/[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 1066 by chpe, Tue Oct 16 15:54:19 2012 UTC
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  is also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. We do not need to select pcre16_printint.c specially, because the  library. We do not need to select pcre16_printint.c specially, because the
59  COMPILE_PCREx macro will already be appropriately set. */  COMPILE_PCREx macro will already be appropriately set. */
# Line 68  COMPILE_PCREx macro will already be appr Line 68  COMPILE_PCREx macro will already be appr
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 77  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL,
88        int, int, int *, int *, branch_chain *, compile_data *, int *);
89    
90    
91    
92  /*************************************************  /*************************************************
93  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 112  overrun before it actually does run off Line 123  overrun before it actually does run off
123    
124  #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */  #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */
125  #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */  #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */
126    #define REQ_MASK       (REQ_CASELESS | REQ_VARY)
127    
128  /* Repeated character flags. */  /* Repeated character flags. */
129    
# Line 490  static const char error_texts[] = Line 502  static const char error_texts[] =
502    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
503    "invalid UTF-16 string\0"    "invalid UTF-16 string\0"
504    /* 75 */    /* 75 */
505    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
506      "character value in \\u.... sequence is too large\0"
507      "invalid UTF-32 string\0"
508    ;    ;
509    
510  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 630  static const pcre_uint8 ebcdic_chartab[] Line 644  static const pcre_uint8 ebcdic_chartab[]
644  #endif  #endif
645    
646    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
647    
648    
649  /*************************************************  /*************************************************
# Line 741  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 749  return (*p == CHAR_RIGHT_CURLY_BRACKET);
749  *************************************************/  *************************************************/
750    
751  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
752  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
753  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
754  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
755  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
756  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
757  sequence.  escape sequence.
758    
759  Arguments:  Arguments:
760    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
761      chptr          points to the data character
762    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
763    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
764    options        the options bits    options        the options bits
765    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
766    
767  Returns:         zero or positive => a data character  Returns:         zero => a data character
768                   negative => a special escape sequence                   positive => a special escape sequence
769                     negative => a back reference
770                   on error, errorcodeptr is set                   on error, errorcodeptr is set
771  */  */
772    
773  static int  static int
774  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
775    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
776  {  {
777  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
778  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
779  const pcre_uchar *ptr = *ptrptr + 1;  const pcre_uchar *ptr = *ptrptr + 1;
780  pcre_int32 c;  pcre_uint32 c;
781    int escape = 0;
782  int i;  int i;
783    
784  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
# Line 784  Otherwise further processing may be requ Line 795  Otherwise further processing may be requ
795  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
796  /* Not alphanumeric */  /* Not alphanumeric */
797  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
798  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; }
799    
800  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
801  /* Not alphanumeric */  /* Not alphanumeric */
802  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)) {}
803  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; }
804  #endif  #endif
805    
806  /* 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 808  else if ((i = escapes[c - 0x48]) != 0)
808  else  else
809    {    {
810    const pcre_uchar *oldptr;    const pcre_uchar *oldptr;
811    BOOL braced, negated;    BOOL braced, negated, overflow;
812      int s;
813    
814    switch (c)    switch (c)
815      {      {
# Line 822  else Line 834  else
834          c = 0;          c = 0;
835          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
836            {            {
837            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
838  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
839            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
840            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 843  else
843            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
844  #endif  #endif
845            }            }
846    
847    #if defined COMPILE_PCRE8
848            if (c > (utf ? 0x10ffff : 0xff))
849    #elif defined COMPILE_PCRE16
850            if (c > (utf ? 0x10ffff : 0xffff))
851    #elif defined COMPILE_PCRE32
852            if (utf && c > 0x10ffff)
853    #endif
854              {
855              *errorcodeptr = ERR76;
856              }
857            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
858          }          }
859        }        }
860      else      else
# Line 857  else Line 881  else
881      (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
882      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
883      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
884      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
885    
886      case CHAR_g:      case CHAR_g:
887      if (isclass) break;      if (isclass) break;
888      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
889        {        {
890        c = -ESC_g;        escape = ESC_g;
891        break;        break;
892        }        }
893    
# Line 876  else Line 900  else
900          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
901        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
902          {          {
903          c = -ESC_k;          escape = ESC_k;
904          break;          break;
905          }          }
906        braced = TRUE;        braced = TRUE;
# Line 892  else Line 916  else
916      else negated = FALSE;      else negated = FALSE;
917    
918      /* The integer range is limited by the machine's int representation. */      /* The integer range is limited by the machine's int representation. */
919      c = 0;      s = 0;
920        overflow = FALSE;
921      while (IS_DIGIT(ptr[1]))      while (IS_DIGIT(ptr[1]))
922        {        {
923        if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */        if (s > INT_MAX / 10 - 1) /* Integer overflow */
924          {          {
925          c = -1;          overflow = TRUE;
926          break;          break;
927          }          }
928        c = c * 10 + *(++ptr) - CHAR_0;        s = s * 10 + (int)(*(++ptr) - CHAR_0);
929        }        }
930      if (((unsigned int)c) > INT_MAX) /* Integer overflow */      if (overflow) /* Integer overflow */
931        {        {
932        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
933          ptr++;          ptr++;
# Line 916  else Line 941  else
941        break;        break;
942        }        }
943    
944      if (c == 0)      if (s == 0)
945        {        {
946        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
947        break;        break;
# Line 924  else Line 949  else
949    
950      if (negated)      if (negated)
951        {        {
952        if (c > bracount)        if (s > bracount)
953          {          {
954          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
955          break;          break;
956          }          }
957        c = bracount - (c - 1);        s = bracount - (s - 1);
958        }        }
959    
960      c = -(ESC_REF + c);      escape = -s;
961      break;      break;
962    
963      /* 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 979  else
979        {        {
980        oldptr = ptr;        oldptr = ptr;
981        /* The integer range is limited by the machine's int representation. */        /* The integer range is limited by the machine's int representation. */
982        c -= CHAR_0;        s = (int)(c -CHAR_0);
983          overflow = FALSE;
984        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
985          {          {
986          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */          if (s > INT_MAX / 10 - 1) /* Integer overflow */
987            {            {
988            c = -1;            overflow = TRUE;
989            break;            break;
990            }            }
991          c = c * 10 + *(++ptr) - CHAR_0;          s = s * 10 + (int)(*(++ptr) - CHAR_0);
992          }          }
993        if (((unsigned int)c) > INT_MAX) /* Integer overflow */        if (overflow) /* Integer overflow */
994          {          {
995          while (IS_DIGIT(ptr[1]))          while (IS_DIGIT(ptr[1]))
996            ptr++;            ptr++;
997          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
998          break;          break;
999          }          }
1000        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1001          {          {
1002          c = -(ESC_REF + c);          escape = -s;
1003          break;          break;
1004          }          }
1005        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 1020  else Line 1046  else
1046          c = 0;          c = 0;
1047          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1048            {            {
1049            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1050  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1051            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1052            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 1064  else
1064        const pcre_uchar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1065    
1066        c = 0;        c = 0;
1067          overflow = FALSE;
1068        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1069          {          {
1070          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1071          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1072    
1073    #ifdef COMPILE_PCRE32
1074            if (c >= 0x10000000l) { overflow = TRUE; break; }
1075    #endif
1076    
1077  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1078          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1079          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 1082  else
1082          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1083  #endif  #endif
1084    
1085  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1086          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1087  #else  #elif defined COMPILE_PCRE16
1088  #ifdef COMPILE_PCRE16          if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1089          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }  #elif defined COMPILE_PCRE32
1090  #endif          if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1091  #endif  #endif
1092          }          }
1093    
1094        if (c < 0)        if (overflow)
1095          {          {
1096          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1097          *errorcodeptr = ERR34;          *errorcodeptr = ERR34;
# Line 1082  else Line 1113  else
1113      c = 0;      c = 0;
1114      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)
1115        {        {
1116        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1117        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1118  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1119        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 1141  else Line 1172  else
1172  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1173  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1174    
1175  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1176       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1177    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1178    
1179  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1180    
1181  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1182    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1183    
1184  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1185    
1186  *ptrptr = ptr;  *ptrptr = ptr;
1187  return c;  *chptr = c;
1188    return escape;
1189  }  }
1190    
   
