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

Diff of /code/trunk/pcre_dfa_exec.c

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

revision 341 by ph10, Sat Apr 19 16:41:04 2008 UTC revision 442 by ph10, Fri Sep 11 10:21:02 2009 UTC
# Line 3  Line 3 
3  *************************************************/  *************************************************/
4    
5  /* PCRE is a library of functions to support regular expressions whose syntax  /* PCRE is a library of functions to support regular expressions whose syntax
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 (but see
7    below for why this module is different).
8    
9                         Written by Philip Hazel                         Written by Philip Hazel
10             Copyright (c) 1997-2008 University of Cambridge             Copyright (c) 1997-2009 University of Cambridge
11    
12  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
13  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 44  FSM). This is NOT Perl- compatible, but Line 45  FSM). This is NOT Perl- compatible, but
45  applications. */  applications. */
46    
47    
48    /* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved
49    the performance of his patterns greatly. I could not use it as it stood, as it
50    was not thread safe, and made assumptions about pattern sizes. Also, it caused
51    test 7 to loop, and test 9 to crash with a segfault.
52    
53    The issue is the check for duplicate states, which is done by a simple linear
54    search up the state list. (Grep for "duplicate" below to find the code.) For
55    many patterns, there will never be many states active at one time, so a simple
56    linear search is fine. In patterns that have many active states, it might be a
57    bottleneck. The suggested code used an indexing scheme to remember which states
58    had previously been used for each character, and avoided the linear search when
59    it knew there was no chance of a duplicate. This was implemented when adding
60    states to the state lists.
61    
62    I wrote some thread-safe, not-limited code to try something similar at the time
63    of checking for duplicates (instead of when adding states), using index vectors
64    on the stack. It did give a 13% improvement with one specially constructed
65    pattern for certain subject strings, but on other strings and on many of the
66    simpler patterns in the test suite it did worse. The major problem, I think,
67    was the extra time to initialize the index. This had to be done for each call
68    of internal_dfa_exec(). (The supplied patch used a static vector, initialized
69    only once - I suspect this was the cause of the problems with the tests.)
70    
71    Overall, I concluded that the gains in some cases did not outweigh the losses
72    in others, so I abandoned this code. */
73    
74    
75    
76  #ifdef HAVE_CONFIG_H  #ifdef HAVE_CONFIG_H
77  #include "config.h"  #include "config.h"
78  #endif  #endif
# Line 60  applications. */ Line 89  applications. */
89  #define SP "                   "  #define SP "                   "
90    
91    
   
92  /*************************************************  /*************************************************
93  *      Code parameters and static tables         *  *      Code parameters and static tables         *
94  *************************************************/  *************************************************/
# Line 223  Arguments: Line 251  Arguments:
251    rlevel            function call recursion level    rlevel            function call recursion level
252    recursing         regex recursive call level    recursing         regex recursive call level
253    
254  Returns:            > 0 => number of match offset pairs placed in offsets  Returns:            > 0 => number of match offset pairs placed in offsets
255                      = 0 => offsets overflowed; longest matches are present                      = 0 => offsets overflowed; longest matches are present
256                       -1 => failed to match                       -1 => failed to match
257                     < -1 => some kind of unexpected problem                     < -1 => some kind of unexpected problem
# Line 389  if (*first_op == OP_REVERSE) Line 417  if (*first_op == OP_REVERSE)
417        current_subject - start_subject : max_back;        current_subject - start_subject : max_back;
418      current_subject -= gone_back;      current_subject -= gone_back;
419      }      }
420    
421      /* Save the earliest consulted character */
422    
423      if (current_subject < md->start_used_ptr)
424        md->start_used_ptr = current_subject;
425    
426    /* Now we can process the individual branches. */    /* Now we can process the individual branches. */
427    
# Line 454  for (;;) Line 487  for (;;)
487    int i, j;    int i, j;
488    int clen, dlen;    int clen, dlen;
489    unsigned int c, d;    unsigned int c, d;
490      int forced_fail = 0;
491      int reached_end = 0;
492    
493    /* Make the new state list into the active state list and empty the    /* Make the new state list into the active state list and empty the
494    new state list. */    new state list. */
# Line 511  for (;;) Line 546  for (;;)
546      stateblock *current_state = active_states + i;      stateblock *current_state = active_states + i;
547      const uschar *code;      const uschar *code;
548      int state_offset = current_state->offset;      int state_offset = current_state->offset;
549      int count, codevalue;      int count, codevalue, rrc;
 #ifdef SUPPORT_UCP  
     int chartype, script;  
 #endif  
550    
551  #ifdef DEBUG  #ifdef DEBUG
552      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);
# Line 546  for (;;) Line 578  for (;;)
578          }          }
579        }        }
580    
581      /* Check for a duplicate state with the same count, and skip if found. */      /* Check for a duplicate state with the same count, and skip if found.
