/[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 345 by ph10, Mon Apr 28 15:10:02 2008 UTC revision 545 by ph10, Wed Jun 16 10:51:15 2010 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) 1997-2010 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 78  never stored, so we push them well clear Line 106  never stored, so we push them well clear
106    
107    
108  /* This table identifies those opcodes that are followed immediately by a  /* This table identifies those opcodes that are followed immediately by a
109  character that is to be tested in some way. This makes is possible to  character that is to be tested in some way. This makes it possible to
110  centralize the loading of these characters. In the case of Type * etc, the  centralize the loading of these characters. In the case of Type * etc, the
111  "character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a  "character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a
112  small value. ***NOTE*** If the start of this table is modified, the two tables  small value. Non-zero values in the table are the offsets from the opcode where
113  that follow must also be modified. */  the character is to be found. ***NOTE*** If the start of this table is
114    modified, the three tables that follow must also be modified. */
115    
116  static const uschar coptable[] = {  static const uschar coptable[] = {
117    0,                             /* End                                    */    0,                             /* End                                    */
118    0, 0, 0, 0, 0,                 /* \A, \G, \K, \B, \b                     */    0, 0, 0, 0, 0,                 /* \A, \G, \K, \B, \b                     */
119    0, 0, 0, 0, 0, 0,              /* \D, \d, \S, \s, \W, \w                 */    0, 0, 0, 0, 0, 0,              /* \D, \d, \S, \s, \W, \w                 */
120    0, 0, 0,                       /* Any, AllAny, Anybyte                   */    0, 0, 0,                       /* Any, AllAny, Anybyte                   */
121    0, 0, 0,                       /* NOTPROP, PROP, EXTUNI                  */    0, 0,                          /* \P, \p                                 */
122    0, 0, 0, 0, 0,                 /* \R, \H, \h, \V, \v                     */    0, 0, 0, 0, 0,                 /* \R, \H, \h, \V, \v                     */
123      0,                             /* \X                                     */
124    0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */    0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */
125    1,                             /* Char                                   */    1,                             /* Char                                   */
126    1,                             /* Charnc                                 */    1,                             /* Charnc                                 */
# Line 127  static const uschar coptable[] = { Line 157  static const uschar coptable[] = {
157    0,                             /* Reverse                                */    0,                             /* Reverse                                */
158    0, 0, 0, 0,                    /* ONCE, BRA, CBRA, COND                  */    0, 0, 0, 0,                    /* ONCE, BRA, CBRA, COND                  */
159    0, 0, 0,                       /* SBRA, SCBRA, SCOND                     */    0, 0, 0,                       /* SBRA, SCBRA, SCOND                     */
160    0,                             /* CREF                                   */    0, 0,                          /* CREF, NCREF                            */
161    0,                             /* RREF                                   */    0, 0,                          /* RREF, NRREF                            */
162    0,                             /* DEF                                    */    0,                             /* DEF                                    */
163    0, 0,                          /* BRAZERO, BRAMINZERO                    */    0, 0,                          /* BRAZERO, BRAMINZERO                    */
164    0, 0, 0, 0,                    /* PRUNE, SKIP, THEN, COMMIT              */    0, 0, 0,                       /* MARK, PRUNE, PRUNE_ARG,                */
165    0, 0, 0                        /* FAIL, ACCEPT, SKIPZERO                 */    0, 0, 0, 0,                    /* SKIP, SKIP_ARG, THEN, THEN_ARG,        */
166      0, 0, 0, 0, 0                  /* COMMIT, FAIL, ACCEPT, CLOSE, SKIPZERO  */
167    };
168    
169    /* This table identifies those opcodes that inspect a character. It is used to
170    remember the fact that a character could have been inspected when the end of
171    the subject is reached. ***NOTE*** If the start of this table is modified, the
172    two tables that follow must also be modified. */
173    
174    static const uschar poptable[] = {
175      0,                             /* End                                    */
176      0, 0, 0, 1, 1,                 /* \A, \G, \K, \B, \b                     */
177      1, 1, 1, 1, 1, 1,              /* \D, \d, \S, \s, \W, \w                 */
178      1, 1, 1,                       /* Any, AllAny, Anybyte                   */
179      1, 1,                          /* \P, \p                                 */
180      1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */
181      1,                             /* \X                                     */
182      0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */
183      1,                             /* Char                                   */
184      1,                             /* Charnc                                 */
185      1,                             /* not                                    */
186      /* Positive single-char repeats                                          */
187      1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */
188      1, 1, 1,                       /* upto, minupto, exact                   */
189      1, 1, 1, 1,                    /* *+, ++, ?+, upto+                      */
190      /* Negative single-char repeats - only for chars < 256                   */
191      1, 1, 1, 1, 1, 1,              /* NOT *, *?, +, +?, ?, ??                */
192      1, 1, 1,                       /* NOT upto, minupto, exact               */
193      1, 1, 1, 1,                    /* NOT *+, ++, ?+, upto+                  */
194      /* Positive type repeats                                                 */
195      1, 1, 1, 1, 1, 1,              /* Type *, *?, +, +?, ?, ??               */
196      1, 1, 1,                       /* Type upto, minupto, exact              */
197      1, 1, 1, 1,                    /* Type *+, ++, ?+, upto+                 */
198      /* Character class & ref repeats                                         */
199      1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */
200      1, 1,                          /* CRRANGE, CRMINRANGE                    */
201      1,                             /* CLASS                                  */
202      1,                             /* NCLASS                                 */
203      1,                             /* XCLASS - variable length               */
204      0,                             /* REF                                    */
205      0,                             /* RECURSE                                */
206      0,                             /* CALLOUT                                */
207      0,                             /* Alt                                    */
208      0,                             /* Ket                                    */
209      0,                             /* KetRmax                                */
210      0,                             /* KetRmin                                */
211      0,                             /* Assert                                 */
212      0,                             /* Assert not                             */
213      0,                             /* Assert behind                          */
214      0,                             /* Assert behind not                      */
215      0,                             /* Reverse                                */
216      0, 0, 0, 0,                    /* ONCE, BRA, CBRA, COND                  */
217      0, 0, 0,                       /* SBRA, SCBRA, SCOND                     */
218      0, 0,                          /* CREF, NCREF                            */
219      0, 0,                          /* RREF, NRREF                            */
220      0,                             /* DEF                                    */
221      0, 0,                          /* BRAZERO, BRAMINZERO                    */
222      0, 0, 0,                       /* MARK, PRUNE, PRUNE_ARG,                */
223      0, 0, 0, 0,                    /* SKIP, SKIP_ARG, THEN, THEN_ARG,        */
224      0, 0, 0, 0, 0                  /* COMMIT, FAIL, ACCEPT, CLOSE, SKIPZERO  */
225  };  };
226    
227  /* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W,  /* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W,
# Line 170  typedef struct stateblock { Line 259  typedef struct stateblock {
259  #define INTS_PER_STATEBLOCK  (sizeof(stateblock)/sizeof(int))  #define INTS_PER_STATEBLOCK  (sizeof(stateblock)/sizeof(int))
260    
261    
262  #ifdef DEBUG  #ifdef PCRE_DEBUG
263  /*************************************************  /*************************************************
264  *             Print character string             *  *             Print character string             *
265  *************************************************/  *************************************************/
# Line 386  if (*first_op == OP_REVERSE) Line 475  if (*first_op == OP_REVERSE)
475    
476      {      {
477      gone_back = (current_subject - max_back < start_subject)?      gone_back = (current_subject - max_back < start_subject)?
