/[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 1124 by chpe, Thu Oct 18 18:34:53 2012 UTC
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  is also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. We do not need to select pcre16_printint.c specially, because the  library. We do not need to select pcre16_printint.c specially, because the
59  COMPILE_PCREx macro will already be appropriately set. */  COMPILE_PCREx macro will already be appropriately set. */
# Line 68  COMPILE_PCREx macro will already be appr Line 68  COMPILE_PCREx macro will already be appr
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 77  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,
88        pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *,
89        compile_data *, int *);
90    
91    
92    
93  /*************************************************  /*************************************************
94  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 110  overrun before it actually does run off Line 122  overrun before it actually does run off
122    
123  /* Private flags added to firstchar and reqchar. */  /* Private flags added to firstchar and reqchar. */
124    
125  #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */  #define REQ_CASELESS    (1 << 0)        /* Indicates caselessness */
126  #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */  #define REQ_VARY        (1 << 1)        /* Reqchar followed non-literal item */
127    /* Negative values for the firstchar and reqchar flags */
128    #define REQ_UNSET       (-2)
129    #define REQ_NONE        (-1)
130    
131  /* Repeated character flags. */  /* Repeated character flags. */
132    
# Line 490  static const char error_texts[] = Line 505  static const char error_texts[] =
505    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
506    "invalid UTF-16 string\0"    "invalid UTF-16 string\0"
507    /* 75 */    /* 75 */
508    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
509      "character value in \\u.... sequence is too large\0"
510      "invalid UTF-32 string\0"
511    ;    ;
512    
513  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 630  static const pcre_uint8 ebcdic_chartab[] Line 647  static const pcre_uint8 ebcdic_chartab[]
647  #endif  #endif
648    
649    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
   
650    
651  /*************************************************  /*************************************************
652  *            Find an error text                  *  *            Find an error text                  *
# Line 741  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 751  return (*p == CHAR_RIGHT_CURLY_BRACKET);
751  *************************************************/  *************************************************/
752    
753  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
754  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
755  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
756  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
757  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
758  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
759  sequence.  escape sequence.
760    
761  Arguments:  Arguments:
762    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
763      chptr          points to the data character
764    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
765    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
766    options        the options bits    options        the options bits
767    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
768    
769  Returns:         zero or positive => a data character  Returns:         zero => a data character
770                   negative => a special escape sequence                   positive => a special escape sequence
771                     negative => a back reference
772                   on error, errorcodeptr is set                   on error, errorcodeptr is set
773  */  */
774    
775  static int  static int
776  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
777    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
778  {  {
779  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
780  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
781  const pcre_uchar *ptr = *ptrptr + 1;  const pcre_uchar *ptr = *ptrptr + 1;
782  pcre_int32 c;  pcre_uint32 c;
783    int escape = 0;
784  int i;  int i;
785    
786  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
# Line 784  Otherwise further processing may be requ Line 797  Otherwise further processing may be requ
797  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
798  /* Not alphanumeric */  /* Not alphanumeric */
799  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
800  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
801    
802  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
803  /* Not alphanumeric */  /* Not alphanumeric */
804  else if (c < 'a' || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
805  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
806  #endif  #endif
807    
808  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
# Line 797  else if ((i = escapes[c - 0x48]) != 0) Line 810  else if ((i = escapes[c - 0x48]) != 0)
810  else  else
811    {    {
812    const pcre_uchar *oldptr;    const pcre_uchar *oldptr;
813    BOOL braced, negated;    BOOL braced, negated, overflow;
814      int s;
815    
816    switch (c)    switch (c)
817      {      {
# Line 822  else Line 836  else
836          c = 0;          c = 0;
837          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
838            {            {
839            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
840  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
841            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
842            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 831  else Line 845  else
845            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
846  #endif  #endif
847            }            }
848    
849    #if defined COMPILE_PCRE8
850            if (c > (utf ? 0x10ffff : 0xff))
851    #elif defined COMPILE_PCRE16
852            if (c > (utf ? 0x10ffff : 0xffff))
853    #elif defined COMPILE_PCRE32
854            if (utf && c > 0x10ffff)
855    #endif
856              {
857              *errorcodeptr = ERR76;
858              }
859            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
860          }          }
861        }        }
862      else      else
# Line 857  else Line 883  else
883      (3) For Oniguruma compatibility we also support \g followed by a name or a      (3) For Oniguruma compatibility we also support \g followed by a name or a
884      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
885      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
886      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
887    
888      case CHAR_g:      case CHAR_g:
889      if (isclass) break;      if (isclass) break;
890      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
891        {        {
892        c = -ESC_g;        escape = ESC_g;
893        break;        break;
894        }        }
895    
# Line 876  else Line 902  else
902          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
903        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
904          {          {
905          c = -ESC_k;          escape = ESC_k;
906          break;          break;
907          }          }
908        braced = TRUE;        braced = TRUE;
# Line 892  else Line 918  else
918      else negated = FALSE;      else negated = FALSE;
919    
920      /* The integer range is limited by the machine's int representation. */      /* The integer range is limited by the machine's int representation. */
921      c = 0;      s = 0;
922        overflow = FALSE;
923      while (IS_DIGIT(ptr[1]))      while (IS_DIGIT(ptr[1]))
924        {        {
925        if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */        if (s > INT_MAX / 10 - 1) /* Integer overflow */
926          {          {
927          c = -1;          overflow = TRUE;
928          break;          break;
929          }          }
930        c = c * 10 + *(++ptr) - CHAR_0;        s = s * 10 + (int)(*(++ptr) - CHAR_0);
931        }        }
932      if (((unsigned int)c) > INT_MAX) /* Integer overflow */      if (overflow) /* Integer overflow */
933        {        {
934        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
935          ptr++;          ptr++;
# Line 916  else Line 943  else
943        break;        break;
944        }        }
945    
946      if (c == 0)      if (s == 0)
947        {        {
948        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
949        break;        break;
# Line 924  else Line 951  else
951    
952      if (negated)      if (negated)
953        {        {
954        if (c > bracount)        if (s > bracount)
955          {          {
956          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
957          break;          break;
958          }          }
959        c = bracount - (c - 1);        s = bracount - (s - 1);
960        }        }
961    
962      c = -(ESC_REF + c);      escape = -s;
963      break;      break;
964    
965      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 954  else Line 981  else
981        {        {
982        oldptr = ptr;        oldptr = ptr;
983        /* The integer range is limited by the machine's int representation. */        /* The integer range is limited by the machine's int representation. */
984        c -= CHAR_0;        s = (int)(c -CHAR_0);
985          overflow = FALSE;
986        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
987          {          {
988          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */          if (s > INT_MAX / 10 - 1) /* Integer overflow */
989            {            {
990            c = -1;            overflow = TRUE;
991            break;            break;
992            }            }
993          c = c * 10 + *(++ptr) - CHAR_0;          s = s * 10 + (int)(*(++ptr) - CHAR_0);
994          }          }
995        if (((unsigned int)c) > INT_MAX) /* Integer overflow */        if (overflow) /* Integer overflow */
996          {          {
997          while (IS_DIGIT(ptr[1]))          while (IS_DIGIT(ptr[1]))
998            ptr++;            ptr++;
999          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1000          break;          break;
1001          }          }
1002        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1003          {          {
1004          c = -(ESC_REF + c);          escape = -s;
1005          break;          break;
1006          }          }
1007        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 1020  else Line 1048  else
1048          c = 0;          c = 0;
1049          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1050            {            {
1051            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1052  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1053            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1054            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 1038  else Line 1066  else
1066        const pcre_uchar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1067    
1068        c = 0;        c = 0;
1069          overflow = FALSE;
1070        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1071          {          {
1072          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1073          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1074    
1075    #ifdef COMPILE_PCRE32
1076            if (c >= 0x10000000l) { overflow = TRUE; break; }
1077    #endif
1078    
1079  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1080          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1081          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 1051  else Line 1084  else
1084          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1085  #endif  #endif
1086    
1087  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1088          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1089  #else  #elif defined COMPILE_PCRE16
1090  #ifdef COMPILE_PCRE16          if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1091          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }  #elif defined COMPILE_PCRE32
1092  #endif          if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1093  #endif  #endif
1094          }          }
1095    
1096        if (c < 0)        if (overflow)
1097          {          {
1098          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1099          *errorcodeptr = ERR34;          *errorcodeptr = ERR34;
# Line 1082  else Line 1115  else
1115      c = 0;      c = 0;
1116      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1117        {        {
1118        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1119        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1120  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1121        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 1141  else Line 1174  else
1174  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1175  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1176    
1177  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1178       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1179    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1180    
1181  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1182    
1183  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1184    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1185    
1186  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1187    
1188  *ptrptr = ptr;  *ptrptr = ptr;
1189  return c;  *chptr = c;
1190    return escape;
1191  }  }
1192    
   
   
1193  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1194  /*************************************************  /*************************************************
1195  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1180  Returns:         type value from ucp_typ Line 1212  Returns:         type value from ucp_typ
1212  static int  static int
1213  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)
1214  {  {
1215  int c, i, bot, top;  pcre_uchar c;
1216    int i, bot, top;
1217  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1218  pcre_uchar name[32];  pcre_uchar name[32];
1219    
# Line 1227  top = PRIV(utt_size); Line 1260  top = PRIV(utt_size);
1260    
1261  while (bot < top)  while (bot < top)
1262    {    {
1263      int r;
1264    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1265    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);
1266    if (c == 0)    if (r == 0)
1267      {      {
1268      *dptr = PRIV(utt)[i].value;      *dptr = PRIV(utt)[i].value;
1269      return PRIV(utt)[i].type;      return PRIV(utt)[i].type;
1270      }      }
1271    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1272    }    }
1273    
1274  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
# Line 1279  int max = -1; Line 1313  int max = -1;
1313  /* 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
1314  an integer overflow. */  an integer overflow. */
1315    
1316  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1317  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1318    {    {
1319    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1294  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1328  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1328    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1329      {      {
1330      max = 0;      max = 0;
1331      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1332      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1333        {        {
1334        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1349  Arguments: Line 1383  Arguments:
1383    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1384    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1385    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1386    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1387    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1388    
1389  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 1425  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1459  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1459      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1460          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1461        {        {
1462        int term;        pcre_uchar term;
1463        const pcre_uchar *thisname;        const pcre_uchar *thisname;
1464        *count += 1;        *count += 1;
1465        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
# Line 1433  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1467  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1467        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1468        thisname = ptr;        thisname = ptr;
1469        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1470        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1471            STRNCMP_UC_UC(name, thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1472          return *count;          return *count;
1473        term++;        term++;
1474        }        }
# Line 1583  Arguments: Line 1617  Arguments:
1617    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1618    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1619    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1620    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1621    
1622  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1623  */  */
# Line 1686  and doing the check at the end; a flag s Line 1720  and doing the check at the end; a flag s
1720    
1721  Arguments:  Arguments:
1722    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1723    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1724    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1725    cd       the "compile data" structure    cd       the "compile data" structure
1726    
# Line 1712  for (;;) Line 1746  for (;;)
1746    {    {
1747    int d;    int d;
1748    pcre_uchar *ce, *cs;    pcre_uchar *ce, *cs;
1749    register int op = *cc;    register pcre_uchar op = *cc;
1750    
1751    switch (op)    switch (op)
1752      {      {
# Line 1832  for (;;) Line 1866  for (;;)
1866      case OP_EXACTI:      case OP_EXACTI:
1867      case OP_NOTEXACT:      case OP_NOTEXACT:
1868      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1869      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1870      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1871  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
1872      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
# Line 1841  for (;;) Line 1875  for (;;)
1875    
1876      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1877      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1878      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)
1879          cc += 2;
1880      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1881      break;      break;
1882    
# Line 1876  for (;;) Line 1911  for (;;)
1911    
1912      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1913    
1914  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1915      case OP_XCLASS:      case OP_XCLASS:
1916      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1917      /* Fall through */      /* Fall through */
# Line 1899  for (;;) Line 1934  for (;;)
1934        case OP_CRRANGE:        case OP_CRRANGE:
1935        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1936        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1937        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1938        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
1939        break;        break;
