375              continue;              continue;
4376    
4377              case ESC_V:              case ESC_V:
4378              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4379                {                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  
4380              continue;              continue;
4381    
4382  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4203  for (;; ptr++) Line 4384  for (;; ptr++)
4384              case ESC_P:              case ESC_P:
4385                {                {
4386                BOOL negated;                BOOL negated;
4387                int pdata;                unsigned int ptype = 0, pdata = 0;
4388                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4389                if (ptype < 0) goto FAILED;                  goto FAILED;
4390                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4391                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4392                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4393                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4226  for (;; ptr++) Line 4406  for (;; ptr++)
4406                goto FAILED;                goto FAILED;
4407                }                }
4408              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4409              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4410              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4411              break;              break;
4412              }              }
4413            }            }
4414    
4415          /* 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).
4416          greater than 256. */          This may be greater than 256. */
4417    
4418            escape = 0;
4419    
4420          }   /* End of backslash handling */          }   /* End of backslash handling */
4421    
4422        /* 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
4423        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
4424        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4425        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4426    
4427        CHECK_RANGE:        CHECK_RANGE:
4428        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4248  for (;; ptr++) Line 4430  for (;; ptr++)
4430          inescq = FALSE;          inescq = FALSE;
4431          ptr += 2;          ptr += 2;
4432          }          }
   
4433        oldptr = ptr;        oldptr = ptr;
4434    
4435        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4436    
4437        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4438    
# Line 4259  for (;; ptr++) Line 4440  for (;; ptr++)
4440    
4441        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4442          {          {
4443          int d;          pcre_uint32 d;
4444          ptr += 2;          ptr += 2;
4445          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4446    
# Line 4275  for (;; ptr++) Line 4456  for (;; ptr++)
4456            break;            break;
4457            }            }
4458    
4459          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          /* Minus (hyphen) at the end of a class is treated as a literal, so put
4460            back the pointer and jump to handle the character that preceded it. */
4461    
4462            if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4463            {            {
4464            ptr = oldptr;            ptr = oldptr;
4465            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4466            }            }
4467    
4468            /* Otherwise, we have a potential range; pick up the next character */
4469    
4470  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4471          if (utf)          if (utf)
4472            {                           /* Braces are required because the */            {                           /* Braces are required because the */
# Line 4296  for (;; ptr++) Line 4482  for (;; ptr++)
4482    
4483          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4484            {            {
4485            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4486              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4487            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4488    
4489            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4490    
4491            if (d < 0)            if (descape != 0)
4492              {              {
4493              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4494                {                {
4495                ptr = oldptr;                ptr = oldptr;
4496                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4497                }                }
4498              }              }
4499            }            }
4500    
4501          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4502          one-character ranges */          one-character ranges. */
4503    
4504          if (d < c)          if (d < c)
4505            {            {
4506            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4507            goto FAILED;            goto FAILED;
4508            }            }
4509            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4510    
4511            /* We have found a character range, so single character optimizations
4512            cannot be done anymore. Any value greater than 1 indicates that there
4513            is more than one character. */
4514    
4515          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          class_one_char = 2;
4516    
4517          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4518    
4519          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4520    
4521          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4522          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. */  
4523    
4524  #ifdef SUPPORT_UCP          continue;   /* Go get the next char in the class */
4525  #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;  
                 }  
4526    
4527                if (occ == ocd)        /* Handle a single character - we can get here for a normal non-escape
4528                  {        char, or after \ that introduces a single character or for an apparent
4529                  *class_uchardata++ = XCL_SINGLE;        range that isn't. Only the value 1 matters for class_one_char, so don't
4530                  }        increase it if it is already 2 or more ... just in case there's a class
4531                else        with a zillion characters in it. */
4532                  {  
4533                  *class_uchardata++ = XCL_RANGE;        CLASS_SINGLE_CHARACTER:
4534                  class_uchardata += PRIV(ord2utf)(occ, class_uchardata);        if (class_one_char < 2) class_one_char++;
4535                  }  
4536                class_uchardata += PRIV(ord2utf)(ocd, class_uchardata);        /* If class_one_char is 1, we have the first single character in the
4537                }        class, and there have been no prior ranges, or XCLASS items generated by
4538              }        escapes. If this is the final character in the class, we can optimize by
4539  #endif  /* SUPPORT_UCP */        turning the item into a 1-character OP_CHAR[I] if it's positive, or
4540          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4541          to be set. Otherwise, there can be no first char if this item is first,
4542          whatever repeat count may follow. In the case of reqchar, save the
4543          previous value for reinstating. */
4544    
4545            /* Now record the original range, possibly modified for UCP caseless        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
4546            overlapping ranges. */          {
4547            ptr++;
4548            zeroreqchar = reqchar;
4549            zeroreqcharflags = reqcharflags;
4550    
4551            *class_uchardata++ = XCL_RANGE;          if (negate_class)
4552  #ifdef SUPPORT_UTF            {
4553  #ifndef COMPILE_PCRE8  #ifdef SUPPORT_UCP
4554            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);  
4555  #endif  #endif
4556  #else /* SUPPORT_UTF */            if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4557            *class_uchardata++ = c;            zerofirstchar = firstchar;
4558            *class_uchardata++ = d;            zerofirstcharflags = firstcharflags;
 #endif /* SUPPORT_UTF */  
4559    
4560            /* With UCP support, we are done. Without UCP support, there is no            /* For caseless UTF-8 mode when UCP support is available, check
4561            caseless matching for UTF characters > 127; we can use the bit map            whether this character has more than one other case. If so, generate
4562            for the smaller ones. As for 16 bit characters without UTF, we            a special OP_NOTPROP item instead of OP_NOTI. */
           can still use  */  
4563    
4564  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4565  #ifndef COMPILE_PCRE8            if (utf && (options & PCRE_CASELESS) != 0 &&
4566            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)  
4567              {              {
4568              if ((options & PCRE_CASELESS) == 0 || c > 127) continue;              *code++ = OP_NOTPROP;
4569              /* Adjust upper limit and fall through to set up the map */              *code++ = PT_CLIST;
4570              d = 127;              *code++ = d;
4571              }              }
4572            else            else
4573              {  #endif
4574              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. */  
4575    
         if (lengthptr == NULL) for (; c <= d; c++)  
           {  
           classbits[c/8] |= (1 << (c&7));  
           if ((options & PCRE_CASELESS) != 0)  
4576              {              {
4577              int uc = cd->fcc[c]; /* flip case */              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4578              classbits[uc/8] |= (1 << (uc&7));  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4579              }              if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4580            }                code += PRIV(ord2utf)(c, code);
4581                else
         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. */  
       if (class_single_char < 2) class_single_char++;  
   
       /* If class_charcount is 1, we saw precisely one character. As long as  
       there were no negated characters >= 128 and there was no use of \p or \P,  
       in other words, no use of any XCLASS features, we can optimize.  
