/[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 1363 by ph10, Tue Oct 1 16:54:40 2013 UTC revision 1495 by ph10, Sat Jul 12 18:22:54 2014 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2013 University of Cambridge             Copyright (c) 1997-2014 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 47  supporting internal functions that are n Line 47  supporting internal functions that are n
47  #endif  #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
50  #define PSSTART start_pattern  /* Field containing processed string start */  #define PSSTART start_pattern  /* Field containing pattern start */
51  #define PSEND   end_pattern    /* Field containing processed string end */  #define PSEND   end_pattern    /* Field containing pattern end */
52    
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
# Line 260  static const verbitem verbs[] = { Line 260  static const verbitem verbs[] = {
260  static const int verbcount = sizeof(verbs)/sizeof(verbitem);  static const int verbcount = sizeof(verbs)/sizeof(verbitem);
261    
262    
263    /* Substitutes for [[:<:]] and [[:>:]], which mean start and end of word in
264    another regex library. */
265    
266    static const pcre_uchar sub_start_of_word[] = {
267      CHAR_BACKSLASH, CHAR_b, CHAR_LEFT_PARENTHESIS, CHAR_QUESTION_MARK,
268      CHAR_EQUALS_SIGN, CHAR_BACKSLASH, CHAR_w, CHAR_RIGHT_PARENTHESIS, '\0' };
269    
270    static const pcre_uchar sub_end_of_word[] = {
271      CHAR_BACKSLASH, CHAR_b, CHAR_LEFT_PARENTHESIS, CHAR_QUESTION_MARK,
272      CHAR_LESS_THAN_SIGN, CHAR_EQUALS_SIGN, CHAR_BACKSLASH, CHAR_w,
273      CHAR_RIGHT_PARENTHESIS, '\0' };
274    
275    
276  /* Tables of names of POSIX character classes and their lengths. The names are  /* Tables of names of POSIX character classes and their lengths. The names are
277  now all in a single string, to reduce the number of relocations when a shared  now all in a single string, to reduce the number of relocations when a shared
278  library is dynamically loaded. The list of lengths is terminated by a zero  library is dynamically loaded. The list of lengths is terminated by a zero
279  length entry. The first three must be alpha, lower, upper, as this is assumed  length entry. The first three must be alpha, lower, upper, as this is assumed
280  for handling case independence. */  for handling case independence. The indices for graph, print, and punct are
281    needed, so identify them. */
282    
283  static const char posix_names[] =  static const char posix_names[] =
284    STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0    STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0
# Line 275  static const char posix_names[] = Line 289  static const char posix_names[] =
289  static const pcre_uint8 posix_name_lengths[] = {  static const pcre_uint8 posix_name_lengths[] = {
290    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };
291    
292    #define PC_GRAPH  8
293    #define PC_PRINT  9
294    #define PC_PUNCT 10
295    
296    
297  /* Table of class bit maps for each POSIX class. Each class is formed from a  /* Table of class bit maps for each POSIX class. Each class is formed from a
298  base map, with an optional addition or removal of another map. Then, for some  base map, with an optional addition or removal of another map. Then, for some
299  classes, there is some additional tweaking: for [:blank:] the vertical space  classes, there is some additional tweaking: for [:blank:] the vertical space
# Line 302  static const int posix_class_maps[] = { Line 321  static const int posix_class_maps[] = {
321    cbit_xdigit,-1,          0              /* xdigit */    cbit_xdigit,-1,          0              /* xdigit */
322  };  };
323    
324  /* Table of substitutes for \d etc when PCRE_UCP is set. The POSIX class  /* Table of substitutes for \d etc when PCRE_UCP is set. They are replaced by
325  substitutes must be in the order of the names, defined above, and there are  Unicode property escapes. */
 both positive and negative cases. NULL means no substitute. */  
326    
327  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
328  static const pcre_uchar string_PNd[]  = {  static const pcre_uchar string_PNd[]  = {
# Line 329  static const pcre_uchar string_pXwd[] = Line 347  static const pcre_uchar string_pXwd[] =
347  static const pcre_uchar *substitutes[] = {  static const pcre_uchar *substitutes[] = {
348    string_PNd,           /* \D */    string_PNd,           /* \D */
349    string_pNd,           /* \d */    string_pNd,           /* \d */
350    string_PXsp,          /* \S */       /* NOTE: Xsp is Perl space */    string_PXsp,          /* \S */   /* Xsp is Perl space, but from 8.34, Perl */
351    string_pXsp,          /* \s */    string_pXsp,          /* \s */   /* space and POSIX space are the same. */
352    string_PXwd,          /* \W */    string_PXwd,          /* \W */
353    string_pXwd           /* \w */    string_pXwd           /* \w */
354  };  };
355    
356    /* The POSIX class substitutes must be in the order of the POSIX class names,
357    defined above, and there are both positive and negative cases. NULL means no
358    general substitute of a Unicode property escape (\p or \P). However, for some
359    POSIX classes (e.g. graph, print, punct) a special property code is compiled
360    directly. */
361    
362  static const pcre_uchar string_pL[] =   {  static const pcre_uchar string_pL[] =   {
363    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
364    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
# Line 382  static const pcre_uchar *posix_substitut Line 406  static const pcre_uchar *posix_substitut
406    NULL,                 /* graph */    NULL,                 /* graph */
407    NULL,                 /* print */    NULL,                 /* print */
408    NULL,                 /* punct */    NULL,                 /* punct */
409    string_pXps,          /* space */    /* NOTE: Xps is POSIX space */    string_pXps,          /* space */   /* Xps is POSIX space, but from 8.34 */
410    string_pXwd,          /* word */    string_pXwd,          /* word  */   /* Perl and POSIX space are the same */
411    NULL,                 /* xdigit */    NULL,                 /* xdigit */
412    /* Negated cases */    /* Negated cases */
413    string_PL,            /* ^alpha */    string_PL,            /* ^alpha */
# Line 397  static const pcre_uchar *posix_substitut Line 421  static const pcre_uchar *posix_substitut
421    NULL,                 /* ^graph */    NULL,                 /* ^graph */
422    NULL,                 /* ^print */    NULL,                 /* ^print */
423    NULL,                 /* ^punct */    NULL,                 /* ^punct */
424    string_PXps,          /* ^space */   /* NOTE: Xps is POSIX space */    string_PXps,          /* ^space */  /* Xps is POSIX space, but from 8.34 */
425    string_PXwd,          /* ^word */    string_PXwd,          /* ^word */   /* Perl and POSIX space are the same */
426    NULL                  /* ^xdigit */    NULL                  /* ^xdigit */
427  };  };
428  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
# Line 462  static const char error_texts[] = Line 486  static const char error_texts[] =
486    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
487    "this version of PCRE is compiled without UTF support\0"    "this version of PCRE is compiled without UTF support\0"
488    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
489    "character value in \\x{...} sequence is too large\0"    "character value in \\x{} or \\o{} is too large\0"
490    /* 35 */    /* 35 */
491    "invalid condition (?(0)\0"    "invalid condition (?(0)\0"
492    "\\C not allowed in lookbehind assertion\0"    "\\C not allowed in lookbehind assertion\0"
# Line 516  static const char error_texts[] = Line 540  static const char error_texts[] =
540    "character value in \\u.... sequence is too large\0"    "character value in \\u.... sequence is too large\0"
541    "invalid UTF-32 string\0"    "invalid UTF-32 string\0"
542    "setting UTF is disabled by the application\0"    "setting UTF is disabled by the application\0"
543      "non-hex character in \\x{} (closing brace missing?)\0"
544      /* 80 */
545      "non-octal character in \\o{} (closing brace missing?)\0"
546      "missing opening brace after \\o\0"
547      "parentheses are too deeply nested\0"
548      "invalid range in character class\0"
549      "group name must start with a non-digit\0"
550      /* 85 */
551      "parentheses are too deeply nested (stack check)\0"
552      "digits missing in \\x{} or \\o{}\0"
553    ;    ;
554    
555  /* 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 773  static const pcre_uint8 posspropstab[3][ Line 807  static const pcre_uint8 posspropstab[3][
807    { ucp_L, ucp_N, ucp_P, ucp_Po }   /* WORD */    { ucp_L, ucp_N, ucp_P, ucp_Po }   /* WORD */
808  };  };
809    
810    /* This table is used when converting repeating opcodes into possessified
811    versions as a result of an explicit possessive quantifier such as ++. A zero
812    value means there is no possessified version - in those cases the item in
813    question must be wrapped in ONCE brackets. The table is truncated at OP_CALLOUT
814    because all relevant opcodes are less than that. */
815    
816    static const pcre_uint8 opcode_possessify[] = {
817      0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,   /* 0 - 15  */
818      0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,   /* 16 - 31 */
819    
820      0,                       /* NOTI */
821      OP_POSSTAR, 0,           /* STAR, MINSTAR */
822      OP_POSPLUS, 0,           /* PLUS, MINPLUS */
823      OP_POSQUERY, 0,          /* QUERY, MINQUERY */
824      OP_POSUPTO, 0,           /* UPTO, MINUPTO */
825      0,                       /* EXACT */
826      0, 0, 0, 0,              /* POS{STAR,PLUS,QUERY,UPTO} */
827    
828      OP_POSSTARI, 0,          /* STARI, MINSTARI */
829      OP_POSPLUSI, 0,          /* PLUSI, MINPLUSI */
830      OP_POSQUERYI, 0,         /* QUERYI, MINQUERYI */
831      OP_POSUPTOI, 0,          /* UPTOI, MINUPTOI */
832      0,                       /* EXACTI */
833      0, 0, 0, 0,              /* POS{STARI,PLUSI,QUERYI,UPTOI} */
834    
835      OP_NOTPOSSTAR, 0,        /* NOTSTAR, NOTMINSTAR */
836      OP_NOTPOSPLUS, 0,        /* NOTPLUS, NOTMINPLUS */
837      OP_NOTPOSQUERY, 0,       /* NOTQUERY, NOTMINQUERY */
838      OP_NOTPOSUPTO, 0,        /* NOTUPTO, NOTMINUPTO */
839      0,                       /* NOTEXACT */
840      0, 0, 0, 0,              /* NOTPOS{STAR,PLUS,QUERY,UPTO} */
841    
842      OP_NOTPOSSTARI, 0,       /* NOTSTARI, NOTMINSTARI */
843      OP_NOTPOSPLUSI, 0,       /* NOTPLUSI, NOTMINPLUSI */
844      OP_NOTPOSQUERYI, 0,      /* NOTQUERYI, NOTMINQUERYI */
845      OP_NOTPOSUPTOI, 0,       /* NOTUPTOI, NOTMINUPTOI */
846      0,                       /* NOTEXACTI */
847      0, 0, 0, 0,              /* NOTPOS{STARI,PLUSI,QUERYI,UPTOI} */
848    
849      OP_TYPEPOSSTAR, 0,       /* TYPESTAR, TYPEMINSTAR */
850      OP_TYPEPOSPLUS, 0,       /* TYPEPLUS, TYPEMINPLUS */
851      OP_TYPEPOSQUERY, 0,      /* TYPEQUERY, TYPEMINQUERY */
852      OP_TYPEPOSUPTO, 0,       /* TYPEUPTO, TYPEMINUPTO */
853      0,                       /* TYPEEXACT */
854      0, 0, 0, 0,              /* TYPEPOS{STAR,PLUS,QUERY,UPTO} */
855    
856      OP_CRPOSSTAR, 0,         /* CRSTAR, CRMINSTAR */
857      OP_CRPOSPLUS, 0,         /* CRPLUS, CRMINPLUS */
858      OP_CRPOSQUERY, 0,        /* CRQUERY, CRMINQUERY */
859      OP_CRPOSRANGE, 0,        /* CRRANGE, CRMINRANGE */
860      0, 0, 0, 0,              /* CRPOS{STAR,PLUS,QUERY,RANGE} */
861    
862      0, 0, 0,                 /* CLASS, NCLASS, XCLASS */
863      0, 0,                    /* REF, REFI */
864      0, 0,                    /* DNREF, DNREFI */
865      0, 0                     /* RECURSE, CALLOUT */
866    };
867    
868    
869    
870  /*************************************************  /*************************************************
# Line 879  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 971  return (*p == CHAR_RIGHT_CURLY_BRACKET);
971  *************************************************/  *************************************************/
972    
973  /* 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
974  positive value for a simple escape such as \n, or 0 for a data character  positive value for a simple escape such as \n, or 0 for a data character which
975  which will be placed in chptr. A backreference to group n is returned as  will be placed in chptr. A backreference to group n is returned as negative n.
976  negative n. When UTF-8 is enabled, a positive value greater than 255 may  When UTF-8 is enabled, a positive value greater than 255 may be returned in
977  be returned in chptr.  chptr. On entry, ptr is pointing at the \. On exit, it is on the final
978  On entry,ptr is pointing at the \. On exit, it is on the final character of the  character of the escape sequence.
 escape sequence.  
979    
980  Arguments:  Arguments:
981    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
982    chptr          points to the data character    chptr          points to a returned data character
983    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
984    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
985    options        the options bits    options        the options bits
# Line 1092  else Line 1183  else
1183      break;      break;
1184    
1185      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
1186      starting with one that is not zero is not straightforward. By experiment,      starting with one that is not zero is not straightforward. Perl has changed
1187      the way Perl works seems to be as follows:      over the years. Nowadays \g{} for backreferences and \o{} for octal are
1188        recommended to avoid the ambiguities in the old syntax.
1189    
1190      Outside a character class, the digits are read as a decimal number. If the      Outside a character class, the digits are read as a decimal number. If the
1191      number is less than 10, or if there are that many previous extracting      number is less than 8 (used to be 10), or if there are that many previous
1192      left brackets, then it is a back reference. Otherwise, up to three octal      extracting left brackets, then it is a back reference. Otherwise, up to
1193      digits are read to form an escaped byte. Thus \123 is likely to be octal      three octal digits are read to form an escaped byte. Thus \123 is likely to
1194      123 (cf \0123, which is octal 012 followed by the literal 3). If the octal      be octal 123 (cf \0123, which is octal 012 followed by the literal 3). If
1195      value is greater than 377, the least significant 8 bits are taken. Inside a      the octal value is greater than 377, the least significant 8 bits are
1196      character class, \ followed by a digit is always an octal number. */      taken. \8 and \9 are treated as the literal characters 8 and 9.
1197    
1198        Inside a character class, \ followed by a digit is always either a literal
1199        8 or 9 or an octal number. */
1200    
1201      case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5:      case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5:
1202      case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:      case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
# Line 1128  else Line 1223  else
1223          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1224          break;          break;
1225          }          }
1226        if (s < 10 || s <= bracount)        if (s < 8 || s <= bracount)  /* Check for back reference */
1227          {          {
1228          escape = -s;          escape = -s;
1229          break;          break;
# Line 1136  else Line 1231  else
1231        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
1232        }        }
1233    
1234      /* Handle an octal number following \. If the first digit is 8 or 9, Perl      /* Handle a digit following \ when the number is not a back reference. If
1235      generates a binary zero byte and treats the digit as a following literal.      the first digit is 8 or 9, Perl used to generate a binary zero byte and
1236      Thus we have to pull back the pointer by one. */      then treat the digit as a following literal. At least by Perl 5.18 this
1237        changed so as not to insert the binary zero. */
1238    
1239      if ((c = *ptr) >= CHAR_8)      if ((c = *ptr) >= CHAR_8) break;
1240        {  
1241        ptr--;      /* Fall through with a digit less than 8 */
       c = 0;  
       break;  
       }  
1242    
1243      /* \0 always starts an octal number, but we may drop through to here with a      /* \0 always starts an octal number, but we may drop through to here with a
1244      larger first octal digit. The original code used just to take the least      larger first octal digit. The original code used just to take the least
# Line 1162  else Line 1255  else
1255  #endif  #endif
1256      break;      break;
1257    
1258      /* \x is complicated. \x{ddd} is a character number which can be greater      /* \o is a relatively new Perl feature, supporting a more general way of
1259      than 0xff in utf or non-8bit mode, but only if the ddd are hex digits.      specifying character codes in octal. The only supported form is \o{ddd}. */
1260      If not, { is treated as a data character. */  
1261        case CHAR_o:
1262        if (ptr[1] != CHAR_LEFT_CURLY_BRACKET) *errorcodeptr = ERR81; else
1263        if (ptr[2] == CHAR_RIGHT_CURLY_BRACKET) *errorcodeptr = ERR86; else
1264          {
1265          ptr += 2;
1266          c = 0;
1267          overflow = FALSE;
1268          while (*ptr >= CHAR_0 && *ptr <= CHAR_7)
1269            {
1270            register pcre_uint32 cc = *ptr++;
1271            if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1272    #ifdef COMPILE_PCRE32
1273            if (c >= 0x20000000l) { overflow = TRUE; break; }
1274    #endif
1275            c = (c << 3) + cc - CHAR_0 ;
1276    #if defined COMPILE_PCRE8
1277            if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1278    #elif defined COMPILE_PCRE16
1279            if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1280    #elif defined COMPILE_PCRE32
1281            if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1282    #endif
1283            }
1284          if (overflow)
1285            {
1286            while (*ptr >= CHAR_0 && *ptr <= CHAR_7) ptr++;
1287            *errorcodeptr = ERR34;
1288            }
1289          else if (*ptr == CHAR_RIGHT_CURLY_BRACKET)
1290            {
1291            if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1292            }
1293          else *errorcodeptr = ERR80;
1294          }
1295        break;
1296    
1297        /* \x is complicated. In JavaScript, \x must be followed by two hexadecimal
1298        numbers. Otherwise it is a lowercase x letter. */
1299    
1300      case CHAR_x:      case CHAR_x:
1301      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1302        {        {
       /* In JavaScript, \x must be followed by two hexadecimal numbers.  
       Otherwise it is a lowercase x letter. */  
1303        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1304          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1305          {          {
# Line 1187  else Line 1316  else
1316  #endif  #endif
1317            }            }
1318          }          }
1319        break;        }    /* End JavaScript handling */
       }  
1320    
1321      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      /* Handle \x in Perl's style. \x{ddd} is a character number which can be
1322        {      greater than 0xff in utf or non-8bit mode, but only if the ddd are hex
1323        const pcre_uchar *pt = ptr + 2;      digits. If not, { used to be treated as a data character. However, Perl
1324        seems to read hex digits up to the first non-such, and ignore the rest, so
1325        that, for example \x{zz} matches a binary zero. This seems crazy, so PCRE
1326        now gives an error. */
1327    
1328        c = 0;      else
1329        overflow = FALSE;        {
1330        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1331          {          {
1332          register pcre_uint32 cc = *pt++;          ptr += 2;
1333          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (*ptr == CHAR_RIGHT_CURLY_BRACKET)
1334              {
1335              *errorcodeptr = ERR86;
1336              break;
1337              }
1338            c = 0;
1339            overflow = FALSE;
1340            while (MAX_255(*ptr) && (digitab[*ptr] & ctype_xdigit) != 0)
1341              {
1342              register pcre_uint32 cc = *ptr++;
1343              if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1344    
1345  #ifdef COMPILE_PCRE32  #ifdef COMPILE_PCRE32
1346          if (c >= 0x10000000l) { overflow = TRUE; break; }            if (c >= 0x10000000l) { overflow = TRUE; break; }
1347  #endif  #endif
1348    
1349  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1350          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1351          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1352  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
1353          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */            if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1354          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1355  #endif  #endif
1356    
1357  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
1358          if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }            if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1359  #elif defined COMPILE_PCRE16  #elif defined COMPILE_PCRE16
1360          if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }            if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1361  #elif defined COMPILE_PCRE32  #elif defined COMPILE_PCRE32
1362          if (utf && c > 0x10ffffU) { overflow = TRUE; break; }            if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1363  #endif  #endif
1364          }            }
1365    
1366        if (overflow)          if (overflow)
1367          {            {
1368          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;            while (MAX_255(*ptr) && (digitab[*ptr] & ctype_xdigit) != 0) ptr++;
1369          *errorcodeptr = ERR34;            *errorcodeptr = ERR34;
1370          }            }
1371    
1372        if (*pt == CHAR_RIGHT_CURLY_BRACKET)          else if (*ptr == CHAR_RIGHT_CURLY_BRACKET)
1373          {            {
1374          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;            if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1375          ptr = pt;            }
         break;  
         }  
1376    
1377        /* If the sequence of hex digits does not end with '}', then we don't          /* If the sequence of hex digits does not end with '}', give an error.
1378        recognize this construct; fall through to the normal \x handling. */          We used just to recognize this construct and fall through to the normal
1379        }          \x handling, but nowadays Perl gives an error, which seems much more
1380            sensible, so we do too. */
1381    
1382      /* Read just a single-byte hex-defined char */          else *errorcodeptr = ERR79;
1383            }   /* End of \x{} processing */
1384    
1385      c = 0;        /* Read a single-byte hex-defined char (up to two hex digits after \x) */
1386      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)  
1387        {        else
1388        pcre_uint32 cc;                          /* Some compilers don't like */          {
1389        cc = *(++ptr);                           /* ++ in initializers */          c = 0;
1390            while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1391              {
1392              pcre_uint32 cc;                          /* Some compilers don't like */
1393              cc = *(++ptr);                           /* ++ in initializers */
1394  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1395        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
1396        c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1397  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
1398        if (cc <= CHAR_z) cc += 64;              /* Convert to upper case */            if (cc <= CHAR_z) cc += 64;              /* Convert to upper case */
1399        c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1400  #endif  #endif
1401        }            }
1402            }     /* End of \xdd handling */
1403          }       /* End of Perl-style \x handling */
1404      break;      break;
1405    
1406      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.
# Line 1443  read_repeat_counts(const pcre_uchar *p, Line 1590  read_repeat_counts(const pcre_uchar *p,
1590  int min = 0;  int min = 0;
1591  int max = -1;  int max = -1;
1592    
1593  /* Read the minimum value and do a paranoid check: a negative value indicates  while (IS_DIGIT(*p))
 an integer overflow. */  
   
 while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);  
 if (min < 0 || min > 65535)  
1594    {    {
1595    *errorcodeptr = ERR5;    min = min * 10 + (int)(*p++ - CHAR_0);
1596    return p;    if (min > 65535)
1597        {
1598        *errorcodeptr = ERR5;
1599        return p;
1600        }
1601    }    }
1602    
 /* Read the maximum value if there is one, and again do a paranoid on its size.  
 Also, max must not be less than min. */  
   
1603  if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else  if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else
1604    {    {
1605    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1606      {      {
1607      max = 0;      max = 0;
1608      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);      while(IS_DIGIT(*p))
     if (max < 0 || max > 65535)  
1609        {        {
1610        *errorcodeptr = ERR5;        max = max * 10 + (int)(*p++ - CHAR_0);
1611        return p;        if (max > 65535)
1612            {
1613            *errorcodeptr = ERR5;
1614            return p;
1615            }
1616        }        }
1617      if (max < min)      if (max < min)
1618        {        {
# Line 1475  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1622  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1622      }      }
1623    }    }
1624    
 /* Fill in the required variables, and pass back the pointer to the terminating  
 '}'. */  
   
1625  *minp = min;  *minp = min;
1626  *maxp = max;  *maxp = max;
1627  return p;  return p;
# Line 1524  for (;;) Line 1668  for (;;)
1668    
1669      case OP_CALLOUT:      case OP_CALLOUT:
1670      case OP_CREF:      case OP_CREF:
1671      case OP_NCREF:      case OP_DNCREF:
1672      case OP_RREF:      case OP_RREF:
1673      case OP_NRREF:      case OP_DNRREF:
1674      case OP_DEF:      case OP_DEF:
1675      code += PRIV(OP_lengths)[*code];      code += PRIV(OP_lengths)[*code];
1676      break;      break;
# Line 1663  for (;;) Line 1807  for (;;)
1807      case OP_COMMIT:      case OP_COMMIT:
1808      case OP_CREF:      case OP_CREF:
1809      case OP_DEF:      case OP_DEF:
1810        case OP_DNCREF:
1811        case OP_DNRREF:
1812      case OP_DOLL:      case OP_DOLL:
1813      case OP_DOLLM:      case OP_DOLLM:
1814      case OP_EOD:      case OP_EOD:
1815      case OP_EODN:      case OP_EODN:
1816      case OP_FAIL:      case OP_FAIL:
     case OP_NCREF:  
     case OP_NRREF:  
1817      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1818      case OP_PRUNE:      case OP_PRUNE:
1819      case OP_REVERSE:      case OP_REVERSE:
# Line 1764  for (;;) Line 1908  for (;;)
1908    
1909      switch (*cc)      switch (*cc)
1910        {        {
       case OP_CRPLUS:  
       case OP_CRMINPLUS:  
1911        case OP_CRSTAR:        case OP_CRSTAR:
1912        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1913          case OP_CRPLUS:
1914          case OP_CRMINPLUS:
1915        case OP_CRQUERY:        case OP_CRQUERY:
1916        case OP_CRMINQUERY:        case OP_CRMINQUERY:
1917          case OP_CRPOSSTAR:
1918          case OP_CRPOSPLUS:
1919          case OP_CRPOSQUERY:
1920        return -1;        return -1;
1921    
1922        case OP_CRRANGE:        case OP_CRRANGE:
1923        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1924          case OP_CRPOSRANGE:
1925        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1926        branchlength += (int)GET2(cc,1);        branchlength += (int)GET2(cc,1);
1927        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
# Line 2366  for (code = first_significant_code(code Line 2514  for (code = first_significant_code(code
2514        case OP_CRMINSTAR:        case OP_CRMINSTAR:
2515        case OP_CRQUERY:        case OP_CRQUERY:
2516        case OP_CRMINQUERY:        case OP_CRMINQUERY:
2517          case OP_CRPOSSTAR:
2518          case OP_CRPOSQUERY:
2519        break;        break;
2520    
2521        default:                   /* Non-repeat => class must match */        default:                   /* Non-repeat => class must match */
2522        case OP_CRPLUS:            /* These repeats aren't empty */        case OP_CRPLUS:            /* These repeats aren't empty */
2523        case OP_CRMINPLUS:        case OP_CRMINPLUS:
2524          case OP_CRPOSPLUS:
2525        return FALSE;        return FALSE;
2526    
2527        case OP_CRRANGE:        case OP_CRRANGE:
2528        case OP_CRMINRANGE:        case OP_CRMINRANGE:
2529          case OP_CRPOSRANGE:
2530        if (GET2(ccode, 1) > 0) return FALSE;  /* Minimum > 0 */        if (GET2(ccode, 1) > 0) return FALSE;  /* Minimum > 0 */
2531        break;        break;
2532        }        }
# Line 2650  switch(ptype) Line 2802  switch(ptype)
2802    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2803            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
2804    
2805    case PT_SPACE:    /* Perl space */    /* Perl space used to exclude VT, but from Perl 5.18 it is included, which
2806    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||    means that Perl space and POSIX space are now identical. PCRE was changed
2807            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)    at release 8.34. */
           == negated;  
2808    
2809      case PT_SPACE:    /* Perl space */
2810    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
2811    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||    switch(c)
2812            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||      {
2813            c == CHAR_FF || c == CHAR_CR)      HSPACE_CASES:
2814            == negated;      VSPACE_CASES:
2815        return negated;
2816    
2817        default:
2818        return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == negated;
2819        }
2820      break;  /* Control never reaches here */
2821    
2822    case PT_WORD:    case PT_WORD:
2823    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
# Line 2708  get_chr_property_list(const pcre_uchar * Line 2866  get_chr_property_list(const pcre_uchar *
2866    const pcre_uint8 *fcc, pcre_uint32 *list)    const pcre_uint8 *fcc, pcre_uint32 *list)
2867  {  {
2868  pcre_uchar c = *code;  pcre_uchar c = *code;
2869    pcre_uchar base;
2870  const pcre_uchar *end;  const pcre_uchar *end;
 const pcre_uint32 *clist_src;  
 pcre_uint32 *clist_dest;  
2871  pcre_uint32 chr;  pcre_uint32 chr;
2872  pcre_uchar base;  
2873    #ifdef SUPPORT_UCP
2874    pcre_uint32 *clist_dest;
2875    const pcre_uint32 *clist_src;
2876    #else
2877    utf = utf;  /* Suppress "unused parameter" compiler warning */
2878    #endif
2879    
2880  list[0] = c;  list[0] = c;
2881  list[1] = FALSE;  list[1] = FALSE;
# Line 2818  switch(c) Line 2981  switch(c)
2981      return code + 2;      return code + 2;
2982      }      }
2983    
2984    /* Convert only if we have anough space. */    /* Convert only if we have enough space. */
2985    
2986    clist_src = PRIV(ucd_caseless_sets) + code[1];    clist_src = PRIV(ucd_caseless_sets) + code[1];
2987    clist_dest = list + 2;    clist_dest = list + 2;
2988    code += 2;    code += 2;
2989    
2990    do {    do {
      /* Early return if there is not enough space. */  
2991       if (clist_dest >= list + 8)       if (clist_dest >= list + 8)
2992         {         {
2993           /* Early return if there is not enough space. This should never
2994           happen, since all clists are shorter than 5 character now. */
2995         list[2] = code[0];         list[2] = code[0];
2996         list[3] = code[1];         list[3] = code[1];
2997         return code;         return code;
2998         }         }
2999       *clist_dest++ = *clist_src;       *clist_dest++ = *clist_src;
3000       }       }
3001     while(*clist_src++ != NOTACHAR);    while(*clist_src++ != NOTACHAR);
3002    
3003    /* Enough space to store all characters. */    /* All characters are stored. The terminating NOTACHAR
3004      is copied form the clist itself. */
3005    
3006    list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT;    list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT;
3007    return code;    return code;
# Line 2846  switch(c) Line 3011  switch(c)
3011    case OP_CLASS:    case OP_CLASS:
3012  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3013    case OP_XCLASS:    case OP_XCLASS:
   
3014    if (c == OP_XCLASS)    if (c == OP_XCLASS)
3015      end = code + GET(code, 0);      end = code + GET(code, 0) - 1;
3016    else    else
3017  #endif  #endif
3018      end = code + 32 / sizeof(pcre_uchar);      end = code + 32 / sizeof(pcre_uchar);
# Line 2859  switch(c) Line 3023  switch(c)
3023      case OP_CRMINSTAR:      case OP_CRMINSTAR:
3024      case OP_CRQUERY:      case OP_CRQUERY:
3025      case OP_CRMINQUERY:      case OP_CRMINQUERY:
3026        case OP_CRPOSSTAR:
3027        case OP_CRPOSQUERY:
3028      list[1] = TRUE;      list[1] = TRUE;
3029      end++;      end++;
3030      break;      break;
3031    
3032        case OP_CRPLUS:
3033        case OP_CRMINPLUS:
3034        case OP_CRPOSPLUS:
3035        end++;
3036        break;
3037    
3038      case OP_CRRANGE:      case OP_CRRANGE:
3039      case OP_CRMINRANGE:      case OP_CRMINRANGE:
3040        case OP_CRPOSRANGE:
3041      list[1] = (GET2(end, 1) == 0);      list[1] = (GET2(end, 1) == 0);
3042      end += 1 + 2 * IMM2_SIZE;      end += 1 + 2 * IMM2_SIZE;
3043      break;      break;
3044      }      }
3045    list[2] = end - code;    list[2] = (pcre_uint32)(end - code);
3046    return end;    return end;
3047    }    }
3048  return NULL;    /* Opcode not accepted */  return NULL;    /* Opcode not accepted */
# Line 2895  Returns:      TRUE if the auto-possessif Line 3068  Returns:      TRUE if the auto-possessif
3068    
3069  static BOOL  static BOOL
3070  compare_opcodes(const pcre_uchar *code, BOOL utf, const compile_data *cd,  compare_opcodes(const pcre_uchar *code, BOOL utf, const compile_data *cd,
3071    const pcre_uint32* base_list)    const pcre_uint32 *base_list, const pcre_uchar *base_end)
3072  {  {
3073  pcre_uchar c;  pcre_uchar c;
3074  pcre_uint32 list[8];  pcre_uint32 list[8];
3075  const pcre_uint32* chr_ptr;  const pcre_uint32 *chr_ptr;
3076  const pcre_uint32* ochr_ptr;  const pcre_uint32 *ochr_ptr;
3077  const pcre_uint32* list_ptr;  const pcre_uint32 *list_ptr;
3078    const pcre_uchar *next_code;
3079    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3080    const pcre_uchar *xclass_flags;
3081    #endif
3082    const pcre_uint8 *class_bitset;
3083    const pcre_uint8 *set1, *set2, *set_end;
3084  pcre_uint32 chr;  pcre_uint32 chr;
3085    BOOL accepted, invert_bits;
3086    BOOL entered_a_group = FALSE;
3087    
3088    /* Note: the base_list[1] contains whether the current opcode has greedy
3089    (represented by a non-zero value) quantifier. This is a different from
3090    other character type lists, which stores here that the character iterator
3091    matches to an empty string (also represented by a non-zero value). */
3092    
3093  for(;;)  for(;;)
3094    {    {
3095      /* All operations move the code pointer forward.
3096      Therefore infinite recursions are not possible. */
3097    
3098    c = *code;    c = *code;
3099    
3100    /* Skip over callouts */    /* Skip over callouts */
# Line 2925  for(;;) Line 3114  for(;;)
3114    switch(c)    switch(c)
3115      {      {
3116      case OP_END:      case OP_END:
3117      /* TRUE only in greedy case. The non-greedy case could be replaced by an      case OP_KETRPOS:
3118      OP_EXACT, but it is probably not worth it. (And note that OP_EXACT uses      /* TRUE only in greedy case. The non-greedy case could be replaced by
3119      more memory, which we cannot get at this stage.) */      an OP_EXACT, but it is probably not worth it. (And note that OP_EXACT
3120        uses more memory, which we cannot get at this stage.) */
3121    
3122      return base_list[1] != 0;      return base_list[1] != 0;
3123    
3124      case OP_KET:      case OP_KET:
3125      /* If the bracket is capturing, and referenced by an OP_RECURSE, the      /* If the bracket is capturing, and referenced by an OP_RECURSE, or
3126      non-greedy case cannot be converted to a possessive form. We do not test      it is an atomic sub-pattern (assert, once, etc.) the non-greedy case
3127      the bracket type at the moment, but we might do it in the future to improve      cannot be converted to a possessive form. */
     this condition. (But note that recursive calls are always atomic.) */  
3128    
3129      if (base_list[1] == 0) return FALSE;      if (base_list[1] == 0) return FALSE;
3130    
3131        switch(*(code - GET(code, 1)))
3132          {
3133          case OP_ASSERT:
3134          case OP_ASSERT_NOT:
3135          case OP_ASSERTBACK:
3136          case OP_ASSERTBACK_NOT:
3137          case OP_ONCE:
3138          case OP_ONCE_NC:
3139          /* Atomic sub-patterns and assertions can always auto-possessify their
3140          last iterator. However, if the group was entered as a result of checking
3141          a previous iterator, this is not possible. */
3142    
3143          return !entered_a_group;
3144          }
3145    
3146        code += PRIV(OP_lengths)[c];
3147        continue;
3148    
3149        case OP_ONCE:
3150        case OP_ONCE_NC:
3151        case OP_BRA:
3152        case OP_CBRA:
3153        next_code = code + GET(code, 1);
3154        code += PRIV(OP_lengths)[c];
3155    
3156        while (*next_code == OP_ALT)
3157          {
3158          if (!compare_opcodes(code, utf, cd, base_list, base_end)) return FALSE;
3159          code = next_code + 1 + LINK_SIZE;
3160          next_code += GET(next_code, 1);
3161          }
3162    
3163        entered_a_group = TRUE;
3164        continue;
3165    
3166        case OP_BRAZERO:
3167        case OP_BRAMINZERO:
3168    
3169        next_code = code + 1;
3170        if (*next_code != OP_BRA && *next_code != OP_CBRA
3171            && *next_code != OP_ONCE && *next_code != OP_ONCE_NC) return FALSE;
3172    
3173        do next_code += GET(next_code, 1); while (*next_code == OP_ALT);
3174    
3175        /* The bracket content will be checked by the
3176        OP_BRA/OP_CBRA case above. */
3177        next_code += 1 + LINK_SIZE;
3178        if (!compare_opcodes(next_code, utf, cd, base_list, base_end))
3179          return FALSE;
3180    
3181      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
3182      continue;      continue;
3183    
3184        default:
3185        break;
3186      }      }
3187    
3188    /* Check for a supported opcode, and load its properties. */    /* Check for a supported opcode, and load its properties. */
# Line 2961  for(;;) Line 3204  for(;;)
3204      list_ptr = base_list;      list_ptr = base_list;
3205      }      }
3206    
3207      /* Character bitsets can also be compared to certain opcodes. */
3208    
3209      else if (base_list[0] == OP_CLASS || list[0] == OP_CLASS
3210    #ifdef COMPILE_PCRE8
3211          /* In 8 bit, non-UTF mode, OP_CLASS and OP_NCLASS are the same. */
3212          || (!utf && (base_list[0] == OP_NCLASS || list[0] == OP_NCLASS))
3213    #endif
3214          )
3215        {
3216    #ifdef COMPILE_PCRE8
3217        if (base_list[0] == OP_CLASS || (!utf && base_list[0] == OP_NCLASS))
3218    #else
3219        if (base_list[0] == OP_CLASS)
3220    #endif
3221          {
3222          set1 = (pcre_uint8 *)(base_end - base_list[2]);
3223          list_ptr = list;
3224          }
3225        else
3226          {
3227          set1 = (pcre_uint8 *)(code - list[2]);
3228          list_ptr = base_list;
3229          }
3230    
3231        invert_bits = FALSE;
3232        switch(list_ptr[0])
3233          {
3234          case OP_CLASS:
3235          case OP_NCLASS:
3236          set2 = (pcre_uint8 *)
3237            ((list_ptr == list ? code : base_end) - list_ptr[2]);
3238          break;
3239    
3240    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3241          case OP_XCLASS:
3242          xclass_flags = (list_ptr == list ? code : base_end) - list_ptr[2] + LINK_SIZE;
3243          if ((*xclass_flags & XCL_HASPROP) != 0) return FALSE;
3244          if ((*xclass_flags & XCL_MAP) == 0)
3245            {
3246            /* No bits are set for characters < 256. */
3247            if (list[1] == 0) return TRUE;
3248            /* Might be an empty repeat. */
3249            continue;
3250            }
3251          set2 = (pcre_uint8 *)(xclass_flags + 1);
3252          break;
3253    #endif
3254    
3255          case OP_NOT_DIGIT:
3256          invert_bits = TRUE;
3257          /* Fall through */
3258          case OP_DIGIT:
3259          set2 = (pcre_uint8 *)(cd->cbits + cbit_digit);
3260          break;
3261    
3262          case OP_NOT_WHITESPACE:
3263          invert_bits = TRUE;
3264          /* Fall through */
3265          case OP_WHITESPACE:
3266          set2 = (pcre_uint8 *)(cd->cbits + cbit_space);
3267          break;
3268    
3269          case OP_NOT_WORDCHAR:
3270          invert_bits = TRUE;
3271          /* Fall through */
3272          case OP_WORDCHAR:
3273          set2 = (pcre_uint8 *)(cd->cbits + cbit_word);
3274          break;
3275    
3276          default:
3277          return FALSE;
3278          }
3279    
3280        /* Because the sets are unaligned, we need
3281        to perform byte comparison here. */
3282        set_end = set1 + 32;
3283        if (invert_bits)
3284          {
3285          do
3286            {
3287            if ((*set1++ & ~(*set2++)) != 0) return FALSE;
3288            }
3289          while (set1 < set_end);
3290          }
3291        else
3292          {
3293          do
3294            {
3295            if ((*set1++ & *set2++) != 0) return FALSE;
3296            }
3297          while (set1 < set_end);
3298          }
3299    
3300        if (list[1] == 0) return TRUE;
3301        /* Might be an empty repeat. */
3302        continue;
3303        }
3304    
3305    /* Some property combinations also acceptable. Unicode property opcodes are    /* Some property combinations also acceptable. Unicode property opcodes are
3306    processed specially; the rest can be handled with a lookup table. */    processed specially; the rest can be handled with a lookup table. */
3307    
# Line 2968  for(;;) Line 3309  for(;;)
3309      {      {
3310      pcre_uint32 leftop, rightop;      pcre_uint32 leftop, rightop;
3311    
     if (list[1] != 0) return FALSE;   /* Must match at least one character */  
3312      leftop = base_list[0];      leftop = base_list[0];
3313      rightop = list[0];      rightop = list[0];
3314    
3315  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3316        accepted = FALSE; /* Always set in non-unicode case. */
3317      if (leftop == OP_PROP || leftop == OP_NOTPROP)      if (leftop == OP_PROP || leftop == OP_NOTPROP)
3318        {        {
3319        if (rightop == OP_EOD) return TRUE;        if (rightop == OP_EOD)
3320        if (rightop == OP_PROP || rightop == OP_NOTPROP)          accepted = TRUE;
3321          else if (rightop == OP_PROP || rightop == OP_NOTPROP)
3322          {          {
3323          int n;          int n;
3324          const pcre_uint8 *p;          const pcre_uint8 *p;
# Line 2997  for(;;) Line 3339  for(;;)
3339          n = propposstab[base_list[2]][list[2]];          n = propposstab[base_list[2]][list[2]];
3340          switch(n)          switch(n)
3341            {            {
3342            case 0: return FALSE;            case 0: break;
3343            case 1: return bothprop;            case 1: accepted = bothprop; break;
3344            case 2: return (base_list[3] == list[3]) != same;            case 2: accepted = (base_list[3] == list[3]) != same; break;
3345            case 3: return !same;            case 3: accepted = !same; break;
3346    
3347            case 4:  /* Left general category, right particular category */            case 4:  /* Left general category, right particular category */
3348            return risprop && catposstab[base_list[3]][list[3]] == same;            accepted = risprop && catposstab[base_list[3]][list[3]] == same;
3349              break;
3350    
3351            case 5:  /* Right general category, left particular category */            case 5:  /* Right general category, left particular category */
3352            return lisprop && catposstab[list[3]][base_list[3]] == same;            accepted = lisprop && catposstab[list[3]][base_list[3]] == same;
3353              break;
3354    
3355            /* This code is logically tricky. Think hard before fiddling with it.            /* This code is logically tricky. Think hard before fiddling with it.
3356            The posspropstab table has four entries per row. Each row relates to            The posspropstab table has four entries per row. Each row relates to
3357            one of PCRE's special properties such as ALNUM or SPACE or WORD.            one of PCRE's special properties such as ALNUM or SPACE or WORD.
3358            Only WORD actually needs all four entries, but using repeats for the            Only WORD actually needs all four entries, but using repeats for the
3359            others means they can all use the same code below.            others means they can all use the same code below.
3360    
3361            The first two entries in each row are Unicode general categories, and            The first two entries in each row are Unicode general categories, and
3362            apply always, because all the characters they include are part of the            apply always, because all the characters they include are part of the
3363            PCRE character set. The third and fourth entries are a general and a            PCRE character set. The third and fourth entries are a general and a
# Line 3023  for(;;) Line 3367  for(;;)
3367            category contains more characters than the specials that are defined            category contains more characters than the specials that are defined
3368            for the property being tested against. Therefore, it cannot be used            for the property being tested against. Therefore, it cannot be used
3369            in a NOTPROP case.            in a NOTPROP case.
3370    
3371            Example: the row for WORD contains ucp_L, ucp_N, ucp_P, ucp_Po.            Example: the row for WORD contains ucp_L, ucp_N, ucp_P, ucp_Po.
3372            Underscore is covered by ucp_P or ucp_Po. */            Underscore is covered by ucp_P or ucp_Po. */
3373    
# Line 3031  for(;;) Line 3375  for(;;)
3375            case 7:  /* Left space vs right general category */            case 7:  /* Left space vs right general category */
3376            case 8:  /* Left word vs right general category */            case 8:  /* Left word vs right general category */
3377            p = posspropstab[n-6];            p = posspropstab[n-6];
3378            return risprop && lisprop ==            accepted = risprop && lisprop ==
3379              (list[3] != p[0] &&              (list[3] != p[0] &&
3380               list[3] != p[1] &&               list[3] != p[1] &&
3381              (list[3] != p[2] || !lisprop));              (list[3] != p[2] || !lisprop));
3382              break;
3383    
3384            case 9:   /* Right alphanum vs left general category */            case 9:   /* Right alphanum vs left general category */
3385            case 10:  /* Right space vs left general category */            case 10:  /* Right space vs left general category */
3386            case 11:  /* Right word vs left general category */            case 11:  /* Right word vs left general category */
3387            p = posspropstab[n-9];            p = posspropstab[n-9];
3388            return lisprop && risprop ==            accepted = lisprop && risprop ==
3389              (base_list[3] != p[0] &&              (base_list[3] != p[0] &&
3390               base_list[3] != p[1] &&               base_list[3] != p[1] &&
3391              (base_list[3] != p[2] || !risprop));              (base_list[3] != p[2] || !risprop));
3392              break;
3393    
3394            case 12:  /* Left alphanum vs right particular category */            case 12:  /* Left alphanum vs right particular category */
3395            case 13:  /* Left space vs right particular category */            case 13:  /* Left space vs right particular category */
3396            case 14:  /* Left word vs right particular category */            case 14:  /* Left word vs right particular category */
3397            p = posspropstab[n-12];            p = posspropstab[n-12];
3398            return risprop && lisprop ==            accepted = risprop && lisprop ==
3399              (catposstab[p[0]][list[3]] &&              (catposstab[p[0]][list[3]] &&
3400               catposstab[p[1]][list[3]] &&               catposstab[p[1]][list[3]] &&
3401              (list[3] != p[3] || !lisprop));              (list[3] != p[3] || !lisprop));
3402              break;
3403    
3404            case 15:  /* Right alphanum vs left particular category */            case 15:  /* Right alphanum vs left particular category */
3405            case 16:  /* Right space vs left particular category */            case 16:  /* Right space vs left particular category */
3406            case 17:  /* Right word vs left particular category */            case 17:  /* Right word vs left particular category */
3407            p = posspropstab[n-15];            p = posspropstab[n-15];
3408            return lisprop && risprop ==            accepted = lisprop && risprop ==
3409              (catposstab[p[0]][base_list[3]] &&              (catposstab[p[0]][base_list[3]] &&
3410               catposstab[p[1]][base_list[3]] &&               catposstab[p[1]][base_list[3]] &&
3411              (base_list[3] != p[3] || !risprop));              (base_list[3] != p[3] || !risprop));
3412              break;
3413            }            }
3414          }          }
       return FALSE;  
3415        }        }
3416    
3417      else      else
3418  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3419    
3420      return leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP &&      accepted = leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP &&
3421             rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP &&             rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP &&
3422             autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP];             autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP];
3423    
3424        if (!accepted) return FALSE;
3425    
3426        if (list[1] == 0) return TRUE;
3427        /* Might be an empty repeat. */
3428        continue;
3429      }      }
3430    
3431    /* Control reaches here only if one of the items is a small character list.    /* Control reaches here only if one of the items is a small character list.
# Line 3186  for(;;) Line 3539  for(;;)
3539        case OP_EOD:    /* Can always possessify before \z */        case OP_EOD:    /* Can always possessify before \z */
3540        break;        break;
3541    
3542    #ifdef SUPPORT_UCP
3543        case OP_PROP:        case OP_PROP:
3544        case OP_NOTPROP:        case OP_NOTPROP:
3545        if (!check_char_prop(chr, list_ptr[2], list_ptr[3],        if (!check_char_prop(chr, list_ptr[2], list_ptr[3],
3546              list_ptr[0] == OP_NOTPROP))              list_ptr[0] == OP_NOTPROP))
3547          return FALSE;          return FALSE;
3548        break;        break;
3549    #endif
       /* The class comparisons work only when the class is the second item  
       of the pair, because there are at present no possessive forms of the  
       class opcodes. Note also that the "code" variable that is used below  
       points after the second item, and that the pointer for the first item  
       is not available, so even if there were possessive forms of the class  
       opcodes, the correct comparison could not be done. */  
3550    
3551        case OP_NCLASS:        case OP_NCLASS:
3552        if (chr > 255) return FALSE;        if (chr > 255) return FALSE;
3553        /* Fall through */        /* Fall through */
3554    
3555        case OP_CLASS:        case OP_CLASS:
       if (list_ptr != list) return FALSE;   /* Class is first opcode */  
3556        if (chr > 255) break;        if (chr > 255) break;
3557        if ((((pcre_uint8 *)(code - list_ptr[2] + 1))[chr >> 3] & (1 << (chr & 7))) != 0)        class_bitset = (pcre_uint8 *)
3558          return FALSE;          ((list_ptr == list ? code : base_end) - list_ptr[2]);
3559          if ((class_bitset[chr >> 3] & (1 << (chr & 7))) != 0) return FALSE;
3560        break;        break;
3561    
3562  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3563        case OP_XCLASS:        case OP_XCLASS:
3564        if (list_ptr != list) return FALSE;   /* Class is first opcode */        if (PRIV(xclass)(chr, (list_ptr == list ? code : base_end) -
3565        if (PRIV(xclass)(chr, code - list_ptr[2] + 1 + LINK_SIZE, utf))            list_ptr[2] + LINK_SIZE, utf)) return FALSE;
         return FALSE;  
3566        break;        break;
3567  #endif  #endif
3568    
# Line 3232  for(;;) Line 3579  for(;;)
3579    if (list[1] == 0) return TRUE;    if (list[1] == 0) return TRUE;
3580    }    }
3581    
3582  return FALSE;  /* Control never reaches here. There used to be a fail-save return FALSE; here,
3583    but some compilers complain about an unreachable statement. */
3584    
3585  }  }
3586    
3587    
# Line 3257  auto_possessify(pcre_uchar *code, BOOL u Line 3606  auto_possessify(pcre_uchar *code, BOOL u
3606  {  {
3607  register pcre_uchar c;  register pcre_uchar c;
3608  const pcre_uchar *end;  const pcre_uchar *end;
3609    pcre_uchar *repeat_opcode;
3610  pcre_uint32 list[8];  pcre_uint32 list[8];
3611    
3612  for (;;)  for (;;)
# Line 3270  for (;;) Line 3620  for (;;)
3620        get_chr_property_list(code, utf, cd->fcc, list) : NULL;        get_chr_property_list(code, utf, cd->fcc, list) : NULL;
3621      list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO;      list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO;
3622    
3623      if (end != NULL && compare_opcodes(end, utf, cd, list))      if (end != NULL && compare_opcodes(end, utf, cd, list, end))
3624        {        {
3625        switch(c)        switch(c)
3626          {          {
# Line 3303  for (;;) Line 3653  for (;;)
3653          break;          break;
3654    
3655          case OP_MINUPTO:          case OP_MINUPTO:
3656          *code += OP_MINUPTO - OP_UPTO;          *code += OP_POSUPTO - OP_MINUPTO;
3657          break;          break;
3658          }          }
3659        }        }
3660      c = *code;      c = *code;
3661      }      }
3662      else if (c == OP_CLASS || c == OP_NCLASS || c == OP_XCLASS)
3663        {
3664    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3665        if (c == OP_XCLASS)
3666          repeat_opcode = code + GET(code, 1);
3667        else
3668    #endif
3669          repeat_opcode = code + 1 + (32 / sizeof(pcre_uchar));
3670    
3671        c = *repeat_opcode;
3672        if (c >= OP_CRSTAR && c <= OP_CRMINRANGE)
3673          {
3674          /* end must not be NULL. */
3675          end = get_chr_property_list(code, utf, cd->fcc, list);
3676    
3677          list[1] = (c & 1) == 0;
3678    
3679          if (compare_opcodes(end, utf, cd, list, end))
3680            {
3681            switch (c)
3682              {
3683              case OP_CRSTAR:
3684              case OP_CRMINSTAR:
3685              *repeat_opcode = OP_CRPOSSTAR;
3686              break;
3687    
3688              case OP_CRPLUS:
3689              case OP_CRMINPLUS:
3690              *repeat_opcode = OP_CRPOSPLUS;
3691              break;
3692    
3693              case OP_CRQUERY:
3694              case OP_CRMINQUERY:
3695              *repeat_opcode = OP_CRPOSQUERY;
3696              break;
3697    
3698              case OP_CRRANGE:
3699              case OP_CRMINRANGE:
3700              *repeat_opcode = OP_CRPOSRANGE;
3701              break;
3702              }
3703            }
3704          }
3705        c = *code;
3706        }
3707    
3708    switch(c)    switch(c)
3709      {      {
# Line 3335  for (;;) Line 3730  for (;;)
3730        code += 2;        code += 2;
3731      break;      break;
3732    
3733    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3734      case OP_XCLASS:      case OP_XCLASS:
3735      code += GET(code, 1);      code += GET(code, 1);
3736      break;      break;
3737    #endif
3738    
3739      case OP_MARK:      case OP_MARK:
3740      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
# Line 3446  class, but [abc[:x\]pqr:]] is (so that a Line 3843  class, but [abc[:x\]pqr:]] is (so that a
3843  below handles the special case of \], but does not try to do any other escape  below handles the special case of \], but does not try to do any other escape
3844  processing. This makes it different from Perl for cases such as [:l\ower:]  processing. This makes it different from Perl for cases such as [:l\ower:]
3845  where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize  where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
3846  "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,  "l\ower". This is a lesser evil than not diagnosing bad classes when Perl does,
3847  I think.  I think.
3848    
3849  A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.  A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.
# Line 3695  for (c = *cptr; c <= d; c++) Line 4092  for (c = *cptr; c <= d; c++)
4092    
4093  if (c > d) return -1;  /* Reached end of range */  if (c > d) return -1;  /* Reached end of range */
4094    
4095    /* Found a character that has a single other case. Search for the end of the
4096    range, which is either the end of the input range, or a character that has zero
4097    or more than one other cases. */
4098    
4099  *ocptr = othercase;  *ocptr = othercase;
4100  next = othercase + 1;  next = othercase + 1;
4101    
4102  for (++c; c <= d; c++)  for (++c; c <= d; c++)
4103    {    {
4104    if (UCD_OTHERCASE(c) != next) break;    if ((co = UCD_CASESET(c)) != 0 || UCD_OTHERCASE(c) != next) break;
4105    next++;    next++;
4106    }    }
4107    
# Line 3738  add_to_class(pcre_uint8 *classbits, pcre Line 4139  add_to_class(pcre_uint8 *classbits, pcre
4139    compile_data *cd, pcre_uint32 start, pcre_uint32 end)    compile_data *cd, pcre_uint32 start, pcre_uint32 end)
4140  {  {
4141  pcre_uint32 c;  pcre_uint32 c;
4142    pcre_uint32 classbits_end = (end <= 0xff ? end : 0xff);
4143  int n8 = 0;  int n8 = 0;
4144    
4145  /* If caseless matching is required, scan the range and process alternate  /* If caseless matching is required, scan the range and process alternate
# Line 3781  if ((options & PCRE_CASELESS) != 0) Line 4183  if ((options & PCRE_CASELESS) != 0)
4183    
4184    /* Not UTF-mode, or no UCP */    /* Not UTF-mode, or no UCP */
4185    
4186    for (c = start; c <= end && c < 256; c++)    for (c = start; c <= classbits_end; c++)
4187      {      {
4188      SETBIT(classbits, cd->fcc[c]);      SETBIT(classbits, cd->fcc[c]);
4189      n8++;      n8++;
# Line 3806  in all cases. */ Line 4208  in all cases. */
4208    
4209  #endif /* COMPILE_PCRE[8|16] */  #endif /* COMPILE_PCRE[8|16] */
4210    
4211  /* If all characters are less than 256, use the bit map. Otherwise use extra  /* Use the bitmap for characters < 256. Otherwise use extra data.*/
 data. */  
4212    
4213  if (end < 0x100)  for (c = start; c <= classbits_end; c++)
4214    {    {
4215    for (c = start; c <= end; c++)    /* Regardless of start, c will always be <= 255. */
4216      {    SETBIT(classbits, c);
4217      n8++;    n8++;
     SETBIT(classbits, c);  
     }  
4218    }    }
4219    
4220  else  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4221    if (start <= 0xff) start = 0xff + 1;
4222    
4223    if (end >= start)
4224    {    {
4225    pcre_uchar *uchardata = *uchardptr;    pcre_uchar *uchardata = *uchardptr;
   
4226  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4227    if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */    if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
4228      {      {
# Line 3861  else Line 4262  else
4262    
4263    *uchardptr = uchardata;   /* Updata extra data pointer */    *uchardptr = uchardata;   /* Updata extra data pointer */
4264    }    }
4265    #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */
4266    
4267  return n8;    /* Number of 8-bit characters */  return n8;    /* Number of 8-bit characters */
4268  }  }
# Line 4082  for (;; ptr++) Line 4484  for (;; ptr++)
4484    BOOL reset_bracount;    BOOL reset_bracount;
4485    int class_has_8bitchar;    int class_has_8bitchar;
4486    int class_one_char;    int class_one_char;
4487    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4488      BOOL xclass_has_prop;
4489    #endif
4490    int newoptions;    int newoptions;
4491    int recno;    int recno;
4492    int refsign;    int refsign;
# Line 4199  for (;; ptr++) Line 4604  for (;; ptr++)
4604          }          }
4605        goto NORMAL_CHAR;        goto NORMAL_CHAR;
4606        }        }
4607        /* Control does not reach here. */
4608      }      }
4609    
4610    /* Fill in length of a previous callout, except when the next thing is    /* In extended mode, skip white space and comments. We need a loop in order
4611    a quantifier. */    to check for more white space and more comments after a comment. */
   
   is_quantifier =  
     c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||  
     (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));  
   
   if (!is_quantifier && previous_callout != NULL &&  
        after_manual_callout-- <= 0)  
     {  
     if (lengthptr == NULL)      /* Don't attempt in pre-compile phase */  
       complete_callout(previous_callout, ptr, cd);  
     previous_callout = NULL;  
     }  
   
   /* In extended mode, skip white space and comments. */  
4612    
4613    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
4614      {      {
4615      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;      for (;;)
     if (c == CHAR_NUMBER_SIGN)  
4616        {        {
4617          while (MAX_255(c) && (cd->ctypes[c] & ctype_space) != 0) c = *(++ptr);
4618          if (c != CHAR_NUMBER_SIGN) break;
4619        ptr++;        ptr++;
4620        while (*ptr != CHAR_NULL)        while (*ptr != CHAR_NULL)
4621          {          {
4622          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr))         /* For non-fixed-length newline cases, */
4623              {                          /* IS_NEWLINE sets cd->nllen. */
4624              ptr += cd->nllen;
4625              break;
4626              }
4627          ptr++;          ptr++;
4628  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4629          if (utf) FORWARDCHAR(ptr);          if (utf) FORWARDCHAR(ptr);
4630  #endif  #endif
4631          }          }
4632        if (*ptr != CHAR_NULL) continue;        c = *ptr;     /* Either NULL or the char after a newline */
   
       /* Else fall through to handle end of string */  
       c = 0;  
4633        }        }
4634      }      }
4635    
4636    /* No auto callout for quantifiers. */    /* See if the next thing is a quantifier. */
4637    
4638    if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier)    is_quantifier =
4639        c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||
4640        (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));
4641    
4642      /* Fill in length of a previous callout, except when the next thing is a
4643      quantifier or when processing a property substitution string in UCP mode. */
4644    
4645      if (!is_quantifier && previous_callout != NULL && nestptr == NULL &&
4646           after_manual_callout-- <= 0)
4647        {
4648        if (lengthptr == NULL)      /* Don't attempt in pre-compile phase */
4649          complete_callout(previous_callout, ptr, cd);
4650        previous_callout = NULL;
4651        }
4652    
4653      /* Create auto callout, except for quantifiers, or while processing property
4654      strings that are substituted for \w etc in UCP mode. */
4655    
4656      if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier && nestptr == NULL)
4657      {      {
4658      previous_callout = code;      previous_callout = code;
4659      code = auto_callout(code, ptr, cd);      code = auto_callout(code, ptr, cd);
4660      }      }
4661    
4662      /* Process the next pattern item. */
4663    
4664    switch(c)    switch(c)
4665      {      {
4666      /* ===================================================================*/      /* ===================================================================*/
4667      case 0:                        /* The branch terminates at string end */      case CHAR_NULL:                /* The branch terminates at string end */
4668      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
4669      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
4670      *firstcharptr = firstchar;      *firstcharptr = firstchar;
# Line 4280  for (;; ptr++) Line 4694  for (;; ptr++)
4694      previous = NULL;      previous = NULL;
4695      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
4696        {        {
4697        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;        if (firstcharflags == REQ_UNSET)
4698            zerofirstcharflags = firstcharflags = REQ_NONE;
4699        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
4700        }        }
4701      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 4328  for (;; ptr++) Line 4743  for (;; ptr++)
4743        }        }
4744      goto NORMAL_CHAR;      goto NORMAL_CHAR;
4745    
4746        /* In another (POSIX) regex library, the ugly syntax [[:<:]] and [[:>:]] is
4747        used for "start of word" and "end of word". As these are otherwise illegal
4748        sequences, we don't break anything by recognizing them. They are replaced
4749        by \b(?=\w) and \b(?<=\w) respectively. Sequences like [a[:<:]] are
4750        erroneous and are handled by the normal code below. */
4751    
4752      case CHAR_LEFT_SQUARE_BRACKET:      case CHAR_LEFT_SQUARE_BRACKET:
4753        if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_STARTWORD, 6) == 0)
4754          {
4755          nestptr = ptr + 7;
4756          ptr = sub_start_of_word - 1;
4757          continue;
4758          }
4759    
4760        if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_ENDWORD, 6) == 0)
4761          {
4762          nestptr = ptr + 7;
4763          ptr = sub_end_of_word - 1;
4764          continue;
4765          }
4766    
4767        /* Handle a real character class. */
4768    
4769      previous = code;      previous = code;
4770    
4771      /* PCRE supports POSIX class stuff inside a class. Perl gives an error if      /* PCRE supports POSIX class stuff inside a class. Perl gives an error if
# Line 4385  for (;; ptr++) Line 4822  for (;; ptr++)
4822    
4823      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4824    
4825        /* Extended class (xclass) will be used when characters > 255
4826        might match. */
4827    
4828    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4829        xclass = FALSE;
4830        class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4831        class_uchardata_base = class_uchardata;   /* Save the start */
4832    #endif
4833    
4834      /* For optimization purposes, we track some properties of the class:      /* For optimization purposes, we track some properties of the class:
4835      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 <
4836      256 character; class_one_char will be 1 if the class contains just one      256 character; class_one_char will be 1 if the class contains just one
4837      character. */      character; xclass_has_prop will be TRUE if unicode property checks
4838        are present in the class. */
4839    
4840      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4841      class_one_char = 0;      class_one_char = 0;
4842    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4843        xclass_has_prop = FALSE;
4844    #endif
4845    
4846      /* 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
4847      temporary bit of memory, in case the class contains fewer than two      temporary bit of memory, in case the class contains fewer than two
# Line 4400  for (;; ptr++) Line 4850  for (;; ptr++)
4850    
4851      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4852    
 #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  
     xclass = FALSE;  
     class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */  
     class_uchardata_base = class_uchardata;   /* Save the start */  
 #endif  
   
4853      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4854      means that an initial ] is taken as a data character. At the start of the      means that an initial ] is taken as a data character. At the start of the
4855      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
# Line 4431  for (;; ptr++) Line 4875  for (;; ptr++)
4875        if (lengthptr != NULL && class_uchardata > class_uchardata_base)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4876          {          {
4877          xclass = TRUE;          xclass = TRUE;
4878          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += (int)(class_uchardata - class_uchardata_base);
4879          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4880          }          }
4881  #endif  #endif
# Line 4493  for (;; ptr++) Line 4937  for (;; ptr++)
4937            posix_class = 0;            posix_class = 0;
4938    
4939          /* 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
4940          different escape sequences that use Unicode properties. */          different escape sequences that use Unicode properties \p or \P. Others
4941            that are not available via \p or \P generate XCL_PROP/XCL_NOTPROP
4942            directly. */
4943    
4944  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4945          if ((options & PCRE_UCP) != 0)          if ((options & PCRE_UCP) != 0)
4946            {            {
4947              unsigned int ptype = 0;
4948            int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);            int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
4949    
4950              /* The posix_substitutes table specifies which POSIX classes can be
4951              converted to \p or \P items. */
4952    
4953            if (posix_substitutes[pc] != NULL)            if (posix_substitutes[pc] != NULL)
4954              {              {
4955              nestptr = tempptr + 1;              nestptr = tempptr + 1;
4956              ptr = posix_substitutes[pc] - 1;              ptr = posix_substitutes[pc] - 1;
4957              continue;              continue;
4958              }              }
4959    
4960              /* There are three other classes that generate special property calls
4961              that are recognized only in an XCLASS. */
4962    
4963              else switch(posix_class)
4964                {
4965                case PC_GRAPH:
4966                ptype = PT_PXGRAPH;
4967                /* Fall through */
4968                case PC_PRINT:
4969                if (ptype == 0) ptype = PT_PXPRINT;
4970                /* Fall through */
4971                case PC_PUNCT:
4972                if (ptype == 0) ptype = PT_PXPUNCT;
4973                *class_uchardata++ = local_negate? XCL_NOTPROP : XCL_PROP;
4974                *class_uchardata++ = ptype;
4975                *class_uchardata++ = 0;
4976                xclass_has_prop = TRUE;
4977                ptr = tempptr + 1;
4978                continue;
4979    
4980                /* For all other POSIX classes, no special action is taken in UCP
4981                mode. Fall through to the non_UCP case. */
4982    
4983                default:
4984                break;
4985                }
4986            }            }
4987  #endif  #endif
4988          /* In the non-UCP case, we build the bit map for the POSIX class in a          /* In the non-UCP case, or when UCP makes no difference, we build the
4989          chunk of local store because we may be adding and subtracting from it,          bit map for the POSIX class in a chunk of local store because we may be
4990          and we don't want to subtract bits that may be in the main map already.          adding and subtracting from it, and we don't want to subtract bits that
4991          At the end we or the result into the bit map that is being built. */          may be in the main map already. At the end we or the result into the
4992            bit map that is being built. */
4993    
4994          posix_class *= 3;          posix_class *= 3;
4995    
# Line 4627  for (;; ptr++) Line 5106  for (;; ptr++)
5106              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
5107              continue;              continue;
5108    
5109              /* Perl 5.004 onwards omits VT from \s, but we must preserve it              /* Perl 5.004 onwards omitted VT from \s, but restored it at Perl
5110              if it was previously set by something earlier in the character              5.18. Before PCRE 8.34, we had to preserve the VT bit if it was
5111              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and              previously set by something earlier in the character class.
5112              EBCDIC, so we lazily just adjust the appropriate bit. */              Luckily, the value of CHAR_VT is 0x0b in both ASCII and EBCDIC, so
5113                we could just adjust the appropriate bit. From PCRE 8.34 we no
5114                longer treat \s and \S specially. */
5115    
5116              case ESC_s:              case ESC_s:
5117              classbits[0] |= cbits[cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
             classbits[1] |= cbits[cbit_space+1] & ~0x08;  
             for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];  
5118              continue;              continue;
5119    
5120              case ESC_S:              case ESC_S:
5121              should_flip_negation = TRUE;              should_flip_negation = TRUE;
5122              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
             classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */  
5123              continue;              continue;
5124    
5125              /* The rest apply in both UCP and non-UCP cases. */              /* The rest apply in both UCP and non-UCP cases. */
# Line 4678  for (;; ptr++) Line 5156  for (;; ptr++)
5156                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
5157                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
5158                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
5159                  xclass_has_prop = TRUE;
5160                class_has_8bitchar--;                /* Undo! */                class_has_8bitchar--;                /* Undo! */
5161                continue;                continue;
5162                }                }
# Line 4763  for (;; ptr++) Line 5242  for (;; ptr++)
5242  #endif  #endif
5243          d = *ptr;  /* Not UTF-8 mode */          d = *ptr;  /* Not UTF-8 mode */
5244    
5245          /* The second part of a range can be a single-character escape, but          /* The second part of a range can be a single-character escape
5246          not any of the other escapes. Perl 5.6 treats a hyphen as a literal          sequence, but not any of the other escapes. Perl treats a hyphen as a
5247          in such circumstances. */          literal in such circumstances. However, in Perl's warning mode, a
5248            warning is given, so PCRE now faults it as it is almost certainly a
5249            mistake on the user's part. */
5250    
5251          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq)
5252            {            {
5253            int descape;            if (d == CHAR_BACKSLASH)
5254            descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);              {
5255            if (*errorcodeptr != 0) goto FAILED;              int descape;
5256                descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
5257                if (*errorcodeptr != 0) goto FAILED;
5258    
5259            /* \b is backspace; any other special means the '-' was literal. */              /* 0 means a character was put into d; \b is backspace; any other
5260                special causes an error. */
5261    
5262            if (descape != 0)              if (descape != 0)
             {  
             if (descape == ESC_b) d = CHAR_BS; else  
5263                {                {
5264                ptr = oldptr;                if (descape == ESC_b) d = CHAR_BS; else
5265                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */                  {
5266                    *errorcodeptr = ERR83;
5267                    goto FAILED;
5268                    }
5269                }                }
5270              }              }
5271    
5272              /* A hyphen followed by a POSIX class is treated in the same way. */
5273    
5274              else if (d == CHAR_LEFT_SQUARE_BRACKET &&
5275                       (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
5276                        ptr[1] == CHAR_EQUALS_SIGN) &&
5277                       check_posix_syntax(ptr, &tempptr))
5278                {
5279                *errorcodeptr = ERR83;
5280                goto FAILED;
5281                }
5282            }            }
5283    
5284          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
# Line 4829  for (;; ptr++) Line 5325  for (;; ptr++)
5325        whatever repeat count may follow. In the case of reqchar, save the        whatever repeat count may follow. In the case of reqchar, save the
5326        previous value for reinstating. */        previous value for reinstating. */
5327    
5328        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)        if (!inescq && class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
5329          {          {
5330          ptr++;          ptr++;
5331          zeroreqchar = reqchar;          zeroreqchar = reqchar;
# Line 4955  for (;; ptr++) Line 5451  for (;; ptr++)
5451        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
5452        code += LINK_SIZE;        code += LINK_SIZE;
5453        *code = negate_class? XCL_NOT:0;        *code = negate_class? XCL_NOT:0;
5454          if (xclass_has_prop) *code |= XCL_HASPROP;
5455    
5456        /* If the map is required, move up the extra data to make room for it;        /* If the map is required, move up the extra data to make room for it;
5457        otherwise just move the code pointer to the end of the extra data. */        otherwise just move the code pointer to the end of the extra data. */
# Line 4964  for (;; ptr++) Line 5461  for (;; ptr++)
5461          *code++ |= XCL_MAP;          *code++ |= XCL_MAP;
5462          memmove(code + (32 / sizeof(pcre_uchar)), code,          memmove(code + (32 / sizeof(pcre_uchar)), code,
5463            IN_UCHARS(class_uchardata - code));            IN_UCHARS(class_uchardata - code));
5464            if (negate_class && !xclass_has_prop)
5465              for (c = 0; c < 32; c++) classbits[c] = ~classbits[c];
5466          memcpy(code, classbits, 32);          memcpy(code, classbits, 32);
5467          code = class_uchardata + (32 / sizeof(pcre_uchar));          code = class_uchardata + (32 / sizeof(pcre_uchar));
5468          }          }
# Line 5046  for (;; ptr++) Line 5545  for (;; ptr++)
5545    
5546      tempcode = previous;      tempcode = previous;
5547    
5548        /* Before checking for a possessive quantifier, we must skip over
5549        whitespace and comments in extended mode because Perl allows white space at
5550        this point. */
5551    
5552        if ((options & PCRE_EXTENDED) != 0)
5553          {
5554          const pcre_uchar *p = ptr + 1;
5555          for (;;)
5556            {
5557            while (MAX_255(*p) && (cd->ctypes[*p] & ctype_space) != 0) p++;
5558            if (*p != CHAR_NUMBER_SIGN) break;
5559            p++;
5560            while (*p != CHAR_NULL)
5561              {
5562              if (IS_NEWLINE(p))         /* For non-fixed-length newline cases, */
5563                {                        /* IS_NEWLINE sets cd->nllen. */
5564                p += cd->nllen;
5565                break;
5566                }
5567              p++;
5568    #ifdef SUPPORT_UTF
5569              if (utf) FORWARDCHAR(p);
5570    #endif
5571              }           /* Loop for comment characters */
5572            }             /* Loop for multiple comments */
5573          ptr = p - 1;    /* Character before the next significant one. */
5574          }
5575    
5576      /* If the next character is '+', we have a possessive quantifier. This      /* If the next character is '+', we have a possessive quantifier. This
5577      implies greediness, whatever the setting of the PCRE_UNGREEDY option.      implies greediness, whatever the setting of the PCRE_UNGREEDY option.
5578      If the next character is '?' this is a minimizing repeat, by default,      If the next character is '?' this is a minimizing repeat, by default,
# Line 5338  for (;; ptr++) Line 5865  for (;; ptr++)
5865      opcodes such as BRA and CBRA, as this is the place where they get converted      opcodes such as BRA and CBRA, as this is the place where they get converted
5866      into the more special varieties such as BRAPOS and SBRA. A test for >=      into the more special varieties such as BRAPOS and SBRA. A test for >=
5867      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,
5868      ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow      ASSERTBACK_NOT, ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND.
5869      repetition of assertions, but now it does, for Perl compatibility. */      Originally, PCRE did not allow repetition of assertions, but now it does,
5870        for Perl compatibility. */
5871    
5872      else if (*previous >= OP_ASSERT && *previous <= OP_COND)      else if (*previous >= OP_ASSERT && *previous <= OP_COND)
5873        {        {
# Line 5357  for (;; ptr++) Line 5885  for (;; ptr++)
5885        /* There is no sense in actually repeating assertions. The only potential        /* There is no sense in actually repeating assertions. The only potential
5886        use of repetition is in cases when the assertion is optional. Therefore,        use of repetition is in cases when the assertion is optional. Therefore,
5887        if the minimum is greater than zero, just ignore the repeat. If the        if the minimum is greater than zero, just ignore the repeat. If the
5888        maximum is not not zero or one, set it to 1. */        maximum is not zero or one, set it to 1. */
5889    
5890        if (*previous < OP_ONCE)    /* Assertion */        if (*previous < OP_ONCE)    /* Assertion */
5891          {          {
# Line 5492  for (;; ptr++) Line 6020  for (;; ptr++)
6020                while (cd->hwm > cd->start_workspace + cd->workspace_size -                while (cd->hwm > cd->start_workspace + cd->workspace_size -
6021                       WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))                       WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))
6022                  {                  {
6023                  int save_offset = save_hwm - cd->start_workspace;                  size_t save_offset = save_hwm - cd->start_workspace;
6024                  int this_offset = this_hwm - cd->start_workspace;                  size_t this_offset = this_hwm - cd->start_workspace;
6025                  *errorcodeptr = expand_workspace(cd);                  *errorcodeptr = expand_workspace(cd);
6026                  if (*errorcodeptr != 0) goto FAILED;                  if (*errorcodeptr != 0) goto FAILED;
6027                  save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;                  save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;
# Line 5574  for (;; ptr++) Line 6102  for (;; ptr++)
6102            while (cd->hwm > cd->start_workspace + cd->workspace_size -            while (cd->hwm > cd->start_workspace + cd->workspace_size -
6103                   WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))                   WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))
6104              {              {
6105              int save_offset = save_hwm - cd->start_workspace;              size_t save_offset = save_hwm - cd->start_workspace;
6106              int this_offset = this_hwm - cd->start_workspace;              size_t this_offset = this_hwm - cd->start_workspace;
6107              *errorcodeptr = expand_workspace(cd);              *errorcodeptr = expand_workspace(cd);
6108              if (*errorcodeptr != 0) goto FAILED;              if (*errorcodeptr != 0) goto FAILED;
6109              save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;              save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;
# Line 5730  for (;; ptr++) Line 6258  for (;; ptr++)
6258        goto FAILED;        goto FAILED;
6259        }        }
6260    
6261      /* If the character following a repeat is '+', or if certain optimization      /* If the character following a repeat is '+', possessive_quantifier is
6262      tests above succeeded, possessive_quantifier is TRUE. For some opcodes,      TRUE. For some opcodes, there are special alternative opcodes for this
6263      there are special alternative opcodes for this case. For anything else, we      case. For anything else, we wrap the entire repeated item inside OP_ONCE
6264      wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+'      brackets. Logically, the '+' notation is just syntactic sugar, taken from
6265      notation is just syntactic sugar, taken from Sun's Java package, but the      Sun's Java package, but the special opcodes can optimize it.
     special opcodes can optimize it.  
6266    
6267      Some (but not all) possessively repeated subpatterns have already been      Some (but not all) possessively repeated subpatterns have already been
6268      completely handled in the code just above. For them, possessive_quantifier      completely handled in the code just above. For them, possessive_quantifier
6269      is always FALSE at this stage.      is always FALSE at this stage. Note that the repeated item starts at
6270        tempcode, not at previous, which might be the first part of a string whose
6271      Note that the repeated item starts at tempcode, not at previous, which      (former) last char we repeated. */
     might be the first part of a string whose (former) last char we repeated.  
   
     Possessifying an 'exact' quantifier has no effect, so we can ignore it. But  
     an 'upto' may follow. We skip over an 'exact' item, and then test the  
     length of what remains before proceeding. */  
6272    
6273      if (possessive_quantifier)      if (possessive_quantifier)
6274        {        {
6275        int len;        int len;
6276    
6277        if (*tempcode == OP_TYPEEXACT)        /* Possessifying an EXACT quantifier has no effect, so we can ignore it.
6278          However, QUERY, STAR, or UPTO may follow (for quantifiers such as {5,6},
6279          {5,}, or {5,10}). We skip over an EXACT item; if the length of what
6280          remains is greater than zero, there's a further opcode that can be
6281          handled. If not, do nothing, leaving the EXACT alone. */
6282    
6283          switch(*tempcode)
6284            {
6285            case OP_TYPEEXACT:
6286          tempcode += PRIV(OP_lengths)[*tempcode] +          tempcode += PRIV(OP_lengths)[*tempcode] +
6287            ((tempcode[1 + IMM2_SIZE] == OP_PROP            ((tempcode[1 + IMM2_SIZE] == OP_PROP
6288            || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);            || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);
6289            break;
6290    
6291        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)          /* CHAR opcodes are used for exacts whose count is 1. */
6292          {  
6293            case OP_CHAR:
6294            case OP_CHARI:
6295            case OP_NOT:
6296            case OP_NOTI:
6297            case OP_EXACT:
6298            case OP_EXACTI:
6299            case OP_NOTEXACT:
6300            case OP_NOTEXACTI:
6301          tempcode += PRIV(OP_lengths)[*tempcode];          tempcode += PRIV(OP_lengths)[*tempcode];
6302  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
6303          if (utf && HAS_EXTRALEN(tempcode[-1]))          if (utf && HAS_EXTRALEN(tempcode[-1]))
6304            tempcode += GET_EXTRALEN(tempcode[-1]);            tempcode += GET_EXTRALEN(tempcode[-1]);
6305  #endif  #endif
6306            break;
6307    
6308            /* For the class opcodes, the repeat operator appears at the end;
6309            adjust tempcode to point to it. */
6310    
6311            case OP_CLASS:
6312            case OP_NCLASS:
6313            tempcode += 1 + 32/sizeof(pcre_uchar);
6314            break;
6315    
6316    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
6317            case OP_XCLASS:
6318            tempcode += GET(tempcode, 1);
6319            break;
6320    #endif
6321          }          }
6322    
6323          /* If tempcode is equal to code (which points to the end of the repeated
6324          item), it means we have skipped an EXACT item but there is no following
6325          QUERY, STAR, or UPTO; the value of len will be 0, and we do nothing. In
6326          all other cases, tempcode will be pointing to the repeat opcode, and will
6327          be less than code, so the value of len will be greater than 0. */
6328    
6329        len = (int)(code - tempcode);        len = (int)(code - tempcode);
6330          if (len > 0)
6331            {
6332            unsigned int repcode = *tempcode;
6333    
6334            /* There is a table for possessifying opcodes, all of which are less
6335            than OP_CALLOUT. A zero entry means there is no possessified version.
6336            */
6337    
6338            if (repcode < OP_CALLOUT && opcode_possessify[repcode] > 0)
6339              *tempcode = opcode_possessify[repcode];
6340    
6341            /* For opcode without a special possessified version, wrap the item in
6342            ONCE brackets. Because we are moving code along, we must ensure that any
6343            pending recursive references are updated. */
6344    
6345            else
6346              {
6347              *code = OP_END;
6348              adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm);
6349              memmove(tempcode + 1 + LINK_SIZE, tempcode, IN_UCHARS(len));
6350              code += 1 + LINK_SIZE;
6351              len += 1 + LINK_SIZE;
6352              tempcode[0] = OP_ONCE;
6353              *code++ = OP_KET;
6354              PUTINC(code, 0, len);
6355              PUT(tempcode, 1, len);
6356              }
6357            }
6358    
6359    #ifdef NEVER
6360        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
6361          {          {
6362          case OP_STAR:  *tempcode = OP_POSSTAR; break;          case OP_STAR:  *tempcode = OP_POSSTAR; break;
# Line 5794  for (;; ptr++) Line 6384  for (;; ptr++)
6384          case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break;          case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break;
6385          case OP_TYPEUPTO:  *tempcode = OP_TYPEPOSUPTO; break;          case OP_TYPEUPTO:  *tempcode = OP_TYPEPOSUPTO; break;
6386    
6387            case OP_CRSTAR:   *tempcode = OP_CRPOSSTAR; break;
6388            case OP_CRPLUS:   *tempcode = OP_CRPOSPLUS; break;
6389            case OP_CRQUERY:  *tempcode = OP_CRPOSQUERY; break;
6390            case OP_CRRANGE:  *tempcode = OP_CRPOSRANGE; break;
6391    
6392          /* Because we are moving code along, we must ensure that any          /* Because we are moving code along, we must ensure that any
6393          pending recursive references are updated. */          pending recursive references are updated. */
6394    
# Line 5809  for (;; ptr++) Line 6404  for (;; ptr++)
6404          PUT(tempcode, 1, len);          PUT(tempcode, 1, len);
6405          break;          break;
6406          }          }
6407    #endif
6408        }        }
6409    
6410      /* In all case we no longer have a previous item. We also set the      /* In all case we no longer have a previous item. We also set the
# Line 5997  for (;; ptr++) Line 6593  for (;; ptr++)
6593          tempptr = ptr;          tempptr = ptr;
6594    
6595          /* A condition can be an assertion, a number (referring to a numbered          /* A condition can be an assertion, a number (referring to a numbered
6596          group), a name (referring to a named group), or 'R', referring to          group's having been set), a name (referring to a named group), or 'R',
6597          recursion. R<digits> and R&name are also permitted for recursion tests.          referring to recursion. R<digits> and R&name are also permitted for
6598            recursion tests.
6599          There are several syntaxes for testing a named group: (?(name)) is used  
6600          by Python; Perl 5.10 onwards uses (?(<name>) or (?('name')).          There are ways of testing a named group: (?(name)) is used by Python;
6601            Perl 5.10 onwards uses (?(<name>) or (?('name')).
6602          There are two unfortunate ambiguities, caused by history. (a) 'R' can  
6603          be the recursive thing or the name 'R' (and similarly for 'R' followed          There is one unfortunate ambiguity, caused by history. 'R' can be the
6604          by digits), and (b) a number could be a name that consists of digits.          recursive thing or the name 'R' (and similarly for 'R' followed by
6605          In both cases, we look for a name first; if not found, we try the other          digits). We look for a name first; if not found, we try the other case.
         cases.  
6606    
6607          For compatibility with auto-callouts, we allow a callout to be          For compatibility with auto-callouts, we allow a callout to be
6608          specified before a condition that is an assertion. First, check for the          specified before a condition that is an assertion. First, check for the
# Line 6031  for (;; ptr++) Line 6626  for (;; ptr++)
6626                 tempptr[2] == CHAR_LESS_THAN_SIGN))                 tempptr[2] == CHAR_LESS_THAN_SIGN))
6627            break;            break;
6628    
6629          /* Most other conditions use OP_CREF (a couple change to OP_RREF          /* Other conditions use OP_CREF/OP_DNCREF/OP_RREF/OP_DNRREF, and all
6630          below), and all need to skip 1+IMM2_SIZE bytes at the start of the group. */          need to skip at least 1+IMM2_SIZE bytes at the start of the group. */
6631    
6632          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
6633          skipbytes = 1+IMM2_SIZE;          skipbytes = 1+IMM2_SIZE;
6634          refsign = -1;          refsign = -1;     /* => not a number */
6635            namelen = -1;     /* => not a name; must set to avoid warning */
6636            name = NULL;      /* Always set to avoid warning */
6637            recno = 0;        /* Always set to avoid warning */
6638    
6639          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
6640    
6641          if (ptr[1] == CHAR_R && ptr[2] == CHAR_AMPERSAND)          ptr++;
6642            if (*ptr == CHAR_R && ptr[1] == CHAR_AMPERSAND)
6643            {            {
6644            terminator = -1;            terminator = -1;
6645            ptr += 2;            ptr += 2;
# Line 6048  for (;; ptr++) Line 6647  for (;; ptr++)
6647            }            }
6648    
6649          /* Check for a test for a named group's having been set, using the Perl          /* Check for a test for a named group's having been set, using the Perl
6650          syntax (?(<name>) or (?('name') */          syntax (?(<name>) or (?('name'), and also allow for the original PCRE
6651            syntax of (?(name) or for (?(+n), (?(-n), and just (?(n). */
6652    
6653          else if (ptr[1] == CHAR_LESS_THAN_SIGN)          else if (*ptr == CHAR_LESS_THAN_SIGN)
6654            {            {
6655            terminator = CHAR_GREATER_THAN_SIGN;            terminator = CHAR_GREATER_THAN_SIGN;
6656            ptr++;            ptr++;
6657            }            }
6658          else if (ptr[1] == CHAR_APOSTROPHE)          else if (*ptr == CHAR_APOSTROPHE)
6659            {            {
6660            terminator = CHAR_APOSTROPHE;            terminator = CHAR_APOSTROPHE;
6661            ptr++;            ptr++;
# Line 6063  for (;; ptr++) Line 6663  for (;; ptr++)
6663          else          else
6664            {            {
6665            terminator = CHAR_NULL;            terminator = CHAR_NULL;
6666            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);            if (*ptr == CHAR_MINUS || *ptr == CHAR_PLUS) refsign = *ptr++;
6667                else if (IS_DIGIT(*ptr)) refsign = 0;
6668            }            }
6669    
6670          /* We now expect to read a name; any thing else is an error */          /* Handle a number */
6671    
6672          if (!MAX_255(ptr[1]) || (cd->ctypes[ptr[1]] & ctype_word) == 0)          if (refsign >= 0)
6673            {            {
6674            ptr += 1;  /* To get the right offset */            while (IS_DIGIT(*ptr))
6675            *errorcodeptr = ERR28;              {
6676            goto FAILED;              recno = recno * 10 + (int)(*ptr - CHAR_0);
6677                ptr++;
6678                }
6679            }            }
6680    
6681          /* Read the name, but also get it as a number if it's all digits */          /* Otherwise we expect to read a name; anything else is an error. When
6682            a name is one of a number of duplicates, a different opcode is used and
6683            it needs more memory. Unfortunately we cannot tell whether a name is a
6684            duplicate in the first pass, so we have to allow for more memory. */
6685    
6686          recno = 0;          else
         name = ++ptr;  
         while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)  
6687            {            {
6688            if (recno >= 0)            if (IS_DIGIT(*ptr))
6689              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;              {
6690            ptr++;              *errorcodeptr = ERR84;
6691                goto FAILED;
6692                }
6693              if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_word) == 0)
6694                {
6695                *errorcodeptr = ERR28;   /* Assertion expected */
6696                goto FAILED;
6697                }
6698              name = ptr++;
6699              while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
6700                {
6701                ptr++;
6702                }
6703              namelen = (int)(ptr - name);
6704              if (lengthptr != NULL && (options & PCRE_DUPNAMES) != 0)
6705                *lengthptr += IMM2_SIZE;
6706            }            }
6707          namelen = (int)(ptr - name);  
6708            /* Check the terminator */
6709    
6710          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||
6711              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
6712            {            {
6713            ptr--;      /* Error offset */            ptr--;                  /* Error offset */
6714            *errorcodeptr = ERR26;            *errorcodeptr = ERR26;  /* Malformed number or name */
6715            goto FAILED;            goto FAILED;
6716            }            }
6717    
# Line 6100  for (;; ptr++) Line 6720  for (;; ptr++)
6720          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
6721    
6722          /* In the real compile we do the work of looking for the actual          /* In the real compile we do the work of looking for the actual
6723          reference. If the string started with "+" or "-" we require the rest to          reference. If refsign is not negative, it means we have a number in
6724          be digits, in which case recno will be set. */          recno. */
6725    
6726          if (refsign > 0)          if (refsign >= 0)
6727            {            {
6728            if (recno <= 0)            if (recno <= 0)
6729              {              {
6730              *errorcodeptr = ERR58;              *errorcodeptr = ERR35;
6731              goto FAILED;              goto FAILED;
6732              }              }
6733            recno = (refsign == CHAR_MINUS)?            if (refsign != 0) recno = (refsign == CHAR_MINUS)?
6734              cd->bracount - recno + 1 : recno +cd->bracount;              cd->bracount - recno + 1 : recno + cd->bracount;
6735            if (recno <= 0 || recno > cd->final_bracount)            if (recno <= 0 || recno > cd->final_bracount)
6736              {              {
6737              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
# Line 6121  for (;; ptr++) Line 6741  for (;; ptr++)
6741            break;            break;
6742            }            }
6743    
6744          /* Otherwise (did not start with "+" or "-"), start by looking for the          /* Otherwise look for the name. */
         name. If we find a name, add one to the opcode to change OP_CREF or  
         OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same,  
         except they record that the reference was originally to a name. The  
         information is used to check duplicate names. */  
6745    
6746          slot = cd->name_table;          slot = cd->name_table;
6747          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 6134  for (;; ptr++) Line 6750  for (;; ptr++)
6750            slot += cd->name_entry_size;            slot += cd->name_entry_size;
6751            }            }
6752    
6753          /* Found the named subpattern */          /* Found the named subpattern. If the name is duplicated, add one to
6754            the opcode to change CREF/RREF into DNCREF/DNRREF and insert
6755            appropriate data values. Otherwise, just insert the unique subpattern
6756            number. */
6757    
6758          if (i < cd->names_found)          if (i < cd->names_found)
6759            {            {
6760            recno = GET2(slot, 0);            int offset = i++;
6761            PUT2(code, 2+LINK_SIZE, recno);            int count = 1;
6762            code[1+LINK_SIZE]++;            recno = GET2(slot, 0);   /* Number from first found */
6763              for (; i < cd->names_found; i++)
6764                {
6765                slot += cd->name_entry_size;
6766                if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) != 0 ||
6767                  (slot+IMM2_SIZE)[namelen] != 0) break;
6768                count++;
6769                }
6770    
6771              if (count > 1)
6772                {
6773                PUT2(code, 2+LINK_SIZE, offset);
6774                PUT2(code, 2+LINK_SIZE+IMM2_SIZE, count);
6775                skipbytes += IMM2_SIZE;
6776                code[1+LINK_SIZE]++;
6777                }
6778              else  /* Not a duplicated name */
6779                {
6780                PUT2(code, 2+LINK_SIZE, recno);
6781                }
6782            }            }
6783    
6784          /* If terminator == CHAR_NULL it means that the name followed directly          /* If terminator == CHAR_NULL it means that the name followed directly
6785          after the opening parenthesis [e.g. (?(abc)...] and in this case there          after the opening parenthesis [e.g. (?(abc)...] and in this case there
6786          are some further alternatives to try. For the cases where terminator !=          are some further alternatives to try. For the cases where terminator !=
6787          0 [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have          CHAR_NULL [things like (?(<name>... or (?('name')... or (?(R&name)... ]
6788          now checked all the possibilities, so give an error. */          we have now checked all the possibilities, so give an error. */
6789    
6790          else if (terminator != CHAR_NULL)          else if (terminator != CHAR_NULL)
6791            {            {
# Line 6184  for (;; ptr++) Line 6822  for (;; ptr++)
6822            skipbytes = 1;            skipbytes = 1;
6823            }            }
6824    
6825          /* Check for the "name" actually being a subpattern number. We are          /* Reference to an unidentified subpattern. */
         in the second pass here, so final_bracount is set. */  
   
         else if (recno > 0 && recno <= cd->final_bracount)  
           {  
           PUT2(code, 2+LINK_SIZE, recno);  
           }  
   
         /* Either an unidentified subpattern, or a reference to (?(0) */  
6826    
6827          else          else
6828            {            {
6829            *errorcodeptr = (recno == 0)? ERR35: ERR15;            *errorcodeptr = ERR15;
6830            goto FAILED;            goto FAILED;
6831            }            }
6832          break;          break;
# Line 6209  for (;; ptr++) Line 6839  for (;; ptr++)
6839          ptr++;          ptr++;
6840          break;          break;
6841    
6842            /* Optimize (?!) to (*FAIL) unless it is quantified - which is a weird
6843            thing to do, but Perl allows all assertions to be quantified, and when
6844            they contain capturing parentheses there may be a potential use for
6845            this feature. Not that that applies to a quantified (?!) but we allow
6846            it for uniformity. */
6847    
6848          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
6849          case CHAR_EXCLAMATION_MARK:            /* Negative lookahead */          case CHAR_EXCLAMATION_MARK:            /* Negative lookahead */
6850          ptr++;          ptr++;
6851          if (*ptr == CHAR_RIGHT_PARENTHESIS)    /* Optimize (?!) */          if (*ptr == CHAR_RIGHT_PARENTHESIS && ptr[1] != CHAR_ASTERISK &&
6852                 ptr[1] != CHAR_PLUS && ptr[1] != CHAR_QUESTION_MARK &&
6853                (ptr[1] != CHAR_LEFT_CURLY_BRACKET || !is_counted_repeat(ptr+2)))
6854            {            {
6855            *code++ = OP_FAIL;            *code++ = OP_FAIL;
6856            previous = NULL;            previous = NULL;
# Line 6309  for (;; ptr++) Line 6946  for (;; ptr++)
6946          terminator = (*ptr == CHAR_LESS_THAN_SIGN)?          terminator = (*ptr == CHAR_LESS_THAN_SIGN)?
6947            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
6948          name = ++ptr;          name = ++ptr;
6949            if (IS_DIGIT(*ptr))
6950              {
6951              *errorcodeptr = ERR84;   /* Group name must start with non-digit */
6952              goto FAILED;
6953              }
6954          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
6955          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
6956    
# Line 6423  for (;; ptr++) Line 7064  for (;; ptr++)
7064    
7065          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
7066          name = ++ptr;          name = ++ptr;
7067            if (IS_DIGIT(*ptr))
7068              {
7069              *errorcodeptr = ERR84;   /* Group name must start with non-digit */
7070              goto FAILED;
7071              }
7072          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
7073          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
7074    
# Line 6470  for (;; ptr++) Line 7116  for (;; ptr++)
7116            /* Count named back references. */            /* Count named back references. */
7117    
7118            if (!is_recurse) cd->namedrefcount++;            if (!is_recurse) cd->namedrefcount++;
7119    
7120              /* If duplicate names are permitted, we have to allow for a named
7121              reference to a duplicated name (this cannot be determined until the
7122              second pass). This needs an extra 16-bit data item. */
7123    
7124              if ((options & PCRE_DUPNAMES) != 0) *lengthptr += IMM2_SIZE;
7125            }            }
7126    
7127          /* In the real compile, search the name table. We check the name          /* In the real compile, search the name table. We check the name
# Line 6516  for (;; ptr++) Line 7168  for (;; ptr++)
7168            for (i++; i < cd->names_found; i++)            for (i++; i < cd->names_found; i++)
7169              {              {
7170              if (STRCMP_UC_UC(slot + IMM2_SIZE, cslot + IMM2_SIZE) != 0) break;              if (STRCMP_UC_UC(slot + IMM2_SIZE, cslot + IMM2_SIZE) != 0) break;
7171    
7172    
7173              count++;              count++;
7174              cslot += cd->name_entry_size;              cslot += cd->name_entry_size;
7175              }              }
# Line 6815  for (;; ptr++) Line 7469  for (;; ptr++)
7469        skipbytes = IMM2_SIZE;        skipbytes = IMM2_SIZE;
7470        }        }
7471    
7472      /* Process nested bracketed regex. Assertions used not to be repeatable,      /* Process nested bracketed regex. First check for parentheses nested too
7473      but this was changed for Perl compatibility, so all kinds can now be      deeply. */
7474      repeated. We copy code into a non-register variable (tempcode) in order to  
7475      be able to pass its address because some compilers complain otherwise. */      if ((cd->parens_depth += 1) > PARENS_NEST_LIMIT)
7476          {
7477          *errorcodeptr = ERR82;
7478          goto FAILED;
7479          }
7480    
7481        /* Assertions used not to be repeatable, but this was changed for Perl
7482        compatibility, so all kinds can now be repeated. We copy code into a
7483        non-register variable (tempcode) in order to be able to pass its address
7484        because some compilers complain otherwise. */
7485    
7486      previous = code;                      /* For handling repetition */      previous = code;                      /* For handling repetition */
7487      *code = bravalue;      *code = bravalue;
# Line 6849  for (;; ptr++) Line 7512  for (;; ptr++)
7512           ))           ))
7513        goto FAILED;        goto FAILED;
7514    
7515        cd->parens_depth -= 1;
7516    
7517      /* If this was an atomic group and there are no capturing groups within it,      /* If this was an atomic group and there are no capturing groups within it,
7518      generate OP_ONCE_NC instead of OP_ONCE. */      generate OP_ONCE_NC instead of OP_ONCE. */
7519    
# Line 7063  for (;; ptr++) Line 7728  for (;; ptr++)
7728        if (escape == ESC_g)        if (escape == ESC_g)
7729          {          {
7730          const pcre_uchar *p;          const pcre_uchar *p;
7731            pcre_uint32 cf;
7732    
7733          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
7734          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
7735            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
7736    
7737          /* These two statements stop the compiler for warning about possibly          /* These two statements stop the compiler for warning about possibly
7738          unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In          unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In
7739          fact, because we actually check for a number below, the paths that          fact, because we do the check for a number below, the paths that
7740          would actually be in error are never taken. */          would actually be in error are never taken. */
7741    
7742          skipbytes = 0;          skipbytes = 0;
7743          reset_bracount = FALSE;          reset_bracount = FALSE;
7744    
7745          /* Test for a name */          /* If it's not a signed or unsigned number, treat it as a name. */
7746    
7747          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          cf = ptr[1];
7748            if (cf != CHAR_PLUS && cf != CHAR_MINUS && !IS_DIGIT(cf))
7749            {            {
           BOOL is_a_number = TRUE;  
           for (p = ptr + 1; *p != CHAR_NULL && *p != (pcre_uchar)terminator; p++)  
             {  
             if (!MAX_255(*p)) { is_a_number = FALSE; break; }  
             if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;  
             if ((cd->ctypes[*p] & ctype_word) == 0) break;  
             }  
           if (*p != (pcre_uchar)terminator)  
             {  
             *errorcodeptr = ERR57;  
             break;  
             }  
           if (is_a_number)  
             {  
             ptr++;  
             goto HANDLE_NUMERICAL_RECURSION;  
             }  
7750            is_recurse = TRUE;            is_recurse = TRUE;
7751            goto NAMED_REF_OR_RECURSE;            goto NAMED_REF_OR_RECURSE;
7752            }            }
7753    
7754          /* Test a signed number in angle brackets or quotes. */          /* Signed or unsigned number (cf = ptr[1]) is known to be plus or minus
7755            or a digit. */
7756    
7757          p = ptr + 2;          p = ptr + 2;
7758          while (IS_DIGIT(*p)) p++;          while (IS_DIGIT(*p)) p++;
# Line 7240  for (;; ptr++) Line 7892  for (;; ptr++)
7892    
7893      /* ===================================================================*/      /* ===================================================================*/
7894      /* Handle a literal character. It is guaranteed not to be whitespace or #      /* Handle a literal character. It is guaranteed not to be whitespace or #
7895      when the extended flag is set. If we are in UTF-8 mode, it may be a      when the extended flag is set. If we are in a UTF mode, it may be a
7896      multi-byte literal character. */      multi-unit literal character. */
7897    
7898      default:      default:
7899      NORMAL_CHAR:      NORMAL_CHAR:
# Line 7406  unsigned int orig_bracount; Line 8058  unsigned int orig_bracount;
8058  unsigned int max_bracount;  unsigned int max_bracount;
8059  branch_chain bc;  branch_chain bc;
8060    
8061    /* If set, call the external function that checks for stack availability. */
8062    
8063    if (PUBL(stack_guard) != NULL && PUBL(stack_guard)())
8064      {
8065      *errorcodeptr= ERR85;
8066      return FALSE;
8067      }
8068    
8069    /* Miscellaneous initialization */
8070    
8071  bc.outer = bcptr;  bc.outer = bcptr;
8072  bc.current_branch = code;  bc.current_branch = code;
8073    
# Line 7605  for (;;) Line 8267  for (;;)
8267    
8268      /* If it was a capturing subpattern, check to see if it contained any      /* If it was a capturing subpattern, check to see if it contained any
8269      recursive back references. If so, we must wrap it in atomic brackets.      recursive back references. If so, we must wrap it in atomic brackets.
8270      In any event, remove the block from the chain. */      Because we are moving code along, we must ensure that any pending recursive
8271        references are updated. In any event, remove the block from the chain. */
8272    
8273      if (capnumber > 0)      if (capnumber > 0)
8274        {        {
8275        if (cd->open_caps->flag)        if (cd->open_caps->flag)
8276          {          {
8277            *code = OP_END;
8278            adjust_recurse(start_bracket, 1 + LINK_SIZE,
8279              (options & PCRE_UTF8) != 0, cd, cd->hwm);
8280          memmove(start_bracket + 1 + LINK_SIZE, start_bracket,          memmove(start_bracket + 1 + LINK_SIZE, start_bracket,
8281            IN_UCHARS(code - start_bracket));            IN_UCHARS(code - start_bracket));
8282          *start_bracket = OP_ONCE;          *start_bracket = OP_ONCE;
# Line 7830  do { Line 8496  do {
8496       switch (*scode)       switch (*scode)
8497         {         {
8498         case OP_CREF:         case OP_CREF:
8499         case OP_NCREF:         case OP_DNCREF:
8500         case OP_RREF:         case OP_RREF:
8501         case OP_NRREF:         case OP_DNRREF:
8502         case OP_DEF:         case OP_DEF:
8503         return FALSE;         return FALSE;
8504    
# Line 8229  PCRE_UTF8 == PCRE_UTF16 == PCRE_UTF32. * Line 8895  PCRE_UTF8 == PCRE_UTF16 == PCRE_UTF32. *
8895      { skipatstart += 6; options |= PCRE_UTF8; continue; }      { skipatstart += 6; options |= PCRE_UTF8; continue; }
8896    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)
8897      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
8898      else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_AUTO_POSSESS_RIGHTPAR, 16) == 0)
8899        { skipatstart += 18; options |= PCRE_NO_AUTO_POSSESS; continue; }
8900    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)
8901      { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; }      { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; }
8902    
# Line 8423  cd->named_group_list_size = NAMED_GROUP_ Line 9091  cd->named_group_list_size = NAMED_GROUP_
9091  cd->start_pattern = (const pcre_uchar *)pattern;  cd->start_pattern = (const pcre_uchar *)pattern;
9092  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));
9093  cd->req_varyopt = 0;  cd->req_varyopt = 0;
9094    cd->parens_depth = 0;
9095  cd->assert_depth = 0;  cd->assert_depth = 0;
9096  cd->max_lookbehind = 0;  cd->max_lookbehind = 0;
9097  cd->external_options = options;  cd->external_options = options;
# Line 8437  outside can help speed up starting point Line 9106  outside can help speed up starting point
9106  ptr += skipatstart;  ptr += skipatstart;
9107  code = cworkspace;  code = cworkspace;
9108  *code = OP_BRA;  *code = OP_BRA;
9109    
9110  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
9111    FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL,    FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL,
9112    cd, &length);    cd, &length);
# Line 8508  field; this time it's used for rememberi Line 9178  field; this time it's used for rememberi
9178  */  */
9179    
9180  cd->final_bracount = cd->bracount;  /* Save for checking forward references */  cd->final_bracount = cd->bracount;  /* Save for checking forward references */
9181    cd->parens_depth = 0;
9182  cd->assert_depth = 0;  cd->assert_depth = 0;
9183  cd->bracount = 0;  cd->bracount = 0;
9184  cd->max_lookbehind = 0;  cd->max_lookbehind = 0;
# Line 8610  subpattern. */ Line 9281  subpattern. */
9281    
9282  if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15;  if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15;
9283    
9284  /* Unless disabled, check whether single character iterators can be  /* Unless disabled, check whether any single character iterators can be
9285  auto-possessified. The function overwrites the appropriate opcode values. */  auto-possessified. The function overwrites the appropriate opcode values, so
9286    the type of the pointer must be cast. NOTE: the intermediate variable "temp" is
9287    used in this code because at least one compiler gives a warning about loss of
9288    "const" attribute if the cast (pcre_uchar *)codestart is used directly in the
9289    function call. */
9290    
9291  if ((options & PCRE_NO_AUTO_POSSESSIFY) == 0)  if ((options & PCRE_NO_AUTO_POSSESS) == 0)
9292    auto_possessify((pcre_uchar *)codestart, utf, cd);    {
9293      pcre_uchar *temp = (pcre_uchar *)codestart;
9294      auto_possessify(temp, utf, cd);
9295      }
9296    
9297  /* If there were any lookbehind assertions that contained OP_RECURSE  /* If there were any lookbehind assertions that contained OP_RECURSE
9298  (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,
# Line 8837  return (pcre32 *)re; Line 9515  return (pcre32 *)re;
9515  }  }
9516    
9517  /* End of pcre_compile.c */  /* End of pcre_compile.c */
9518    

Legend:
Removed from v.1363  
changed lines
  Added in v.1495

  ViewVC Help
Powered by ViewVC 1.1.5