/[pcre]/code/trunk/pcre_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_compile.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 905 by zherczeg, Mon Jan 23 19:26:03 2012 UTC revision 1147 by chpe, Sat Oct 20 20:17:03 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 489  static const char error_texts[] = Line 504  static const char error_texts[] =
504    "too many forward references\0"    "too many forward references\0"
505    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
506    "invalid UTF-16 string\0"    "invalid UTF-16 string\0"
507      /* 75 */
508      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
509      "character value in \\u.... sequence is too large\0"
510      "invalid UTF-32 string\0"
511    ;    ;
512    
513  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 628  static const pcre_uint8 ebcdic_chartab[] Line 647  static const pcre_uint8 ebcdic_chartab[]
647  #endif  #endif
648    
649    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
   
650    
651  /*************************************************  /*************************************************
652  *            Find an error text                  *  *            Find an error text                  *
# Line 739  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 751  return (*p == CHAR_RIGHT_CURLY_BRACKET);
751  *************************************************/  *************************************************/
752    
753  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
754  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
755  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
756  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
757  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
758  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
759  sequence.  escape sequence.
760    
761  Arguments:  Arguments:
762    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
763      chptr          points to the data character
764    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
765    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
766    options        the options bits    options        the options bits
767    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
768    
769  Returns:         zero or positive => a data character  Returns:         zero => a data character
770                   negative => a special escape sequence                   positive => a special escape sequence
771                     negative => a back reference
772                   on error, errorcodeptr is set                   on error, errorcodeptr is set
773  */  */
774    
775  static int  static int
776  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
777    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
778  {  {
779  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
780  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
781  const pcre_uchar *ptr = *ptrptr + 1;  const pcre_uchar *ptr = *ptrptr + 1;
782  pcre_int32 c;  pcre_uint32 c;
783    int escape = 0;
784  int i;  int i;
785    
786  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
# Line 782  Otherwise further processing may be requ Line 797  Otherwise further processing may be requ
797  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
798  /* Not alphanumeric */  /* Not alphanumeric */
799  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
800  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) { 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 795  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 820  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 829  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 855  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 874  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 890  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 914  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 922  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 952  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 1018  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 1036  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 1049  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 1080  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 1139  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 1169  escape sequence. Line 1203  escape sequence.
1203  Argument:  Argument:
1204    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1205    negptr         points to a boolean that is set TRUE for negation else FALSE    negptr         points to a boolean that is set TRUE for negation else FALSE
1206    dptr           points to an int that is set to the detailed property value    ptypeptr       points to an unsigned int that is set to the type value
1207      pdataptr       points to an unsigned int that is set to the detailed property value
1208    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1209    
1210  Returns:         type value from ucp_type_table, or -1 for an invalid type  Returns:         TRUE if the type value was found, or FALSE for an invalid type
1211  */  */
1212    
1213  static int  static BOOL
1214  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1215      unsigned int *pdataptr, int *errorcodeptr)
1216  {  {
1217  int c, i, bot, top;  pcre_uchar c;
1218    int i, bot, top;
1219  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1220  pcre_uchar name[32];  pcre_uchar name[32];
1221    
# Line 1225  top = PRIV(utt_size); Line 1262  top = PRIV(utt_size);
1262    
1263  while (bot < top)  while (bot < top)
1264    {    {
1265      int r;
1266    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1267    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1268    if (c == 0)    if (r == 0)
1269      {      {
1270      *dptr = PRIV(utt)[i].value;      *ptypeptr = PRIV(utt)[i].type;
1271      return PRIV(utt)[i].type;      *pdataptr = PRIV(utt)[i].value;
1272        return TRUE;
1273      }      }
1274    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1275    }    }
1276    
1277  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1278  *ptrptr = ptr;  *ptrptr = ptr;
1279  return -1;  return FALSE;
1280    
1281  ERROR_RETURN:  ERROR_RETURN:
1282  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1283  *ptrptr = ptr;  *ptrptr = ptr;
1284  return -1;  return FALSE;
1285  }  }
1286  #endif  #endif
1287    
# Line 1277  int max = -1; Line 1316  int max = -1;
1316  /* Read the minimum value and do a paranoid check: a negative value indicates  /* Read the minimum value and do a paranoid check: a negative value indicates
1317  an integer overflow. */  an integer overflow. */
1318    
1319  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1320  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1321    {    {
1322    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1292  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1331  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1331    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1332      {      {
1333      max = 0;      max = 0;
1334      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1335      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1336        {        {
1337        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1347  Arguments: Line 1386  Arguments:
1386    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1387    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1388    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1389    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1390    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1391    
1392  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
# Line 1423  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1462  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1462      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1463          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1464        {        {
1465        int term;        pcre_uchar term;
1466        const pcre_uchar *thisname;        const pcre_uchar *thisname;
1467        *count += 1;        *count += 1;
1468        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
# Line 1431  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1470  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1470        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1471        thisname = ptr;        thisname = ptr;
1472        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1473        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1474            STRNCMP_UC_UC(name, thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1475          return *count;          return *count;
1476        term++;        term++;
1477        }        }
# Line 1581  Arguments: Line 1620  Arguments:
1620    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1621    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1622    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1623    utf          TRUE if we are in UTF-8 / UTF-16 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1624    
1625  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1626  */  */
# Line 1684  and doing the check at the end; a flag s Line 1723  and doing the check at the end; a flag s
1723    
1724  Arguments:  Arguments:
1725    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1726    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1727    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1728    cd       the "compile data" structure    cd       the "compile data" structure
1729    
# Line 1710  for (;;) Line 1749  for (;;)
1749    {    {
1750    int d;    int d;
1751    pcre_uchar *ce, *cs;    pcre_uchar *ce, *cs;
1752    register int op = *cc;    register pcre_uchar op = *cc;
1753    
1754    switch (op)    switch (op)
1755      {      {
# Line 1830  for (;;) Line 1869  for (;;)
1869      case OP_EXACTI:      case OP_EXACTI:
1870      case OP_NOTEXACT:      case OP_NOTEXACT:
1871      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1872      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1873      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1874  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
1875      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
# Line 1839  for (;;) Line 1878  for (;;)
1878    
1879      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1880      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1881      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)
1882          cc += 2;
1883      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1884      break;      break;
1885    
# Line 1874  for (;;) Line 1914  for (;;)
1914    
1915      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1916    
1917  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1918      case OP_XCLASS:      case OP_XCLASS:
     cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];  
     /* Fall through */  
1919  #endif  #endif
   
1920      case OP_CLASS:      case OP_CLASS:
1921      case OP_NCLASS:      case OP_NCLASS:
1922      cc += PRIV(OP_lengths)[OP_CLASS];  
1923        switch (op)
1924          {
1925    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1926          case OP_XCLASS:
1927          cc += GET(cc, 1);
1928          break;
1929    #endif
1930    
1931          case OP_CLASS:
1932          case OP_NCLASS:
1933          cc += PRIV(OP_lengths)[OP_CLASS];
1934          break;
1935          }
1936    
1937      switch (*cc)      switch (*cc)
1938        {        {
# Line 1897  for (;;) Line 1947  for (;;)
1947        case OP_CRRANGE:        case OP_CRRANGE:
1948        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1949        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1950        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1951        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
1952        break;        break;
1953    
# Line 2013  length. Line 2063  length.
2063    
2064  Arguments:  Arguments:
2065    code        points to start of expression    code        points to start of expression
2066    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2067    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2068    
2069  Returns:      pointer to the opcode for the bracket, or NULL if not found  Returns:      pointer to the opcode for the bracket, or NULL if not found
# Line 2024  PRIV(find_bracket)(const pcre_uchar *cod Line 2074  PRIV(find_bracket)(const pcre_uchar *cod
2074  {  {
2075  for (;;)  for (;;)
2076    {    {
2077    register int c = *code;    register pcre_uchar c = *code;
2078    
2079    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2080    
# Line 2047  for (;;) Line 2097  for (;;)
2097    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2098             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2099      {      {
2100      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2101      if (n == number) return (pcre_uchar *)code;      if (n == number) return (pcre_uchar *)code;
2102      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
2103      }      }
# Line 2077  for (;;) Line 2127  for (;;)
2127        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2128        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2129        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2130        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2131          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2132        break;        break;
2133    
2134        case OP_MARK:        case OP_MARK:
# Line 2100  for (;;) Line 2150  for (;;)
2150    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
2151    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2152    
2153  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2154      if (utf) switch(c)      if (utf) switch(c)
2155        {        {
2156        case OP_CHAR:        case OP_CHAR:
# Line 2152  instance of OP_RECURSE. Line 2202  instance of OP_RECURSE.
2202    
2203  Arguments:  Arguments:
2204    code        points to start of expression    code        points to start of expression
2205    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2206    
2207  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
2208  */  */
# Line 2162  find_recurse(const pcre_uchar *code, BOO Line 2212  find_recurse(const pcre_uchar *code, BOO
2212  {  {
2213  for (;;)  for (;;)
2214    {    {
2215    register int c = *code;    register pcre_uchar c = *code;
2216    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2217    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2218    
# Line 2197  for (;;) Line 2247  for (;;)
2247        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2248        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2249        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2250        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2251          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2252        break;        break;
2253    
2254        case OP_MARK:        case OP_MARK:
# Line 2220  for (;;) Line 2270  for (;;)
2270      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
2271      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2272    
2273  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2274      if (utf) switch(c)      if (utf) switch(c)
2275        {        {
2276        case OP_CHAR:        case OP_CHAR:
2277        case OP_CHARI:        case OP_CHARI:
2278          case OP_NOT:
2279          case OP_NOTI:
2280        case OP_EXACT:        case OP_EXACT:
2281        case OP_EXACTI:        case OP_EXACTI:
2282          case OP_NOTEXACT:
2283          case OP_NOTEXACTI:
2284        case OP_UPTO:        case OP_UPTO:
2285        case OP_UPTOI:        case OP_UPTOI:
2286          case OP_NOTUPTO:
2287          case OP_NOTUPTOI:
2288        case OP_MINUPTO:        case OP_MINUPTO:
2289        case OP_MINUPTOI:        case OP_MINUPTOI:
2290          case OP_NOTMINUPTO:
2291          case OP_NOTMINUPTOI:
2292        case OP_POSUPTO:        case OP_POSUPTO:
2293        case OP_POSUPTOI:        case OP_POSUPTOI:
2294          case OP_NOTPOSUPTO:
2295          case OP_NOTPOSUPTOI:
2296        case OP_STAR:        case OP_STAR:
2297        case OP_STARI:        case OP_STARI:
2298          case OP_NOTSTAR:
2299          case OP_NOTSTARI:
2300        case OP_MINSTAR:        case OP_MINSTAR:
2301        case OP_MINSTARI:        case OP_MINSTARI:
2302          case OP_NOTMINSTAR:
2303          case OP_NOTMINSTARI:
2304        case OP_POSSTAR:        case OP_POSSTAR:
2305        case OP_POSSTARI:        case OP_POSSTARI:
2306          case OP_NOTPOSSTAR:
2307          case OP_NOTPOSSTARI:
2308        case OP_PLUS:        case OP_PLUS:
2309        case OP_PLUSI:        case OP_PLUSI:
2310          case OP_NOTPLUS:
2311          case OP_NOTPLUSI:
2312        case OP_MINPLUS:        case OP_MINPLUS:
2313        case OP_MINPLUSI:        case OP_MINPLUSI:
2314          case OP_NOTMINPLUS:
2315          case OP_NOTMINPLUSI:
2316        case OP_POSPLUS:        case OP_POSPLUS:
2317        case OP_POSPLUSI:        case OP_POSPLUSI:
2318          case OP_NOTPOSPLUS:
2319          case OP_NOTPOSPLUSI:
2320        case OP_QUERY:        case OP_QUERY:
2321        case OP_QUERYI:        case OP_QUERYI:
2322          case OP_NOTQUERY:
2323          case OP_NOTQUERYI:
2324        case OP_MINQUERY:        case OP_MINQUERY:
2325        case OP_MINQUERYI:        case OP_MINQUERYI:
2326          case OP_NOTMINQUERY:
2327          case OP_NOTMINQUERYI:
2328        case OP_POSQUERY:        case OP_POSQUERY:
2329        case OP_POSQUERYI:        case OP_POSQUERYI:
2330          case OP_NOTPOSQUERY:
2331          case OP_NOTPOSQUERYI:
2332        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2333        break;        break;
2334        }        }
# Line 2278  bracket whose current branch will alread Line 2356  bracket whose current branch will alread
2356  Arguments:  Arguments:
2357    code        points to start of search    code        points to start of search
2358    endcode     points to where to stop    endcode     points to where to stop
2359    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2360    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2361    
2362  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2288  static BOOL Line 2366  static BOOL
2366  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2367    BOOL utf, compile_data *cd)    BOOL utf, compile_data *cd)
2368  {  {
2369  register int c;  register pcre_uchar c;
2370  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2371       code < endcode;       code < endcode;
2372       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
# Line 2322  for (code = first_significant_code(code Line 2400  for (code = first_significant_code(code
2400      /* Test for forward reference */      /* Test for forward reference */
2401    
2402      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2403        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2404    
2405      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2406    
# Line 2495  for (code = first_significant_code(code Line 2573  for (code = first_significant_code(code
2573      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2574      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2575      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2576      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2577        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2578      break;      break;
2579    
2580      /* End of branch */      /* End of branch */
# Line 2511  for (code = first_significant_code(code Line 2589  for (code = first_significant_code(code
2589      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2590      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2591    
2592  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2593      case OP_STAR:      case OP_STAR:
2594      case OP_STARI:      case OP_STARI:
2595      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2577  Arguments: Line 2655  Arguments:
2655    code        points to start of the recursion    code        points to start of the recursion
2656    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2657    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2658    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2659    cd          pointers to tables etc    cd          pointers to tables etc
2660    
2661  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
# Line 2643  Returns:   TRUE or FALSE Line 2721  Returns:   TRUE or FALSE
2721  static BOOL  static BOOL
2722  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2723  {  {
2724  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2725  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2726  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2727    {    {
# Line 2692  register int yield = 0; Line 2770  register int yield = 0;
2770  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2771    {    {
2772    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2773      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2774    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2775    yield++;    yield++;
2776    }    }
# Line 2724  value in the reference (which is a group Line 2802  value in the reference (which is a group
2802  Arguments:  Arguments:
2803    group      points to the start of the group    group      points to the start of the group
2804    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2805    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2806    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2807    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
2808    
# Line 2747  while ((ptr = (pcre_uchar *)find_recurse Line 2825  while ((ptr = (pcre_uchar *)find_recurse
2825    
2826    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2827      {      {
2828      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2829      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2830        {        {
2831        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2760  while ((ptr = (pcre_uchar *)find_recurse Line 2838  while ((ptr = (pcre_uchar *)find_recurse
2838    
2839    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2840      {      {
2841      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2842      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2843      }      }
2844    
# Line 2828  PUT(previous_callout, 2 + LINK_SIZE, len Line 2906  PUT(previous_callout, 2 + LINK_SIZE, len
2906  *************************************************/  *************************************************/
2907    
2908  /* 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
2909  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
2910  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
2911  start address.  start address. A character with multiple other cases is returned on its own
2912    with a special return value.
