/[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 1120 by chpe, Tue Oct 16 15:57:38 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      //if (escape == 0) switch(op_code)
3137    {    {
   case OP_CHAR:  
3138  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3139    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3140  #else  #else
3141    c = *previous;    c = *previous;
3142  #endif  #endif
3143    return c != next;    }
3144    
3145    /* 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
3146    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3147    
3148    case OP_CHARI:  if (escape == 0)
3149  #ifdef SUPPORT_UTF    {
3150    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3151  #else    case, which maps to the special PT_CLIST property. Check this first. */
3152    c = *previous;  
 #endif  
   if (c == next) return FALSE;  
 #ifdef SUPPORT_UTF  
   if (utf)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3153  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3154      othercase = UCD_OTHERCASE((unsigned int)next);    if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3155  #else      {
3156      othercase = NOTACHAR;      int ocs = UCD_CASESET(next);
3157  #endif      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
     return (unsigned int)c != othercase;  
3158      }      }
   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;  
3159  #endif  #endif
   return c == next;  
3160    
3161    case OP_NOTI:    switch(op_code)
3162        {
3163        case OP_CHAR:
3164        return c != next;
3165    
3166        /* For CHARI (caseless character) we must check the other case. If we have
3167        Unicode property support, we can use it to test the other case of
3168        high-valued characters. We know that next can have only one other case,
3169        because multi-other-case characters are dealt with above. */
3170    
3171        case OP_CHARI:
3172        if (c == next) return FALSE;
3173  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3174    GETCHARTEST(c, previous);      if (utf)
3175          {
3176          pcre_uint32 othercase;
3177          if (next < 128) othercase = cd->fcc[next]; else
3178    #ifdef SUPPORT_UCP
3179          othercase = UCD_OTHERCASE(next);
3180  #else  #else
3181    c = *previous;        othercase = NOTACHAR;
3182  #endif  #endif
3183    if (c == next) return TRUE;        return c != othercase;
3184          }
3185        else
3186    #endif  /* SUPPORT_UTF */
3187        return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3188    
3189        case OP_NOT:
3190        return c == next;
3191    
3192        case OP_NOTI:
3193        if (c == next) return TRUE;
3194  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3195    if (utf)      if (utf)
3196      {        {
3197      unsigned int othercase;        pcre_uint32 othercase;
3198      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3199  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3200      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3201  #else  #else
3202      othercase = NOTACHAR;        othercase = NOTACHAR;
3203  #endif  #endif
3204      return (unsigned int)c == othercase;        return c == othercase;
3205      }        }
3206    else      else
3207  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3208    return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3209    
3210    /* 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.
3211    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. */
3212    
3213    case OP_DIGIT:      case OP_DIGIT:
3214    return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3215    
3216    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3217    return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3218    
3219    case OP_WHITESPACE:      case OP_WHITESPACE:
3220    return next > 255 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3221    
3222    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3223    return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3224    
3225    case OP_WORDCHAR:      case OP_WORDCHAR:
3226    return next > 255 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3227    
3228    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3229    return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3230    
3231    case OP_HSPACE:      case OP_HSPACE:
3232    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3233    switch(next)      switch(next)
3234      {        {
3235      case 0x09:        HSPACE_CASES:
3236      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;  
     }  
3237    
3238    case OP_ANYNL:        default:
3239    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3240    case OP_NOT_VSPACE:        }
3241    switch(next)  
3242      {      case OP_ANYNL:
3243      case 0x0a:      case OP_VSPACE:
3244      case 0x0b:      case OP_NOT_VSPACE:
3245      case 0x0c:      switch(next)
3246      case 0x0d:        {
3247      case 0x85:        VSPACE_CASES:
3248      case 0x2028:        return op_code == OP_NOT_VSPACE;
3249      case 0x2029:  
3250      return op_code == OP_NOT_VSPACE;        default:
3251      default:        return op_code != OP_NOT_VSPACE;
3252      return op_code != OP_NOT_VSPACE;        }
     }  
3253    
3254  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3255    case OP_PROP:      case OP_PROP:
3256    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, (int)previous[0], (int)previous[1], FALSE);
3257    
3258    case OP_NOTPROP:      case OP_NOTPROP:
3259    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, (int)previous[0], (int)previous[1], TRUE);
3260  #endif  #endif
3261    
3262    default:      default:
3263    return FALSE;      return FALSE;
3264        }
3265    }    }
3266    
   
3267  /* 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
3268  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
3269  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 3274  switch(op_code)
3274    {    {
3275    case OP_CHAR:    case OP_CHAR:
3276    case OP_CHARI:    case OP_CHARI:
3277  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3278      {      {
3279      case ESC_d:      case ESC_d:
3280      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
# Line 3248  switch(op_code) Line 3298  switch(op_code)
3298      case ESC_H:      case ESC_H:
3299      switch(c)      switch(c)
3300        {        {
3301        case 0x09:        HSPACE_CASES:
3302        case 0x20:        return escape != ESC_h;
3303        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;  
3304        default:        default:
3305        return -next == ESC_h;        return escape == ESC_h;
3306        }        }
3307    
3308      case ESC_v:      case ESC_v:
3309      case ESC_V:      case ESC_V:
3310      switch(c)      switch(c)
3311        {        {
3312        case 0x0a:        VSPACE_CASES:
3313        case 0x0b:        return escape != ESC_v;
3314        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3315        default:        default:
3316        return -next == ESC_v;        return escape == ESC_v;
3317        }        }
3318    
3319      /* 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
3320      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3321      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3322    
3323  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3324      case ESC_du:      case ESC_du:
# Line 3301  switch(op_code) Line 3329  switch(op_code)
3329      case ESC_SU:      case ESC_SU:
3330        {        {
3331        int temperrorcode = 0;        int temperrorcode = 0;
3332        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3333        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3334        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3335        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3336        }        }
# Line 3329  switch(op_code) Line 3357  switch(op_code)
3357    
3358        /* Do the property check. */        /* Do the property check. */
3359    
3360        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3361        }        }
3362  #endif  #endif
3363    
# Line 3344  switch(op_code) Line 3372  switch(op_code)
3372    these op-codes are never generated.) */    these op-codes are never generated.) */
3373    
3374    case OP_DIGIT:    case OP_DIGIT:
3375    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3376           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3377    
3378    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3379    return next == -ESC_d;    return escape == ESC_d;
3380    
3381    case OP_WHITESPACE:    case OP_WHITESPACE:
3382    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3383    
3384    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3385    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;
3386    
3387    case OP_HSPACE:    case OP_HSPACE:
3388    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3389           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3390    
3391    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3392    return next == -ESC_h;    return escape == ESC_h;
3393    
3394    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3395    case OP_ANYNL:    case OP_ANYNL:
3396    case OP_VSPACE:    case OP_VSPACE:
3397    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3398    
3399    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3400    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3401    
3402    case OP_WORDCHAR:    case OP_WORDCHAR:
3403    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3404           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3405    
3406    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3407    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3408    
3409    default:    default:
3410    return FALSE;    return FALSE;
# Line 3388  switch(op_code) Line 3416  switch(op_code)
3416    
3417    
3418  /*************************************************  /*************************************************
3419    *        Add a character or range to a class     *
3420    *************************************************/
3421    
3422    /* This function packages up the logic of adding a character or range of
3423    characters to a class. The character values in the arguments will be within the
3424    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3425    mutually recursive with the function immediately below.
3426    
3427    Arguments:
3428      classbits     the bit map for characters < 256
3429      uchardptr     points to the pointer for extra data
3430      options       the options word
3431      cd            contains pointers to tables etc.