478        current_subject - start_subject : max_back;        (int)(current_subject - start_subject) : max_back;
479      current_subject -= gone_back;      current_subject -= gone_back;
480      }      }
481    
482      /* Save the earliest consulted character */
483    
484      if (current_subject < md->start_used_ptr)
485        md->start_used_ptr = current_subject;
486    
487    /* Now we can process the individual branches. */    /* Now we can process the individual branches. */
488    
489    end_code = this_start_code;    end_code = this_start_code;
# Line 398  if (*first_op == OP_REVERSE) Line 492  if (*first_op == OP_REVERSE)
492      int back = GET(end_code, 2+LINK_SIZE);      int back = GET(end_code, 2+LINK_SIZE);
493      if (back <= gone_back)      if (back <= gone_back)
494        {        {
495        int bstate = end_code - start_code + 2 + 2*LINK_SIZE;        int bstate = (int)(end_code - start_code + 2 + 2*LINK_SIZE);
496        ADD_NEW_DATA(-bstate, 0, gone_back - back);        ADD_NEW_DATA(-bstate, 0, gone_back - back);
497        }        }
498      end_code += GET(end_code, 1);      end_code += GET(end_code, 1);
# Line 434  else Line 528  else
528        ((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA)? 2:0);        ((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA)? 2:0);
529      do      do
530        {        {
531        ADD_NEW(end_code - start_code + length, 0);        ADD_NEW((int)(end_code - start_code + length), 0);
532        end_code += GET(end_code, 1);        end_code += GET(end_code, 1);
533        length = 1 + LINK_SIZE;        length = 1 + LINK_SIZE;
534        }        }
# Line 454  for (;;) Line 548  for (;;)
548    int i, j;    int i, j;
549    int clen, dlen;    int clen, dlen;
550    unsigned int c, d;    unsigned int c, d;
551      int forced_fail = 0;
552      BOOL could_continue = FALSE;
553    
554    /* 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
555    new state list. */    new state list. */
# Line 467  for (;;) Line 563  for (;;)
563    workspace[0] ^= 1;              /* Remember for the restarting feature */    workspace[0] ^= 1;              /* Remember for the restarting feature */
564    workspace[1] = active_count;    workspace[1] = active_count;
565    
566  #ifdef DEBUG  #ifdef PCRE_DEBUG
567    printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP);    printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP);
568    pchars((uschar *)ptr, strlen((char *)ptr), stdout);    pchars((uschar *)ptr, strlen((char *)ptr), stdout);
569    printf("\"\n");    printf("\"\n");
# Line 511  for (;;) Line 607  for (;;)
607      stateblock *current_state = active_states + i;      stateblock *current_state = active_states + i;
608      const uschar *code;      const uschar *code;
609      int state_offset = current_state->offset;      int state_offset = current_state->offset;
610      int count, codevalue;      int count, codevalue, rrc;
 #ifdef SUPPORT_UCP  
     int chartype, script;  
 #endif  
611    
612  #ifdef DEBUG  #ifdef PCRE_DEBUG
613      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);
614      if (clen == 0) printf("EOL\n");      if (clen == 0) printf("EOL\n");
615        else if (c > 32 && c < 127) printf("'%c'\n", c);        else if (c > 32 && c < 127) printf("'%c'\n", c);
# Line 546  for (;;) Line 639  for (;;)
639          }          }
640        }        }
641    
642      /* 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.
643        See the note at the head of this module about the possibility of improving
644        performance here. */
645    
646      for (j = 0; j < i; j++)      for (j = 0; j < i; j++)
647        {        {
# Line 563  for (;;) Line 658  for (;;)
658      code = start_code + state_offset;      code = start_code + state_offset;
659      codevalue = *code;      codevalue = *code;
660    
661        /* If this opcode inspects a character, but we are at the end of the
662        subject, remember the fact for use when testing for a partial match. */
663    
664        if (clen == 0 && poptable[codevalue] != 0)
665          could_continue = TRUE;
666    
667      /* If this opcode is followed by an inline character, load it. It is      /* If this opcode is followed by an inline character, load it. It is
668      tempting to test for the presence of a subject character here, but that      tempting to test for the presence of a subject character here, but that
669      is wrong, because sometimes zero repetitions of the subject are      is wrong, because sometimes zero repetitions of the subject are
# Line 609  for (;;) Line 710  for (;;)
710    
711      switch (codevalue)      switch (codevalue)
712        {        {
713    /* ========================================================================== */
714          /* These cases are never obeyed. This is a fudge that causes a compile-
715          time error if the vectors coptable or poptable, which are indexed by
716          opcode, are not the correct length. It seems to be the only way to do
717          such a check at compile time, as the sizeof() operator does not work
718          in the C preprocessor. */
719    
720          case OP_TABLE_LENGTH:
721          case OP_TABLE_LENGTH +
722            ((sizeof(coptable) == OP_TABLE_LENGTH) &&
723             (sizeof(poptable) == OP_TABLE_LENGTH)):
724          break;
725    
726  /* ========================================================================== */  /* ========================================================================== */
727        /* 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
728        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
729        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
730          start of the subject, save the match data, shifting up all previous
731        matches so we always have the longest first. */        matches so we always have the longest first. */
732    
733        case OP_KET:        case OP_KET:
# Line 627  for (;;) Line 741  for (;;)
741            ADD_ACTIVE(state_offset - GET(code, 1), 0);            ADD_ACTIVE(state_offset - GET(code, 1), 0);
742            }            }
743          }          }
744        else if (ptr > current_subject || (md->moptions & PCRE_NOTEMPTY) == 0)        else
745          {          {
746          if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;          if (ptr > current_subject ||
747            else if (match_count > 0 && ++match_count * 2 >= offsetcount)              ((md->moptions & PCRE_NOTEMPTY) == 0 &&
748              match_count = 0;                ((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 ||
749          count = ((match_count == 0)? offsetcount : match_count * 2) - 2;                  current_subject > start_subject + md->start_offset)))
750          if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));            {
751          if (offsetcount >= 2)            if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
752            {              else if (match_count > 0 && ++match_count * 2 >= offsetcount)
753            offsets[0] = current_subject - start_subject;                match_count = 0;
754            offsets[1] = ptr - start_subject;            count = ((match_count == 0)? offsetcount : match_count * 2) - 2;
755            DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,            if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));
756              offsets[1] - offsets[0], current_subject));            if (offsetcount >= 2)
757            }              {
758          if ((md->moptions & PCRE_DFA_SHORTEST) != 0)              offsets[0] = (int)(current_subject - start_subject);
759            {              offsets[1] = (int)(ptr - start_subject);
760            DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"              DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,
761              "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,                offsets[1] - offsets[0], current_subject));
762              match_count, rlevel*2-2, SP));              }
763            return match_count;            if ((md->moptions & PCRE_DFA_SHORTEST) != 0)
764                {
765                DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"
766                  "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,
767                  match_count, rlevel*2-2, SP));
768                return match_count;
769                }
770            }            }
771          }          }
772        break;        break;
# Line 658  for (;;) Line 778  for (;;)
778        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
779        case OP_ALT:        case OP_ALT:
780        do { code += GET(code, 1); } while (*code == OP_ALT);        do { code += GET(code, 1); } while (*code == OP_ALT);
781        ADD_ACTIVE(code - start_code, 0);        ADD_ACTIVE((int)(code - start_code), 0);
