/[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 964 by ph10, Fri May 4 13:03:39 2012 UTC revision 1231 by ph10, Fri Dec 7 12:05:53 2012 UTC
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  is also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. We do not need to select pcre16_printint.c specially, because the  library. We do not need to select pcre16_printint.c specially, because the
59  COMPILE_PCREx macro will already be appropriately set. */  COMPILE_PCREx macro will already be appropriately set. */
# Line 68  COMPILE_PCREx macro will already be appr Line 68  COMPILE_PCREx macro will already be appr
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 77  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, 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 490  static const char error_texts[] = Line 505  static const char error_texts[] =
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 */    /* 75 */
508    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
509      "character value in \\u.... sequence is too large\0"
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 630  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 657  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 741  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 775  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 784  Otherwise further processing may be requ Line 797  Otherwise further processing may be requ
797  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
798  /* Not alphanumeric */  /* Not alphanumeric */
799  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
800  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
801    
802  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
803  /* Not alphanumeric */  /* Not alphanumeric */
804  else if (c < 'a' || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
805  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
806  #endif  #endif
807    
808  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
# Line 797  else if ((i = escapes[c - 0x48]) != 0) Line 810  else if ((i = escapes[c - 0x48]) != 0)
810  else  else
811    {    {
812    const pcre_uchar *oldptr;    const pcre_uchar *oldptr;
813    BOOL braced, negated;    BOOL braced, negated, overflow;
814      int s;
815    
816    switch (c)    switch (c)
817      {      {
# Line 822  else Line 836  else
836          c = 0;          c = 0;
837          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
838            {            {
839            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
840  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
841            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
842            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 831  else Line 845  else
845            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
846  #endif  #endif
847            }            }
848    
849    #if defined COMPILE_PCRE8
850            if (c > (utf ? 0x10ffff : 0xff))
851    #elif defined COMPILE_PCRE16
852            if (c > (utf ? 0x10ffff : 0xffff))
853    #elif defined COMPILE_PCRE32
854            if (utf && c > 0x10ffff)
855    #endif
856              {
857              *errorcodeptr = ERR76;
858              }
859            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
860          }          }
861        }        }
862      else      else
# Line 857  else Line 883  else
883      (3) For Oniguruma compatibility we also support \g followed by a name or a      (3) For Oniguruma compatibility we also support \g followed by a name or a
884      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
885      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
886      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
887    
888      case CHAR_g:      case CHAR_g:
889      if (isclass) break;      if (isclass) break;
890      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
891        {        {
892        c = -ESC_g;        escape = ESC_g;
893        break;        break;
894        }        }
895    
# Line 872  else Line 898  else
898      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
899        {        {
900        const pcre_uchar *p;        const pcre_uchar *p;
901        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++)
902          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
903        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
904          {          {
905          c = -ESC_k;          escape = ESC_k;
906          break;          break;
907          }          }
908        braced = TRUE;        braced = TRUE;
# Line 892  else Line 918  else
918      else negated = FALSE;      else negated = FALSE;
919    
920      /* The integer range is limited by the machine's int representation. */      /* The integer range is limited by the machine's int representation. */
921      c = 0;      s = 0;
922        overflow = FALSE;
923      while (IS_DIGIT(ptr[1]))      while (IS_DIGIT(ptr[1]))
924        {        {
925        if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */        if (s > INT_MAX / 10 - 1) /* Integer overflow */
926          {          {
927          c = -1;          overflow = TRUE;
928          break;          break;
929          }          }
930        c = c * 10 + *(++ptr) - CHAR_0;        s = s * 10 + (int)(*(++ptr) - CHAR_0);
931        }        }
932      if (((unsigned int)c) > INT_MAX) /* Integer overflow */      if (overflow) /* Integer overflow */
933        {        {
934        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
935          ptr++;          ptr++;
# Line 916  else Line 943  else
943        break;        break;
944        }        }
945    
946      if (c == 0)      if (s == 0)
947        {        {
948        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
949        break;        break;
# Line 924  else Line 951  else
951    
952      if (negated)      if (negated)
953        {        {
954        if (c > bracount)        if (s > bracount)
955          {          {
956          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
957          break;          break;
958          }          }
959        c = bracount - (c - 1);        s = bracount - (s - 1);
960        }        }
961    
962      c = -(ESC_REF + c);      escape = -s;
963      break;      break;
964    
965      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 954  else Line 981  else
981        {        {
982        oldptr = ptr;        oldptr = ptr;
983        /* The integer range is limited by the machine's int representation. */        /* The integer range is limited by the machine's int representation. */
984        c -= CHAR_0;        s = (int)(c -CHAR_0);
985          overflow = FALSE;
986        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
987          {          {
988          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */          if (s > INT_MAX / 10 - 1) /* Integer overflow */
989            {            {
990            c = -1;            overflow = TRUE;
991            break;            break;
992            }            }
993          c = c * 10 + *(++ptr) - CHAR_0;          s = s * 10 + (int)(*(++ptr) - CHAR_0);
994          }          }
995        if (((unsigned int)c) > INT_MAX) /* Integer overflow */        if (overflow) /* Integer overflow */
996          {          {
997          while (IS_DIGIT(ptr[1]))          while (IS_DIGIT(ptr[1]))
998            ptr++;            ptr++;
999          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1000          break;          break;
1001          }          }
1002        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1003          {          {
1004          c = -(ESC_REF + c);          escape = -s;
1005          break;          break;
1006          }          }
1007        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 1020  else Line 1048  else
1048          c = 0;          c = 0;
1049          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1050            {            {
1051            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1052  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1053            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1054            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 1038  else Line 1066  else
1066        const pcre_uchar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1067    
1068        c = 0;        c = 0;
1069          overflow = FALSE;
1070        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1071          {          {
1072          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1073          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1074    
1075    #ifdef COMPILE_PCRE32
1076            if (c >= 0x10000000l) { overflow = TRUE; break; }
1077    #endif
1078    
1079  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1080          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1081          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 1051  else Line 1084  else
1084          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1085  #endif  #endif
1086    
1087  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1088          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1089  #else  #elif defined COMPILE_PCRE16
1090  #ifdef COMPILE_PCRE16          if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1091          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }  #elif defined COMPILE_PCRE32
1092  #endif          if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1093  #endif  #endif
1094          }          }
1095    
1096        if (c < 0)        if (overflow)
1097          {          {
1098          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1099          *errorcodeptr = ERR34;          *errorcodeptr = ERR34;
# Line 1082  else Line 1115  else
1115      c = 0;      c = 0;
1116      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1117        {        {
1118        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1119        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1120  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1121        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 1101  else Line 1134  else
1134    
1135      case CHAR_c:      case CHAR_c:
1136      c = *(++ptr);      c = *(++ptr);
1137      if (c == 0)      if (c == CHAR_NULL)
1138        {        {
1139        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1140        break;        break;
# Line 1141  else Line 1174  else
1174  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1175  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1176    
1177  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1178       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1179    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1180    
1181  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1182    
1183  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1184    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1185    
1186  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1187    
1188  *ptrptr = ptr;  *ptrptr = ptr;
1189  return c;  *chptr = c;
1190    return escape;
1191  }  }
1192    
   
   
1193  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1194  /*************************************************  /*************************************************
1195  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1171  escape sequence. Line 1203  escape sequence.