582        See the note at the head of this module about the possibility of improving
583        performance here. */
584    
585      for (j = 0; j < i; j++)      for (j = 0; j < i; j++)
586        {        {
# Line 613  for (;;) Line 647  for (;;)
647  /* ========================================================================== */  /* ========================================================================== */
648        /* Reached a closing bracket. If not at the end of the pattern, carry        /* Reached a closing bracket. If not at the end of the pattern, carry
649        on with the next opcode. Otherwise, unless we have an empty string and        on with the next opcode. Otherwise, unless we have an empty string and
650        PCRE_NOTEMPTY is set, save the match data, shifting up all previous        PCRE_NOTEMPTY is set, or PCRE_NOTEMPTY_ATSTART is set and we are at the
651          start of the subject, save the match data, shifting up all previous
652        matches so we always have the longest first. */        matches so we always have the longest first. */
653    
654        case OP_KET:        case OP_KET:
# Line 627  for (;;) Line 662  for (;;)
662            ADD_ACTIVE(state_offset - GET(code, 1), 0);            ADD_ACTIVE(state_offset - GET(code, 1), 0);
663            }            }
664          }          }
665        else if (ptr > current_subject || (md->moptions & PCRE_NOTEMPTY) == 0)        else
666          {          {
667          if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;          reached_end++;    /* Count branches that reach the end */
668            else if (match_count > 0 && ++match_count * 2 >= offsetcount)          if (ptr > current_subject ||
669              match_count = 0;              ((md->moptions & PCRE_NOTEMPTY) == 0 &&
670          count = ((match_count == 0)? offsetcount : match_count * 2) - 2;                ((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 ||
671          if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));                  current_subject > start_subject + md->start_offset)))
672          if (offsetcount >= 2)            {
673            {            if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
674            offsets[0] = current_subject - start_subject;              else if (match_count > 0 && ++match_count * 2 >= offsetcount)
675            offsets[1] = ptr - start_subject;                match_count = 0;
676            DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,            count = ((match_count == 0)? offsetcount : match_count * 2) - 2;
677              offsets[1] - offsets[0], current_subject));            if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));
678            }            if (offsetcount >= 2)
679          if ((md->moptions & PCRE_DFA_SHORTEST) != 0)              {
680            {              offsets[0] = current_subject - start_subject;
681            DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"              offsets[1] = ptr - start_subject;
682              "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,              DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,
683              match_count, rlevel*2-2, SP));                offsets[1] - offsets[0], current_subject));
684            return match_count;              }
685            }            if ((md->moptions & PCRE_DFA_SHORTEST) != 0)
686                {
687                DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"
688                  "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,
689                  match_count, rlevel*2-2, SP));
690                return match_count;
691                }
692              }
693          }          }
694        break;        break;
695    
# Line 739  for (;;) Line 781  for (;;)
781    
782        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
783        case OP_ANY:        case OP_ANY:
784        if (clen > 0 && ((ims & PCRE_DOTALL) != 0 || !IS_NEWLINE(ptr)))        if (clen > 0 && !IS_NEWLINE(ptr))
785          { ADD_NEW(state_offset + 1, 0); }          { ADD_NEW(state_offset + 1, 0); }
786        break;        break;
787    
# Line 760  for (;;) Line 802  for (;;)
802        if ((md->moptions & PCRE_NOTEOL) == 0)        if ((md->moptions & PCRE_NOTEOL) == 0)
803          {          {
804          if (clen == 0 ||          if (clen == 0 ||
805              (IS_NEWLINE(ptr) &&              ((md->poptions & PCRE_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr) &&
806                 ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen)                 ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen)
807              ))              ))
808            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 797  for (;;) Line 839  for (;;)
839          if (ptr > start_subject)          if (ptr > start_subject)
840            {            {
841            const uschar *temp = ptr - 1;            const uschar *temp = ptr - 1;
842              if (temp < md->start_used_ptr) md->start_used_ptr = temp;
843  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
844            if (utf8) BACKCHAR(temp);            if (utf8) BACKCHAR(temp);
845  #endif  #endif
# Line 805  for (;;) Line 848  for (;;)
848            }            }
849          else left_word = 0;          else left_word = 0;
850    
851          if (clen > 0) right_word = c < 256 && (ctypes[c] & ctype_word) != 0;          if (clen > 0)
852            else right_word = 0;            right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
853            else              /* This is a fudge to ensure that if this is the */
854              {               /* last item in the pattern, we don't count it as */
855              reached_end--;  /* reached, thus disabling a partial match. */
856              right_word = 0;
857              }
858    
859          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))
860            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 825  for (;;) Line 873  for (;;)
873        if (clen > 0)        if (clen > 0)
874          {          {
875          BOOL OK;          BOOL OK;
876          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
877          switch(code[1])          switch(code[1])
878            {            {
879            case PT_ANY:            case PT_ANY:
# Line 833  for (;;) Line 881  for (;;)
881            break;            break;
882    
883            case PT_LAMP:            case PT_LAMP:
884            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
885            break;            break;
886    
887            case PT_GC:            case PT_GC:
888            OK = category == code[2];            OK = _pcre_ucp_gentype[prop->chartype] == code[2];
889            break;            break;
890    
891            case PT_PC:            case PT_PC:
892            OK = chartype == code[2];            OK = prop->chartype == code[2];
893            break;            break;
894    
895            case PT_SC:            case PT_SC:
896            OK = script == code[2];            OK = prop->script == code[2];
897            break;            break;
898    
899            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 877  for (;;) Line 925  for (;;)
925          {          {
926          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
927              (c < 256 &&              (c < 256 &&
928                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
929                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
930            {            {
931            if (count > 0 && codevalue == OP_TYPEPOSPLUS)            if (count > 0 && codevalue == OP_TYPEPOSPLUS)
# Line 903  for (;;) Line 948  for (;;)
948          {          {
949          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
950              (c < 256 &&              (c < 256 &&
951                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
952                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
953            {            {
954            if (codevalue == OP_TYPEPOSQUERY)            if (codevalue == OP_TYPEPOSQUERY)
# Line 928  for (;;) Line 970  for (;;)
970          {          {
971          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
972              (c < 256 &&              (c < 256 &&
973                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
974                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
975            {            {
976            if (codevalue == OP_TYPEPOSSTAR)            if (codevalue == OP_TYPEPOSSTAR)
# Line 951  for (;;) Line 990  for (;;)
990          {          {
991          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
992              (c < 256 &&              (c < 256 &&
993                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
994                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
995            {            {
996            if (++count >= GET2(code, 1))            if (++count >= GET2(code, 1))
# Line 975  for (;;) Line 1011  for (;;)
1011          {          {
1012          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1013              (c < 256 &&              (c < 256 &&
1014                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1015                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1016            {            {
1017            if (codevalue == OP_TYPEPOSUPTO)            if (codevalue == OP_TYPEPOSUPTO)
# Line 1009  for (;;) Line 1042  for (;;)
1042        if (clen > 0)        if (clen > 0)
1043          {          {
1044          BOOL OK;          BOOL OK;
1045          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1046          switch(code[2])          switch(code[2])
1047            {            {
1048            case PT_ANY:            case PT_ANY:
# Line 1017  for (;;) Line 1050  for (;;)
1050            break;            break;
1051    
1052            case PT_LAMP:            case PT_LAMP:
1053            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1054            break;            break;
1055    
1056            case PT_GC:            case PT_GC:
1057            OK = category == code[3];            OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1058            break;            break;
1059    
1060            case PT_PC:            case PT_PC:
1061            OK = chartype == code[3];            OK = prop->chartype == code[3];
1062            break;            break;
1063    
1064            case PT_SC:            case PT_SC:
1065            OK = script == code[3];            OK = prop->script == code[3];
1066            break;            break;
1067    
1068            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1058  for (;;) Line 1091  for (;;)
1091        case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:        case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:
1092        count = current_state->count;  /* Already matched */        count = current_state->count;  /* Already matched */
1093        if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }        if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1094        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1095          {          {
1096          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1097          int ncount = 0;          int ncount = 0;
# Line 1072  for (;;) Line 1105  for (;;)
1105            int nd;            int nd;
1106            int ndlen = 1;            int ndlen = 1;
1107            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1108            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1109            ncount++;            ncount++;
1110            nptr += ndlen;            nptr += ndlen;
1111            }            }
# Line 1231  for (;;) Line 1264  for (;;)
1264        if (clen > 0)        if (clen > 0)
1265          {          {
1266          BOOL OK;          BOOL OK;
1267          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1268          switch(code[2])          switch(code[2])
1269            {            {
1270            case PT_ANY:            case PT_ANY:
# Line 1239  for (;;) Line 1272  for (;;)
1272            break;            break;
1273    
1274            case PT_LAMP:            case PT_LAMP:
1275            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1276            break;            break;
1277    
1278            case PT_GC:            case PT_GC:
1279            OK = category == code[3];            OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1280            break;            break;
1281    
1282            case PT_PC:            case PT_PC:
1283            OK = chartype == code[3];            OK = prop->chartype == code[3];
1284            break;            break;
1285    
1286            case PT_SC:            case PT_SC:
1287            OK = script == code[3];            OK = prop->script == code[3];
1288            break;            break;
1289    
1290            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1289  for (;;) Line 1322  for (;;)
1322        QS2:        QS2:
1323    
1324        ADD_ACTIVE(state_offset + 2, 0);        ADD_ACTIVE(state_offset + 2, 0);
1325        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1326          {          {
1327          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1328          int ncount = 0;          int ncount = 0;
# Line 1304  for (;;) Line 1337  for (;;)
1337            int nd;            int nd;
1338            int ndlen = 1;            int ndlen = 1;
1339            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1340            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1341            ncount++;            ncount++;
1342            nptr += ndlen;            nptr += ndlen;
1343            }            }
# Line 1478  for (;;) Line 1511  for (;;)
1511        if (clen > 0)        if (clen > 0)
1512          {          {
1513          BOOL OK;          BOOL OK;
1514          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1515          switch(code[4])          switch(code[4])
1516            {            {
1517            case PT_ANY:            case PT_ANY:
# Line 1486  for (;;) Line 1519  for (;;)
1519            break;            break;
1520    
1521            case PT_LAMP:            case PT_LAMP:
1522            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1523            break;            break;
1524    
1525            case PT_GC:            case PT_GC:
1526            OK = category == code[5];            OK = _pcre_ucp_gentype[prop->chartype] == code[5];
1527            break;            break;
1528    
1529            case PT_PC:            case PT_PC:
1530            OK = chartype == code[5];            OK = prop->chartype == code[5];
1531            break;            break;
1532    
1533            case PT_SC:            case PT_SC:
1534            OK = script == code[5];            OK = prop->script == code[5];
1535            break;            break;
1536    
1537            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1531  for (;;) Line 1564  for (;;)
1564        if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)        if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)
1565          { ADD_ACTIVE(state_offset + 4, 0); }          { ADD_ACTIVE(state_offset + 4, 0); }
1566        count = current_state->count;  /* Number already matched */        count = current_state->count;  /* Number already matched */
1567        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1568          {          {
1569          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1570          int ncount = 0;          int ncount = 0;
# Line 1545  for (;;) Line 1578  for (;;)
1578            int nd;            int nd;
1579            int ndlen = 1;            int ndlen = 1;
1580            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1581            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1582            ncount++;            ncount++;
1583            nptr += ndlen;            nptr += ndlen;
1584            }            }
# Line 1725  for (;;) Line 1758  for (;;)
1758            other case of the character. */            other case of the character. */
1759    
1760  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1761            othercase = _pcre_ucp_othercase(c);            othercase = UCD_OTHERCASE(c);