782        break;        break;
783    
784        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
# Line 666  for (;;) Line 786  for (;;)
786        case OP_SBRA:        case OP_SBRA:
787        do        do
788          {          {
789          ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0);          ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
790          code += GET(code, 1);          code += GET(code, 1);
791          }          }
792        while (*code == OP_ALT);        while (*code == OP_ALT);
# Line 675  for (;;) Line 795  for (;;)
795        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
796        case OP_CBRA:        case OP_CBRA:
797        case OP_SCBRA:        case OP_SCBRA:
798        ADD_ACTIVE(code - start_code + 3 + LINK_SIZE,  0);        ADD_ACTIVE((int)(code - start_code + 3 + LINK_SIZE),  0);
799        code += GET(code, 1);        code += GET(code, 1);
800        while (*code == OP_ALT)        while (*code == OP_ALT)
801          {          {
802          ADD_ACTIVE(code - start_code + 1 + LINK_SIZE,  0);          ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE),  0);
803          code += GET(code, 1);          code += GET(code, 1);
804          }          }
805        break;        break;
# Line 690  for (;;) Line 810  for (;;)
810        ADD_ACTIVE(state_offset + 1, 0);        ADD_ACTIVE(state_offset + 1, 0);
811        code += 1 + GET(code, 2);        code += 1 + GET(code, 2);
812        while (*code == OP_ALT) code += GET(code, 1);        while (*code == OP_ALT) code += GET(code, 1);
813        ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0);        ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
814        break;        break;
815    
816        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
817        case OP_SKIPZERO:        case OP_SKIPZERO:
818        code += 1 + GET(code, 2);        code += 1 + GET(code, 2);
819        while (*code == OP_ALT) code += GET(code, 1);        while (*code == OP_ALT) code += GET(code, 1);
820        ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0);        ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
821        break;        break;
822    
823        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
# Line 760  for (;;) Line 880  for (;;)
880        if ((md->moptions & PCRE_NOTEOL) == 0)        if ((md->moptions & PCRE_NOTEOL) == 0)
881          {          {
882          if (clen == 0 ||          if (clen == 0 ||
883              (IS_NEWLINE(ptr) &&              ((md->poptions & PCRE_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr) &&
884                 ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen)                 ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen)
885              ))              ))
886            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 797  for (;;) Line 917  for (;;)
917          if (ptr > start_subject)          if (ptr > start_subject)
918            {            {
919            const uschar *temp = ptr - 1;            const uschar *temp = ptr - 1;
920              if (temp < md->start_used_ptr) md->start_used_ptr = temp;
921  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
922            if (utf8) BACKCHAR(temp);            if (utf8) BACKCHAR(temp);
923  #endif  #endif
924            GETCHARTEST(d, temp);            GETCHARTEST(d, temp);
925    #ifdef SUPPORT_UCP
926              if ((md->poptions & PCRE_UCP) != 0)
927                {
928                if (d == '_') left_word = TRUE; else
929                  {
930                  int cat = UCD_CATEGORY(d);
931                  left_word = (cat == ucp_L || cat == ucp_N);
932                  }
933                }
934              else
935    #endif
936            left_word = d < 256 && (ctypes[d] & ctype_word) != 0;            left_word = d < 256 && (ctypes[d] & ctype_word) != 0;
937            }            }
938          else left_word = 0;          else left_word = FALSE;
939    
940          if (clen > 0) right_word = c < 256 && (ctypes[c] & ctype_word) != 0;          if (clen > 0)
941            else right_word = 0;            {
942    #ifdef SUPPORT_UCP
943              if ((md->poptions & PCRE_UCP) != 0)
944                {
945                if (c == '_') right_word = TRUE; else
946                  {
947                  int cat = UCD_CATEGORY(c);
948                  right_word = (cat == ucp_L || cat == ucp_N);
949                  }
950                }
951              else
952    #endif
953              right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
954              }
955            else right_word = FALSE;
956    
957          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))
958            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 825  for (;;) Line 971  for (;;)
971        if (clen > 0)        if (clen > 0)
972          {          {
973          BOOL OK;          BOOL OK;
974          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
975          switch(code[1])          switch(code[1])
976            {            {
977            case PT_ANY:            case PT_ANY:
# Line 833  for (;;) Line 979  for (;;)
979            break;            break;
980    
981            case PT_LAMP:            case PT_LAMP:
982            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
983                   prop->chartype == ucp_Lt;
984            break;            break;
985    
986            case PT_GC:            case PT_GC:
987            OK = category == code[2];            OK = _pcre_ucp_gentype[prop->chartype] == code[2];
988            break;            break;
989    
990            case PT_PC:            case PT_PC:
991            OK = chartype == code[2];            OK = prop->chartype == code[2];
992            break;            break;
993    
994            case PT_SC:            case PT_SC:
995            OK = script == code[2];            OK = prop->script == code[2];
996              break;
997    
998              /* These are specials for combination cases. */
999    
1000              case PT_ALNUM:
1001              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1002                   _pcre_ucp_gentype[prop->chartype] == ucp_N;
1003              break;
1004    
1005              case PT_SPACE:    /* Perl space */
1006              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1007                   c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1008              break;
1009    
1010              case PT_PXSPACE:  /* POSIX space */
1011              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1012                   c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1013                   c == CHAR_FF || c == CHAR_CR;
1014              break;
1015    
1016              case PT_WORD:
1017              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1018                   _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1019                   c == CHAR_UNDERSCORE;
1020            break;            break;
1021    
1022            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 994  for (;;) Line 1165  for (;;)
1165        if (clen > 0)        if (clen > 0)
1166          {          {
1167          BOOL OK;          BOOL OK;
1168          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1169          switch(code[2])          switch(code[2])
1170            {            {
1171            case PT_ANY:            case PT_ANY:
# Line 1002  for (;;) Line 1173  for (;;)
1173            break;            break;
1174    
1175            case PT_LAMP:            case PT_LAMP:
1176            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1177                prop->chartype == ucp_Lt;
1178            break;            break;
1179    
1180            case PT_GC:            case PT_GC:
1181            OK = category == code[3];            OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1182            break;            break;
1183    
1184            case PT_PC:            case PT_PC:
1185            OK = chartype == code[3];            OK = prop->chartype == code[3];
1186            break;            break;
1187    
1188            case PT_SC:            case PT_SC:
1189            OK = script == code[3];            OK = prop->script == code[3];
1190              break;
1191    
1192              /* These are specials for combination cases. */
1193    
1194              case PT_ALNUM:
1195              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1196                   _pcre_ucp_gentype[prop->chartype] == ucp_N;
1197              break;
1198    
1199              case PT_SPACE:    /* Perl space */
1200              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1201                   c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1202              break;
1203    
1204              case PT_PXSPACE:  /* POSIX space */