   
1191  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1192  /*************************************************  /*************************************************
1193  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1180  Returns:         type value from ucp_typ Line 1210  Returns:         type value from ucp_typ
1210  static int  static int
1211  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)
1212  {  {
1213  int c, i, bot, top;  pcre_uchar c;
1214    int i, bot, top;
1215  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1216  pcre_uchar name[32];  pcre_uchar name[32];
1217    
# Line 1227  top = PRIV(utt_size); Line 1258  top = PRIV(utt_size);
1258    
1259  while (bot < top)  while (bot < top)
1260    {    {
1261      int r;
1262    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1263    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);
1264    if (c == 0)    if (r == 0)
1265      {      {
1266      *dptr = PRIV(utt)[i].value;      *dptr = PRIV(utt)[i].value;
1267      return PRIV(utt)[i].type;      return PRIV(utt)[i].type;
1268      }      }
1269    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1270    }    }
1271    
1272  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
# Line 1279  int max = -1; Line 1311  int max = -1;
1311  /* 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
1312  an integer overflow. */  an integer overflow. */
1313    
1314  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1315  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1316    {    {
1317    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1294  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1326  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1326    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1327      {      {
1328      max = 0;      max = 0;
1329      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1330      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1331        {        {
1332        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1349  Arguments: Line 1381  Arguments:
1381    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1382    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1383    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1384    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1385    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1386    
1387  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 1583  Arguments: Line 1615  Arguments:
1615    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1616    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1617    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1618    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1619    
1620  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1621  */  */
# Line 1686  and doing the check at the end; a flag s Line 1718  and doing the check at the end; a flag s
1718    
1719  Arguments:  Arguments:
1720    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1721    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1722    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1723    cd       the "compile data" structure    cd       the "compile data" structure
1724    
# Line 1820  for (;;) Line 1852  for (;;)
1852      case OP_NOTI:      case OP_NOTI:
1853      branchlength++;      branchlength++;
1854      cc += 2;      cc += 2;
1855  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1856      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1857  #endif  #endif
1858      break;      break;
# Line 1834  for (;;) Line 1866  for (;;)
1866      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1867      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1868      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1869  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1870      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1871  #endif  #endif
1872      break;      break;
1873    
1874      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1875      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1876      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)
1877          cc += 2;
1878      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1879      break;      break;
1880    
# Line 1876  for (;;) Line 1909  for (;;)
1909    
1910      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1911    
1912  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1913      case OP_XCLASS:      case OP_XCLASS:
1914      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1915      /* Fall through */      /* Fall through */
# Line 2015  length. Line 2048  length.
2048    
2049  Arguments:  Arguments:
2050    code        points to start of expression    code        points to start of expression
2051    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2052    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2053    
2054  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 2079  for (;;) Line 2112  for (;;)
2112        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2113        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2114        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2115        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2116          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2117        break;        break;
2118    
2119        case OP_MARK:        case OP_MARK:
# Line 2102  for (;;) Line 2135  for (;;)
2135    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
2136    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2137    
2138  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2139      if (utf) switch(c)      if (utf) switch(c)
2140        {        {
2141        case OP_CHAR:        case OP_CHAR:
# Line 2154  instance of OP_RECURSE. Line 2187  instance of OP_RECURSE.
2187    
2188  Arguments:  Arguments:
2189    code        points to start of expression    code        points to start of expression
2190    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2191    
2192  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
2193  */  */
# Line 2199  for (;;) Line 2232  for (;;)
2232        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2233        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2234        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2235        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2236          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2237        break;        break;
2238    
2239        case OP_MARK:        case OP_MARK:
# Line 2222  for (;;) Line 2255  for (;;)
2255      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
2256      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2257    
2258  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2259      if (utf) switch(c)      if (utf) switch(c)
2260        {        {
2261        case OP_CHAR:        case OP_CHAR:
# Line 2308  bracket whose current branch will alread Line 2341  bracket whose current branch will alread
2341  Arguments:  Arguments:
2342    code        points to start of search    code        points to start of search
2343    endcode     points to where to stop    endcode     points to where to stop
2344    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2345    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2346    
2347  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2525  for (code = first_significant_code(code Line 2558  for (code = first_significant_code(code
2558      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2559      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2560      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2561      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2562        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2563      break;      break;
2564    
2565      /* End of branch */      /* End of branch */
# Line 2541  for (code = first_significant_code(code Line 2574  for (code = first_significant_code(code
2574      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2575      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2576    
2577  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2578      case OP_STAR:      case OP_STAR:
2579      case OP_STARI:      case OP_STARI:
2580      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2607  Arguments: Line 2640  Arguments:
2640    code        points to start of the recursion    code        points to start of the recursion
2641    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2642    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2643    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2644    cd          pointers to tables etc    cd          pointers to tables etc
2645    
2646  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2754  value in the reference (which is a group Line 2787  value in the reference (which is a group
2787  Arguments:  Arguments:
2788    group      points to the start of the group    group      points to the start of the group
2789    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2790    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2791    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2792    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
2793    
# Line 2858  PUT(previous_callout, 2 + LINK_SIZE, len Line 2891  PUT(previous_callout, 2 + LINK_SIZE, len
2891  *************************************************/  *************************************************/
2892    
2893  /* 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
2894  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
2895  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
2896  start address.  start address. A character with multiple other cases is returned on its own
2897    with a special return value.
2898    
2899  Arguments:  Arguments:
2900    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2868  Arguments: Line 2902  Arguments:
2902    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2903    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2904    
2905  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2906                   0 when a range is returned
2907                  >0 the CASESET offset for char with multiple other cases
2908                    in this case, ocptr contains the original
2909  */  */
2910    
2911  static BOOL  static int
2912  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2913    unsigned int *odptr)    pcre_uint32 *odptr)
2914  {  {
2915  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2916    int co;
2917    
2918    /* Find the first character that has an other case. If it has multiple other
2919    cases, return its case offset value. */
2920    
2921  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2922    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2923      if ((co = UCD_CASESET(c)) != 0)
2924        {
2925        *ocptr = c++;   /* Character that has the set */
2926        *cptr = c;      /* Rest of input range */
2927        return co;
2928        }
2929      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2930      }
2931    
2932  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2933    
2934  *ocptr = othercase;  *ocptr = othercase;
2935  next = othercase + 1;  next = othercase + 1;
# Line 2891  for (++c; c <= d; c++) Line 2940  for (++c; c <= d; c++)
2940    next++;    next++;
2941    }    }
2942    
2943  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2944  *cptr = c;  *cptr = c;             /* Rest of input range */
2945    return 0;
 return TRUE;  
2946  }  }
2947    
2948    
# Line 2916  Returns:       TRUE if auto-possessifyin Line 2964  Returns:       TRUE if auto-possessifyin
2964  */  */
2965    
2966  static BOOL  static BOOL
2967  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(pcre_uint32 c, int ptype, int pdata, BOOL negated)
2968  {  {
2969    #ifdef SUPPORT_UCP
2970    const pcre_uint32 *p;
2971    #endif
2972    
2973  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2974    
2975  switch(ptype)  switch(ptype)
2976    {    {
2977    case PT_LAMP:    case PT_LAMP:
# Line 2956  switch(ptype) Line 3009  switch(ptype)
3009    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3010            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3011            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3012    
3013    #ifdef SUPPORT_UCP
3014      case PT_CLIST:
3015      p = PRIV(ucd_caseless_sets) + prop->caseset;
3016      for (;;)
3017        {
3018        if ((unsigned int)c < *p) return !negated;
3019        if ((unsigned int)c == *p++) return negated;
3020        }
3021      break;  /* Control never reaches here */
3022    #endif
3023    }    }
3024    
3025  return FALSE;  return FALSE;
3026  }  }
3027  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2973  sense to automatically possessify the re Line 3038  sense to automatically possessify the re
3038    
3039  Arguments:  Arguments:
3040    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3041    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3042    ptr           next character in pattern    ptr           next character in pattern
3043    options       options bits    options       options bits
3044    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2985  static BOOL Line 3050  static BOOL
3050  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3051    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3052  {  {
3053  pcre_int32 c, next;  pcre_uint32 c = NOTACHAR;
3054    pcre_uint32 next;
3055    int escape;
3056  int op_code = *previous++;  int op_code = *previous++;
3057    
3058  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 3017  value is a character, a negative value i Line 3084  value is a character, a negative value i
3084  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3085    {    {
3086    int temperrorcode = 0;    int temperrorcode = 0;
3087    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3088    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3089    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3090    }    }
3091  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
3092    {    {
3093      escape = 0;
3094  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3095    if (utf) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3096  #endif  #endif
# Line 3059  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3127  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3127    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)
3128      return FALSE;      return FALSE;
3129    
3130  /* 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. */  
3131    
3132  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3133        op_code == OP_NOT || op_code == OP_NOTI)
3134      //if (escape == 0) switch(op_code)
3135    {    {
   case OP_CHAR:  
3136  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3137    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3138  #else  #else
3139    c = *previous;    c = *previous;
3140  #endif  #endif
3141    return c != next;    }
3142    
3143    /* 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
3144    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3145    
3146    case OP_CHARI:  if (escape == 0)
3147  #ifdef SUPPORT_UTF    {
3148    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3149  #else    case, which maps to the special PT_CLIST property. Check this first. */
3150    c = *previous;  
 #endif  
   if (c == next) return FALSE;  
 #ifdef SUPPORT_UTF  
   if (utf)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3151  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3152      othercase = UCD_OTHERCASE((unsigned int)next);    if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3153  #else      {
3154      othercase = NOTACHAR;      int ocs = UCD_CASESET(next);
3155  #endif      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
     return (unsigned int)c != othercase;  
3156      }      }
   else  
 #endif  /* SUPPORT_UTF */  
   return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */  
   