   
       In UTF-8 mode, we can optimize the negative case only if there were no  
       characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR  
       operate on single-bytes characters only. This is an historical hangover.  
       Maybe one day we can tidy these opcodes to handle multi-byte characters.  
   
       The optimization throws away the bit map. We turn the item into a  
       1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
       Note that OP_NOT[I] does not support multibyte characters. In the positive  
       case, it can cause firstchar to be set. Otherwise, there can be no first  
       char if this item is first, whatever repeat count may follow. In the case  
       of reqchar, save the previous value for reinstating. */  
   
 #ifdef SUPPORT_UTF  
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET  
         && (!utf || !negate_class || c < (MAX_VALUE_FOR_SINGLE_CHAR + 1)))  
 #else  
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)  
4582  #endif  #endif
4583          {                *code++ = c;
4584          ptr++;              }
         zeroreqchar = reqchar;  
4585    
4586          /* The OP_NOT[I] opcodes work on single characters only. */            /* We are finished with this character class */
4587    
4588          if (negate_class)            goto END_CLASS;
           {  
           if (firstchar == REQ_UNSET) firstchar = REQ_NONE;  
           zerofirstchar = firstchar;  
           *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
           *code++ = c;  
           goto NOT_CHAR;  
4589            }            }
4590    
4591          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4592          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4593    
4594  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4595          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4596            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4597          else          else
# Line 4536  for (;; ptr++) Line 4603  for (;; ptr++)
4603          goto ONE_CHAR;          goto ONE_CHAR;
4604          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4605    
4606        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4607          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 */  
4608    
4609        /* Handle a single-byte character */        class_has_8bitchar +=
4610          {          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));  
           }  
         }  
4611        }        }
4612    
4613      /* 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.
4614      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
4615      string. */      string. */
4616    
4617      while (((c = *(++ptr)) != 0 ||      while (((c = *(++ptr)) != CHAR_NULL ||
4618             (nestptr != NULL &&             (nestptr != NULL &&
4619               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != CHAR_NULL)) &&
4620             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
4621    
4622      /* Check for missing terminating ']' */      /* Check for missing terminating ']' */
4623    
4624      if (c == 0)      if (c == CHAR_NULL)
4625        {        {
4626        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
4627        goto FAILED;        goto FAILED;
4628        }        }
4629    
4630        /* We will need an XCLASS if data has been placed in class_uchardata. In
4631        the second phase this is a sufficient test. However, in the pre-compile
4632        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4633        only if the very last character in the class needs XCLASS will it contain
4634        anything at this point. For this reason, xclass gets set TRUE above when
4635        uchar_classdata is emptied, and that's why this code is the way it is here
4636        instead of just doing a test on class_uchardata below. */
4637    
4638    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4639        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4640    #endif
4641    
4642      /* 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
4643      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4644      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
4645    
4646      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4647      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4648        zerofirstcharflags = firstcharflags;
4649      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4650        zeroreqcharflags = reqcharflags;
4651    
4652      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4653      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 4674  for (;; ptr++) Line 4703  for (;; ptr++)
4703        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4704        }        }
4705      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4706      NOT_CHAR:  
4707        END_CLASS:
4708      break;      break;
4709    
4710    
# Line 4712  for (;; ptr++) Line 4742  for (;; ptr++)
4742      if (repeat_min == 0)      if (repeat_min == 0)
4743        {        {
4744        firstchar = zerofirstchar;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4745          firstcharflags = zerofirstcharflags;
4746        reqchar = zeroreqchar;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4747          reqcharflags = zeroreqcharflags;
4748        }        }
4749    
4750      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4775  for (;; ptr++) Line 4807  for (;; ptr++)
4807    
4808      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4809    
4810      /* If previous was a character match, abolish the item and generate a      /* If previous was a character or negated character match, abolish the item
4811      repeat item instead. If a char item has a minumum of more than one, ensure      and generate a repeat item instead. If a char item has a minimum of more
4812      that it is set in reqchar - it might not be if a sequence such as x{3} is      than one, ensure that it is set in reqchar - it might not be if a sequence
4813      the first thing in a branch because the x will have gone into firstchar      such as x{3} is the first thing in a branch because the x will have gone
4814      instead.  */      into firstchar instead.  */
4815    
4816      if (*previous == OP_CHAR || *previous == OP_CHARI)      if (*previous == OP_CHAR || *previous == OP_CHARI
4817        {          || *previous == OP_NOT || *previous == OP_NOTI)
4818        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;        {
4819          switch (*previous)
4820            {
4821            default: /* Make compiler happy. */
4822            case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
4823            case OP_CHARI: op_type = OP_STARI - OP_STAR; break;
4824            case OP_NOT:   op_type = OP_NOTSTAR - OP_STAR; break;
4825            case OP_NOTI:  op_type = OP_NOTSTARI - OP_STAR; break;
4826            }
4827    
4828        /* Deal with UTF characters that take up more than one character. It's        /* Deal with UTF characters that take up more than one character. It's
4829        easier to write this out separately than try to macrify it. Use c to        easier to write this out separately than try to macrify it. Use c to
4830        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
4831        it's a length rather than a small character. */        it's a length rather than a small character. */
4832    
4833  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4834        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4835          {          {
4836          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4806  for (;; ptr++) Line 4846  for (;; ptr++)
4846        with UTF disabled, or for a single character UTF character. */        with UTF disabled, or for a single character UTF character. */
4847          {          {
4848          c = code[-1];          c = code[-1];
4849          if (repeat_min > 1) reqchar = c | req_caseopt | cd->req_varyopt;          if (*previous <= OP_CHARI && repeat_min > 1)
4850              {
4851              reqchar = c;
4852              reqcharflags = req_caseopt | cd->req_varyopt;
4853              }
4854          }          }
4855    
4856        /* 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 4825  for (;; ptr++) Line 4869  for (;; ptr++)
4869        goto OUTPUT_SINGLE_REPEAT;   /* Code shared with single character types */        goto OUTPUT_SINGLE_REPEAT;   /* Code shared with single character types */
4870        }        }
4871    