1203  Argument:  Argument:
1204    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1205    negptr         points to a boolean that is set TRUE for negation else FALSE    negptr         points to a boolean that is set TRUE for negation else FALSE
1206    dptr           points to an int that is set to the detailed property value    ptypeptr       points to an unsigned int that is set to the type value
1207      pdataptr       points to an unsigned int that is set to the detailed property value
1208    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1209    
1210  Returns:         type value from ucp_type_table, or -1 for an invalid type  Returns:         TRUE if the type value was found, or FALSE for an invalid type
1211  */  */
1212    
1213  static int  static BOOL
1214  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1215      unsigned int *pdataptr, int *errorcodeptr)
1216  {  {
1217  int c, i, bot, top;  pcre_uchar c;
1218    int i, bot, top;
1219  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1220  pcre_uchar name[32];  pcre_uchar name[32];
1221    
1222  c = *(++ptr);  c = *(++ptr);
1223  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1224    
1225  *negptr = FALSE;  *negptr = FALSE;
1226    
# Line 1202  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1237  if (c == CHAR_LEFT_CURLY_BRACKET)
1237    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1238      {      {
1239      c = *(++ptr);      c = *(++ptr);
1240      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1241      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1242      name[i] = c;      name[i] = c;
1243      }      }
# Line 1227  top = PRIV(utt_size); Line 1262  top = PRIV(utt_size);
1262    
1263  while (bot < top)  while (bot < top)
1264    {    {
1265      int r;
1266    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1267    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1268    if (c == 0)    if (r == 0)
1269      {      {
1270      *dptr = PRIV(utt)[i].value;      *ptypeptr = PRIV(utt)[i].type;
1271      return PRIV(utt)[i].type;      *pdataptr = PRIV(utt)[i].value;
1272        return TRUE;
1273      }      }
1274    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1275    }    }
1276    
1277  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1278  *ptrptr = ptr;  *ptrptr = ptr;
1279  return -1;  return FALSE;
1280    
1281  ERROR_RETURN:  ERROR_RETURN:
1282  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1283  *ptrptr = ptr;  *ptrptr = ptr;
1284  return -1;  return FALSE;
1285  }  }
1286  #endif  #endif
1287    
# Line 1279  int max = -1; Line 1316  int max = -1;
1316  /* Read the minimum value and do a paranoid check: a negative value indicates  /* Read the minimum value and do a paranoid check: a negative value indicates
1317  an integer overflow. */  an integer overflow. */
1318    
1319  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1320  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1321    {    {
1322    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1294  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1331  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1331    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1332      {      {
1333      max = 0;      max = 0;
1334      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1335      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1336        {        {
1337        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1349  Arguments: Line 1386  Arguments:
1386    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1387    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1388    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1389    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1390    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1391    
1392  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
# Line 1395  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1432  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1432    
1433    else if (ptr[2] == CHAR_NUMBER_SIGN)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1434      {      {
1435      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;      for (ptr += 3; *ptr != CHAR_NULL; ptr++)
1436          if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1437      goto FAIL_EXIT;      goto FAIL_EXIT;
1438      }      }
1439    
# Line 1408  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1446  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1446      ptr += 2;      ptr += 2;
1447      if (ptr[1] != CHAR_QUESTION_MARK)      if (ptr[1] != CHAR_QUESTION_MARK)
1448        {        {
1449        while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;        while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1450        if (*ptr != 0) ptr++;        if (*ptr != CHAR_NULL) ptr++;
1451        }        }
1452      }      }
1453    
# Line 1425  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1463  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1463      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1464          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1465        {        {
1466        int term;        pcre_uchar term;
1467        const pcre_uchar *thisname;        const pcre_uchar *thisname;
1468        *count += 1;        *count += 1;
1469        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
# Line 1433  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1471  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1471        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1472        thisname = ptr;        thisname = ptr;
1473        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1474        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1475            STRNCMP_UC_UC(name, thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1476          return *count;          return *count;
1477        term++;        term++;
1478        }        }
# Line 1452  for (; ptr < cd->end_pattern; ptr++) Line 1490  for (; ptr < cd->end_pattern; ptr++)
1490    
1491    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1492      {      {
1493      if (*(++ptr) == 0) goto FAIL_EXIT;      if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1494      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1495        {        {
1496        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1497        if (*ptr == 0) goto FAIL_EXIT;        if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1498        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1499        }        }
1500      continue;      continue;
# Line 1500  for (; ptr < cd->end_pattern; ptr++) Line 1538  for (; ptr < cd->end_pattern; ptr++)
1538    
1539      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1540        {        {
1541        if (*ptr == 0) return -1;        if (*ptr == CHAR_NULL) return -1;
1542        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1543          {          {
1544          if (*(++ptr) == 0) goto FAIL_EXIT;          if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1545          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1546            {            {
1547            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1548            if (*ptr == 0) goto FAIL_EXIT;            if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1549            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1550            }            }
1551          continue;          continue;
# Line 1521  for (; ptr < cd->end_pattern; ptr++) Line 1559  for (; ptr < cd->end_pattern; ptr++)
1559    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1560      {      {
1561      ptr++;      ptr++;
1562      while (*ptr != 0)      while (*ptr != CHAR_NULL)
1563        {        {
1564        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1565        ptr++;        ptr++;
# Line 1529  for (; ptr < cd->end_pattern; ptr++) Line 1567  for (; ptr < cd->end_pattern; ptr++)
1567        if (utf) FORWARDCHAR(ptr);        if (utf) FORWARDCHAR(ptr);
1568  #endif  #endif
1569        }        }
1570      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1571      continue;      continue;
1572      }      }
1573    
# Line 1539  for (; ptr < cd->end_pattern; ptr++) Line 1577  for (; ptr < cd->end_pattern; ptr++)
1577      {      {
1578      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1579      if (rc > 0) return rc;      if (rc > 0) return rc;
1580      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1581      }      }
1582    
1583    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
# Line 1583  Arguments: Line 1621  Arguments:
1621    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1622    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1623    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1624    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1625    
1626  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1627  */  */
# Line 1604  matching closing parens. That is why we Line 1642  matching closing parens. That is why we
1642  for (;;)  for (;;)
1643    {    {
1644    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1645    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == CHAR_NULL) break;
1646    }    }
1647    
1648  return rc;  return rc;
# Line 1686  and doing the check at the end; a flag s Line 1724  and doing the check at the end; a flag s
1724    
1725  Arguments:  Arguments:
1726    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1727    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1728    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1729    cd       the "compile data" structure    cd       the "compile data" structure
1730    
# Line 1712  for (;;) Line 1750  for (;;)
1750    {    {
1751    int d;    int d;
1752    pcre_uchar *ce, *cs;    pcre_uchar *ce, *cs;
1753    register int op = *cc;    register pcre_uchar op = *cc;
1754    
1755    switch (op)    switch (op)
1756      {      {
# Line 1832  for (;;) Line 1870  for (;;)
1870      case OP_EXACTI:      case OP_EXACTI:
1871      case OP_NOTEXACT:      case OP_NOTEXACT:
1872      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1873      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1874      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1875  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
1876      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
# Line 1841  for (;;) Line 1879  for (;;)
1879    
1880      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1881      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1882      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)
1883          cc += 2;
1884      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1885      break;      break;
1886    
# Line 1876  for (;;) Line 1915  for (;;)
1915    
1916      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1917    
 #if defined SUPPORT_UTF || defined COMPILE_PCRE16  
     case OP_XCLASS:  
     cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];  
     /* Fall through */  
 #endif  
   
1918      case OP_CLASS:      case OP_CLASS:
1919      case OP_NCLASS:      case OP_NCLASS:
1920    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1921        case OP_XCLASS:
1922        /* The original code caused an unsigned overflow in 64 bit systems,
1923        so now we use a conditional statement. */
1924        if (op == OP_XCLASS)
1925          cc += GET(cc, 1);
1926        else
1927          cc += PRIV(OP_lengths)[OP_CLASS];
1928    #else
1929      cc += PRIV(OP_lengths)[OP_CLASS];      cc += PRIV(OP_lengths)[OP_CLASS];
1930    #endif
1931    
1932      switch (*cc)      switch (*cc)
1933        {        {
# Line 1899  for (;;) Line 1942  for (;;)
1942        case OP_CRRANGE:        case OP_CRRANGE:
1943        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1944        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1945        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1946        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
1947        break;        break;
1948    
# Line 2015  length. Line 2058  length.
2058    
2059  Arguments:  Arguments:
2060    code        points to start of expression    code        points to start of expression
2061    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2062    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2063    
2064  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 2026  PRIV(find_bracket)(const pcre_uchar *cod Line 2069  PRIV(find_bracket)(const pcre_uchar *cod
2069  {  {
2070  for (;;)  for (;;)
2071    {    {
2072    register int c = *code;    register pcre_uchar c = *code;
2073    
2074    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2075    
# Line 2049  for (;;) Line 2092  for (;;)
2092    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2093             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2094      {      {
2095      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2096      if (n == number) return (pcre_uchar *)code;      if (n == number) return (pcre_uchar *)code;
2097      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
2098      }      }
# Line 2079  for (;;) Line 2122  for (;;)
2122        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2123        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2124        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2125        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2126          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2127        break;        break;
2128    
2129        case OP_MARK:        case OP_MARK:
# Line 2102  for (;;) Line 2145  for (;;)
2145    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
2146    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2147    
2148  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2149      if (utf) switch(c)      if (utf) switch(c)
2150        {        {
2151        case OP_CHAR:        case OP_CHAR:
# Line 2154  instance of OP_RECURSE. Line 2197  instance of OP_RECURSE.
2197    
2198  Arguments:  Arguments:
2199    code        points to start of expression    code        points to start of expression
2200    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2201    
2202  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
2203  */  */
# Line 2164  find_recurse(const pcre_uchar *code, BOO Line 2207  find_recurse(const pcre_uchar *code, BOO
2207  {  {
2208  for (;;)  for (;;)
2209    {    {
2210    register int c = *code;    register pcre_uchar c = *code;
2211    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2212    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2213    
# Line 2199  for (;;) Line 2242  for (;;)
2242        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2243        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2244        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2245        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2246          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2247        break;        break;
2248    
2249        case OP_MARK:        case OP_MARK:
# Line 2222  for (;;) Line 2265  for (;;)
2265      by a multi-byte character. The length in the table is a minimum, so we have      by a multi-byte character. The length in the table is a minimum, so we have
2266      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2267    
2268  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2269      if (utf) switch(c)      if (utf) switch(c)
2270        {        {
2271        case OP_CHAR:        case OP_CHAR:
# Line 2308  bracket whose current branch will alread Line 2351  bracket whose current branch will alread
2351  Arguments:  Arguments:
2352    code        points to start of search    code        points to start of search
2353    endcode     points to where to stop    endcode     points to where to stop
2354    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2355    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2356    
2357  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2318  static BOOL Line 2361  static BOOL
2361  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2362    BOOL utf, compile_data *cd)    BOOL utf, compile_data *cd)
2363  {  {
2364  register int c;  register pcre_uchar c;
2365  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2366       code < endcode;       code < endcode;
2367       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
# Line 2352  for (code = first_significant_code(code Line 2395  for (code = first_significant_code(code
2395      /* Test for forward reference */      /* Test for forward reference */
2396    
2397      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2398        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2399    
2400      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2401    
# Line 2525  for (code = first_significant_code(code Line 2568  for (code = first_significant_code(code
2568      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2569      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2570      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2571      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2572        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2573      break;      break;
2574    
2575      /* End of branch */      /* End of branch */
# Line 2541  for (code = first_significant_code(code Line 2584  for (code = first_significant_code(code
2584      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2585      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2586    
2587  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2588      case OP_STAR:      case OP_STAR:
2589      case OP_STARI:      case OP_STARI:
2590      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2607  Arguments: Line 2650  Arguments:
2650    code        points to start of the recursion    code        points to start of the recursion
2651    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2652    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2653    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2654    cd          pointers to tables etc    cd          pointers to tables etc
2655    
2656  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2673  Returns:   TRUE or FALSE Line 2716  Returns:   TRUE or FALSE
2716  static BOOL  static BOOL
2717  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2718  {  {
2719  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2720  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2721  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
2722    {    {
2723    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2724      ptr++;      ptr++;
# Line 2722  register int yield = 0; Line 2765  register int yield = 0;
2765  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2766    {    {
2767    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2768      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2769    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2770    yield++;    yield++;
2771    }    }
# Line 2754  value in the reference (which is a group Line 2797  value in the reference (which is a group
2797  Arguments:  Arguments:
2798    group      points to the start of the group    group      points to the start of the group
2799    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2800    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2801    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2802    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
2803    
# Line 2777  while ((ptr = (pcre_uchar *)find_recurse Line 2820  while ((ptr = (pcre_uchar *)find_recurse
2820    
2821    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2822      {      {
2823      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2824      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2825        {        {
2826        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2790  while ((ptr = (pcre_uchar *)find_recurse Line 2833  while ((ptr = (pcre_uchar *)find_recurse
2833    
2834    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2835      {      {
2836      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2837      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2838      }      }
2839    
# Line 2858  PUT(previous_callout, 2 + LINK_SIZE, len Line 2901  PUT(previous_callout, 2 + LINK_SIZE, len
2901  *************************************************/  *************************************************/
2902    
2903  /* 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
2904  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
2905  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
2906  start address.  start address. A character with multiple other cases is returned on its own
2907    with a special return value.
2908    
2909  Arguments:  Arguments:
2910    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2868  Arguments: Line 2912  Arguments:
2912    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2913    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2914    
2915  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2916                   0 when a range is returned
2917                  >0 the CASESET offset for char with multiple other cases
2918                    in this case, ocptr contains the original
2919  */  */
2920    
2921  static BOOL  static int
2922  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2923    unsigned int *odptr)    pcre_uint32 *odptr)
2924  {  {
2925  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2926    unsigned int co;
2927    
2928    /* Find the first character that has an other case. If it has multiple other
2929    cases, return its case offset value. */
2930    
2931  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2932    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2933      if ((co = UCD_CASESET(c)) != 0)
2934        {
2935        *ocptr = c++;   /* Character that has the set */
2936        *cptr = c;      /* Rest of input range */
2937        return (int)co;
2938        }
2939      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2940      }
2941    
2942  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2943    
2944  *ocptr = othercase;  *ocptr = othercase;
2945  next = othercase + 1;  next = othercase + 1;
# Line 2891  for (++c; c <= d; c++) Line 2950  for (++c; c <= d; c++)
2950    next++;    next++;
2951    }    }
2952    
2953  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2954  *cptr = c;  *cptr = c;             /* Rest of input range */
2955    return 0;
 return TRUE;  
2956  }  }
2957    
2958    
# Line 2916  Returns:       TRUE if auto-possessifyin Line 2974  Returns:       TRUE if auto-possessifyin
2974  */  */
2975    
2976  static BOOL  static BOOL
2977  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)
2978  {  {
2979    #ifdef SUPPORT_UCP
2980    const pcre_uint32 *p;
2981    #endif
2982    
2983  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2984    
2985  switch(ptype)  switch(ptype)
2986    {    {
2987    case PT_LAMP:    case PT_LAMP:
# Line 2956  switch(ptype) Line 3019  switch(ptype)
3019    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3020            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3021            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3022    
3023    #ifdef SUPPORT_UCP
3024      case PT_CLIST:
3025      p = PRIV(ucd_caseless_sets) + prop->caseset;
3026      for (;;)
3027        {
3028        if (c < *p) return !negated;
3029        if (c == *p++) return negated;
3030        }
3031      break;  /* Control never reaches here */
3032    #endif
3033    }    }
3034    
3035  return FALSE;  return FALSE;
3036  }  }
3037  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2973  sense to automatically possessify the re Line 3048  sense to automatically possessify the re
3048    
3049  Arguments:  Arguments:
3050    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3051    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3052    ptr           next character in pattern    ptr           next character in pattern
3053    options       options bits    options       options bits
3054    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2985  static BOOL Line 3060  static BOOL
3060  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3061    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3062  {  {
3063  pcre_int32 c, next;  pcre_uint32 c = NOTACHAR;
3064  int op_code = *previous++;  pcre_uint32 next;
3065    int escape;
3066    pcre_uchar op_code = *previous++;
3067    
3068  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3069    
# Line 2998  if ((options & PCRE_EXTENDED) != 0) Line 3075  if ((options & PCRE_EXTENDED) != 0)
3075      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3076        {        {
3077        ptr++;        ptr++;
3078        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3079          {          {
3080          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3081          ptr++;          ptr++;
# Line 3017  value is a character, a negative value i Line 3094  value is a character, a negative value i
3094  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3095    {    {
3096    int temperrorcode = 0;    int temperrorcode = 0;
3097    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3098    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3099    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3100    }    }
3101  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
3102    {    {
3103      escape = 0;
3104  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3105    if (utf) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3106  #endif  #endif
# Line 3040  if ((options & PCRE_EXTENDED) != 0) Line 3118  if ((options & PCRE_EXTENDED) != 0)
3118      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3119        {        {
3120        ptr++;        ptr++;
3121        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3122          {          {
3123          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3124          ptr++;          ptr++;
# Line 3059  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3137  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3137    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)
3138      return FALSE;      return FALSE;
3139    
3140  /* 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. */  
3141    
3142  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3143        op_code == OP_NOT || op_code == OP_NOTI)
3144    {    {
   case OP_CHAR:  
3145  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3146    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3147  #else  #else
3148    c = *previous;    c = *previous;
3149  #endif  #endif
3150    return c != next;    }
3151    
3152    /* 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
3153    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3154    
3155    case OP_CHARI:  if (escape == 0)
3156  #ifdef SUPPORT_UTF    {
3157    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3158  #else    case, which maps to the special PT_CLIST property. Check this first. */
3159    c = *previous;  
3160    #ifdef SUPPORT_UCP
3161      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3162        {
3163        unsigned int ocs = UCD_CASESET(next);
3164        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3165        }
3166  #endif  #endif
3167    if (c == next) return FALSE;  
3168  #ifdef SUPPORT_UTF    switch(op_code)
   if (utf)  
3169      {      {
3170      unsigned int othercase;      case OP_CHAR:
3171      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3172    
3173        /* For CHARI (caseless character) we must check the other case. If we have
3174        Unicode property support, we can use it to test the other case of
3175        high-valued characters. We know that next can have only one other case,
3176        because multi-other-case characters are dealt with above. */
3177    
3178        case OP_CHARI:
3179        if (c == next) return FALSE;
3180    #ifdef SUPPORT_UTF
3181        if (utf)
3182          {
3183          pcre_uint32 othercase;
3184          if (next < 128) othercase = cd->fcc[next]; else
3185  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3186      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3187  #else  #else
3188      othercase = NOTACHAR;        othercase = NOTACHAR;
3189  #endif  #endif
3190      return (unsigned int)c != othercase;        return c != othercase;
3191      }        }
3192    else      else
3193  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3194    return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3195    
3196    case OP_NOT:      case OP_NOT:
3197  #ifdef SUPPORT_UTF      return c == next;
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   return c == next;  
3198    
3199    case OP_NOTI:      case OP_NOTI:
3200  #ifdef SUPPORT_UTF      if (c == next) return TRUE;
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   if (c == next) return TRUE;  
3201  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3202    if (utf)      if (utf)
3203      {        {
3204      unsigned int othercase;        pcre_uint32 othercase;
3205      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3206  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3207      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3208  #else  #else
3209      othercase = NOTACHAR;        othercase = NOTACHAR;
3210  #endif  #endif
3211      return (unsigned int)c == othercase;        return c == othercase;
3212      }        }
3213    else      else
3214  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3215    return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3216    
3217    /* 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.