   case OP_NOT:  
 #ifdef SUPPORT_UTF  
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
3157  #endif  #endif
   return c == next;  
3158    
3159    case OP_NOTI:    switch(op_code)
3160        {
3161        case OP_CHAR:
3162        return c != next;
3163    
3164        /* For CHARI (caseless character) we must check the other case. If we have
3165        Unicode property support, we can use it to test the other case of
3166        high-valued characters. We know that next can have only one other case,
3167        because multi-other-case characters are dealt with above. */
3168    
3169        case OP_CHARI:
3170        if (c == next) return FALSE;
3171  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3172    GETCHARTEST(c, previous);      if (utf)
3173          {
3174          pcre_uint32 othercase;
3175          if (next < 128) othercase = cd->fcc[next]; else
3176    #ifdef SUPPORT_UCP
3177          othercase = UCD_OTHERCASE(next);
3178  #else  #else
3179    c = *previous;        othercase = NOTACHAR;
3180  #endif  #endif
3181    if (c == next) return TRUE;        return c != othercase;
3182          }
3183        else
3184    #endif  /* SUPPORT_UTF */
3185        return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3186    
3187        case OP_NOT:
3188        return c == next;
3189    
3190        case OP_NOTI:
3191        if (c == next) return TRUE;
3192  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3193    if (utf)      if (utf)
3194      {        {
3195      unsigned int othercase;        pcre_uint32 othercase;
3196      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3197  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3198      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3199  #else  #else
3200      othercase = NOTACHAR;        othercase = NOTACHAR;
3201  #endif  #endif
3202      return (unsigned int)c == othercase;        return c == othercase;
3203      }        }
3204    else      else
3205  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3206    return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3207    
3208    /* 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.
3209    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. */
3210    
3211    case OP_DIGIT:      case OP_DIGIT:
3212    return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3213    
3214    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3215    return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3216    
3217    case OP_WHITESPACE:      case OP_WHITESPACE:
3218    return next > 255 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3219    
3220    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3221    return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3222    
3223    case OP_WORDCHAR:      case OP_WORDCHAR:
3224    return next > 255 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3225    
3226    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3227    return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3228    
3229    case OP_HSPACE:      case OP_HSPACE:
3230    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3231    switch(next)      switch(next)
3232      {        {
3233      case 0x09:        HSPACE_CASES:
3234      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;  
     }  
3235    
3236    case OP_ANYNL:        default:
3237    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3238    case OP_NOT_VSPACE:        }
3239    switch(next)  
3240      {      case OP_ANYNL:
3241      case 0x0a:      case OP_VSPACE:
3242      case 0x0b:      case OP_NOT_VSPACE:
3243      case 0x0c:      switch(next)
3244      case 0x0d:        {
3245      case 0x85:        VSPACE_CASES:
3246      case 0x2028:        return op_code == OP_NOT_VSPACE;
3247      case 0x2029:  
3248      return op_code == OP_NOT_VSPACE;        default:
3249      default:        return op_code != OP_NOT_VSPACE;
3250      return op_code != OP_NOT_VSPACE;        }
     }  
3251    
3252  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3253    case OP_PROP:      case OP_PROP:
3254    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3255    
3256    case OP_NOTPROP:      case OP_NOTPROP:
3257    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3258  #endif  #endif
3259    
3260    default:      default:
3261    return FALSE;      return FALSE;
3262        }
3263    }    }
3264    
   