1762  #else  #else
1763            othercase = NOTACHAR;            othercase = NOTACHAR;
1764  #endif  #endif
# Line 1750  for (;;) Line 1783  for (;;)
1783        to wait for them to pass before continuing. */        to wait for them to pass before continuing. */
1784    
1785        case OP_EXTUNI:        case OP_EXTUNI:
1786        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1787          {          {
1788          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1789          int ncount = 0;          int ncount = 0;
# Line 1758  for (;;) Line 1791  for (;;)
1791            {            {
1792            int nclen = 1;            int nclen = 1;
1793            GETCHARLEN(c, nptr, nclen);            GETCHARLEN(c, nptr, nclen);
1794            if (_pcre_ucp_findprop(c, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(c) != ucp_M) break;
1795            ncount++;            ncount++;
1796            nptr += nclen;            nptr += nclen;
1797            }            }
# Line 1926  for (;;) Line 1959  for (;;)
1959            if (utf8 && d >= 128)            if (utf8 && d >= 128)
1960              {              {
1961  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1962              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
1963  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
1964              }              }
1965            else            else
# Line 1964  for (;;) Line 1997  for (;;)
1997            if (utf8 && d >= 128)            if (utf8 && d >= 128)
1998              {              {
1999  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2000              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2001  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2002              }              }
2003            else            else
# Line 2000  for (;;) Line 2033  for (;;)
2033            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2034              {              {
2035  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2036              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2037  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2038              }              }
2039            else            else
# Line 2032  for (;;) Line 2065  for (;;)
2065            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2066              {              {
2067  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2068              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2069  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2070              }              }
2071            else            else
# Line 2067  for (;;) Line 2100  for (;;)
2100            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2101              {              {
2102  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2103              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2104  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2105              }              }
2106            else            else
# Line 2175  for (;;) Line 2208  for (;;)
2208    
2209  /* ========================================================================== */  /* ========================================================================== */
2210        /* These are the opcodes for fancy brackets of various kinds. We have        /* These are the opcodes for fancy brackets of various kinds. We have
2211        to use recursion in order to handle them. The "always failing" assersion        to use recursion in order to handle them. The "always failing" assertion
2212        (?!) is optimised when compiling to OP_FAIL, so we have to support that,        (?!) is optimised to OP_FAIL when compiling, so we have to support that,
2213        though the other "backtracking verbs" are not supported. */        though the other "backtracking verbs" are not supported. */
2214    
2215        case OP_FAIL:        case OP_FAIL:
2216        break;        forced_fail++;    /* Count FAILs for multiple states */
2217          break;
2218    
2219        case OP_ASSERT:        case OP_ASSERT:
2220        case OP_ASSERT_NOT:        case OP_ASSERT_NOT:
# Line 2218  for (;;) Line 2252  for (;;)
2252          {          {
2253          int local_offsets[1000];          int local_offsets[1000];
2254          int local_workspace[1000];          int local_workspace[1000];
2255          int condcode = code[LINK_SIZE+1];          int codelink = GET(code, 1);
2256            int condcode;
2257    
2258            /* Because of the way auto-callout works during compile, a callout item
2259            is inserted between OP_COND and an assertion condition. This does not
2260            happen for the other conditions. */
2261    
2262            if (code[LINK_SIZE+1] == OP_CALLOUT)
2263              {
2264              rrc = 0;
2265              if (pcre_callout != NULL)
2266                {
2267                pcre_callout_block cb;
2268                cb.version          = 1;   /* Version 1 of the callout block */
2269                cb.callout_number   = code[LINK_SIZE+2];
2270                cb.offset_vector    = offsets;
2271                cb.subject          = (PCRE_SPTR)start_subject;
2272                cb.subject_length   = end_subject - start_subject;
2273                cb.start_match      = current_subject - start_subject;
2274                cb.current_position = ptr - start_subject;
2275                cb.pattern_position = GET(code, LINK_SIZE + 3);