3218    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. */
3219    
3220    case OP_DIGIT:      case OP_DIGIT:
3221    return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3222    
3223    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3224    return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3225    
3226    case OP_WHITESPACE:      case OP_WHITESPACE:
3227    return next > 255 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3228    
3229    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3230    return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3231    
3232    case OP_WORDCHAR:      case OP_WORDCHAR:
3233    return next > 255 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3234    
3235    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3236    return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3237    
3238    case OP_HSPACE:      case OP_HSPACE:
3239    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3240    switch(next)      switch(next)
3241      {        {
3242      case 0x09:        HSPACE_CASES:
3243      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;  
     }  
3244    
3245    case OP_ANYNL:        default:
3246    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3247    case OP_NOT_VSPACE:        }
3248    switch(next)  
3249      {      case OP_ANYNL:
3250      case 0x0a:      case OP_VSPACE:
3251      case 0x0b:      case OP_NOT_VSPACE:
3252      case 0x0c:      switch(next)
3253      case 0x0d:        {
3254      case 0x85:        VSPACE_CASES:
3255      case 0x2028:        return op_code == OP_NOT_VSPACE;
3256      case 0x2029:  
3257      return op_code == OP_NOT_VSPACE;        default:
3258      default:        return op_code != OP_NOT_VSPACE;
3259      return op_code != OP_NOT_VSPACE;        }
     }  
3260    
3261  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3262    case OP_PROP:      case OP_PROP:
3263    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3264    
3265    case OP_NOTPROP:      case OP_NOTPROP:
3266    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3267  #endif  #endif
3268    
3269    default:      default:
3270    return FALSE;      return FALSE;
3271        }
3272    }    }
3273    
   
3274  /* 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
3275  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
3276  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 3219  switch(op_code) Line 3281  switch(op_code)
3281    {    {
3282    case OP_CHAR:    case OP_CHAR:
3283    case OP_CHARI:    case OP_CHARI:
3284  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3285      {      {
3286      case ESC_d:      case ESC_d:
3287      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
# Line 3248  switch(op_code) Line 3305  switch(op_code)
3305      case ESC_H:      case ESC_H:
3306      switch(c)      switch(c)
3307        {        {
3308        case 0x09:        HSPACE_CASES:
3309        case 0x20:        return escape != ESC_h;
3310        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;  
3311        default:        default:
3312        return -next == ESC_h;        return escape == ESC_h;
3313        }        }
3314    
3315      case ESC_v:      case ESC_v:
3316      case ESC_V:      case ESC_V:
3317      switch(c)      switch(c)
3318        {        {
3319        case 0x0a:        VSPACE_CASES:
3320        case 0x0b:        return escape != ESC_v;
3321        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3322        default:        default:
3323        return -next == ESC_v;        return escape == ESC_v;
3324        }        }
3325    
3326      /* 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
3327      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3328      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3329    
3330  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3331      case ESC_du:      case ESC_du:
# Line 3301  switch(op_code) Line 3336  switch(op_code)
3336      case ESC_SU:      case ESC_SU:
3337        {        {
3338        int temperrorcode = 0;        int temperrorcode = 0;
3339        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3340        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3341        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3342        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3343        }        }
# Line 3311  switch(op_code) Line 3346  switch(op_code)
3346      case ESC_p:      case ESC_p:
3347      case ESC_P:      case ESC_P:
3348        {        {
3349        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3350          int errorcodeptr;
3351        BOOL negated;        BOOL negated;
3352    
3353        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3354        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3355        if (ptype < 0) return FALSE;          return FALSE;
3356        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3357    
3358        /* 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 3329  switch(op_code) Line 3365  switch(op_code)
3365    
3366        /* Do the property check. */        /* Do the property check. */
3367    
3368        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3369        }        }
3370  #endif  #endif
3371    
# Line 3344  switch(op_code) Line 3380  switch(op_code)
3380    these op-codes are never generated.) */    these op-codes are never generated.) */
3381    
3382    case OP_DIGIT:    case OP_DIGIT:
3383    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3384           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3385    
3386    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3387    return next == -ESC_d;    return escape == ESC_d;
3388    
3389    case OP_WHITESPACE:    case OP_WHITESPACE:
3390    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3391    
3392    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3393    return next == -ESC_s || next == -ESC_h || next == -ESC_v || next == -ESC_R;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3394    
3395    case OP_HSPACE:    case OP_HSPACE:
3396    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3397           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3398    
3399    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3400    return next == -ESC_h;    return escape == ESC_h;
3401    
3402    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3403    case OP_ANYNL:    case OP_ANYNL:
3404    case OP_VSPACE:    case OP_VSPACE:
3405    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3406    
3407      case OP_NOT_VSPACE:
3408      return escape == ESC_v || escape == ESC_R;
3409    
3410      case OP_WORDCHAR:
3411      return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3412             escape == ESC_v || escape == ESC_R;
3413    
3414      case OP_NOT_WORDCHAR:
3415      return escape == ESC_w || escape == ESC_d;
3416    
3417      default:
3418      return FALSE;
3419      }
3420    
3421    /* Control does not reach here */
3422    }
3423    
3424    
3425    
3426    /*************************************************
3427    *        Add a character or range to a class     *
3428    *************************************************/
3429    
3430    /* This function packages up the logic of adding a character or range of
3431    characters to a class. The character values in the arguments will be within the
3432    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3433    mutually recursive with the function immediately below.
3434    
3435    Arguments:
3436      classbits     the bit map for characters < 256
3437      uchardptr     points to the pointer for extra data
3438      options       the options word
3439      cd            contains pointers to tables etc.
3440      start         start of range character
3441      end           end of range character
3442    
3443    Returns:        the number of < 256 characters added
3444                    the pointer to extra data is updated
3445    */
3446    
3447    static int
3448    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3449      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3450    {
3451    pcre_uint32 c;
3452    int n8 = 0;
3453    
3454    /* If caseless matching is required, scan the range and process alternate
3455    cases. In Unicode, there are 8-bit characters that have alternate cases that
3456    are greater than 255 and vice-versa. Sometimes we can just extend the original
3457    range. */
3458    
3459    if ((options & PCRE_CASELESS) != 0)
3460      {
3461    #ifdef SUPPORT_UCP
3462      if ((options & PCRE_UTF8) != 0)
3463        {
3464        int rc;
3465        pcre_uint32 oc, od;
3466    
3467        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3468        c = start;
3469    
3470        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3471          {
3472          /* Handle a single character that has more than one other case. */
3473    
3474          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3475            PRIV(ucd_caseless_sets) + rc, oc);
3476    
3477          /* Do nothing if the other case range is within the original range. */
3478    
3479          else if (oc >= start && od <= end) continue;
3480    
3481          /* Extend the original range if there is overlap, noting that if oc < c, we
3482          can't have od > end because a subrange is always shorter than the basic
3483          range. Otherwise, use a recursive call to add the additional range. */
3484    
3485          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3486          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3487          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3488          }
3489        }
3490      else
3491    #endif  /* SUPPORT_UCP */
3492    
3493      /* Not UTF-mode, or no UCP */
3494    
3495      for (c = start; c <= end && c < 256; c++)
3496        {
3497        SETBIT(classbits, cd->fcc[c]);
3498        n8++;
3499        }
3500      }
3501    
3502    /* Now handle the original range. Adjust the final value according to the bit
3503    length - this means that the same lists of (e.g.) horizontal spaces can be used
3504    in all cases. */
3505    
3506    #if defined COMPILE_PCRE8
3507    #ifdef SUPPORT_UTF
3508      if ((options & PCRE_UTF8) == 0)
3509    #endif
3510      if (end > 0xff) end = 0xff;
3511    
3512    #elif defined COMPILE_PCRE16
3513    #ifdef SUPPORT_UTF
3514      if ((options & PCRE_UTF16) == 0)
3515    #endif
3516      if (end > 0xffff) end = 0xffff;
3517    
3518    #endif /* COMPILE_PCRE[8|16] */
3519    
3520    /* If all characters are less than 256, use the bit map. Otherwise use extra
3521    data. */
3522    
3523    if (end < 0x100)
3524      {
3525      for (c = start; c <= end; c++)
3526        {
3527        n8++;
3528        SETBIT(classbits, c);
3529        }
3530      }
3531    
3532    else
3533      {
3534      pcre_uchar *uchardata = *uchardptr;
3535    
3536    #ifdef SUPPORT_UTF
3537      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3538        {
3539        if (start < end)
3540          {
3541          *uchardata++ = XCL_RANGE;
3542          uchardata += PRIV(ord2utf)(start, uchardata);
3543          uchardata += PRIV(ord2utf)(end, uchardata);
3544          }
3545        else if (start == end)
3546          {
3547          *uchardata++ = XCL_SINGLE;
3548          uchardata += PRIV(ord2utf)(start, uchardata);
3549          }
3550        }
3551      else
3552    #endif  /* SUPPORT_UTF */
3553    
3554      /* Without UTF support, character values are constrained by the bit length,
3555      and can only be > 256 for 16-bit and 32-bit libraries. */
3556    
3557    #ifdef COMPILE_PCRE8
3558        {}
3559    #else
3560      if (start < end)
3561        {
3562        *uchardata++ = XCL_RANGE;
3563        *uchardata++ = start;
3564        *uchardata++ = end;
3565        }
3566      else if (start == end)
3567        {
3568        *uchardata++ = XCL_SINGLE;
3569        *uchardata++ = start;
3570        }
3571    #endif
3572    
3573      *uchardptr = uchardata;   /* Updata extra data pointer */
3574      }
3575    
3576    return n8;    /* Number of 8-bit characters */
3577    }
3578    
3579    
3580    
3581    
3582    /*************************************************
3583    *        Add a list of characters to a class     *
3584    *************************************************/
3585    
3586    /* This function is used for adding a list of case-equivalent characters to a
3587    class, and also for adding a list of horizontal or vertical whitespace. If the
3588    list is in order (which it should be), ranges of characters are detected and
3589    handled appropriately. This function is mutually recursive with the function
3590    above.