1940    
# Line 2015  length. Line 2050  length.
2050    
2051  Arguments:  Arguments:
2052    code        points to start of expression    code        points to start of expression
2053    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2054    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2055    
2056  Returns:      pointer to the opcode for the bracket, or NULL if not found  Returns:      pointer to the opcode for the bracket, or NULL if not found
# Line 2026  PRIV(find_bracket)(const pcre_uchar *cod Line 2061  PRIV(find_bracket)(const pcre_uchar *cod
2061  {  {
2062  for (;;)  for (;;)
2063    {    {
2064    register int c = *code;    register pcre_uchar c = *code;
2065    
2066    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2067    
# Line 2049  for (;;) Line 2084  for (;;)
2084    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2085             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2086      {      {
2087      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2088      if (n == number) return (pcre_uchar *)code;      if (n == number) return (pcre_uchar *)code;
2089      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
2090      }      }
# Line 2079  for (;;) Line 2114  for (;;)
2114        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2115        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2116        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2117        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2118          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2119        break;        break;
2120    
2121        case OP_MARK:        case OP_MARK:
# Line 2102  for (;;) Line 2137  for (;;)
2137    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
2138    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2139    
2140  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2141      if (utf) switch(c)      if (utf) switch(c)
2142        {        {
2143        case OP_CHAR:        case OP_CHAR:
# Line 2154  instance of OP_RECURSE. Line 2189  instance of OP_RECURSE.
2189    
2190  Arguments:  Arguments:
2191    code        points to start of expression    code        points to start of expression
2192    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2193    
2194  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
2195  */  */
# Line 2164  find_recurse(const pcre_uchar *code, BOO Line 2199  find_recurse(const pcre_uchar *code, BOO
2199  {  {
2200  for (;;)  for (;;)
2201    {    {
2202    register int c = *code;    register pcre_uchar c = *code;
2203    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2204    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2205    
# Line 2199  for (;;) Line 2234  for (;;)
2234        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2235        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2236        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2237        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2238          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2239        break;        break;
2240    
2241        case OP_MARK:        case OP_MARK:
# Line 2222  for (;;) Line 2257  for (;;)
2257      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
2258      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2259    
2260  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2261      if (utf) switch(c)      if (utf) switch(c)
2262        {        {
2263        case OP_CHAR:        case OP_CHAR:
# Line 2308  bracket whose current branch will alread Line 2343  bracket whose current branch will alread
2343  Arguments:  Arguments:
2344    code        points to start of search    code        points to start of search
2345    endcode     points to where to stop    endcode     points to where to stop
2346    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2347    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2348    
2349  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2318  static BOOL Line 2353  static BOOL
2353  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2354    BOOL utf, compile_data *cd)    BOOL utf, compile_data *cd)
2355  {  {
2356  register int c;  register pcre_uchar c;
2357  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2358       code < endcode;       code < endcode;
2359       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
# Line 2352  for (code = first_significant_code(code Line 2387  for (code = first_significant_code(code
2387      /* Test for forward reference */      /* Test for forward reference */
2388    
2389      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2390        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2391    
2392      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2393    
# Line 2525  for (code = first_significant_code(code Line 2560  for (code = first_significant_code(code
2560      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2561      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2562      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2563      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2564        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2565      break;      break;
2566    
2567      /* End of branch */      /* End of branch */
# Line 2541  for (code = first_significant_code(code Line 2576  for (code = first_significant_code(code
2576      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2577      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2578    
2579  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2580      case OP_STAR:      case OP_STAR:
2581      case OP_STARI:      case OP_STARI:
2582      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2607  Arguments: Line 2642  Arguments:
2642    code        points to start of the recursion    code        points to start of the recursion
2643    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2644    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2645    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2646    cd          pointers to tables etc    cd          pointers to tables etc
2647    
2648  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2673  Returns:   TRUE or FALSE Line 2708  Returns:   TRUE or FALSE
2708  static BOOL  static BOOL
2709  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2710  {  {
2711  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2712  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2713  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2714    {    {
# Line 2722  register int yield = 0; Line 2757  register int yield = 0;
2757  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2758    {    {
2759    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2760      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2761    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2762    yield++;    yield++;
2763    }    }
# Line 2754  value in the reference (which is a group Line 2789  value in the reference (which is a group
2789  Arguments:  Arguments:
2790    group      points to the start of the group    group      points to the start of the group
2791    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2792    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2793    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2794    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
2795    
# Line 2777  while ((ptr = (pcre_uchar *)find_recurse Line 2812  while ((ptr = (pcre_uchar *)find_recurse
2812    
2813    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2814      {      {
2815      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2816      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2817        {        {
2818        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2790  while ((ptr = (pcre_uchar *)find_recurse Line 2825  while ((ptr = (pcre_uchar *)find_recurse
2825    
2826    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2827      {      {
2828      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2829      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2830      }      }
2831    
# Line 2858  PUT(previous_callout, 2 + LINK_SIZE, len Line 2893  PUT(previous_callout, 2 + LINK_SIZE, len
2893  *************************************************/  *************************************************/
2894    
2895  /* 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
2896  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
2897  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
2898  start address.  start address. A character with multiple other cases is returned on its own
2899    with a special return value.
2900    
2901  Arguments:  Arguments:
2902    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2868  Arguments: Line 2904  Arguments:
2904    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2905    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2906    
2907  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2908                   0 when a range is returned
2909                  >0 the CASESET offset for char with multiple other cases
2910                    in this case, ocptr contains the original
2911  */  */
2912    
2913  static BOOL  static int
2914  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2915    unsigned int *odptr)    pcre_uint32 *odptr)
2916  {  {
2917  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2918    int co;
2919    
2920    /* Find the first character that has an other case. If it has multiple other
2921    cases, return its case offset value. */
2922    
2923  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2924    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2925      if ((co = UCD_CASESET(c)) != 0)
2926        {
2927        *ocptr = c++;   /* Character that has the set */
2928        *cptr = c;      /* Rest of input range */
2929        return co;
2930        }
2931      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2932      }
2933    
2934  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2935    
2936  *ocptr = othercase;  *ocptr = othercase;
2937  next = othercase + 1;  next = othercase + 1;
# Line 2891  for (++c; c <= d; c++) Line 2942  for (++c; c <= d; c++)
2942    next++;    next++;
2943    }    }
2944    
2945  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2946  *cptr = c;  *cptr = c;             /* Rest of input range */
2947    return 0;
 return TRUE;  
2948  }  }
2949    
2950    
# Line 2916  Returns:       TRUE if auto-possessifyin Line 2966  Returns:       TRUE if auto-possessifyin
2966  */  */
2967    
2968  static BOOL  static BOOL
2969  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(pcre_uint32 c, int ptype, int pdata, BOOL negated)
2970  {  {
2971    #ifdef SUPPORT_UCP
2972    const pcre_uint32 *p;
2973    #endif
2974    
2975  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2976    
2977  switch(ptype)  switch(ptype)
2978    {    {
2979    case PT_LAMP:    case PT_LAMP:
# Line 2956  switch(ptype) Line 3011  switch(ptype)
3011    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3012            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3013            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3014    
3015    #ifdef SUPPORT_UCP
3016      case PT_CLIST:
3017      p = PRIV(ucd_caseless_sets) + prop->caseset;
3018      for (;;)
3019        {
3020        if ((unsigned int)c < *p) return !negated;
3021        if ((unsigned int)c == *p++) return negated;
3022        }
3023      break;  /* Control never reaches here */
3024    #endif
3025    }    }
3026    
3027  return FALSE;  return FALSE;
3028  }  }
3029  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2973  sense to automatically possessify the re Line 3040  sense to automatically possessify the re
3040    
3041  Arguments:  Arguments:
3042    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3043    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3044    ptr           next character in pattern    ptr           next character in pattern
3045    options       options bits    options       options bits
3046    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2985  static BOOL Line 3052  static BOOL
3052  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3053    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3054  {  {
3055  pcre_int32 c, next;  pcre_uint32 c = NOTACHAR;
3056  int op_code = *previous++;  pcre_uint32 next;
3057    int escape;
3058    pcre_uchar op_code = *previous++;
3059    
3060  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3061    
# Line 3017  value is a character, a negative value i Line 3086  value is a character, a negative value i
3086  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3087    {    {
3088    int temperrorcode = 0;    int temperrorcode = 0;
3089    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3090    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3091    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3092    }    }
3093  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
3094    {    {
3095      escape = 0;
3096  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3097    if (utf) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3098  #endif  #endif
# Line 3059  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3129  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3129    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)
3130      return FALSE;      return FALSE;
3131    
3132  /* 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. */  
3133    
3134  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3135        op_code == OP_NOT || op_code == OP_NOTI)
3136    {    {
   case OP_CHAR:  
3137  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3138    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3139  #else  #else
3140    c = *previous;    c = *previous;
3141  #endif  #endif
3142    return c != next;    }
3143    
3144    /* 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
3145    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3146    
3147    case OP_CHARI:  if (escape == 0)
3148  #ifdef SUPPORT_UTF    {
3149    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3150  #else    case, which maps to the special PT_CLIST property. Check this first. */
3151    c = *previous;  
 #endif  
   if (c == next) return FALSE;  
 #ifdef SUPPORT_UTF  
   if (utf)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3152  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3153      othercase = UCD_OTHERCASE((unsigned int)next);    if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3154  #else      {
3155      othercase = NOTACHAR;      int ocs = UCD_CASESET(next);
3156  #endif      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
     return (unsigned int)c != othercase;  
3157      }      }
   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;  
3158  #endif  #endif
   return c == next;  
3159    
3160    case OP_NOTI:    switch(op_code)
3161        {
3162        case OP_CHAR:
3163        return c != next;
3164    
3165        /* For CHARI (caseless character) we must check the other case. If we have
3166        Unicode property support, we can use it to test the other case of
3167        high-valued characters. We know that next can have only one other case,
3168        because multi-other-case characters are dealt with above. */
3169    
3170        case OP_CHARI:
3171        if (c == next) return FALSE;
3172  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3173    GETCHARTEST(c, previous);      if (utf)
3174          {
3175          pcre_uint32 othercase;
3176          if (next < 128) othercase = cd->fcc[next]; else
3177    #ifdef SUPPORT_UCP
3178          othercase = UCD_OTHERCASE(next);
3179  #else  #else
3180    c = *previous;        othercase = NOTACHAR;
3181  #endif  #endif
3182    if (c == next) return TRUE;        return c != othercase;
3183          }
3184        else
3185    #endif  /* SUPPORT_UTF */
3186        return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3187    
3188        case OP_NOT:
3189        return c == next;
3190    
3191        case OP_NOTI:
3192        if (c == next) return TRUE;
3193  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3194    if (utf)      if (utf)
3195      {        {
3196      unsigned int othercase;        pcre_uint32 othercase;
3197      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3198  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3199      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3200  #else  #else
3201      othercase = NOTACHAR;        othercase = NOTACHAR;
3202  #endif  #endif
3203      return (unsigned int)c == othercase;        return c == othercase;
3204      }        }
3205    else      else
3206  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3207    return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3208    
3209    /* 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.
3210    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. */
3211    
3212    case OP_DIGIT:      case OP_DIGIT:
3213    return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3214    
3215    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3216    return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3217    
3218    case OP_WHITESPACE:      case OP_WHITESPACE:
3219    return next > 255 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3220    
3221    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3222    return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3223    
3224    case OP_WORDCHAR:      case OP_WORDCHAR:
3225    return next > 255 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3226    
3227    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3228    return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3229    
3230    case OP_HSPACE:      case OP_HSPACE:
3231    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3232    switch(next)      switch(next)
3233      {        {
3234      case 0x09:        HSPACE_CASES:
3235      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;  
     }  
3236    
3237    case OP_ANYNL:        default:
3238    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3239    case OP_NOT_VSPACE:        }
3240    switch(next)  
3241      {      case OP_ANYNL:
3242      case 0x0a:      case OP_VSPACE:
3243      case 0x0b:      case OP_NOT_VSPACE:
3244      case 0x0c:      switch(next)
3245      case 0x0d:        {
3246      case 0x85:        VSPACE_CASES:
3247      case 0x2028:        return op_code == OP_NOT_VSPACE;
3248      case 0x2029:  
3249      return op_code == OP_NOT_VSPACE;        default:
3250      default:        return op_code != OP_NOT_VSPACE;
3251      return op_code != OP_NOT_VSPACE;        }
     }  
3252    
3253  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3254    case OP_PROP:      case OP_PROP:
3255    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, (int)previous[0], (int)previous[1], FALSE);
3256    
3257    case OP_NOTPROP:      case OP_NOTPROP:
3258    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, (int)previous[0], (int)previous[1], TRUE);
3259  #endif  #endif
3260    
3261    default:      default:
3262    return FALSE;      return FALSE;
3263        }
3264    }    }
3265    
   
3266  /* 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
3267  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
3268  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 3273  switch(op_code)
3273    {    {
3274    case OP_CHAR:    case OP_CHAR:
3275    case OP_CHARI:    case OP_CHARI:
3276  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3277      {      {
3278      case ESC_d:      case ESC_d:
3279      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
# Line 3248  switch(op_code) Line 3297  switch(op_code)
3297      case ESC_H:      case ESC_H:
3298      switch(c)      switch(c)
3299        {        {
3300        case 0x09:        HSPACE_CASES:
3301        case 0x20:        return escape != ESC_h;
3302        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;  
3303        default:        default:
3304        return -next == ESC_h;        return escape == ESC_h;
3305        }        }
3306    
3307      case ESC_v:      case ESC_v:
3308      case ESC_V:      case ESC_V:
3309      switch(c)      switch(c)
3310        {        {
3311        case 0x0a:        VSPACE_CASES:
3312        case 0x0b:        return escape != ESC_v;
3313        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3314        default:        default:
3315        return -next == ESC_v;        return escape == ESC_v;
3316        }        }
3317    
3318      /* 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
3319      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3320      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3321    
3322  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3323      case ESC_du:      case ESC_du:
# Line 3301  switch(op_code) Line 3328  switch(op_code)
3328      case ESC_SU:      case ESC_SU:
3329        {        {
3330        int temperrorcode = 0;        int temperrorcode = 0;
3331        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3332        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3333        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3334        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3335        }        }
# Line 3329  switch(op_code) Line 3356  switch(op_code)
3356    
3357        /* Do the property check. */        /* Do the property check. */
3358    
3359        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3360        }        }
3361  #endif  #endif
3362    
# Line 3344  switch(op_code) Line 3371  switch(op_code)
3371    these op-codes are never generated.) */    these op-codes are never generated.) */
3372    
3373    case OP_DIGIT:    case OP_DIGIT:
3374    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3375           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3376    
3377    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3378    return next == -ESC_d;    return escape == ESC_d;
3379    
3380    case OP_WHITESPACE:    case OP_WHITESPACE:
3381    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3382    
3383    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3384    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;
3385    
3386    case OP_HSPACE:    case OP_HSPACE:
3387    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3388           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3389    
3390    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3391    return next == -ESC_h;    return escape == ESC_h;
3392    
3393    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3394    case OP_ANYNL:    case OP_ANYNL:
3395    case OP_VSPACE:    case OP_VSPACE:
3396    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3397    
3398    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3399    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3400    
3401    case OP_WORDCHAR:    case OP_WORDCHAR:
3402    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3403           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3404    
3405    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3406    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3407    
3408    default:    default:
3409    return FALSE;    return FALSE;
# Line 3388  switch(op_code) Line 3415  switch(op_code)
3415    
3416    
3417  /*************************************************  /*************************************************
3418    *        Add a character or range to a class     *
3419    *************************************************/
3420    
3421    /* This function packages up the logic of adding a character or range of
3422    characters to a class. The character values in the arguments will be within the
3423    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3424    mutually recursive with the function immediately below.
3425    
3426    Arguments:
3427      classbits     the bit map for characters < 256
3428      uchardptr     points to the pointer for extra data
3429      options       the options word
3430      cd            contains pointers to tables etc.
3431      start         start of range character
3432      end           end of range character
3433    
3434    Returns:        the number of < 256 characters added
3435                    the pointer to extra data is updated
3436    */
3437    
3438    static int
3439    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3440      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3441    {
3442    pcre_uint32 c;
3443    int n8 = 0;
3444    
3445    /* If caseless matching is required, scan the range and process alternate
3446    cases. In Unicode, there are 8-bit characters that have alternate cases that
3447    are greater than 255 and vice-versa. Sometimes we can just extend the original
3448    range. */
3449    
3450    if ((options & PCRE_CASELESS) != 0)
3451      {
3452    #ifdef SUPPORT_UCP
3453      if ((options & PCRE_UTF8) != 0)
3454        {
3455        int rc;
3456        pcre_uint32 oc, od;
3457    
3458        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3459        c = start;
3460    
3461        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3462          {
3463          /* Handle a single character that has more than one other case. */
3464    
3465          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3466            PRIV(ucd_caseless_sets) + rc, oc);
3467    
3468          /* Do nothing if the other case range is within the original range. */
3469    
3470          else if (oc >= start && od <= end) continue;
3471    
3472          /* Extend the original range if there is overlap, noting that if oc < c, we
3473          can't have od > end because a subrange is always shorter than the basic
3474          range. Otherwise, use a recursive call to add the additional range. */
3475    
3476          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3477          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3478          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3479          }
3480        }
3481      else
3482    #endif  /* SUPPORT_UCP */
3483    
3484      /* Not UTF-mode, or no UCP */
3485    
3486      for (c = start; c <= end && c < 256; c++)
3487        {
3488        SETBIT(classbits, cd->fcc[c]);
3489        n8++;
3490        }
3491      }
3492    
3493    /* Now handle the original range. Adjust the final value according to the bit
3494    length - this means that the same lists of (e.g.) horizontal spaces can be used
3495    in all cases. */
3496    
3497    #if defined COMPILE_PCRE8
3498    #ifdef SUPPORT_UTF
3499      if ((options & PCRE_UTF8) == 0)
3500    #endif
3501      if (end > 0xff) end = 0xff;
3502    
3503    #elif defined COMPILE_PCRE16
3504    #ifdef SUPPORT_UTF
3505      if ((options & PCRE_UTF16) == 0)
3506    #endif
3507      if (end > 0xffff) end = 0xffff;
3508    
3509    #endif /* COMPILE_PCRE[8|16] */
3510    
3511    /* If all characters are less than 256, use the bit map. Otherwise use extra
3512    data. */
3513    
3514    if (end < 0x100)
3515      {
3516      for (c = start; c <= end; c++)
3517        {
3518        n8++;
3519        SETBIT(classbits, c);
3520        }
3521      }
3522    
3523    else
3524      {
3525      pcre_uchar *uchardata = *uchardptr;
3526    
3527    #ifdef SUPPORT_UTF
3528      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3529        {
3530        if (start < end)
3531          {
3532          *uchardata++ = XCL_RANGE;
3533          uchardata += PRIV(ord2utf)(start, uchardata);
3534          uchardata += PRIV(ord2utf)(end, uchardata);
3535          }
3536        else if (start == end)
3537          {
3538          *uchardata++ = XCL_SINGLE;
3539          uchardata += PRIV(ord2utf)(start, uchardata);
3540          }
3541        }
3542      else
3543    #endif  /* SUPPORT_UTF */
3544    
3545      /* Without UTF support, character values are constrained by the bit length,
3546      and can only be > 256 for 16-bit and 32-bit libraries. */
3547    
3548    #ifdef COMPILE_PCRE8
3549        {}
3550    #else
3551      if (start < end)
3552        {
3553        *uchardata++ = XCL_RANGE;
3554        *uchardata++ = start;
3555        *uchardata++ = end;
3556        }
3557      else if (start == end)
3558        {
3559        *uchardata++ = XCL_SINGLE;
3560        *uchardata++ = start;
3561        }
3562    #endif
3563    
3564      *uchardptr = uchardata;   /* Updata extra data pointer */
3565      }
3566    
3567    return n8;    /* Number of 8-bit characters */
3568    }
3569    
3570    
3571    
3572    
3573    /*************************************************
3574    *        Add a list of characters to a class     *
3575    *************************************************/
3576    
3577    /* This function is used for adding a list of case-equivalent characters to a
3578    class, and also for adding a list of horizontal or vertical whitespace. If the
3579    list is in order (which it should be), ranges of characters are detected and
3580    handled appropriately. This function is mutually recursive with the function
3581    above.
3582    
3583    Arguments:
3584      classbits     the bit map for characters < 256
3585      uchardptr     points to the pointer for extra data
3586      options       the options word
3587      cd            contains pointers to tables etc.
3588      p             points to row of 32-bit values, terminated by NOTACHAR
3589      except        character to omit; this is used when adding lists of
3590                      case-equivalent characters to avoid including the one we
3591                      already know about
3592    
3593    Returns:        the number of < 256 characters added
3594                    the pointer to extra data is updated
3595    */
3596    
3597    static int
3598    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3599      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3600    {
3601    int n8 = 0;
3602    while (p[0] < NOTACHAR)
3603      {
3604      int n = 0;
3605      if (p[0] != except)
3606        {
3607        while(p[n+1] == p[0] + n + 1) n++;
3608        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3609        }
3610      p += n + 1;
3611      }
3612    return n8;
3613    }
3614    
3615    
3616    
3617    /*************************************************
3618    *    Add characters not in a list to a class     *
3619    *************************************************/
3620    
3621    /* This function is used for adding the complement of a list of horizontal or
3622    vertical whitespace to a class. The list must be in order.
3623    
3624    Arguments:
3625      classbits     the bit map for characters < 256
3626      uchardptr     points to the pointer for extra data
3627      options       the options word
3628      cd            contains pointers to tables etc.
3629      p             points to row of 32-bit values, terminated by NOTACHAR
3630    
3631    Returns:        the number of < 256 characters added
3632                    the pointer to extra data is updated
3633    */
3634    
3635    static int
3636    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3637      int options, compile_data *cd, const pcre_uint32 *p)
3638    {
3639    BOOL utf = (options & PCRE_UTF8) != 0;
3640    int n8 = 0;
3641    if (p[0] > 0)
3642      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3643    while (p[0] < NOTACHAR)
3644      {
3645      while (p[1] == p[0] + 1) p++;
3646      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3647        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3648      p++;
3649      }
3650    return n8;
3651    }
3652    
3653    
3654    
3655    /*************************************************
3656  *           Compile one branch                   *  *           Compile one branch                   *
3657  *************************************************/  *************************************************/
3658    
# Line 3402  Arguments: Line 3667  Arguments:
3667    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3668    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3669    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3670    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3671    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3672      reqcharptr     place to put the last required character
3673      reqcharflagsptr place to put the last required character flags, or a negative number
3674    bcptr          points to current branch chain    bcptr          points to current branch chain
3675    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3676    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3416  Returns:         TRUE on success Line 3683  Returns:         TRUE on success
3683    
3684  static BOOL  static BOOL
3685  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3686    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3687    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3688      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3689      branch_chain *bcptr, int cond_depth,
3690    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3691  {  {
3692  int repeat_type, op_type;  int repeat_type, op_type;
3693  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3694  int bravalue = 0;  int bravalue = 0;
3695  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3696  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
3697  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
3698    pcre_uint32 zeroreqchar, zerofirstchar;
3699    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3700  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3701  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3702  int after_manual_callout = 0;  int after_manual_callout = 0;
3703  int length_prevgroup = 0;  int length_prevgroup = 0;
3704  register int c;  register pcre_uint32 c;
3705    int escape;
3706  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3707  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3708  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3450  must not do this for other options (e.g. Line 3722  must not do this for other options (e.g.
3722  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3723    
3724  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3725  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3726  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3727    #ifndef COMPILE_PCRE32
3728  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3729    #endif
3730  #else  #else
3731  BOOL utf = FALSE;  BOOL utf = FALSE;
3732  #endif  #endif
3733    
3734  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3735    class_uchardata always so that it can be passed to add_to_class() always,
3736    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3737    alternative calls for the different cases. */
3738    
3739    pcre_uchar *class_uchardata;
3740  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3741  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3742  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3743  #endif  #endif
3744    
# Line 3484  to take the zero repeat into account. Th Line 3761  to take the zero repeat into account. Th
3761  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3762  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3763    
3764  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3765    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3766    
3767  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
3768  or zero, according to the current setting of the caseless flag. The  or zero, according to the current setting of the caseless flag. The
# Line 3505  for (;; ptr++) Line 3783  for (;; ptr++)
3783    BOOL is_recurse;    BOOL is_recurse;
3784    BOOL reset_bracount;    BOOL reset_bracount;
3785    int class_has_8bitchar;    int class_has_8bitchar;
3786    int class_single_char;    int class_one_char;
3787    int newoptions;    int newoptions;
3788    int recno;    int recno;
3789    int refsign;    int refsign;
3790    int skipbytes;    int skipbytes;
3791    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
3792    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
3793    int terminator;    int terminator;
3794    int mclength;    int mclength;
3795    int tempbracount;    int tempbracount;
3796      pcre_uint32 ec;
3797    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3798    
3799    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3677  for (;; ptr++) Line 3956  for (;; ptr++)
3956      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3957      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3958      *firstcharptr = firstchar;      *firstcharptr = firstchar;
3959        *firstcharflagsptr = firstcharflags;
3960      *reqcharptr = reqchar;      *reqcharptr = reqchar;
3961        *reqcharflagsptr = reqcharflags;
3962      *codeptr = code;      *codeptr = code;
3963      *ptrptr = ptr;      *ptrptr = ptr;
3964      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3701  for (;; ptr++) Line 3982  for (;; ptr++)
3982      previous = NULL;      previous = NULL;
3983      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3984        {        {
3985        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3986        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3987        }        }
3988      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3716  for (;; ptr++) Line 3997  for (;; ptr++)
3997      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
3998    
3999      case CHAR_DOT:      case CHAR_DOT:
4000      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4001      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4002        zerofirstcharflags = firstcharflags;
4003      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4004        zeroreqcharflags = reqcharflags;
4005      previous = code;      previous = code;
4006      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4007      break;      break;
# Line 3792  for (;; ptr++) Line 4075  for (;; ptr++)
4075          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4076        {        {
4077        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4078        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4079        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4080          zerofirstcharflags = firstcharflags;
4081        break;        break;
4082        }        }
4083    
# Line 3803  for (;; ptr++) Line 4087  for (;; ptr++)
4087    
4088      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4089    
4090      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4091      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 <
4092      < 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
4093      a single character. */      character. */
4094    
4095      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4096      class_single_char = 0;      class_one_char = 0;
4097    
4098      /* 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
4099      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
4100      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
4101      */      map. */
4102    
4103      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4104    
4105  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4106      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4107      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4108      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4109  #endif  #endif
4110    
4111      /* 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 3843  for (;; ptr++) Line 4127  for (;; ptr++)
4127        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4128        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
4129        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4130        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4131          however. */
4132    
4133        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4134          {          {
4135            xclass = TRUE;
4136          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4137          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4138          }          }
# Line 3905  for (;; ptr++) Line 4191  for (;; ptr++)
4191          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4192          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4193    
4194          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4195            posix_class = 0;            posix_class = 0;
4196    
4197          /* 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 4234  for (;; ptr++)
4234              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4235            }            }
4236    
4237          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4238          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4239    
4240          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3964  for (;; ptr++) Line 4250  for (;; ptr++)
4250            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4251    
4252          ptr = tempptr + 1;          ptr = tempptr + 1;
4253          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4254          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4255          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4256          class_single_char = 2;          class_one_char = 2;
4257          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4258          }          }
4259    
# Line 3975  for (;; ptr++) Line 4261  for (;; ptr++)
4261        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
4262        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
4263        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4264        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4265        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4266        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4267        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4268    
4269        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4270          {          {
4271          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4272    
4273          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4274    
4275          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4276          else if (-c == ESC_N)            /* \N is not supported in a class */            c = ec;
4277            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4278            else if (escape == ESC_N)            /* \N is not supported in a class */
4279            {            {
4280            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4281            goto FAILED;            goto FAILED;
4282            }            }
4283          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4284            {            {
4285            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4286              {              {
# Line 4000  for (;; ptr++) Line 4289  for (;; ptr++)
4289            else inescq = TRUE;            else inescq = TRUE;
4290            continue;            continue;
4291            }            }
4292          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4293    
4294          if (c < 0)          else
4295            {            {
4296            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4297            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4298            class_has_8bitchar++;            class_has_8bitchar++;
4299            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4300            class_single_char += 2;            class_one_char += 2;
4301    
4302            switch (-c)            switch (escape)
4303              {              {
4304  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4305              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 4309  for (;; ptr++)
4309              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4310              case ESC_SU:              case ESC_SU:
4311              nestptr = ptr;              nestptr = ptr;
4312              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4313              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4314              continue;              continue;
4315  #endif  #endif
# Line 4044  for (;; ptr++) Line 4333  for (;; ptr++)
4333    
4334              /* 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
4335              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4336              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4337                EBCDIC, so we lazily just adjust the appropriate bit. */
4338    
4339              case ESC_s:              case ESC_s:
4340              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4057  for (;; ptr++) Line 4347  for (;; ptr++)
4347              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4348              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4349              continue;              continue;
4350    
4351                /* The rest apply in both UCP and non-UCP cases. */
4352    
4353              case ESC_h:              case ESC_h:
4354              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4355              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  
4356              continue;              continue;
4357    
4358              case ESC_H:              case ESC_H:
4359              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4360                {                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  
4361              continue;              continue;
4362    
4363              case ESC_v:              case ESC_v:
4364              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4365              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  
4366              continue;              continue;
4367    
4368              case ESC_V:              case ESC_V:
4369              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4370                {                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  
4371              continue;              continue;
4372    
4373  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4242  for (;; ptr++) Line 4378  for (;; ptr++)
4378                int pdata;                int pdata;
4379                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4380                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4381                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4382                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4383                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4384                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4262  for (;; ptr++) Line 4397  for (;; ptr++)
4397                goto FAILED;                goto FAILED;
4398                }                }
4399              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4400              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4401              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4402              break;              break;
4403              }              }
4404            }            }
4405    
4406          /* 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).
