/[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 236 by ph10, Tue Sep 11 12:57:06 2007 UTC revision 498 by ph10, Wed Mar 3 19:29:38 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) 1997-2007 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 81  never stored, so we push them well clear Line 109  never stored, so we push them well clear
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 is 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 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,                          /* Any, 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 uschar coptable[] = { Line 157  static 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, 0,                    /* PRUNE, SKIP, THEN, COMMIT              */
165    0, 0                           /* FAIL, ACCEPT                           */    0, 0, 0, 0                     /* FAIL, ACCEPT, CLOSE, SKIPZERO          */
166    };
167    
168    /* This table identifies those opcodes that inspect a character. It is used to
169    remember the fact that a character could have been inspected when the end of
170    the subject is reached. ***NOTE*** If the start of this table is modified, the
171    two tables that follow must also be modified. */
172    
173    static const uschar poptable[] = {
174      0,                             /* End                                    */
175      0, 0, 0, 1, 1,                 /* \A, \G, \K, \B, \b                     */
176      1, 1, 1, 1, 1, 1,              /* \D, \d, \S, \s, \W, \w                 */
177      1, 1, 1,                       /* Any, AllAny, Anybyte                   */
178      1, 1,                          /* \P, \p                                 */
179      1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */
180      1,                             /* \X                                     */
181      0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */
182      1,                             /* Char                                   */
183      1,                             /* Charnc                                 */
184      1,                             /* not                                    */
185      /* Positive single-char repeats                                          */
186      1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */
187      1, 1, 1,                       /* upto, minupto, exact                   */
188      1, 1, 1, 1,                    /* *+, ++, ?+, upto+                      */
189      /* Negative single-char repeats - only for chars < 256                   */
190      1, 1, 1, 1, 1, 1,              /* NOT *, *?, +, +?, ?, ??                */
191      1, 1, 1,                       /* NOT upto, minupto, exact               */
192      1, 1, 1, 1,                    /* NOT *+, ++, ?+, upto+                  */
193      /* Positive type repeats                                                 */
194      1, 1, 1, 1, 1, 1,              /* Type *, *?, +, +?, ?, ??               */
195      1, 1, 1,                       /* Type upto, minupto, exact              */
196      1, 1, 1, 1,                    /* Type *+, ++, ?+, upto+                 */
197      /* Character class & ref repeats                                         */
198      1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */
199      1, 1,                          /* CRRANGE, CRMINRANGE                    */
200      1,                             /* CLASS                                  */
201      1,                             /* NCLASS                                 */
202      1,                             /* XCLASS - variable length               */
203      0,                             /* REF                                    */
204      0,                             /* RECURSE                                */
205      0,                             /* CALLOUT                                */
206      0,                             /* Alt                                    */
207      0,                             /* Ket                                    */
208      0,                             /* KetRmax                                */
209      0,                             /* KetRmin                                */
210      0,                             /* Assert                                 */
211      0,                             /* Assert not                             */
212      0,                             /* Assert behind                          */
213      0,                             /* Assert behind not                      */
214      0,                             /* Reverse                                */
215      0, 0, 0, 0,                    /* ONCE, BRA, CBRA, COND                  */
216      0, 0, 0,                       /* SBRA, SCBRA, SCOND                     */
217      0, 0,                          /* CREF, NCREF                            */
218      0, 0,                          /* RREF, NRREF                            */
219      0,                             /* DEF                                    */
220      0, 0,                          /* BRAZERO, BRAMINZERO                    */
221      0, 0, 0, 0,                    /* PRUNE, SKIP, THEN, COMMIT              */
222      0, 0, 0, 0                     /* FAIL, ACCEPT, CLOSE, SKIPZERO          */
223  };  };
224    
225  /* 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,
226  and \w */  and \w */
227    
228  static uschar toptable1[] = {  static const uschar toptable1[] = {
229    0, 0, 0, 0, 0, 0,    0, 0, 0, 0, 0, 0,
230    ctype_digit, ctype_digit,    ctype_digit, ctype_digit,
231    ctype_space, ctype_space,    ctype_space, ctype_space,
232    ctype_word,  ctype_word,    ctype_word,  ctype_word,
233    0                               /* OP_ANY */    0, 0                            /* OP_ANY, OP_ALLANY */
234  };  };
235    
236  static uschar toptable2[] = {  static const uschar toptable2[] = {
237    0, 0, 0, 0, 0, 0,    0, 0, 0, 0, 0, 0,
238    ctype_digit, 0,    ctype_digit, 0,
239    ctype_space, 0,    ctype_space, 0,
240    ctype_word,  0,    ctype_word,  0,
241    1                               /* OP_ANY */    1, 1                            /* OP_ANY, OP_ALLANY */
242  };  };
243    
244    
# Line 170  typedef struct stateblock { Line 257  typedef struct stateblock {
257  #define INTS_PER_STATEBLOCK  (sizeof(stateblock)/sizeof(int))  #define INTS_PER_STATEBLOCK  (sizeof(stateblock)/sizeof(int))