1205              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1206                   c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1207                   c == CHAR_FF || c == CHAR_CR;
1208              break;
1209    
1210              case PT_WORD:
1211              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1212                   _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1213                   c == CHAR_UNDERSCORE;
1214            break;            break;
1215    
1216            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1043  for (;;) Line 1239  for (;;)
1239        case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:        case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:
1240        count = current_state->count;  /* Already matched */        count = current_state->count;  /* Already matched */
1241        if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }        if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1242        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1243          {          {
1244          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1245          int ncount = 0;          int ncount = 0;
# Line 1057  for (;;) Line 1253  for (;;)
1253            int nd;            int nd;
1254            int ndlen = 1;            int ndlen = 1;
1255            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1256            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1257            ncount++;            ncount++;
1258            nptr += ndlen;            nptr += ndlen;
1259            }            }
# Line 1216  for (;;) Line 1412  for (;;)
1412        if (clen > 0)        if (clen > 0)
1413          {          {
1414          BOOL OK;          BOOL OK;
1415          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1416          switch(code[2])          switch(code[2])
1417            {            {
1418            case PT_ANY:            case PT_ANY:
# Line 1224  for (;;) Line 1420  for (;;)
1420            break;            break;
1421    
1422            case PT_LAMP:            case PT_LAMP:
1423            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1424                prop->chartype == ucp_Lt;
1425            break;            break;
1426    
1427            case PT_GC:            case PT_GC:
1428            OK = category == code[3];            OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1429            break;            break;
1430    
1431            case PT_PC:            case PT_PC:
1432            OK = chartype == code[3];            OK = prop->chartype == code[3];
1433            break;            break;
1434    
1435            case PT_SC:            case PT_SC:
1436            OK = script == code[3];            OK = prop->script == code[3];
1437              break;
1438    
1439              /* These are specials for combination cases. */
1440    
1441              case PT_ALNUM:
1442              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1443                   _pcre_ucp_gentype[prop->chartype] == ucp_N;
1444              break;
1445    
1446              case PT_SPACE:    /* Perl space */
1447              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1448                   c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1449              break;
1450    
1451              case PT_PXSPACE:  /* POSIX space */
1452              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1453                   c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1454                   c == CHAR_FF || c == CHAR_CR;
1455              break;
1456    
1457              case PT_WORD:
1458              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1459                   _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1460                   c == CHAR_UNDERSCORE;
1461            break;            break;
1462    
1463            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1274  for (;;) Line 1495  for (;;)
1495        QS2:        QS2:
1496    
1497        ADD_ACTIVE(state_offset + 2, 0);        ADD_ACTIVE(state_offset + 2, 0);
1498        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1499          {          {
1500          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1501          int ncount = 0;          int ncount = 0;
# Line 1289  for (;;) Line 1510  for (;;)
1510            int nd;            int nd;
1511            int ndlen = 1;            int ndlen = 1;
1512            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1513            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1514            ncount++;            ncount++;
1515            nptr += ndlen;            nptr += ndlen;
1516            }            }
# Line 1463  for (;;) Line 1684  for (;;)
1684        if (clen > 0)        if (clen > 0)
1685          {          {
1686          BOOL OK;          BOOL OK;
1687          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1688          switch(code[4])          switch(code[4])
1689            {            {
1690            case PT_ANY:            case PT_ANY:
# Line 1471  for (;;) Line 1692  for (;;)
1692            break;            break;
1693    
1694            case PT_LAMP:            case PT_LAMP:
1695            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1696                prop->chartype == ucp_Lt;
1697            break;            break;
1698    
1699            case PT_GC:            case PT_GC:
1700            OK = category == code[5];            OK = _pcre_ucp_gentype[prop->chartype] == code[5];
1701            break;            break;
1702    
1703            case PT_PC:            case PT_PC:
1704            OK = chartype == code[5];            OK = prop->chartype == code[5];
1705            break;            break;
1706    
1707            case PT_SC:            case PT_SC:
1708            OK = script == code[5];            OK = prop->script == code[5];
1709              break;
1710    
1711              /* These are specials for combination cases. */
1712    
1713              case PT_ALNUM:
1714              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1715                   _pcre_ucp_gentype[prop->chartype] == ucp_N;
1716              break;
1717    
1718              case PT_SPACE:    /* Perl space */
1719              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1720                   c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1721              break;
1722    
1723              case PT_PXSPACE:  /* POSIX space */
1724              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1725                   c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1726                   c == CHAR_FF || c == CHAR_CR;
1727              break;
1728    
1729              case PT_WORD:
1730              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1731                   _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1732                   c == CHAR_UNDERSCORE;
1733            break;            break;
1734    
1735            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1516  for (;;) Line 1762  for (;;)
1762        if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)        if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)
1763          { ADD_ACTIVE(state_offset + 4, 0); }          { ADD_ACTIVE(state_offset + 4, 0); }
1764        count = current_state->count;  /* Number already matched */        count = current_state->count;  /* Number already matched */
1765        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1766          {          {
1767          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1768          int ncount = 0;          int ncount = 0;
# Line 1530  for (;;) Line 1776  for (;;)
1776            int nd;            int nd;
1777            int ndlen = 1;            int ndlen = 1;
1778            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1779            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1780            ncount++;            ncount++;
1781            nptr += ndlen;            nptr += ndlen;
1782            }            }
# Line 1710  for (;;) Line 1956  for (;;)
1956            other case of the character. */            other case of the character. */
1957    
1958  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1959            othercase = _pcre_ucp_othercase(c);            othercase = UCD_OTHERCASE(c);
1960  #else  #else
1961            othercase = NOTACHAR;            othercase = NOTACHAR;
1962  #endif  #endif
# Line 1735  for (;;) Line 1981  for (;;)
1981        to wait for them to pass before continuing. */        to wait for them to pass before continuing. */
1982    
1983        case OP_EXTUNI:        case OP_EXTUNI:
1984        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1985          {          {