3591    
3592    Arguments:
3593      classbits     the bit map for characters < 256
3594      uchardptr     points to the pointer for extra data
3595      options       the options word
3596      cd            contains pointers to tables etc.
3597      p             points to row of 32-bit values, terminated by NOTACHAR
3598      except        character to omit; this is used when adding lists of
3599                      case-equivalent characters to avoid including the one we
3600                      already know about
3601    
3602    Returns:        the number of < 256 characters added
3603                    the pointer to extra data is updated
3604    */
3605    
3606    static int
3607    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3608      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3609    {
3610    int n8 = 0;
3611    while (p[0] < NOTACHAR)
3612      {
3613      int n = 0;
3614      if (p[0] != except)
3615        {
3616        while(p[n+1] == p[0] + n + 1) n++;
3617        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3618        }
3619      p += n + 1;
3620      }
3621    return n8;
3622    }
3623    
3624    
3625    
3626    /*************************************************
3627    *    Add characters not in a list to a class     *
3628    *************************************************/
3629    
3630    case OP_NOT_VSPACE:  /* This function is used for adding the complement of a list of horizontal or
3631    return next == -ESC_v || next == -ESC_R;  vertical whitespace to a class. The list must be in order.
3632    
3633    case OP_WORDCHAR:  Arguments:
3634    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    classbits     the bit map for characters < 256
3635           next == -ESC_v || next == -ESC_R;    uchardptr     points to the pointer for extra data
3636      options       the options word
3637      cd            contains pointers to tables etc.
3638      p             points to row of 32-bit values, terminated by NOTACHAR
3639    
3640    case OP_NOT_WORDCHAR:  Returns:        the number of < 256 characters added
3641    return next == -ESC_w || next == -ESC_d;                  the pointer to extra data is updated
3642    */
3643    
3644    default:  static int
3645    return FALSE;  add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3646      int options, compile_data *cd, const pcre_uint32 *p)
3647    {
3648    BOOL utf = (options & PCRE_UTF8) != 0;
3649    int n8 = 0;
3650    if (p[0] > 0)
3651      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3652    while (p[0] < NOTACHAR)
3653      {
3654      while (p[1] == p[0] + 1) p++;
3655      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3656        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3657      p++;
3658    }    }
3659    return n8;
 /* Control does not reach here */  
3660  }  }
3661    
3662    
# Line 3402  Arguments: Line 3676  Arguments:
3676    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3677    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3678    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3679    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3680    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3681      reqcharptr     place to put the last required character
3682      reqcharflagsptr place to put the last required character flags, or a negative number
3683    bcptr          points to current branch chain    bcptr          points to current branch chain
3684    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3685    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3416  Returns:         TRUE on success Line 3692  Returns:         TRUE on success
3692    
3693  static BOOL  static BOOL
3694  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3695    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3696    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3697      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3698      branch_chain *bcptr, int cond_depth,
3699    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3700  {  {
3701  int repeat_type, op_type;  int repeat_type, op_type;
3702  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3703  int bravalue = 0;  int bravalue = 0;
3704  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3705  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
3706  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
3707    pcre_uint32 zeroreqchar, zerofirstchar;
3708    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3709  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3710  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3711  int after_manual_callout = 0;  int after_manual_callout = 0;
3712  int length_prevgroup = 0;  int length_prevgroup = 0;
3713  register int c;  register pcre_uint32 c;
3714    int escape;
3715  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3716  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3717  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3450  must not do this for other options (e.g. Line 3731  must not do this for other options (e.g.
3731  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3732    
3733  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3734  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3735  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3736    #ifndef COMPILE_PCRE32
3737  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3738    #endif
3739  #else  #else
3740  BOOL utf = FALSE;  BOOL utf = FALSE;
3741  #endif  #endif
3742    
3743  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3744    class_uchardata always so that it can be passed to add_to_class() always,
3745    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3746    alternative calls for the different cases. */
3747    
3748    pcre_uchar *class_uchardata;
3749  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3750  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3751  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3752  #endif  #endif
3753    
# Line 3484  to take the zero repeat into account. Th Line 3770  to take the zero repeat into account. Th
3770  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3771  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3772    
3773  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3774    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3775    
3776  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
3777  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 3505  for (;; ptr++) Line 3792  for (;; ptr++)
3792    BOOL is_recurse;    BOOL is_recurse;
3793    BOOL reset_bracount;    BOOL reset_bracount;
3794    int class_has_8bitchar;    int class_has_8bitchar;
3795    int class_single_char;    int class_one_char;
3796    int newoptions;    int newoptions;
3797    int recno;    int recno;
3798    int refsign;    int refsign;
3799    int skipbytes;    int skipbytes;
3800    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
3801    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
3802    int terminator;    int terminator;
3803    int mclength;    unsigned int mclength;
3804    int tempbracount;    unsigned int tempbracount;
3805      pcre_uint32 ec;
3806    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3807    
3808    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3524  for (;; ptr++) Line 3812  for (;; ptr++)
3812    /* 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
3813    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
3814    
3815    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
3816      {      {
3817      ptr = nestptr;      ptr = nestptr;
3818      nestptr = NULL;      nestptr = NULL;
# Line 3599  for (;; ptr++) Line 3887  for (;; ptr++)
3887    
3888    /* 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 */
3889    
3890    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
3891      {      {
3892      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3893        {        {
# Line 3647  for (;; ptr++) Line 3935  for (;; ptr++)
3935      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3936        {        {
3937        ptr++;        ptr++;
3938        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3939          {          {
3940          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3941          ptr++;          ptr++;
# Line 3655  for (;; ptr++) Line 3943  for (;; ptr++)
3943          if (utf) FORWARDCHAR(ptr);          if (utf) FORWARDCHAR(ptr);
3944  #endif  #endif
3945          }          }
3946        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
3947    
3948        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
3949        c = 0;        c = 0;
# Line 3677  for (;; ptr++) Line 3965  for (;; ptr++)
3965      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3966      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3967      *firstcharptr = firstchar;      *firstcharptr = firstchar;
3968        *firstcharflagsptr = firstcharflags;
3969      *reqcharptr = reqchar;      *reqcharptr = reqchar;
3970        *reqcharflagsptr = reqcharflags;
3971      *codeptr = code;      *codeptr = code;
3972      *ptrptr = ptr;      *ptrptr = ptr;
3973      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3701  for (;; ptr++) Line 3991  for (;; ptr++)
3991      previous = NULL;      previous = NULL;
3992      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3993        {        {
3994        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3995        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3996        }        }
3997      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3716  for (;; ptr++) Line 4006  for (;; ptr++)
4006      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4007    
4008      case CHAR_DOT:      case CHAR_DOT:
4009      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4010      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4011        zerofirstcharflags = firstcharflags;
4012      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4013        zeroreqcharflags = reqcharflags;
4014      previous = code;      previous = code;
4015      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4016      break;      break;
# Line 3792  for (;; ptr++) Line 4084  for (;; ptr++)
4084          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4085        {        {
4086        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4087        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4088        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4089          zerofirstcharflags = firstcharflags;
4090        break;        break;
4091        }        }
4092    
# Line 3803  for (;; ptr++) Line 4096  for (;; ptr++)
4096    
4097      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4098    
4099      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4100      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 <
4101      < 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
4102      a single character. */      character. */
4103    
4104      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4105      class_single_char = 0;      class_one_char = 0;
4106    
4107      /* 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
4108      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
4109      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
4110      */      map. */
4111    
4112      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4113    
4114  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4115      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4116      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4117      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4118  #endif  #endif
4119    
4120      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4121      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
4122      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4123    
4124      if (c != 0) do      if (c != CHAR_NULL) do
4125        {        {
4126        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4127    
# Line 3843  for (;; ptr++) Line 4136  for (;; ptr++)
4136        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4137        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
4138        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4139        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4140          however. */
4141    
4142        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4143          {          {
4144            xclass = TRUE;
4145          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4146          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4147          }          }
# Line 3948  for (;; ptr++) Line 4243  for (;; ptr++)
4243              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4244            }            }
4245    
4246          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4247          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4248    
4249          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3964  for (;; ptr++) Line 4259  for (;; ptr++)
4259            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4260    
4261          ptr = tempptr + 1;          ptr = tempptr + 1;
4262          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4263          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4264          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4265          class_single_char = 2;          class_one_char = 2;
4266          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4267          }          }
4268    
# Line 3975  for (;; ptr++) Line 4270  for (;; ptr++)
4270        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
4271        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
4272        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4273        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4274        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4275        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4276        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4277    
4278        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4279          {          {
4280          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4281    
4282          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4283    
4284          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4285          else if (-c == ESC_N)            /* \N is not supported in a class */            c = ec;
4286            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4287            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 4000  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 4020  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 4044  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 4058  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 4239  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 4262  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 4284  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 4295  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 4311  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 4332  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
4582    #endif
4583                  *code++ = c;
4584              }              }
           }  
   
         continue;   /* Go get the next char in the class */  
         }  
4585    
4586        /* Handle a lone single character - we can get here for a normal            /* We are finished with this character class */
       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 was no use of \p or \P, in other words, no use of any XCLASS  
       features, we can optimize.  