     /* If previous was a single negated character ([^a] or similar), we use  
     one of the special opcodes, replacing it. The code is shared with single-  
     character repeats by setting opt_type to add a suitable offset into  
     repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI  
     are currently used only for single-byte chars. */  
   
     else if (*previous == OP_NOT || *previous == OP_NOTI)  
       {  
       op_type = ((*previous == OP_NOT)? OP_NOTSTAR : OP_NOTSTARI) - OP_STAR;  
       c = previous[1];  
       if (!possessive_quantifier &&  
           repeat_max < 0 &&  
           check_auto_possessive(previous, utf, ptr + 1, options, cd))  
         {  
         repeat_type = 0;    /* Force greedy */  
         possessive_quantifier = TRUE;  
         }  
       goto OUTPUT_SINGLE_REPEAT;  
       }  
   
4872      /* If previous was a character type match (\d or similar), abolish it and      /* If previous was a character type match (\d or similar), abolish it and
4873      create a suitable repeat item. The code is shared with single-character      create a suitable repeat item. The code is shared with single-character
4874      repeats by setting op_type to add a suitable offset into repeat_type. Note      repeats by setting op_type to add a suitable offset into repeat_type. Note
# Line 4883  for (;; ptr++) Line 4907  for (;; ptr++)
4907    
4908        if (repeat_max == 0) goto END_REPEAT;        if (repeat_max == 0) goto END_REPEAT;
4909    
       /*--------------------------------------------------------------------*/  
       /* 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; */  
       /*--------------------------------------------------------------------*/  
   
4910        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
4911    
4912        repeat_type += op_type;        repeat_type += op_type;
# Line 4945  for (;; ptr++) Line 4959  for (;; ptr++)
4959    
4960          if (repeat_max < 0)          if (repeat_max < 0)
4961            {            {
4962  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4963            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4964              {              {
4965              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4970  for (;; ptr++) Line 4984  for (;; ptr++)
4984    
4985          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4986            {            {
4987  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4988            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4989              {              {
4990              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5000  for (;; ptr++) Line 5014  for (;; ptr++)
5014    
5015        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
5016    
5017  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5018        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
5019          {          {
5020          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5039  for (;; ptr++) Line 5053  for (;; ptr++)
5053          goto END_REPEAT;          goto END_REPEAT;
5054          }          }
5055    
       /*--------------------------------------------------------------------*/  
       /* 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; */  
       /*--------------------------------------------------------------------*/  
   
5056        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
5057          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
5058        else if (repeat_min == 1 && repeat_max == -1)        else if (repeat_min == 1 && repeat_max == -1)
# Line 5208  for (;; ptr++) Line 5212  for (;; ptr++)
5212    
5213            else            else
5214              {              {
5215              if (groupsetfirstchar && reqchar < 0) reqchar = firstchar;              if (groupsetfirstchar && reqcharflags < 0)
5216                  {
5217                  reqchar = firstchar;
5218                  reqcharflags = firstcharflags;
5219                  }
5220    
5221              for (i = 1; i < repeat_min; i++)              for (i = 1; i < repeat_min; i++)
5222                {                {
# Line 5583  for (;; ptr++) Line 5591  for (;; ptr++)
5591        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
5592          {          {
5593          arg = ++ptr;          arg = ++ptr;
5594          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5595          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5596            if ((unsigned int)arglen > MAX_MARK)
5597              {
5598              *errorcodeptr = ERR75;
5599              goto FAILED;
5600              }
5601          }          }
5602    
5603        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5600  for (;; ptr++) Line 5613  for (;; ptr++)
5613          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5614              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5615            {            {
5616              int setverb;
5617    
5618            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5619            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5620    
# Line 5617  for (;; ptr++) Line 5632  for (;; ptr++)
5632                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5633                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5634                }                }
5635              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5636                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5637    
5638              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5639              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
5640              }              }
5641    
5642            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5632  for (;; ptr++) Line 5648  for (;; ptr++)
5648                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5649                goto FAILED;                goto FAILED;
5650                }                }
5651              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5652              }              }
5653    
5654            else            else
# Line 5643  for (;; ptr++) Line 5658  for (;; ptr++)
5658                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5659                goto FAILED;                goto FAILED;
5660                }                }
5661              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5662              *code++ = arglen;              *code++ = arglen;
5663              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5664              code += arglen;              code += arglen;
5665              *code++ = 0;              *code++ = 0;
5666              }              }
5667    
5668              switch (setverb)
5669                {
5670                case OP_THEN:
5671                case OP_THEN_ARG:
5672                cd->external_flags |= PCRE_HASTHEN;
5673                break;
5674    
5675                case OP_PRUNE:
5676                case OP_PRUNE_ARG:
5677                case OP_SKIP:
5678                case OP_SKIP_ARG:
5679                cd->had_pruneorskip = TRUE;
5680                break;
5681                }
5682    
5683            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5684            }            }
5685    
# Line 5676  for (;; ptr++) Line 5705  for (;; ptr++)
5705          {          {
5706          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */
5707          ptr++;          ptr++;
5708          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5709          if (*ptr == 0)          if (*ptr == CHAR_NULL)
5710            {            {
5711            *errorcodeptr = ERR18;            *errorcodeptr = ERR18;
5712            goto FAILED;            goto FAILED;
# Line 5700  for (;; ptr++) Line 5729  for (;; ptr++)
5729          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5730          case CHAR_LEFT_PARENTHESIS:          case CHAR_LEFT_PARENTHESIS:
5731          bravalue = OP_COND;       /* Conditional group */          bravalue = OP_COND;       /* Conditional group */
5732            tempptr = ptr;
5733    
5734          /* A condition can be an assertion, a number (referring to a numbered          /* A condition can be an assertion, a number (referring to a numbered
5735          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 5712  for (;; ptr++) Line 5742  for (;; ptr++)
5742          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
5743          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.