3265  /* 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
3266  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
3267  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 3272  switch(op_code)
3272    {    {
3273    case OP_CHAR:    case OP_CHAR:
3274    case OP_CHARI:    case OP_CHARI:
3275  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3276      {      {
3277      case ESC_d:      case ESC_d:
3278      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
# Line 3248  switch(op_code) Line 3296  switch(op_code)
3296      case ESC_H:      case ESC_H:
3297      switch(c)      switch(c)
3298        {        {
3299        case 0x09:        HSPACE_CASES:
3300        case 0x20:        return escape != ESC_h;
3301        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;  
3302        default:        default:
3303        return -next == ESC_h;        return escape == ESC_h;
3304        }        }
3305    
3306      case ESC_v:      case ESC_v:
3307      case ESC_V:      case ESC_V:
3308      switch(c)      switch(c)
3309        {        {
3310        case 0x0a:        VSPACE_CASES:
3311        case 0x0b:        return escape != ESC_v;
3312        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3313        default:        default:
3314        return -next == ESC_v;        return escape == ESC_v;
3315        }        }
3316    
3317      /* 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
3318      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3319      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3320    
3321  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3322      case ESC_du:      case ESC_du:
# Line 3301  switch(op_code) Line 3327  switch(op_code)
3327      case ESC_SU:      case ESC_SU:
3328        {        {
3329        int temperrorcode = 0;        int temperrorcode = 0;
3330        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3331        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3332        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3333        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3334        }        }
# Line 3329  switch(op_code) Line 3355  switch(op_code)
3355    
3356        /* Do the property check. */        /* Do the property check. */
3357    
3358        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3359        }        }
3360  #endif  #endif
3361    
# Line 3344  switch(op_code) Line 3370  switch(op_code)
3370    these op-codes are never generated.) */    these op-codes are never generated.) */
3371    
3372    case OP_DIGIT:    case OP_DIGIT:
3373    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3374           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3375    
3376    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3377    return next == -ESC_d;    return escape == ESC_d;
3378    
3379    case OP_WHITESPACE:    case OP_WHITESPACE:
3380    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3381    
3382    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3383    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;
3384    
3385    case OP_HSPACE:    case OP_HSPACE:
3386    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3387           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3388    
3389    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3390    return next == -ESC_h;    return escape == ESC_h;
3391    
3392    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3393    case OP_ANYNL:    case OP_ANYNL:
3394    case OP_VSPACE:    case OP_VSPACE:
3395    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3396    
3397    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3398    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3399    
3400    case OP_WORDCHAR:    case OP_WORDCHAR:
3401    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3402           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3403    
3404    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3405    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3406    
3407    default:    default:
3408    return FALSE;    return FALSE;
# Line 3388  switch(op_code) Line 3414  switch(op_code)
3414    
3415    
3416  /*************************************************  /*************************************************
3417    *        Add a character or range to a class     *
3418    *************************************************/
3419    
3420    /* This function packages up the logic of adding a character or range of
3421    characters to a class. The character values in the arguments will be within the
3422    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3423    mutually recursive with the function immediately below.
3424    
3425    Arguments:
3426      classbits     the bit map for characters < 256
3427      uchardptr     points to the pointer for extra data
3428      options       the options word
3429      cd            contains pointers to tables etc.
3430      start         start of range character
3431      end           end of range character
3432    
3433    Returns:        the number of < 256 characters added
3434                    the pointer to extra data is updated
3435    */
3436    
3437    static int
3438    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3439      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3440    {
3441    pcre_uint32 c;
3442    int n8 = 0;
3443    
3444    /* If caseless matching is required, scan the range and process alternate
3445    cases. In Unicode, there are 8-bit characters that have alternate cases that
3446    are greater than 255 and vice-versa. Sometimes we can just extend the original
3447    range. */
3448    
3449    if ((options & PCRE_CASELESS) != 0)
3450      {
3451    #ifdef SUPPORT_UCP
3452      if ((options & PCRE_UTF8) != 0)
3453        {
3454        int rc;
3455        pcre_uint32 oc, od;
3456    
3457        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3458        c = start;
3459    
3460        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3461          {
3462          /* Handle a single character that has more than one other case. */
3463    
3464          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3465            PRIV(ucd_caseless_sets) + rc, oc);
3466    
3467          /* Do nothing if the other case range is within the original range. */
3468    
3469          else if (oc >= start && od <= end) continue;
3470    
3471          /* Extend the original range if there is overlap, noting that if oc < c, we
3472          can't have od > end because a subrange is always shorter than the basic
3473          range. Otherwise, use a recursive call to add the additional range. */
3474    
3475          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3476          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3477          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3478          }
3479        }
3480      else
3481    #endif  /* SUPPORT_UCP */
3482    
3483      /* Not UTF-mode, or no UCP */
3484    
3485      for (c = start; c <= end && c < 256; c++)
3486        {
3487        SETBIT(classbits, cd->fcc[c]);
3488        n8++;
3489        }
3490      }
3491    
3492    /* Now handle the original range. Adjust the final value according to the bit
3493    length - this means that the same lists of (e.g.) horizontal spaces can be used
3494    in all cases. */
3495    
3496    #if defined COMPILE_PCRE8
3497    #ifdef SUPPORT_UTF
3498      if ((options & PCRE_UTF8) == 0)
3499    #endif
3500      if (end > 0xff) end = 0xff;
3501    
3502    #elif defined COMPILE_PCRE16
3503    #ifdef SUPPORT_UTF
3504      if ((options & PCRE_UTF16) == 0)
3505    #endif
3506      if (end > 0xffff) end = 0xffff;
3507    
3508    #endif /* COMPILE_PCRE[8|16] */
3509    
3510    /* If all characters are less than 256, use the bit map. Otherwise use extra
3511    data. */
3512    
3513    if (end < 0x100)
3514      {
3515      for (c = start; c <= end; c++)
3516        {
3517        n8++;
3518        SETBIT(classbits, c);
3519        }
3520      }
3521    
3522    else
3523      {
3524      pcre_uchar *uchardata = *uchardptr;
3525    
3526    #ifdef SUPPORT_UTF
3527      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3528        {
3529        if (start < end)
3530          {
3531          *uchardata++ = XCL_RANGE;
3532          uchardata += PRIV(ord2utf)(start, uchardata);
3533          uchardata += PRIV(ord2utf)(end, uchardata);
3534          }
3535        else if (start == end)
3536          {
3537          *uchardata++ = XCL_SINGLE;
3538          uchardata += PRIV(ord2utf)(start, uchardata);
3539          }
3540        }
3541      else
3542    #endif  /* SUPPORT_UTF */
3543    
3544      /* Without UTF support, character values are constrained by the bit length,
3545      and can only be > 256 for 16-bit and 32-bit libraries. */
3546    
3547    #ifdef COMPILE_PCRE8
3548        {}
3549    #else
3550      if (start < end)
3551        {
3552        *uchardata++ = XCL_RANGE;
3553        *uchardata++ = start;
3554        *uchardata++ = end;
3555        }
3556      else if (start == end)
3557        {
3558        *uchardata++ = XCL_SINGLE;
3559        *uchardata++ = start;
3560        }
3561    #endif
3562    
3563      *uchardptr = uchardata;   /* Updata extra data pointer */
3564      }
3565    
3566    return n8;    /* Number of 8-bit characters */
3567    }
3568    
3569    
3570    
3571    
3572    /*************************************************
3573    *        Add a list of characters to a class     *
3574    *************************************************/
3575    
3576    /* This function is used for adding a list of case-equivalent characters to a
3577    class, and also for adding a list of horizontal or vertical whitespace. If the
3578    list is in order (which it should be), ranges of characters are detected and
3579    handled appropriately. This function is mutually recursive with the function
3580    above.
3581    
3582    Arguments:
3583      classbits     the bit map for characters < 256
3584      uchardptr     points to the pointer for extra data
3585      options       the options word
3586      cd            contains pointers to tables etc.
3587      p             points to row of 32-bit values, terminated by NOTACHAR
3588      except        character to omit; this is used when adding lists of
3589                      case-equivalent characters to avoid including the one we
3590                      already know about
3591    
3592    Returns:        the number of < 256 characters added
3593                    the pointer to extra data is updated
3594    */
3595    
3596    static int
3597    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3598      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3599    {
3600    int n8 = 0;
3601    while (p[0] < NOTACHAR)
3602      {
3603      int n = 0;
3604      if (p[0] != except)
3605        {
3606        while(p[n+1] == p[0] + n + 1) n++;
3607        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3608        }
3609      p += n + 1;
3610      }
3611    return n8;
3612    }
3613    
3614    
3615    
3616    /*************************************************
3617    *    Add characters not in a list to a class     *
3618    *************************************************/
3619    
3620    /* This function is used for adding the complement of a list of horizontal or
3621    vertical whitespace to a class. The list must be in order.
3622    
3623    Arguments:
3624      classbits     the bit map for characters < 256
3625      uchardptr     points to the pointer for extra data
3626      options       the options word
3627      cd            contains pointers to tables etc.
3628      p             points to row of 32-bit values, terminated by NOTACHAR
3629    
3630    Returns:        the number of < 256 characters added
3631                    the pointer to extra data is updated
3632    */
3633    
3634    static int
3635    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3636      int options, compile_data *cd, const pcre_uint32 *p)
3637    {
3638    BOOL utf = (options & PCRE_UTF8) != 0;
3639    int n8 = 0;
3640    if (p[0] > 0)
3641      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3642    while (p[0] < NOTACHAR)
3643      {
3644      while (p[1] == p[0] + 1) p++;
3645      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3646        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3647      p++;
3648      }
3649    return n8;
3650    }
3651    
3652    
3653    
3654    /*************************************************
3655  *           Compile one branch                   *  *           Compile one branch                   *
3656  *************************************************/  *************************************************/
3657    
# Line 3430  pcre_int32 req_caseopt, reqvary, tempreq Line 3694  pcre_int32 req_caseopt, reqvary, tempreq
3694  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3695  int after_manual_callout = 0;  int after_manual_callout = 0;
3696  int length_prevgroup = 0;  int length_prevgroup = 0;
3697  register int c;  register pcre_uint32 c;
3698    int escape;
3699  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3700  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3701  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3450  must not do this for other options (e.g. Line 3715  must not do this for other options (e.g.
3715  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3716    
3717  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3718  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3719  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3720  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3721  #else  #else
3722  BOOL utf = FALSE;  BOOL utf = FALSE;
3723  #endif  #endif
3724    
3725  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3726    class_uchardata always so that it can be passed to add_to_class() always,
3727    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3728    alternative calls for the different cases. */
3729    
3730    pcre_uchar *class_uchardata;
3731  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3732  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3733  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3734  #endif  #endif
3735    
# Line 3505  for (;; ptr++) Line 3773  for (;; ptr++)
3773    BOOL is_recurse;    BOOL is_recurse;
3774    BOOL reset_bracount;    BOOL reset_bracount;
3775    int class_has_8bitchar;    int class_has_8bitchar;
3776    int class_single_char;    int class_one_char;
3777    int newoptions;    int newoptions;
3778    int recno;    int recno;
3779    int refsign;    int refsign;
# Line 3515  for (;; ptr++) Line 3783  for (;; ptr++)
3783    int terminator;    int terminator;
3784    int mclength;    int mclength;
3785    int tempbracount;    int tempbracount;
3786      int ec; // FIXMEchpe pcre_uint32
3787    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3788    
3789    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3803  for (;; ptr++) Line 4072  for (;; ptr++)
4072    
4073      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4074    
4075      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4076      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 <
4077      < 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
4078      a single character. */      character. */
4079    
4080      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4081      class_single_char = 0;      class_one_char = 0;
4082    
4083      /* 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
4084      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
4085      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
4086      */      map. */
4087    
4088      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4089    
4090  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4091      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4092      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4093      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4094  #endif  #endif
4095    
4096      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
# Line 3832  for (;; ptr++) Line 4101  for (;; ptr++)
4101        {        {
4102        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4103    
4104  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4105        if (utf && HAS_EXTRALEN(c))        if (utf && HAS_EXTRALEN(c))
4106          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4107          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
# Line 3843  for (;; ptr++) Line 4112  for (;; ptr++)
4112        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4113        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
4114        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4115        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4116          however. */
4117    
4118        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4119          {          {
4120            xclass = TRUE;
4121          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4122          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4123          }          }
# Line 3905  for (;; ptr++) Line 4176  for (;; ptr++)
4176          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4177          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4178    
4179          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4180            posix_class = 0;            posix_class = 0;
4181    
4182          /* When PCRE_UCP is set, some of the POSIX classes are converted to          /* When PCRE_UCP is set, some of the POSIX classes are converted to
# Line 3948  for (;; ptr++) Line 4219  for (;; ptr++)
4219              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4220            }            }
4221    
4222          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4223          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4224    
4225          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3964  for (;; ptr++) Line 4235  for (;; ptr++)
4235            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4236    
4237          ptr = tempptr + 1;          ptr = tempptr + 1;
4238          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4239          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4240          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4241          class_single_char = 2;          class_one_char = 2;
4242          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4243          }          }
4244    
# Line 3975  for (;; ptr++) Line 4246  for (;; ptr++)
4246        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
4247        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
4248        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4249        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4250        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4251        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4252        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4253    
4254        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4255          {          {
4256          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4257    
4258          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4259    
4260          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4261          else if (-c == ESC_N)            /* \N is not supported in a class */            c = ec;
4262            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4263            else if (escape == ESC_N)            /* \N is not supported in a class */
4264            {            {
4265            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4266            goto FAILED;            goto FAILED;
4267            }            }
4268          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4269            {            {
4270            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4271              {              {
# Line 4000  for (;; ptr++) Line 4274  for (;; ptr++)
4274            else inescq = TRUE;            else inescq = TRUE;
4275            continue;            continue;
4276            }            }
4277          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4278    
4279          if (c < 0)          else
4280            {            {
4281            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4282            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4283            class_has_8bitchar++;            class_has_8bitchar++;
4284            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4285            class_single_char += 2;            class_one_char += 2;
4286    
4287            switch (-c)            switch (escape)
4288              {              {
4289  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4290              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 4294  for (;; ptr++)
4294              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4295              case ESC_SU:              case ESC_SU:
4296              nestptr = ptr;              nestptr = ptr;
4297              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4298              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4299              continue;              continue;
4300  #endif  #endif
# Line 4044  for (;; ptr++) Line 4318  for (;; ptr++)
4318    
4319              /* 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
4320              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4321              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4322                EBCDIC, so we lazily just adjust the appropriate bit. */
4323    
4324              case ESC_s:              case ESC_s:
4325              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4057  for (;; ptr++) Line 4332  for (;; ptr++)
4332              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4333              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4334              continue;              continue;
4335    
4336                /* The rest apply in both UCP and non-UCP cases. */
4337    
4338              case ESC_h:              case ESC_h:
4339              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4340              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  
4341              continue;              continue;
4342    
4343              case ESC_H:              case ESC_H:
4344              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4345                {                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  
4346              continue;              continue;
4347    
4348              case ESC_v:              case ESC_v:
4349              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4350              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  
4351              continue;              continue;
4352    
4353              case ESC_V:              case ESC_V:
4354              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4355                {                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  
4356              continue;              continue;
4357    
4358  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4242  for (;; ptr++) Line 4363  for (;; ptr++)
4363                int pdata;                int pdata;
4364                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4365                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4366                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4367                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4368                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4369                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4262  for (;; ptr++) Line 4382  for (;; ptr++)
4382                goto FAILED;                goto FAILED;
4383                }                }
4384              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4385              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4386              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4387              break;              break;
4388              }              }
4389            }            }
4390    
4391          /* 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).
4392          greater than 256. */          This may be greater than 256. */
4393    
4394            escape = 0;
4395    
4396          }   /* End of backslash handling */          }   /* End of backslash handling */
4397    
4398        /* 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
4399        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
4400        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4401        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4402    
4403        CHECK_RANGE:        CHECK_RANGE:
4404        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4284  for (;; ptr++) Line 4406  for (;; ptr++)
4406          inescq = FALSE;          inescq = FALSE;
4407          ptr += 2;          ptr += 2;
4408          }          }
   