   
       The optimization throws away the bit map. We turn the item into a  
       1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
       In the positive case, it can cause firstchar to be set. Otherwise, there  
       can be no first char if this item is first, whatever repeat count may  
       follow. In the case of reqchar, save the previous value for reinstating. */  
4587    
4588        if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)            goto END_CLASS;
         {  
         ptr++;  
         zeroreqchar = reqchar;  
   
         if (negate_class)  
           {  
           if (firstchar == REQ_UNSET) firstchar = REQ_NONE;  
           zerofirstchar = firstchar;  
           *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
 #ifdef SUPPORT_UTF  
           if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)  
             code += PRIV(ord2utf)(c, code);  
           else  
 #endif  
             *code++ = c;  
           goto NOT_CHAR;  
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 4565  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 4703  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 4741  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 4813  for (;; ptr++) Line 4816  for (;; ptr++)
4816      if (*previous == OP_CHAR || *previous == OP_CHARI      if (*previous == OP_CHAR || *previous == OP_CHARI
4817          || *previous == OP_NOT || *previous == OP_NOTI)          || *previous == OP_NOT || *previous == OP_NOTI)
4818        {        {
4819        switch (*previous)        switch (*previous)
4820          {          {
4821          default: /* Make compiler happy. */          default: /* Make compiler happy. */
4822          case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;          case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
# Line 4827  for (;; ptr++) Line 4830  for (;; ptr++)
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 4844  for (;; ptr++) Line 4847  for (;; ptr++)
4847          {          {
4848          c = code[-1];          c = code[-1];
4849          if (*previous <= OP_CHARI && repeat_min > 1)          if (*previous <= OP_CHARI && repeat_min > 1)
4850            reqchar = c | req_caseopt | cd->req_varyopt;            {
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 4901  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 4963  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 4988  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 5018  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 5057  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 5226  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 5593  for (;; ptr++) Line 5583  for (;; ptr++)
5583        ptr++;        ptr++;
5584        while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;        while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;
5585        namelen = (int)(ptr - name);        namelen = (int)(ptr - name);
5586    
5587        /* It appears that Perl allows any characters whatsoever, other than        /* It appears that Perl allows any characters whatsoever, other than
5588        a closing parenthesis, to appear in arguments, so we no longer insist on        a closing parenthesis, to appear in arguments, so we no longer insist on
5589        letters, digits, and underscores. */        letters, digits, and underscores. */
# Line 5601  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 (arglen > (int)MAX_MARK)          if ((unsigned int)arglen > MAX_MARK)
5597            {            {
5598            *errorcodeptr = ERR75;            *errorcodeptr = ERR75;
5599            goto FAILED;            goto FAILED;
5600            }            }
5601          }          }
5602    
5603        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5623  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 5640  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 5655  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 5666  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 5699  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 5776  for (;; ptr++) Line 5782  for (;; ptr++)
5782            }            }
5783          else          else
5784            {            {
5785            terminator = 0;            terminator = CHAR_NULL;
5786            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);
5787            }            }
5788    
# Line 5796  for (;; ptr++) Line 5802  for (;; ptr++)
5802          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
5803            {            {
5804            if (recno >= 0)            if (recno >= 0)
5805              recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1;              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;
5806            ptr++;            ptr++;
5807            }            }
5808          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
5809    
5810          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||
5811              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
5812            {            {
5813            ptr--;      /* Error offset */            ptr--;      /* Error offset */
# Line 5866  for (;; ptr++) Line 5872  for (;; ptr++)
5872            code[1+LINK_SIZE]++;            code[1+LINK_SIZE]++;
5873            }            }
5874    
5875          /* If terminator == 0 it means that the name followed directly after          /* If terminator == CHAR_NULL it means that the name followed directly
5876          the opening parenthesis [e.g. (?(abc)...] and in this case there are          after the opening parenthesis [e.g. (?(abc)...] and in this case there
5877          some further alternatives to try. For the cases where terminator != 0          are some further alternatives to try. For the cases where terminator !=
5878          [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have          0 [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have
5879          now checked all the possibilities, so give an error. */          now checked all the possibilities, so give an error. */
5880    
5881          else if (terminator != 0)          else if (terminator != CHAR_NULL)
5882            {            {
5883            *errorcodeptr = ERR15;            *errorcodeptr = ERR15;
5884            goto FAILED;            goto FAILED;
# Line 6041  for (;; ptr++) Line 6047  for (;; ptr++)
6047    
6048            if (lengthptr != NULL)            if (lengthptr != NULL)
6049              {              {
6050              if (*ptr != terminator)              if (*ptr != (pcre_uchar)terminator)
6051                {                {
6052                *errorcodeptr = ERR42;                *errorcodeptr = ERR42;
6053                goto FAILED;                goto FAILED;
# Line 6183  for (;; ptr++) Line 6189  for (;; ptr++)
6189              *errorcodeptr = ERR62;              *errorcodeptr = ERR62;
6190              goto FAILED;              goto FAILED;
6191              }              }
6192            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6193              {              {
6194              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
6195              goto FAILED;              goto FAILED;
# Line 6289  for (;; ptr++) Line 6295  for (;; ptr++)
6295            while(IS_DIGIT(*ptr))            while(IS_DIGIT(*ptr))
6296              recno = recno * 10 + *ptr++ - CHAR_0;              recno = recno * 10 + *ptr++ - CHAR_0;
6297    
6298            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6299              {              {
6300              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
6301              goto FAILED;              goto FAILED;
# Line 6393  for (;; ptr++) Line 6399  for (;; ptr++)
6399    
6400          /* Can't determine a first byte now */          /* Can't determine a first byte now */
6401    
6402          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6403          continue;          continue;
6404    
6405    
# Line 6527  for (;; ptr++) Line 6533  for (;; ptr++)
6533           cond_depth +           cond_depth +
6534             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
6535           &subfirstchar,                   /* For possible first char */           &subfirstchar,                   /* For possible first char */
6536             &subfirstcharflags,
6537           &subreqchar,                     /* For possible last char */           &subreqchar,                     /* For possible last char */
6538             &subreqcharflags,
6539           bcptr,                           /* Current branch chain */           bcptr,                           /* Current branch chain */
6540           cd,                              /* Tables block */           cd,                              /* Tables block */
6541           (lengthptr == NULL)? NULL :      /* Actual compile phase */           (lengthptr == NULL)? NULL :      /* Actual compile phase */
# Line 6588  for (;; ptr++) Line 6596  for (;; ptr++)
6596            *errorcodeptr = ERR27;            *errorcodeptr = ERR27;
6597            goto FAILED;            goto FAILED;
6598            }            }
6599          if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;          if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE;
6600          }          }
6601        }        }
6602    
# Line 6637  for (;; ptr++) Line 6645  for (;; ptr++)
6645      back off. */      back off. */
6646    
6647      zeroreqchar = reqchar;      zeroreqchar = reqchar;
6648        zeroreqcharflags = reqcharflags;
6649      zerofirstchar = firstchar;      zerofirstchar = firstchar;
6650        zerofirstcharflags = firstcharflags;
6651      groupsetfirstchar = FALSE;      groupsetfirstchar = FALSE;
6652    
6653      if (bravalue >= OP_ONCE)      if (bravalue >= OP_ONCE)
# Line 6648  for (;; ptr++) Line 6658  for (;; ptr++)
6658        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
6659        repeat forces firstchar to "none". */        repeat forces firstchar to "none". */
6660    
6661        if (firstchar == REQ_UNSET)        if (firstcharflags == REQ_UNSET)
6662          {          {
6663          if (subfirstchar >= 0)          if (subfirstcharflags >= 0)
6664            {            {
6665            firstchar = subfirstchar;            firstchar = subfirstchar;
6666              firstcharflags = subfirstcharflags;
6667            groupsetfirstchar = TRUE;            groupsetfirstchar = TRUE;
6668            }            }
6669          else firstchar = REQ_NONE;          else firstcharflags = REQ_NONE;
6670          zerofirstchar = REQ_NONE;          zerofirstcharflags = REQ_NONE;
6671          }          }
6672    
6673        /* If firstchar was previously set, convert the subpattern's firstchar        /* If firstchar was previously set, convert the subpattern's firstchar
6674        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
6675        existence beforehand. */        existence beforehand. */
6676    
6677        else if (subfirstchar >= 0 && subreqchar < 0)        else if (subfirstcharflags >= 0 && subreqcharflags < 0)
6678          subreqchar = subfirstchar | tempreqvary;          {
6679            subreqchar = subfirstchar;
6680            subreqcharflags = subfirstcharflags | tempreqvary;
6681            }
6682    
6683        /* 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
6684        really the first byte - see above), set it. */        really the first byte - see above), set it. */
6685    
6686        if (subreqchar >= 0) reqchar = subreqchar;        if (subreqcharflags >= 0)
6687            {
6688            reqchar = subreqchar;
6689            reqcharflags = subreqcharflags;
6690            }
6691        }        }
6692    
6693      /* 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 6680  for (;; ptr++) Line 6698  for (;; ptr++)
6698      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
6699      firstchar, looking for an asserted first char. */      firstchar, looking for an asserted first char. */
6700    
6701      else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;      else if (bravalue == OP_ASSERT && subreqcharflags >= 0)
6702          {
6703          reqchar = subreqchar;
6704          reqcharflags = subreqcharflags;
6705          }
6706      break;     /* End of processing '(' */      break;     /* End of processing '(' */
6707    
6708    
# Line 6688  for (;; ptr++) Line 6710  for (;; ptr++)
6710      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
6711      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