2913    
2914  Arguments:  Arguments:
2915    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2838  Arguments: Line 2917  Arguments:
2917    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2918    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2919    
2920  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2921                   0 when a range is returned
2922                  >0 the CASESET offset for char with multiple other cases
2923                    in this case, ocptr contains the original
2924  */  */
2925    
2926  static BOOL  static int
2927  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2928    unsigned int *odptr)    pcre_uint32 *odptr)
2929  {  {
2930  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2931    unsigned int co;
2932    
2933    /* Find the first character that has an other case. If it has multiple other
2934    cases, return its case offset value. */
2935    
2936  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2937    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2938      if ((co = UCD_CASESET(c)) != 0)
2939        {
2940        *ocptr = c++;   /* Character that has the set */
2941        *cptr = c;      /* Rest of input range */
2942        return (int)co;
2943        }
2944      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2945      }
2946    
2947  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2948    
2949  *ocptr = othercase;  *ocptr = othercase;
2950  next = othercase + 1;  next = othercase + 1;
# Line 2861  for (++c; c <= d; c++) Line 2955  for (++c; c <= d; c++)
2955    next++;    next++;
2956    }    }
2957    
2958  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2959  *cptr = c;  *cptr = c;             /* Rest of input range */
2960    return 0;
 return TRUE;  
2961  }  }
2962    
2963    
# Line 2886  Returns:       TRUE if auto-possessifyin Line 2979  Returns:       TRUE if auto-possessifyin
2979  */  */
2980    
2981  static BOOL  static BOOL
2982  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata, BOOL negated)
2983  {  {
2984    #ifdef SUPPORT_UCP
2985    const pcre_uint32 *p;
2986    #endif
2987    
2988  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2989    
2990  switch(ptype)  switch(ptype)
2991    {    {
2992    case PT_LAMP:    case PT_LAMP:
# Line 2926  switch(ptype) Line 3024  switch(ptype)
3024    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3025            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3026            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3027    
3028    #ifdef SUPPORT_UCP
3029      case PT_CLIST:
3030      p = PRIV(ucd_caseless_sets) + prop->caseset;
3031      for (;;)
3032        {
3033        if ((unsigned int)c < *p) return !negated;
3034        if ((unsigned int)c == *p++) return negated;
3035        }
3036      break;  /* Control never reaches here */
3037    #endif
3038    }    }
3039    
3040  return FALSE;  return FALSE;
3041  }  }
3042  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2943  sense to automatically possessify the re Line 3053  sense to automatically possessify the re
3053    
3054  Arguments:  Arguments:
3055    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3056    utf           TRUE in UTF-8 / UTF-16 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3057    ptr           next character in pattern    ptr           next character in pattern
3058    options       options bits    options       options bits
3059    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2955  static BOOL Line 3065  static BOOL
3065  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3066    const pcre_uchar *ptr, int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3067  {  {
3068  pcre_int32 c, next;  pcre_uint32 c = NOTACHAR;
3069  int op_code = *previous++;  pcre_uint32 next;
3070    int escape;
3071    pcre_uchar op_code = *previous++;
3072    
3073  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3074    
# Line 2987  value is a character, a negative value i Line 3099  value is a character, a negative value i
3099  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3100    {    {
3101    int temperrorcode = 0;    int temperrorcode = 0;
3102    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3103    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3104    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3105    }    }
3106  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
3107    {    {
3108      escape = 0;
3109  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3110    if (utf) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3111  #endif  #endif
# Line 3029  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 3142  if (*ptr == CHAR_ASTERISK || *ptr == CHA
3142    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)
3143      return FALSE;      return FALSE;
3144    
3145  /* 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. */  
3146    
3147  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3148        op_code == OP_NOT || op_code == OP_NOTI)
3149    {    {
   case OP_CHAR:  
3150  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3151    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3152  #else  #else
3153    c = *previous;    c = *previous;
3154  #endif  #endif
3155    return c != next;    }
3156    
3157    /* 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
3158    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3159    
3160    case OP_CHARI:  if (escape == 0)
3161  #ifdef SUPPORT_UTF    {
3162    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3163  #else    case, which maps to the special PT_CLIST property. Check this first. */
3164    c = *previous;  
3165    #ifdef SUPPORT_UCP
3166      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3167        {
3168        unsigned int ocs = UCD_CASESET(next);
3169        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3170        }
3171  #endif  #endif
3172    if (c == next) return FALSE;  
3173  #ifdef SUPPORT_UTF    switch(op_code)
   if (utf)  
3174      {      {
3175      unsigned int othercase;      case OP_CHAR:
3176      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3177    
3178        /* For CHARI (caseless character) we must check the other case. If we have
3179        Unicode property support, we can use it to test the other case of
3180        high-valued characters. We know that next can have only one other case,
3181        because multi-other-case characters are dealt with above. */
3182    
3183        case OP_CHARI:
3184        if (c == next) return FALSE;
3185    #ifdef SUPPORT_UTF
3186        if (utf)
3187          {
3188          pcre_uint32 othercase;
3189          if (next < 128) othercase = cd->fcc[next]; else
3190  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3191      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3192  #else  #else
3193      othercase = NOTACHAR;        othercase = NOTACHAR;
3194  #endif  #endif
3195      return (unsigned int)c != othercase;        return c != othercase;
3196      }        }
3197    else      else
3198  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3199    return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3200    
3201    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3202    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3203    an XCLASS instead. */  
3204        case OP_NOTI:
3205    case OP_NOT:      if (c == next) return TRUE;
   return (c = *previous) == next;  
   
   case OP_NOTI:  
   if ((c = *previous) == next) return TRUE;  
3206  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3207    if (utf)      if (utf)
3208      {        {
3209      unsigned int othercase;        pcre_uint32 othercase;
3210      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3211  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3212      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3213  #else  #else
3214      othercase = NOTACHAR;        othercase = NOTACHAR;
3215  #endif  #endif
3216      return (unsigned int)c == othercase;        return c == othercase;
3217      }        }
3218    else      else
3219  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
3220    return (c == (int)(TABLE_GET((unsigned int)next, cd->fcc, next)));  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3221    
3222    /* 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.
3223    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. */
3224    
3225    case OP_DIGIT:      case OP_DIGIT:
3226    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3227    
3228    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3229    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3230    
3231    case OP_WHITESPACE:      case OP_WHITESPACE:
3232    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3233    
3234    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3235    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3236    
3237    case OP_WORDCHAR:      case OP_WORDCHAR:
3238    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3239    
3240    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3241    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3242    
3243    case OP_HSPACE:      case OP_HSPACE:
3244    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3245    switch(next)      switch(next)
3246      {        {
3247      case 0x09:        HSPACE_CASES:
3248      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;  
     }  
3249    
3250    case OP_ANYNL:        default:
3251    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3252    case OP_NOT_VSPACE:        }
3253    switch(next)  
3254      {      case OP_ANYNL:
3255      case 0x0a:      case OP_VSPACE:
3256      case 0x0b:      case OP_NOT_VSPACE:
3257      case 0x0c:      switch(next)
3258      case 0x0d:        {
3259      case 0x85:        VSPACE_CASES:
3260      case 0x2028:        return op_code == OP_NOT_VSPACE;
3261      case 0x2029:  
3262      return op_code == OP_NOT_VSPACE;        default:
3263      default:        return op_code != OP_NOT_VSPACE;
3264      return op_code != OP_NOT_VSPACE;        }
     }  
3265    
3266  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3267    case OP_PROP:      case OP_PROP:
3268    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3269    
3270    case OP_NOTPROP:      case OP_NOTPROP:
3271    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3272  #endif  #endif
3273    
3274    default:      default:
3275    return FALSE;      return FALSE;
3276        }
3277    }    }
3278    
   
3279  /* 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
3280  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
3281  generated only when PCRE_UCP is *not* set, that is, when only ASCII  generated only when PCRE_UCP is *not* set, that is, when only ASCII
# Line 3183  switch(op_code) Line 3286  switch(op_code)
3286    {    {
3287    case OP_CHAR:    case OP_CHAR:
3288    case OP_CHARI:    case OP_CHARI:
3289  #ifdef SUPPORT_UTF    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3290      {      {
3291      case ESC_d:      case ESC_d:
3292      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3293    
3294      case ESC_D:      case ESC_D:
3295      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3296    
3297      case ESC_s:      case ESC_s:
3298      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3299    
3300      case ESC_S:      case ESC_S:
3301      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3302    
3303      case ESC_w:      case ESC_w:
3304      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3305    
3306      case ESC_W:      case ESC_W:
3307      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3308    
3309      case ESC_h:      case ESC_h:
3310      case ESC_H:      case ESC_H:
3311      switch(c)      switch(c)
3312        {        {
3313        case 0x09:        HSPACE_CASES:
3314        case 0x20:        return escape != ESC_h;
3315        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;  
3316        default:        default:
3317        return -next == ESC_h;        return escape == ESC_h;
3318        }        }
3319    
3320      case ESC_v:      case ESC_v:
3321      case ESC_V:      case ESC_V:
3322      switch(c)      switch(c)
3323        {        {
3324        case 0x0a:        VSPACE_CASES:
3325        case 0x0b:        return escape != ESC_v;
3326        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3327        default:        default:
3328        return -next == ESC_v;        return escape == ESC_v;
3329        }        }
3330    
3331      /* 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
3332      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3333      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3334    
3335  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3336      case ESC_du:      case ESC_du:
# Line 3265  switch(op_code) Line 3341  switch(op_code)
3341      case ESC_SU:      case ESC_SU:
3342        {        {
3343        int temperrorcode = 0;        int temperrorcode = 0;
3344        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3345        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3346        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3347        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3348        }        }
# Line 3275  switch(op_code) Line 3351  switch(op_code)
3351      case ESC_p:      case ESC_p:
3352      case ESC_P:      case ESC_P:
3353        {        {
3354        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3355          int errorcodeptr;
3356        BOOL negated;        BOOL negated;
3357    
3358        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3359        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3360        if (ptype < 0) return FALSE;          return FALSE;
3361        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3362    
3363        /* If the property item is optional, we have to give up. (When generated        /* If the property item is optional, we have to give up. (When generated
# Line 3293  switch(op_code) Line 3370  switch(op_code)
3370    
3371        /* Do the property check. */        /* Do the property check. */
3372    
3373        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3374        }        }
3375  #endif  #endif
3376    
# Line 3308  switch(op_code) Line 3385  switch(op_code)
3385    these op-codes are never generated.) */    these op-codes are never generated.) */
3386    
3387    case OP_DIGIT:    case OP_DIGIT:
3388    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3389           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3390    
3391    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3392    return next == -ESC_d;    return escape == ESC_d;
3393    
3394    case OP_WHITESPACE:    case OP_WHITESPACE:
3395    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3396    
3397    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3398    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3399    
3400    case OP_HSPACE:    case OP_HSPACE:
3401    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3402           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3403    
3404    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3405    return next == -ESC_h;    return escape == ESC_h;
3406    
3407    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3408    case OP_ANYNL:    case OP_ANYNL:
3409    case OP_VSPACE:    case OP_VSPACE:
3410    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3411    
3412    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3413    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3414    
3415    case OP_WORDCHAR:    case OP_WORDCHAR:
3416    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3417           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3418    
3419    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3420    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3421    
3422    default:    default:
3423    return FALSE;    return FALSE;
# Line 3352  switch(op_code) Line 3429  switch(op_code)
3429    
3430    
3431  /*************************************************  /*************************************************
3432    *        Add a character or range to a class     *
3433    *************************************************/
3434    
3435    /* This function packages up the logic of adding a character or range of
3436    characters to a class. The character values in the arguments will be within the
3437    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3438    mutually recursive with the function immediately below.