4407          greater than 256. */          This may be greater than 256. */
4408    
4409            escape = 0;
4410    
4411          }   /* End of backslash handling */          }   /* End of backslash handling */
4412    
4413        /* 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
4414        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
4415        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4416        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4417    
4418        CHECK_RANGE:        CHECK_RANGE:
4419        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4284  for (;; ptr++) Line 4421  for (;; ptr++)
4421          inescq = FALSE;          inescq = FALSE;
4422          ptr += 2;          ptr += 2;
4423          }          }
   
4424        oldptr = ptr;        oldptr = ptr;
4425    
4426        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4427    
4428        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4429    
# Line 4295  for (;; ptr++) Line 4431  for (;; ptr++)
4431    
4432        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4433          {          {
4434          int d;          pcre_uint32 d;
4435          ptr += 2;          ptr += 2;
4436          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4437    
# Line 4310  for (;; ptr++) Line 4446  for (;; ptr++)
4446            inescq = TRUE;            inescq = TRUE;
4447            break;            break;
4448            }            }
4449    
4450            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4451            back the pointer and jump to handle the character that preceded it. */
4452    
4453          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4454            {            {
4455            ptr = oldptr;            ptr = oldptr;
4456            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4457            }            }
4458    
4459            /* Otherwise, we have a potential range; pick up the next character */
4460    
4461  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4462          if (utf)          if (utf)
# Line 4332  for (;; ptr++) Line 4473  for (;; ptr++)
4473    
4474          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4475            {            {
4476            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4477              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4478            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4479    
4480            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4481    
4482            if (d < 0)            if (descape != 0)
4483              {              {
4484              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4485                {                {
4486                ptr = oldptr;                ptr = oldptr;
4487                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4488                }                }
4489              }              }
4490            }            }
4491    
4492          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4493          one-character ranges */          one-character ranges. */
4494    
4495          if (d < c)          if (d < c)
4496            {            {
4497            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4498            goto FAILED;            goto FAILED;
4499            }            }
4500            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4501    
4502          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4503            cannot be done anymore. Any value greater than 1 indicates that there
4504            is more than one character. */
4505    
4506            class_one_char = 2;
4507    
4508          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4509    
4510          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4511    
4512          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4513          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4514          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));  
             }  
           }  
   
4515          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4516          }          }
4517    
4518        /* 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
4519        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
4520        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4521          increase it if it is already 2 or more ... just in case there's a class
4522        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4523    
4524        /* Only the value of 1 matters for class_single_char. */        CLASS_SINGLE_CHARACTER:
4525          if (class_one_char < 2) class_one_char++;
4526        if (class_single_char < 2) class_single_char++;  
4527          /* If class_one_char is 1, we have the first single character in the
4528          class, and there have been no prior ranges, or XCLASS items generated by
4529          escapes. If this is the final character in the class, we can optimize by
4530          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4531          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4532          to be set. Otherwise, there can be no first char if this item is first,
4533          whatever repeat count may follow. In the case of reqchar, save the
4534          previous value for reinstating. */
4535    
4536        /* If class_charcount is 1, we saw precisely one character. As long as        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       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)  
4537          {          {
4538          ptr++;          ptr++;
4539          zeroreqchar = reqchar;          zeroreqchar = reqchar;
4540            zeroreqcharflags = reqcharflags;
4541    
4542          if (negate_class)          if (negate_class)
4543            {            {
4544            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;  #ifdef SUPPORT_UCP
4545              int d;
4546    #endif
4547              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4548            zerofirstchar = firstchar;            zerofirstchar = firstchar;
4549            *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;            zerofirstcharflags = firstcharflags;
4550  #ifdef SUPPORT_UTF  
4551            if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)            /* For caseless UTF-8 mode when UCP support is available, check
4552              code += PRIV(ord2utf)(c, code);            whether this character has more than one other case. If so, generate
4553            else            a special OP_NOTPROP item instead of OP_NOTI. */
4554    
4555    #ifdef SUPPORT_UCP
4556              if (utf && (options & PCRE_CASELESS) != 0 &&
4557                  (d = UCD_CASESET(c)) != 0)
4558                {
4559                *code++ = OP_NOTPROP;
4560                *code++ = PT_CLIST;
4561                *code++ = d;
4562                }
4563              else
4564  #endif  #endif
4565              *code++ = c;            /* Char has only one other case, or UCP not available */
4566            goto NOT_CHAR;  
4567                {
4568                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4569    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4570                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4571                  code += PRIV(ord2utf)(c, code);
4572                else
4573    #endif
4574                  *code++ = c;
4575                }
4576    
4577              /* We are finished with this character class */
4578    
4579              goto END_CLASS;
4580            }            }
4581    
4582          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4583          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4584    
4585  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4586          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4587            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4588          else          else
# Line 4564  for (;; ptr++) Line 4593  for (;; ptr++)
4593            }            }
4594          goto ONE_CHAR;          goto ONE_CHAR;
4595          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4596    
4597        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4598          has been generated. Add this character to the class. */
4599  #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
4600        if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))        class_has_8bitchar +=
4601  #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));  
           }  
         }  
4602        }        }
4603    
4604      /* 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 4618  for (;; ptr++)
4618        goto FAILED;        goto FAILED;
4619        }        }
4620    
4621        /* We will need an XCLASS if data has been placed in class_uchardata. In
4622        the second phase this is a sufficient test. However, in the pre-compile
4623        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4624        only if the very last character in the class needs XCLASS will it contain
4625        anything at this point. For this reason, xclass gets set TRUE above when
4626        uchar_classdata is emptied, and that's why this code is the way it is here
4627        instead of just doing a test on class_uchardata below. */
4628    
4629    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4630        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4631    #endif
4632    
4633      /* 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
4634      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4635      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
4636    
4637      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4638      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4639        zerofirstcharflags = firstcharflags;
4640      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4641        zeroreqcharflags = reqcharflags;
4642    
4643      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4644      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
# Line 4703  for (;; ptr++) Line 4694  for (;; ptr++)
4694        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4695        }        }
4696      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4697      NOT_CHAR:  
4698        END_CLASS:
4699      break;      break;
4700    
4701    
# Line 4741  for (;; ptr++) Line 4733  for (;; ptr++)
4733      if (repeat_min == 0)      if (repeat_min == 0)
4734        {        {
4735        firstchar = zerofirstchar;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4736          firstcharflags = zerofirstcharflags;
4737        reqchar = zeroreqchar;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4738          reqcharflags = zeroreqcharflags;
4739        }        }
4740    
4741      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4813  for (;; ptr++) Line 4807  for (;; ptr++)
4807      if (*previous == OP_CHAR || *previous == OP_CHARI      if (*previous == OP_CHAR || *previous == OP_CHARI
4808          || *previous == OP_NOT || *previous == OP_NOTI)          || *previous == OP_NOT || *previous == OP_NOTI)
4809        {        {
4810        switch (*previous)        switch (*previous)
4811          {          {
4812          default: /* Make compiler happy. */          default: /* Make compiler happy. */
4813          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 4821  for (;; ptr++)
4821        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
4822        it's a length rather than a small character. */        it's a length rather than a small character. */
4823    
4824  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4825        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4826          {          {
4827          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4844  for (;; ptr++) Line 4838  for (;; ptr++)
4838          {          {
4839          c = code[-1];          c = code[-1];
4840          if (*previous <= OP_CHARI && repeat_min > 1)          if (*previous <= OP_CHARI && repeat_min > 1)
4841            reqchar = c | req_caseopt | cd->req_varyopt;            {
4842              reqchar = c;
4843              reqcharflags = req_caseopt | cd->req_varyopt;
4844              }
4845          }          }
4846    
4847        /* If the repetition is unlimited, it pays to see if the next thing on        /* If the repetition is unlimited, it pays to see if the next thing on
# Line 4963  for (;; ptr++) Line 4960  for (;; ptr++)
4960    
4961          if (repeat_max < 0)          if (repeat_max < 0)
4962            {            {
4963  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4964            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4965              {              {
4966              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4988  for (;; ptr++) Line 4985  for (;; ptr++)
4985    
4986          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4987            {            {
4988  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4989            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4990              {              {
4991              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5018  for (;; ptr++) Line 5015  for (;; ptr++)
5015    
5016        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
5017    
5018  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5019        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
5020          {          {
5021          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5226  for (;; ptr++) Line 5223  for (;; ptr++)
5223    
5224            else            else
5225              {              {
5226              if (groupsetfirstchar && reqchar < 0) reqchar = firstchar;              if (groupsetfirstchar && reqcharflags < 0)
5227                  {
5228                  reqchar = firstchar;
5229                  reqcharflags = firstcharflags;
5230                  }
5231    
5232              for (i = 1; i < repeat_min; i++)              for (i = 1; i < repeat_min; i++)
5233                {                {
# Line 5593  for (;; ptr++) Line 5594  for (;; ptr++)
5594        ptr++;        ptr++;
5595        while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;        while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;
5596        namelen = (int)(ptr - name);        namelen = (int)(ptr - name);
5597    
5598        /* It appears that Perl allows any characters whatsoever, other than        /* It appears that Perl allows any characters whatsoever, other than
5599        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
5600        letters, digits, and underscores. */        letters, digits, and underscores. */
# Line 5603  for (;; ptr++) Line 5604  for (;; ptr++)
5604          arg = ++ptr;          arg = ++ptr;
5605          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5606          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5607          if (arglen > (int)MAX_MARK)          if ((unsigned int)arglen > MAX_MARK)
5608            {            {
5609            *errorcodeptr = ERR75;            *errorcodeptr = ERR75;
5610            goto FAILED;            goto FAILED;
5611            }            }
5612          }          }
5613    
5614        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5623  for (;; ptr++) Line 5624  for (;; ptr++)
5624          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5625              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5626            {            {
5627              int setverb;
5628    
5629            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5630            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5631    
# Line 5640  for (;; ptr++) Line 5643  for (;; ptr++)
5643                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5644                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5645                }                }
5646              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5647                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5648    
5649              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5650              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
5651              }              }
5652    
5653            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5655  for (;; ptr++) Line 5659  for (;; ptr++)
5659                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5660                goto FAILED;                goto FAILED;
5661                }                }
5662              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5663              }              }
5664    
5665            else            else
# Line 5666  for (;; ptr++) Line 5669  for (;; ptr++)
5669                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5670                goto FAILED;                goto FAILED;
5671                }                }
5672              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5673              *code++ = arglen;              *code++ = arglen;
5674              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5675              code += arglen;              code += arglen;
5676              *code++ = 0;              *code++ = 0;
5677              }              }
5678    
5679              switch (setverb)
5680                {
5681                case OP_THEN:
5682                case OP_THEN_ARG:
5683                cd->external_flags |= PCRE_HASTHEN;
5684                break;
5685    
5686                case OP_PRUNE:
5687                case OP_PRUNE_ARG:
5688                case OP_SKIP:
5689                case OP_SKIP_ARG:
5690                cd->had_pruneorskip = TRUE;
5691                break;
5692                }
5693    
5694            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5695            }            }
5696    
# Line 5796  for (;; ptr++) Line 5813  for (;; ptr++)
5813          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
5814            {            {
5815            if (recno >= 0)            if (recno >= 0)
5816              recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1;              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;
5817            ptr++;            ptr++;
5818            }            }
5819          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
# Line 6393  for (;; ptr++) Line 6410  for (;; ptr++)
6410    
6411          /* Can't determine a first byte now */          /* Can't determine a first byte now */
6412    
6413          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6414          continue;          continue;
6415    
6416    
# Line 6527  for (;; ptr++) Line 6544  for (;; ptr++)
6544           cond_depth +           cond_depth +
6545             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
6546           &subfirstchar,                   /* For possible first char */           &subfirstchar,                   /* For possible first char */
6547             &subfirstcharflags,
6548           &subreqchar,                     /* For possible last char */           &subreqchar,                     /* For possible last char */
6549             &subreqcharflags,
6550           bcptr,                           /* Current branch chain */           bcptr,                           /* Current branch chain */
6551           cd,                              /* Tables block */           cd,                              /* Tables block */
6552           (lengthptr == NULL)? NULL :      /* Actual compile phase */           (lengthptr == NULL)? NULL :      /* Actual compile phase */
# Line 6588  for (;; ptr++) Line 6607  for (;; ptr++)
6607            *errorcodeptr = ERR27;            *errorcodeptr = ERR27;
6608            goto FAILED;            goto FAILED;
6609            }            }
6610          if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;          if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE;
6611          }          }
6612        }        }
6613    
# Line 6637  for (;; ptr++) Line 6656  for (;; ptr++)
6656      back off. */      back off. */
6657    
6658      zeroreqchar = reqchar;      zeroreqchar = reqchar;
6659        zeroreqcharflags = reqcharflags;
6660      zerofirstchar = firstchar;      zerofirstchar = firstchar;
6661        zerofirstcharflags = firstcharflags;
6662      groupsetfirstchar = FALSE;      groupsetfirstchar = FALSE;
6663    
6664      if (bravalue >= OP_ONCE)      if (bravalue >= OP_ONCE)
# Line 6648  for (;; ptr++) Line 6669  for (;; ptr++)
6669        no firstchar, set "none" for the whole branch. In both cases, a zero        no firstchar, set "none" for the whole branch. In both cases, a zero
6670        repeat forces firstchar to "none". */        repeat forces firstchar to "none". */
6671    
6672        if (firstchar == REQ_UNSET)        if (firstcharflags == REQ_UNSET)
6673          {          {
6674          if (subfirstchar >= 0)          if (subfirstcharflags >= 0)
6675            {            {
6676            firstchar = subfirstchar;            firstchar = subfirstchar;
6677              firstcharflags = subfirstcharflags;
6678            groupsetfirstchar = TRUE;            groupsetfirstchar = TRUE;
6679            }            }
6680          else firstchar = REQ_NONE;          else firstcharflags = REQ_NONE;
6681          zerofirstchar = REQ_NONE;          zerofirstcharflags = REQ_NONE;
6682          }          }
6683    
6684        /* If firstchar was previously set, convert the subpattern's firstchar        /* If firstchar was previously set, convert the subpattern's firstchar
6685        into reqchar if there wasn't one, using the vary flag that was in        into reqchar if there wasn't one, using the vary flag that was in
6686        existence beforehand. */        existence beforehand. */
6687    
6688        else if (subfirstchar >= 0 && subreqchar < 0)        else if (subfirstcharflags >= 0 && subreqcharflags < 0)
6689          subreqchar = subfirstchar | tempreqvary;          {
6690            subreqchar = subfirstchar;
6691            subreqcharflags = subfirstcharflags | tempreqvary;
6692            }
6693    
6694        /* If the subpattern set a required byte (or set a first byte that isn't        /* If the subpattern set a required byte (or set a first byte that isn't
6695        really the first byte - see above), set it. */        really the first byte - see above), set it. */
6696    
6697        if (subreqchar >= 0) reqchar = subreqchar;        if (subreqcharflags >= 0)
6698            {
6699            reqchar = subreqchar;
6700            reqcharflags = subreqcharflags;
6701            }
6702        }        }
6703    
6704      /* For a forward assertion, we take the reqchar, if set. This can be      /* For a forward assertion, we take the reqchar, if set. This can be
# Line 6680  for (;; ptr++) Line 6709  for (;; ptr++)
6709      of a firstchar. This is overcome by a scan at the end if there's no      of a firstchar. This is overcome by a scan at the end if there's no
6710      firstchar, looking for an asserted first char. */      firstchar, looking for an asserted first char. */
6711    
6712      else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;      else if (bravalue == OP_ASSERT && subreqcharflags >= 0)
6713          {
6714          reqchar = subreqchar;
6715          reqcharflags = subreqcharflags;
6716          }
6717      break;     /* End of processing '(' */      break;     /* End of processing '(' */
6718    
6719    
# Line 6688  for (;; ptr++) Line 6721  for (;; ptr++)
6721      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
6722      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
6723      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
6724      are ESC_REF plus the reference number. Only back references and those types      are negative the reference number. Only back references and those types
6725      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
6726      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
6727      ever created. */      ever created. */
6728    
6729      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6730      tempptr = ptr;      tempptr = ptr;
6731      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6732    
6733      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6734    
6735      if (c < 0)      if (escape == 0)
6736          c = ec;
6737        else
6738        {        {
6739        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6740          {          {
6741          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6742            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6708  for (;; ptr++) Line 6744  for (;; ptr++)
6744          continue;          continue;
6745          }          }
6746    
6747        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6748    
6749        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6750        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6751    
6752        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6753          firstchar = REQ_NONE;          firstcharflags = REQ_NONE;
6754    
6755        /* 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. */
6756    
6757        zerofirstchar = firstchar;        zerofirstchar = firstchar;
6758          zerofirstcharflags = firstcharflags;
6759        zeroreqchar = reqchar;        zeroreqchar = reqchar;
6760          zeroreqcharflags = reqcharflags;
6761    
6762        /* \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'
6763        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6764        -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 <
6765        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
6766        -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
6767        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6768    
6769        if (-c == ESC_g)        if (escape == ESC_g)
6770          {          {
6771          const pcre_uchar *p;          const pcre_uchar *p;
6772          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 6822  for (;; ptr++)
6822        /* \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).