6712      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
6713      are ESC_REF plus the reference number. Only back references and those types      are negative the reference number. Only back references and those types
6714      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
6715      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
6716      ever created. */      ever created. */
6717    
6718      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6719      tempptr = ptr;      tempptr = ptr;
6720      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6721    
6722      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6723    
6724      if (c < 0)      if (escape == 0)
6725          c = ec;
6726        else
6727        {        {
6728        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6729          {          {
6730          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6731            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6708  for (;; ptr++) Line 6733  for (;; ptr++)
6733          continue;          continue;
6734          }          }
6735    
6736        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6737    
6738        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6739        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6740    
6741        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6742          firstchar = REQ_NONE;          firstcharflags = REQ_NONE;
6743    
6744        /* 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. */
6745    
6746        zerofirstchar = firstchar;        zerofirstchar = firstchar;
6747          zerofirstcharflags = firstcharflags;
6748        zeroreqchar = reqchar;        zeroreqchar = reqchar;
6749          zeroreqcharflags = reqcharflags;
6750    
6751        /* \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'
6752        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6753        -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 <
6754        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
6755        -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
6756        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6757    
6758        if (-c == ESC_g)        if (escape == ESC_g)
6759          {          {
6760          const pcre_uchar *p;          const pcre_uchar *p;
6761          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
# Line 6748  for (;; ptr++) Line 6775  for (;; ptr++)
6775          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
6776            {            {
6777            BOOL is_a_number = TRUE;            BOOL is_a_number = TRUE;
6778            for (p = ptr + 1; *p != 0 && *p != terminator; p++)            for (p = ptr + 1; *p != CHAR_NULL && *p != (pcre_uchar)terminator; p++)
6779              {              {
6780              if (!MAX_255(*p)) { is_a_number = FALSE; break; }              if (!MAX_255(*p)) { is_a_number = FALSE; break; }
6781              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;
6782              if ((cd->ctypes[*p] & ctype_word) == 0) break;              if ((cd->ctypes[*p] & ctype_word) == 0) break;
6783              }              }
6784            if (*p != terminator)            if (*p != (pcre_uchar)terminator)
6785              {              {
6786              *errorcodeptr = ERR57;              *errorcodeptr = ERR57;
6787              break;              break;
# Line 6772  for (;; ptr++) Line 6799  for (;; ptr++)
6799    
6800          p = ptr + 2;          p = ptr + 2;
6801          while (IS_DIGIT(*p)) p++;          while (IS_DIGIT(*p)) p++;
6802          if (*p != terminator)          if (*p != (pcre_uchar)terminator)
6803            {            {
6804            *errorcodeptr = ERR57;            *errorcodeptr = ERR57;
6805            break;            break;
# Line 6784  for (;; ptr++) Line 6811  for (;; ptr++)
6811        /* \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).
6812        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6813    
6814        if (-c == ESC_k)        if (escape == ESC_k)
6815          {          {
6816          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6817            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))
# Line 6803  for (;; ptr++) Line 6830  for (;; ptr++)
6830        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
6831        ':' later. */        ':' later. */
6832    
6833        if (-c >= ESC_REF)        if (escape < 0)
6834          {          {
6835          open_capitem *oc;          open_capitem *oc;
6836          recno = -c - ESC_REF;          recno = -escape;
6837    
6838          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6839          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6840          previous = code;          previous = code;
6841          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6842          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
# Line 6833  for (;; ptr++) Line 6860  for (;; ptr++)
6860        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6861    
6862  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6863        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6864          {          {
6865          BOOL negated;          BOOL negated;
6866          int pdata;          unsigned int ptype = 0, pdata = 0;
6867          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
6868          if (ptype < 0) goto FAILED;            goto FAILED;
6869          previous = code;          previous = code;
6870          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6871          *code++ = ptype;          *code++ = ptype;
6872          *code++ = pdata;          *code++ = pdata;
6873          }          }
# Line 6849  for (;; ptr++) Line 6876  for (;; ptr++)
6876        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6877        allowed. */        allowed. */
6878    
6879        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6880          {          {
6881          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6882          goto FAILED;          goto FAILED;
# Line 6859  for (;; ptr++) Line 6886  for (;; ptr++)
6886        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6887        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
6888        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
6889        Unicode property tests. Note that \b and \B do a one-character        Unicode property tests. Note that \b and \B do a one-character
6890        lookbehind. */        lookbehind. */
6891    
6892        else        else
6893          {          {
6894          if ((-c == ESC_b || -c == ESC_B) && cd->max_lookbehind == 0)          if ((escape == ESC_b || escape == ESC_B) && cd->max_lookbehind == 0)
6895            cd->max_lookbehind = 1;            cd->max_lookbehind = 1;
6896  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6897          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6898            {            {
6899            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6900            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6901            }            }
6902          else          else
6903  #endif  #endif
# Line 6878  for (;; ptr++) Line 6905  for (;; ptr++)
6905          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6906    
6907            {            {
6908            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
6909            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
6910            }            }
6911          }          }
6912        continue;        continue;
# Line 6889  for (;; ptr++) Line 6916  for (;; ptr++)
6916      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
6917      handle it as a data character. */      handle it as a data character. */
6918    
6919  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6920      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6921        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6922      else      else
# Line 6922  for (;; ptr++) Line 6949  for (;; ptr++)
6949    
6950      ONE_CHAR:      ONE_CHAR:
6951      previous = code;      previous = code;
6952    
6953        /* For caseless UTF-8 mode when UCP support is available, check whether
6954        this character has more than one other case. If so, generate a special
6955        OP_PROP item instead of OP_CHARI. */
6956    
6957    #ifdef SUPPORT_UCP
6958        if (utf && (options & PCRE_CASELESS) != 0)
6959          {
6960          GETCHAR(c, mcbuffer);
6961          if ((c = UCD_CASESET(c)) != 0)
6962            {
6963            *code++ = OP_PROP;
6964            *code++ = PT_CLIST;
6965            *code++ = c;
6966            if (firstcharflags == REQ_UNSET) firstcharflags = zerofirstcharflags = REQ_NONE;
6967            break;
6968            }
6969          }
6970    #endif
6971    
6972        /* Caseful matches, or not one of the multicase characters. */
6973    
6974      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6975      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6976    
# Line 6935  for (;; ptr++) Line 6984  for (;; ptr++)
6984      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
6985      repeat. */      repeat. */
6986    
6987      if (firstchar == REQ_UNSET)      if (firstcharflags == REQ_UNSET)
6988        {        {
6989        zerofirstchar = REQ_NONE;        zerofirstcharflags = REQ_NONE;
6990        zeroreqchar = reqchar;        zeroreqchar = reqchar;
6991          zeroreqcharflags = reqcharflags;
6992    
6993        /* 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
6994        only if it is not to be matched caselessly. */        only if it is not to be matched caselessly. */
# Line 6946  for (;; ptr++) Line 6996  for (;; ptr++)
6996        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
6997          {          {
6998          firstchar = mcbuffer[0] | req_caseopt;          firstchar = mcbuffer[0] | req_caseopt;
6999          if (mclength != 1) reqchar = code[-1] | cd->req_varyopt;          firstchar = mcbuffer[0];
7000            firstcharflags = req_caseopt;
7001    
7002            if (mclength != 1)
7003              {
7004              reqchar = code[-1];
7005              reqcharflags = cd->req_varyopt;
7006              }
7007          }          }
7008        else firstchar = reqchar = REQ_NONE;        else firstcharflags = reqcharflags = REQ_NONE;
7009        }        }
7010    
7011      /* 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 6957  for (;; ptr++) Line 7014  for (;; ptr++)
7014      else      else
7015        {        {
7016        zerofirstchar = firstchar;        zerofirstchar = firstchar;
7017          zerofirstcharflags = firstcharflags;
7018        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7019          zeroreqcharflags = reqcharflags;
7020        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7021          reqchar = code[-1] | req_caseopt | cd->req_varyopt;          {
7022            reqchar = code[-1];
7023            reqcharflags = req_caseopt | cd->req_varyopt;
7024            }
7025        }        }
7026    
7027      break;            /* End of literal character handling */      break;            /* End of literal character handling */
# Line 6978  return FALSE; Line 7040  return FALSE;
7040    
7041    
7042    
   
7043  /*************************************************  /*************************************************
7044  *     Compile sequence of alternatives           *  *     Compile sequence of alternatives           *
7045  *************************************************/  *************************************************/
# Line 6999  Arguments: Line 7060  Arguments:
7060    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
7061    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
7062    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
7063    firstcharptr   place to put the first required character, or a negative number    firstcharptr    place to put the first required character
7064    reqcharptr     place to put the last required character, or a negative number    firstcharflagsptr place to put the first character flags, or a negative number
7065      reqcharptr     place to put the last required character
7066      reqcharflagsptr place to put the last required character flags, or a negative number
7067    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
7068    cd             points to the data block with tables pointers etc.    cd             points to the data block with tables pointers etc.