3439    
3440    Arguments:
3441      classbits     the bit map for characters < 256
3442      uchardptr     points to the pointer for extra data
3443      options       the options word
3444      cd            contains pointers to tables etc.
3445      start         start of range character
3446      end           end of range character
3447    
3448    Returns:        the number of < 256 characters added
3449                    the pointer to extra data is updated
3450    */
3451    
3452    static int
3453    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3454      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3455    {
3456    pcre_uint32 c;
3457    int n8 = 0;
3458    
3459    /* If caseless matching is required, scan the range and process alternate
3460    cases. In Unicode, there are 8-bit characters that have alternate cases that
3461    are greater than 255 and vice-versa. Sometimes we can just extend the original
3462    range. */
3463    
3464    if ((options & PCRE_CASELESS) != 0)
3465      {
3466    #ifdef SUPPORT_UCP
3467      if ((options & PCRE_UTF8) != 0)
3468        {
3469        int rc;
3470        pcre_uint32 oc, od;
3471    
3472        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3473        c = start;
3474    
3475        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3476          {
3477          /* Handle a single character that has more than one other case. */
3478    
3479          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3480            PRIV(ucd_caseless_sets) + rc, oc);
3481    
3482          /* Do nothing if the other case range is within the original range. */
3483    
3484          else if (oc >= start && od <= end) continue;
3485    
3486          /* Extend the original range if there is overlap, noting that if oc < c, we
3487          can't have od > end because a subrange is always shorter than the basic
3488          range. Otherwise, use a recursive call to add the additional range. */
3489    
3490          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3491          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3492          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3493          }
3494        }
3495      else
3496    #endif  /* SUPPORT_UCP */
3497    
3498      /* Not UTF-mode, or no UCP */
3499    
3500      for (c = start; c <= end && c < 256; c++)
3501        {
3502        SETBIT(classbits, cd->fcc[c]);
3503        n8++;
3504        }
3505      }
3506    
3507    /* Now handle the original range. Adjust the final value according to the bit
3508    length - this means that the same lists of (e.g.) horizontal spaces can be used
3509    in all cases. */
3510    
3511    #if defined COMPILE_PCRE8
3512    #ifdef SUPPORT_UTF
3513      if ((options & PCRE_UTF8) == 0)
3514    #endif
3515      if (end > 0xff) end = 0xff;
3516    
3517    #elif defined COMPILE_PCRE16
3518    #ifdef SUPPORT_UTF
3519      if ((options & PCRE_UTF16) == 0)
3520    #endif
3521      if (end > 0xffff) end = 0xffff;
3522    
3523    #endif /* COMPILE_PCRE[8|16] */
3524    
3525    /* If all characters are less than 256, use the bit map. Otherwise use extra
3526    data. */
3527    
3528    if (end < 0x100)
3529      {
3530      for (c = start; c <= end; c++)
3531        {
3532        n8++;
3533        SETBIT(classbits, c);
3534        }
3535      }
3536    
3537    else
3538      {
3539      pcre_uchar *uchardata = *uchardptr;
3540    
3541    #ifdef SUPPORT_UTF
3542      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3543        {
3544        if (start < end)
3545          {
3546          *uchardata++ = XCL_RANGE;
3547          uchardata += PRIV(ord2utf)(start, uchardata);
3548          uchardata += PRIV(ord2utf)(end, uchardata);
3549          }
3550        else if (start == end)
3551          {
3552          *uchardata++ = XCL_SINGLE;
3553          uchardata += PRIV(ord2utf)(start, uchardata);
3554          }
3555        }
3556      else
3557    #endif  /* SUPPORT_UTF */
3558    
3559      /* Without UTF support, character values are constrained by the bit length,
3560      and can only be > 256 for 16-bit and 32-bit libraries. */
3561    
3562    #ifdef COMPILE_PCRE8
3563        {}
3564    #else
3565      if (start < end)
3566        {
3567        *uchardata++ = XCL_RANGE;
3568        *uchardata++ = start;
3569        *uchardata++ = end;
3570        }
3571      else if (start == end)
3572        {
3573        *uchardata++ = XCL_SINGLE;
3574        *uchardata++ = start;
3575        }
3576    #endif
3577    
3578      *uchardptr = uchardata;   /* Updata extra data pointer */
3579      }
3580    
3581    return n8;    /* Number of 8-bit characters */
3582    }
3583    
3584    
3585    
3586    
3587    /*************************************************
3588    *        Add a list of characters to a class     *
3589    *************************************************/
3590    
3591    /* This function is used for adding a list of case-equivalent characters to a
3592    class, and also for adding a list of horizontal or vertical whitespace. If the
3593    list is in order (which it should be), ranges of characters are detected and
3594    handled appropriately. This function is mutually recursive with the function
3595    above.
3596    
3597    Arguments:
3598      classbits     the bit map for characters < 256
3599      uchardptr     points to the pointer for extra data
3600      options       the options word
3601      cd            contains pointers to tables etc.
3602      p             points to row of 32-bit values, terminated by NOTACHAR
3603      except        character to omit; this is used when adding lists of
3604                      case-equivalent characters to avoid including the one we
3605                      already know about
3606    
3607    Returns:        the number of < 256 characters added
3608                    the pointer to extra data is updated
3609    */
3610    
3611    static int
3612    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3613      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3614    {
3615    int n8 = 0;
3616    while (p[0] < NOTACHAR)
3617      {
3618      int n = 0;
3619      if (p[0] != except)
3620        {
3621        while(p[n+1] == p[0] + n + 1) n++;
3622        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3623        }
3624      p += n + 1;
3625      }
3626    return n8;
3627    }
3628    
3629    
3630    
3631    /*************************************************
3632    *    Add characters not in a list to a class     *
3633    *************************************************/
3634    
3635    /* This function is used for adding the complement of a list of horizontal or
3636    vertical whitespace to a class. The list must be in order.
3637    
3638    Arguments:
3639      classbits     the bit map for characters < 256
3640      uchardptr     points to the pointer for extra data
3641      options       the options word
3642      cd            contains pointers to tables etc.
3643      p             points to row of 32-bit values, terminated by NOTACHAR
3644    
3645    Returns:        the number of < 256 characters added
3646                    the pointer to extra data is updated
3647    */
3648    
3649    static int
3650    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3651      int options, compile_data *cd, const pcre_uint32 *p)
3652    {
3653    BOOL utf = (options & PCRE_UTF8) != 0;
3654    int n8 = 0;
3655    if (p[0] > 0)
3656      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3657    while (p[0] < NOTACHAR)
3658      {
3659      while (p[1] == p[0] + 1) p++;
3660      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3661        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3662      p++;
3663      }
3664    return n8;
3665    }
3666    
3667    
3668    
3669    /*************************************************
3670  *           Compile one branch                   *  *           Compile one branch                   *
3671  *************************************************/  *************************************************/
3672    
# Line 3366  Arguments: Line 3681  Arguments:
3681    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3682    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3683    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3684    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3685    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3686      reqcharptr     place to put the last required character
3687      reqcharflagsptr place to put the last required character flags, or a negative number
3688    bcptr          points to current branch chain    bcptr          points to current branch chain
3689    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3690    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3380  Returns:         TRUE on success Line 3697  Returns:         TRUE on success
3697    
3698  static BOOL  static BOOL
3699  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3700    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3701    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3702      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3703      branch_chain *bcptr, int cond_depth,
3704    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3705  {  {
3706  int repeat_type, op_type;  int repeat_type, op_type;
3707  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3708  int bravalue = 0;  int bravalue = 0;
3709  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3710  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
3711  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
3712    pcre_uint32 zeroreqchar, zerofirstchar;
3713    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3714  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3715  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3716  int after_manual_callout = 0;  int after_manual_callout = 0;
3717  int length_prevgroup = 0;  int length_prevgroup = 0;
3718  register int c;  register pcre_uint32 c;
3719    int escape;
3720  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3721  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
3722  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3414  must not do this for other options (e.g. Line 3736  must not do this for other options (e.g.