6823        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6824    
6825        if (-c == ESC_k)        if (escape == ESC_k)
6826          {          {
6827          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6828            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 6841  for (;; ptr++)
6841        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
6842        ':' later. */        ':' later. */
6843    
6844        if (-c >= ESC_REF)        if (escape < 0)
6845          {          {
6846          open_capitem *oc;          open_capitem *oc;
6847          recno = -c - ESC_REF;          recno = -escape;
6848    
6849          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6850          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6851          previous = code;          previous = code;
6852          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6853          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
# Line 6833  for (;; ptr++) Line 6871  for (;; ptr++)
6871        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6872    
6873  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6874        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6875          {          {
6876          BOOL negated;          BOOL negated;
6877          int pdata;          int pdata;
6878          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
6879          if (ptype < 0) goto FAILED;          if (ptype < 0) goto FAILED;
6880          previous = code;          previous = code;
6881          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6882          *code++ = ptype;          *code++ = ptype;
6883          *code++ = pdata;          *code++ = pdata;
6884          }          }
# Line 6849  for (;; ptr++) Line 6887  for (;; ptr++)
6887        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6888        allowed. */        allowed. */
6889    
6890        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6891          {          {
6892          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6893          goto FAILED;          goto FAILED;
# Line 6859  for (;; ptr++) Line 6897  for (;; ptr++)
6897        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6898        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
6899        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
6900        Unicode property tests. Note that \b and \B do a one-character        Unicode property tests. Note that \b and \B do a one-character
6901        lookbehind. */        lookbehind. */
6902    
6903        else        else
6904          {          {
6905          if ((-c == ESC_b || -c == ESC_B) && cd->max_lookbehind == 0)          if ((escape == ESC_b || escape == ESC_B) && cd->max_lookbehind == 0)
6906            cd->max_lookbehind = 1;            cd->max_lookbehind = 1;
6907  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6908          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6909            {            {
6910            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6911            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6912            }            }
6913          else          else
6914  #endif  #endif
# Line 6878  for (;; ptr++) Line 6916  for (;; ptr++)
6916          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6917    
6918            {            {
6919            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
6920            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
6921            }            }
6922          }          }
6923        continue;        continue;
# Line 6889  for (;; ptr++) Line 6927  for (;; ptr++)
6927      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
6928      handle it as a data character. */      handle it as a data character. */
6929    
6930  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6931      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6932        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6933      else      else
# Line 6922  for (;; ptr++) Line 6960  for (;; ptr++)
6960    
6961      ONE_CHAR:      ONE_CHAR:
6962      previous = code;      previous = code;
6963    
6964        /* For caseless UTF-8 mode when UCP support is available, check whether
6965        this character has more than one other case. If so, generate a special
6966        OP_PROP item instead of OP_CHARI. */
6967    
6968    #ifdef SUPPORT_UCP
6969        if (utf && (options & PCRE_CASELESS) != 0)
6970          {
6971          GETCHAR(c, mcbuffer);
6972          if ((c = UCD_CASESET(c)) != 0)
6973            {
6974            *code++ = OP_PROP;
6975            *code++ = PT_CLIST;
6976            *code++ = c;
6977            if (firstcharflags == REQ_UNSET) firstcharflags = zerofirstcharflags = REQ_NONE;
6978            break;
6979            }
6980          }
6981    #endif
6982    
6983        /* Caseful matches, or not one of the multicase characters. */
6984    
6985      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6986      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6987    
# Line 6935  for (;; ptr++) Line 6995  for (;; ptr++)
6995      Otherwise, leave the firstchar value alone, and don't change it on a zero      Otherwise, leave the firstchar value alone, and don't change it on a zero
6996      repeat. */      repeat. */
6997    
6998      if (firstchar == REQ_UNSET)      if (firstcharflags == REQ_UNSET)
6999        {        {
7000        zerofirstchar = REQ_NONE;        zerofirstcharflags = REQ_NONE;
7001        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7002          zeroreqcharflags = reqcharflags;
7003    
7004        /* If the character is more than one byte long, we can set firstchar        /* If the character is more than one byte long, we can set firstchar
7005        only if it is not to be matched caselessly. */        only if it is not to be matched caselessly. */
# Line 6946  for (;; ptr++) Line 7007  for (;; ptr++)
7007        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7008          {          {
7009          firstchar = mcbuffer[0] | req_caseopt;          firstchar = mcbuffer[0] | req_caseopt;
7010          if (mclength != 1) reqchar = code[-1] | cd->req_varyopt;          firstchar = mcbuffer[0];
7011            firstcharflags = req_caseopt;
7012    
7013            if (mclength != 1)
7014              {
7015              reqchar = code[-1];
7016              reqcharflags = cd->req_varyopt;
7017              }
7018          }          }
7019        else firstchar = reqchar = REQ_NONE;        else firstcharflags = reqcharflags = REQ_NONE;
7020        }        }
7021    
7022      /* firstchar was previously set; we can set reqchar only if the length is      /* firstchar was previously set; we can set reqchar only if the length is
# Line 6957  for (;; ptr++) Line 7025  for (;; ptr++)
7025      else      else
7026        {        {
7027        zerofirstchar = firstchar;        zerofirstchar = firstchar;
7028          zerofirstcharflags = firstcharflags;
7029        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7030          zeroreqcharflags = reqcharflags;
7031        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7032          reqchar = code[-1] | req_caseopt | cd->req_varyopt;          {
7033            reqchar = code[-1];
7034            reqcharflags = req_caseopt | cd->req_varyopt;
7035            }
7036        }        }
7037    
7038      break;            /* End of literal character handling */      break;            /* End of literal character handling */
# Line 6978  return FALSE; Line 7051  return FALSE;
7051    
7052    
7053    
   
7054  /*************************************************  /*************************************************
7055  *     Compile sequence of alternatives           *  *     Compile sequence of alternatives           *
7056  *************************************************/  *************************************************/
# Line 6999  Arguments: Line 7071  Arguments:
7071    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
7072    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
7073    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
7074    firstcharptr   place to put the first required character, or a negative number    firstcharptr    place to put the first required character
7075    reqcharptr     place to put the last required character, or a negative number    firstcharflagsptr place to put the first character flags, or a negative number
7076      reqcharptr     place to put the last required character
7077      reqcharflagsptr place to put the last required character flags, or a negative number
7078    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
7079    cd             points to the data block with tables pointers etc.    cd             points to the data block with tables pointers etc.