5744          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
5745          cases. */          cases.
5746    
5747            For compatibility with auto-callouts, we allow a callout to be
5748            specified before a condition that is an assertion. First, check for the
5749            syntax of a callout; if found, adjust the temporary pointer that is
5750            used to check for an assertion condition. That's all that is needed! */
5751    
5752            if (ptr[1] == CHAR_QUESTION_MARK && ptr[2] == CHAR_C)
5753              {
5754              for (i = 3;; i++) if (!IS_DIGIT(ptr[i])) break;
5755              if (ptr[i] == CHAR_RIGHT_PARENTHESIS)
5756                tempptr += i + 1;
5757              }
5758    
5759          /* For conditions that are assertions, check the syntax, and then exit          /* For conditions that are assertions, check the syntax, and then exit
5760          the switch. This will take control down to where bracketed groups,          the switch. This will take control down to where bracketed groups,
5761          including assertions, are processed. */          including assertions, are processed. */
5762    
5763          if (ptr[1] == CHAR_QUESTION_MARK && (ptr[2] == CHAR_EQUALS_SIGN ||          if (tempptr[1] == CHAR_QUESTION_MARK &&
5764              ptr[2] == CHAR_EXCLAMATION_MARK || ptr[2] == CHAR_LESS_THAN_SIGN))                (tempptr[2] == CHAR_EQUALS_SIGN ||
5765                   tempptr[2] == CHAR_EXCLAMATION_MARK ||
5766                   tempptr[2] == CHAR_LESS_THAN_SIGN))
5767            break;            break;
5768    
5769          /* 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 5753  for (;; ptr++) Line 5797  for (;; ptr++)
5797            }            }
5798          else          else
5799            {            {
5800            terminator = 0;            terminator = CHAR_NULL;
5801            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);
5802            }            }
5803    
# Line 5773  for (;; ptr++) Line 5817  for (;; ptr++)
5817          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
5818            {            {
5819            if (recno >= 0)            if (recno >= 0)
5820              recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1;              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;
5821            ptr++;            ptr++;
5822            }            }
5823          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
5824    
5825          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||
5826              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
5827            {            {
5828            ptr--;      /* Error offset */            ptr--;      /* Error offset */
# Line 5843  for (;; ptr++) Line 5887  for (;; ptr++)
5887            code[1+LINK_SIZE]++;            code[1+LINK_SIZE]++;
5888            }            }
5889    
5890          /* If terminator == 0 it means that the name followed directly after          /* If terminator == CHAR_NULL it means that the name followed directly
5891          the opening parenthesis [e.g. (?(abc)...] and in this case there are          after the opening parenthesis [e.g. (?(abc)...] and in this case there
5892          some further alternatives to try. For the cases where terminator != 0          are some further alternatives to try. For the cases where terminator !=
5893          [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have          0 [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have
5894          now checked all the possibilities, so give an error. */          now checked all the possibilities, so give an error. */
5895    
5896          else if (terminator != 0)          else if (terminator != CHAR_NULL)
5897            {            {
5898            *errorcodeptr = ERR15;            *errorcodeptr = ERR15;
5899            goto FAILED;            goto FAILED;
# Line 6018  for (;; ptr++) Line 6062  for (;; ptr++)
6062    
6063            if (lengthptr != NULL)            if (lengthptr != NULL)
6064              {              {
6065              if (*ptr != terminator)              if (*ptr != (pcre_uchar)terminator)
6066                {                {
6067                *errorcodeptr = ERR42;                *errorcodeptr = ERR42;
6068                goto FAILED;                goto FAILED;
# Line 6160  for (;; ptr++) Line 6204  for (;; ptr++)
6204              *errorcodeptr = ERR62;              *errorcodeptr = ERR62;
6205              goto FAILED;              goto FAILED;
6206              }              }
6207            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6208              {              {
6209              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
6210              goto FAILED;              goto FAILED;
# Line 6266  for (;; ptr++) Line 6310  for (;; ptr++)
6310            while(IS_DIGIT(*ptr))            while(IS_DIGIT(*ptr))
6311              recno = recno * 10 + *ptr++ - CHAR_0;              recno = recno * 10 + *ptr++ - CHAR_0;
6312    
6313            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6314              {              {
6315              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
6316              goto FAILED;              goto FAILED;
# Line 6370  for (;; ptr++) Line 6414  for (;; ptr++)
6414    
6415          /* Can't determine a first byte now */          /* Can't determine a first byte now */
6416    
6417          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6418          continue;          continue;
6419    
6420    
# Line 6504  for (;; ptr++) Line 6548  for (;; ptr++)
6548           cond_depth +           cond_depth +
6549             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
6550           &subfirstchar,                   /* For possible first char */           &subfirstchar,                   /* For possible first char */
6551             &subfirstcharflags,
6552           &subreqchar,                     /* For possible last char */           &subreqchar,                     /* For possible last char */
6553             &subreqcharflags,
6554           bcptr,                           /* Current branch chain */           bcptr,                           /* Current branch chain */
6555           cd,                              /* Tables block */           cd,                              /* Tables block */
6556           (lengthptr == NULL)? NULL :      /* Actual compile phase */           (lengthptr == NULL)? NULL :      /* Actual compile phase */
# Line 6565  for (;; ptr++) Line 6611  for (;; ptr++)
6611            *errorcodeptr = ERR27;            *errorcodeptr = ERR27;
6612            goto FAILED;            goto FAILED;
6613            }            }
6614          if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;          if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE;
6615          }          }
6616        }        }
6617    
# Line 6614  for (;; ptr++) Line 6660  for (;; ptr++)
6660      back off. */      back off. */
6661    
6662      zeroreqchar = reqchar;      zeroreqchar = reqchar;