4409        oldptr = ptr;        oldptr = ptr;
4410    
4411        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4412    
4413        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4414    
# Line 4310  for (;; ptr++) Line 4431  for (;; ptr++)
4431            inescq = TRUE;            inescq = TRUE;
4432            break;            break;
4433            }            }
4434    
4435            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4436            back the pointer and jump to handle the character that preceded it. */
4437    
4438          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4439            {            {
4440            ptr = oldptr;            ptr = oldptr;
4441            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4442            }            }
4443    
4444            /* Otherwise, we have a potential range; pick up the next character */
4445    
4446  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4447          if (utf)          if (utf)
# Line 4332  for (;; ptr++) Line 4458  for (;; ptr++)
4458    
4459          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4460            {            {
4461            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4462              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4463            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4464    
4465            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4466    
4467            if (d < 0)            if (descape != 0)
4468              {              {
4469              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4470                {                {
4471                ptr = oldptr;                ptr = oldptr;
4472                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4473                }                }
4474              }              }
4475            }            }
4476    
4477          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4478          one-character ranges */          one-character ranges. */
4479    
4480          if (d < c)          if (d < c)
4481            {            {
4482            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4483            goto FAILED;            goto FAILED;
4484            }            }
4485            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4486    
4487          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4488            cannot be done anymore. Any value greater than 1 indicates that there
4489            is more than one character. */
4490    
4491            class_one_char = 2;
4492    
4493          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4494    
4495          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4496    
4497          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4498          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4499          class_single_char = 2;  
   
         /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless  
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
         if ((d > 255) || (utf && ((options & PCRE_CASELESS) != 0 && d > 127)))  
 #elif defined  SUPPORT_UTF  
         if (utf && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
 #elif !(defined COMPILE_PCRE8)  
         if (d > 255)  
 #endif  
 #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)  
           {  
           xclass = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
   
 #ifdef SUPPORT_UCP  
 #ifndef COMPILE_PCRE8  
           if (utf && (options & PCRE_CASELESS) != 0)  
 #else  
           if ((options & PCRE_CASELESS) != 0)  
 #endif  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
   
               if (occ == ocd)  
                 {  
                 *class_uchardata++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_uchardata++ = XCL_RANGE;  
                 class_uchardata += PRIV(ord2utf)(occ, class_uchardata);  
                 }  
               class_uchardata += PRIV(ord2utf)(ocd, class_uchardata);  
               }  
             }  
 #endif  /* SUPPORT_UCP */  
   
           /* Now record the original range, possibly modified for UCP caseless  
           overlapping ranges. */  
   
           *class_uchardata++ = XCL_RANGE;  
 #ifdef SUPPORT_UTF  
 #ifndef COMPILE_PCRE8  
           if (utf)  
             {  
             class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
             class_uchardata += PRIV(ord2utf)(d, class_uchardata);  
             }  
           else  
             {  
             *class_uchardata++ = c;  
             *class_uchardata++ = d;  
             }  
 #else  
           class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
           class_uchardata += PRIV(ord2utf)(d, class_uchardata);  
 #endif  
 #else /* SUPPORT_UTF */  
           *class_uchardata++ = c;  
           *class_uchardata++ = d;  
 #endif /* SUPPORT_UTF */  
   
           /* With UCP support, we are done. Without UCP support, there is no  
           caseless matching for UTF characters > 127; we can use the bit map  
           for the smaller ones. As for 16 bit characters without UTF, we  
           can still use  */  
   
 #ifdef SUPPORT_UCP  
 #ifndef COMPILE_PCRE8  
           if (utf)  
 #endif  
             continue;    /* With next character in the class */  
 #endif  /* SUPPORT_UCP */  
   
 #if defined SUPPORT_UTF && !defined(SUPPORT_UCP) && !(defined COMPILE_PCRE8)  
           if (utf)  
             {  
             if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
             /* Adjust upper limit and fall through to set up the map */  
             d = 127;  
             }  
           else  
             {  
             if (c > 255) continue;  
             /* Adjust upper limit and fall through to set up the map */  
             d = 255;  
             }  
 #elif defined SUPPORT_UTF && !defined(SUPPORT_UCP)  
           if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
           /* Adjust upper limit and fall through to set up the map */  
           d = 127;  
 #else  
           if (c > 255) continue;  
           /* Adjust upper limit and fall through to set up the map */  
           d = 255;  
 #endif  /* SUPPORT_UTF && !SUPPORT_UCP && !COMPILE_PCRE8 */  
           }  
 #endif  /* SUPPORT_UTF || !COMPILE_PCRE8 */  
   
         /* We use the bit map for 8 bit mode, or when the characters fall  
         partially or entirely to [0-255] ([0-127] for UCP) ranges. */  
   
         class_has_8bitchar = 1;  
   
         /* We can save a bit of time by skipping this in the pre-compile. */  
   
         if (lengthptr == NULL) for (; c <= d; c++)  
           {  
           classbits[c/8] |= (1 << (c&7));  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             int uc = cd->fcc[c]; /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
           }  
   
4500          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4501          }          }
4502    
4503        /* Handle a lone single character - we can get here for a normal        /* Handle a single character - we can get here for a normal non-escape
4504        non-escape char, or after \ that introduces a single character or for an        char, or after \ that introduces a single character or for an apparent
4505        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4506          increase it if it is already 2 or more ... just in case there's a class
4507        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4508    
4509          CLASS_SINGLE_CHARACTER:
4510          if (class_one_char < 2) class_one_char++;
4511    
4512          /* If class_one_char is 1, we have the first single character in the
4513          class, and there have been no prior ranges, or XCLASS items generated by
4514          escapes. If this is the final character in the class, we can optimize by
4515          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4516          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4517          to be set. Otherwise, there can be no first char if this item is first,
4518          whatever repeat count may follow. In the case of reqchar, save the
4519          previous value for reinstating. */
4520    
4521        /* Only the value of 1 matters for class_single_char. */        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
   