7080    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
# Line 7012  Returns:         TRUE on success Line 7086  Returns:         TRUE on success
7086  static BOOL  static BOOL
7087  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,
7088    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
7089    int cond_depth, pcre_int32 *firstcharptr, pcre_int32 *reqcharptr,    int cond_depth,
7090      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
7091      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
7092    branch_chain *bcptr, compile_data *cd, int *lengthptr)    branch_chain *bcptr, compile_data *cd, int *lengthptr)
7093  {  {
7094  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
# Line 7022  pcre_uchar *start_bracket = code; Line 7098  pcre_uchar *start_bracket = code;
7098  pcre_uchar *reverse_count = NULL;  pcre_uchar *reverse_count = NULL;
7099  open_capitem capitem;  open_capitem capitem;
7100  int capnumber = 0;  int capnumber = 0;
7101  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7102  pcre_int32 branchfirstchar, branchreqchar;  pcre_int32 firstcharflags, reqcharflags;
7103    pcre_uint32 branchfirstchar, branchreqchar;
7104    pcre_int32 branchfirstcharflags, branchreqcharflags;
7105  int length;  int length;
7106  int orig_bracount;  int orig_bracount;
7107  int max_bracount;  int max_bracount;
# Line 7032  branch_chain bc; Line 7110  branch_chain bc;
7110  bc.outer = bcptr;  bc.outer = bcptr;
7111  bc.current_branch = code;  bc.current_branch = code;
7112    
7113  firstchar = reqchar = REQ_UNSET;  firstchar = reqchar = 0;
7114    firstcharflags = reqcharflags = REQ_UNSET;
7115    
7116  /* Accumulate the length for use in the pre-compile phase. Start with the  /* Accumulate the length for use in the pre-compile phase. Start with the
7117  length of the BRA and KET and any extra bytes that are required at the  length of the BRA and KET and any extra bytes that are required at the
# Line 7092  for (;;) Line 7171  for (;;)
7171    into the length. */    into the length. */
7172    
7173    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,
7174          &branchreqchar, &bc, cond_depth, cd,          &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc,
7175          (lengthptr == NULL)? NULL : &length))          cond_depth, cd, (lengthptr == NULL)? NULL : &length))
7176      {      {
7177      *ptrptr = ptr;      *ptrptr = ptr;
7178      return FALSE;      return FALSE;
# Line 7114  for (;;) Line 7193  for (;;)
7193      if (*last_branch != OP_ALT)      if (*last_branch != OP_ALT)
7194        {        {
7195        firstchar = branchfirstchar;        firstchar = branchfirstchar;
7196          firstcharflags = branchfirstcharflags;
7197        reqchar = branchreqchar;        reqchar = branchreqchar;
7198          reqcharflags = branchreqcharflags;
7199        }        }
7200    
7201      /* If this is not the first branch, the first char and reqchar have to      /* If this is not the first branch, the first char and reqchar have to
# Line 7128  for (;;) Line 7209  for (;;)
7209        we have to abandon the firstchar for the regex, but if there was        we have to abandon the firstchar for the regex, but if there was
7210        previously no reqchar, it takes on the value of the old firstchar. */        previously no reqchar, it takes on the value of the old firstchar. */
7211    
7212        if (firstchar >= 0 && firstchar != branchfirstchar)        if (firstcharflags >= 0 &&
7213              (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar))
7214          {          {
7215          if (reqchar < 0) reqchar = firstchar;          if (reqcharflags < 0)
7216          firstchar = REQ_NONE;            {
7217              reqchar = firstchar;
7218              reqcharflags = firstcharflags;
7219              }
7220            firstcharflags = REQ_NONE;
7221          }          }
7222    
7223        /* If we (now or from before) have no firstchar, a firstchar from the        /* If we (now or from before) have no firstchar, a firstchar from the
7224        branch becomes a reqchar if there isn't a branch reqchar. */        branch becomes a reqchar if there isn't a branch reqchar. */
7225    
7226        if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0)        if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0)
7227            branchreqchar = branchfirstchar;          {
7228            branchreqchar = branchfirstchar;
7229            branchreqcharflags = branchfirstcharflags;
7230            }
7231    
7232        /* Now ensure that the reqchars match */        /* Now ensure that the reqchars match */
7233    
7234        if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY))        if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) ||
7235          reqchar = REQ_NONE;            reqchar != branchreqchar)
7236        else reqchar |= branchreqchar;   /* To "or" REQ_VARY */          reqcharflags = REQ_NONE;
7237          else
7238            {
7239            reqchar = branchreqchar;
7240            reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */
7241            }
7242        }        }
7243    
7244      /* If lookbehind, check that this branch matches a fixed-length string, and      /* If lookbehind, check that this branch matches a fixed-length string, and
# Line 7173  for (;;) Line 7267  for (;;)
7267          *ptrptr = ptr;          *ptrptr = ptr;
7268          return FALSE;          return FALSE;
7269          }          }
7270        else        else
7271          {          {
7272          if (fixed_length > cd->max_lookbehind)          if (fixed_length > cd->max_lookbehind)
7273            cd->max_lookbehind = fixed_length;            cd->max_lookbehind = fixed_length;
7274          PUT(reverse_count, 0, fixed_length);          PUT(reverse_count, 0, fixed_length);
7275          }          }
7276        }        }
7277      }      }
# Line 7240  for (;;) Line 7334  for (;;)
7334      *codeptr = code;      *codeptr = code;
7335      *ptrptr = ptr;      *ptrptr = ptr;
7336      *firstcharptr = firstchar;      *firstcharptr = firstchar;
7337        *firstcharflagsptr = firstcharflags;
7338      *reqcharptr = reqchar;      *reqcharptr = reqchar;
7339        *reqcharflagsptr = reqcharflags;
7340      if (lengthptr != NULL)      if (lengthptr != NULL)
7341        {        {
7342        if (OFLOW_MAX - *lengthptr < length)        if (OFLOW_MAX - *lengthptr < length)
# Line 7310  and the highest back reference was great Line 7406  and the highest back reference was great
7406  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
7407  of the more common cases more precisely.  of the more common cases more precisely.
7408    
7409    ... A second exception is when the .* appears inside an atomic group, because
7410    this prevents the number of characters it matches from being adjusted.
7411    
7412  Arguments:  Arguments:
7413    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7414    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
7415                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7416                    the less precise approach                    the less precise approach
7417    backref_map    the back reference bitmap    cd             points to the compile data block
7418      atomcount      atomic group level
7419    
7420  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7421  */  */
7422    
7423  static BOOL  static BOOL
7424  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7425    unsigned int backref_map)    compile_data *cd, int atomcount)
7426  {  {
7427  do {  do {
7428     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7334  do { Line 7434  do {
7434     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7435         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7436       {       {
7437       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7438       }       }
7439    
7440     /* Capturing brackets */     /* Capturing brackets */
# Line 7344  do { Line 7444  do {
7444       {       {
7445       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7446       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7447       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7448         }
7449    
7450       /* Positive forward assertions and conditions */
7451    
7452       else if (op == OP_ASSERT || op == OP_COND)
7453         {
7454         if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7455       }       }
7456    
7457     /* Other brackets */     /* Atomic groups */
7458    
7459     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ONCE || op == OP_ONCE_NC)
             op == OP_COND)  
7460       {       {
7461       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7462           return FALSE;
7463       }       }
7464    
7465     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7466     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
7467       group. */
7468    
7469     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7470               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7471       {       {
7472       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7473             atomcount > 0 || cd->had_pruneorskip)
7474         return FALSE;         return FALSE;
7475       }       }
7476    
7477     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7478    
7479     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;
7480    
7481     code += GET(code, 1);     code += GET(code, 1);
7482     }     }
7483  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7385  return TRUE; Line 7495  return TRUE;
7495  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
7496  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
7497  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7498  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7499    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7500    count, because once again the assumption no longer holds.
7501    
7502  Arguments:  Arguments:
7503    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7504    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
7505                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7506                    the less precise approach                    the less precise approach
7507    backref_map    the back reference bitmap    cd             points to the compile data
7508      atomcount      atomic group level
7509    
7510  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7511  */  */
7512    
7513  static BOOL  static BOOL
7514  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7515    unsigned int backref_map)    compile_data *cd, int atomcount)
7516  {  {
7517  do {  do {
7518     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7425  do { Line 7538  do {
7538         return FALSE;         return FALSE;
7539    
7540         default:     /* Assertion */         default:     /* Assertion */
7541         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7542         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7543         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7544         break;         break;
# Line 7439  do { Line 7552  do {
7552     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7553         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7554       {       {
7555       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7556       }       }
7557    
7558     /* Capturing brackets */     /* Capturing brackets */
# Line 7449  do { Line 7562  do {
7562       {       {
7563       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7564       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7565       if (!is_startline(scode, new_map, backref_map)) return FALSE;       if (!is_startline(scode, new_map, cd, atomcount)) return FALSE;
7566       }       }
7567    
7568     /* Other brackets */     /* Positive forward assertions */
7569    
7570       else if (op == OP_ASSERT)
7571         {
7572         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7573         }
7574    
7575       /* Atomic brackets */
7576    
7577     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)     else if (op == OP_ONCE || op == OP_ONCE_NC)
7578       {       {
7579       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE;
7580       }       }
7581    
7582     /* .* means "start at start or after \n" if it isn't in brackets that     /* .* means "start at start or after \n" if it isn't in atomic brackets or
7583     may be referenced. */     brackets that may be referenced, as long as the pattern does not contain
7584       *PRUNE or *SKIP, because these break the feature. Consider, for example,
7585       /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the
7586       start of a line. */
7587    
7588     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)
7589       {       {
7590       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 ||
7591             atomcount > 0 || cd->had_pruneorskip)
7592           return FALSE;
7593       }       }
7594    
7595     /* Check for explicit circumflex */     /* Check for explicit circumflex; anything else gives a FALSE result. Note
7596       in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC
7597       because the number of characters matched by .* cannot be adjusted inside
7598       them. */
7599    
7600     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;