6663        zeroreqcharflags = reqcharflags;
6664      zerofirstchar = firstchar;      zerofirstchar = firstchar;
6665        zerofirstcharflags = firstcharflags;
6666      groupsetfirstchar = FALSE;      groupsetfirstchar = FALSE;
6667    
6668      if (bravalue >= OP_ONCE)      if (bravalue >= OP_ONCE)
# Line 6625  for (;; ptr++) Line 6673  for (;; ptr++)
6673        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
6674        repeat forces firstchar to "none". */        repeat forces firstchar to "none". */
6675    
6676        if (firstchar == REQ_UNSET)        if (firstcharflags == REQ_UNSET)
6677          {          {
6678          if (subfirstchar >= 0)          if (subfirstcharflags >= 0)
6679            {            {
6680            firstchar = subfirstchar;            firstchar = subfirstchar;
6681              firstcharflags = subfirstcharflags;
6682            groupsetfirstchar = TRUE;            groupsetfirstchar = TRUE;
6683            }            }
6684          else firstchar = REQ_NONE;          else firstcharflags = REQ_NONE;
6685          zerofirstchar = REQ_NONE;          zerofirstcharflags = REQ_NONE;
6686          }          }
6687    
6688        /* If firstchar was previously set, convert the subpattern's firstchar        /* If firstchar was previously set, convert the subpattern's firstchar
6689        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
6690        existence beforehand. */        existence beforehand. */
6691    
6692        else if (subfirstchar >= 0 && subreqchar < 0)        else if (subfirstcharflags >= 0 && subreqcharflags < 0)
6693          subreqchar = subfirstchar | tempreqvary;          {
6694            subreqchar = subfirstchar;
6695            subreqcharflags = subfirstcharflags | tempreqvary;
6696            }
6697    
6698        /* 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
6699        really the first byte - see above), set it. */        really the first byte - see above), set it. */
6700    
6701        if (subreqchar >= 0) reqchar = subreqchar;        if (subreqcharflags >= 0)
6702            {
6703            reqchar = subreqchar;
6704            reqcharflags = subreqcharflags;
6705            }
6706        }        }
6707    
6708      /* 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 6657  for (;; ptr++) Line 6713  for (;; ptr++)
6713      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
6714      firstchar, looking for an asserted first char. */      firstchar, looking for an asserted first char. */
6715    
6716      else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;      else if (bravalue == OP_ASSERT && subreqcharflags >= 0)
6717          {
6718          reqchar = subreqchar;
6719          reqcharflags = subreqcharflags;
6720          }
6721      break;     /* End of processing '(' */      break;     /* End of processing '(' */
6722    
6723    
# Line 6665  for (;; ptr++) Line 6725  for (;; ptr++)
6725      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
6726      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
6727      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
6728      are ESC_REF plus the reference number. Only back references and those types      are negative the reference number. Only back references and those types
6729      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
6730      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
6731      ever created. */      ever created. */
6732    
6733      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6734      tempptr = ptr;      tempptr = ptr;
6735      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6736      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6737    
6738      if (c < 0)      if (escape == 0)                  /* The escape coded a single character */
6739          c = ec;
6740        else
6741        {        {
6742        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6743          {          {
6744          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6745            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6685  for (;; ptr++) Line 6747  for (;; ptr++)
6747          continue;          continue;
6748          }          }
6749    
6750        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6751    
6752        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6753        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6754    
6755        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6756          firstchar = REQ_NONE;          firstcharflags = REQ_NONE;
6757    
6758        /* 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. */
6759    
6760        zerofirstchar = firstchar;        zerofirstchar = firstchar;
6761          zerofirstcharflags = firstcharflags;
6762        zeroreqchar = reqchar;        zeroreqchar = reqchar;
6763          zeroreqcharflags = reqcharflags;
6764    
6765        /* \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'
6766        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6767        -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 <
6768        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
6769        -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
6770        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6771    
6772        if (-c == ESC_g)        if (escape == ESC_g)
6773          {          {
6774          const pcre_uchar *p;          const pcre_uchar *p;
6775          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
# Line 6725  for (;; ptr++) Line 6789  for (;; ptr++)
6789          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
6790            {            {
6791            BOOL is_a_number = TRUE;            BOOL is_a_number = TRUE;
6792            for (p = ptr + 1; *p != 0 && *p != terminator; p++)            for (p = ptr + 1; *p != CHAR_NULL && *p != (pcre_uchar)terminator; p++)
6793              {              {
6794              if (!MAX_255(*p)) { is_a_number = FALSE; break; }              if (!MAX_255(*p)) { is_a_number = FALSE; break; }
6795              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;
6796              if ((cd->ctypes[*p] & ctype_word) == 0) break;              if ((cd->ctypes[*p] & ctype_word) == 0) break;
6797              }              }
6798            if (*p != terminator)            if (*p != (pcre_uchar)terminator)
6799              {              {
6800              *errorcodeptr = ERR57;              *errorcodeptr = ERR57;
6801              break;              break;
# Line 6749  for (;; ptr++) Line 6813  for (;; ptr++)
6813    
6814          p = ptr + 2;          p = ptr + 2;
6815          while (IS_DIGIT(*p)) p++;          while (IS_DIGIT(*p)) p++;
6816          if (*p != terminator)          if (*p != (pcre_uchar)terminator)
6817            {            {
6818            *errorcodeptr = ERR57;            *errorcodeptr = ERR57;
6819            break;            break;
# Line 6761  for (;; ptr++) Line 6825  for (;; ptr++)
6825        /* \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).