3736  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3737    
3738  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3739  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3740  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3741    #ifndef COMPILE_PCRE32
3742  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
3743    #endif
3744  #else  #else
3745  BOOL utf = FALSE;  BOOL utf = FALSE;
3746  #endif  #endif
3747    
3748  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3749    class_uchardata always so that it can be passed to add_to_class() always,
3750    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3751    alternative calls for the different cases. */
3752    
3753    pcre_uchar *class_uchardata;
3754  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3755  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
3756  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
3757  #endif  #endif
3758    
# Line 3448  to take the zero repeat into account. Th Line 3775  to take the zero repeat into account. Th
3775  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3776  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3777    
3778  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3779    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3780    
3781  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
3782  or zero, according to the current setting of the caseless flag. The  or zero, according to the current setting of the caseless flag. The
# Line 3469  for (;; ptr++) Line 3797  for (;; ptr++)
3797    BOOL is_recurse;    BOOL is_recurse;
3798    BOOL reset_bracount;    BOOL reset_bracount;
3799    int class_has_8bitchar;    int class_has_8bitchar;
3800    int class_single_char;    int class_one_char;
3801    int newoptions;    int newoptions;
3802    int recno;    int recno;
3803    int refsign;    int refsign;
3804    int skipbytes;    int skipbytes;
3805    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
3806    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
3807    int terminator;    int terminator;
3808    int mclength;    unsigned int mclength;
3809    int tempbracount;    unsigned int tempbracount;
3810      pcre_uint32 ec;
3811    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
3812    
3813    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3641  for (;; ptr++) Line 3970  for (;; ptr++)
3970      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3971      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3972      *firstcharptr = firstchar;      *firstcharptr = firstchar;
3973        *firstcharflagsptr = firstcharflags;
3974      *reqcharptr = reqchar;      *reqcharptr = reqchar;
3975        *reqcharflagsptr = reqcharflags;
3976      *codeptr = code;      *codeptr = code;
3977      *ptrptr = ptr;      *ptrptr = ptr;
3978      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3665  for (;; ptr++) Line 3996  for (;; ptr++)
3996      previous = NULL;      previous = NULL;
3997      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3998        {        {
3999        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4000        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
4001        }        }
4002      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3680  for (;; ptr++) Line 4011  for (;; ptr++)
4011      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4012    
4013      case CHAR_DOT:      case CHAR_DOT:
4014      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4015      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4016        zerofirstcharflags = firstcharflags;
4017      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4018        zeroreqcharflags = reqcharflags;
4019      previous = code;      previous = code;
4020      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4021      break;      break;
# Line 3756  for (;; ptr++) Line 4089  for (;; ptr++)
4089          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4090        {        {
4091        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4092        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4093        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4094          zerofirstcharflags = firstcharflags;
4095        break;        break;
4096        }        }
4097    
# Line 3767  for (;; ptr++) Line 4101  for (;; ptr++)
4101    
4102      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4103    
4104      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4105      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 <
4106      < 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
4107      a single character. */      character. */
4108    
4109      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4110      class_single_char = 0;      class_one_char = 0;
4111    
4112      /* 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
4113      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
4114      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
4115      */      map. */
4116    
4117      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4118    
4119  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4120      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4121      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4122      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4123  #endif  #endif
4124    
4125      /* 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 3807  for (;; ptr++) Line 4141  for (;; ptr++)
4141        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4142        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
4143        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4144        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4145          however. */
4146    
4147        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4148          {          {
4149            xclass = TRUE;
4150          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4151          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4152          }          }
# Line 3869  for (;; ptr++) Line 4205  for (;; ptr++)
4205          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4206          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4207    
4208          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4209            posix_class = 0;            posix_class = 0;
4210    
4211          /* 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 3912  for (;; ptr++) Line 4248  for (;; ptr++)
4248              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4249            }            }
4250    
4251          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4252          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4253    
4254          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3928  for (;; ptr++) Line 4264  for (;; ptr++)
4264            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4265    
4266          ptr = tempptr + 1;          ptr = tempptr + 1;
4267          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4268          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4269          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4270          class_single_char = 2;          class_one_char = 2;
4271          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4272          }          }
4273    
# Line 3939  for (;; ptr++) Line 4275  for (;; ptr++)
4275        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
4276        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
4277        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4278        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4279        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4280        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4281        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4282    
4283        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4284          {          {
4285          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4286    
4287          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4288    
4289          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4290          else if (-c == ESC_N)            /* \N is not supported in a class */            c = ec;
4291            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4292            else if (escape == ESC_N)            /* \N is not supported in a class */
4293            {            {
4294            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4295            goto FAILED;            goto FAILED;
4296            }            }
4297          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4298            {            {
4299            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4300              {              {
# Line 3964  for (;; ptr++) Line 4303  for (;; ptr++)
4303            else inescq = TRUE;            else inescq = TRUE;
4304            continue;            continue;
4305            }            }
4306          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4307    
4308          if (c < 0)          else
4309            {            {
4310            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4311            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4312            class_has_8bitchar++;            class_has_8bitchar++;
4313            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4314            class_single_char += 2;            class_one_char += 2;
4315    
4316            switch (-c)            switch (escape)
4317              {              {
4318  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4319              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3984  for (;; ptr++) Line 4323  for (;; ptr++)
4323              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4324              case ESC_SU:              case ESC_SU:
4325              nestptr = ptr;              nestptr = ptr;
4326              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4327              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4328              continue;              continue;
4329  #endif  #endif
# Line 4008  for (;; ptr++) Line 4347  for (;; ptr++)
4347    
4348              /* 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
4349              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4350              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4351                EBCDIC, so we lazily just adjust the appropriate bit. */
4352    
4353              case ESC_s:              case ESC_s:
4354              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4021  for (;; ptr++) Line 4361  for (;; ptr++)
4361              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4362              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4363              continue;              continue;
4364    
4365                /* The rest apply in both UCP and non-UCP cases. */
4366    
4367              case ESC_h:              case ESC_h:
4368              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4369              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  
4370              continue;              continue;
4371    
4372              case ESC_H:              case ESC_H:
4373              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4374                {                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  
4375              continue;              continue;
4376    
4377              case ESC_v:              case ESC_v:
4378              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4379              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  
4380              continue;              continue;
4381    
4382              case ESC_V:              case ESC_V:
4383              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4384                {                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  
4385              continue;              continue;
4386    
4387  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4203  for (;; ptr++) Line 4389  for (;; ptr++)
4389              case ESC_P:              case ESC_P:
4390                {                {
4391                BOOL negated;                BOOL negated;
4392                int pdata;                unsigned int ptype = 0, pdata = 0;
4393                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4394                if (ptype < 0) goto FAILED;                  goto FAILED;
4395                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4396                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4397                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4398                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4226  for (;; ptr++) Line 4411  for (;; ptr++)
4411                goto FAILED;                goto FAILED;
4412                }                }
4413              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4414              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4415              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4416              break;              break;
4417              }              }
4418            }            }
4419    
4420          /* 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).
4421          greater than 256. */          This may be greater than 256. */
4422    
4423            escape = 0;
4424    
4425          }   /* End of backslash handling */          }   /* End of backslash handling */
4426    
4427        /* 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
4428        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
4429        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4430        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4431    
4432        CHECK_RANGE:        CHECK_RANGE:
4433        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4248  for (;; ptr++) Line 4435  for (;; ptr++)
4435          inescq = FALSE;          inescq = FALSE;
4436          ptr += 2;          ptr += 2;
4437          }          }
   
4438        oldptr = ptr;        oldptr = ptr;
4439    
4440        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4441    
4442        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4443    
# Line 4259  for (;; ptr++) Line 4445  for (;; ptr++)
4445    
4446        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4447          {          {
4448          int d;          pcre_uint32 d;
4449          ptr += 2;          ptr += 2;
4450          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4451    
# Line 4274  for (;; ptr++) Line 4460  for (;; ptr++)
4460            inescq = TRUE;            inescq = TRUE;
4461            break;            break;
4462            }            }
4463    
4464            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4465            back the pointer and jump to handle the character that preceded it. */
4466    
4467          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4468            {            {
4469            ptr = oldptr;            ptr = oldptr;
4470            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4471            }            }
4472    
4473            /* Otherwise, we have a potential range; pick up the next character */
4474    
4475  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4476          if (utf)          if (utf)
# Line 4296  for (;; ptr++) Line 4487  for (;; ptr++)
4487    
4488          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4489            {            {
4490            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4491              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4492            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4493    
4494            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4495    
4496            if (d < 0)            if (descape != 0)
4497              {              {
4498              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4499                {                {
4500                ptr = oldptr;                ptr = oldptr;
4501                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4502                }                }
4503              }              }
4504            }            }
4505    
4506          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4507          one-character ranges */          one-character ranges. */
4508    
4509          if (d < c)          if (d < c)
4510            {            {
4511            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4512            goto FAILED;            goto FAILED;
4513            }            }
4514            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4515    
4516          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4517            cannot be done anymore. Any value greater than 1 indicates that there
4518            is more than one character. */
4519    
4520            class_one_char = 2;
4521    
4522          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4523    
4524          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4525    
4526          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4527          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4528          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));  
             }  
           }  
   
4529          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4530          }          }
4531    
4532        /* 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
4533        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
4534        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4535          increase it if it is already 2 or more ... just in case there's a class
4536        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4537    
4538        /* Only the value of 1 matters for class_single_char. */        CLASS_SINGLE_CHARACTER:
4539        if (class_single_char < 2) class_single_char++;        if (class_one_char < 2) class_one_char++;
4540    
4541        /* If class_charcount is 1, we saw precisely one character. As long as        /* If class_one_char is 1, we have the first single character in the
4542        there were no negated characters >= 128 and there was no use of \p or \P,        class, and there have been no prior ranges, or XCLASS items generated by
4543        in other words, no use of any XCLASS features, we can optimize.        escapes. If this is the final character in the class, we can optimize by