       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. */  
   
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)  
4522          {          {
4523          ptr++;          ptr++;
4524          zeroreqchar = reqchar;          zeroreqchar = reqchar;
4525    
4526          if (negate_class)          if (negate_class)
4527            {            {
4528    #ifdef SUPPORT_UCP
4529                // FIXMEchpe pcreuint32?
4530              int d;
4531    #endif
4532            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4533            zerofirstchar = firstchar;            zerofirstchar = firstchar;
4534            *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
4535  #ifdef SUPPORT_UTF            /* For caseless UTF-8 mode when UCP support is available, check
4536            if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)            whether this character has more than one other case. If so, generate
4537              code += PRIV(ord2utf)(c, code);            a special OP_NOTPROP item instead of OP_NOTI. */
4538            else  
4539    #ifdef SUPPORT_UCP
4540              if (utf && (options & PCRE_CASELESS) != 0 &&
4541                  (d = UCD_CASESET(c)) != 0)
4542                {
4543                *code++ = OP_NOTPROP;
4544                *code++ = PT_CLIST;
4545                *code++ = d;
4546                }
4547              else
4548  #endif  #endif
4549              *code++ = c;            /* Char has only one other case, or UCP not available */
4550            goto NOT_CHAR;  
4551                {
4552                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4553    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4554                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4555                  code += PRIV(ord2utf)(c, code);
4556                else
4557    #endif
4558                  *code++ = c;
4559                }
4560    
4561              /* We are finished with this character class */
4562    
4563              goto END_CLASS;
4564            }            }
4565    
4566          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4567          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4568    
4569  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4570          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4571            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4572          else          else
# Line 4564  for (;; ptr++) Line 4577  for (;; ptr++)
4577            }            }
4578          goto ONE_CHAR;          goto ONE_CHAR;
4579          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4580    
4581        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4582          has been generated. Add this character to the class. */
4583  #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
4584        if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))        class_has_8bitchar +=
4585  #elif defined SUPPORT_UTF          add_to_class(classbits, &class_uchardata, options, cd, c, c);
       if (utf && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
 #elif !(defined COMPILE_PCRE8)  
       if (c > 255)  
 #endif  
   
 #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)  
         {  
         xclass = TRUE;  
         *class_uchardata++ = XCL_SINGLE;  
 #ifdef SUPPORT_UTF  
 #ifndef COMPILE_PCRE8  
         /* In non 8 bit mode, we can get here even if we are not in UTF mode. */  
         if (!utf)  
           *class_uchardata++ = c;  
         else  
 #endif  
           class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
 #else /* SUPPORT_UTF */  
         *class_uchardata++ = c;  
 #endif /* SUPPORT_UTF */  
   
 #ifdef SUPPORT_UCP  
 #ifdef COMPILE_PCRE8  
         if ((options & PCRE_CASELESS) != 0)  
 #else  
         /* In non 8 bit mode, we can get here even if we are not in UTF mode. */  
         if (utf && (options & PCRE_CASELESS) != 0)  
 #endif  
           {  
           unsigned int othercase;  
           if ((int)(othercase = UCD_OTHERCASE(c)) != c)  
             {  
             *class_uchardata++ = XCL_SINGLE;  
             class_uchardata += PRIV(ord2utf)(othercase, class_uchardata);  
             }  
           }  
 #endif  /* SUPPORT_UCP */  
   