6826        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6827    
6828        if (-c == ESC_k)        if (escape == ESC_k)
6829          {          {
6830          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6831            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))
# Line 6780  for (;; ptr++) Line 6844  for (;; ptr++)
6844        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
6845        ':' later. */        ':' later. */
6846    
6847        if (-c >= ESC_REF)        if (escape < 0)
6848          {          {
6849          open_capitem *oc;          open_capitem *oc;
6850          recno = -c - ESC_REF;          recno = -escape;
6851    
6852          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6853          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6854          previous = code;          previous = code;
6855          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6856          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
# Line 6810  for (;; ptr++) Line 6874  for (;; ptr++)
6874        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6875    
6876  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6877        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6878          {          {
6879          BOOL negated;          BOOL negated;
6880          int pdata;          unsigned int ptype = 0, pdata = 0;
6881          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
6882          if (ptype < 0) goto FAILED;            goto FAILED;
6883          previous = code;          previous = code;
6884          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6885          *code++ = ptype;          *code++ = ptype;
6886          *code++ = pdata;          *code++ = pdata;
6887          }          }
# Line 6826  for (;; ptr++) Line 6890  for (;; ptr++)
6890        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6891        allowed. */        allowed. */
6892    
6893        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6894          {          {
6895          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6896          goto FAILED;          goto FAILED;
# Line 6836  for (;; ptr++) Line 6900  for (;; ptr++)
6900        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6901        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
6902        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
6903        Unicode property tests. */        Unicode property tests. Note that \b and \B do a one-character
6904          lookbehind, and \A also behaves as if it does. */
6905    
6906        else        else
6907          {          {
6908            if ((escape == ESC_b || escape == ESC_B || escape == ESC_A) &&
6909                 cd->max_lookbehind == 0)
6910              cd->max_lookbehind = 1;
6911  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6912          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6913            {            {
6914            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6915            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6916            }            }
6917          else          else
6918  #endif  #endif
# Line 6852  for (;; ptr++) Line 6920  for (;; ptr++)
6920          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6921    
6922            {            {
6923            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
6924            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
6925            }            }
6926          }          }
6927        continue;        continue;
# Line 6863  for (;; ptr++) Line 6931  for (;; ptr++)
6931      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
6932      handle it as a data character. */      handle it as a data character. */
6933    
6934  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6935      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6936        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6937      else      else
# Line 6896  for (;; ptr++) Line 6964  for (;; ptr++)
6964    
6965      ONE_CHAR:      ONE_CHAR:
6966      previous = code;      previous = code;
6967    
6968        /* For caseless UTF-8 mode when UCP support is available, check whether
6969        this character has more than one other case. If so, generate a special
6970        OP_PROP item instead of OP_CHARI. */
6971    
6972    #ifdef SUPPORT_UCP
6973        if (utf && (options & PCRE_CASELESS) != 0)
6974          {
6975          GETCHAR(c, mcbuffer);
6976          if ((c = UCD_CASESET(c)) != 0)
6977            {
6978            *code++ = OP_PROP;
6979            *code++ = PT_CLIST;
6980            *code++ = c;
6981            if (firstcharflags == REQ_UNSET) firstcharflags = zerofirstcharflags = REQ_NONE;
6982            break;
6983            }
6984          }
6985    #endif
6986    
6987        /* Caseful matches, or not one of the multicase characters. */
6988    
6989      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6990      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6991    
# Line 6909  for (;; ptr++) Line 6999  for (;; ptr++)
6999      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
7000      repeat. */      repeat. */
7001    
7002      if (firstchar == REQ_UNSET)      if (firstcharflags == REQ_UNSET)
7003        {        {
7004        zerofirstchar = REQ_NONE;        zerofirstcharflags = REQ_NONE;
7005        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7006          zeroreqcharflags = reqcharflags;
7007    
7008        /* 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
7009        only if it is not to be matched caselessly. */        only if it is not to be matched caselessly. */
# Line 6920  for (;; ptr++) Line 7011  for (;; ptr++)
7011        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7012          {          {
7013          firstchar = mcbuffer[0] | req_caseopt;          firstchar = mcbuffer[0] | req_caseopt;
7014          if (mclength != 1) reqchar = code[-1] | cd->req_varyopt;          firstchar = mcbuffer[0];
7015            firstcharflags = req_caseopt;
7016    
7017            if (mclength != 1)
7018              {
7019              reqchar = code[-1];
7020              reqcharflags = cd->req_varyopt;
7021              }
7022          }          }
7023        else firstchar = reqchar = REQ_NONE;        else firstcharflags = reqcharflags = REQ_NONE;
7024        }        }
7025    
7026      /* 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 6931  for (;; ptr++) Line 7029  for (;; ptr++)
7029      else      else
7030        {        {
7031        zerofirstchar = firstchar;        zerofirstchar = firstchar;
7032          zerofirstcharflags = firstcharflags;
7033        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7034          zeroreqcharflags = reqcharflags;
7035        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7036          reqchar = code[-1] | req_caseopt | cd->req_varyopt;          {
7037            reqchar = code[-1];
7038            reqcharflags = req_caseopt | cd->req_varyopt;
7039            }
7040        }        }
7041    
7042      break;            /* End of literal character handling */      break;            /* End of literal character handling */
# Line 6952  return FALSE; Line 7055  return FALSE;
7055    
7056    
7057    
   
7058  /*************************************************  /*************************************************
7059  *     Compile sequence of alternatives           *  *     Compile sequence of alternatives           *
7060  *************************************************/  *************************************************/
# Line 6973  Arguments: Line 7075  Arguments:
7075    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
7076    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
7077    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
7078    firstcharptr   place to put the first required character, or a negative number    firstcharptr    place to put the first required character
7079    reqcharptr     place to put the last required character, or a negative number    firstcharflagsptr place to put the first character flags, or a negative number
7080      reqcharptr     place to put the last required character
7081      reqcharflagsptr place to put the last required character flags, or a negative number
7082    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
7083    cd             points to the data block with tables pointers etc.    cd             points to the data block with tables pointers etc.