258    
259    
260  #ifdef DEBUG  #ifdef PCRE_DEBUG
261  /*************************************************  /*************************************************
262  *             Print character string             *  *             Print character string             *
263  *************************************************/  *************************************************/
# Line 223  Arguments: Line 310  Arguments:
310    rlevel            function call recursion level    rlevel            function call recursion level
311    recursing         regex recursive call level    recursing         regex recursive call level
312    
313  Returns:            > 0 =>  Returns:            > 0 => number of match offset pairs placed in offsets
314                      = 0 =>                      = 0 => offsets overflowed; longest matches are present
315                       -1 => failed to match                       -1 => failed to match
316                     < -1 => some kind of unexpected problem                     < -1 => some kind of unexpected problem
317    
# Line 390  if (*first_op == OP_REVERSE) Line 477  if (*first_op == OP_REVERSE)
477      current_subject -= gone_back;      current_subject -= gone_back;
478      }      }
479    
480      /* Save the earliest consulted character */
481    
482      if (current_subject < md->start_used_ptr)
483        md->start_used_ptr = current_subject;
484    
485    /* Now we can process the individual branches. */    /* Now we can process the individual branches. */
486    
487    end_code = this_start_code;    end_code = this_start_code;
# Line 454  for (;;) Line 546  for (;;)
546    int i, j;    int i, j;
547    int clen, dlen;    int clen, dlen;
548    unsigned int c, d;    unsigned int c, d;
549      int forced_fail = 0;
550      BOOL could_continue = FALSE;
551    
552    /* 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
553    new state list. */    new state list. */
# Line 467  for (;;) Line 561  for (;;)
561    workspace[0] ^= 1;              /* Remember for the restarting feature */    workspace[0] ^= 1;              /* Remember for the restarting feature */
562    workspace[1] = active_count;    workspace[1] = active_count;
563    
564  #ifdef DEBUG  #ifdef PCRE_DEBUG
565    printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP);    printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP);
566    pchars((uschar *)ptr, strlen((char *)ptr), stdout);    pchars((uschar *)ptr, strlen((char *)ptr), stdout);
567    printf("\"\n");    printf("\"\n");
# Line 511  for (;;) Line 605  for (;;)
605      stateblock *current_state = active_states + i;      stateblock *current_state = active_states + i;
606      const uschar *code;      const uschar *code;
607      int state_offset = current_state->offset;      int state_offset = current_state->offset;
608      int count, codevalue;      int count, codevalue, rrc;
 #ifdef SUPPORT_UCP  
     int chartype, script;  
 #endif  
609    
610  #ifdef DEBUG  #ifdef PCRE_DEBUG
611      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);
612      if (clen == 0) printf("EOL\n");      if (clen == 0) printf("EOL\n");
613        else if (c > 32 && c < 127) printf("'%c'\n", c);        else if (c > 32 && c < 127) printf("'%c'\n", c);
# Line 546  for (;;) Line 637  for (;;)
637          }          }
638        }        }
639    
640      /* 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.
641        See the note at the head of this module about the possibility of improving
642        performance here. */
643    
644      for (j = 0; j < i; j++)      for (j = 0; j < i; j++)
645        {        {
# Line 563  for (;;) Line 656  for (;;)
656      code = start_code + state_offset;      code = start_code + state_offset;
657      codevalue = *code;      codevalue = *code;
658    
659        /* If this opcode inspects a character, but we are at the end of the
660        subject, remember the fact for use when testing for a partial match. */
661    
662        if (clen == 0 && poptable[codevalue] != 0)
663          could_continue = TRUE;
664    
665      /* 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
666      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
667      is wrong, because sometimes zero repetitions of the subject are      is wrong, because sometimes zero repetitions of the subject are
# Line 609  for (;;) Line 708  for (;;)
708    
709      switch (codevalue)      switch (codevalue)
710        {        {
711    /* ========================================================================== */
712          /* These cases are never obeyed. This is a fudge that causes a compile-
713          time error if the vectors coptable or poptable, which are indexed by
714          opcode, are not the correct length. It seems to be the only way to do
715          such a check at compile time, as the sizeof() operator does not work
716          in the C preprocessor. */
717    
718          case OP_TABLE_LENGTH:
719          case OP_TABLE_LENGTH +
720            ((sizeof(coptable) == OP_TABLE_LENGTH) &&
721             (sizeof(poptable) == OP_TABLE_LENGTH)):
722          break;
723    
724  /* ========================================================================== */  /* ========================================================================== */
725        /* 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
726        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
727        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
728          start of the subject, save the match data, shifting up all previous
729        matches so we always have the longest first. */        matches so we always have the longest first. */
730    
731        case OP_KET:        case OP_KET:
# Line 627  for (;;) Line 739  for (;;)
739            ADD_ACTIVE(state_offset - GET(code, 1), 0);            ADD_ACTIVE(state_offset - GET(code, 1), 0);
740            }            }
741          }          }
742        else if (ptr > current_subject || (md->moptions & PCRE_NOTEMPTY) == 0)        else
743          {          {
744          if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;          if (ptr > current_subject ||
745            else if (match_count > 0 && ++match_count * 2 >= offsetcount)              ((md->moptions & PCRE_NOTEMPTY) == 0 &&