2276                cb.next_item_length = GET(code, 3 + 2*LINK_SIZE);
2277                cb.capture_top      = 1;
2278                cb.capture_last     = -1;
2279                cb.callout_data     = md->callout_data;
2280                if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc;   /* Abandon */
2281                }
2282              if (rrc > 0) break;                      /* Fail this thread */
2283              code += _pcre_OP_lengths[OP_CALLOUT];    /* Skip callout data */
2284              }
2285    
2286            condcode = code[LINK_SIZE+1];
2287    
2288          /* Back reference conditions are not supported */          /* Back reference conditions are not supported */
2289    
# Line 2227  for (;;) Line 2292  for (;;)
2292          /* The DEFINE condition is always false */          /* The DEFINE condition is always false */
2293    
2294          if (condcode == OP_DEF)          if (condcode == OP_DEF)
2295            {            { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
           ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0);  
           }  
2296    
2297          /* The only supported version of OP_RREF is for the value RREF_ANY,          /* The only supported version of OP_RREF is for the value RREF_ANY,
2298          which means "test if in any recursion". We can't test for specifically          which means "test if in any recursion". We can't test for specifically
# Line 2239  for (;;) Line 2302  for (;;)
2302            {            {
2303            int value = GET2(code, LINK_SIZE+2);            int value = GET2(code, LINK_SIZE+2);
2304            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;
2305            if (recursing > 0) { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }            if (recursing > 0)
2306              else { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); }              { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }
2307              else { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2308            }            }
2309    
2310          /* Otherwise, the condition is an assertion */          /* Otherwise, the condition is an assertion */
# Line 2270  for (;;) Line 2334  for (;;)
2334                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))
2335              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
2336            else            else
2337              { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); }              { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2338            }            }
2339          }          }
2340        break;        break;
# Line 2422  for (;;) Line 2486  for (;;)
2486        /* Handle callouts */        /* Handle callouts */
2487    
2488        case OP_CALLOUT:        case OP_CALLOUT:
2489          rrc = 0;
2490        if (pcre_callout != NULL)        if (pcre_callout != NULL)
2491          {          {
         int rrc;  
2492          pcre_callout_block cb;          pcre_callout_block cb;
2493          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 1;   /* Version 1 of the callout block */
2494          cb.callout_number   = code[1];          cb.callout_number   = code[1];
# Line 2439  for (;;) Line 2503  for (;;)
2503          cb.capture_last     = -1;          cb.capture_last     = -1;
2504          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
2505          if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc;   /* Abandon */          if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc;   /* Abandon */
         if (rrc == 0) { ADD_ACTIVE(state_offset + 2 + 2*LINK_SIZE, 0); }  
2506          }          }
2507          if (rrc == 0)
2508            { ADD_ACTIVE(state_offset + _pcre_OP_lengths[OP_CALLOUT], 0); }
2509        break;        break;
2510    
2511    
# Line 2456  for (;;) Line 2521  for (;;)
2521    /* We have finished the processing at the current subject character. If no    /* We have finished the processing at the current subject character. If no
2522    new states have been set for the next character, we have found all the    new states have been set for the next character, we have found all the
2523    matches that we are going to find. If we are at the top level and partial    matches that we are going to find. If we are at the top level and partial
2524    matching has been requested, check for appropriate conditions. */    matching has been requested, check for appropriate conditions. The "forced_
2525      fail" variable counts the number of (*F) encountered for the character. If it
2526      is equal to the original active_count (saved in workspace[1]) it means that
2527      (*F) was found on every active state. In this case we don't want to give a
2528      partial match. */
2529    
2530    if (new_count <= 0)    if (new_count <= 0)
2531      {      {
2532      if (match_count < 0 &&                     /* No matches found */      if (rlevel == 1 &&                               /* Top level, and */
2533          rlevel == 1 &&                         /* Top level match function */          reached_end != workspace[1] &&               /* Not all reached end */
2534          (md->moptions & PCRE_PARTIAL) != 0 &&  /* Want partial matching */          forced_fail != workspace[1] &&               /* Not all forced fail & */
2535          ptr >= end_subject &&                  /* Reached end of subject */          (                                            /* either... */