1986          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1987          int ncount = 0;          int ncount = 0;
# Line 1743  for (;;) Line 1989  for (;;)
1989            {            {
1990            int nclen = 1;            int nclen = 1;
1991            GETCHARLEN(c, nptr, nclen);            GETCHARLEN(c, nptr, nclen);
1992            if (_pcre_ucp_findprop(c, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(c) != ucp_M) break;
1993            ncount++;            ncount++;
1994            nptr += nclen;            nptr += nclen;
1995            }            }
# Line 1911  for (;;) Line 2157  for (;;)
2157            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2158              {              {
2159  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2160              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2161  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2162              }              }
2163            else            else
# Line 1949  for (;;) Line 2195  for (;;)
2195            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2196              {              {
2197  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2198              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2199  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2200              }              }
2201            else            else
# Line 1985  for (;;) Line 2231  for (;;)
2231            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2232              {              {
2233  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2234              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2235  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2236              }              }
2237            else            else
# Line 2017  for (;;) Line 2263  for (;;)
2263            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2264              {              {
2265  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2266              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2267  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2268              }              }
2269            else            else
# Line 2052  for (;;) Line 2298  for (;;)
2298            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2299              {              {
2300  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2301              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2302  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2303              }              }
2304            else            else
# Line 2113  for (;;) Line 2359  for (;;)
2359          points to the byte after the end of the class. If there is a          points to the byte after the end of the class. If there is a
2360          quantifier, this is where it will be. */          quantifier, this is where it will be. */
2361    
2362          next_state_offset = ecode - start_code;          next_state_offset = (int)(ecode - start_code);
2363    
2364          switch (*ecode)          switch (*ecode)
2365            {            {
# Line 2160  for (;;) Line 2406  for (;;)
2406    
2407  /* ========================================================================== */  /* ========================================================================== */
2408        /* These are the opcodes for fancy brackets of various kinds. We have        /* These are the opcodes for fancy brackets of various kinds. We have
2409        to use recursion in order to handle them. The "always failing" assersion        to use recursion in order to handle them. The "always failing" assertion
2410        (?!) 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,
2411        though the other "backtracking verbs" are not supported. */        though the other "backtracking verbs" are not supported. */
2412    
2413        case OP_FAIL:        case OP_FAIL:
2414          forced_fail++;    /* Count FAILs for multiple states */
2415        break;        break;
2416    
2417        case OP_ASSERT:        case OP_ASSERT:
# Line 2183  for (;;) Line 2430  for (;;)
2430            md,                                   /* static match data */            md,                                   /* static match data */
2431            code,                                 /* this subexpression's code */            code,                                 /* this subexpression's code */
2432            ptr,                                  /* where we currently are */            ptr,                                  /* where we currently are */
2433            ptr - start_subject,                  /* start offset */            (int)(ptr - start_subject),           /* start offset */
2434            local_offsets,                        /* offset vector */            local_offsets,                        /* offset vector */
2435            sizeof(local_offsets)/sizeof(int),    /* size of same */            sizeof(local_offsets)/sizeof(int),    /* size of same */
2436            local_workspace,                      /* workspace vector */            local_workspace,                      /* workspace vector */
# Line 2192  for (;;) Line 2439  for (;;)
2439            rlevel,                               /* function recursion level */            rlevel,                               /* function recursion level */
2440            recursing);                           /* pass on regex recursion */            recursing);                           /* pass on regex recursion */
2441    
2442            if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2443          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))
2444              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); }
2445          }          }
2446        break;        break;
2447    
# Line 2203  for (;;) Line 2451  for (;;)
2451          {          {
2452          int local_offsets[1000];          int local_offsets[1000];
2453          int local_workspace[1000];          int local_workspace[1000];
2454          int condcode = code[LINK_SIZE+1];          int codelink = GET(code, 1);
2455            int condcode;
2456    
2457            /* Because of the way auto-callout works during compile, a callout item
2458            is inserted between OP_COND and an assertion condition. This does not
2459            happen for the other conditions. */
2460    
2461            if (code[LINK_SIZE+1] == OP_CALLOUT)
2462              {
2463              rrc = 0;
2464              if (pcre_callout != NULL)
2465                {
2466                pcre_callout_block cb;
2467                cb.version          = 1;   /* Version 1 of the callout block */
2468                cb.callout_number   = code[LINK_SIZE+2];
2469                cb.offset_vector    = offsets;
2470                cb.subject          = (PCRE_SPTR)start_subject;
2471                cb.subject_length   = (int)(end_subject - start_subject);
2472                cb.start_match      = (int)(current_subject - start_subject);
2473                cb.current_position = (int)(ptr - start_subject);
2474                cb.pattern_position = GET(code, LINK_SIZE + 3);
2475                cb.next_item_length = GET(code, 3 + 2*LINK_SIZE);
2476                cb.capture_top      = 1;
2477                cb.capture_last     = -1;
2478                cb.callout_data     = md->callout_data;
2479                if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc;   /* Abandon */
2480                }
2481              if (rrc > 0) break;                      /* Fail this thread */
2482              code += _pcre_OP_lengths[OP_CALLOUT];    /* Skip callout data */
2483              }
2484    
2485            condcode = code[LINK_SIZE+1];
2486    
2487          /* Back reference conditions are not supported */          /* Back reference conditions are not supported */
2488    
2489          if (condcode == OP_CREF) return PCRE_ERROR_DFA_UCOND;          if (condcode == OP_CREF || condcode == OP_NCREF)
2490              return PCRE_ERROR_DFA_UCOND;
2491    
2492          /* The DEFINE condition is always false */          /* The DEFINE condition is always false */
2493    
2494          if (condcode == OP_DEF)          if (condcode == OP_DEF)
2495            {            { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
           ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0);  
           }  
2496    
2497          /* 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,
2498          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
2499          recursed groups. */          recursed groups. */
2500    
2501          else if (condcode == OP_RREF)          else if (condcode == OP_RREF || condcode == OP_NRREF)
2502            {            {
2503            int value = GET2(code, LINK_SIZE+2);            int value = GET2(code, LINK_SIZE+2);
2504            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;