7601    
# Line 7495  we return that char, otherwise -1. Line 7623  we return that char, otherwise -1.
7623    
7624  Arguments:  Arguments:
7625    code       points to start of expression (the bracket)    code       points to start of expression (the bracket)
7626      flags       points to the first char flags, or to REQ_NONE
7627    inassert   TRUE if in an assertion    inassert   TRUE if in an assertion
7628    
7629  Returns:     -1 or the fixed first char  Returns:     the fixed first char, or 0 with REQ_NONE in flags
7630  */  */
7631    
7632  static int  static pcre_uint32
7633  find_firstassertedchar(const pcre_uchar *code, BOOL inassert)  find_firstassertedchar(const pcre_uchar *code, pcre_int32 *flags,
7634      BOOL inassert)
7635  {  {
7636  register int c = -1;  register pcre_uint32 c = 0;
7637    int cflags = REQ_NONE;
7638    
7639    *flags = REQ_NONE;
7640  do {  do {
7641     int d;     pcre_uint32 d;
7642       int dflags;
7643     int xl = (*code == OP_CBRA || *code == OP_SCBRA ||     int xl = (*code == OP_CBRA || *code == OP_SCBRA ||
7644               *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0;               *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0;
7645     const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,     const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,
7646       TRUE);       TRUE);
7647     register int op = *scode;     register pcre_uchar op = *scode;
7648    
7649     switch(op)     switch(op)
7650       {       {
7651       default:       default:
7652       return -1;       return 0;
7653    
7654       case OP_BRA:       case OP_BRA:
7655       case OP_BRAPOS:       case OP_BRAPOS:
# Line 7527  do { Line 7661  do {
7661       case OP_ONCE:       case OP_ONCE:
7662       case OP_ONCE_NC:       case OP_ONCE_NC:
7663       case OP_COND:       case OP_COND:
7664       if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0)       d = find_firstassertedchar(scode, &dflags, op == OP_ASSERT);
7665         return -1;       if (dflags < 0)
7666       if (c < 0) c = d; else if (c != d) return -1;         return 0;
7667         if (cflags < 0) { c = d; cflags = dflags; } else if (c != d || cflags != dflags) return 0;
7668       break;       break;
7669    
7670       case OP_EXACT:       case OP_EXACT:
# Line 7540  do { Line 7675  do {
7675       case OP_PLUS:       case OP_PLUS:
7676       case OP_MINPLUS:       case OP_MINPLUS:
7677       case OP_POSPLUS:       case OP_POSPLUS:
7678       if (!inassert) return -1;       if (!inassert) return 0;
7679       if (c < 0) c = scode[1];       if (cflags < 0) { c = scode[1]; cflags = 0; }
7680         else if (c != scode[1]) return -1;         else if (c != scode[1]) return 0;
7681       break;       break;
7682    
7683       case OP_EXACTI:       case OP_EXACTI:
# Line 7553  do { Line 7688  do {
7688       case OP_PLUSI:       case OP_PLUSI:
7689       case OP_MINPLUSI:       case OP_MINPLUSI:
7690       case OP_POSPLUSI:       case OP_POSPLUSI:
7691       if (!inassert) return -1;       if (!inassert) return 0;
7692       if (c < 0) c = scode[1] | REQ_CASELESS;       if (cflags < 0) { c = scode[1]; cflags = REQ_CASELESS; }
7693         else if (c != scode[1]) return -1;         else if (c != scode[1]) return 0;
7694       break;       break;
7695       }       }
7696    
7697     code += GET(code, 1);     code += GET(code, 1);
7698     }     }
7699  while (*code == OP_ALT);  while (*code == OP_ALT);
7700    
7701    *flags = cflags;
7702  return c;  return c;
7703  }  }
7704    
# Line 7589  Returns:        pointer to compiled data Line 7726  Returns:        pointer to compiled data
7726                  with errorptr and erroroffset set                  with errorptr and erroroffset set
7727  */  */
7728    
7729  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7730  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7731  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
7732    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7733  #else  #elif defined COMPILE_PCRE16
7734  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7735  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,
7736    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7737    #elif defined COMPILE_PCRE32
7738    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7739    pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr,
7740      int *erroroffset, const unsigned char *tables)
7741  #endif  #endif
7742  {  {
7743  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7744  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7745  #else  #elif defined COMPILE_PCRE16
7746  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7747    #elif defined COMPILE_PCRE32
7748    return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7749  #endif  #endif
7750  }  }
7751    
7752    
7753  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7754  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7755  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
7756    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7757  #else  #elif defined COMPILE_PCRE16
7758  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7759  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,
7760    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7761    #elif defined COMPILE_PCRE32
7762    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7763    pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr,
7764      const char **errorptr, int *erroroffset, const unsigned char *tables)
7765  #endif  #endif
7766  {  {
7767  REAL_PCRE *re;  REAL_PCRE *re;
7768  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
7769  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7770    pcre_int32 firstcharflags, reqcharflags;
7771  int newline;  int newline;
7772  int errorcode = 0;  int errorcode = 0;
7773  int skipatstart = 0;  int skipatstart = 0;
# Line 7700  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7848  while (ptr[skipatstart] == CHAR_LEFT_PAR
7848    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7849      { skipatstart += 8; options |= PCRE_UTF16; continue; }      { skipatstart += 8; options |= PCRE_UTF16; continue; }
7850  #endif  #endif
7851    #ifdef COMPILE_PCRE32
7852      if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7853        { skipatstart += 8; options |= PCRE_UTF32; continue; }
7854    #endif
7855    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)
7856      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
7857    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 7880  while (ptr[skipatstart] == CHAR_LEFT_PAR
7880    else break;    else break;
7881    }    }
7882    
7883  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */
7884  utf = (options & PCRE_UTF8) != 0;  utf = (options & PCRE_UTF8) != 0;
7885    
7886  /* 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 7892  not used here. */
7892  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&
7893       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)
7894    {    {
7895  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7896    errorcode = ERR44;    errorcode = ERR44;
7897  #else  #elif defined COMPILE_PCRE16
7898    errorcode = ERR74;    errorcode = ERR74;
7899    #elif defined COMPILE_PCRE32
7900      errorcode = ERR77;
7901  #endif  #endif
7902    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;
7903    }    }
# Line 7863  ptr += skipatstart; Line 8017  ptr += skipatstart;
8017  code = cworkspace;  code = cworkspace;
8018  *code = OP_BRA;  *code = OP_BRA;
8019  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
8020    FALSE, 0, 0, &firstchar, &reqchar, NULL, cd, &length);    FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL,
8021      cd, &length);
8022  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
8023    
8024  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 7907  re->name_count = cd->names_found; Line 8062  re->name_count = cd->names_found;
8062  re->ref_count = 0;  re->ref_count = 0;
8063  re->tables = (tables == PRIV(default_tables))? NULL : tables;  re->tables = (tables == PRIV(default_tables))? NULL : tables;
8064  re->nullpad = NULL;  re->nullpad = NULL;
8065    #ifdef COMPILE_PCRE32
8066    re->dummy1 = re->dummy2 = 0;
8067    #endif
8068    
8069  /* 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
8070  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 8084  cd->start_code = codestart;
8084  cd->hwm = (pcre_uchar *)(cd->start_workspace);  cd->hwm = (pcre_uchar *)(cd->start_workspace);
8085  cd->req_varyopt = 0;  cd->req_varyopt = 0;
8086  cd->had_accept = FALSE;  cd->had_accept = FALSE;
8087    cd->had_pruneorskip = FALSE;
8088  cd->check_lookbehind = FALSE;  cd->check_lookbehind = FALSE;
8089  cd->open_caps = NULL;  cd->open_caps = NULL;
8090    
# Line 7937  ptr = (const pcre_uchar *)pattern + skip Line 8096  ptr = (const pcre_uchar *)pattern + skip
8096  code = (pcre_uchar *)codestart;  code = (pcre_uchar *)codestart;
8097  *code = OP_BRA;  *code = OP_BRA;
8098  (void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0,  (void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0,
8099    &firstchar, &reqchar, NULL, cd, NULL);    &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL, cd, NULL);
8100  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
8101  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
8102  re->max_lookbehind = cd->max_lookbehind;  re->max_lookbehind = cd->max_lookbehind;
8103  re->flags = cd->external_flags | PCRE_MODE;  re->flags = cd->external_flags | PCRE_MODE;
8104    
8105  if (cd->had_accept) reqchar = REQ_NONE;   /* Must disable after (*ACCEPT) */  if (cd->had_accept)
8106      {
8107      reqchar = 0;              /* Must disable after (*ACCEPT) */
8108      reqcharflags = REQ_NONE;
8109      }
8110    
8111  /* If not reached end of pattern on success, there's an excess bracket. */  /* If not reached end of pattern on success, there's an excess bracket. */
8112    
# Line 7993  if (errorcode == 0 && re->top_backref > Line 8156  if (errorcode == 0 && re->top_backref >
8156    
8157  /* If there were any lookbehind assertions that contained OP_RECURSE  /* If there were any lookbehind assertions that contained OP_RECURSE
8158  (recursions or subroutine calls), a flag is set for them to be checked here,  (recursions or subroutine calls), a flag is set for them to be checked here,
8159  because they may contain forward references. Actual recursions can't be fixed  because they may contain forward references. Actual recursions cannot be fixed
8160  length, but subroutine calls can. It is done like this so that those without  length, but subroutine calls can. It is done like this so that those without
8161  OP_RECURSE that are not fixed length get a diagnosic with a useful offset. The  OP_RECURSE that are not fixed length get a diagnosic with a useful offset. The
8162  exceptional ones forgo this. We scan the pattern to check that they are fixed  exceptional ones forgo this. We scan the pattern to check that they are fixed
# Line 8049  if (errorcode != 0) Line 8212  if (errorcode != 0)
8212    }    }
8213    
8214  /* 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
8215  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
8216  as starting with .* when DOTALL is set).  as starting with non-atomic .* when DOTALL is set and there are no occurrences
8217    of *PRUNE or *SKIP.
8218    
8219  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
8220  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
8221  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches
8222  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
8223  */  non-DOTALL matches when *PRUNE and SKIP are not present. */
8224    
8225  if ((re->options & PCRE_ANCHORED) == 0)  if ((re->options & PCRE_ANCHORED) == 0)
8226    {    {
8227    if (is_anchored(codestart, 0, cd->backref_map))    if (is_anchored(codestart, 0, cd, 0)) re->options |= PCRE_ANCHORED;
     re->options |= PCRE_ANCHORED;  
8228    else    else
8229      {      {
8230      if (firstchar < 0)      if (firstcharflags < 0)
8231        firstchar = find_firstassertedchar(codestart, FALSE);        firstchar = find_firstassertedchar(codestart, &firstcharflags, FALSE);
8232      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */      if (firstcharflags >= 0)   /* Remove caseless flag for non-caseable chars */
8233        {        {
8234  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8235        re->first_char = firstchar & 0xff;        re->first_char = firstchar & 0xff;
8236  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8237        re->first_char = firstchar & 0xffff;        re->first_char = firstchar & 0xffff;
8238    #elif defined COMPILE_PCRE32
8239          re->first_char = firstchar;
8240  #endif  #endif
8241  #endif        if ((firstcharflags & REQ_CASELESS) != 0)
       if ((firstchar & REQ_CASELESS) != 0)  
8242          {          {
8243  #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)  #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)
8244          /* We ignore non-ASCII first chars in 8 bit mode. */          /* We ignore non-ASCII first chars in 8 bit mode. */
# Line 8098  if ((re->options & PCRE_ANCHORED) == 0) Line 8261  if ((re->options & PCRE_ANCHORED) == 0)
8261    
8262        re->flags |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
8263        }        }
8264      else if (is_startline(codestart, 0, cd->backref_map))  
8265        re->flags |= PCRE_STARTLINE;      else if (is_startline(codestart, 0, cd, 0)) re->flags |= PCRE_STARTLINE;
8266      }      }
8267    }    }
8268    
# Line 8107  if ((re->options & PCRE_ANCHORED) == 0) Line 8270  if ((re->options & PCRE_ANCHORED) == 0)
8270  variable length item in the regex. Remove the caseless flag for non-caseable  variable length item in the regex. Remove the caseless flag for non-caseable
8271  bytes. */  bytes. */
8272    
8273  if (reqchar >= 0 &&  if (reqcharflags >= 0 &&
8274       ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0))       ((re->options & PCRE_ANCHORED) == 0 || (reqcharflags & REQ_VARY) != 0))
8275    {    {
8276  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8277    re->req_char = reqchar & 0xff;    re->req_char = reqchar & 0xff;
8278  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8279    re->req_char = reqchar & 0xffff;    re->req_char = reqchar & 0xffff;
8280    #elif defined COMPILE_PCRE32
8281      re->req_char = reqchar;
8282  #endif  #endif
8283  #endif    if ((reqcharflags & REQ_CASELESS) != 0)
   if ((reqchar & REQ_CASELESS) != 0)  
8284      {      {
8285  #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)  #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)
8286      /* We ignore non-ASCII first chars in 8 bit mode. */      /* We ignore non-ASCII first chars in 8 bit mode. */
# Line 8167  if ((re->flags & PCRE_REQCHSET) != 0) Line 8330  if ((re->flags & PCRE_REQCHSET) != 0)
8330      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
8331    }    }
8332    
8333  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8334  pcre_printint((pcre *)re, stdout, TRUE);  pcre_printint((pcre *)re, stdout, TRUE);
8335  #else  #elif defined COMPILE_PCRE16
8336  pcre16_printint((pcre *)re, stdout, TRUE);  pcre16_printint((pcre *)re, stdout, TRUE);
8337    #elif defined COMPILE_PCRE32
8338    pcre32_printint((pcre *)re, stdout, TRUE);
8339  #endif  #endif
8340    
8341  /* 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 8351  if (code - codestart > length)
8351    }    }
8352  #endif   /* PCRE_DEBUG */  #endif   /* PCRE_DEBUG */
8353    
8354  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8355  return (pcre *)re;  return (pcre *)re;
8356  #else  #elif defined COMPILE_PCRE16
8357  return (pcre16 *)re;  return (pcre16 *)re;
8358    #elif defined COMPILE_PCRE32
8359    return (pcre32 *)re;
8360  #endif  #endif
8361  }  }
8362    

Legend:
Removed from v.964  
changed lines
  Added in v.1124

  ViewVC Help
Powered by ViewVC 1.1.5