3432      start         start of range character
3433      end           end of range character
3434    
3435    Returns:        the number of < 256 characters added
3436                    the pointer to extra data is updated
3437    */
3438    
3439    static int
3440    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3441      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3442    {
3443    pcre_uint32 c;
3444    int n8 = 0;
3445    
3446    /* If caseless matching is required, scan the range and process alternate
3447    cases. In Unicode, there are 8-bit characters that have alternate cases that
3448    are greater than 255 and vice-versa. Sometimes we can just extend the original
3449    range. */
3450    
3451    if ((options & PCRE_CASELESS) != 0)
3452      {
3453    #ifdef SUPPORT_UCP
3454      if ((options & PCRE_UTF8) != 0)
3455        {
3456        int rc;
3457        pcre_uint32 oc, od;
3458    
3459        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3460        c = start;
3461    
3462        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3463          {
3464          /* Handle a single character that has more than one other case. */
3465    
3466          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3467            PRIV(ucd_caseless_sets) + rc, oc);
3468    
3469          /* Do nothing if the other case range is within the original range. */
3470    
3471          else if (oc >= start && od <= end) continue;
3472    
3473          /* Extend the original range if there is overlap, noting that if oc < c, we
3474          can't have od > end because a subrange is always shorter than the basic
3475          range. Otherwise, use a recursive call to add the additional range. */
3476    
3477          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3478          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3479          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3480          }
3481        }
3482      else
3483    #endif  /* SUPPORT_UCP */
3484    
3485      /* Not UTF-mode, or no UCP */
3486    
3487      for (c = start; c <= end && c < 256; c++)
3488        {
3489        SETBIT(classbits, cd->fcc[c]);
3490        n8++;
3491        }
3492      }
3493    
3494    /* Now handle the original range. Adjust the final value according to the bit
3495    length - this means that the same lists of (e.g.) horizontal spaces can be used
3496    in all cases. */
3497    
3498    #if defined COMPILE_PCRE8
3499    #ifdef SUPPORT_UTF
3500      if ((options & PCRE_UTF8) == 0)
3501    #endif
3502      if (end > 0xff) end = 0xff;
3503    
3504    #elif defined COMPILE_PCRE16
3505    #ifdef SUPPORT_UTF
3506      if ((options & PCRE_UTF16) == 0)
3507    #endif
3508      if (end > 0xffff) end = 0xffff;
3509    
3510    #endif /* COMPILE_PCRE[8|16] */
3511    
3512    /* If all characters are less than 256, use the bit map. Otherwise use extra
3513    data. */
3514    
3515    if (end < 0x100)
3516      {
3517      for (c = start; c <= end; c++)
3518        {
3519        n8++;
3520        SETBIT(classbits, c);
3521        }
3522      }
3523    
3524    else
3525      {
3526      pcre_uchar *uchardata = *uchardptr;
3527    
3528    #ifdef SUPPORT_UTF
3529      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3530        {
3531        if (start < end)
3532          {
3533          *uchardata++ = XCL_RANGE;
3534          uchardata += PRIV(ord2utf)(start, uchardata);
3535          uchardata += PRIV(ord2utf)(end, uchardata);
3536          }
3537        else if (start == end)
3538          {
3539          *uchardata++ = XCL_SINGLE;
3540          uchardata += PRIV(ord2utf)(start, uchardata);
3541          }
3542        }
3543      else
3544    #endif  /* SUPPORT_UTF */
3545    
3546      /* Without UTF support, character values are constrained by the bit length,
3547      and can only be > 256 for 16-bit and 32-bit libraries. */
3548    
3549    #ifdef COMPILE_PCRE8
3550        {}
3551    #else
3552      if (start < end)
3553        {
3554        *uchardata++ = XCL_RANGE;
3555        *uchardata++ = start;
3556        *uchardata++ = end;
3557        }
3558      else if (start == end)
3559        {
3560        *uchardata++ = XCL_SINGLE;
3561        *uchardata++ = start;
3562        }
3563    #endif
3564    
3565      *uchardptr = uchardata;   /* Updata extra data pointer */
3566      }
3567    
3568    return n8;    /* Number of 8-bit characters */
3569    }
3570    
3571    
3572    
3573    
3574    /*************************************************
3575    *        Add a list of characters to a class     *
3576    *************************************************/
3577    
3578    /* This function is used for adding a list of case-equivalent characters to a
3579    class, and also for adding a list of horizontal or vertical whitespace. If the
3580    list is in order (which it should be), ranges of characters are detected and
3581    handled appropriately. This function is mutually recursive with the function
3582    above.
3583    
3584    Arguments:
3585      classbits     the bit map for characters < 256
3586      uchardptr     points to the pointer for extra data
3587      options       the options word
3588      cd            contains pointers to tables etc.
3589      p             points to row of 32-bit values, terminated by NOTACHAR
3590      except        character to omit; this is used when adding lists of
3591                      case-equivalent characters to avoid including the one we
3592                      already know about
3593    
3594    Returns:        the number of < 256 characters added
3595                    the pointer to extra data is updated
3596    */
3597    
3598    static int
3599    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3600      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3601    {
3602    int n8 = 0;
3603    while (p[0] < NOTACHAR)
3604      {
3605      int n = 0;
3606      if (p[0] != except)
3607        {
3608        while(p[n+1] == p[0] + n + 1) n++;
3609        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3610        }
3611      p += n + 1;
3612      }
3613    return n8;
3614    }
3615    
3616    
3617    
3618    /*************************************************
3619    *    Add characters not in a list to a class     *
3620    *************************************************/
3621    
3622    /* This function is used for adding the complement of a list of horizontal or
3623    vertical whitespace to a class. The list must be in order.
3624    
3625    Arguments:
3626      classbits     the bit map for characters < 256
3627      uchardptr     points to the pointer for extra data
3628      options       the options word
3629      cd            contains pointers to tables etc.
3630      p             points to row of 32-bit values, terminated by NOTACHAR
3631    
3632    Returns:        the number of < 256 characters added
3633                    the pointer to extra data is updated
3634    */
3635    
3636    static int
3637    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3638      int options, compile_data *cd, const pcre_uint32 *p)
3639    {
3640    BOOL utf = (options & PCRE_UTF8) != 0;
3641    int n8 = 0;
3642    if (p[0] > 0)
3643      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3644    while (p[0] < NOTACHAR)
3645      {
3646      while (p[1] == p[0] + 1) p++;
3647      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3648        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3649      p++;
3650      }
3651    return n8;
3652    }
3653    
3654    
3655    
3656    /*************************************************
3657  *           Compile one branch                   *  *           Compile one branch                   *
3658  *************************************************/  *************************************************/
3659    
# Line 3402  Arguments: Line 3668  Arguments:
3668    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3669    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3670    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3671    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3672    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3673      reqcharptr     place to put the last required character
3674      reqcharflagsptr place to put the last required character flags, or a negative number
3675    bcptr          points to current branch chain    bcptr          points to current branch chain
3676    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3677    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3416  Returns:         TRUE on success Line 3684  Returns:         TRUE on success
3684    
3685  static BOOL  static BOOL
3686  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3687    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3688    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3689      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3690      branch_chain *bcptr, int cond_depth,
3691    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3692  {  {
3693  int repeat_type, op_type;  int repeat_type, op_type;
3694  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3695  int bravalue = 0;  int bravalue = 0;
3696  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3697  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
3698  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
3699    pcre_uint32 zeroreqchar, zerofirstchar;
3700    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3701  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3702  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3703  int after_manual_callout = 0;  int after_manual_callout = 0;
3704  int length_prevgroup = 0;  int length_prevgroup = 0;
3705  register int c;  register pcre_uint32 c;
3706    int escape;
3707  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3708  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3709  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3450  must not do this for other options (e.g. Line 3723  must not do this for other options (e.g.