2536          ptr > current_subject)                 /* Matched non-empty string */          (md->moptions & PCRE_PARTIAL_HARD) != 0      /* Hard partial */
2537            ||                                           /* or... */
2538            ((md->moptions & PCRE_PARTIAL_SOFT) != 0 &&  /* Soft partial and */
2539             match_count < 0)                            /* no matches */
2540            ) &&                                         /* And... */
2541            ptr >= end_subject &&                     /* Reached end of subject */
2542            ptr > current_subject)                    /* Matched non-empty string */
2543        {        {
2544        if (offsetcount >= 2)        if (offsetcount >= 2)
2545          {          {
2546          offsets[0] = current_subject - start_subject;          offsets[0] = md->start_used_ptr - start_subject;
2547          offsets[1] = end_subject - start_subject;          offsets[1] = end_subject - start_subject;
2548          }          }
2549        match_count = PCRE_ERROR_PARTIAL;        match_count = PCRE_ERROR_PARTIAL;
# Line 2523  Returns:          > 0 => number of match Line 2598  Returns:          > 0 => number of match
2598                   < -1 => some kind of unexpected problem                   < -1 => some kind of unexpected problem
2599  */  */
2600    
2601  PCRE_EXP_DEFN int  PCRE_EXP_DEFN int PCRE_CALL_CONVENTION
2602  pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,  pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,
2603    const char *subject, int length, int start_offset, int options, int *offsets,    const char *subject, int length, int start_offset, int options, int *offsets,
2604    int offsetcount, int *workspace, int wscount)    int offsetcount, int *workspace, int wscount)
# Line 2610  md->start_code = (const uschar *)argumen Line 2685  md->start_code = (const uschar *)argumen
2685      re->name_table_offset + re->name_count * re->name_entry_size;      re->name_table_offset + re->name_count * re->name_entry_size;
2686  md->start_subject = (const unsigned char *)subject;  md->start_subject = (const unsigned char *)subject;
2687  md->end_subject = end_subject;  md->end_subject = end_subject;
2688    md->start_offset = start_offset;
2689  md->moptions = options;  md->moptions = options;
2690  md->poptions = re->options;  md->poptions = re->options;
2691    
# Line 2632  switch ((((options & PCRE_NEWLINE_BITS) Line 2708  switch ((((options & PCRE_NEWLINE_BITS)
2708           PCRE_NEWLINE_BITS)           PCRE_NEWLINE_BITS)
2709    {    {
2710    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Compile-time default */
2711    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = CHAR_CR; break;
2712    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = CHAR_NL; break;
2713    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
2714         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break;
2715    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
2716    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
2717    default: return PCRE_ERROR_BADNEWLINE;    default: return PCRE_ERROR_BADNEWLINE;
# Line 2731  if ((re->flags & PCRE_REQCHSET) != 0) Line 2807  if ((re->flags & PCRE_REQCHSET) != 0)
2807    }    }
2808    
2809  /* Call the main matching function, looping for a non-anchored regex after a  /* Call the main matching function, looping for a non-anchored regex after a
2810  failed match. Unless restarting, optimize by moving to the first match  failed match. If not restarting, perform certain optimizations at the start of
2811  character if possible, when not anchored. Then unless wanting a partial match,  a match. */
 check for a required later character. */  
2812    
2813  for (;;)  for (;;)
2814    {    {
# Line 2743  for (;;) Line 2818  for (;;)
2818      {      {
2819      const uschar *save_end_subject = end_subject;      const uschar *save_end_subject = end_subject;
2820    
2821      /* Advance to a unique first char if possible. If firstline is TRUE, the      /* If firstline is TRUE, the start of the match is constrained to the first
2822      start of the match is constrained to the first line of a multiline string.      line of a multiline string. Implement this by temporarily adjusting
2823      Implement this by temporarily adjusting end_subject so that we stop      end_subject so that we stop scanning at a newline. If the match fails at
2824      scanning at a newline. If the match fails at the newline, later code breaks      the newline, later code breaks this loop. */
     this loop. */  
2825    
2826      if (firstline)      if (firstline)
2827        {        {
2828        const uschar *t = current_subject;        USPTR t = current_subject;
2829    #ifdef SUPPORT_UTF8
2830          if (utf8)
2831            {
2832            while (t < md->end_subject && !IS_NEWLINE(t))
2833              {
2834              t++;
2835              while (t < end_subject && (*t & 0xc0) == 0x80) t++;
2836              }
2837            }
2838          else
2839    #endif
2840        while (t < md->end_subject && !IS_NEWLINE(t)) t++;        while (t < md->end_subject && !IS_NEWLINE(t)) t++;
2841        end_subject = t;        end_subject = t;
2842        }        }
2843    
2844      if (first_byte >= 0)      /* There are some optimizations that avoid running the match if a known
2845        starting point is not found, or if a known later character is not present.