         }  
       else  
 #endif  /* SUPPORT_UTF || COMPILE_PCRE16 */  
   
       /* Handle a single-byte character */  
         {  
         class_has_8bitchar = 1;  
         classbits[c/8] |= (1 << (c&7));  
         if ((options & PCRE_CASELESS) != 0)  
           {  
           c = cd->fcc[c]; /* flip case */  
           classbits[c/8] |= (1 << (c&7));  
           }  
         }  
4586        }        }
4587    
4588      /* Loop until ']' reached. This "while" is the end of the "do" far above.      /* Loop until ']' reached. This "while" is the end of the "do" far above.
# Line 4641  for (;; ptr++) Line 4602  for (;; ptr++)
4602        goto FAILED;        goto FAILED;
4603        }        }
4604    
4605        /* We will need an XCLASS if data has been placed in class_uchardata. In
4606        the second phase this is a sufficient test. However, in the pre-compile
4607        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4608        only if the very last character in the class needs XCLASS will it contain
4609        anything at this point. For this reason, xclass gets set TRUE above when
4610        uchar_classdata is emptied, and that's why this code is the way it is here
4611        instead of just doing a test on class_uchardata below. */
4612    
4613    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4614        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4615    #endif
4616    
4617      /* 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
4618      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4619      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
# Line 4703  for (;; ptr++) Line 4676  for (;; ptr++)
4676        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4677        }        }
4678      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4679      NOT_CHAR:  
4680        END_CLASS:
4681      break;      break;
4682    
4683    
# Line 4813  for (;; ptr++) Line 4787  for (;; ptr++)
4787      if (*previous == OP_CHAR || *previous == OP_CHARI      if (*previous == OP_CHAR || *previous == OP_CHARI
4788          || *previous == OP_NOT || *previous == OP_NOTI)          || *previous == OP_NOT || *previous == OP_NOTI)
4789        {        {
4790        switch (*previous)        switch (*previous)
4791          {          {
4792          default: /* Make compiler happy. */          default: /* Make compiler happy. */
4793          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 4801  for (;; ptr++)
4801        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
4802        it's a length rather than a small character. */        it's a length rather than a small character. */
4803    
4804  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4805        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4806          {          {
4807          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4963  for (;; ptr++) Line 4937  for (;; ptr++)
4937    
4938          if (repeat_max < 0)          if (repeat_max < 0)
4939            {            {
4940  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4941            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4942              {              {
4943              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4988  for (;; ptr++) Line 4962  for (;; ptr++)
4962    
4963          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4964            {            {
4965  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4966            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4967              {              {
4968              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5018  for (;; ptr++) Line 4992  for (;; ptr++)
4992    
4993        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
4994    
4995  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4996        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
4997          {          {
4998          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5505  for (;; ptr++) Line 5479  for (;; ptr++)
5479        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)
5480          {          {
5481          tempcode += PRIV(OP_lengths)[*tempcode];          tempcode += PRIV(OP_lengths)[*tempcode];
5482  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5483          if (utf && HAS_EXTRALEN(tempcode[-1]))          if (utf && HAS_EXTRALEN(tempcode[-1]))
5484            tempcode += GET_EXTRALEN(tempcode[-1]);            tempcode += GET_EXTRALEN(tempcode[-1]);
5485  #endif  #endif
# Line 5593  for (;; ptr++) Line 5567  for (;; ptr++)
5567        ptr++;        ptr++;
5568        while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;        while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;
5569        namelen = (int)(ptr - name);        namelen = (int)(ptr - name);
5570    
5571        /* It appears that Perl allows any characters whatsoever, other than        /* It appears that Perl allows any characters whatsoever, other than
5572        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
5573        letters, digits, and underscores. */        letters, digits, and underscores. */
# Line 5603  for (;; ptr++) Line 5577  for (;; ptr++)
5577          arg = ++ptr;          arg = ++ptr;
5578          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5579          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5580          if (arglen > (int)MAX_MARK)          if ((unsigned int)arglen > MAX_MARK)
5581            {            {
5582            *errorcodeptr = ERR75;            *errorcodeptr = ERR75;
5583            goto FAILED;            goto FAILED;
5584            }            }
5585          }          }
5586    
5587        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5623  for (;; ptr++) Line 5597  for (;; ptr++)
5597          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5598              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5599            {            {
5600              int setverb;
5601    
5602            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5603            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5604    
# Line 5640  for (;; ptr++) Line 5616  for (;; ptr++)
5616                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5617                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5618                }                }
5619              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5620                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5621    
5622              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5623              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
# Line 5655  for (;; ptr++) Line 5632  for (;; ptr++)
5632                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5633                goto FAILED;                goto FAILED;
5634                }                }
5635              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5636              }              }
5637    
5638            else            else
# Line 5666  for (;; ptr++) Line 5642  for (;; ptr++)
5642                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5643                goto FAILED;                goto FAILED;
5644                }                }
5645              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5646              *code++ = arglen;              *code++ = arglen;
5647              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5648              code += arglen;              code += arglen;
5649              *code++ = 0;              *code++ = 0;
5650              }              }
5651    
5652              switch (setverb)
5653                {
5654                case OP_THEN:
5655                case OP_THEN_ARG:
5656                cd->external_flags |= PCRE_HASTHEN;
5657                break;
5658    
5659                case OP_PRUNE:
5660                case OP_PRUNE_ARG:
5661                case OP_SKIP:
5662                case OP_SKIP_ARG:
5663                cd->had_pruneorskip = TRUE;
5664                break;
5665                }
5666    
5667            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5668            }            }
5669    
# Line 6688  for (;; ptr++) Line 6678  for (;; ptr++)
6678      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
6679      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
6680      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
6681      are ESC_REF plus the reference number. Only back references and those types      are negative the reference number. Only back references and those types
6682      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
6683      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
6684      ever created. */      ever created. */
6685    
6686      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6687      tempptr = ptr;      tempptr = ptr;
6688      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6689    
6690      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6691    
6692      if (c < 0)      if (escape == 0)
6693          c = ec;
6694        else
6695        {        {
6696        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6697          {          {
6698          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6699            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6708  for (;; ptr++) Line 6701  for (;; ptr++)
6701          continue;          continue;
6702          }          }
6703    
6704        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6705    
6706        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6707        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6708    
6709        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstchar == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6710          firstchar = REQ_NONE;          firstchar = REQ_NONE;
6711    
6712        /* Set values to reset to if this is followed by a zero repeat. */        /* Set values to reset to if this is followed by a zero repeat. */
# Line 6723  for (;; ptr++) Line 6716  for (;; ptr++)
6716    
6717        /* \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'
6718        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6719        -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 <
6720        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
6721        -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
6722        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6723    
6724        if (-c == ESC_g)        if (escape == ESC_g)
6725          {          {
6726          const pcre_uchar *p;          const pcre_uchar *p;
6727          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
# Line 6784  for (;; ptr++) Line 6777  for (;; ptr++)
6777        /* \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).
6778        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6779    
6780        if (-c == ESC_k)        if (escape == ESC_k)
6781          {          {
6782          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6783            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 6796  for (;; ptr++)
6796        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
6797        ':' later. */        ':' later. */
6798    
6799        if (-c >= ESC_REF)        if (escape < 0)
6800          {          {
6801          open_capitem *oc;          open_capitem *oc;
6802          recno = -c - ESC_REF;          recno = -escape;
6803    
6804          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6805          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
# Line 6833  for (;; ptr++) Line 6826  for (;; ptr++)
6826        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6827    
6828  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6829        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6830          {          {
6831          BOOL negated;          BOOL negated;
6832          int pdata;          int pdata;
6833          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
6834          if (ptype < 0) goto FAILED;          if (ptype < 0) goto FAILED;
6835          previous = code;          previous = code;
6836          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6837          *code++ = ptype;          *code++ = ptype;
6838          *code++ = pdata;          *code++ = pdata;
6839          }          }
# Line 6849  for (;; ptr++) Line 6842  for (;; ptr++)
6842        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6843        allowed. */        allowed. */
6844    
6845        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6846          {          {
6847          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6848          goto FAILED;          goto FAILED;
# Line 6859  for (;; ptr++) Line 6852  for (;; ptr++)
6852        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6853        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
6854        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
6855        Unicode property tests. Note that \b and \B do a one-character        Unicode property tests. Note that \b and \B do a one-character
6856        lookbehind. */        lookbehind. */
6857    
6858        else        else
6859          {          {
6860          if ((-c == ESC_b || -c == ESC_B) && cd->max_lookbehind == 0)          if ((escape == ESC_b || escape == ESC_B) && cd->max_lookbehind == 0)
6861            cd->max_lookbehind = 1;            cd->max_lookbehind = 1;
6862  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6863          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6864            {            {
6865            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6866            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6867            }            }
6868          else          else
6869  #endif  #endif
# Line 6878  for (;; ptr++) Line 6871  for (;; ptr++)
6871          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6872    
6873            {            {
6874            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
6875            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
6876            }            }
6877          }          }
6878        continue;        continue;
# Line 6889  for (;; ptr++) Line 6882  for (;; ptr++)
6882      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
6883      handle it as a data character. */      handle it as a data character. */
6884    
6885  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6886      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6887        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6888      else      else
# Line 6912  for (;; ptr++) Line 6905  for (;; ptr++)
6905      mclength = 1;      mclength = 1;
6906      mcbuffer[0] = c;      mcbuffer[0] = c;
6907    
6908  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6909      if (utf && HAS_EXTRALEN(c))      if (utf && HAS_EXTRALEN(c))
6910        ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr));        ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr));
6911  #endif  #endif
# Line 6922  for (;; ptr++) Line 6915  for (;; ptr++)
6915    
6916      ONE_CHAR:      ONE_CHAR:
6917      previous = code;      previous = code;
6918    
6919        /* For caseless UTF-8 mode when UCP support is available, check whether
6920        this character has more than one other case. If so, generate a special
6921        OP_PROP item instead of OP_CHARI. */
6922    
6923    #ifdef SUPPORT_UCP
6924        if (utf && (options & PCRE_CASELESS) != 0)
6925          {
6926          GETCHAR(c, mcbuffer);
6927          if ((c = UCD_CASESET(c)) != 0)
6928            {
6929            *code++ = OP_PROP;
6930            *code++ = PT_CLIST;
6931            *code++ = c;
6932            if (firstchar == REQ_UNSET) firstchar = zerofirstchar = REQ_NONE;
6933            break;
6934            }
6935          }
6936    #endif
6937    
6938        /* Caseful matches, or not one of the multicase characters. */
6939    
6940      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6941      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6942    
# Line 7173  for (;;) Line 7188  for (;;)
7188          *ptrptr = ptr;          *ptrptr = ptr;
7189          return FALSE;          return FALSE;
7190          }          }
7191        else        else
7192          {          {
7193          if (fixed_length > cd->max_lookbehind)          if (fixed_length > cd->max_lookbehind)
7194            cd->max_lookbehind = fixed_length;            cd->max_lookbehind = fixed_length;
7195          PUT(reverse_count, 0, fixed_length);          PUT(reverse_count, 0, fixed_length);
7196          }          }
7197        }        }
7198      }      }