3723  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3724    
3725  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3726  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3727  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3728    #ifndef COMPILE_PCRE32
3729  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3730    #endif
3731  #else  #else
3732  BOOL utf = FALSE;  BOOL utf = FALSE;
3733  #endif  #endif
3734    
3735  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3736    class_uchardata always so that it can be passed to add_to_class() always,
3737    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3738    alternative calls for the different cases. */
3739    
3740    pcre_uchar *class_uchardata;
3741  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3742  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3743  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3744  #endif  #endif
3745    
# Line 3484  to take the zero repeat into account. Th Line 3762  to take the zero repeat into account. Th
3762  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3763  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3764    
3765  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3766    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3767    
3768  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
3769  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 3784  for (;; ptr++)
3784    BOOL is_recurse;    BOOL is_recurse;
3785    BOOL reset_bracount;    BOOL reset_bracount;
3786    int class_has_8bitchar;    int class_has_8bitchar;
3787    int class_single_char;    int class_one_char;
3788    int newoptions;    int newoptions;
3789    int recno;    int recno;
3790    int refsign;    int refsign;
3791    int skipbytes;    int skipbytes;
3792    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
3793    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
3794    int terminator;    int terminator;
3795    int mclength;    int mclength;
3796    int tempbracount;    int tempbracount;
3797      pcre_uint32 ec;
3798    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3799    
3800    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3677  for (;; ptr++) Line 3957  for (;; ptr++)
3957      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3958      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3959      *firstcharptr = firstchar;      *firstcharptr = firstchar;
3960        *firstcharflagsptr = firstcharflags;
3961      *reqcharptr = reqchar;      *reqcharptr = reqchar;
3962        *reqcharflagsptr = reqcharflags;
3963      *codeptr = code;      *codeptr = code;
3964      *ptrptr = ptr;      *ptrptr = ptr;
3965      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3701  for (;; ptr++) Line 3983  for (;; ptr++)
3983      previous = NULL;      previous = NULL;
3984      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3985        {        {
3986        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3987        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3988        }        }
3989      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3716  for (;; ptr++) Line 3998  for (;; ptr++)
3998      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
3999    
4000      case CHAR_DOT:      case CHAR_DOT:
4001      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4002      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4003        zerofirstcharflags = firstcharflags;
4004      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4005        zeroreqcharflags = reqcharflags;
4006      previous = code;      previous = code;
4007      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4008      break;      break;
# Line 3792  for (;; ptr++) Line 4076  for (;; ptr++)
4076          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4077        {        {
4078        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4079        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4080        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4081          zerofirstcharflags = firstcharflags;
4082        break;        break;
4083        }        }
4084    
# Line 3803  for (;; ptr++) Line 4088  for (;; ptr++)
4088    
4089      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4090    
4091      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4092      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 <
4093      < 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
4094      a single character. */      character. */
4095    
4096      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4097      class_single_char = 0;      class_one_char = 0;
4098    
4099      /* 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
4100      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
4101      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
4102      */      map. */
4103    
4104      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4105    
4106  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4107      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4108      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4109      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4110  #endif  #endif
4111    
4112      /* 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 4128  for (;; ptr++)
4128        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4129        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
4130        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4131        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4132          however. */
4133    
4134        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4135          {          {
4136            xclass = TRUE;
4137          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4138          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4139          }          }
# Line 3905  for (;; ptr++) Line 4192  for (;; ptr++)
4192          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4193          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4194    
4195          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4196            posix_class = 0;            posix_class = 0;
4197    
4198          /* 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 4235  for (;; ptr++)
4235              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4236            }            }
4237    
4238          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4239          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4240    
4241          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3964  for (;; ptr++) Line 4251  for (;; ptr++)
4251            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4252    
4253          ptr = tempptr + 1;          ptr = tempptr + 1;
4254          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4255          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4256          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4257          class_single_char = 2;          class_one_char = 2;
4258          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4259          }          }
4260    
# Line 3975  for (;; ptr++) Line 4262  for (;; ptr++)
4262        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
4263        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
4264        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4265        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4266        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4267        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4268        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4269    
4270        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4271          {          {
4272          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4273    
4274          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4275    
4276          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4277          else if (-c == ESC_N)            /* \N is not supported in a class */            c = ec;
4278            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4279            else if (escape == ESC_N)            /* \N is not supported in a class */
4280            {            {
4281            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4282            goto FAILED;            goto FAILED;
4283            }            }
4284          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4285            {            {
4286            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4287              {              {
# Line 4000  for (;; ptr++) Line 4290  for (;; ptr++)
4290            else inescq = TRUE;            else inescq = TRUE;
4291            continue;            continue;
4292            }            }
4293          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4294    
4295          if (c < 0)          else
4296            {            {
4297            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4298            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4299            class_has_8bitchar++;            class_has_8bitchar++;
4300            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4301            class_single_char += 2;            class_one_char += 2;
4302    
4303            switch (-c)            switch (escape)
4304              {              {
4305  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4306              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 4310  for (;; ptr++)
4310              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4311              case ESC_SU:              case ESC_SU:
4312              nestptr = ptr;              nestptr = ptr;
4313              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4314              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4315              continue;              continue;
4316  #endif  #endif
# Line 4044  for (;; ptr++) Line 4334  for (;; ptr++)
4334    
4335              /* 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
4336              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4337              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4338                EBCDIC, so we lazily just adjust the appropriate bit. */
4339    
4340              case ESC_s:              case ESC_s:
4341              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4057  for (;; ptr++) Line 4348  for (;; ptr++)
4348              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4349              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4350              continue;              continue;
4351    
4352                /* The rest apply in both UCP and non-UCP cases. */
4353    
4354              case ESC_h:              case ESC_h:
4355              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4356              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  
4357              continue;              continue;
4358    
4359              case ESC_H:              case ESC_H:
4360              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4361                {                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  
4362              continue;              continue;
4363    
4364              case ESC_v:              case ESC_v:
4365              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4366              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  
4367              continue;              continue;
4368    
4369              case ESC_V:              case ESC_V:
4370              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4371                {                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  
4372              continue;              continue;
4373    
4374  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4242  for (;; ptr++) Line 4379  for (;; ptr++)
4379                int pdata;                int pdata;
4380                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4381                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4382                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4383                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4384                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4385                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4262  for (;; ptr++) Line 4398  for (;; ptr++)
4398                goto FAILED;                goto FAILED;
4399                }                }
4400              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4401              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4402              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4403              break;              break;
4404              }              }
4405            }            }
4406    
4407          /* 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).
4408          greater than 256. */          This may be greater than 256. */
4409    
4410            escape = 0;
4411    
4412          }   /* End of backslash handling */          }   /* End of backslash handling */
4413    
4414        /* 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
4415        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
4416        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4417        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4418    
4419        CHECK_RANGE:        CHECK_RANGE:
4420        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4284  for (;; ptr++) Line 4422  for (;; ptr++)
4422          inescq = FALSE;          inescq = FALSE;
4423          ptr += 2;          ptr += 2;
4424          }          }
   
4425        oldptr = ptr;        oldptr = ptr;
4426    
4427        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4428    
4429        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4430    
# Line 4295  for (;; ptr++) Line 4432  for (;; ptr++)
4432    
4433        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4434          {          {
4435          int d;          pcre_uint32 d;
4436          ptr += 2;          ptr += 2;
4437          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4438    
# Line 4310  for (;; ptr++) Line 4447  for (;; ptr++)
4447            inescq = TRUE;            inescq = TRUE;
4448            break;            break;
4449            }            }
4450    
4451            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4452            back the pointer and jump to handle the character that preceded it. */
4453    
4454          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4455            {            {
4456            ptr = oldptr;            ptr = oldptr;
4457            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4458            }            }
4459    
4460            /* Otherwise, we have a potential range; pick up the next character */
4461    
4462  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4463          if (utf)          if (utf)
# Line 4332  for (;; ptr++) Line 4474  for (;; ptr++)
4474    
4475          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4476            {            {
4477            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4478              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4479            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4480    
4481            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4482    
4483            if (d < 0)            if (descape != 0)
4484              {              {
4485              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4486                {                {
4487                ptr = oldptr;                ptr = oldptr;
4488                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4489                }                }
4490              }              }
4491            }            }
4492    
4493          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4494          one-character ranges */          one-character ranges. */
4495    
4496          if (d < c)          if (d < c)
4497            {            {
4498            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4499            goto FAILED;            goto FAILED;
4500            }            }
4501            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4502    
4503          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4504            cannot be done anymore. Any value greater than 1 indicates that there
4505            is more than one character. */
4506    
4507            class_one_char = 2;
4508    
4509          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4510    
4511          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4512    
4513          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4514          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4515          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));  
             }  
           }  
   
4516          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4517          }          }
4518    
4519        /* 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
4520        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
4521        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4522          increase it if it is already 2 or more ... just in case there's a class
4523        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4524    
4525        /* Only the value of 1 matters for class_single_char. */        CLASS_SINGLE_CHARACTER:
4526          if (class_one_char < 2) class_one_char++;
4527        if (class_single_char < 2) class_single_char++;  
4528          /* If class_one_char is 1, we have the first single character in the
4529          class, and there have been no prior ranges, or XCLASS items generated by
4530          escapes. If this is the final character in the class, we can optimize by
4531          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4532          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4533          to be set. Otherwise, there can be no first char if this item is first,
4534          whatever repeat count may follow. In the case of reqchar, save the
4535          previous value for reinstating. */
4536    
4537        /* 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)  
4538          {          {
4539          ptr++;          ptr++;
4540          zeroreqchar = reqchar;          zeroreqchar = reqchar;
4541            zeroreqcharflags = reqcharflags;
4542    
4543          if (negate_class)          if (negate_class)
4544            {            {
4545            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;  #ifdef SUPPORT_UCP
4546              int d;
4547    #endif
4548              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4549            zerofirstchar = firstchar;            zerofirstchar = firstchar;
4550            *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;            zerofirstcharflags = firstcharflags;
4551  #ifdef SUPPORT_UTF  
4552            if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)            /* For caseless UTF-8 mode when UCP support is available, check
4553              code += PRIV(ord2utf)(c, code);            whether this character has more than one other case. If so, generate
4554            else            a special OP_NOTPROP item instead of OP_NOTI. */
4555    
4556    #ifdef SUPPORT_UCP
4557              if (utf && (options & PCRE_CASELESS) != 0 &&
4558                  (d = UCD_CASESET(c)) != 0)
4559                {
4560                *code++ = OP_NOTPROP;
4561                *code++ = PT_CLIST;
4562                *code++ = d;
4563                }
4564              else
4565  #endif  #endif
4566              *code++ = c;            /* Char has only one other case, or UCP not available */
4567            goto NOT_CHAR;  
4568                {
4569                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4570    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4571                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4572                  code += PRIV(ord2utf)(c, code);
4573                else
4574    #endif
4575                  *code++ = c;
4576                }
4577    
4578              /* We are finished with this character class */
4579    
4580              goto END_CLASS;
4581            }            }
4582    
4583          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4584          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4585    
4586  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4587          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4588            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4589          else          else
# Line 4564  for (;; ptr++) Line 4594  for (;; ptr++)
4594            }            }
4595          goto ONE_CHAR;          goto ONE_CHAR;
4596          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4597    
4598        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4599          has been generated. Add this character to the class. */
4600  #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
4601        if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))        class_has_8bitchar +=
4602  #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));  
           }  
         }  
4603        }        }
4604    
4605      /* 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 4619  for (;; ptr++)
4619        goto FAILED;        goto FAILED;
4620        }        }
4621    
4622        /* We will need an XCLASS if data has been placed in class_uchardata. In
4623        the second phase this is a sufficient test. However, in the pre-compile
4624        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4625        only if the very last character in the class needs XCLASS will it contain
4626        anything at this point. For this reason, xclass gets set TRUE above when
4627        uchar_classdata is emptied, and that's why this code is the way it is here
4628        instead of just doing a test on class_uchardata below. */
4629    
4630    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4631        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4632    #endif
4633    
4634      /* 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
4635      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4636      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
4637    
4638      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4639      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4640        zerofirstcharflags = firstcharflags;
4641      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4642        zeroreqcharflags = reqcharflags;
4643    
4644      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4645      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 4695  for (;; ptr++)
4695        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4696        }        }
4697      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4698      NOT_CHAR:  
4699        END_CLASS:
4700      break;      break;
4701    
4702    
# Line 4741  for (;; ptr++) Line 4734  for (;; ptr++)
4734      if (repeat_min == 0)      if (repeat_min == 0)
4735        {        {
4736        firstchar = zerofirstchar;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4737          firstcharflags = zerofirstcharflags;
4738        reqchar = zeroreqchar;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4739          reqcharflags = zeroreqcharflags;
4740        }        }
4741    
4742      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4813  for (;; ptr++) Line 4808  for (;; ptr++)
4808      if (*previous == OP_CHAR || *previous == OP_CHARI      if (*previous == OP_CHAR || *previous == OP_CHARI
4809          || *previous == OP_NOT || *previous == OP_NOTI)          || *previous == OP_NOT || *previous == OP_NOTI)
4810        {        {
4811        switch (*previous)        switch (*previous)
4812          {          {
4813          default: /* Make compiler happy. */          default: /* Make compiler happy. */
4814          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 4822  for (;; ptr++)
4822        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
4823        it's a length rather than a small character. */        it's a length rather than a small character. */
4824    
4825  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4826        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4827          {          {
4828          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4844  for (;; ptr++) Line 4839  for (;; ptr++)
4839          {          {
4840          c = code[-1];          c = code[-1];
4841          if (*previous <= OP_CHARI && repeat_min > 1)          if (*previous <= OP_CHARI && repeat_min > 1)
4842            reqchar = c | req_caseopt | cd->req_varyopt;            {
4843              reqchar = c;
4844              reqcharflags = req_caseopt | cd->req_varyopt;
4845              }
4846          }          }
4847    
4848        /* 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 4961  for (;; ptr++)
4961    
4962          if (repeat_max < 0)          if (repeat_max < 0)
4963            {            {
4964  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4965            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4966              {              {
4967              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4988  for (;; ptr++) Line 4986  for (;; ptr++)
4986    
4987          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4988            {            {
4989  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4990            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4991              {              {
4992              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5018  for (;; ptr++) Line 5016  for (;; ptr++)
5016    
5017        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
5018    
5019  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5020        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
5021          {          {
5022          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5226  for (;; ptr++) Line 5224  for (;; ptr++)
5224    
5225            else            else
5226              {              {
5227              if (groupsetfirstchar && reqchar < 0) reqchar = firstchar;              if (groupsetfirstchar && reqcharflags < 0)
5228                  {
5229                  reqchar = firstchar;
5230                  reqcharflags = firstcharflags;
5231                  }
5232    
5233              for (i = 1; i < repeat_min; i++)              for (i = 1; i < repeat_min; i++)
5234                {                {
# Line 5593  for (;; ptr++) Line 5595  for (;; ptr++)
5595        ptr++;        ptr++;
5596        while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;        while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;
5597        namelen = (int)(ptr - name);        namelen = (int)(ptr - name);
5598    
5599        /* It appears that Perl allows any characters whatsoever, other than        /* It appears that Perl allows any characters whatsoever, other than
5600        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
5601        letters, digits, and underscores. */        letters, digits, and underscores. */
# Line 5603  for (;; ptr++) Line 5605  for (;; ptr++)
5605          arg = ++ptr;          arg = ++ptr;
5606          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5607          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5608          if (arglen > (int)MAX_MARK)          if ((unsigned int)arglen > MAX_MARK)
5609            {            {
5610            *errorcodeptr = ERR75;            *errorcodeptr = ERR75;
5611            goto FAILED;            goto FAILED;
5612            }            }
5613          }          }
5614    
5615        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5623  for (;; ptr++) Line 5625  for (;; ptr++)
5625          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5626              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5627            {            {
5628              int setverb;
5629    
5630            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5631            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5632    
# Line 5640  for (;; ptr++) Line 5644  for (;; ptr++)
5644                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5645                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5646                }                }
5647              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5648                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5649    
5650              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5651              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
5652              }              }
5653    
5654            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5655  for (;; ptr++) Line 5660  for (;; ptr++)
5660                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5661                goto FAILED;                goto FAILED;
5662                }                }
5663              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5664              }              }
5665    
5666            else            else
# Line 5666  for (;; ptr++) Line 5670  for (;; ptr++)
5670                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5671                goto FAILED;                goto FAILED;
5672                }                }
5673              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5674              *code++ = arglen;              *code++ = arglen;
5675              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5676              code += arglen;              code += arglen;
5677              *code++ = 0;              *code++ = 0;
5678              }              }
5679    
5680              switch (setverb)
5681                {
5682                case OP_THEN:
5683                case OP_THEN_ARG:
5684                cd->external_flags |= PCRE_HASTHEN;
5685                break;
5686    
5687                case OP_PRUNE:
5688                case OP_PRUNE_ARG:
5689                case OP_SKIP:
5690                case OP_SKIP_ARG:
5691                cd->had_pruneorskip = TRUE;
5692                break;
5693                }
5694    
5695            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5696            }            }
5697    
# Line 5796  for (;; ptr++) Line 5814  for (;; ptr++)
5814          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
5815            {            {
5816            if (recno >= 0)            if (recno >= 0)
5817              recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1;              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;
5818            ptr++;            ptr++;
5819            }            }
5820          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
# Line 6393  for (;; ptr++) Line 6411  for (;; ptr++)
6411    
6412          /* Can't determine a first byte now */          /* Can't determine a first byte now */
6413    
6414          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6415          continue;          continue;
6416    
6417    
# Line 6527  for (;; ptr++) Line 6545  for (;; ptr++)
6545           cond_depth +           cond_depth +
6546             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
6547           &subfirstchar,                   /* For possible first char */           &subfirstchar,                   /* For possible first char */
6548             &subfirstcharflags,
6549           &subreqchar,                     /* For possible last char */           &subreqchar,                     /* For possible last char */
6550             &subreqcharflags,
6551           bcptr,                           /* Current branch chain */           bcptr,                           /* Current branch chain */
6552           cd,                              /* Tables block */           cd,                              /* Tables block */
6553           (lengthptr == NULL)? NULL :      /* Actual compile phase */           (lengthptr == NULL)? NULL :      /* Actual compile phase */
# Line 6588  for (;; ptr++) Line 6608  for (;; ptr++)
6608            *errorcodeptr = ERR27;            *errorcodeptr = ERR27;
6609            goto FAILED;            goto FAILED;
6610            }            }
6611          if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;          if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE;
6612          }          }
6613        }        }
6614    
# Line 6637  for (;; ptr++) Line 6657  for (;; ptr++)
6657      back off. */      back off. */
6658    
6659      zeroreqchar = reqchar;      zeroreqchar = reqchar;
6660        zeroreqcharflags = reqcharflags;
6661      zerofirstchar = firstchar;      zerofirstchar = firstchar;
6662        zerofirstcharflags = firstcharflags;
6663      groupsetfirstchar = FALSE;      groupsetfirstchar = FALSE;
6664    
6665      if (bravalue >= OP_ONCE)      if (bravalue >= OP_ONCE)
# Line 6648  for (;; ptr++) Line 6670  for (;; ptr++)
6670        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
6671        repeat forces firstchar to "none". */        repeat forces firstchar to "none". */
6672    
6673        if (firstchar == REQ_UNSET)        if (firstcharflags == REQ_UNSET)
6674          {          {
6675          if (subfirstchar >= 0)          if (subfirstcharflags >= 0)
6676            {            {
6677            firstchar = subfirstchar;            firstchar = subfirstchar;
6678              firstcharflags = subfirstcharflags;
6679            groupsetfirstchar = TRUE;            groupsetfirstchar = TRUE;
6680            }            }
6681          else firstchar = REQ_NONE;          else firstcharflags = REQ_NONE;
6682          zerofirstchar = REQ_NONE;          zerofirstcharflags = REQ_NONE;
6683          }          }
6684    
6685        /* If firstchar was previously set, convert the subpattern's firstchar        /* If firstchar was previously set, convert the subpattern's firstchar
6686        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
6687        existence beforehand. */        existence beforehand. */
6688    
6689        else if (subfirstchar >= 0 && subreqchar < 0)        else if (subfirstcharflags >= 0 && subreqcharflags < 0)
6690          subreqchar = subfirstchar | tempreqvary;          {
6691            subreqchar = subfirstchar;
6692            subreqcharflags = subfirstcharflags | tempreqvary;
6693            }
6694    
6695        /* 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
6696        really the first byte - see above), set it. */        really the first byte - see above), set it. */
6697    
6698        if (subreqchar >= 0) reqchar = subreqchar;        if (subreqcharflags >= 0)
6699            {
6700            reqchar = subreqchar;
6701            reqcharflags = subreqcharflags;
6702            }
6703        }        }
6704    
6705      /* 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 6710  for (;; ptr++)
6710      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
6711      firstchar, looking for an asserted first char. */      firstchar, looking for an asserted first char. */
6712    
6713      else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;      else if (bravalue == OP_ASSERT && subreqcharflags >= 0)
6714          {
6715          reqchar = subreqchar;
6716          reqcharflags = subreqcharflags;
6717          }
6718      break;     /* End of processing '(' */      break;     /* End of processing '(' */
6719    
6720    
# Line 6688  for (;; ptr++) Line 6722  for (;; ptr++)
6722      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
6723      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
6724      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
6725      are ESC_REF plus the reference number. Only back references and those types      are negative the reference number. Only back references and those types
6726      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
6727      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
6728      ever created. */      ever created. */
6729    
6730      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6731      tempptr = ptr;      tempptr = ptr;
6732      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6733    
6734      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6735    
6736      if (c < 0)      if (escape == 0)
6737          c = ec;
6738        else
6739        {        {
6740        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6741          {          {
6742          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6743            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6708  for (;; ptr++) Line 6745  for (;; ptr++)
6745          continue;          continue;
6746          }          }
6747    
6748        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6749    
6750        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6751        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6752    
6753        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6754          firstchar = REQ_NONE;          firstcharflags = REQ_NONE;
6755    
6756        /* 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. */
6757    
6758        zerofirstchar = firstchar;        zerofirstchar = firstchar;
6759          zerofirstcharflags = firstcharflags;
6760        zeroreqchar = reqchar;        zeroreqchar = reqchar;
6761          zeroreqcharflags = reqcharflags;
6762    
6763        /* \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'
6764        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6765        -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 <
6766        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
6767        -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
6768        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6769    
6770        if (-c == ESC_g)        if (escape == ESC_g)
6771          {          {
6772          const pcre_uchar *p;          const pcre_uchar *p;
6773          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 6823  for (;; ptr++)
6823        /* \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).
6824        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6825    
6826        if (-c == ESC_k)        if (escape == ESC_k)
6827          {          {
6828          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6829            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 6842  for (;; ptr++)
6842        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
6843        ':' later. */        ':' later. */
6844    
6845        if (-c >= ESC_REF)        if (escape < 0)
6846          {          {
6847          open_capitem *oc;          open_capitem *oc;
6848          recno = -c - ESC_REF;          recno = -escape;
6849    
6850          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6851          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6852          previous = code;          previous = code;
6853          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6854          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
# Line 6833  for (;; ptr++) Line 6872  for (;; ptr++)
6872        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6873    
6874  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6875        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6876          {          {
6877          BOOL negated;          BOOL negated;
6878          int pdata;          int pdata;
6879          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
6880          if (ptype < 0) goto FAILED;          if (ptype < 0) goto FAILED;
6881          previous = code;          previous = code;
6882          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6883          *code++ = ptype;          *code++ = ptype;
6884          *code++ = pdata;          *code++ = pdata;
6885          }          }
# Line 6849  for (;; ptr++) Line 6888  for (;; ptr++)
6888        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6889        allowed. */        allowed. */
6890    
6891        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6892          {          {
6893          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6894          goto FAILED;          goto FAILED;
# Line 6859  for (;; ptr++) Line 6898  for (;; ptr++)
6898        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6899        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
6900        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
6901        Unicode property tests. Note that \b and \B do a one-character        Unicode property tests. Note that \b and \B do a one-character
6902        lookbehind. */        lookbehind. */
6903    
6904        else        else
6905          {          {
6906          if ((-c == ESC_b || -c == ESC_B) && cd->max_lookbehind == 0)          if ((escape == ESC_b || escape == ESC_B) && cd->max_lookbehind == 0)
6907            cd->max_lookbehind = 1;            cd->max_lookbehind = 1;
6908  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6909          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6910            {            {
6911            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6912            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6913            }            }
6914          else          else
6915  #endif  #endif
# Line 6878  for (;; ptr++) Line 6917  for (;; ptr++)
6917          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6918    
6919            {            {
6920            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
6921            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
6922            }            }
6923          }          }
6924        continue;        continue;
# Line 6889  for (;; ptr++) Line 6928  for (;; ptr++)
6928      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
6929      handle it as a data character. */      handle it as a data character. */
6930    
6931  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6932      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6933        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6934      else      else
# Line 6922  for (;; ptr++) Line 6961  for (;; ptr++)
6961    
6962      ONE_CHAR:      ONE_CHAR:
6963      previous = code;      previous = code;
6964    
6965        /* For caseless UTF-8 mode when UCP support is available, check whether
6966        this character has more than one other case. If so, generate a special
6967        OP_PROP item instead of OP_CHARI. */
6968    
6969    #ifdef SUPPORT_UCP
6970        if (utf && (options & PCRE_CASELESS) != 0)
6971          {
6972          GETCHAR(c, mcbuffer);
6973          if ((c = UCD_CASESET(c)) != 0)
6974            {
6975            *code++ = OP_PROP;
6976            *code++ = PT_CLIST;
6977            *code++ = c;
6978            if (firstcharflags == REQ_UNSET) firstcharflags = zerofirstcharflags = REQ_NONE;
6979            break;
6980            }
6981          }
6982    #endif
6983    
6984        /* Caseful matches, or not one of the multicase characters. */
6985    
6986      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
6987      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
6988    
# Line 6935  for (;; ptr++) Line 6996  for (;; ptr++)
6996      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
6997      repeat. */      repeat. */
6998    
6999      if (firstchar == REQ_UNSET)      if (firstcharflags == REQ_UNSET)
7000        {        {
7001        zerofirstchar = REQ_NONE;        zerofirstcharflags = REQ_NONE;
7002        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7003          zeroreqcharflags = reqcharflags;
7004    
7005        /* 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
7006        only if it is not to be matched caselessly. */        only if it is not to be matched caselessly. */
# Line 6946  for (;; ptr++) Line 7008  for (;; ptr++)
7008        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7009          {          {
7010          firstchar = mcbuffer[0] | req_caseopt;          firstchar = mcbuffer[0] | req_caseopt;
7011          if (mclength != 1) reqchar = code[-1] | cd->req_varyopt;          firstchar = mcbuffer[0];
7012            firstcharflags = req_caseopt;
7013    
7014            if (mclength != 1)
7015              {
7016              reqchar = code[-1];
7017              reqcharflags = cd->req_varyopt;
7018              }
7019          }          }
7020        else firstchar = reqchar = REQ_NONE;        else firstcharflags = reqcharflags = REQ_NONE;
7021        }        }
7022    
7023      /* 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 7026  for (;; ptr++)
7026      else      else
7027        {        {
7028        zerofirstchar = firstchar;        zerofirstchar = firstchar;
7029          zerofirstcharflags = firstcharflags;
7030        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7031          zeroreqcharflags = reqcharflags;
7032        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7033          reqchar = code[-1] | req_caseopt | cd->req_varyopt;          {
7034            reqchar = code[-1];
7035            reqcharflags = req_caseopt | cd->req_varyopt;
7036            }
7037        }        }
7038    
7039      break;            /* End of literal character handling */      break;            /* End of literal character handling */
# Line 6978  return FALSE; Line 7052  return FALSE;
7052    
7053    
7054    
   
7055  /*************************************************  /*************************************************
7056  *     Compile sequence of alternatives           *  *     Compile sequence of alternatives           *
7057  *************************************************/  *************************************************/
# Line 6999  Arguments: Line 7072  Arguments:
7072    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
7073    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
7074    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
7075    firstcharptr   place to put the first required character, or a negative number    firstcharptr    place to put the first required character
7076    reqcharptr     place to put the last required character, or a negative number    firstcharflagsptr place to put the first character flags, or a negative number
7077      reqcharptr     place to put the last required character
7078      reqcharflagsptr place to put the last required character flags, or a negative number
7079    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
7080    cd             points to the data block with tables pointers etc.    cd             points to the data block with tables pointers etc.
7081    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
# Line 7012  Returns:         TRUE on success Line 7087  Returns:         TRUE on success
7087  static BOOL  static BOOL
7088  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,
7089    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
7090    int cond_depth, pcre_int32 *firstcharptr, pcre_int32 *reqcharptr,    int cond_depth,
7091      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
7092      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
7093    branch_chain *bcptr, compile_data *cd, int *lengthptr)    branch_chain *bcptr, compile_data *cd, int *lengthptr)
7094  {  {
7095  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
# Line 7022  pcre_uchar *start_bracket = code; Line 7099  pcre_uchar *start_bracket = code;
7099  pcre_uchar *reverse_count = NULL;  pcre_uchar *reverse_count = NULL;
7100  open_capitem capitem;  open_capitem capitem;
7101  int capnumber = 0;  int capnumber = 0;
7102  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7103  pcre_int32 branchfirstchar, branchreqchar;  pcre_int32 firstcharflags, reqcharflags;
7104    pcre_uint32 branchfirstchar, branchreqchar;
7105    pcre_int32 branchfirstcharflags, branchreqcharflags;
7106  int length;  int length;
7107  int orig_bracount;  int orig_bracount;
7108  int max_bracount;  int max_bracount;
# Line 7032  branch_chain bc; Line 7111  branch_chain bc;
7111  bc.outer = bcptr;  bc.outer = bcptr;
7112  bc.current_branch = code;  bc.current_branch = code;
7113    
7114  firstchar = reqchar = REQ_UNSET;  firstchar = reqchar = 0;
7115    firstcharflags = reqcharflags = REQ_UNSET;
7116    
7117  /* 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
7118  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 7172  for (;;)
7172    into the length. */    into the length. */
7173    
7174    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,
7175          &branchreqchar, &bc, cond_depth, cd,          &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc,
7176          (lengthptr == NULL)? NULL : &length))          cond_depth, cd, (lengthptr == NULL)? NULL : &length))
7177      {      {
7178      *ptrptr = ptr;      *ptrptr = ptr;
7179      return FALSE;      return FALSE;
# Line 7114  for (;;) Line 7194  for (;;)
7194      if (*last_branch != OP_ALT)      if (*last_branch != OP_ALT)
7195        {        {
7196        firstchar = branchfirstchar;        firstchar = branchfirstchar;
7197          firstcharflags = branchfirstcharflags;
7198        reqchar = branchreqchar;        reqchar = branchreqchar;
7199          reqcharflags = branchreqcharflags;
7200        }        }
7201    
7202      /* 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 7210  for (;;)
7210        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
7211        previously no reqchar, it takes on the value of the old firstchar. */        previously no reqchar, it takes on the value of the old firstchar. */
7212    
7213        if (firstchar >= 0 && firstchar != branchfirstchar)        if (firstcharflags >= 0 &&
7214              (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar))
7215          {          {
7216          if (reqchar < 0) reqchar = firstchar;          if (reqcharflags < 0)
7217          firstchar = REQ_NONE;            {
7218              reqchar = firstchar;
7219              reqcharflags = firstcharflags;
7220              }
7221            firstcharflags = REQ_NONE;
7222          }          }
7223    
7224        /* 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
7225        branch becomes a reqchar if there isn't a branch reqchar. */        branch becomes a reqchar if there isn't a branch reqchar. */
7226    
7227        if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0)        if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0)
7228            branchreqchar = branchfirstchar;          {
7229            branchreqchar = branchfirstchar;
7230            branchreqcharflags = branchfirstcharflags;
7231            }
7232    
7233        /* Now ensure that the reqchars match */        /* Now ensure that the reqchars match */
7234    
7235        if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY))        if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) ||
7236          reqchar = REQ_NONE;            reqchar != branchreqchar)
7237        else reqchar |= branchreqchar;   /* To "or" REQ_VARY */          reqcharflags = REQ_NONE;
7238          else
7239            {
7240            reqchar = branchreqchar;
7241            reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */
7242            }
7243        }        }
7244    
7245      /* 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 7268  for (;;)
7268          *ptrptr = ptr;          *ptrptr = ptr;
7269          return FALSE;          return FALSE;
7270          }          }
7271        else        else
7272          {          {
7273          if (fixed_length > cd->max_lookbehind)          if (fixed_length > cd->max_lookbehind)
7274            cd->max_lookbehind = fixed_length;            cd->max_lookbehind = fixed_length;
7275          PUT(reverse_count, 0, fixed_length);          PUT(reverse_count, 0, fixed_length);
7276          }          }
7277        }        }
7278      }      }
# Line 7240  for (;;) Line 7335  for (;;)
7335      *codeptr = code;      *codeptr = code;
7336      *ptrptr = ptr;      *ptrptr = ptr;
7337      *firstcharptr = firstchar;      *firstcharptr = firstchar;
7338        *firstcharflagsptr = firstcharflags;
7339      *reqcharptr = reqchar;      *reqcharptr = reqchar;
7340        *reqcharflagsptr = reqcharflags;
7341      if (lengthptr != NULL)      if (lengthptr != NULL)
7342        {        {
7343        if (OFLOW_MAX - *lengthptr < length)        if (OFLOW_MAX - *lengthptr < length)
# Line 7310  and the highest back reference was great Line 7407  and the highest back reference was great
7407  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
7408  of the more common cases more precisely.  of the more common cases more precisely.
7409    
7410    ... A second exception is when the .* appears inside an atomic group, because
7411    this prevents the number of characters it matches from being adjusted.