2505            if (recursing > 0) { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }            if (recursing > 0)
2506              else { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); }              { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }
2507              else { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2508            }            }
2509    
2510          /* Otherwise, the condition is an assertion */          /* Otherwise, the condition is an assertion */
# Line 2242  for (;;) Line 2521  for (;;)
2521              md,                                   /* fixed match data */              md,                                   /* fixed match data */
2522              asscode,                              /* this subexpression's code */              asscode,                              /* this subexpression's code */
2523              ptr,                                  /* where we currently are */              ptr,                                  /* where we currently are */
2524              ptr - start_subject,                  /* start offset */              (int)(ptr - start_subject),           /* start offset */
2525              local_offsets,                        /* offset vector */              local_offsets,                        /* offset vector */
2526              sizeof(local_offsets)/sizeof(int),    /* size of same */              sizeof(local_offsets)/sizeof(int),    /* size of same */
2527              local_workspace,                      /* workspace vector */              local_workspace,                      /* workspace vector */
# Line 2251  for (;;) Line 2530  for (;;)
2530              rlevel,                               /* function recursion level */              rlevel,                               /* function recursion level */
2531              recursing);                           /* pass on regex recursion */              recursing);                           /* pass on regex recursion */
2532    
2533              if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2534            if ((rc >= 0) ==            if ((rc >= 0) ==
2535                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))
2536              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); }
2537            else            else
2538              { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); }              { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2539            }            }
2540          }          }
2541        break;        break;
# Line 2274  for (;;) Line 2554  for (;;)
2554            md,                                   /* fixed match data */            md,                                   /* fixed match data */
2555            start_code + GET(code, 1),            /* this subexpression's code */            start_code + GET(code, 1),            /* this subexpression's code */
2556            ptr,                                  /* where we currently are */            ptr,                                  /* where we currently are */
2557            ptr - start_subject,                  /* start offset */            (int)(ptr - start_subject),           /* start offset */
2558            local_offsets,                        /* offset vector */            local_offsets,                        /* offset vector */
2559            sizeof(local_offsets)/sizeof(int),    /* size of same */            sizeof(local_offsets)/sizeof(int),    /* size of same */
2560            local_workspace,                      /* workspace vector */            local_workspace,                      /* workspace vector */
# Line 2326  for (;;) Line 2606  for (;;)
2606            md,                                   /* fixed match data */            md,                                   /* fixed match data */
2607            code,                                 /* this subexpression's code */            code,                                 /* this subexpression's code */
2608            ptr,                                  /* where we currently are */            ptr,                                  /* where we currently are */
2609            ptr - start_subject,                  /* start offset */            (int)(ptr - start_subject),           /* start offset */
2610            local_offsets,                        /* offset vector */            local_offsets,                        /* offset vector */
2611            sizeof(local_offsets)/sizeof(int),    /* size of same */            sizeof(local_offsets)/sizeof(int),    /* size of same */
2612            local_workspace,                      /* workspace vector */            local_workspace,                      /* workspace vector */
# Line 2343  for (;;) Line 2623  for (;;)
2623    
2624            do { end_subpattern += GET(end_subpattern, 1); }            do { end_subpattern += GET(end_subpattern, 1); }
2625              while (*end_subpattern == OP_ALT);              while (*end_subpattern == OP_ALT);
2626            next_state_offset = end_subpattern - start_code + LINK_SIZE + 1;            next_state_offset =
2627                (int)(end_subpattern - start_code + LINK_SIZE + 1);
2628    
2629            /* If the end of this subpattern is KETRMAX or KETRMIN, we must            /* If the end of this subpattern is KETRMAX or KETRMIN, we must
2630            arrange for the repeat state also to be added to the relevant list.            arrange for the repeat state also to be added to the relevant list.
# Line 2351  for (;;) Line 2632  for (;;)
2632    
2633            repeat_state_offset = (*end_subpattern == OP_KETRMAX ||            repeat_state_offset = (*end_subpattern == OP_KETRMAX ||
2634                                   *end_subpattern == OP_KETRMIN)?                                   *end_subpattern == OP_KETRMIN)?
2635              end_subpattern - start_code - GET(end_subpattern, 1) : -1;              (int)(end_subpattern - start_code - GET(end_subpattern, 1)) : -1;
2636    
2637            /* If we have matched an empty string, add the next state at the            /* If we have matched an empty string, add the next state at the
2638            current character pointer. This is important so that the duplicate            current character pointer. This is important so that the duplicate
# Line 2407  for (;;) Line 2688  for (;;)
2688        /* Handle callouts */        /* Handle callouts */
2689    
2690        case OP_CALLOUT:        case OP_CALLOUT:
2691          rrc = 0;
2692        if (pcre_callout != NULL)        if (pcre_callout != NULL)
2693          {          {
         int rrc;  
2694          pcre_callout_block cb;          pcre_callout_block cb;
2695          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 1;   /* Version 1 of the callout block */
2696          cb.callout_number   = code[1];          cb.callout_number   = code[1];
2697          cb.offset_vector    = offsets;          cb.offset_vector    = offsets;
2698          cb.subject          = (PCRE_SPTR)start_subject;          cb.subject          = (PCRE_SPTR)start_subject;
2699          cb.subject_length   = end_subject - start_subject;          cb.subject_length   = (int)(end_subject - start_subject);
2700          cb.start_match      = current_subject - start_subject;          cb.start_match      = (int)(current_subject - start_subject);
2701          cb.current_position = ptr - start_subject;          cb.current_position = (int)(ptr - start_subject);
2702          cb.pattern_position = GET(code, 2);          cb.pattern_position = GET(code, 2);
2703          cb.next_item_length = GET(code, 2 + LINK_SIZE);          cb.next_item_length = GET(code, 2 + LINK_SIZE);
2704          cb.capture_top      = 1;          cb.capture_top      = 1;
2705          cb.capture_last     = -1;          cb.capture_last     = -1;
2706          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
2707          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); }  
2708          }          }
2709          if (rrc == 0)
2710            { ADD_ACTIVE(state_offset + _pcre_OP_lengths[OP_CALLOUT], 0); }
2711        break;        break;
2712    
2713    
# Line 2441  for (;;) Line 2723  for (;;)
2723    /* We have finished the processing at the current subject character. If no    /* We have finished the processing at the current subject character. If no
2724    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
2725    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
2726    matching has been requested, check for appropriate conditions. */    matching has been requested, check for appropriate conditions.
2727    
2728      The "forced_ fail" variable counts the number of (*F) encountered for the
2729      character. If it is equal to the original active_count (saved in
2730      workspace[1]) it means that (*F) was found on every active state. In this
2731      case we don't want to give a partial match.
2732    
2733      The "could_continue" variable is true if a state could have continued but
2734      for the fact that the end of the subject was reached. */
2735    
2736    if (new_count <= 0)    if (new_count <= 0)
2737      {      {
2738      if (match_count < 0 &&                     /* No matches found */      if (rlevel == 1 &&                               /* Top level, and */
2739          rlevel == 1 &&                         /* Top level match function */          could_continue &&                            /* Some could go on */
2740          (md->moptions & PCRE_PARTIAL) != 0 &&  /* Want partial matching */          forced_fail != workspace[1] &&               /* Not all forced fail & */
2741          ptr >= end_subject &&                  /* Reached end of subject */          (                                            /* either... */
2742          ptr > current_subject)                 /* Matched non-empty string */          (md->moptions & PCRE_PARTIAL_HARD) != 0      /* Hard partial */
2743            ||                                           /* or... */
2744            ((md->moptions & PCRE_PARTIAL_SOFT) != 0 &&  /* Soft partial and */
2745             match_count < 0)                            /* no matches */
2746            ) &&                                         /* And... */
2747            ptr >= end_subject &&                     /* Reached end of subject */
2748            ptr > current_subject)                    /* Matched non-empty string */
2749        {        {
2750        if (offsetcount >= 2)        if (offsetcount >= 2)
2751          {          {
2752          offsets[0] = current_subject - start_subject;          offsets[0] = (int)(md->start_used_ptr - start_subject);
2753          offsets[1] = end_subject - start_subject;          offsets[1] = (int)(end_subject - start_subject);
2754          }          }
2755        match_count = PCRE_ERROR_PARTIAL;        match_count = PCRE_ERROR_PARTIAL;
2756        }        }
# Line 2508  Returns: > 0 => number of match Line 2804  Returns: > 0 => number of match
2804                   < -1 => some kind of unexpected problem                   < -1 => some kind of unexpected problem
2805  */  */
2806    
2807  PCRE_EXP_DEFN int  PCRE_EXP_DEFN int PCRE_CALL_CONVENTION
2808  pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,  pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,
2809    const char *subject, int length, int start_offset, int options, int *offsets,    const char *subject, int length, int start_offset, int options, int *offsets,
2810    int offsetcount, int *workspace, int wscount)    int offsetcount, int *workspace, int wscount)
# Line 2595  md->start_code = (const uschar *)argumen Line 2891  md->start_code = (const uschar *)argumen
2891      re->name_table_offset + re->name_count * re->name_entry_size;      re->name_table_offset + re->name_count * re->name_entry_size;
2892  md->start_subject = (const unsigned char *)subject;  md->start_subject = (const unsigned char *)subject;
2893  md->end_subject = end_subject;  md->end_subject = end_subject;
2894    md->start_offset = start_offset;
2895  md->moptions = options;  md->moptions = options;
2896  md->poptions = re->options;  md->poptions = re->options;
2897    
# Line 2617  switch ((((options & PCRE_NEWLINE_BITS) Line 2914  switch ((((options & PCRE_NEWLINE_BITS)
2914           PCRE_NEWLINE_BITS)           PCRE_NEWLINE_BITS)
2915    {    {
2916    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Compile-time default */
2917    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = CHAR_CR; break;
2918    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = CHAR_NL; break;
2919    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
2920         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break;
2921    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
2922    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
2923    default: return PCRE_ERROR_BADNEWLINE;    default: return PCRE_ERROR_BADNEWLINE;
# Line 2699  if (!anchored) Line 2996  if (!anchored)
2996      }      }
2997    else    else
2998      {      {
2999      if (startline && study != NULL &&      if (!startline && study != NULL &&
3000           (study->options & PCRE_STUDY_MAPPED) != 0)           (study->flags & PCRE_STUDY_MAPPED) != 0)
3001        start_bits = study->start_bits;        start_bits = study->start_bits;
3002      }      }
3003    }    }
# Line 2716  if ((re->flags & PCRE_REQCHSET) != 0) Line 3013  if ((re->flags & PCRE_REQCHSET) != 0)
3013    }    }
3014    
3015  /* 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
3016  failed match. Unless restarting, optimize by moving to the first match  failed match. If not restarting, perform certain optimizations at the start of
3017  character if possible, when not anchored. Then unless wanting a partial match,  a match. */
 check for a required later character. */  
3018    
3019  for (;;)  for (;;)
3020    {    {
# Line 2728  for (;;) Line 3024  for (;;)
3024      {      {
3025      const uschar *save_end_subject = end_subject;      const uschar *save_end_subject = end_subject;
3026    
3027      /* 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
3028      start of the match is constrained to the first line of a multiline string.      line of a multiline string. Implement this by temporarily adjusting
3029      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
3030      scanning at a newline. If the match fails at the newline, later code breaks      the newline, later code breaks this loop. */
     this loop. */  
3031    
3032      if (firstline)      if (firstline)
3033        {        {
3034        const uschar *t = current_subject;        USPTR t = current_subject;
3035    #ifdef SUPPORT_UTF8
3036          if (utf8)
3037            {
3038            while (t < md->end_subject && !IS_NEWLINE(t))
3039              {
3040              t++;
3041              while (t < end_subject && (*t & 0xc0) == 0x80) t++;
3042              }
3043            }
3044          else
3045    #endif
3046        while (t < md->end_subject && !IS_NEWLINE(t)) t++;        while (t < md->end_subject && !IS_NEWLINE(t)) t++;
3047        end_subject = t;        end_subject = t;
3048        }        }
3049    
3050      if (first_byte >= 0)      /* There are some optimizations that avoid running the match if a known
3051        starting point is not found. However, there is an option that disables
3052        these, for testing and for ensuring that all callouts do actually occur. */
3053    
3054        if ((options & PCRE_NO_START_OPTIMIZE) == 0)
3055        {        {
3056        if (first_byte_caseless)        /* Advance to a known first byte. */
         while (current_subject < end_subject &&  
                lcc[*current_subject] != first_byte)  
           current_subject++;  
       else  
         while (current_subject < end_subject && *current_subject != first_byte)  
           current_subject++;  
       }  
3057    
3058      /* Or to just after a linebreak for a multiline match if possible */        if (first_byte >= 0)
3059            {
3060            if (first_byte_caseless)
3061              while (current_subject < end_subject &&
3062                     lcc[*current_subject] != first_byte)
3063                current_subject++;
3064            else
3065              while (current_subject < end_subject &&
3066                     *current_subject != first_byte)
3067                current_subject++;
3068            }
3069    
3070      else if (startline)        /* Or to just after a linebreak for a multiline match if possible */
3071        {  
3072        if (current_subject > md->start_subject + start_offset)        else if (startline)
3073          {          {
3074          while (current_subject <= end_subject && !WAS_NEWLINE(current_subject))          if (current_subject > md->start_subject + start_offset)
3075            current_subject++;            {
3076    #ifdef SUPPORT_UTF8
3077              if (utf8)
3078                {
3079                while (current_subject < end_subject &&
3080                       !WAS_NEWLINE(current_subject))
3081                  {
3082                  current_subject++;
3083                  while(current_subject < end_subject &&
3084                        (*current_subject & 0xc0) == 0x80)
3085                    current_subject++;
3086                  }
3087                }
3088              else
3089    #endif
3090              while (current_subject < end_subject && !WAS_NEWLINE(current_subject))
3091                current_subject++;
3092    
3093          /* 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
3094          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
3095          character. */            more character. */
3096    
3097          if (current_subject[-1] == '\r' &&            if (current_subject[-1] == CHAR_CR &&
3098               (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&                 (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
3099               current_subject < end_subject &&                 current_subject < end_subject &&
3100               *current_subject == '\n')                 *current_subject == CHAR_NL)
3101            current_subject++;              current_subject++;
3102              }
3103          }          }
       }  
3104    
3105      /* Or to a non-unique first char after study */        /* Or to a non-unique first char after study */
3106    
3107      else if (start_bits != NULL)        else if (start_bits != NULL)
       {  
       while (current_subject < end_subject)  
3108          {          {
3109          register unsigned int c = *current_subject;          while (current_subject < end_subject)
3110          if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++;            {
3111              register unsigned int c = *current_subject;
3112              if ((start_bits[c/8] & (1 << (c&7))) == 0)
3113                {
3114                current_subject++;
3115    #ifdef SUPPORT_UTF8
3116                if (utf8)
3117                  while(current_subject < end_subject &&
3118                        (*current_subject & 0xc0) == 0x80) current_subject++;
3119    #endif
3120                }
3121            else break;            else break;
3122              }
3123          }          }
3124        }        }
3125    
3126      /* Restore fudged end_subject */      /* Restore fudged end_subject */
3127    
3128      end_subject = save_end_subject;      end_subject = save_end_subject;
     }  
   
   /* If req_byte is set, we know that that character must appear in the subject  
   for the match to succeed. If the first character is set, req_byte must be  
   later in the subject; otherwise the test starts at the match point. This  
   optimization can save a huge amount of work in patterns with nested unlimited  
   repeats that aren't going to match. Writing separate code for cased/caseless  
   versions makes it go faster, as does using an autoincrement and backing off  
   on a match.  
   
   HOWEVER: when the subject string is very, very long, searching to its end can  
   take a long time, and give bad performance on quite ordinary patterns. This  
   showed up when somebody was matching /^C/ on a 32-megabyte string... so we  
   don't do this when the string is sufficiently long.  
   
   ALSO: this processing is disabled when partial matching is requested.  
   */  
   
   if (req_byte >= 0 &&  
       end_subject - current_subject < REQ_BYTE_MAX &&  
       (options & PCRE_PARTIAL) == 0)  
     {  
     register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);  
3129    
3130      /* We don't need to repeat the search if we haven't yet reached the      /* The following two optimizations are disabled for partial matching or if
3131      place we found it at last time. */      disabling is explicitly requested (and of course, by the test above, this
3132        code is not obeyed when restarting after a partial match). */
3133    
3134      if (p > req_byte_ptr)      if ((options & PCRE_NO_START_OPTIMIZE) == 0 &&
3135            (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT)) == 0)
3136        {        {
3137        if (req_byte_caseless)        /* If the pattern was studied, a minimum subject length may be set. This
3138          {        is a lower bound; no actual string of that length may actually match the
3139          while (p < end_subject)        pattern. Although the value is, strictly, in characters, we treat it as
3140            {        bytes to avoid spending too much time in this optimization. */
3141            register int pp = *p++;  
3142            if (pp == req_byte || pp == req_byte2) { p--; break; }        if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 &&
3143            }            (pcre_uint32)(end_subject - current_subject) < study->minlength)
3144          }          return PCRE_ERROR_NOMATCH;
3145        else  
3146          /* If req_byte is set, we know that that character must appear in the
3147          subject for the match to succeed. If the first character is set, req_byte
3148          must be later in the subject; otherwise the test starts at the match
3149          point. This optimization can save a huge amount of work in patterns with
3150          nested unlimited repeats that aren't going to match. Writing separate
3151          code for cased/caseless versions makes it go faster, as does using an
3152          autoincrement and backing off on a match.
3153    
3154          HOWEVER: when the subject string is very, very long, searching to its end
3155          can take a long time, and give bad performance on quite ordinary
3156          patterns. This showed up when somebody was matching /^C/ on a 32-megabyte
3157          string... so we don't do this when the string is sufficiently long. */
3158    
3159          if (req_byte >= 0 && end_subject - current_subject < REQ_BYTE_MAX)
3160          {          {
3161          while (p < end_subject)          register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);
3162    
3163            /* We don't need to repeat the search if we haven't yet reached the
3164            place we found it at last time. */
3165    
3166            if (p > req_byte_ptr)
3167            {            {
3168            if (*p++ == req_byte) { p--; break; }            if (req_byte_caseless)
3169            }              {
3170          }              while (p < end_subject)
3171                  {
3172                  register int pp = *p++;
3173                  if (pp == req_byte || pp == req_byte2) { p--; break; }
3174                  }
3175                }
3176              else
3177                {
3178                while (p < end_subject)
3179                  {
3180                  if (*p++ == req_byte) { p--; break; }
3181                  }
3182                }
3183    
3184        /* If we can't find the required character, break the matching loop,            /* If we can't find the required character, break the matching loop,
3185        which will cause a return or PCRE_ERROR_NOMATCH. */            which will cause a return or PCRE_ERROR_NOMATCH. */
3186    
3187        if (p >= end_subject) break;            if (p >= end_subject) break;
3188    
3189        /* If we have found the required character, save the point where we            /* If we have found the required character, save the point where we
3190        found it, so that we don't search again next time round the loop if            found it, so that we don't search again next time round the loop if
3191        the start hasn't passed this character yet. */            the start hasn't passed this character yet. */
3192    
3193        req_byte_ptr = p;            req_byte_ptr = p;
3194              }
3195            }
3196        }        }
3197      }      }   /* End of optimizations that are done when not restarting */
3198    
3199    /* OK, now we can do the business */    /* OK, now we can do the business */
3200    
3201      md->start_used_ptr = current_subject;
3202    
3203    rc = internal_dfa_exec(    rc = internal_dfa_exec(
3204      md,                                /* fixed match data */      md,                                /* fixed match data */
3205      md->start_code,                    /* this subexpression's code */      md->start_code,                    /* this subexpression's code */
# Line 2882  for (;;) Line 3234  for (;;)
3234    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
3235    or ANY or ANYCRLF, advance the match position by one more character. */    or ANY or ANYCRLF, advance the match position by one more character. */
3236    
3237    if (current_subject[-1] == '\r' &&    if (current_subject[-1] == CHAR_CR &&
3238        current_subject < end_subject &&        current_subject < end_subject &&
3239        *current_subject == '\n' &&        *current_subject == CHAR_NL &&
3240        (re->flags & PCRE_HASCRORLF) == 0 &&        (re->flags & PCRE_HASCRORLF) == 0 &&
3241          (md->nltype == NLTYPE_ANY ||          (md->nltype == NLTYPE_ANY ||
3242           md->nltype == NLTYPE_ANYCRLF ||           md->nltype == NLTYPE_ANYCRLF ||
Legend:
Removed from v.345  
changed lines
  Added in v.545
webmaster@exim.org
 ViewVC Help
Powered by ViewVC 1.1.12