4544          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4545        In UTF-8 mode, we can optimize the negative case only if there were no        OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4546        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR        to be set. Otherwise, there can be no first char if this item is first,
4547        operate on single-bytes characters only. This is an historical hangover.        whatever repeat count may follow. In the case of reqchar, save the
4548        Maybe one day we can tidy these opcodes to handle multi-byte characters.        previous value for reinstating. */
   
       The optimization throws away the bit map. We turn the item into a  
       1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
       Note that OP_NOT[I] does not support multibyte characters. In the positive  
       case, it can cause firstchar to be set. Otherwise, there can be no first  
       char if this item is first, whatever repeat count may follow. In the case  
       of reqchar, save the previous value for reinstating. */  
4549    
4550  #ifdef SUPPORT_UTF        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET  
         && (!utf || !negate_class || c < (MAX_VALUE_FOR_SINGLE_CHAR + 1)))  
 #else  
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)  
 #endif  
4551          {          {
4552          ptr++;          ptr++;
4553          zeroreqchar = reqchar;          zeroreqchar = reqchar;
4554            zeroreqcharflags = reqcharflags;
         /* The OP_NOT[I] opcodes work on single characters only. */  
4555    
4556          if (negate_class)          if (negate_class)
4557            {            {
4558            if (firstchar == REQ_UNSET) firstchar = REQ_NONE;  #ifdef SUPPORT_UCP
4559              int d;
4560    #endif
4561              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4562            zerofirstchar = firstchar;            zerofirstchar = firstchar;
4563            *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;            zerofirstcharflags = firstcharflags;
4564            *code++ = c;  
4565            goto NOT_CHAR;            /* For caseless UTF-8 mode when UCP support is available, check
4566              whether this character has more than one other case. If so, generate
4567              a special OP_NOTPROP item instead of OP_NOTI. */
4568    
4569    #ifdef SUPPORT_UCP
4570              if (utf && (options & PCRE_CASELESS) != 0 &&
4571                  (d = UCD_CASESET(c)) != 0)
4572                {
4573                *code++ = OP_NOTPROP;
4574                *code++ = PT_CLIST;
4575                *code++ = d;
4576                }
4577              else
4578    #endif
4579              /* Char has only one other case, or UCP not available */
4580    
4581                {
4582                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4583    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4584                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4585                  code += PRIV(ord2utf)(c, code);
4586                else
4587    #endif
4588                  *code++ = c;
4589                }
4590    
4591              /* We are finished with this character class */
4592    
4593              goto END_CLASS;
4594            }            }
4595    
4596          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4597          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4598    
4599  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4600          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4601            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4602          else          else
# Line 4535  for (;; ptr++) Line 4607  for (;; ptr++)
4607            }            }
4608          goto ONE_CHAR;          goto ONE_CHAR;
4609          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4610    
4611        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4612          has been generated. Add this character to the class. */
4613  #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
4614        if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))        class_has_8bitchar +=
4615  #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));  
           }  
         }  
4616        }        }
4617    
4618      /* 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 4612  for (;; ptr++) Line 4632  for (;; ptr++)
4632        goto FAILED;        goto FAILED;
4633        }        }
4634    
4635        /* We will need an XCLASS if data has been placed in class_uchardata. In
4636        the second phase this is a sufficient test. However, in the pre-compile
4637        phase, class_uchardata gets emptied to prevent workspace overflow, so it
4638        only if the very last character in the class needs XCLASS will it contain
4639        anything at this point. For this reason, xclass gets set TRUE above when
4640        uchar_classdata is emptied, and that's why this code is the way it is here
4641        instead of just doing a test on class_uchardata below. */
4642    
4643    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4644        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4645    #endif
4646    
4647      /* 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
4648      setting, whatever the repeat count. Any reqchar setting must remain      setting, whatever the repeat count. Any reqchar setting must remain
4649      unchanged after any kind of repeat. */      unchanged after any kind of repeat. */
4650    
4651      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4652      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4653        zerofirstcharflags = firstcharflags;
4654      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4655        zeroreqcharflags = reqcharflags;
4656    
4657      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4658      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
# Line 4674  for (;; ptr++) Line 4708  for (;; ptr++)
4708        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4709        }        }
4710      code += 32 / sizeof(pcre_uchar);      code += 32 / sizeof(pcre_uchar);
4711      NOT_CHAR:  
4712        END_CLASS:
4713      break;      break;
4714    
4715    
# Line 4712  for (;; ptr++) Line 4747  for (;; ptr++)
4747      if (repeat_min == 0)      if (repeat_min == 0)
4748        {        {
4749        firstchar = zerofirstchar;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4750          firstcharflags = zerofirstcharflags;
4751        reqchar = zeroreqchar;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4752          reqcharflags = zeroreqcharflags;
4753        }        }
4754    
4755      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4775  for (;; ptr++) Line 4812  for (;; ptr++)
4812    
4813      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4814    
4815      /* If previous was a character match, abolish the item and generate a      /* If previous was a character or negated character match, abolish the item
4816      repeat item instead. If a char item has a minumum of more than one, ensure      and generate a repeat item instead. If a char item has a minimum of more
4817      that it is set in reqchar - it might not be if a sequence such as x{3} is      than one, ensure that it is set in reqchar - it might not be if a sequence
4818      the first thing in a branch because the x will have gone into firstchar      such as x{3} is the first thing in a branch because the x will have gone
4819      instead.  */      into firstchar instead.  */
4820    
4821      if (*previous == OP_CHAR || *previous == OP_CHARI)      if (*previous == OP_CHAR || *previous == OP_CHARI
4822        {          || *previous == OP_NOT || *previous == OP_NOTI)
4823        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;        {
4824          switch (*previous)
4825            {
4826            default: /* Make compiler happy. */
4827            case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
4828            case OP_CHARI: op_type = OP_STARI - OP_STAR; break;
4829            case OP_NOT:   op_type = OP_NOTSTAR - OP_STAR; break;
4830            case OP_NOTI:  op_type = OP_NOTSTARI - OP_STAR; break;
4831            }
4832    
4833        /* Deal with UTF characters that take up more than one character. It's        /* Deal with UTF characters that take up more than one character. It's
4834        easier to write this out separately than try to macrify it. Use c to        easier to write this out separately than try to macrify it. Use c to
4835        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
4836        it's a length rather than a small character. */        it's a length rather than a small character. */
4837    
4838  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4839        if (utf && NOT_FIRSTCHAR(code[-1]))        if (utf && NOT_FIRSTCHAR(code[-1]))
4840          {          {
4841          pcre_uchar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
# Line 4806  for (;; ptr++) Line 4851  for (;; ptr++)
4851        with UTF disabled, or for a single character UTF character. */        with UTF disabled, or for a single character UTF character. */
4852          {          {
4853          c = code[-1];          c = code[-1];
4854          if (repeat_min > 1) reqchar = c | req_caseopt | cd->req_varyopt;          if (*previous <= OP_CHARI && repeat_min > 1)
4855              {
4856              reqchar = c;
4857              reqcharflags = req_caseopt | cd->req_varyopt;
4858              }
4859          }          }
4860    
4861        /* If the repetition is unlimited, it pays to see if the next thing on        /* If the repetition is unlimited, it pays to see if the next thing on
# Line 4825  for (;; ptr++) Line 4874  for (;; ptr++)
4874        goto OUTPUT_SINGLE_REPEAT;   /* Code shared with single character types */        goto OUTPUT_SINGLE_REPEAT;   /* Code shared with single character types */
4875        }        }
4876    
     /* If previous was a single negated character ([^a] or similar), we use  
     one of the special opcodes, replacing it. The code is shared with single-  
     character repeats by setting opt_type to add a suitable offset into  
     repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI  
     are currently used only for single-byte chars. */  
   
     else if (*previous == OP_NOT || *previous == OP_NOTI)  
       {  
       op_type = ((*previous == OP_NOT)? OP_NOTSTAR : OP_NOTSTARI) - OP_STAR;  
       c = previous[1];  
       if (!possessive_quantifier &&  
           repeat_max < 0 &&  
           check_auto_possessive(previous, utf, ptr + 1, options, cd))  
         {  
         repeat_type = 0;    /* Force greedy */  
         possessive_quantifier = TRUE;  
         }  
       goto OUTPUT_SINGLE_REPEAT;  
       }  
   
4877      /* If previous was a character type match (\d or similar), abolish it and      /* If previous was a character type match (\d or similar), abolish it and
4878      create a suitable repeat item. The code is shared with single-character      create a suitable repeat item. The code is shared with single-character
4879      repeats by setting op_type to add a suitable offset into repeat_type. Note      repeats by setting op_type to add a suitable offset into repeat_type. Note
# Line 4945  for (;; ptr++) Line 4974  for (;; ptr++)
4974    
4975          if (repeat_max < 0)          if (repeat_max < 0)
4976            {            {
4977  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4978            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
4979              {              {
4980              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 4970  for (;; ptr++) Line 4999  for (;; ptr++)
4999    
5000          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
5001            {            {
5002  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5003            if (utf && (c & UTF_LENGTH) != 0)            if (utf && (c & UTF_LENGTH) != 0)
5004              {              {
5005              memcpy(code, utf_chars, IN_UCHARS(c & 7));              memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5000  for (;; ptr++) Line 5029  for (;; ptr++)
5029    
5030        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
5031    
5032  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
5033        if (utf && (c & UTF_LENGTH) != 0)        if (utf && (c & UTF_LENGTH) != 0)
5034          {          {
5035          memcpy(code, utf_chars, IN_UCHARS(c & 7));          memcpy(code, utf_chars, IN_UCHARS(c & 7));
# Line 5208  for (;; ptr++) Line 5237  for (;; ptr++)
5237    
5238            else            else
5239              {              {
5240              if (groupsetfirstchar && reqchar < 0) reqchar = firstchar;              if (groupsetfirstchar && reqcharflags < 0)
5241                  {
5242                  reqchar = firstchar;
5243                  reqcharflags = firstcharflags;
5244                  }
5245    
5246              for (i = 1; i < repeat_min; i++)              for (i = 1; i < repeat_min; i++)
5247                {                {
# Line 5585  for (;; ptr++) Line 5618  for (;; ptr++)
5618          arg = ++ptr;          arg = ++ptr;
5619          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5620          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5621            if ((unsigned int)arglen > MAX_MARK)
5622              {
5623              *errorcodeptr = ERR75;
5624              goto FAILED;
5625              }
5626          }          }
5627    
5628        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
# Line 5600  for (;; ptr++) Line 5638  for (;; ptr++)
5638          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5639              STRNCMP_UC_C8(name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5640            {            {
5641              int setverb;
5642    
5643            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5644            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5645    
# Line 5617  for (;; ptr++) Line 5657  for (;; ptr++)
5657                *code++ = OP_CLOSE;                *code++ = OP_CLOSE;
5658                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5659                }                }
5660              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              setverb = *code++ =
5661                  (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5662    
5663              /* Do not set firstchar after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5664              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
5665              }              }
5666    
5667            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5632  for (;; ptr++) Line 5673  for (;; ptr++)
5673                *errorcodeptr = ERR66;                *errorcodeptr = ERR66;
5674                goto FAILED;                goto FAILED;
5675                }                }
5676              *code = verbs[i].op;              setverb = *code++ = verbs[i].op;
             if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;  
5677              }              }
5678    
5679            else            else
# Line 5643  for (;; ptr++) Line 5683  for (;; ptr++)
5683                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5684                goto FAILED;                goto FAILED;
5685                }                }
5686              *code = verbs[i].op_arg;              setverb = *code++ = verbs[i].op_arg;
             if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;  
5687              *code++ = arglen;              *code++ = arglen;
5688              memcpy(code, arg, IN_UCHARS(arglen));              memcpy(code, arg, IN_UCHARS(arglen));
5689              code += arglen;              code += arglen;
5690              *code++ = 0;              *code++ = 0;
5691              }              }
5692    
5693              switch (setverb)
5694                {
5695                case OP_THEN:
5696                case OP_THEN_ARG:
5697                cd->external_flags |= PCRE_HASTHEN;
5698                break;
5699    
5700                case OP_PRUNE:
5701                case OP_PRUNE_ARG:
5702                case OP_SKIP:
5703                case OP_SKIP_ARG:
5704                cd->had_pruneorskip = TRUE;
5705                break;
5706                }
5707    
5708            break;  /* Found verb, exit loop */            break;  /* Found verb, exit loop */
5709            }            }
5710    
# Line 5773  for (;; ptr++) Line 5827  for (;; ptr++)
5827          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
5828            {            {
5829            if (recno >= 0)            if (recno >= 0)
5830              recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1;              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;
5831            ptr++;            ptr++;
5832            }            }
5833          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
5834    
5835          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||
5836              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
5837            {            {
5838            ptr--;      /* Error offset */            ptr--;      /* Error offset */
# Line 6018  for (;; ptr++) Line 6072  for (;; ptr++)
6072    
6073            if (lengthptr != NULL)            if (lengthptr != NULL)
6074              {              {
6075              if (*ptr != terminator)              if (*ptr != (pcre_uchar)terminator)
6076                {                {
6077                *errorcodeptr = ERR42;                *errorcodeptr = ERR42;
6078                goto FAILED;                goto FAILED;
# Line 6160  for (;; ptr++) Line 6214  for (;; ptr++)
6214              *errorcodeptr = ERR62;              *errorcodeptr = ERR62;
6215              goto FAILED;              goto FAILED;
6216              }              }
6217            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6218              {              {
6219              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
6220              goto FAILED;              goto FAILED;
# Line 6266  for (;; ptr++) Line 6320  for (;; ptr++)
6320            while(IS_DIGIT(*ptr))            while(IS_DIGIT(*ptr))
6321              recno = recno * 10 + *ptr++ - CHAR_0;              recno = recno * 10 + *ptr++ - CHAR_0;
6322    
6323            if (*ptr != terminator)            if (*ptr != (pcre_uchar)terminator)
6324              {              {
6325              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
6326              goto FAILED;              goto FAILED;
# Line 6370  for (;; ptr++) Line 6424  for (;; ptr++)
6424    
6425          /* Can't determine a first byte now */          /* Can't determine a first byte now */
6426    
6427          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6428          continue;          continue;
6429    
6430    
# Line 6504  for (;; ptr++) Line 6558  for (;; ptr++)
6558           cond_depth +           cond_depth +
6559             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
6560           &subfirstchar,                   /* For possible first char */           &subfirstchar,                   /* For possible first char */
6561             &subfirstcharflags,
6562           &subreqchar,                     /* For possible last char */           &subreqchar,                     /* For possible last char */
6563             &subreqcharflags,
6564           bcptr,                           /* Current branch chain */           bcptr,                           /* Current branch chain */
6565           cd,                              /* Tables block */           cd,                              /* Tables block */
6566           (lengthptr == NULL)? NULL :      /* Actual compile phase */           (lengthptr == NULL)? NULL :      /* Actual compile phase */
# Line 6565  for (;; ptr++) Line 6621  for (;; ptr++)
6621            *errorcodeptr = ERR27;            *errorcodeptr = ERR27;
6622            goto FAILED;            goto FAILED;
6623            }            }
6624          if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;          if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE;
6625          }          }
6626        }        }
6627    
# Line 6614  for (;; ptr++) Line 6670  for (;; ptr++)
6670      back off. */      back off. */
6671    
6672      zeroreqchar = reqchar;      zeroreqchar = reqchar;
6673        zeroreqcharflags = reqcharflags;
6674      zerofirstchar = firstchar;      zerofirstchar = firstchar;
6675        zerofirstcharflags = firstcharflags;
6676      groupsetfirstchar = FALSE;      groupsetfirstchar = FALSE;
6677    
6678      if (bravalue >= OP_ONCE)      if (bravalue >= OP_ONCE)
# Line 6625  for (;; ptr++) Line 6683  for (;; ptr++)
6683        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
6684        repeat forces firstchar to "none". */        repeat forces firstchar to "none". */
6685    
6686        if (firstchar == REQ_UNSET)        if (firstcharflags == REQ_UNSET)
6687          {          {
6688          if (subfirstchar >= 0)          if (subfirstcharflags >= 0)
6689            {            {
6690            firstchar = subfirstchar;            firstchar = subfirstchar;
6691              firstcharflags = subfirstcharflags;
6692            groupsetfirstchar = TRUE;            groupsetfirstchar = TRUE;
6693            }            }
6694          else firstchar = REQ_NONE;          else firstcharflags = REQ_NONE;
6695          zerofirstchar = REQ_NONE;          zerofirstcharflags = REQ_NONE;
6696          }          }
6697    
6698        /* If firstchar was previously set, convert the subpattern's firstchar        /* If firstchar was previously set, convert the subpattern's firstchar
6699        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
6700        existence beforehand. */        existence beforehand. */
6701    
6702        else if (subfirstchar >= 0 && subreqchar < 0)        else if (subfirstcharflags >= 0 && subreqcharflags < 0)
6703          subreqchar = subfirstchar | tempreqvary;          {
6704            subreqchar = subfirstchar;
6705            subreqcharflags = subfirstcharflags | tempreqvary;
6706            }
6707    
6708        /* 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
6709        really the first byte - see above), set it. */        really the first byte - see above), set it. */
6710    
6711        if (subreqchar >= 0) reqchar = subreqchar;        if (subreqcharflags >= 0)
6712            {
6713            reqchar = subreqchar;
6714            reqcharflags = subreqcharflags;
6715            }
6716        }        }
6717    
6718      /* For a forward assertion, we take the reqchar, if set. This can be      /* For a forward assertion, we take the reqchar, if set. This can be
# Line 6657  for (;; ptr++) Line 6723  for (;; ptr++)
6723      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
6724      firstchar, looking for an asserted first char. */      firstchar, looking for an asserted first char. */
6725    
6726      else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;      else if (bravalue == OP_ASSERT && subreqcharflags >= 0)
6727          {
6728          reqchar = subreqchar;
6729          reqcharflags = subreqcharflags;
6730          }
6731      break;     /* End of processing '(' */      break;     /* End of processing '(' */
6732    
6733    
# Line 6665  for (;; ptr++) Line 6735  for (;; ptr++)
6735      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
6736      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
6737      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
6738      are ESC_REF plus the reference number. Only back references and those types      are negative the reference number. Only back references and those types
6739      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
6740      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
6741      ever created. */      ever created. */
6742    
6743      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
6744      tempptr = ptr;      tempptr = ptr;
6745      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE);
6746    
6747      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
6748    
6749      if (c < 0)      if (escape == 0)
6750          c = ec;
6751        else
6752        {        {
6753        if (-c == ESC_Q)            /* Handle start of quoted string */        if (escape == ESC_Q)            /* Handle start of quoted string */
6754          {          {
6755          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
6756            ptr += 2;               /* avoid empty string */            ptr += 2;               /* avoid empty string */
# Line 6685  for (;; ptr++) Line 6758  for (;; ptr++)
6758          continue;          continue;
6759          }          }
6760    
6761        if (-c == ESC_E) continue;  /* Perl ignores an orphan \E */        if (escape == ESC_E) continue;  /* Perl ignores an orphan \E */
6762    
6763        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6764        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6765    
6766        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z)
6767          firstchar = REQ_NONE;          firstcharflags = REQ_NONE;
6768    
6769        /* 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. */
6770    
6771        zerofirstchar = firstchar;        zerofirstchar = firstchar;
6772          zerofirstcharflags = firstcharflags;
6773        zeroreqchar = reqchar;        zeroreqchar = reqchar;
6774          zeroreqcharflags = reqcharflags;
6775    
6776        /* \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'
6777        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
6778        -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 <
6779        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
6780        -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
6781        that is a synonym for a named back reference). */        that is a synonym for a named back reference). */
6782    
6783        if (-c == ESC_g)        if (escape == ESC_g)
6784          {          {
6785          const pcre_uchar *p;          const pcre_uchar *p;
6786          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
# Line 6725  for (;; ptr++) Line 6800  for (;; ptr++)
6800          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
6801            {            {
6802            BOOL is_a_number = TRUE;            BOOL is_a_number = TRUE;
6803            for (p = ptr + 1; *p != 0 && *p != terminator; p++)            for (p = ptr + 1; *p != 0 && *p != (pcre_uchar)terminator; p++)
6804              {              {
6805              if (!MAX_255(*p)) { is_a_number = FALSE; break; }              if (!MAX_255(*p)) { is_a_number = FALSE; break; }
6806              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;              if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;
6807              if ((cd->ctypes[*p] & ctype_word) == 0) break;              if ((cd->ctypes[*p] & ctype_word) == 0) break;
6808              }              }
6809            if (*p != terminator)            if (*p != (pcre_uchar)terminator)
6810              {              {
6811              *errorcodeptr = ERR57;              *errorcodeptr = ERR57;
6812              break;              break;
# Line 6749  for (;; ptr++) Line 6824  for (;; ptr++)
6824    
6825          p = ptr + 2;          p = ptr + 2;
6826          while (IS_DIGIT(*p)) p++;          while (IS_DIGIT(*p)) p++;
6827          if (*p != terminator)          if (*p != (pcre_uchar)terminator)
6828            {            {
6829            *errorcodeptr = ERR57;            *errorcodeptr = ERR57;
6830            break;            break;
# Line 6761  for (;; ptr++) Line 6836  for (;; ptr++)
6836        /* \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).
6837        We also support \k{name} (.NET syntax).  */        We also support \k{name} (.NET syntax).  */
6838    
6839        if (-c == ESC_k)        if (escape == ESC_k)
6840          {          {
6841          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&          if ((ptr[1] != CHAR_LESS_THAN_SIGN &&
6842            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))
# Line 6780  for (;; ptr++) Line 6855  for (;; ptr++)
6855        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
6856        ':' later. */        ':' later. */
6857    
6858        if (-c >= ESC_REF)        if (escape < 0)
6859          {          {
6860          open_capitem *oc;          open_capitem *oc;
6861          recno = -c - ESC_REF;          recno = -escape;
6862    
6863          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6864          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;          if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
6865          previous = code;          previous = code;
6866          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6867          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
# Line 6810  for (;; ptr++) Line 6885  for (;; ptr++)
6885        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
6886    
6887  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6888        else if (-c == ESC_P || -c == ESC_p)        else if (escape == ESC_P || escape == ESC_p)
6889          {          {
6890          BOOL negated;          BOOL negated;
6891          int pdata;          unsigned int ptype = 0, pdata = 0;
6892          int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);          if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
6893          if (ptype < 0) goto FAILED;            goto FAILED;
6894          previous = code;          previous = code;
6895          *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP;          *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP;
6896          *code++ = ptype;          *code++ = ptype;
6897          *code++ = pdata;          *code++ = pdata;
6898          }          }
# Line 6826  for (;; ptr++) Line 6901  for (;; ptr++)
6901        /* If Unicode properties are not supported, \X, \P, and \p are not        /* If Unicode properties are not supported, \X, \P, and \p are not
6902        allowed. */        allowed. */
6903    
6904        else if (-c == ESC_X || -c == ESC_P || -c == ESC_p)        else if (escape == ESC_X || escape == ESC_P || escape == ESC_p)
6905          {          {
6906          *errorcodeptr = ERR45;          *errorcodeptr = ERR45;
6907          goto FAILED;          goto FAILED;
# Line 6836  for (;; ptr++) Line 6911  for (;; ptr++)
6911        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6912        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
6913        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
6914        Unicode property tests. */        Unicode property tests. Note that \b and \B do a one-character
6915          lookbehind. */
6916    
6917        else        else
6918          {          {
6919            if ((escape == ESC_b || escape == ESC_B) && cd->max_lookbehind == 0)
6920              cd->max_lookbehind = 1;
6921  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6922          if (-c >= ESC_DU && -c <= ESC_wu)          if (escape >= ESC_DU && escape <= ESC_wu)
6923            {            {
6924            nestptr = ptr + 1;                   /* Where to resume */            nestptr = ptr + 1;                   /* Where to resume */
6925            ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */            ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
6926            }            }
6927          else          else
6928  #endif  #endif
# Line 6852  for (;; ptr++) Line 6930  for (;; ptr++)
6930          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6931    
6932            {            {
6933            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (escape > ESC_b && escape < ESC_Z)? code : NULL;
6934            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape;
6935            }            }
6936          }          }
6937        continue;        continue;
# Line 6863  for (;; ptr++) Line 6941  for (;; ptr++)
6941      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
6942      handle it as a data character. */      handle it as a data character. */
6943    
6944  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
6945      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)      if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
6946        mclength = PRIV(ord2utf)(c, mcbuffer);        mclength = PRIV(ord2utf)(c, mcbuffer);
6947      else      else
# Line 6896  for (;; ptr++) Line 6974  for (;; ptr++)
6974    
6975      ONE_CHAR:      ONE_CHAR:
6976      previous = code;      previous = code;
6977    
6978        /* For caseless UTF-8 mode when UCP support is available, check whether
6979        this character has more than one other case. If so, generate a special
6980        OP_PROP item instead of OP_CHARI. */
6981    
6982    #ifdef SUPPORT_UCP
6983        if (utf && (options & PCRE_CASELESS) != 0)
6984          {
6985          GETCHAR(c, mcbuffer);
6986          if ((c = UCD_CASESET(c)) != 0)
6987            {
6988            *code++ = OP_PROP;
6989            *code++ = PT_CLIST;
6990            *code++ = c;
6991            if (firstcharflags == REQ_UNSET) firstcharflags = zerofirstcharflags = REQ_NONE;
6992            break;
6993            }
6994          }
6995    #endif
6996    
6997        /* Caseful matches, or not one of the multicase characters. */
6998    
6999      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR;
7000      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
7001    
# Line 6909  for (;; ptr++) Line 7009  for (;; ptr++)
7009      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
7010      repeat. */      repeat. */
7011    
7012      if (firstchar == REQ_UNSET)      if (firstcharflags == REQ_UNSET)
7013        {        {
7014        zerofirstchar = REQ_NONE;        zerofirstcharflags = REQ_NONE;
7015        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7016          zeroreqcharflags = reqcharflags;
7017    
7018        /* 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
7019        only if it is not to be matched caselessly. */        only if it is not to be matched caselessly. */
# Line 6920  for (;; ptr++) Line 7021  for (;; ptr++)
7021        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7022          {          {
7023          firstchar = mcbuffer[0] | req_caseopt;          firstchar = mcbuffer[0] | req_caseopt;
7024          if (mclength != 1) reqchar = code[-1] | cd->req_varyopt;          firstchar = mcbuffer[0];
7025            firstcharflags = req_caseopt;
7026    
7027            if (mclength != 1)
7028              {
7029              reqchar = code[-1];
7030              reqcharflags = cd->req_varyopt;
7031              }
7032          }          }
7033        else firstchar = reqchar = REQ_NONE;        else firstcharflags = reqcharflags = REQ_NONE;
7034        }        }
7035    
7036      /* firstchar was previously set; we can set reqchar only if the length is      /* firstchar was previously set; we can set reqchar only if the length is
# Line 6931  for (;; ptr++) Line 7039  for (;; ptr++)
7039      else      else
7040        {        {
7041        zerofirstchar = firstchar;        zerofirstchar = firstchar;
7042          zerofirstcharflags = firstcharflags;
7043        zeroreqchar = reqchar;        zeroreqchar = reqchar;
7044          zeroreqcharflags = reqcharflags;
7045        if (mclength == 1 || req_caseopt == 0)        if (mclength == 1 || req_caseopt == 0)
7046          reqchar = code[-1] | req_caseopt | cd->req_varyopt;          {
7047            reqchar = code[-1];
7048            reqcharflags = req_caseopt | cd->req_varyopt;
7049            }
7050        }        }
7051    
7052      break;            /* End of literal character handling */      break;            /* End of literal character handling */
# Line 6952  return FALSE; Line 7065  return FALSE;
7065    
7066    
7067    
   
7068  /*************************************************  /*************************************************
7069  *     Compile sequence of alternatives           *  *     Compile sequence of alternatives           *
7070  *************************************************/  *************************************************/
# Line 6973  Arguments: Line 7085  Arguments:
7085    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
7086    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
7087    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
7088    firstcharptr   place to put the first required character, or a negative number    firstcharptr    place to put the first required character
7089    reqcharptr     place to put the last required character, or a negative number    firstcharflagsptr place to put the first character flags, or a negative number
7090      reqcharptr     place to put the last required character
7091      reqcharflagsptr place to put the last required character flags, or a negative number
7092    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
7093    cd             points to the data block with tables pointers etc.    cd             points to the data block with tables pointers etc.
7094    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
# Line 6986  Returns:         TRUE on success Line 7100  Returns:         TRUE on success
7100  static BOOL  static BOOL
7101  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,  compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr,
7102    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
7103    int cond_depth, pcre_int32 *firstcharptr, pcre_int32 *reqcharptr,    int cond_depth,
7104      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
7105      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
7106    branch_chain *bcptr, compile_data *cd, int *lengthptr)    branch_chain *bcptr, compile_data *cd, int *lengthptr)
7107  {  {
7108  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
# Line 6996  pcre_uchar *start_bracket = code; Line 7112  pcre_uchar *start_bracket = code;
7112  pcre_uchar *reverse_count = NULL;  pcre_uchar *reverse_count = NULL;
7113  open_capitem capitem;  open_capitem capitem;
7114  int capnumber = 0;  int capnumber = 0;
7115  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7116  pcre_int32 branchfirstchar, branchreqchar;  pcre_int32 firstcharflags, reqcharflags;
7117    pcre_uint32 branchfirstchar, branchreqchar;
7118    pcre_int32 branchfirstcharflags, branchreqcharflags;
7119  int length;  int length;
7120  int orig_bracount;  unsigned int orig_bracount;
7121  int max_bracount;  unsigned int max_bracount;
7122  branch_chain bc;  branch_chain bc;
7123    
7124  bc.outer = bcptr;  bc.outer = bcptr;
7125  bc.current_branch = code;  bc.current_branch = code;
7126    
7127  firstchar = reqchar = REQ_UNSET;  firstchar = reqchar = 0;
7128    firstcharflags = reqcharflags = REQ_UNSET;
7129    
7130  /* 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
7131  length of the BRA and KET and any extra bytes that are required at the  length of the BRA and KET and any extra bytes that are required at the
# Line 7066  for (;;) Line 7185  for (;;)
7185    into the length. */    into the length. */
7186    
7187    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,
7188          &branchreqchar, &bc, cond_depth, cd,          &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc,
7189          (lengthptr == NULL)? NULL : &length))          cond_depth, cd, (lengthptr == NULL)? NULL : &length))
7190      {      {
7191      *ptrptr = ptr;      *ptrptr = ptr;
7192      return FALSE;      return FALSE;
# Line 7088  for (;;) Line 7207  for (;;)
7207      if (*last_branch != OP_ALT)      if (*last_branch != OP_ALT)
7208        {        {
7209        firstchar = branchfirstchar;        firstchar = branchfirstchar;
7210          firstcharflags = branchfirstcharflags;
7211        reqchar = branchreqchar;        reqchar = branchreqchar;
7212          reqcharflags = branchreqcharflags;
7213        }        }
7214    
7215      /* If this is not the first branch, the first char and reqchar have to      /* If this is not the first branch, the first char and reqchar have to
# Line 7102  for (;;) Line 7223  for (;;)
7223        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
7224        previously no reqchar, it takes on the value of the old firstchar. */        previously no reqchar, it takes on the value of the old firstchar. */
7225    
7226        if (firstchar >= 0 && firstchar != branchfirstchar)        if (firstcharflags >= 0 &&
7227              (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar))
7228          {          {
7229          if (reqchar < 0) reqchar = firstchar;          if (reqcharflags < 0)
7230          firstchar = REQ_NONE;            {
7231              reqchar = firstchar;
7232              reqcharflags = firstcharflags;
7233              }
7234            firstcharflags = REQ_NONE;
7235          }          }
7236    
7237        /* 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
7238        branch becomes a reqchar if there isn't a branch reqchar. */        branch becomes a reqchar if there isn't a branch reqchar. */
7239    
7240        if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0)        if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0)
7241            branchreqchar = branchfirstchar;          {
7242            branchreqchar = branchfirstchar;
7243            branchreqcharflags = branchfirstcharflags;
7244            }
7245    
7246        /* Now ensure that the reqchars match */        /* Now ensure that the reqchars match */
7247    
7248        if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY))        if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) ||
7249          reqchar = REQ_NONE;            reqchar != branchreqchar)
7250        else reqchar |= branchreqchar;   /* To "or" REQ_VARY */          reqcharflags = REQ_NONE;
7251          else
7252            {
7253            reqchar = branchreqchar;
7254            reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */
7255            }
7256        }        }
7257    
7258      /* If lookbehind, check that this branch matches a fixed-length string, and      /* If lookbehind, check that this branch matches a fixed-length string, and
# Line 7147  for (;;) Line 7281  for (;;)
7281          *ptrptr = ptr;          *ptrptr = ptr;
7282          return FALSE;          return FALSE;
7283          }          }
7284        else { PUT(reverse_count, 0, fixed_length); }        else
7285            {
7286            if (fixed_length > cd->max_lookbehind)
7287              cd->max_lookbehind = fixed_length;
7288            PUT(reverse_count, 0, fixed_length);
7289            }
7290        }        }
7291      }      }
7292    
# Line 7209  for (;;) Line 7348  for (;;)
7348      *codeptr = code;      *codeptr = code;
7349      *ptrptr = ptr;      *ptrptr = ptr;
7350      *firstcharptr = firstchar;      *firstcharptr = firstchar;
7351        *firstcharflagsptr = firstcharflags;
7352      *reqcharptr = reqchar;      *reqcharptr = reqchar;
7353        *reqcharflagsptr = reqcharflags;
7354      if (lengthptr != NULL)      if (lengthptr != NULL)
7355        {        {
7356        if (OFLOW_MAX - *lengthptr < length)        if (OFLOW_MAX - *lengthptr < length)
# Line 7279  and the highest back reference was great Line 7420  and the highest back reference was great
7420  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
7421  of the more common cases more precisely.  of the more common cases more precisely.
7422    
7423    ... A second exception is when the .* appears inside an atomic group, because
7424    this prevents the number of characters it matches from being adjusted.
7425    
7426  Arguments:  Arguments:
7427    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7428    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
7429                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7430                    the less precise approach                    the less precise approach
7431    backref_map    the back reference bitmap    cd             points to the compile data block
7432      atomcount      atomic group level
7433    
7434  Returns:     TRUE or FALSE  Returns:     TRUE or FALSE
7435  */  */
7436    
7437  static BOOL  static BOOL
7438  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,  is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
7439    unsigned int backref_map)    compile_data *cd, int atomcount)
7440  {  {
7441  do {  do {
7442     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7303  do { Line 7448  do {
7448     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7449         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7450       {       {
7451       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7452       }       }
7453    
7454     /* Capturing brackets */     /* Capturing brackets */
# Line 7313  do { Line 7458  do {
7458       {       {
7459       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7460       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7461       if (!is_anchored(scode, new_map, backref_map)) return FALSE;       if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE;
7462       }       }
7463    
7464     /* Other brackets */     /* Positive forward assertions and conditions */
7465    
7466     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||     else if (op == OP_ASSERT || op == OP_COND)
             op == OP_COND)  
7467       {       {
7468       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE;
7469         }
7470    
7471       /* Atomic groups */
7472    
7473       else if (op == OP_ONCE || op == OP_ONCE_NC)
7474         {
7475         if (!is_anchored(scode, bracket_map, cd, atomcount + 1))
7476           return FALSE;
7477       }       }
7478    
7479     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
7480     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
7481       group. */
7482    
7483     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
7484               op == OP_TYPEPOSSTAR))               op == OP_TYPEPOSSTAR))
7485       {       {
7486       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)       if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 ||
7487             atomcount > 0 || cd->had_pruneorskip)
7488         return FALSE;         return FALSE;
7489       }       }
7490    
7491     /* Check for explicit anchoring */     /* Check for explicit anchoring */
7492    
7493     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;
7494    
7495     code += GET(code, 1);     code += GET(code, 1);
7496     }     }
7497  while (*code == OP_ALT);   /* Loop for each alternative */  while (*code == OP_ALT);   /* Loop for each alternative */
# Line 7354  return TRUE; Line 7509  return TRUE;
7509  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
7510  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
7511  have to take account of back references to capturing brackets that contain .*  have to take account of back references to capturing brackets that contain .*
7512  because in that case we can't make the assumption.  because in that case we can't make the assumption. Also, the appearance of .*
7513    inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not
7514    count, because once again the assumption no longer holds.
7515    
7516  Arguments:  Arguments:
7517    code           points to start of expression (the bracket)    code           points to start of expression (the bracket)
7518    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
7519                    handles up to substring 31; after that we just have to take                    handles up to substring 31; after that we just have to take
7520                    the less precise approach                    the less precise approach
7521    backref_map    the back reference bitmap    cd             points to the compile data
7522      atomcount      atomic group level
7523    
7524  Returns:         TRUE or FALSE  Returns:         TRUE or FALSE
7525  */  */
7526    
7527  static BOOL  static BOOL
7528  is_startline(const pcre_uchar *code, unsigned int bracket_map,  is_startline(const pcre_uchar *code, unsigned int bracket_map,
7529    unsigned int backref_map)    compile_data *cd, int atomcount)
7530  {  {
7531  do {  do {
7532     const pcre_uchar *scode = first_significant_code(     const pcre_uchar *scode = first_significant_code(
# Line 7394  do { Line 7552  do {
7552         return FALSE;         return FALSE;
7553    
7554         default:     /* Assertion */         default:     /* Assertion */
7555         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7556         do scode += GET(scode, 1); while (*scode == OP_ALT);         do scode += GET(scode, 1); while (*scode == OP_ALT);
7557         scode += 1 + LINK_SIZE;         scode += 1 + LINK_SIZE;
7558         break;         break;
# Line 7408  do { Line 7566  do {
7566     if (op == OP_BRA  || op == OP_BRAPOS ||     if (op == OP_BRA  || op == OP_BRAPOS ||
7567         op == OP_SBRA || op == OP_SBRAPOS)         op == OP_SBRA || op == OP_SBRAPOS)
7568       {       {
7569       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7570       }       }
7571    
7572     /* Capturing brackets */     /* Capturing brackets */
# Line 7418  do { Line 7576  do {
7576       {       {
7577       int n = GET2(scode, 1+LINK_SIZE);       int n = GET2(scode, 1+LINK_SIZE);
7578       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);       int new_map = bracket_map | ((n < 32)? (1 << n) : 1);
7579       if (!is_startline(scode, new_map, backref_map)) return FALSE;       if (!is_startline(scode, new_map, cd, atomcount)) return FALSE;
7580       }       }
7581    
7582     /* Other brackets */     /* Positive forward assertions */
7583    
7584       else if (op == OP_ASSERT)
7585         {
7586         if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE;
7587         }
7588    
7589       /* Atomic brackets */
7590    
7591     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)     else if (op == OP_ONCE || op == OP_ONCE_NC)
7592       {       {
7593       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE;
7594       }       }
7595    
7596     /* .* 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
7597     may be referenced. */     brackets that may be referenced, as long as the pattern does not contain
7598       *PRUNE or *SKIP, because these break the feature. Consider, for example,
7599       /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the
7600       start of a line. */
7601    
7602     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)     else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)
7603       {       {
7604       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 ||
7605             atomcount > 0 || cd->had_pruneorskip)
7606           return FALSE;
7607       }       }
7608    
7609     /* Check for explicit circumflex */     /* Check for explicit circumflex; anything else gives a FALSE result. Note
7610       in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC
7611       because the number of characters matched by .* cannot be adjusted inside
7612       them. */
7613    
7614     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;     else if (op != OP_CIRC && op != OP_CIRCM) return FALSE;
7615    
# Line 7464  we return that char, otherwise -1. Line 7637  we return that char, otherwise -1.
7637    
7638  Arguments:  Arguments:
7639    code       points to start of expression (the bracket)    code       points to start of expression (the bracket)
7640      flags       points to the first char flags, or to REQ_NONE
7641    inassert   TRUE if in an assertion    inassert   TRUE if in an assertion
7642    
7643  Returns:     -1 or the fixed first char  Returns:     the fixed first char, or 0 with REQ_NONE in flags
7644  */  */
7645    
7646  static int  static pcre_uint32
7647  find_firstassertedchar(const pcre_uchar *code, BOOL inassert)  find_firstassertedchar(const pcre_uchar *code, pcre_int32 *flags,
7648      BOOL inassert)
7649  {  {
7650  register int c = -1;  register pcre_uint32 c = 0;
7651    int cflags = REQ_NONE;
7652    
7653    *flags = REQ_NONE;
7654  do {  do {
7655     int d;     pcre_uint32 d;
7656       int dflags;
7657     int xl = (*code == OP_CBRA || *code == OP_SCBRA ||     int xl = (*code == OP_CBRA || *code == OP_SCBRA ||
7658               *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0;               *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0;
7659     const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,     const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,
7660       TRUE);       TRUE);
7661     register int op = *scode;     register pcre_uchar op = *scode;
7662    
7663     switch(op)     switch(op)
7664       {       {
7665       default:       default:
7666       return -1;       return 0;
7667    
7668       case OP_BRA:       case OP_BRA:
7669       case OP_BRAPOS:       case OP_BRAPOS:
# Line 7496  do { Line 7675  do {
7675       case OP_ONCE:       case OP_ONCE:
7676       case OP_ONCE_NC:       case OP_ONCE_NC:
7677       case OP_COND:       case OP_COND:
7678       if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0)       d = find_firstassertedchar(scode, &dflags, op == OP_ASSERT);
7679         return -1;       if (dflags < 0)
7680       if (c < 0) c = d; else if (c != d) return -1;         return 0;
7681         if (cflags < 0) { c = d; cflags = dflags; } else if (c != d || cflags != dflags) return 0;
7682       break;       break;
7683    
7684       case OP_EXACT:       case OP_EXACT:
# Line 7509  do { Line 7689  do {
7689       case OP_PLUS:       case OP_PLUS:
7690       case OP_MINPLUS:       case OP_MINPLUS:
7691       case OP_POSPLUS:       case OP_POSPLUS:
7692       if (!inassert) return -1;       if (!inassert) return 0;
7693       if (c < 0) c = scode[1];       if (cflags < 0) { c = scode[1]; cflags = 0; }
7694         else if (c != scode[1]) return -1;         else if (c != scode[1]) return 0;
7695       break;       break;
7696    
7697       case OP_EXACTI:       case OP_EXACTI:
# Line 7522  do { Line 7702  do {
7702       case OP_PLUSI:       case OP_PLUSI:
7703       case OP_MINPLUSI:       case OP_MINPLUSI:
7704       case OP_POSPLUSI:       case OP_POSPLUSI:
7705       if (!inassert) return -1;       if (!inassert) return 0;
7706       if (c < 0) c = scode[1] | REQ_CASELESS;       if (cflags < 0) { c = scode[1]; cflags = REQ_CASELESS; }
7707         else if (c != scode[1]) return -1;         else if (c != scode[1]) return 0;
7708       break;       break;
7709       }       }
7710    
7711     code += GET(code, 1);     code += GET(code, 1);
7712     }     }
7713  while (*code == OP_ALT);  while (*code == OP_ALT);
7714    
7715    *flags = cflags;
7716  return c;  return c;
7717  }  }
7718    
# Line 7558  Returns:        pointer to compiled data Line 7740  Returns:        pointer to compiled data
7740                  with errorptr and erroroffset set                  with errorptr and erroroffset set
7741  */  */
7742    
7743  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7744  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7745  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
7746    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7747  #else  #elif defined COMPILE_PCRE16
7748  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7749  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,  pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,
7750    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
7751    #elif defined COMPILE_PCRE32
7752    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7753    pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr,
7754      int *erroroffset, const unsigned char *tables)
7755  #endif  #endif
7756  {  {
7757  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7758  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7759  #else  #elif defined COMPILE_PCRE16
7760  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);  return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7761    #elif defined COMPILE_PCRE32
7762    return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
7763  #endif  #endif
7764  }  }
7765    
7766    
7767  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7768  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
7769  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
7770    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7771  #else  #elif defined COMPILE_PCRE16
7772  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION  PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
7773  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,  pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,
7774    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
7775    #elif defined COMPILE_PCRE32
7776    PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION
7777    pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr,
7778      const char **errorptr, int *erroroffset, const unsigned char *tables)
7779  #endif  #endif
7780  {  {
7781  REAL_PCRE *re;  REAL_PCRE *re;
7782  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
7783  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
7784    pcre_int32 firstcharflags, reqcharflags;
7785  int newline;  int newline;
7786  int errorcode = 0;  int errorcode = 0;
7787  int skipatstart = 0;  int skipatstart = 0;
# Line 7669  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7862  while (ptr[skipatstart] == CHAR_LEFT_PAR
7862    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)    if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7863      { skipatstart += 8; options |= PCRE_UTF16; continue; }      { skipatstart += 8; options |= PCRE_UTF16; continue; }
7864  #endif  #endif
7865    #ifdef COMPILE_PCRE32
7866      if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
7867        { skipatstart += 8; options |= PCRE_UTF32; continue; }
7868    #endif
7869    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)
7870      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
7871    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 7697  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 7894  while (ptr[skipatstart] == CHAR_LEFT_PAR
7894    else break;    else break;
7895    }    }
7896    
7897  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */
7898  utf = (options & PCRE_UTF8) != 0;  utf = (options & PCRE_UTF8) != 0;
7899    
7900  /* 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 7709  not used here. */ Line 7906  not used here. */
7906  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&
7907       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)       (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)
7908    {    {
7909  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
7910    errorcode = ERR44;    errorcode = ERR44;
7911  #else  #elif defined COMPILE_PCRE16
7912    errorcode = ERR74;    errorcode = ERR74;
7913    #elif defined COMPILE_PCRE32
7914      errorcode = ERR77;
7915  #endif  #endif
7916    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;
7917    }    }
# Line 7817  cd->start_pattern = (const pcre_uchar *) Line 8016  cd->start_pattern = (const pcre_uchar *)
8016  cd->end_pattern = (const pcre_uchar *)(pattern + STRLEN_UC((const pcre_uchar *)pattern));  cd->end_pattern = (const pcre_uchar *)(pattern + STRLEN_UC((const pcre_uchar *)pattern));
8017  cd->req_varyopt = 0;  cd->req_varyopt = 0;
8018  cd->assert_depth = 0;  cd->assert_depth = 0;
8019    cd->max_lookbehind = 0;
8020  cd->external_options = options;  cd->external_options = options;
8021  cd->external_flags = 0;  cd->external_flags = 0;
8022  cd->open_caps = NULL;  cd->open_caps = NULL;
# Line 7831  ptr += skipatstart; Line 8031  ptr += skipatstart;
8031  code = cworkspace;  code = cworkspace;
8032  *code = OP_BRA;  *code = OP_BRA;
8033  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
8034    FALSE, 0, 0, &firstchar, &reqchar, NULL, cd, &length);    FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL,
8035      cd, &length);
8036  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
8037    
8038  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 7867  re->magic_number = MAGIC_NUMBER; Line 8068  re->magic_number = MAGIC_NUMBER;
8068  re->size = (int)size;  re->size = (int)size;
8069  re->options = cd->external_options;  re->options = cd->external_options;
8070  re->flags = cd->external_flags;  re->flags = cd->external_flags;
 re->dummy1 = 0;  
8071  re->first_char = 0;  re->first_char = 0;
8072  re->req_char = 0;  re->req_char = 0;
8073  re->name_table_offset = sizeof(REAL_PCRE) / sizeof(pcre_uchar);  re->name_table_offset = sizeof(REAL_PCRE) / sizeof(pcre_uchar);
# Line 7876  re->name_count = cd->names_found; Line 8076  re->name_count = cd->names_found;
8076  re->ref_count = 0;  re->ref_count = 0;
8077  re->tables = (tables == PRIV(default_tables))? NULL : tables;  re->tables = (tables == PRIV(default_tables))? NULL : tables;
8078  re->nullpad = NULL;  re->nullpad = NULL;
8079    #ifdef COMPILE_PCRE32
8080    re->dummy1 = re->dummy2 = 0;
8081    #endif
8082    
8083  /* 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
8084  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 7887  field; this time it's used for rememberi Line 8090  field; this time it's used for rememberi
8090  cd->final_bracount = cd->bracount;  /* Save for checking forward references */  cd->final_bracount = cd->bracount;  /* Save for checking forward references */
8091  cd->assert_depth = 0;  cd->assert_depth = 0;
8092  cd->bracount = 0;  cd->bracount = 0;
8093    cd->max_lookbehind = 0;
8094  cd->names_found = 0;  cd->names_found = 0;
8095  cd->name_table = (pcre_uchar *)re + re->name_table_offset;  cd->name_table = (pcre_uchar *)re + re->name_table_offset;
8096  codestart = cd->name_table + re->name_entry_size * re->name_count;  codestart = cd->name_table + re->name_entry_size * re->name_count;
# Line 7894  cd->start_code = codestart; Line 8098  cd->start_code = codestart;
8098  cd->hwm = (pcre_uchar *)(cd->start_workspace);  cd->hwm = (pcre_uchar *)(cd->start_workspace);
8099  cd->req_varyopt = 0;  cd->req_varyopt = 0;
8100  cd->had_accept = FALSE;  cd->had_accept = FALSE;
8101    cd->had_pruneorskip = FALSE;
8102  cd->check_lookbehind = FALSE;  cd->check_lookbehind = FALSE;
8103  cd->open_caps = NULL;  cd->open_caps = NULL;
8104    
# Line 7905  ptr = (const pcre_uchar *)pattern + skip Line 8110  ptr = (const pcre_uchar *)pattern + skip
8110  code = (pcre_uchar *)codestart;  code = (pcre_uchar *)codestart;
8111  *code = OP_BRA;  *code = OP_BRA;
8112  (void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0,  (void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0,
8113    &firstchar, &reqchar, NULL, cd, NULL);    &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL, cd, NULL);
8114  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
8115  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
8116    re->max_lookbehind = cd->max_lookbehind;
8117  re->flags = cd->external_flags | PCRE_MODE;  re->flags = cd->external_flags | PCRE_MODE;
8118    
8119  if (cd->had_accept) reqchar = REQ_NONE;   /* Must disable after (*ACCEPT) */  if (cd->had_accept)
8120      {
8121      reqchar = 0;              /* Must disable after (*ACCEPT) */
8122      reqcharflags = REQ_NONE;
8123      }
8124    
8125  /* 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. */
8126    
# Line 7960  if (errorcode == 0 && re->top_backref > Line 8170  if (errorcode == 0 && re->top_backref >
8170    
8171  /* If there were any lookbehind assertions that contained OP_RECURSE  /* If there were any lookbehind assertions that contained OP_RECURSE
8172  (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,
8173  because they may contain forward references. Actual recursions can't be fixed  because they may contain forward references. Actual recursions cannot be fixed
8174  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
8175  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
8176  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 7995  if (cd->check_lookbehind) Line 8205  if (cd->check_lookbehind)
8205                      (fixed_length == -4)? ERR70 : ERR25;                      (fixed_length == -4)? ERR70 : ERR25;
8206          break;          break;
8207          }          }
8208          if (fixed_length > cd->max_lookbehind) cd->max_lookbehind = fixed_length;
8209        PUT(cc, 1, fixed_length);        PUT(cc, 1, fixed_length);
8210        }        }
8211      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 8015  if (errorcode != 0) Line 8226  if (errorcode != 0)
8226    }    }
8227    
8228  /* 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
8229  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
8230  as starting with .* when DOTALL is set).  as starting with non-atomic .* when DOTALL is set and there are no occurrences
8231    of *PRUNE or *SKIP.
8232    
8233  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
8234  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
8235  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches  PCRE_STARTLINE flag. This is helpful for multiline matches when all branches
8236  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
8237  */  non-DOTALL matches when *PRUNE and SKIP are not present. */
8238    
8239  if ((re->options & PCRE_ANCHORED) == 0)  if ((re->options & PCRE_ANCHORED) == 0)
8240    {    {
8241    if (is_anchored(codestart, 0, cd->backref_map))    if (is_anchored(codestart, 0, cd, 0)) re->options |= PCRE_ANCHORED;
     re->options |= PCRE_ANCHORED;  
8242    else    else
8243      {      {
8244      if (firstchar < 0)      if (firstcharflags < 0)
8245        firstchar = find_firstassertedchar(codestart, FALSE);        firstchar = find_firstassertedchar(codestart, &firstcharflags, FALSE);
8246      if (firstchar >= 0)   /* Remove caseless flag for non-caseable chars */      if (firstcharflags >= 0)   /* Remove caseless flag for non-caseable chars */
8247        {        {
8248  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
8249        re->first_char = firstchar & 0xff;        re->first_char = firstchar & 0xff;
8250  #else  #elif defined COMPILE_PCRE16