7069    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
# Line 7012  Returns:         TRUE on success Line 7075  Returns:         TRUE on success
7075  static BOOL  static BOOL
7076  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,
7077    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
7078    int cond_depth, pcre_int32 *firstcharptr, pcre_int32 *reqcharptr,    int cond_depth,
7079      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
7080      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
7081    branch_chain *bcptr, compile_data *cd, int *lengthptr)    branch_chain *bcptr, compile_data *cd, int *lengthptr)
7082  {  {
7083  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
# Line 7022  pcre_uchar *start_bracket = code; Line 7087  pcre_uchar *start_bracket = code;
7087  pcre_uchar *reverse_count = NULL;  pcre_uchar *reverse_count = NULL;
7088  open_capitem capitem;  open_capitem capitem;
7089  int capnumber = 0;  int capnumber = 0;
7090  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7091  pcre_int32 branchfirstchar, branchreqchar;  pcre_int32 firstcharflags, reqcharflags;
7092    pcre_uint32 branchfirstchar, branchreqchar;
7093    pcre_int32 branchfirstcharflags, branchreqcharflags;
7094  int length;  int length;
7095  int orig_bracount;  unsigned int orig_bracount;
7096  int max_bracount;  unsigned int max_bracount;
7097  branch_chain bc;  branch_chain bc;
7098    
7099  bc.outer = bcptr;  bc.outer = bcptr;
7100  bc.current_branch = code;  bc.current_branch = code;
7101    
7102  firstchar = reqchar = REQ_UNSET;  firstchar = reqchar = 0;
7103    firstcharflags = reqcharflags = REQ_UNSET;
7104    
7105  /* 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
7106  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 7092  for (;;) Line 7160  for (;;)
7160    into the length. */    into the length. */
7161    
7162    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,
7163          &branchreqchar, &bc, cond_depth, cd,          &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc,
7164          (lengthptr == NULL)? NULL : &length))          cond_depth, cd, (lengthptr == NULL)? NULL : &length))
7165      {      {
7166      *ptrptr = ptr;      *ptrptr = ptr;
7167      return FALSE;      return FALSE;
# Line 7114  for (;;) Line 7182  for (;;)
7182      if (*last_branch != OP_ALT)      if (*last_branch != OP_ALT)
7183        {        {
7184        firstchar = branchfirstchar;        firstchar = branchfirstchar;
7185          firstcharflags = branchfirstcharflags;
7186        reqchar = branchreqchar;        reqchar = branchreqchar;
7187          reqcharflags = branchreqcharflags;
7188        }        }
7189    
7190      /* 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 7128  for (;;) Line 7198  for (;;)
7198        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
7199        previously no reqchar, it takes on the value of the old firstchar. */        previously no reqchar, it takes on the value of the old firstchar. */
7200    
7201        if (firstchar >= 0 && firstchar != branchfirstchar)        if (firstcharflags >= 0 &&
7202              (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar))
7203          {          {
7204          if (reqchar < 0) reqchar = firstchar;          if (reqcharflags < 0)
7205          firstchar = REQ_NONE;            {
7206              reqchar = firstchar;
7207              reqcharflags = firstcharflags;
7208              }
7209            firstcharflags = REQ_NONE;
7210          }          }
7211    
7212        /* 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
7213        branch becomes a reqchar if there isn't a branch reqchar. */        branch becomes a reqchar if there isn't a branch reqchar. */
7214    
7215        if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0)        if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0)
7216            branchreqchar = branchfirstchar;          {
7217            branchreqchar = branchfirstchar;
7218            branchreqcharflags = branchfirstcharflags;
7219            }
7220    
7221        /* Now ensure that the reqchars match */        /* Now ensure that the reqchars match */
7222    
7223        if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY))        if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) ||
7224          reqchar = REQ_NONE;            reqchar != branchreqchar)
7225        else reqchar |= branchreqchar;   /* To "or" REQ_VARY */          reqcharflags = REQ_NONE;
7226          else
7227            {
7228            reqchar = branchreqchar;
7229            reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */
7230            }
7231        }        }
7232    
7233      /* 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 7173  for (;;) Line 7256  for (;;)
7256          *ptrptr = ptr;          *ptrptr = ptr;
7257          return FALSE;          return FALSE;
7258          }          }
7259        else        else
7260          {          {
7261          if (fixed_length > cd->max_lookbehind)          if (fixed_length > cd->max_lookbehind)
7262            cd->max_lookbehind = fixed_length;            cd->max_lookbehind = fixed_length;
7263          PUT(reverse_count, 0, fixed_length);          PUT(reverse_count, 0, fixed_length);
7264          }          }
7265        }        }
7266      }      }
# Line 7240  for (;;) Line 7323  for (;;)
7323      *codeptr = code;      *codeptr = code;
7324      *ptrptr = ptr;      *ptrptr = ptr;
7325      *firstcharptr = firstchar;      *firstcharptr = firstchar;
7326        *firstcharflagsptr = firstcharflags;
7327      *reqcharptr = reqchar;      *reqcharptr = reqchar;
7328        *reqcharflagsptr = reqcharflags;
7329      if (lengthptr != NULL)      if (lengthptr != NULL)
7330        {        {
7331        if (OFLOW_MAX - *lengthptr < length)        if (OFLOW_MAX - *lengthptr < length)
# Line 7310  and the highest back reference was great Line 7395  and the highest back reference was great
7395  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
7396  of the more common cases more precisely.  of the more common cases more precisely.
7397    
7398    ... A second exception is when the .* appears inside an atomic group, because
7399    this prevents the number of characters it matches from being adjusted.
7400    
7401  Arguments:  Arguments:
7402    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7403    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
7404                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7405                    the less precise approach                    the less precise approach
7406    backref_map    the back reference bitmap    cd             points to the compile data block
7407      atomcount      atomic group level
7408    
7409  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7410  */  */
7411    
7412  static BOOL  static BOOL
7413  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7414    unsigned int backref_map)    compile_data *cd, int atomcount)
7415  {  {
7416  do {  do {
7417     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7334  do { Line 7423  do {
7423     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7424         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7425       {       {
7426       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7427       }       }
7428    
7429     /* Capturing brackets */     /* Capturing brackets */
# Line 7344  do { Line 7433  do {
7433       {       {
7434       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7435       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7436       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7437       }       }
7438    
7439     /* Other brackets */     /* Positive forward assertions and conditions */
7440    
7441     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ASSERT || op == OP_COND)
             op == OP_COND)  
7442       {       {
7443       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7444         }
7445    
7446       /* Atomic groups */
7447    
7448       else if (op == OP_ONCE || op == OP_ONCE_NC)
7449         {
7450         if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7451           return FALSE;
7452       }       }
7453    
7454     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7455     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
7456       group. */
7457    
7458     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7459               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7460       {       {
7461       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7462             atomcount > 0 || cd->had_pruneorskip)
7463         return FALSE;         return FALSE;
7464       }       }
7465    
7466     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7467    
7468     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;
7469    
7470     code += GET(code, 1);     code += GET(code, 1);
7471     }     }
7472  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7385  return TRUE; Line 7484  return TRUE;
7484  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
7485  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
7486  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7487  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7488    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7489    count, because once again the assumption no longer holds.
7490    
7491  Arguments:  Arguments:
7492    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7493    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
7494                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7495                    the less precise approach                    the less precise approach
7496    backref_map    the back reference bitmap    cd             points to the compile data
7497      atomcount      atomic group level
7498    
7499  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7500  */  */
7501    
7502  static BOOL  static BOOL
7503  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7504    unsigned int backref_map)    compile_data *cd, int atomcount)
7505  {  {
7506  do {  do {
7507     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7425  do { Line 7527  do {
7527         return FALSE;         return FALSE;
7528    
7529         default:     /* Assertion */         default:     /* Assertion */
7530         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7531         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7532         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7533         break;         break;
# Line 7439  do { Line 7541  do {
7541     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7542         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7543       {       {
7544       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7545       }       }
7546    
7547     /* Capturing brackets */     /* Capturing brackets */
# Line 7449  do { Line 7551  do {
7551       {       {
7552       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode,