/[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 1502 by ph10, Mon Sep 15 13:56:18 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 2226  for (code = first_significant_code(code Line 2374  for (code = first_significant_code(code
2374    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2375      {      {
2376      const pcre_uchar *scode = cd->start_code + GET(code, 1);      const pcre_uchar *scode = cd->start_code + GET(code, 1);
2377        const pcre_uchar *endgroup = scode;
2378      BOOL empty_branch;      BOOL empty_branch;
2379    
2380      /* Test for forward reference or uncompleted reference. This is disabled      /* Test for forward reference or uncompleted reference. This is disabled
# Line 2240  for (code = first_significant_code(code Line 2389  for (code = first_significant_code(code
2389        if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */        if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2390        }        }
2391    
2392      /* If we are scanning a completed pattern, there are no forward references      /* If the reference is to a completed group, we need to detect whether this
2393      and all groups are complete. We need to detect whether this is a recursive      is a recursive call, as otherwise there will be an infinite loop. If it is
2394      call, as otherwise there will be an infinite loop. If it is a recursion,      a recursion, just skip over it. Simple recursions are easily detected. For
2395      just skip over it. Simple recursions are easily detected. For mutual      mutual recursions we keep a chain on the stack. */
     recursions we keep a chain on the stack. */  
2396    
2397        do endgroup += GET(endgroup, 1); while (*endgroup == OP_ALT);
2398        if (code >= scode && code <= endgroup) continue;  /* Simple recursion */
2399      else      else
2400        {        {
2401        recurse_check *r = recurses;        recurse_check *r = recurses;
       const pcre_uchar *endgroup = scode;  
   
       do endgroup += GET(endgroup, 1); while (*endgroup == OP_ALT);  
       if (code >= scode && code <= endgroup) continue;  /* Simple recursion */  
   
2402        for (r = recurses; r != NULL; r = r->prev)        for (r = recurses; r != NULL; r = r->prev)
2403          if (r->group == scode) break;          if (r->group == scode) break;
2404        if (r != NULL) continue;   /* Mutual recursion */        if (r != NULL) continue;   /* Mutual recursion */
# Line 2366  for (code = first_significant_code(code Line 2511  for (code = first_significant_code(code
2511        case OP_CRMINSTAR:        case OP_CRMINSTAR:
2512        case OP_CRQUERY:        case OP_CRQUERY:
2513        case OP_CRMINQUERY:        case OP_CRMINQUERY:
2514          case OP_CRPOSSTAR:
2515          case OP_CRPOSQUERY:
2516        break;        break;
2517    
2518        default:                   /* Non-repeat => class must match */        default:                   /* Non-repeat => class must match */
2519        case OP_CRPLUS:            /* These repeats aren't empty */        case OP_CRPLUS:            /* These repeats aren't empty */
2520        case OP_CRMINPLUS:        case OP_CRMINPLUS:
2521          case OP_CRPOSPLUS:
2522        return FALSE;        return FALSE;
2523    
2524        case OP_CRRANGE:        case OP_CRRANGE:
2525        case OP_CRMINRANGE:        case OP_CRMINRANGE:
2526          case OP_CRPOSRANGE:
2527        if (GET2(ccode, 1) > 0) return FALSE;  /* Minimum > 0 */        if (GET2(ccode, 1) > 0) return FALSE;  /* Minimum > 0 */
2528        break;        break;
2529        }        }
# Line 2650  switch(ptype) Line 2799  switch(ptype)
2799    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2800            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
2801    
2802    case PT_SPACE:    /* Perl space */    /* Perl space used to exclude VT, but from Perl 5.18 it is included, which
2803    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||    means that Perl space and POSIX space are now identical. PCRE was changed
2804            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)    at release 8.34. */
           == negated;  
2805    
2806      case PT_SPACE:    /* Perl space */
2807    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
2808    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||    switch(c)
2809            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||      {
2810            c == CHAR_FF || c == CHAR_CR)      HSPACE_CASES:
2811            == negated;      VSPACE_CASES:
2812        return negated;
2813    
2814        default:
2815        return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == negated;
2816        }
2817      break;  /* Control never reaches here */
2818    
2819    case PT_WORD:    case PT_WORD:
2820    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 2863  get_chr_property_list(const pcre_uchar *
2863    const pcre_uint8 *fcc, pcre_uint32 *list)    const pcre_uint8 *fcc, pcre_uint32 *list)
2864  {  {
2865  pcre_uchar c = *code;  pcre_uchar c = *code;
2866    pcre_uchar base;
2867  const pcre_uchar *end;  const pcre_uchar *end;
 const pcre_uint32 *clist_src;  
 pcre_uint32 *clist_dest;  
2868  pcre_uint32 chr;  pcre_uint32 chr;
2869  pcre_uchar base;  
2870    #ifdef SUPPORT_UCP
2871    pcre_uint32 *clist_dest;
2872    const pcre_uint32 *clist_src;
2873    #else
2874    utf = utf;  /* Suppress "unused parameter" compiler warning */
2875    #endif
2876    
2877  list[0] = c;  list[0] = c;
2878  list[1] = FALSE;  list[1] = FALSE;
# Line 2818  switch(c) Line 2978  switch(c)
2978      return code + 2;      return code + 2;
2979      }      }
2980    
2981    /* Convert only if we have anough space. */    /* Convert only if we have enough space. */
2982    
2983    clist_src = PRIV(ucd_caseless_sets) + code[1];    clist_src = PRIV(ucd_caseless_sets) + code[1];
2984    clist_dest = list + 2;    clist_dest = list + 2;
2985    code += 2;    code += 2;
2986    
2987    do {    do {
      /* Early return if there is not enough space. */  
2988       if (clist_dest >= list + 8)       if (clist_dest >= list + 8)
2989         {         {
2990           /* Early return if there is not enough space. This should never
2991           happen, since all clists are shorter than 5 character now. */
2992         list[2] = code[0];         list[2] = code[0];
2993         list[3] = code[1];         list[3] = code[1];
2994         return code;         return code;
2995         }         }
2996       *clist_dest++ = *clist_src;       *clist_dest++ = *clist_src;
2997       }       }
2998     while(*clist_src++ != NOTACHAR);    while(*clist_src++ != NOTACHAR);
2999    
3000    /* Enough space to store all characters. */    /* All characters are stored. The terminating NOTACHAR
3001      is copied form the clist itself. */
3002    
3003    list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT;    list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT;
3004    return code;    return code;
# Line 2846  switch(c) Line 3008  switch(c)
3008    case OP_CLASS:    case OP_CLASS:
3009  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3010    case OP_XCLASS:    case OP_XCLASS:
   
3011    if (c == OP_XCLASS)    if (c == OP_XCLASS)
3012      end = code + GET(code, 0);      end = code + GET(code, 0) - 1;
3013    else    else
3014  #endif  #endif
3015      end = code + 32 / sizeof(pcre_uchar);      end = code + 32 / sizeof(pcre_uchar);
# Line 2859  switch(c) Line 3020  switch(c)
3020      case OP_CRMINSTAR:      case OP_CRMINSTAR:
3021      case OP_CRQUERY:      case OP_CRQUERY:
3022      case OP_CRMINQUERY:      case OP_CRMINQUERY:
3023        case OP_CRPOSSTAR:
3024        case OP_CRPOSQUERY:
3025      list[1] = TRUE;      list[1] = TRUE;
3026      end++;      end++;
3027      break;      break;
3028    
3029        case OP_CRPLUS:
3030        case OP_CRMINPLUS:
3031        case OP_CRPOSPLUS:
3032        end++;
3033        break;
3034    
3035      case OP_CRRANGE:      case OP_CRRANGE:
3036      case OP_CRMINRANGE:      case OP_CRMINRANGE:
3037        case OP_CRPOSRANGE:
3038      list[1] = (GET2(end, 1) == 0);      list[1] = (GET2(end, 1) == 0);
3039      end += 1 + 2 * IMM2_SIZE;      end += 1 + 2 * IMM2_SIZE;
3040      break;      break;
3041      }      }
3042    list[2] = end - code;    list[2] = (pcre_uint32)(end - code);
3043    return end;    return end;
3044    }    }
3045  return NULL;    /* Opcode not accepted */  return NULL;    /* Opcode not accepted */
# Line 2895  Returns:      TRUE if the auto-possessif Line 3065  Returns:      TRUE if the auto-possessif
3065    
3066  static BOOL  static BOOL
3067  compare_opcodes(const pcre_uchar *code, BOOL utf, const compile_data *cd,  compare_opcodes(const pcre_uchar *code, BOOL utf, const compile_data *cd,
3068    const pcre_uint32* base_list)    const pcre_uint32 *base_list, const pcre_uchar *base_end)
3069  {  {
3070  pcre_uchar c;  pcre_uchar c;
3071  pcre_uint32 list[8];  pcre_uint32 list[8];
3072  const pcre_uint32* chr_ptr;  const pcre_uint32 *chr_ptr;
3073  const pcre_uint32* ochr_ptr;  const pcre_uint32 *ochr_ptr;
3074  const pcre_uint32* list_ptr;  const pcre_uint32 *list_ptr;
3075    const pcre_uchar *next_code;
3076    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3077    const pcre_uchar *xclass_flags;
3078    #endif
3079    const pcre_uint8 *class_bitset;
3080    const pcre_uint8 *set1, *set2, *set_end;
3081  pcre_uint32 chr;  pcre_uint32 chr;
3082    BOOL accepted, invert_bits;
3083    BOOL entered_a_group = FALSE;
3084    
3085    /* Note: the base_list[1] contains whether the current opcode has greedy
3086    (represented by a non-zero value) quantifier. This is a different from
3087    other character type lists, which stores here that the character iterator
3088    matches to an empty string (also represented by a non-zero value). */
3089    
3090  for(;;)  for(;;)
3091    {    {
3092      /* All operations move the code pointer forward.
3093      Therefore infinite recursions are not possible. */
3094    
3095    c = *code;    c = *code;
3096    
3097    /* Skip over callouts */    /* Skip over callouts */
# Line 2925  for(;;) Line 3111  for(;;)
3111    switch(c)    switch(c)
3112      {      {
3113      case OP_END:      case OP_END:
3114      /* TRUE only in greedy case. The non-greedy case could be replaced by an      case OP_KETRPOS:
3115      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
3116      more memory, which we cannot get at this stage.) */      an OP_EXACT, but it is probably not worth it. (And note that OP_EXACT
3117        uses more memory, which we cannot get at this stage.) */
3118    
3119      return base_list[1] != 0;      return base_list[1] != 0;
3120    
3121      case OP_KET:      case OP_KET:
3122      /* If the bracket is capturing, and referenced by an OP_RECURSE, the      /* If the bracket is capturing, and referenced by an OP_RECURSE, or
3123      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
3124      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.) */  
3125    
3126      if (base_list[1] == 0) return FALSE;      if (base_list[1] == 0) return FALSE;
3127    
3128        switch(*(code - GET(code, 1)))
3129          {
3130          case OP_ASSERT:
3131          case OP_ASSERT_NOT:
3132          case OP_ASSERTBACK:
3133          case OP_ASSERTBACK_NOT:
3134          case OP_ONCE:
3135          case OP_ONCE_NC:
3136          /* Atomic sub-patterns and assertions can always auto-possessify their
3137          last iterator. However, if the group was entered as a result of checking
3138          a previous iterator, this is not possible. */
3139    
3140          return !entered_a_group;
3141          }
3142    
3143        code += PRIV(OP_lengths)[c];
3144        continue;
3145    
3146        case OP_ONCE:
3147        case OP_ONCE_NC:
3148        case OP_BRA:
3149        case OP_CBRA:
3150        next_code = code + GET(code, 1);
3151        code += PRIV(OP_lengths)[c];
3152    
3153        while (*next_code == OP_ALT)
3154          {
3155          if (!compare_opcodes(code, utf, cd, base_list, base_end)) return FALSE;
3156          code = next_code + 1 + LINK_SIZE;
3157          next_code += GET(next_code, 1);
3158          }
3159    
3160        entered_a_group = TRUE;
3161        continue;
3162    
3163        case OP_BRAZERO:
3164        case OP_BRAMINZERO:
3165    
3166        next_code = code + 1;
3167        if (*next_code != OP_BRA && *next_code != OP_CBRA
3168            && *next_code != OP_ONCE && *next_code != OP_ONCE_NC) return FALSE;
3169    
3170        do next_code += GET(next_code, 1); while (*next_code == OP_ALT);
3171    
3172        /* The bracket content will be checked by the
3173        OP_BRA/OP_CBRA case above. */
3174        next_code += 1 + LINK_SIZE;
3175        if (!compare_opcodes(next_code, utf, cd, base_list, base_end))
3176          return FALSE;
3177    
3178      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
3179      continue;      continue;
3180    
3181        default:
3182        break;
3183      }      }
3184    
3185    /* Check for a supported opcode, and load its properties. */    /* Check for a supported opcode, and load its properties. */
# Line 2961  for(;;) Line 3201  for(;;)
3201      list_ptr = base_list;      list_ptr = base_list;
3202      }      }
3203    
3204      /* Character bitsets can also be compared to certain opcodes. */
3205    
3206      else if (base_list[0] == OP_CLASS || list[0] == OP_CLASS
3207    #ifdef COMPILE_PCRE8
3208          /* In 8 bit, non-UTF mode, OP_CLASS and OP_NCLASS are the same. */
3209          || (!utf && (base_list[0] == OP_NCLASS || list[0] == OP_NCLASS))
3210    #endif
3211          )
3212        {
3213    #ifdef COMPILE_PCRE8
3214        if (base_list[0] == OP_CLASS || (!utf && base_list[0] == OP_NCLASS))
3215    #else
3216        if (base_list[0] == OP_CLASS)
3217    #endif
3218          {
3219          set1 = (pcre_uint8 *)(base_end - base_list[2]);
3220          list_ptr = list;
3221          }
3222        else
3223          {
3224          set1 = (pcre_uint8 *)(code - list[2]);
3225          list_ptr = base_list;
3226          }
3227    
3228        invert_bits = FALSE;
3229        switch(list_ptr[0])
3230          {
3231          case OP_CLASS:
3232          case OP_NCLASS:
3233          set2 = (pcre_uint8 *)
3234            ((list_ptr == list ? code : base_end) - list_ptr[2]);
3235          break;
3236    
3237    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3238          case OP_XCLASS:
3239          xclass_flags = (list_ptr == list ? code : base_end) - list_ptr[2] + LINK_SIZE;
3240          if ((*xclass_flags & XCL_HASPROP) != 0) return FALSE;
3241          if ((*xclass_flags & XCL_MAP) == 0)
3242            {
3243            /* No bits are set for characters < 256. */
3244            if (list[1] == 0) return TRUE;
3245            /* Might be an empty repeat. */
3246            continue;
3247            }
3248          set2 = (pcre_uint8 *)(xclass_flags + 1);
3249          break;
3250    #endif
3251    
3252          case OP_NOT_DIGIT:
3253          invert_bits = TRUE;
3254          /* Fall through */
3255          case OP_DIGIT:
3256          set2 = (pcre_uint8 *)(cd->cbits + cbit_digit);
3257          break;
3258    
3259          case OP_NOT_WHITESPACE:
3260          invert_bits = TRUE;
3261          /* Fall through */
3262          case OP_WHITESPACE:
3263          set2 = (pcre_uint8 *)(cd->cbits + cbit_space);
3264          break;
3265    
3266          case OP_NOT_WORDCHAR:
3267          invert_bits = TRUE;
3268          /* Fall through */
3269          case OP_WORDCHAR:
3270          set2 = (pcre_uint8 *)(cd->cbits + cbit_word);
3271          break;
3272    
3273          default:
3274          return FALSE;
3275          }
3276    
3277        /* Because the sets are unaligned, we need
3278        to perform byte comparison here. */
3279        set_end = set1 + 32;
3280        if (invert_bits)
3281          {
3282          do
3283            {
3284            if ((*set1++ & ~(*set2++)) != 0) return FALSE;
3285            }
3286          while (set1 < set_end);
3287          }
3288        else
3289          {
3290          do
3291            {
3292            if ((*set1++ & *set2++) != 0) return FALSE;
3293            }
3294          while (set1 < set_end);
3295          }
3296    
3297        if (list[1] == 0) return TRUE;
3298        /* Might be an empty repeat. */
3299        continue;
3300        }
3301    
3302    /* Some property combinations also acceptable. Unicode property opcodes are    /* Some property combinations also acceptable. Unicode property opcodes are
3303    processed specially; the rest can be handled with a lookup table. */    processed specially; the rest can be handled with a lookup table. */
3304    
# Line 2968  for(;;) Line 3306  for(;;)
3306      {      {
3307      pcre_uint32 leftop, rightop;      pcre_uint32 leftop, rightop;
3308    
     if (list[1] != 0) return FALSE;   /* Must match at least one character */  
3309      leftop = base_list[0];      leftop = base_list[0];
3310      rightop = list[0];      rightop = list[0];
3311    
3312  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3313        accepted = FALSE; /* Always set in non-unicode case. */
3314      if (leftop == OP_PROP || leftop == OP_NOTPROP)      if (leftop == OP_PROP || leftop == OP_NOTPROP)
3315        {        {
3316        if (rightop == OP_EOD) return TRUE;        if (rightop == OP_EOD)
3317        if (rightop == OP_PROP || rightop == OP_NOTPROP)          accepted = TRUE;
3318          else if (rightop == OP_PROP || rightop == OP_NOTPROP)
3319          {          {
3320          int n;          int n;
3321          const pcre_uint8 *p;          const pcre_uint8 *p;
# Line 2997  for(;;) Line 3336  for(;;)
3336          n = propposstab[base_list[2]][list[2]];          n = propposstab[base_list[2]][list[2]];
3337          switch(n)          switch(n)
3338            {            {
3339            case 0: return FALSE;            case 0: break;
3340            case 1: return bothprop;            case 1: accepted = bothprop; break;
3341            case 2: return (base_list[3] == list[3]) != same;            case 2: accepted = (base_list[3] == list[3]) != same; break;
3342            case 3: return !same;            case 3: accepted = !same; break;
3343    
3344            case 4:  /* Left general category, right particular category */            case 4:  /* Left general category, right particular category */
3345            return risprop && catposstab[base_list[3]][list[3]] == same;            accepted = risprop && catposstab[base_list[3]][list[3]] == same;
3346              break;
3347    
3348            case 5:  /* Right general category, left particular category */            case 5:  /* Right general category, left particular category */
3349            return lisprop && catposstab[list[3]][base_list[3]] == same;            accepted = lisprop && catposstab[list[3]][base_list[3]] == same;
3350              break;
3351    
3352            /* This code is logically tricky. Think hard before fiddling with it.            /* This code is logically tricky. Think hard before fiddling with it.
3353            The posspropstab table has four entries per row. Each row relates to            The posspropstab table has four entries per row. Each row relates to
3354            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.
3355            Only WORD actually needs all four entries, but using repeats for the            Only WORD actually needs all four entries, but using repeats for the
3356            others means they can all use the same code below.            others means they can all use the same code below.
3357    
3358            The first two entries in each row are Unicode general categories, and            The first two entries in each row are Unicode general categories, and
3359            apply always, because all the characters they include are part of the            apply always, because all the characters they include are part of the
3360            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 3364  for(;;)
3364            category contains more characters than the specials that are defined            category contains more characters than the specials that are defined
3365            for the property being tested against. Therefore, it cannot be used            for the property being tested against. Therefore, it cannot be used
3366            in a NOTPROP case.            in a NOTPROP case.
3367    
3368            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.
3369            Underscore is covered by ucp_P or ucp_Po. */            Underscore is covered by ucp_P or ucp_Po. */
3370    
# Line 3031  for(;;) Line 3372  for(;;)
3372            case 7:  /* Left space vs right general category */            case 7:  /* Left space vs right general category */
3373            case 8:  /* Left word vs right general category */            case 8:  /* Left word vs right general category */
3374            p = posspropstab[n-6];            p = posspropstab[n-6];
3375            return risprop && lisprop ==            accepted = risprop && lisprop ==
3376              (list[3] != p[0] &&              (list[3] != p[0] &&
3377               list[3] != p[1] &&               list[3] != p[1] &&
3378              (list[3] != p[2] || !lisprop));              (list[3] != p[2] || !lisprop));
3379              break;
3380    
3381            case 9:   /* Right alphanum vs left general category */            case 9:   /* Right alphanum vs left general category */
3382            case 10:  /* Right space vs left general category */            case 10:  /* Right space vs left general category */
3383            case 11:  /* Right word vs left general category */            case 11:  /* Right word vs left general category */
3384            p = posspropstab[n-9];            p = posspropstab[n-9];
3385            return lisprop && risprop ==            accepted = lisprop && risprop ==
3386              (base_list[3] != p[0] &&              (base_list[3] != p[0] &&
3387               base_list[3] != p[1] &&               base_list[3] != p[1] &&
3388              (base_list[3] != p[2] || !risprop));              (base_list[3] != p[2] || !risprop));
3389              break;
3390    
3391            case 12:  /* Left alphanum vs right particular category */            case 12:  /* Left alphanum vs right particular category */
3392            case 13:  /* Left space vs right particular category */            case 13:  /* Left space vs right particular category */
3393            case 14:  /* Left word vs right particular category */            case 14:  /* Left word vs right particular category */
3394            p = posspropstab[n-12];            p = posspropstab[n-12];
3395            return risprop && lisprop ==            accepted = risprop && lisprop ==
3396              (catposstab[p[0]][list[3]] &&              (catposstab[p[0]][list[3]] &&
3397               catposstab[p[1]][list[3]] &&               catposstab[p[1]][list[3]] &&
3398              (list[3] != p[3] || !lisprop));              (list[3] != p[3] || !lisprop));
3399              break;
3400    
3401            case 15:  /* Right alphanum vs left particular category */            case 15:  /* Right alphanum vs left particular category */
3402            case 16:  /* Right space vs left particular category */            case 16:  /* Right space vs left particular category */
3403            case 17:  /* Right word vs left particular category */            case 17:  /* Right word vs left particular category */
3404            p = posspropstab[n-15];            p = posspropstab[n-15];
3405            return lisprop && risprop ==            accepted = lisprop && risprop ==
3406              (catposstab[p[0]][base_list[3]] &&              (catposstab[p[0]][base_list[3]] &&
3407               catposstab[p[1]][base_list[3]] &&               catposstab[p[1]][base_list[3]] &&
3408              (base_list[3] != p[3] || !risprop));              (base_list[3] != p[3] || !risprop));
3409              break;
3410            }            }
3411          }          }
       return FALSE;  
3412        }        }
3413    
3414      else      else
3415  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3416    
3417      return leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP &&      accepted = leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP &&
3418             rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP &&             rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP &&
3419             autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP];             autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP];
3420    
3421        if (!accepted) return FALSE;
3422    
3423        if (list[1] == 0) return TRUE;
3424        /* Might be an empty repeat. */
3425        continue;
3426      }      }
3427    
3428    /* 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 3536  for(;;)
3536        case OP_EOD:    /* Can always possessify before \z */        case OP_EOD:    /* Can always possessify before \z */
3537        break;        break;
3538    
3539    #ifdef SUPPORT_UCP
3540        case OP_PROP:        case OP_PROP:
3541        case OP_NOTPROP:        case OP_NOTPROP:
3542        if (!check_char_prop(chr, list_ptr[2], list_ptr[3],        if (!check_char_prop(chr, list_ptr[2], list_ptr[3],
3543              list_ptr[0] == OP_NOTPROP))              list_ptr[0] == OP_NOTPROP))
3544          return FALSE;          return FALSE;
3545        break;        break;
3546    #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. */  
3547    
3548        case OP_NCLASS:        case OP_NCLASS:
3549        if (chr > 255) return FALSE;        if (chr > 255) return FALSE;
3550        /* Fall through */        /* Fall through */
3551    
3552        case OP_CLASS:        case OP_CLASS:
       if (list_ptr != list) return FALSE;   /* Class is first opcode */  
3553        if (chr > 255) break;        if (chr > 255) break;
3554        if ((((pcre_uint8 *)(code - list_ptr[2] + 1))[chr >> 3] & (1 << (chr & 7))) != 0)        class_bitset = (pcre_uint8 *)
3555          return FALSE;          ((list_ptr == list ? code : base_end) - list_ptr[2]);
3556          if ((class_bitset[chr >> 3] & (1 << (chr & 7))) != 0) return FALSE;
3557        break;        break;
3558    
3559  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3560        case OP_XCLASS:        case OP_XCLASS:
3561        if (list_ptr != list) return FALSE;   /* Class is first opcode */        if (PRIV(xclass)(chr, (list_ptr == list ? code : base_end) -
3562        if (PRIV(xclass)(chr, code - list_ptr[2] + 1 + LINK_SIZE, utf))            list_ptr[2] + LINK_SIZE, utf)) return FALSE;
         return FALSE;  
3563        break;        break;
3564  #endif  #endif
3565    
# Line 3232  for(;;) Line 3576  for(;;)
3576    if (list[1] == 0) return TRUE;    if (list[1] == 0) return TRUE;
3577    }    }
3578    
3579  return FALSE;  /* Control never reaches here. There used to be a fail-save return FALSE; here,
3580    but some compilers complain about an unreachable statement. */
3581    
3582  }  }
3583    
3584    
# Line 3257  auto_possessify(pcre_uchar *code, BOOL u Line 3603  auto_possessify(pcre_uchar *code, BOOL u
3603  {  {
3604  register pcre_uchar c;  register pcre_uchar c;
3605  const pcre_uchar *end;  const pcre_uchar *end;
3606    pcre_uchar *repeat_opcode;
3607  pcre_uint32 list[8];  pcre_uint32 list[8];
3608    
3609  for (;;)  for (;;)
# Line 3270  for (;;) Line 3617  for (;;)
3617        get_chr_property_list(code, utf, cd->fcc, list) : NULL;        get_chr_property_list(code, utf, cd->fcc, list) : NULL;
3618      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;
3619    
3620      if (end != NULL && compare_opcodes(end, utf, cd, list))      if (end != NULL && compare_opcodes(end, utf, cd, list, end))
3621        {        {
3622        switch(c)        switch(c)
3623          {          {
# Line 3303  for (;;) Line 3650  for (;;)
3650          break;          break;
3651    
3652          case OP_MINUPTO:          case OP_MINUPTO:
3653          *code += OP_MINUPTO - OP_UPTO;          *code += OP_POSUPTO - OP_MINUPTO;
3654          break;          break;
3655          }          }
3656        }        }
3657      c = *code;      c = *code;
3658      }      }
3659      else if (c == OP_CLASS || c == OP_NCLASS || c == OP_XCLASS)
3660        {
3661    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3662        if (c == OP_XCLASS)
3663          repeat_opcode = code + GET(code, 1);
3664        else
3665    #endif
3666          repeat_opcode = code + 1 + (32 / sizeof(pcre_uchar));
3667    
3668        c = *repeat_opcode;
3669        if (c >= OP_CRSTAR && c <= OP_CRMINRANGE)
3670          {
3671          /* end must not be NULL. */
3672          end = get_chr_property_list(code, utf, cd->fcc, list);
3673    
3674          list[1] = (c & 1) == 0;
3675    
3676          if (compare_opcodes(end, utf, cd, list, end))
3677            {
3678            switch (c)
3679              {
3680              case OP_CRSTAR:
3681              case OP_CRMINSTAR:
3682              *repeat_opcode = OP_CRPOSSTAR;
3683              break;
3684    
3685              case OP_CRPLUS:
3686              case OP_CRMINPLUS:
3687              *repeat_opcode = OP_CRPOSPLUS;
3688              break;
3689    
3690              case OP_CRQUERY:
3691              case OP_CRMINQUERY:
3692              *repeat_opcode = OP_CRPOSQUERY;
3693              break;
3694    
3695              case OP_CRRANGE:
3696              case OP_CRMINRANGE:
3697              *repeat_opcode = OP_CRPOSRANGE;
3698              break;
3699              }
3700            }
3701          }
3702        c = *code;
3703        }
3704    
3705    switch(c)    switch(c)
3706      {      {
# Line 3335  for (;;) Line 3727  for (;;)
3727        code += 2;        code += 2;
3728      break;      break;
3729    
3730    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3731      case OP_XCLASS:      case OP_XCLASS:
3732      code += GET(code, 1);      code += GET(code, 1);
3733      break;      break;
3734    #endif
3735    
3736      case OP_MARK:      case OP_MARK:
3737      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
# Line 3446  class, but [abc[:x\]pqr:]] is (so that a Line 3840  class, but [abc[:x\]pqr:]] is (so that a
3840  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
3841  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:]
3842  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
3843  "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,
3844  I think.  I think.
3845    
3846  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 4089  for (c = *cptr; c <= d; c++)
4089    
4090  if (c > d) return -1;  /* Reached end of range */  if (c > d) return -1;  /* Reached end of range */
4091    
4092    /* Found a character that has a single other case. Search for the end of the
4093    range, which is either the end of the input range, or a character that has zero
4094    or more than one other cases. */
4095    
4096  *ocptr = othercase;  *ocptr = othercase;
4097  next = othercase + 1;  next = othercase + 1;
4098    
4099  for (++c; c <= d; c++)  for (++c; c <= d; c++)
4100    {    {
4101    if (UCD_OTHERCASE(c) != next) break;    if ((co = UCD_CASESET(c)) != 0 || UCD_OTHERCASE(c) != next) break;
4102    next++;    next++;
4103    }    }
4104    
# Line 3738  add_to_class(pcre_uint8 *classbits, pcre Line 4136  add_to_class(pcre_uint8 *classbits, pcre
4136    compile_data *cd, pcre_uint32 start, pcre_uint32 end)    compile_data *cd, pcre_uint32 start, pcre_uint32 end)
4137  {  {
4138  pcre_uint32 c;  pcre_uint32 c;
4139    pcre_uint32 classbits_end = (end <= 0xff ? end : 0xff);
4140  int n8 = 0;  int n8 = 0;
4141    
4142  /* 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 4180  if ((options & PCRE_CASELESS) != 0)
4180    
4181    /* Not UTF-mode, or no UCP */    /* Not UTF-mode, or no UCP */
4182    
4183    for (c = start; c <= end && c < 256; c++)    for (c = start; c <= classbits_end; c++)
4184      {      {
4185      SETBIT(classbits, cd->fcc[c]);      SETBIT(classbits, cd->fcc[c]);
4186      n8++;      n8++;
# Line 3806  in all cases. */ Line 4205  in all cases. */
4205    
4206  #endif /* COMPILE_PCRE[8|16] */  #endif /* COMPILE_PCRE[8|16] */
4207    
4208  /* 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. */  
4209    
4210  if (end < 0x100)  for (c = start; c <= classbits_end; c++)
4211    {    {
4212    for (c = start; c <= end; c++)    /* Regardless of start, c will always be <= 255. */
4213      {    SETBIT(classbits, c);
4214      n8++;    n8++;
     SETBIT(classbits, c);  
     }  
4215    }    }
4216    
4217  else  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4218    if (start <= 0xff) start = 0xff + 1;
4219    
4220    if (end >= start)
4221    {    {
4222    pcre_uchar *uchardata = *uchardptr;    pcre_uchar *uchardata = *uchardptr;
   
4223  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4224    if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */    if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
4225      {      {
# Line 3861  else Line 4259  else
4259    
4260    *uchardptr = uchardata;   /* Updata extra data pointer */    *uchardptr = uchardata;   /* Updata extra data pointer */
4261    }    }
4262    #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */
4263    
4264  return n8;    /* Number of 8-bit characters */  return n8;    /* Number of 8-bit characters */
4265  }  }
# Line 4082  for (;; ptr++) Line 4481  for (;; ptr++)
4481    BOOL reset_bracount;    BOOL reset_bracount;
4482    int class_has_8bitchar;    int class_has_8bitchar;
4483    int class_one_char;    int class_one_char;
4484    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4485      BOOL xclass_has_prop;
4486    #endif
4487    int newoptions;    int newoptions;
4488    int recno;    int recno;
4489    int refsign;    int refsign;
# Line 4199  for (;; ptr++) Line 4601  for (;; ptr++)
4601          }          }
4602        goto NORMAL_CHAR;        goto NORMAL_CHAR;
4603        }        }
4604        /* Control does not reach here. */
4605      }      }
4606    
4607    /* 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
4608    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. */  
4609    
4610    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
4611      {      {
4612      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;      for (;;)
     if (c == CHAR_NUMBER_SIGN)  
4613        {        {
4614          while (MAX_255(c) && (cd->ctypes[c] & ctype_space) != 0) c = *(++ptr);
4615          if (c != CHAR_NUMBER_SIGN) break;
4616        ptr++;        ptr++;
4617        while (*ptr != CHAR_NULL)        while (*ptr != CHAR_NULL)
4618          {          {
4619          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr))         /* For non-fixed-length newline cases, */
4620              {                          /* IS_NEWLINE sets cd->nllen. */
4621              ptr += cd->nllen;
4622              break;
4623              }
4624          ptr++;          ptr++;
4625  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4626          if (utf) FORWARDCHAR(ptr);          if (utf) FORWARDCHAR(ptr);
4627  #endif  #endif
4628          }          }
4629        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;  
4630        }        }
4631      }      }
4632    
4633    /* No auto callout for quantifiers. */    /* See if the next thing is a quantifier. */
4634    
4635      is_quantifier =
4636        c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||
4637        (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));
4638    
4639      /* Fill in length of a previous callout, except when the next thing is a
4640      quantifier or when processing a property substitution string in UCP mode. */
4641    
4642    if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier)    if (!is_quantifier && previous_callout != NULL && nestptr == NULL &&
4643           after_manual_callout-- <= 0)
4644        {
4645        if (lengthptr == NULL)      /* Don't attempt in pre-compile phase */
4646          complete_callout(previous_callout, ptr, cd);
4647        previous_callout = NULL;
4648        }
4649    
4650      /* Create auto callout, except for quantifiers, or while processing property
4651      strings that are substituted for \w etc in UCP mode. */
4652    
4653      if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier && nestptr == NULL)
4654      {      {
4655      previous_callout = code;      previous_callout = code;
4656      code = auto_callout(code, ptr, cd);      code = auto_callout(code, ptr, cd);
4657      }      }
4658    
4659      /* Process the next pattern item. */
4660    
4661    switch(c)    switch(c)
4662      {      {
4663      /* ===================================================================*/      /* ===================================================================*/
4664      case 0:                        /* The branch terminates at string end */      case CHAR_NULL:                /* The branch terminates at string end */
4665      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
4666      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
4667      *firstcharptr = firstchar;      *firstcharptr = firstchar;
# Line 4280  for (;; ptr++) Line 4691  for (;; ptr++)
4691      previous = NULL;      previous = NULL;
4692      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
4693        {        {
4694        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;        if (firstcharflags == REQ_UNSET)
4695            zerofirstcharflags = firstcharflags = REQ_NONE;
4696        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
4697        }        }
4698      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 4328  for (;; ptr++) Line 4740  for (;; ptr++)
4740        }        }
4741      goto NORMAL_CHAR;      goto NORMAL_CHAR;
4742    
4743        /* In another (POSIX) regex library, the ugly syntax [[:<:]] and [[:>:]] is
4744        used for "start of word" and "end of word". As these are otherwise illegal
4745        sequences, we don't break anything by recognizing them. They are replaced
4746        by \b(?=\w) and \b(?<=\w) respectively. Sequences like [a[:<:]] are
4747        erroneous and are handled by the normal code below. */
4748    
4749      case CHAR_LEFT_SQUARE_BRACKET:      case CHAR_LEFT_SQUARE_BRACKET:
4750        if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_STARTWORD, 6) == 0)
4751          {
4752          nestptr = ptr + 7;
4753          ptr = sub_start_of_word - 1;
4754          continue;
4755          }
4756    
4757        if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_ENDWORD, 6) == 0)
4758          {
4759          nestptr = ptr + 7;
4760          ptr = sub_end_of_word - 1;
4761          continue;
4762          }
4763    
4764        /* Handle a real character class. */
4765    
4766      previous = code;      previous = code;
4767    
4768      /* 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 4819  for (;; ptr++)
4819    
4820      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4821    
4822        /* Extended class (xclass) will be used when characters > 255
4823        might match. */
4824    
4825    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4826        xclass = FALSE;
4827        class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4828        class_uchardata_base = class_uchardata;   /* Save the start */
4829    #endif
4830    
4831      /* For optimization purposes, we track some properties of the class:      /* For optimization purposes, we track some properties of the class:
4832      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 <
4833      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
4834      character. */      character; xclass_has_prop will be TRUE if unicode property checks
4835        are present in the class. */
4836    
4837      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4838      class_one_char = 0;      class_one_char = 0;
4839    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4840        xclass_has_prop = FALSE;
4841    #endif
4842    
4843      /* 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
4844      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 4847  for (;; ptr++)
4847    
4848      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4849    
 #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  
   
4850      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4851      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
4852      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
# Line 4431  for (;; ptr++) Line 4872  for (;; ptr++)
4872        if (lengthptr != NULL && class_uchardata > class_uchardata_base)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4873          {          {
4874          xclass = TRUE;          xclass = TRUE;
4875          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += (int)(class_uchardata - class_uchardata_base);
4876          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4877          }          }
4878  #endif  #endif
# Line 4493  for (;; ptr++) Line 4934  for (;; ptr++)
4934            posix_class = 0;            posix_class = 0;
4935    
4936          /* 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
4937          different escape sequences that use Unicode properties. */          different escape sequences that use Unicode properties \p or \P. Others
4938            that are not available via \p or \P generate XCL_PROP/XCL_NOTPROP
4939            directly. */
4940    
4941  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4942          if ((options & PCRE_UCP) != 0)          if ((options & PCRE_UCP) != 0)
4943            {            {
4944              unsigned int ptype = 0;
4945            int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);            int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
4946    
4947              /* The posix_substitutes table specifies which POSIX classes can be
4948              converted to \p or \P items. */
4949    
4950            if (posix_substitutes[pc] != NULL)            if (posix_substitutes[pc] != NULL)
4951              {              {
4952              nestptr = tempptr + 1;              nestptr = tempptr + 1;
4953              ptr = posix_substitutes[pc] - 1;              ptr = posix_substitutes[pc] - 1;
4954              continue;              continue;
4955              }              }
4956    
4957              /* There are three other classes that generate special property calls
4958              that are recognized only in an XCLASS. */
4959    
4960              else switch(posix_class)
4961                {
4962                case PC_GRAPH:
4963                ptype = PT_PXGRAPH;
4964                /* Fall through */
4965                case PC_PRINT:
4966                if (ptype == 0) ptype = PT_PXPRINT;
4967                /* Fall through */
4968                case PC_PUNCT:
4969                if (ptype == 0) ptype = PT_PXPUNCT;
4970                *class_uchardata++ = local_negate? XCL_NOTPROP : XCL_PROP;
4971                *class_uchardata++ = ptype;
4972                *class_uchardata++ = 0;
4973                xclass_has_prop = TRUE;
4974                ptr = tempptr + 1;
4975                continue;
4976    
4977                /* For all other POSIX classes, no special action is taken in UCP
4978                mode. Fall through to the non_UCP case. */
4979    
4980                default:
4981                break;
4982                }
4983            }            }
4984  #endif  #endif
4985          /* 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
4986          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
4987          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
4988          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
4989            bit map that is being built. */
4990    
4991          posix_class *= 3;          posix_class *= 3;
4992    
# Line 4627  for (;; ptr++) Line 5103  for (;; ptr++)
5103              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
5104              continue;              continue;
5105    
5106              /* 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
5107              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
5108              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and              previously set by something earlier in the character class.
5109              EBCDIC, so we lazily just adjust the appropriate bit. */              Luckily, the value of CHAR_VT is 0x0b in both ASCII and EBCDIC, so
5110                we could just adjust the appropriate bit. From PCRE 8.34 we no
5111                longer treat \s and \S specially. */
5112    
5113              case ESC_s:              case ESC_s:
5114              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];  
5115              continue;              continue;
5116    
5117              case ESC_S:              case ESC_S:
5118              should_flip_negation = TRUE;              should_flip_negation = TRUE;
5119              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 */  
5120              continue;              continue;
5121    
5122              /* 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 5153  for (;; ptr++)
5153                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
5154                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
5155                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
5156                  xclass_has_prop = TRUE;
5157                class_has_8bitchar--;                /* Undo! */                class_has_8bitchar--;                /* Undo! */
5158                continue;                continue;
5159                }                }
# Line 4763  for (;; ptr++) Line 5239  for (;; ptr++)
5239  #endif  #endif
5240          d = *ptr;  /* Not UTF-8 mode */          d = *ptr;  /* Not UTF-8 mode */
5241    
5242          /* 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
5243          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
5244          in such circumstances. */          literal in such circumstances. However, in Perl's warning mode, a
5245            warning is given, so PCRE now faults it as it is almost certainly a
5246            mistake on the user's part. */
5247    
5248          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq)
5249            {            {
5250            int descape;            if (d == CHAR_BACKSLASH)
5251            descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);              {
5252            if (*errorcodeptr != 0) goto FAILED;              int descape;
5253                descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
5254                if (*errorcodeptr != 0) goto FAILED;
5255    
5256            /* \b is backspace; any other special means the '-' was literal. */              /* 0 means a character was put into d; \b is backspace; any other
5257                special causes an error. */
5258    
5259            if (descape != 0)              if (descape != 0)
             {  
             if (descape == ESC_b) d = CHAR_BS; else  
5260                {                {
5261                ptr = oldptr;                if (descape == ESC_b) d = CHAR_BS; else
5262                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */                  {
5263                    *errorcodeptr = ERR83;
5264                    goto FAILED;
5265                    }
5266                }                }
5267              }              }
5268    
5269              /* A hyphen followed by a POSIX class is treated in the same way. */
5270    
5271              else if (d == CHAR_LEFT_SQUARE_BRACKET &&
5272                       (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
5273                        ptr[1] == CHAR_EQUALS_SIGN) &&
5274                       check_posix_syntax(ptr, &tempptr))
5275                {
5276                *errorcodeptr = ERR83;
5277                goto FAILED;
5278                }
5279            }            }
5280    
5281          /* 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 5322  for (;; ptr++)
5322        whatever repeat count may follow. In the case of reqchar, save the        whatever repeat count may follow. In the case of reqchar, save the
5323        previous value for reinstating. */        previous value for reinstating. */
5324    
5325        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)        if (!inescq && class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
5326          {          {
5327          ptr++;          ptr++;
5328          zeroreqchar = reqchar;          zeroreqchar = reqchar;
# Line 4955  for (;; ptr++) Line 5448  for (;; ptr++)
5448        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
5449        code += LINK_SIZE;        code += LINK_SIZE;
5450        *code = negate_class? XCL_NOT:0;        *code = negate_class? XCL_NOT:0;
5451          if (xclass_has_prop) *code |= XCL_HASPROP;
5452    
5453        /* 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;
5454        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 5458  for (;; ptr++)
5458          *code++ |= XCL_MAP;          *code++ |= XCL_MAP;
5459          memmove(code + (32 / sizeof(pcre_uchar)), code,          memmove(code + (32 / sizeof(pcre_uchar)), code,
5460            IN_UCHARS(class_uchardata - code));            IN_UCHARS(class_uchardata - code));
5461            if (negate_class && !xclass_has_prop)
5462              for (c = 0; c < 32; c++) classbits[c] = ~classbits[c];
5463          memcpy(code, classbits, 32);          memcpy(code, classbits, 32);
5464          code = class_uchardata + (32 / sizeof(pcre_uchar));          code = class_uchardata + (32 / sizeof(pcre_uchar));
5465          }          }
# Line 5046  for (;; ptr++) Line 5542  for (;; ptr++)
5542    
5543      tempcode = previous;      tempcode = previous;
5544    
5545        /* Before checking for a possessive quantifier, we must skip over
5546        whitespace and comments in extended mode because Perl allows white space at
5547        this point. */
5548    
5549        if ((options & PCRE_EXTENDED) != 0)
5550          {
5551          const pcre_uchar *p = ptr + 1;
5552          for (;;)
5553            {
5554            while (MAX_255(*p) && (cd->ctypes[*p] & ctype_space) != 0) p++;
5555            if (*p != CHAR_NUMBER_SIGN) break;
5556            p++;
5557            while (*p != CHAR_NULL)
5558              {
5559              if (IS_NEWLINE(p))         /* For non-fixed-length newline cases, */
5560                {                        /* IS_NEWLINE sets cd->nllen. */
5561                p += cd->nllen;
5562                break;
5563                }
5564              p++;
5565    #ifdef SUPPORT_UTF
5566              if (utf) FORWARDCHAR(p);
5567    #endif
5568              }           /* Loop for comment characters */
5569            }             /* Loop for multiple comments */
5570          ptr = p - 1;    /* Character before the next significant one. */
5571          }
5572    
5573      /* If the next character is '+', we have a possessive quantifier. This      /* If the next character is '+', we have a possessive quantifier. This
5574      implies greediness, whatever the setting of the PCRE_UNGREEDY option.      implies greediness, whatever the setting of the PCRE_UNGREEDY option.
5575      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 5862  for (;; ptr++)
5862      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
5863      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 >=
5864      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,
5865      ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow      ASSERTBACK_NOT, ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND.
5866      repetition of assertions, but now it does, for Perl compatibility. */      Originally, PCRE did not allow repetition of assertions, but now it does,
5867        for Perl compatibility. */
5868    
5869      else if (*previous >= OP_ASSERT && *previous <= OP_COND)      else if (*previous >= OP_ASSERT && *previous <= OP_COND)
5870        {        {
# Line 5357  for (;; ptr++) Line 5882  for (;; ptr++)
5882        /* There is no sense in actually repeating assertions. The only potential        /* There is no sense in actually repeating assertions. The only potential
5883        use of repetition is in cases when the assertion is optional. Therefore,        use of repetition is in cases when the assertion is optional. Therefore,
5884        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
5885        maximum is not not zero or one, set it to 1. */        maximum is not zero or one, set it to 1. */
5886    
5887        if (*previous < OP_ONCE)    /* Assertion */        if (*previous < OP_ONCE)    /* Assertion */
5888          {          {
# Line 5492  for (;; ptr++) Line 6017  for (;; ptr++)
6017                while (cd->hwm > cd->start_workspace + cd->workspace_size -                while (cd->hwm > cd->start_workspace + cd->workspace_size -
6018                       WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))                       WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))
6019                  {                  {
6020                  int save_offset = save_hwm - cd->start_workspace;                  size_t save_offset = save_hwm - cd->start_workspace;
6021                  int this_offset = this_hwm - cd->start_workspace;                  size_t this_offset = this_hwm - cd->start_workspace;
6022                  *errorcodeptr = expand_workspace(cd);                  *errorcodeptr = expand_workspace(cd);
6023                  if (*errorcodeptr != 0) goto FAILED;                  if (*errorcodeptr != 0) goto FAILED;
6024                  save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;                  save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;
# Line 5574  for (;; ptr++) Line 6099  for (;; ptr++)
6099            while (cd->hwm > cd->start_workspace + cd->workspace_size -            while (cd->hwm > cd->start_workspace + cd->workspace_size -
6100                   WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))                   WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))
6101              {              {
6102              int save_offset = save_hwm - cd->start_workspace;              size_t save_offset = save_hwm - cd->start_workspace;
6103              int this_offset = this_hwm - cd->start_workspace;              size_t this_offset = this_hwm - cd->start_workspace;
6104              *errorcodeptr = expand_workspace(cd);              *errorcodeptr = expand_workspace(cd);
6105              if (*errorcodeptr != 0) goto FAILED;              if (*errorcodeptr != 0) goto FAILED;
6106              save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;              save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;
# Line 5730  for (;; ptr++) Line 6255  for (;; ptr++)
6255        goto FAILED;        goto FAILED;
6256        }        }
6257    
6258      /* If the character following a repeat is '+', or if certain optimization      /* If the character following a repeat is '+', possessive_quantifier is
6259      tests above succeeded, possessive_quantifier is TRUE. For some opcodes,      TRUE. For some opcodes, there are special alternative opcodes for this
6260      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
6261      wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+'      brackets. Logically, the '+' notation is just syntactic sugar, taken from
6262      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.  
6263    
6264      Some (but not all) possessively repeated subpatterns have already been      Some (but not all) possessively repeated subpatterns have already been
6265      completely handled in the code just above. For them, possessive_quantifier      completely handled in the code just above. For them, possessive_quantifier
6266      is always FALSE at this stage.      is always FALSE at this stage. Note that the repeated item starts at
6267        tempcode, not at previous, which might be the first part of a string whose
6268      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. */  
6269    
6270      if (possessive_quantifier)      if (possessive_quantifier)
6271        {        {
6272        int len;        int len;
6273    
6274        if (*tempcode == OP_TYPEEXACT)        /* Possessifying an EXACT quantifier has no effect, so we can ignore it.
6275          However, QUERY, STAR, or UPTO may follow (for quantifiers such as {5,6},
6276          {5,}, or {5,10}). We skip over an EXACT item; if the length of what
6277          remains is greater than zero, there's a further opcode that can be
6278          handled. If not, do nothing, leaving the EXACT alone. */
6279    
6280          switch(*tempcode)
6281            {
6282            case OP_TYPEEXACT:
6283          tempcode += PRIV(OP_lengths)[*tempcode] +          tempcode += PRIV(OP_lengths)[*tempcode] +
6284            ((tempcode[1 + IMM2_SIZE] == OP_PROP            ((tempcode[1 + IMM2_SIZE] == OP_PROP
6285            || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);            || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);
6286            break;
6287    
6288        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)          /* CHAR opcodes are used for exacts whose count is 1. */
6289          {  
6290            case OP_CHAR:
6291            case OP_CHARI:
6292            case OP_NOT:
6293            case OP_NOTI:
6294            case OP_EXACT:
6295            case OP_EXACTI:
6296            case OP_NOTEXACT:
6297            case OP_NOTEXACTI:
6298          tempcode += PRIV(OP_lengths)[*tempcode];          tempcode += PRIV(OP_lengths)[*tempcode];
6299  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
6300          if (utf && HAS_EXTRALEN(tempcode[-1]))          if (utf && HAS_EXTRALEN(tempcode[-1]))
6301            tempcode += GET_EXTRALEN(tempcode[-1]);            tempcode += GET_EXTRALEN(tempcode[-1]);
6302  #endif  #endif
6303            break;
6304    
6305            /* For the class opcodes, the repeat operator appears at the end;
6306            adjust tempcode to point to it. */
6307    
6308            case OP_CLASS:
6309            case OP_NCLASS:
6310            tempcode += 1 + 32/sizeof(pcre_uchar);
6311            break;
6312    
6313    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
6314            case OP_XCLASS:
6315            tempcode += GET(tempcode, 1);
6316            break;
6317    #endif
6318          }          }
6319    
6320          /* If tempcode is equal to code (which points to the end of the repeated
6321          item), it means we have skipped an EXACT item but there is no following
6322          QUERY, STAR, or UPTO; the value of len will be 0, and we do nothing. In
6323          all other cases, tempcode will be pointing to the repeat opcode, and will
6324          be less than code, so the value of len will be greater than 0. */
6325    
6326        len = (int)(code - tempcode);        len = (int)(code - tempcode);
6327          if (len > 0)
6328            {
6329            unsigned int repcode = *tempcode;
6330    
6331            /* There is a table for possessifying opcodes, all of which are less
6332            than OP_CALLOUT. A zero entry means there is no possessified version.
6333            */
6334    
6335            if (repcode < OP_CALLOUT && opcode_possessify[repcode] > 0)
6336              *tempcode = opcode_possessify[repcode];
6337    
6338            /* For opcode without a special possessified version, wrap the item in
6339            ONCE brackets. Because we are moving code along, we must ensure that any
6340            pending recursive references are updated. */
6341    
6342            else
6343              {
6344              *code = OP_END;
6345              adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm);
6346              memmove(tempcode + 1 + LINK_SIZE, tempcode, IN_UCHARS(len));
6347              code += 1 + LINK_SIZE;
6348              len += 1 + LINK_SIZE;
6349              tempcode[0] = OP_ONCE;
6350              *code++ = OP_KET;
6351              PUTINC(code, 0, len);
6352              PUT(tempcode, 1, len);
6353              }
6354            }
6355    
6356    #ifdef NEVER
6357        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
6358          {          {
6359          case OP_STAR:  *tempcode = OP_POSSTAR; break;          case OP_STAR:  *tempcode = OP_POSSTAR; break;
# Line 5794  for (;; ptr++) Line 6381  for (;; ptr++)
6381          case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break;          case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break;
6382          case OP_TYPEUPTO:  *tempcode = OP_TYPEPOSUPTO; break;          case OP_TYPEUPTO:  *tempcode = OP_TYPEPOSUPTO; break;
6383    
6384            case OP_CRSTAR:   *tempcode = OP_CRPOSSTAR; break;
6385            case OP_CRPLUS:   *tempcode = OP_CRPOSPLUS; break;
6386            case OP_CRQUERY:  *tempcode = OP_CRPOSQUERY; break;
6387            case OP_CRRANGE:  *tempcode = OP_CRPOSRANGE; break;
6388    
6389          /* Because we are moving code along, we must ensure that any          /* Because we are moving code along, we must ensure that any
6390          pending recursive references are updated. */          pending recursive references are updated. */
6391    
# Line 5809  for (;; ptr++) Line 6401  for (;; ptr++)
6401          PUT(tempcode, 1, len);          PUT(tempcode, 1, len);
6402          break;          break;
6403          }          }
6404    #endif
6405        }        }
6406    
6407      /* 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 6590  for (;; ptr++)
6590          tempptr = ptr;          tempptr = ptr;
6591    
6592          /* A condition can be an assertion, a number (referring to a numbered          /* A condition can be an assertion, a number (referring to a numbered
6593          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',
6594          recursion. R<digits> and R&name are also permitted for recursion tests.          referring to recursion. R<digits> and R&name are also permitted for
6595            recursion tests.
6596          There are several syntaxes for testing a named group: (?(name)) is used  
6597          by Python; Perl 5.10 onwards uses (?(<name>) or (?('name')).          There are ways of testing a named group: (?(name)) is used by Python;
6598            Perl 5.10 onwards uses (?(<name>) or (?('name')).
6599          There are two unfortunate ambiguities, caused by history. (a) 'R' can  
6600          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
6601          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
6602          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.  
6603    
6604          For compatibility with auto-callouts, we allow a callout to be          For compatibility with auto-callouts, we allow a callout to be
6605          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 6623  for (;; ptr++)
6623                 tempptr[2] == CHAR_LESS_THAN_SIGN))                 tempptr[2] == CHAR_LESS_THAN_SIGN))
6624            break;            break;
6625    
6626          /* 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
6627          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. */
6628    
6629          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
6630          skipbytes = 1+IMM2_SIZE;          skipbytes = 1+IMM2_SIZE;
6631          refsign = -1;          refsign = -1;     /* => not a number */
6632            namelen = -1;     /* => not a name; must set to avoid warning */
6633            name = NULL;      /* Always set to avoid warning */
6634            recno = 0;        /* Always set to avoid warning */
6635    
6636          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
6637    
6638          if (ptr[1] == CHAR_R && ptr[2] == CHAR_AMPERSAND)          ptr++;
6639            if (*ptr == CHAR_R && ptr[1] == CHAR_AMPERSAND)
6640            {            {
6641            terminator = -1;            terminator = -1;
6642            ptr += 2;            ptr += 2;
# Line 6048  for (;; ptr++) Line 6644  for (;; ptr++)
6644            }            }
6645    
6646          /* 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
6647          syntax (?(<name>) or (?('name') */          syntax (?(<name>) or (?('name'), and also allow for the original PCRE
6648            syntax of (?(name) or for (?(+n), (?(-n), and just (?(n). */
6649    
6650          else if (ptr[1] == CHAR_LESS_THAN_SIGN)          else if (*ptr == CHAR_LESS_THAN_SIGN)
6651            {            {
6652            terminator = CHAR_GREATER_THAN_SIGN;            terminator = CHAR_GREATER_THAN_SIGN;
6653            ptr++;            ptr++;
6654            }            }
6655          else if (ptr[1] == CHAR_APOSTROPHE)          else if (*ptr == CHAR_APOSTROPHE)
6656            {            {
6657            terminator = CHAR_APOSTROPHE;            terminator = CHAR_APOSTROPHE;
6658            ptr++;            ptr++;
# Line 6063  for (;; ptr++) Line 6660  for (;; ptr++)
6660          else          else
6661            {            {
6662            terminator = CHAR_NULL;            terminator = CHAR_NULL;
6663            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);            if (*ptr == CHAR_MINUS || *ptr == CHAR_PLUS) refsign = *ptr++;
6664                else if (IS_DIGIT(*ptr)) refsign = 0;
6665            }            }
6666    
6667          /* We now expect to read a name; any thing else is an error */          /* Handle a number */
6668    
6669          if (!MAX_255(ptr[1]) || (cd->ctypes[ptr[1]] & ctype_word) == 0)          if (refsign >= 0)
6670            {            {
6671            ptr += 1;  /* To get the right offset */            while (IS_DIGIT(*ptr))
6672            *errorcodeptr = ERR28;              {
6673            goto FAILED;              recno = recno * 10 + (int)(*ptr - CHAR_0);
6674                ptr++;
6675                }
6676            }            }
6677    
6678          /* 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
6679            a name is one of a number of duplicates, a different opcode is used and
6680            it needs more memory. Unfortunately we cannot tell whether a name is a
6681            duplicate in the first pass, so we have to allow for more memory. */
6682    
6683          recno = 0;          else
         name = ++ptr;  
         while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)  
6684            {            {
6685            if (recno >= 0)            if (IS_DIGIT(*ptr))
6686              recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1;              {
6687            ptr++;              *errorcodeptr = ERR84;
6688                goto FAILED;
6689                }
6690              if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_word) == 0)
6691                {
6692                *errorcodeptr = ERR28;   /* Assertion expected */
6693                goto FAILED;
6694                }
6695              name = ptr++;
6696              while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
6697                {
6698                ptr++;
6699                }
6700              namelen = (int)(ptr - name);
6701              if (lengthptr != NULL && (options & PCRE_DUPNAMES) != 0)
6702                *lengthptr += IMM2_SIZE;
6703            }            }
6704          namelen = (int)(ptr - name);  
6705            /* Check the terminator */
6706    
6707          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||          if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) ||
6708              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
6709            {            {
6710            ptr--;      /* Error offset */            ptr--;                  /* Error offset */
6711            *errorcodeptr = ERR26;            *errorcodeptr = ERR26;  /* Malformed number or name */
6712            goto FAILED;            goto FAILED;
6713            }            }
6714    
# Line 6100  for (;; ptr++) Line 6717  for (;; ptr++)
6717          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
6718    
6719          /* 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
6720          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
6721          be digits, in which case recno will be set. */          recno. */
6722    
6723          if (refsign > 0)          if (refsign >= 0)
6724            {            {
6725            if (recno <= 0)            if (recno <= 0)
6726              {              {
6727              *errorcodeptr = ERR58;              *errorcodeptr = ERR35;
6728              goto FAILED;              goto FAILED;
6729              }              }
6730            recno = (refsign == CHAR_MINUS)?            if (refsign != 0) recno = (refsign == CHAR_MINUS)?
6731              cd->bracount - recno + 1 : recno +cd->bracount;              cd->bracount - recno + 1 : recno + cd->bracount;
6732            if (recno <= 0 || recno > cd->final_bracount)            if (recno <= 0 || recno > cd->final_bracount)
6733              {              {
6734              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
# Line 6121  for (;; ptr++) Line 6738  for (;; ptr++)
6738            break;            break;
6739            }            }
6740    
6741          /* 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. */  
6742    
6743          slot = cd->name_table;          slot = cd->name_table;
6744          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 6134  for (;; ptr++) Line 6747  for (;; ptr++)
6747            slot += cd->name_entry_size;            slot += cd->name_entry_size;
6748            }            }
6749    
6750          /* Found the named subpattern */          /* Found the named subpattern. If the name is duplicated, add one to
6751            the opcode to change CREF/RREF into DNCREF/DNRREF and insert
6752            appropriate data values. Otherwise, just insert the unique subpattern
6753            number. */
6754    
6755          if (i < cd->names_found)          if (i < cd->names_found)
6756            {            {
6757            recno = GET2(slot, 0);            int offset = i++;
6758            PUT2(code, 2+LINK_SIZE, recno);            int count = 1;
6759            code[1+LINK_SIZE]++;            recno = GET2(slot, 0);   /* Number from first found */
6760              for (; i < cd->names_found; i++)
6761                {
6762                slot += cd->name_entry_size;
6763                if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) != 0 ||
6764                  (slot+IMM2_SIZE)[namelen] != 0) break;
6765                count++;
6766                }
6767    
6768              if (count > 1)
6769                {
6770                PUT2(code, 2+LINK_SIZE, offset);
6771                PUT2(code, 2+LINK_SIZE+IMM2_SIZE, count);
6772                skipbytes += IMM2_SIZE;
6773                code[1+LINK_SIZE]++;
6774                }
6775              else  /* Not a duplicated name */
6776                {
6777                PUT2(code, 2+LINK_SIZE, recno);
6778                }
6779            }            }
6780    
6781          /* If terminator == CHAR_NULL it means that the name followed directly          /* If terminator == CHAR_NULL it means that the name followed directly
6782          after the opening parenthesis [e.g. (?(abc)...] and in this case there          after the opening parenthesis [e.g. (?(abc)...] and in this case there
6783          are some further alternatives to try. For the cases where terminator !=          are some further alternatives to try. For the cases where terminator !=
6784          0 [things like (?(<name>... or (?('name')... or (?(R&name)... ] we have          CHAR_NULL [things like (?(<name>... or (?('name')... or (?(R&name)... ]
6785          now checked all the possibilities, so give an error. */          we have now checked all the possibilities, so give an error. */
6786    
6787          else if (terminator != CHAR_NULL)          else if (terminator != CHAR_NULL)
6788            {            {
# Line 6184  for (;; ptr++) Line 6819  for (;; ptr++)
6819            skipbytes = 1;            skipbytes = 1;
6820            }            }
6821    
6822          /* 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) */  
6823    
6824          else          else
6825            {            {
6826            *errorcodeptr = (recno == 0)? ERR35: ERR15;            *errorcodeptr = ERR15;
6827            goto FAILED;            goto FAILED;
6828            }            }
6829          break;          break;
# Line 6209  for (;; ptr++) Line 6836  for (;; ptr++)
6836          ptr++;          ptr++;
6837          break;          break;
6838    
6839            /* Optimize (?!) to (*FAIL) unless it is quantified - which is a weird
6840            thing to do, but Perl allows all assertions to be quantified, and when
6841            they contain capturing parentheses there may be a potential use for
6842            this feature. Not that that applies to a quantified (?!) but we allow
6843            it for uniformity. */
6844    
6845          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
6846          case CHAR_EXCLAMATION_MARK:            /* Negative lookahead */          case CHAR_EXCLAMATION_MARK:            /* Negative lookahead */
6847          ptr++;          ptr++;
6848          if (*ptr == CHAR_RIGHT_PARENTHESIS)    /* Optimize (?!) */          if (*ptr == CHAR_RIGHT_PARENTHESIS && ptr[1] != CHAR_ASTERISK &&
6849                 ptr[1] != CHAR_PLUS && ptr[1] != CHAR_QUESTION_MARK &&
6850                (ptr[1] != CHAR_LEFT_CURLY_BRACKET || !is_counted_repeat(ptr+2)))
6851            {            {
6852            *code++ = OP_FAIL;            *code++ = OP_FAIL;
6853            previous = NULL;            previous = NULL;
# Line 6309  for (;; ptr++) Line 6943  for (;; ptr++)
6943          terminator = (*ptr == CHAR_LESS_THAN_SIGN)?          terminator = (*ptr == CHAR_LESS_THAN_SIGN)?
6944            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
6945          name = ++ptr;          name = ++ptr;
6946            if (IS_DIGIT(*ptr))
6947              {
6948              *errorcodeptr = ERR84;   /* Group name must start with non-digit */
6949              goto FAILED;
6950              }
6951          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
6952          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
6953    
# Line 6423  for (;; ptr++) Line 7061  for (;; ptr++)
7061    
7062          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
7063          name = ++ptr;          name = ++ptr;
7064            if (IS_DIGIT(*ptr))
7065              {
7066              *errorcodeptr = ERR84;   /* Group name must start with non-digit */
7067              goto FAILED;
7068              }
7069          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
7070          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
7071    
# Line 6470  for (;; ptr++) Line 7113  for (;; ptr++)
7113            /* Count named back references. */            /* Count named back references. */
7114    
7115            if (!is_recurse) cd->namedrefcount++;            if (!is_recurse) cd->namedrefcount++;
7116    
7117              /* If duplicate names are permitted, we have to allow for a named
7118              reference to a duplicated name (this cannot be determined until the
7119              second pass). This needs an extra 16-bit data item. */
7120    
7121              if ((options & PCRE_DUPNAMES) != 0) *lengthptr += IMM2_SIZE;
7122            }            }
7123    
7124          /* 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 7165  for (;; ptr++)
7165            for (i++; i < cd->names_found; i++)            for (i++; i < cd->names_found; i++)
7166              {              {
7167              if (STRCMP_UC_UC(slot + IMM2_SIZE, cslot + IMM2_SIZE) != 0) break;              if (STRCMP_UC_UC(slot + IMM2_SIZE, cslot + IMM2_SIZE) != 0) break;
7168    
7169    
7170              count++;              count++;
7171              cslot += cd->name_entry_size;              cslot += cd->name_entry_size;
7172              }              }
# Line 6815  for (;; ptr++) Line 7466  for (;; ptr++)
7466        skipbytes = IMM2_SIZE;        skipbytes = IMM2_SIZE;
7467        }        }
7468    
7469      /* Process nested bracketed regex. Assertions used not to be repeatable,      /* Process nested bracketed regex. First check for parentheses nested too
7470      but this was changed for Perl compatibility, so all kinds can now be      deeply. */
7471      repeated. We copy code into a non-register variable (tempcode) in order to  
7472      be able to pass its address because some compilers complain otherwise. */      if ((cd->parens_depth += 1) > PARENS_NEST_LIMIT)
7473          {
7474          *errorcodeptr = ERR82;
7475          goto FAILED;
7476          }
7477    
7478        /* Assertions used not to be repeatable, but this was changed for Perl
7479        compatibility, so all kinds can now be repeated. We copy code into a
7480        non-register variable (tempcode) in order to be able to pass its address
7481        because some compilers complain otherwise. */
7482    
7483      previous = code;                      /* For handling repetition */      previous = code;                      /* For handling repetition */
7484      *code = bravalue;      *code = bravalue;
# Line 6849  for (;; ptr++) Line 7509  for (;; ptr++)
7509           ))           ))
7510        goto FAILED;        goto FAILED;
7511    
7512        cd->parens_depth -= 1;
7513    
7514      /* 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,
7515      generate OP_ONCE_NC instead of OP_ONCE. */      generate OP_ONCE_NC instead of OP_ONCE. */
7516    
# Line 7063  for (;; ptr++) Line 7725  for (;; ptr++)
7725        if (escape == ESC_g)        if (escape == ESC_g)
7726          {          {
7727          const pcre_uchar *p;          const pcre_uchar *p;
7728            pcre_uint32 cf;
7729    
7730          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
7731          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
7732            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
7733    
7734          /* These two statements stop the compiler for warning about possibly          /* These two statements stop the compiler for warning about possibly
7735          unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In          unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In
7736          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
7737          would actually be in error are never taken. */          would actually be in error are never taken. */
7738    
7739          skipbytes = 0;          skipbytes = 0;
7740          reset_bracount = FALSE;          reset_bracount = FALSE;
7741    
7742          /* Test for a name */          /* If it's not a signed or unsigned number, treat it as a name. */
7743    
7744          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          cf = ptr[1];
7745            if (cf != CHAR_PLUS && cf != CHAR_MINUS && !IS_DIGIT(cf))
7746            {            {
           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;  
             }  
7747            is_recurse = TRUE;            is_recurse = TRUE;
7748            goto NAMED_REF_OR_RECURSE;            goto NAMED_REF_OR_RECURSE;
7749            }            }
7750    
7751          /* Test a signed number in angle brackets or quotes. */          /* Signed or unsigned number (cf = ptr[1]) is known to be plus or minus
7752            or a digit. */
7753    
7754          p = ptr + 2;          p = ptr + 2;
7755          while (IS_DIGIT(*p)) p++;          while (IS_DIGIT(*p)) p++;
# Line 7240  for (;; ptr++) Line 7889  for (;; ptr++)
7889    
7890      /* ===================================================================*/      /* ===================================================================*/
7891      /* Handle a literal character. It is guaranteed not to be whitespace or #      /* Handle a literal character. It is guaranteed not to be whitespace or #
7892      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
7893      multi-byte literal character. */      multi-unit literal character. */
7894    
7895      default:      default:
7896      NORMAL_CHAR:      NORMAL_CHAR:
# Line 7406  unsigned int orig_bracount; Line 8055  unsigned int orig_bracount;
8055  unsigned int max_bracount;  unsigned int max_bracount;
8056  branch_chain bc;  branch_chain bc;
8057    
8058    /* If set, call the external function that checks for stack availability. */
8059    
8060    if (PUBL(stack_guard) != NULL && PUBL(stack_guard)())
8061      {
8062      *errorcodeptr= ERR85;
8063      return FALSE;
8064      }
8065    
8066    /* Miscellaneous initialization */
8067    
8068  bc.outer = bcptr;  bc.outer = bcptr;
8069  bc.current_branch = code;  bc.current_branch = code;
8070    
# Line 7605  for (;;) Line 8264  for (;;)
8264    
8265      /* 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
8266      recursive back references. If so, we must wrap it in atomic brackets.      recursive back references. If so, we must wrap it in atomic brackets.
8267      In any event, remove the block from the chain. */      Because we are moving code along, we must ensure that any pending recursive
8268        references are updated. In any event, remove the block from the chain. */
8269    
8270      if (capnumber > 0)      if (capnumber > 0)
8271        {        {
8272        if (cd->open_caps->flag)        if (cd->open_caps->flag)
8273          {          {
8274            *code = OP_END;
8275            adjust_recurse(start_bracket, 1 + LINK_SIZE,
8276              (options & PCRE_UTF8) != 0, cd, cd->hwm);
8277          memmove(start_bracket + 1 + LINK_SIZE, start_bracket,          memmove(start_bracket + 1 + LINK_SIZE, start_bracket,
8278            IN_UCHARS(code - start_bracket));            IN_UCHARS(code - start_bracket));
8279          *start_bracket = OP_ONCE;          *start_bracket = OP_ONCE;
# Line 7830  do { Line 8493  do {
8493       switch (*scode)       switch (*scode)
8494         {         {
8495         case OP_CREF:         case OP_CREF:
8496         case OP_NCREF:         case OP_DNCREF:
8497         case OP_RREF:         case OP_RREF:
8498         case OP_NRREF:         case OP_DNRREF:
8499         case OP_DEF:         case OP_DEF:
8500         return FALSE;         return FALSE;
8501    
# Line 8229  PCRE_UTF8 == PCRE_UTF16 == PCRE_UTF32. * Line 8892  PCRE_UTF8 == PCRE_UTF16 == PCRE_UTF32. *
8892      { skipatstart += 6; options |= PCRE_UTF8; continue; }      { skipatstart += 6; options |= PCRE_UTF8; continue; }
8893    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)
8894      { skipatstart += 6; options |= PCRE_UCP; continue; }      { skipatstart += 6; options |= PCRE_UCP; continue; }
8895      else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_AUTO_POSSESS_RIGHTPAR, 16) == 0)
8896        { skipatstart += 18; options |= PCRE_NO_AUTO_POSSESS; continue; }
8897    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)
8898      { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; }      { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; }
8899    
# Line 8423  cd->named_group_list_size = NAMED_GROUP_ Line 9088  cd->named_group_list_size = NAMED_GROUP_
9088  cd->start_pattern = (const pcre_uchar *)pattern;  cd->start_pattern = (const pcre_uchar *)pattern;
9089  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));
9090  cd->req_varyopt = 0;  cd->req_varyopt = 0;
9091    cd->parens_depth = 0;
9092  cd->assert_depth = 0;  cd->assert_depth = 0;
9093  cd->max_lookbehind = 0;  cd->max_lookbehind = 0;
9094  cd->external_options = options;  cd->external_options = options;
# Line 8437  outside can help speed up starting point Line 9103  outside can help speed up starting point
9103  ptr += skipatstart;  ptr += skipatstart;
9104  code = cworkspace;  code = cworkspace;
9105  *code = OP_BRA;  *code = OP_BRA;
9106    
9107  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
9108    FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL,    FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL,
9109    cd, &length);    cd, &length);
# Line 8508  field; this time it's used for rememberi Line 9175  field; this time it's used for rememberi
9175  */  */
9176    
9177  cd->final_bracount = cd->bracount;  /* Save for checking forward references */  cd->final_bracount = cd->bracount;  /* Save for checking forward references */
9178    cd->parens_depth = 0;
9179  cd->assert_depth = 0;  cd->assert_depth = 0;
9180  cd->bracount = 0;  cd->bracount = 0;
9181  cd->max_lookbehind = 0;  cd->max_lookbehind = 0;
# Line 8610  subpattern. */ Line 9278  subpattern. */
9278    
9279  if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15;  if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15;
9280    
9281  /* Unless disabled, check whether single character iterators can be  /* Unless disabled, check whether any single character iterators can be
9282  auto-possessified. The function overwrites the appropriate opcode values. */  auto-possessified. The function overwrites the appropriate opcode values, so
9283    the type of the pointer must be cast. NOTE: the intermediate variable "temp" is
9284    used in this code because at least one compiler gives a warning about loss of
9285    "const" attribute if the cast (pcre_uchar *)codestart is used directly in the
9286    function call. */
9287    
9288  if ((options & PCRE_NO_AUTO_POSSESSIFY) == 0)  if ((options & PCRE_NO_AUTO_POSSESS) == 0)
9289    auto_possessify((pcre_uchar *)codestart, utf, cd);    {
9290      pcre_uchar *temp = (pcre_uchar *)codestart;
9291      auto_possessify(temp, utf, cd);
9292      }
9293    
9294  /* If there were any lookbehind assertions that contained OP_RECURSE  /* If there were any lookbehind assertions that contained OP_RECURSE
9295  (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 9512  return (pcre32 *)re;
9512  }  }
9513    
9514  /* End of pcre_compile.c */  /* End of pcre_compile.c */
9515    

Legend:
Removed from v.1363  
changed lines
  Added in v.1502

  ViewVC Help
Powered by ViewVC 1.1.5