/[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 455 by ph10, Sat Sep 26 19:12:32 2009 UTC revision 510 by ph10, Sat Mar 27 17:45:29 2010 UTC
# Line 7  and semantics are as close as possible t Line 7  and semantics are as close as possible t
7  below for why this module is different).  below for why this module is different).
8    
9                         Written by Philip Hazel                         Written by Philip Hazel
10             Copyright (c) 1997-2009 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 45  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  /* 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  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  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.  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  The issue is the check for duplicate states, which is done by a simple linear
# Line 68  was the extra time to initialize the ind Line 68  was the extra time to initialize the ind
68  of internal_dfa_exec(). (The supplied patch used a static vector, initialized  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.)  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  Overall, I concluded that the gains in some cases did not outweigh the losses
72  in others, so I abandoned this code. */  in others, so I abandoned this code. */
73    
74    
# Line 106  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 155  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 198  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 417  if (*first_op == OP_REVERSE) Line 478  if (*first_op == OP_REVERSE)
478        current_subject - start_subject : max_back;        current_subject - start_subject : max_back;
479      current_subject -= gone_back;      current_subject -= gone_back;
480      }      }
481    
482    /* Save the earliest consulted character */    /* Save the earliest consulted character */
483    
484    if (current_subject < md->start_used_ptr)    if (current_subject < md->start_used_ptr)
485      md->start_used_ptr = current_subject;      md->start_used_ptr = current_subject;
486    
487    /* Now we can process the individual branches. */    /* Now we can process the individual branches. */
488    
# Line 488  for (;;) Line 549  for (;;)
549    int clen, dlen;    int clen, dlen;
550    unsigned int c, d;    unsigned int c, d;
551    int forced_fail = 0;    int forced_fail = 0;
552    int reached_end = 0;    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 502  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 548  for (;;) Line 609  for (;;)
609      int state_offset = current_state->offset;      int state_offset = current_state->offset;
610      int count, codevalue, rrc;      int count, codevalue, rrc;
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 578  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      See the note at the head of this module about the possibility of improving
644      performance here. */      performance here. */
645    
# Line 597  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 643  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, or PCRE_NOTEMPTY_ATSTART is set and we are at the        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        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    
# Line 662  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        else
745          {          {
746          reached_end++;    /* Count branches that reach the end */          if (ptr > current_subject ||
         if (ptr > current_subject ||  
747              ((md->moptions & PCRE_NOTEMPTY) == 0 &&              ((md->moptions & PCRE_NOTEMPTY) == 0 &&
748                ((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 ||                ((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 ||
749                  current_subject > start_subject + md->start_offset)))                  current_subject > start_subject + md->start_offset)))
# Line 689  for (;;) Line 767  for (;;)
767                match_count, rlevel*2-2, SP));                match_count, rlevel*2-2, SP));
768              return match_count;              return match_count;
769              }              }
770            }            }
771          }          }
772        break;        break;
773    
# Line 839  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;            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
# Line 848  for (;;) Line 926  for (;;)
926            }            }
927          else left_word = 0;          else left_word = 0;
928    
929          if (clen > 0)          if (clen > 0)
930            right_word = c < 256 && (ctypes[c] & ctype_word) != 0;            right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
931          else              /* This is a fudge to ensure that if this is the */          else right_word = 0;
           {               /* last item in the pattern, we don't count it as */  
           reached_end--;  /* reached, thus disabling a partial match. */  
           right_word = 0;  
           }  
932    
933          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))
934            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 2241  for (;;) Line 2315  for (;;)
2315            rlevel,                               /* function recursion level */            rlevel,                               /* function recursion level */
2316            recursing);                           /* pass on regex recursion */            recursing);                           /* pass on regex recursion */
2317    
2318            if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2319          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))
2320              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
2321          }          }
# Line 2287  for (;;) Line 2362  for (;;)
2362    
2363          /* Back reference conditions are not supported */          /* Back reference conditions are not supported */
2364    
2365          if (condcode == OP_CREF) return PCRE_ERROR_DFA_UCOND;          if (condcode == OP_CREF || condcode == OP_NCREF)
2366              return PCRE_ERROR_DFA_UCOND;
2367    
2368          /* The DEFINE condition is always false */          /* The DEFINE condition is always false */
2369    
# Line 2298  for (;;) Line 2374  for (;;)
2374          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
2375          recursed groups. */          recursed groups. */
2376    
2377          else if (condcode == OP_RREF)          else if (condcode == OP_RREF || condcode == OP_NRREF)
2378            {            {
2379            int value = GET2(code, LINK_SIZE+2);            int value = GET2(code, LINK_SIZE+2);
2380            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;
# Line 2330  for (;;) Line 2406  for (;;)
2406              rlevel,                               /* function recursion level */              rlevel,                               /* function recursion level */
2407              recursing);                           /* pass on regex recursion */              recursing);                           /* pass on regex recursion */
2408    
2409              if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2410            if ((rc >= 0) ==            if ((rc >= 0) ==
2411                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))
2412              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
# Line 2521  for (;;) Line 2598  for (;;)
2598    /* We have finished the processing at the current subject character. If no    /* We have finished the processing at the current subject character. If no
2599    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
2600    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
2601    matching has been requested, check for appropriate conditions. The "forced_    matching has been requested, check for appropriate conditions.
2602    fail" variable counts the number of (*F) encountered for the character. If it  
2603    is equal to the original active_count (saved in workspace[1]) it means that    The "forced_ fail" variable counts the number of (*F) encountered for the
2604    (*F) was found on every active state. In this case we don't want to give a    character. If it is equal to the original active_count (saved in
2605    partial match. */    workspace[1]) it means that (*F) was found on every active state. In this
2606      case we don't want to give a partial match.
2607    
2608      The "could_continue" variable is true if a state could have continued but
2609      for the fact that the end of the subject was reached. */
2610    
2611    if (new_count <= 0)    if (new_count <= 0)
2612      {      {
2613      if (rlevel == 1 &&                               /* Top level, and */      if (rlevel == 1 &&                               /* Top level, and */
2614          reached_end != workspace[1] &&               /* Not all reached end */          could_continue &&                            /* Some could go on */
2615          forced_fail != workspace[1] &&               /* Not all forced fail & */          forced_fail != workspace[1] &&               /* Not all forced fail & */
2616          (                                            /* either... */          (                                            /* either... */
2617          (md->moptions & PCRE_PARTIAL_HARD) != 0      /* Hard partial */          (md->moptions & PCRE_PARTIAL_HARD) != 0      /* Hard partial */
# Line 2651  if (extra_data != NULL) Line 2732  if (extra_data != NULL)
2732    if ((flags & PCRE_EXTRA_TABLES) != 0)    if ((flags & PCRE_EXTRA_TABLES) != 0)
2733      md->tables = extra_data->tables;      md->tables = extra_data->tables;
2734    }    }
2735    
2736  /* Check that the first field in the block is the magic number. If it is not,  /* Check that the first field in the block is the magic number. If it is not,
2737  test for a regex that was compiled on a host of opposite endianness. If this is  test for a regex that was compiled on a host of opposite endianness. If this is
2738  the case, flipped values are put in internal_re and internal_study if there was  the case, flipped values are put in internal_re and internal_study if there was
# Line 2913  for (;;) Line 2994  for (;;)
2994    
2995      end_subject = save_end_subject;      end_subject = save_end_subject;
2996    
2997      /* The following two optimizations are disabled for partial matching or if      /* The following two optimizations are disabled for partial matching or if
2998      disabling is explicitly requested (and of course, by the test above, this      disabling is explicitly requested (and of course, by the test above, this
2999      code is not obeyed when restarting after a partial match). */      code is not obeyed when restarting after a partial match). */
3000    
3001      if ((options & PCRE_NO_START_OPTIMIZE) == 0 &&      if ((options & PCRE_NO_START_OPTIMIZE) == 0 &&
3002          (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT)) == 0)          (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT)) == 0)
3003        {        {
3004        /* If the pattern was studied, a minimum subject length may be set. This        /* If the pattern was studied, a minimum subject length may be set. This
3005        is a lower bound; no actual string of that length may actually match the        is a lower bound; no actual string of that length may actually match the
3006        pattern. Although the value is, strictly, in characters, we treat it as        pattern. Although the value is, strictly, in characters, we treat it as
3007        bytes to avoid spending too much time in this optimization. */        bytes to avoid spending too much time in this optimization. */
3008    
3009        if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 &&        if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 &&
3010            end_subject - current_subject < study->minlength)            (pcre_uint32)(end_subject - current_subject) < study->minlength)
3011          return PCRE_ERROR_NOMATCH;          return PCRE_ERROR_NOMATCH;
3012    
3013        /* If req_byte is set, we know that that character must appear in the        /* If req_byte is set, we know that that character must appear in the
3014        subject for the match to succeed. If the first character is set, req_byte        subject for the match to succeed. If the first character is set, req_byte
3015        must be later in the subject; otherwise the test starts at the match        must be later in the subject; otherwise the test starts at the match
# Line 2936  for (;;) Line 3017  for (;;)
3017        nested unlimited repeats that aren't going to match. Writing separate        nested unlimited repeats that aren't going to match. Writing separate
3018        code for cased/caseless versions makes it go faster, as does using an        code for cased/caseless versions makes it go faster, as does using an
3019        autoincrement and backing off on a match.        autoincrement and backing off on a match.
3020    
3021        HOWEVER: when the subject string is very, very long, searching to its end        HOWEVER: when the subject string is very, very long, searching to its end
3022        can take a long time, and give bad performance on quite ordinary        can take a long time, and give bad performance on quite ordinary
3023        patterns. This showed up when somebody was matching /^C/ on a 32-megabyte        patterns. This showed up when somebody was matching /^C/ on a 32-megabyte
3024        string... so we don't do this when the string is sufficiently long. */        string... so we don't do this when the string is sufficiently long. */
3025    
3026        if (req_byte >= 0 && end_subject - current_subject < REQ_BYTE_MAX)        if (req_byte >= 0 && end_subject - current_subject < REQ_BYTE_MAX)
3027          {          {
3028          register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);          register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);
3029    
3030          /* We don't need to repeat the search if we haven't yet reached the          /* We don't need to repeat the search if we haven't yet reached the
3031          place we found it at last time. */          place we found it at last time. */
3032    
3033          if (p > req_byte_ptr)          if (p > req_byte_ptr)
3034            {            {
3035            if (req_byte_caseless)            if (req_byte_caseless)
# Line 2966  for (;;) Line 3047  for (;;)
3047                if (*p++ == req_byte) { p--; break; }                if (*p++ == req_byte) { p--; break; }
3048                }                }
3049              }              }
3050    
3051            /* If we can't find the required character, break the matching loop,            /* If we can't find the required character, break the matching loop,
3052            which will cause a return or PCRE_ERROR_NOMATCH. */            which will cause a return or PCRE_ERROR_NOMATCH. */
3053    
3054            if (p >= end_subject) break;            if (p >= end_subject) break;
3055    
3056            /* If we have found the required character, save the point where we            /* If we have found the required character, save the point where we
3057            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
3058            the start hasn't passed this character yet. */            the start hasn't passed this character yet. */
3059    
3060            req_byte_ptr = p;            req_byte_ptr = p;
3061            }            }
3062          }          }
3063        }        }
3064      }   /* End of optimizations that are done when not restarting */      }   /* End of optimizations that are done when not restarting */
3065    
3066    /* OK, now we can do the business */    /* OK, now we can do the business */
3067    
3068    md->start_used_ptr = current_subject;    md->start_used_ptr = current_subject;
3069    
3070    rc = internal_dfa_exec(    rc = internal_dfa_exec(
3071      md,                                /* fixed match data */      md,                                /* fixed match data */
3072      md->start_code,                    /* this subexpression's code */      md->start_code,                    /* this subexpression's code */

Legend:
Removed from v.455  
changed lines
  Added in v.510

  ViewVC Help
Powered by ViewVC 1.1.5