# Line 7310  and the highest back reference was great Line 7325  and the highest back reference was great
7325  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
7326  of the more common cases more precisely.  of the more common cases more precisely.
7327    
7328    ... A second exception is when the .* appears inside an atomic group, because
7329    this prevents the number of characters it matches from being adjusted.
7330    
7331  Arguments:  Arguments:
7332    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7333    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
7334                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7335                    the less precise approach                    the less precise approach
7336    backref_map    the back reference bitmap    cd             points to the compile data block
7337      atomcount      atomic group level
7338    
7339  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7340  */  */
7341    
7342  static BOOL  static BOOL
7343  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7344    unsigned int backref_map)    compile_data *cd, int atomcount)
7345  {  {
7346  do {  do {
7347     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7334  do { Line 7353  do {
7353     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7354         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7355       {       {
7356       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7357       }       }
7358    
7359     /* Capturing brackets */     /* Capturing brackets */
# Line 7344  do { Line 7363  do {
7363       {       {
7364       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7365       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7366       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7367       }       }
7368    
7369     /* Other brackets */     /* Positive forward assertions and conditions */
7370    
7371     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ASSERT || op == OP_COND)
             op == OP_COND)  
7372       {       {
7373       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7374         }
7375    
7376       /* Atomic groups */
7377    
7378       else if (op == OP_ONCE || op == OP_ONCE_NC)
7379         {
7380         if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7381           return FALSE;
7382       }       }
7383    
7384     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7385     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
7386       group. */
7387    
7388     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7389               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7390       {       {
7391       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7392             atomcount > 0 || cd->had_pruneorskip)
7393         return FALSE;         return FALSE;
7394       }       }
7395    
7396     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7397    
7398     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;
7399    
7400     code += GET(code, 1);     code += GET(code, 1);
7401     }     }
7402  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7385  return TRUE; Line 7414  return TRUE;
7414  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
7415  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
7416  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7417  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7418    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7419    count, because once again the assumption no longer holds.
7420    
7421  Arguments:  Arguments:
7422    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7423    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
7424                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7425                    the less precise approach                    the less precise approach
7426    backref_map    the back reference bitmap    cd             points to the compile data
7427      atomcount      atomic group level
7428    
7429  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7430  */  */
7431    
7432  static BOOL  static BOOL
7433  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7434    unsigned int backref_map)    compile_data *cd, int atomcount)
7435  {  {
7436  do {  do {
7437     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7425  do { Line 7457  do {
7457         return FALSE;         return FALSE;
7458    
7459         default:     /* Assertion */         default:     /* Assertion */
7460         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7461         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7462         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7463         break;         break;
# Line 7439  do { Line 7471  do {
7471     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7472         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7473       {       {
7474       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7475       }       }
7476    
7477     /* Capturing brackets */     /* Capturing brackets */
# Line 7449  do { Line 7481  do {
7481       {       {
7482       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7483       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7484       if (!is_startline(scode, new_map, backref_map)) return FALSE;       if (!is_startline(scode, new_map, cd, atomcount)) return FALSE;
7485       }       }
7486    
7487     /* Other brackets */     /* Positive forward assertions */
7488    
7489     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)     else if (op == OP_ASSERT)
7490       {       {
7491       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7492       }       }
7493    
7494       /* Atomic brackets */
7495    
7496     /* .* means "start at start or after \n" if it isn't in brackets that     else if (op == OP_ONCE || op == OP_ONCE_NC)
7497     may be referenced. */       {
7498         if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE;
7499         }
7500    
7501       /* .* means "start at start or after \n" if it isn't in atomic brackets or
7502       brackets that may be referenced, as long as the pattern does not contain
7503       *PRUNE or *SKIP, because these break the feature. Consider, for example,
7504       /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the
7505       start of a line. */
7506    
7507     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)
7508       {       {
7509       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 ||
7510             atomcount > 0 || cd->had_pruneorskip)
7511           return FALSE;
7512       }       }
7513    
7514     /* Check for explicit circumflex */     /* Check for explicit circumflex; anything else gives a FALSE result. Note
7515       in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC
7516       because the number of characters matched by .* cannot be adjusted inside
7517       them. */
7518    
7519     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;
7520    
# Line 7589  Returns:        pointer to compiled data Line 7636  Returns:        pointer to compiled data
7636                  with errorptr and erroroffset set                  with errorptr and erroroffset set
7637  */  */
7638    
7639  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7640  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7641  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
7642    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7643  #else  #elif defined COMPILE_PCRE16
7644  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7645  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,
7646    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7647    #elif defined COMPILE_PCRE32
7648    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7649    pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr,
7650      int *erroroffset, const unsigned char *tables)
7651  #endif  #endif
7652  {  {
7653  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7654  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7655  #else  #elif defined COMPILE_PCRE16
7656  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7657    #elif defined COMPILE_PCRE32
7658    return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7659  #endif  #endif
7660  }  }
7661    
7662    
7663  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7664  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7665  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
7666    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7667  #else  #elif defined COMPILE_PCRE16
7668  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7669  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,
7670    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7671    #elif defined COMPILE_PCRE32
7672    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7673    pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr,
7674      const char **errorptr, int *erroroffset, const unsigned char *tables)
7675  #endif  #endif
7676  {  {
7677  REAL_PCRE *re;  REAL_PCRE *re;
# Line 7700  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7757  while (ptr[skipatstart] == CHAR_LEFT_PAR
7757    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7758      { skipatstart += 8; options |= PCRE_UTF16; continue; }      { skipatstart += 8; options |= PCRE_UTF16; continue; }
7759  #endif  #endif
7760    #ifdef COMPILE_PCRE32
7761      if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7762        { skipatstart += 8; options |= PCRE_UTF32; continue; }
7763    #endif
7764    else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UCP_RIGHTPAR, 4) == 0)    else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UCP_RIGHTPAR, 4) == 0)
7765      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
7766    else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_START_OPT_RIGHTPAR, 13) == 0)    else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_START_OPT_RIGHTPAR, 13) == 0)
# Line 7728  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7789  while (ptr[skipatstart] == CHAR_LEFT_PAR
7789    else break;    else break;
7790    }    }
7791    
7792  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */
7793  utf = (options & PCRE_UTF8) != 0;  utf = (options & PCRE_UTF8) != 0;
7794    
7795  /* Can't support UTF unless PCRE has been compiled to include the code. The  /* Can't support UTF unless PCRE has been compiled to include the code. The
# Line 7740  not used here. */ Line 7801  not used here. */
7801  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&
7802       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)
7803    {    {
7804  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7805    errorcode = ERR44;    errorcode = ERR44;
7806  #else  #elif defined COMPILE_PCRE16
7807    errorcode = ERR74;    errorcode = ERR74;
7808    #elif defined COMPILE_PCRE32
7809      errorcode = ERR77;
7810  #endif  #endif
7811    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;
7812    }    }
# Line 7907  re->name_count = cd->names_found; Line 7970  re->name_count = cd->names_found;
7970  re->ref_count = 0;  re->ref_count = 0;
7971  re->tables = (tables == PRIV(default_tables))? NULL : tables;  re->tables = (tables == PRIV(default_tables))? NULL : tables;
7972  re->nullpad = NULL;  re->nullpad = NULL;
7973    #ifdef COMPILE_PCRE32
7974    re->dummy1 = re->dummy2 = 0;
7975    #endif
7976    
7977  /* The starting points of the name/number translation table and of the code are  /* The starting points of the name/number translation table and of the code are
7978  passed around in the compile data block. The start/end pattern and initial  passed around in the compile data block. The start/end pattern and initial
# Line 7926  cd->start_code = codestart; Line 7992  cd->start_code = codestart;
7992  cd->hwm = (pcre_uchar *)(cd->start_workspace);  cd->hwm = (pcre_uchar *)(cd->start_workspace);
7993  cd->req_varyopt = 0;  cd->req_varyopt = 0;
7994  cd->had_accept = FALSE;  cd->had_accept = FALSE;
7995    cd->had_pruneorskip = FALSE;
7996  cd->check_lookbehind = FALSE;  cd->check_lookbehind = FALSE;
7997  cd->open_caps = NULL;  cd->open_caps = NULL;
7998    
# Line 8049  if (errorcode != 0) Line 8116  if (errorcode != 0)
8116    }    }
8117    
8118  /* If the anchored option was not passed, set the flag if we can determine that  /* If the anchored option was not passed, set the flag if we can determine that
8119  the pattern is anchored by virtue of ^ characters or \A or anything else (such  the pattern is anchored by virtue of ^ characters or \A or anything else, such
8120  as starting with .* when DOTALL is set).  as starting with non-atomic .* when DOTALL is set and there are no occurrences
8121    of *PRUNE or *SKIP.
8122    
8123  Otherwise, if we know what the first byte has to be, save it, because that  Otherwise, if we know what the first byte has to be, save it, because that
8124  speeds up unanchored matches no end. If not, see if we can set the  speeds up unanchored matches no end. If not, see if we can set the
8125  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches
8126  start with ^. and also when all branches start with .* for non-DOTALL matches.  start with ^. and also when all branches start with non-atomic .* for
8127  */  non-DOTALL matches when *PRUNE and SKIP are not present. */
8128    
8129  if ((re->options & PCRE_ANCHORED) == 0)  if ((re->options & PCRE_ANCHORED) == 0)
8130    {    {
8131    if (is_anchored(codestart, 0, cd->backref_map))    if (is_anchored(codestart, 0, cd, 0)) re->options |= PCRE_ANCHORED;
     re->options |= PCRE_ANCHORED;  
8132    else    else
8133      {      {
8134      if (firstchar < 0)      if (firstchar < 0)
8135        firstchar = find_firstassertedchar(codestart, FALSE);        firstchar = find_firstassertedchar(codestart, FALSE);
8136      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */
8137        {        {
8138  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8139        re->first_char = firstchar & 0xff;        re->first_char = firstchar & 0xff;
8140  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8141        re->first_char = firstchar & 0xffff;        re->first_char = firstchar & 0xffff;
8142  #endif  #elif defined COMPILE_PCRE32
8143          re->first_char = firstchar & ~REQ_MASK;
8144  #endif  #endif
8145        if ((firstchar & REQ_CASELESS) != 0)        if ((firstchar & REQ_CASELESS) != 0)
8146          {          {
# Line 8098  if ((re->options & PCRE_ANCHORED) == 0) Line 8165  if ((re->options & PCRE_ANCHORED) == 0)
8165    
8166        re->flags |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
8167        }        }
8168      else if (is_startline(codestart, 0, cd->backref_map))  
8169        re->flags |= PCRE_STARTLINE;      else if (is_startline(codestart, 0, cd, 0)) re->flags |= PCRE_STARTLINE;
8170      }      }
8171    }    }
8172    
# Line 8110  bytes. */ Line 8177  bytes. */
8177  if (reqchar >= 0 &&  if (reqchar >= 0 &&
8178       ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0))       ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0))
8179    {    {
8180  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8181    re->req_char = reqchar & 0xff;    re->req_char = reqchar & 0xff;
8182  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8183    re->req_char = reqchar & 0xffff;    re->req_char = reqchar & 0xffff;
8184  #endif  #elif defined COMPILE_PCRE32
8185      re->req_char = reqchar & ~REQ_MASK;
8186  #endif  #endif
8187    if ((reqchar & REQ_CASELESS) != 0)    if ((reqchar & REQ_CASELESS) != 0)
8188      {      {
# Line 8167  if ((re->flags & PCRE_REQCHSET) != 0) Line 8234  if ((re->flags & PCRE_REQCHSET) != 0)
8234      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
8235    }    }
8236    
8237  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8238  pcre_printint((pcre *)re, stdout, TRUE);  pcre_printint((pcre *)re, stdout, TRUE);
8239  #else  #elif defined COMPILE_PCRE16
8240  pcre16_printint((pcre *)re, stdout, TRUE);  pcre16_printint((pcre *)re, stdout, TRUE);
8241    #elif defined COMPILE_PCRE32
8242    pcre32_printint((pcre *)re, stdout, TRUE);
8243  #endif  #endif
8244    
8245  /* This check is done here in the debugging case so that the code that  /* This check is done here in the debugging case so that the code that
# Line 8186  if (code - codestart > length) Line 8255  if (code - codestart > length)
8255    }    }
8256  #endif   /* PCRE_DEBUG */  #endif   /* PCRE_DEBUG */
8257    
8258  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8259  return (pcre *)re;  return (pcre *)re;
8260  #else  #elif defined COMPILE_PCRE16
8261  return (pcre16 *)re;  return (pcre16 *)re;
8262    #elif defined COMPILE_PCRE32
8263    return (pcre32 *)re;
8264  #endif  #endif
8265  }  }
8266    

Legend:
Removed from v.964  
changed lines
  Added in v.1066

  ViewVC Help
Powered by ViewVC 1.1.5