7084    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
# Line 6986  Returns:         TRUE on success Line 7090  Returns:         TRUE on success
7090  static BOOL  static BOOL
7091  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,
7092    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
7093    int cond_depth, pcre_int32 *firstcharptr, pcre_int32 *reqcharptr,    int cond_depth,
7094      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
7095      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
7096    branch_chain *bcptr, compile_data *cd, int *lengthptr)    branch_chain *bcptr, compile_data *cd, int *lengthptr)
7097  {  {
7098  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
# Line 6996  pcre_uchar *start_bracket = code; Line 7102  pcre_uchar *start_bracket = code;
7102  pcre_uchar *reverse_count = NULL;  pcre_uchar *reverse_count = NULL;
7103  open_capitem capitem;  open_capitem capitem;
7104  int capnumber = 0;  int capnumber = 0;
7105  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7106  pcre_int32 branchfirstchar, branchreqchar;  pcre_int32 firstcharflags, reqcharflags;
7107    pcre_uint32 branchfirstchar, branchreqchar;
7108    pcre_int32 branchfirstcharflags, branchreqcharflags;
7109  int length;  int length;
7110  int orig_bracount;  unsigned int orig_bracount;
7111  int max_bracount;  unsigned int max_bracount;
7112  branch_chain bc;  branch_chain bc;
7113    
7114  bc.outer = bcptr;  bc.outer = bcptr;
7115  bc.current_branch = code;  bc.current_branch = code;
7116    
7117  firstchar = reqchar = REQ_UNSET;  firstchar = reqchar = 0;
7118    firstcharflags = reqcharflags = REQ_UNSET;
7119    
7120  /* 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
7121  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 7066  for (;;) Line 7175  for (;;)
7175    into the length. */    into the length. */
7176    
7177    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,
7178          &branchreqchar, &bc, cond_depth, cd,          &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc,
7179          (lengthptr == NULL)? NULL : &length))          cond_depth, cd, (lengthptr == NULL)? NULL : &length))
7180      {      {
7181      *ptrptr = ptr;      *ptrptr = ptr;
7182      return FALSE;      return FALSE;
# Line 7088  for (;;) Line 7197  for (;;)
7197      if (*last_branch != OP_ALT)      if (*last_branch != OP_ALT)
7198        {        {
7199        firstchar = branchfirstchar;        firstchar = branchfirstchar;
7200          firstcharflags = branchfirstcharflags;
7201        reqchar = branchreqchar;        reqchar = branchreqchar;
7202          reqcharflags = branchreqcharflags;
7203        }        }
7204    
7205      /* 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 7102  for (;;) Line 7213  for (;;)
7213        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
7214        previously no reqchar, it takes on the value of the old firstchar. */        previously no reqchar, it takes on the value of the old firstchar. */
7215    
7216        if (firstchar >= 0 && firstchar != branchfirstchar)        if (firstcharflags >= 0 &&
7217              (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar))
7218          {          {
7219          if (reqchar < 0) reqchar = firstchar;          if (reqcharflags < 0)
7220          firstchar = REQ_NONE;            {
7221              reqchar = firstchar;
7222              reqcharflags = firstcharflags;
7223              }
7224            firstcharflags = REQ_NONE;
7225          }          }
7226    
7227        /* 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
7228        branch becomes a reqchar if there isn't a branch reqchar. */        branch becomes a reqchar if there isn't a branch reqchar. */
7229    
7230        if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0)        if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0)
7231            branchreqchar = branchfirstchar;          {
7232            branchreqchar = branchfirstchar;
7233            branchreqcharflags = branchfirstcharflags;
7234            }
7235    
7236        /* Now ensure that the reqchars match */        /* Now ensure that the reqchars match */
7237    
7238        if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY))        if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) ||
7239          reqchar = REQ_NONE;            reqchar != branchreqchar)
7240        else reqchar |= branchreqchar;   /* To "or" REQ_VARY */          reqcharflags = REQ_NONE;
7241          else
7242            {
7243            reqchar = branchreqchar;
7244            reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */
7245            }
7246        }        }
7247    
7248      /* 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 7147  for (;;) Line 7271  for (;;)
7271          *ptrptr = ptr;          *ptrptr = ptr;
7272          return FALSE;          return FALSE;
7273          }          }
7274        else { PUT(reverse_count, 0, fixed_length); }        else
7275            {
7276            if (fixed_length > cd->max_lookbehind)
7277              cd->max_lookbehind = fixed_length;
7278            PUT(reverse_count, 0, fixed_length);
7279            }
7280        }        }
7281      }      }
7282    
# Line 7209  for (;;) Line 7338  for (;;)
7338      *codeptr = code;      *codeptr = code;
7339      *ptrptr = ptr;      *ptrptr = ptr;
7340      *firstcharptr = firstchar;      *firstcharptr = firstchar;
7341        *firstcharflagsptr = firstcharflags;
7342      *reqcharptr = reqchar;      *reqcharptr = reqchar;
7343        *reqcharflagsptr = reqcharflags;
7344      if (lengthptr != NULL)      if (lengthptr != NULL)
7345        {        {
7346        if (OFLOW_MAX - *lengthptr < length)        if (OFLOW_MAX - *lengthptr < length)
# Line 7279  and the highest back reference was great Line 7410  and the highest back reference was great
7410  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
7411  of the more common cases more precisely.  of the more common cases more precisely.
7412    
7413    ... A second exception is when the .* appears inside an atomic group, because
7414    this prevents the number of characters it matches from being adjusted.
7415    
7416  Arguments:  Arguments:
7417    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7418    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
7419                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7420                    the less precise approach                    the less precise approach
7421    backref_map    the back reference bitmap    cd             points to the compile data block
7422      atomcount      atomic group level
7423    
7424  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7425  */  */
7426    
7427  static BOOL  static BOOL
7428  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7429    unsigned int backref_map)    compile_data *cd, int atomcount)
7430  {  {
7431  do {  do {
7432     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7303  do { Line 7438  do {
7438     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7439         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7440       {       {
7441       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7442       }       }
7443    
7444     /* Capturing brackets */     /* Capturing brackets */
# Line 7313  do { Line 7448  do {
7448       {       {
7449       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7450       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7451       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7452       }       }
7453    
7454     /* Other brackets */     /* Positive forward assertions and conditions */
7455    
7456     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ASSERT || op == OP_COND)
             op == OP_COND)  
7457       {       {
7458       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7459         }
7460    
7461       /* Atomic groups */
7462    
7463       else if (op == OP_ONCE || op == OP_ONCE_NC)
7464         {
7465         if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7466           return FALSE;
7467       }       }
7468    
7469     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7470     it isn't in brackets that are or may be referenced. */     it isn't in brackets that are or may be referenced or inside an atomic
7471       group. */
7472    
7473     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7474               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7475       {       {
7476       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7477             atomcount > 0 || cd->had_pruneorskip)
7478         return FALSE;         return FALSE;
7479       }       }
7480    
7481     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7482    
7483     else if (op != OP_SOD && op != OP_SOM && op != OP_CIRC) return FALSE;     else if (op != OP_SOD && op != OP_SOM && op != OP_CIRC) return FALSE;
7484    
7485     code += GET(code, 1);     code += GET(code, 1);
7486     }     }
7487  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7354  return TRUE; Line 7499  return TRUE;
7499  matching and for non-DOTALL patterns that start with .* (which must start at  matching and for non-DOTALL patterns that start with .* (which must start at
7500  the beginning or after \n). As in the case of is_anchored() (see above), we  the beginning or after \n). As in the case of is_anchored() (see above), we
7501  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7502  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7503    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7504    count, because once again the assumption no longer holds.
7505    
7506  Arguments:  Arguments:
7507    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7508    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
7509                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7510                    the less precise approach                    the less precise approach
7511    backref_map    the back reference bitmap    cd             points to the compile data
7512      atomcount      atomic group level
7513    
7514  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7515  */  */
7516    
7517  static BOOL  static BOOL
7518  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7519    unsigned int backref_map)    compile_data *cd, int atomcount)
7520  {  {
7521  do {  do {
7522     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7394  do { Line 7542  do {
7542         return FALSE;         return FALSE;
7543    
7544         default:     /* Assertion */         default:     /* Assertion */
7545         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7546         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7547         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7548         break;         break;
# Line 7408  do { Line 7556  do {
7556     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7557         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7558       {       {
7559       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7560       }       }
7561    
7562     /* Capturing brackets */     /* Capturing brackets */
# Line 7418  do { Line 7566  do {
7566       {       {
7567       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7568       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7569       if (!is_startline(scode, new_map, backref_map)) return FALSE;       if (!is_startline(scode, new_map, cd, atomcount)) return FALSE;
7570         }
7571    
7572       /* Positive forward assertions */
7573    
7574       else if (op == OP_ASSERT)
7575         {
7576         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7577       }       }
7578    
7579     /* Other brackets */     /* Atomic brackets */
7580    
7581     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)     else if (op == OP_ONCE || op == OP_ONCE_NC)
7582       {       {
7583       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE;
7584       }       }
7585    
7586     /* .* means "start at start or after \n" if it isn't in brackets that     /* .* means "start at start or after \n" if it isn't in atomic brackets or
7587     may be referenced. */     brackets that may be referenced, as long as the pattern does not contain
7588       *PRUNE or *SKIP, because these break the feature. Consider, for example,
7589       /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the
7590       start of a line. */
7591    
7592     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)
7593       {       {
7594       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 ||
7595             atomcount > 0 || cd->had_pruneorskip)
7596           return FALSE;
7597       }       }
7598    
7599     /* Check for explicit circumflex */     /* Check for explicit circumflex; anything else gives a FALSE result. Note
7600       in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC
7601   &n