2846        However, there is an option that disables these, for testing and for
2847        ensuring that all callouts do actually occur. */
2848    
2849        if ((options & PCRE_NO_START_OPTIMIZE) == 0)
2850        {        {
       if (first_byte_caseless)  
         while (current_subject < end_subject &&  
                lcc[*current_subject] != first_byte)  
           current_subject++;  
       else  
         while (current_subject < end_subject && *current_subject != first_byte)  
           current_subject++;  
       }  
2851    
2852      /* Or to just after a linebreak for a multiline match if possible */        /* Advance to a known first byte. */
2853    
2854      else if (startline)        if (first_byte >= 0)
2855        {          {
2856        if (current_subject > md->start_subject + start_offset)          if (first_byte_caseless)
2857              while (current_subject < end_subject &&
2858                     lcc[*current_subject] != first_byte)
2859                current_subject++;
2860            else
2861              while (current_subject < end_subject &&
2862                     *current_subject != first_byte)
2863                current_subject++;
2864            }
2865    
2866          /* Or to just after a linebreak for a multiline match if possible */
2867    
2868          else if (startline)
2869          {          {
2870          while (current_subject <= end_subject && !WAS_NEWLINE(current_subject))          if (current_subject > md->start_subject + start_offset)
2871            current_subject++;            {
2872    #ifdef SUPPORT_UTF8
2873              if (utf8)
2874                {
2875                while (current_subject < end_subject &&
2876                       !WAS_NEWLINE(current_subject))
2877                  {
2878                  current_subject++;
2879                  while(current_subject < end_subject &&
2880                        (*current_subject & 0xc0) == 0x80)
2881                    current_subject++;
2882                  }
2883                }
2884              else
2885    #endif
2886              while (current_subject < end_subject && !WAS_NEWLINE(current_subject))
2887                current_subject++;
2888    
2889          /* If we have just passed a CR and the newline option is ANY or            /* If we have just passed a CR and the newline option is ANY or
2890          ANYCRLF, and we are now at a LF, advance the match position by one more            ANYCRLF, and we are now at a LF, advance the match position by one
2891          character. */            more character. */
2892    
2893          if (current_subject[-1] == '\r' &&            if (current_subject[-1] == CHAR_CR &&
2894               (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&                 (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
2895               current_subject < end_subject &&                 current_subject < end_subject &&
2896               *current_subject == '\n')                 *current_subject == CHAR_NL)
2897            current_subject++;              current_subject++;
2898              }
2899          }          }
       }  
2900    
2901      /* Or to a non-unique first char after study */        /* Or to a non-unique first char after study */
2902    
2903      else if (start_bits != NULL)        else if (start_bits != NULL)
       {  
       while (current_subject < end_subject)  
2904          {          {
2905          register unsigned int c = *current_subject;          while (current_subject < end_subject)
2906          if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++;            {
2907            else break;            register unsigned int c = *current_subject;
2908              if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++;
2909                else break;
2910              }
2911          }          }
2912        }        }
2913    
# Line 2818  for (;;) Line 2929  for (;;)
2929    showed up when somebody was matching /^C/ on a 32-megabyte string... so we    showed up when somebody was matching /^C/ on a 32-megabyte string... so we
2930    don't do this when the string is sufficiently long.    don't do this when the string is sufficiently long.
2931    
2932    ALSO: this processing is disabled when partial matching is requested.    ALSO: this processing is disabled when partial matching is requested, and can
2933    */    also be explicitly deactivated. Furthermore, we have to disable when
2934      restarting after a partial match, because the required character may have
2935      already been matched. */
2936    
2937    if (req_byte >= 0 &&    if ((options & PCRE_NO_START_OPTIMIZE) == 0 &&
2938          req_byte >= 0 &&
2939        end_subject - current_subject < REQ_BYTE_MAX &&        end_subject - current_subject < REQ_BYTE_MAX &&
2940        (options & PCRE_PARTIAL) == 0)        (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_DFA_RESTART)) == 0)
2941      {      {
2942      register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);      register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);
2943    
# Line 2863  for (;;) Line 2977  for (;;)
2977    
2978    /* OK, now we can do the business */    /* OK, now we can do the business */
2979    
2980      md->start_used_ptr = current_subject;
2981    
2982    rc = internal_dfa_exec(    rc = internal_dfa_exec(
2983      md,                                /* fixed match data */      md,                                /* fixed match data */
2984      md->start_code,                    /* this subexpression's code */      md->start_code,                    /* this subexpression's code */
# Line 2897  for (;;) Line 3013  for (;;)
3013    not contain any explicit matches for \r or \n, and the newline option is CRLF    not contain any explicit matches for \r or \n, and the newline option is CRLF
3014    or ANY or ANYCRLF, advance the match position by one more character. */    or ANY or ANYCRLF, advance the match position by one more character. */
3015    
3016    if (current_subject[-1] == '\r' &&    if (current_subject[-1] == CHAR_CR &&
3017        current_subject < end_subject &&        current_subject < end_subject &&
3018        *current_subject == '\n' &&        *current_subject == CHAR_NL &&
3019        (re->flags & PCRE_HASCRORLF) == 0 &&        (re->flags & PCRE_HASCRORLF) == 0 &&
3020          (md->nltype == NLTYPE_ANY ||          (md->nltype == NLTYPE_ANY ||
3021           md->nltype == NLTYPE_ANYCRLF ||           md->nltype == NLTYPE_ANYCRLF ||
Legend:
Removed from v.341  
changed lines
  Added in v.442
  ViewVC Help
Powered by ViewVC 1.1.5