7412    
7413  Arguments:  Arguments:
7414    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7415    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
7416                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7417                    the less precise approach                    the less precise approach
7418    backref_map    the back reference bitmap    cd             points to the compile data block
7419      atomcount      atomic group level
7420    
7421  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7422  */  */
7423    
7424  static BOOL  static BOOL
7425  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7426    unsigned int backref_map)    compile_data *cd, int atomcount)
7427  {  {
7428  do {  do {
7429     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7334  do { Line 7435  do {
7435     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7436         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7437       {       {
7438       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7439       }       }
7440    
7441     /* Capturing brackets */     /* Capturing brackets */
# Line 7344  do { Line 7445  do {
7445       {       {
7446       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7447       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7448       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7449         }
7450    
7451       /* Positive forward assertions and conditions */
7452    
7453       else if (op == OP_ASSERT || op == OP_COND)
7454         {
7455         if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7456       }       }
7457    
7458     /* Other brackets */     /* Atomic groups */
7459    
7460     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ONCE || op == OP_ONCE_NC)
             op == OP_COND)  
7461       {       {
7462       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7463           return FALSE;
7464       }       }
7465    
7466     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7467     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
7468       group. */
7469    
7470     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7471               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7472       {       {
7473       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7474             atomcount > 0 || cd->had_pruneorskip)
7475         return FALSE;         return FALSE;
7476       }       }
7477    
7478     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7479    
7480     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;
7481    
7482     code += GET(code, 1);     code += GET(code, 1);
7483     }     }
7484  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7385  return TRUE; Line 7496  return TRUE;
7496  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
7497  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
7498  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7499  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7500    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7501    count, because once again the assumption no longer holds.
7502    
7503  Arguments:  Arguments:
7504    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7505    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
7506                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7507                    the less precise approach                    the less precise approach
7508    backref_map    the back reference bitmap    cd             points to the compile data
7509      atomcount      atomic group level
7510    
7511  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7512  */  */
7513    
7514  static BOOL  static BOOL
7515  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7516    unsigned int backref_map)    compile_data *cd, int atomcount)
7517  {  {
7518  do {  do {
7519     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7425  do { Line 7539  do {
7539         return FALSE;         return FALSE;
7540    
7541         default:     /* Assertion */         default:     /* Assertion */
7542         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7543         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7544         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7545         break;         break;
# Line 7439  do { Line 7553  do {
7553     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7554         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7555       {       {
7556       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7557       }       }
7558    
7559     /* Capturing brackets */     /* Capturing brackets */
# Line 7449  do { Line 7563  do {
7563       {       {
7564       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7565       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7566       if (!is_startline(scode, new_map, backref_map)) return FALSE;       if (!is_startline(scode, new_map, cd, atomcount)) return FALSE;
7567       }       }
7568    
7569     /* Other brackets */     /* Positive forward assertions */
7570    
7571       else if (op == OP_ASSERT)
7572         {
7573         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7574         }
7575    
7576       /* Atomic brackets */
7577    
7578     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)     else if (op == OP_ONCE || op == OP_ONCE_NC)
7579       {       {
7580       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE;
7581       }       }
7582    
7583     /* .* 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
7584     may be referenced. */     brackets that may be referenced, as long as the pattern does not contain
7585       *PRUNE or *SKIP, because these break the feature. Consider, for example,
7586       /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the
7587       start of a line. */
7588    
7589     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)
7590       {       {
7591       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 ||
7592             atomcount > 0 || cd->had_pruneorskip)
7593           return FALSE;
7594       }       }
7595    
7596     /* Check for explicit circumflex */     /* Check for explicit circumflex; anything else gives a FALSE result. Note
7597       in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC
7598       because the number of characters matched by .* cannot be adjusted inside
7599       them. */
7600    
7601     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;
7602    
# Line 7495  we return that char, otherwise -1. Line 7624  we return that char, otherwise -1.
7624    
7625  Arguments:  Arguments:
7626    code       points to start of expression (the bracket)    code       points to start of expression (the bracket)
7627      flags       points to the first char flags, or to REQ_NONE
7628    inassert   TRUE if in an assertion    inassert   TRUE if in an assertion
7629    
7630  Returns:     -1 or the fixed first char  Returns:     the fixed first char, or 0 with REQ_NONE in flags
7631  */  */
7632    
7633  static int  static pcre_uint32
7634  find_firstassertedchar(const pcre_uchar *code, BOOL inassert)  find_firstassertedchar(const pcre_uchar *code, pcre_int32 *flags,
7635      BOOL inassert)
7636  {  {
7637  register int c = -1;  register pcre_uint32 c = 0;
7638    int cflags = REQ_NONE;
7639    
7640    *flags = REQ_NONE;
7641  do {  do {
7642     int d;     pcre_uint32 d;
7643       int dflags;
7644     int xl = (*code == OP_CBRA || *code == OP_SCBRA ||     int xl = (*code == OP_CBRA || *code == OP_SCBRA ||
7645               *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0;               *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0;
7646     const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,     const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,
7647       TRUE);       TRUE);
7648     register int op = *scode;     register pcre_uchar op = *scode;
7649    
7650     switch(op)     switch(op)
7651       {       {
7652       default:       default:
7653       return -1;       return 0;
7654    
7655       case OP_BRA:       case OP_BRA:
7656       case OP_BRAPOS:       case OP_BRAPOS:
# Line 7527  do { Line 7662  do {
7662       case OP_ONCE:       case OP_ONCE:
7663       case OP_ONCE_NC:       case OP_ONCE_NC:
7664       case OP_COND:       case OP_COND:
7665       if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0)       d = find_firstassertedchar(scode, &dflags, op == OP_ASSERT);
7666         return -1;       if (dflags < 0)
7667       if (c < 0) c = d; else if (c != d) return -1;         return 0;
7668         if (cflags < 0) { c = d; cflags = dflags; } else if (c != d || cflags != dflags) return 0;
7669       break;       break;
7670    
7671       case OP_EXACT:       case OP_EXACT:
# Line 7540  do { Line 7676  do {
7676       case OP_PLUS:       case OP_PLUS:
7677       case OP_MINPLUS:       case OP_MINPLUS:
7678       case OP_POSPLUS:       case OP_POSPLUS:
7679       if (!inassert) return -1;       if (!inassert) return 0;
7680       if (c < 0) c = scode[1];       if (cflags < 0) { c = scode[1]; cflags = 0; }
7681         else if (c != scode[1]) return -1;         else if (c != scode[1]) return 0;
7682       break;       break;
7683    
7684       case OP_EXACTI:       case OP_EXACTI:
# Line 7553  do { Line 7689  do {
7689       case OP_PLUSI:       case OP_PLUSI:
7690       case OP_MINPLUSI:       case OP_MINPLUSI:
7691       case OP_POSPLUSI:       case OP_POSPLUSI:
7692       if (!inassert) return -1;       if (!inassert) return 0;
7693       if (c < 0) c = scode[1] | REQ_CASELESS;       if (cflags < 0) { c = scode[1]; cflags = REQ_CASELESS; }
7694         else if (c != scode[1]) return -1;         else if (c != scode[1]) return 0;
7695       break;       break;
7696       }       }
7697    
7698     code += GET(code, 1);     code += GET(code, 1);
7699     }     }
7700  while (*code == OP_ALT);  while (*code == OP_ALT);
7701    
7702    *flags = cflags;
7703  return c;  return c;
7704  }  }
7705    
# Line 7589  Returns:        pointer to compiled data Line 7727  Returns:        pointer to compiled data
7727                  with errorptr and erroroffset set                  with errorptr and erroroffset set
7728  */  */
7729    
7730  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7731  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7732  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
7733    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7734  #else  #elif defined COMPILE_PCRE16
7735  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7736  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,
7737    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7738    #elif defined COMPILE_PCRE32
7739    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7740    pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr,
7741      int *erroroffset, const unsigned char *tables)
7742  #endif  #endif
7743  {  {
7744  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7745  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7746  #else  #elif defined COMPILE_PCRE16
7747  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7748    #elif defined COMPILE_PCRE32
7749    return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7750  #endif  #endif
7751  }  }
7752    
7753    
7754  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7755  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7756  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
7757    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7758  #else  #elif defined COMPILE_PCRE16
7759  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7760  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,
7761    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7762    #elif defined COMPILE_PCRE32
7763    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7764    pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr,
7765      const char **errorptr, int *erroroffset, const unsigned char *tables)
7766  #endif  #endif
7767  {  {
7768  REAL_PCRE *re;  REAL_PCRE *re;
7769  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
7770  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7771    pcre_int32 firstcharflags, reqcharflags;
7772  int newline;  int newline;
7773  int errorcode = 0;  int errorcode = 0;
7774  int skipatstart = 0;  int skipatstart = 0;
# Line 7700  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7849  while (ptr[skipatstart] == CHAR_LEFT_PAR
7849    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7850      { skipatstart += 8; options |= PCRE_UTF16; continue; }      { skipatstart += 8; options |= PCRE_UTF16; continue; }
7851  #endif  #endif
7852    #ifdef COMPILE_PCRE32
7853      if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7854        { skipatstart += 8; options |= PCRE_UTF32; continue; }
7855    #endif
7856    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)
7857      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
7858    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 7881  while (ptr[skipatstart] == CHAR_LEFT_PAR
7881    else break;    else break;
7882    }    }
7883    
7884  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */
7885  utf = (options & PCRE_UTF8) != 0;  utf = (options & PCRE_UTF8) != 0;
7886    
7887  /* 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 7893  not used here. */
7893  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&
7894       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)
7895    {    {
7896  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7897    errorcode = ERR44;    errorcode = ERR44;
7898  #else  #elif defined COMPILE_PCRE16
7899    errorcode = ERR74;    errorcode = ERR74;
7900    #elif defined COMPILE_PCRE32
7901      errorcode = ERR77;
7902  #endif  #endif
7903    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;
7904    }    }
# Line 7863  ptr += skipatstart; Line 8018  ptr += skipatstart;
8018  code = cworkspace;  code = cworkspace;
8019  *code = OP_BRA;  *code = OP_BRA;
8020  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
8021    FALSE, 0, 0, &firstchar, &reqchar, NULL, cd, &length);    FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL,
8022      cd, &length);
8023  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
8024    
8025  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 8063  re->name_count = cd->names_found;
8063  re->ref_count = 0;  re->ref_count = 0;
8064  re->tables = (tables == PRIV(default_tables))? NULL : tables;  re->tables = (tables == PRIV(default_tables))? NULL : tables;
8065  re->nullpad = NULL;  re->nullpad = NULL;
8066    #ifdef COMPILE_PCRE32
8067    re->dummy1 = re->dummy2 = 0;
8068    #endif
8069    
8070  /* 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
8071  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 8085  cd->start_code = codestart;
8085  cd->hwm = (pcre_uchar *)(cd->start_workspace);  cd->hwm = (pcre_uchar *)(cd->start_workspace);
8086  cd->req_varyopt = 0;  cd->req_varyopt = 0;
8087  cd->had_accept = FALSE;  cd->had_accept = FALSE;
8088    cd->had_pruneorskip = FALSE;
8089  cd->check_lookbehind = FALSE;  cd->check_lookbehind = FALSE;
8090  cd->open_caps = NULL;  cd->open_caps = NULL;
8091    
# Line 7937  ptr = (const pcre_uchar *)pattern + skip Line 8097  ptr = (const pcre_uchar *)pattern + skip
8097  code = (pcre_uchar *)codestart;  code = (pcre_uchar *)codestart;
8098  *code = OP_BRA;  *code = OP_BRA;
8099  (void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0,  (void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0,
8100    &firstchar, &reqchar, NULL, cd, NULL);    &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL, cd, NULL);
8101  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
8102  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
8103  re->max_lookbehind = cd->max_lookbehind;  re->max_lookbehind = cd->max_lookbehind;
8104  re->flags = cd->external_flags | PCRE_MODE;  re->flags = cd->external_flags | PCRE_MODE;
8105    
8106  if (cd->had_accept) reqchar = REQ_NONE;   /* Must disable after (*ACCEPT) */  if (cd->had_accept)
8107      {
8108      reqchar = 0;              /* Must disable after (*ACCEPT) */
8109      reqcharflags = REQ_NONE;
8110      }
8111    
8112  /* 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. */
8113    
# Line 7993  if (errorcode == 0 && re->top_backref > Line 8157  if (errorcode == 0 && re->top_backref >
8157    
8158  /* If there were any lookbehind assertions that contained OP_RECURSE  /* If there were any lookbehind assertions that contained OP_RECURSE
8159  (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,
8160  because they may contain forward references. Actual recursions can't be fixed  because they may contain forward references. Actual recursions cannot be fixed
8161  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
8162  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
8163  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 8213  if (errorcode != 0)
8213    }    }
8214    
8215  /* 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
8216  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
8217  as starting with .* when DOTALL is set).  as starting with non-atomic .* when DOTALL is set and there are no occurrences
8218    of *PRUNE or *SKIP.
8219    
8220  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
8221  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
8222  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches
8223  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
8224  */  non-DOTALL matches when *PRUNE and SKIP are not present. */
8225    
8226  if ((re->options & PCRE_ANCHORED) == 0)  if ((re->options & PCRE_ANCHORED) == 0)
8227    {    {
8228    if (is_anchored(codestart, 0, cd->backref_map))    if (is_anchored(codestart, 0, cd, 0)) re->options |= PCRE_ANCHORED;
     re->options |= PCRE_ANCHORED;  
8229    else    else
8230      {      {
8231      if (firstchar < 0)      if (firstcharflags < 0)
8232        firstchar = find_firstassertedchar(codestart, FALSE);        firstchar = find_firstassertedchar(codestart, &firstcharflags, FALSE);
8233      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */      if (firstcharflags >= 0)   /* Remove caseless flag for non-caseable chars */
8234        {        {
8235  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8236        re->first_char = firstchar & 0xff;        re->first_char = firstchar & 0xff;
8237  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8238        re->first_char = firstchar & 0xffff;        re->first_char = firstchar & 0xffff;
8239    #elif defined COMPILE_PCRE32
8240          re->first_char = firstchar;
8241  #endif  #endif
8242  #endif        if ((firstcharflags & REQ_CASELESS) != 0)
       if ((firstchar & REQ_CASELESS) != 0)  
8243          {          {
8244  #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)  #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)
8245          /* 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 8262  if ((re->options & PCRE_ANCHORED) == 0)
8262    
8263        re->flags |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
8264        }        }
8265      else if (is_startline(codestart, 0, cd->backref_map))  
8266        re->flags |= PCRE_STARTLINE;      else if (is_startline(codestart, 0, cd, 0)) re->flags |= PCRE_STARTLINE;
8267      }      }
8268    }    }
8269    
# Line 8107  if ((re->options & PCRE_ANCHORED) == 0) Line 8271  if ((re->options & PCRE_ANCHORED) == 0)
8271  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
8272  bytes. */  bytes. */
8273    
8274  if (reqchar >= 0 &&  if (reqcharflags >= 0 &&
8275       ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0))       ((re->options & PCRE_ANCHORED) == 0 || (reqcharflags & REQ_VARY) != 0))
8276    {    {
8277  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8278    re->req_char = reqchar & 0xff;    re->req_char = reqchar & 0xff;
8279  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
8280    re->req_char = reqchar & 0xffff;    re->req_char = reqchar & 0xffff;
8281    #elif defined COMPILE_PCRE32
8282      re->req_char = reqchar;
8283  #endif  #endif
8284  #endif    if ((reqcharflags & REQ_CASELESS) != 0)
   if ((reqchar & REQ_CASELESS) != 0)  
8285      {      {
8286  #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)  #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)
8287      /* 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 8331  if ((re->flags & PCRE_REQCHSET) != 0)
8331      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
8332    }    }
8333    
8334  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8335  pcre_printint((pcre *)re, stdout, TRUE);  pcre_printint((pcre *)re, stdout, TRUE);
8336  #else  #elif defined COMPILE_PCRE16
8337  pcre16_printint((pcre *)re, stdout, TRUE);  pcre16_printint((pcre *)re, stdout, TRUE);
8338    #elif defined COMPILE_PCRE32
8339    pcre32_printint((pcre *)re, stdout, TRUE);
8340  #endif  #endif
8341    
8342  /* 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 8352  if (code - codestart > length)
8352    }    }
8353  #endif   /* PCRE_DEBUG */  #endif   /* PCRE_DEBUG */
8354    
8355  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8356  return (pcre *)re;  return (pcre *)re;
8357  #else  #elif defined COMPILE_PCRE16
8358  return (pcre16 *)re;  return (pcre16 *)re;
8359    #elif defined COMPILE_PCRE32
8360    return (pcre32 *)re;
8361  #endif  #endif
8362  }  }
8363    

Legend:
Removed from v.964  
changed lines
  Added in v.1120

  ViewVC Help
Powered by ViewVC 1.1.5