746              match_count = 0;                ((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 ||
747          count = ((match_count == 0)? offsetcount : match_count * 2) - 2;                  current_subject > start_subject + md->start_offset)))
748          if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));            {
749          if (offsetcount >= 2)            if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
750            {              else if (match_count > 0 && ++match_count * 2 >= offsetcount)
751            offsets[0] = current_subject - start_subject;                match_count = 0;
752            offsets[1] = ptr - start_subject;            count = ((match_count == 0)? offsetcount : match_count * 2) - 2;
753            DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,            if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));
754              offsets[1] - offsets[0], current_subject));            if (offsetcount >= 2)
755            }              {
756          if ((md->moptions & PCRE_DFA_SHORTEST) != 0)              offsets[0] = current_subject - start_subject;
757            {              offsets[1] = ptr - start_subject;
758            DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"              DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,
759              "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,                offsets[1] - offsets[0], current_subject));
760              match_count, rlevel*2-2, SP));              }
761            return match_count;            if ((md->moptions & PCRE_DFA_SHORTEST) != 0)
762                {
763                DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"
764                  "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,
765                  match_count, rlevel*2-2, SP));
766                return match_count;
767                }
768            }            }
769          }          }
770        break;        break;
# Line 694  for (;;) Line 812  for (;;)
812        break;        break;
813    
814        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
815          case OP_SKIPZERO:
816          code += 1 + GET(code, 2);
817          while (*code == OP_ALT) code += GET(code, 1);
818          ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0);
819          break;
820    
821          /*-----------------------------------------------------------------*/
822        case OP_CIRC:        case OP_CIRC:
823        if ((ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0) ||        if ((ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0) ||
824            ((ims & PCRE_MULTILINE) != 0 &&            ((ims & PCRE_MULTILINE) != 0 &&
# Line 732  for (;;) Line 857  for (;;)
857    
858        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
859        case OP_ANY:        case OP_ANY:
860        if (clen > 0 && ((ims & PCRE_DOTALL) != 0 || !IS_NEWLINE(ptr)))        if (clen > 0 && !IS_NEWLINE(ptr))
861            { ADD_NEW(state_offset + 1, 0); }
862          break;
863    
864          /*-----------------------------------------------------------------*/
865          case OP_ALLANY:
866          if (clen > 0)
867          { ADD_NEW(state_offset + 1, 0); }          { ADD_NEW(state_offset + 1, 0); }
868        break;        break;
869    
# Line 747  for (;;) Line 878  for (;;)
878        if ((md->moptions & PCRE_NOTEOL) == 0)        if ((md->moptions & PCRE_NOTEOL) == 0)
879          {          {
880          if (clen == 0 ||          if (clen == 0 ||
881              (IS_NEWLINE(ptr) &&              ((md->poptions & PCRE_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr) &&
882                 ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen)                 ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen)
883              ))              ))
884            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 784  for (;;) Line 915  for (;;)
915          if (ptr > start_subject)          if (ptr > start_subject)
916            {            {
917            const uschar *temp = ptr - 1;            const uschar *temp = ptr - 1;
918              if (temp < md->start_used_ptr) md->start_used_ptr = temp;
919  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
920            if (utf8) BACKCHAR(temp);            if (utf8) BACKCHAR(temp);
921  #endif  #endif
# Line 792  for (;;) Line 924  for (;;)
924            }            }
925          else left_word = 0;          else left_word = 0;
926    
927          if (clen > 0) right_word = c < 256 && (ctypes[c] & ctype_word) != 0;          if (clen > 0)
928            else right_word = 0;            right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
929            else right_word = 0;
930    
931          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))
932            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 812  for (;;) Line 945  for (;;)
945        if (clen > 0)        if (clen > 0)
946          {          {
947          BOOL OK;          BOOL OK;
948          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
949          switch(code[1])          switch(code[1])
950            {            {
951            case PT_ANY:            case PT_ANY:
# Line 820  for (;;) Line 953  for (;;)
953            break;            break;
954    
955            case PT_LAMP:            case PT_LAMP:
956            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
957            break;            break;
958    
959            case PT_GC:            case PT_GC:
960            OK = category == code[2];            OK = _pcre_ucp_gentype[prop->chartype] == code[2];
961            break;            break;
962    
963            case PT_PC:            case PT_PC:
964            OK = chartype == code[2];            OK = prop->chartype == code[2];
965            break;            break;
966    
967            case PT_SC:            case PT_SC:
968            OK = script == code[2];            OK = prop->script == code[2];
969            break;            break;
970    
971            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 852  for (;;) Line 985  for (;;)
985  /* ========================================================================== */  /* ========================================================================== */
986        /* These opcodes likewise inspect the subject character, but have an        /* These opcodes likewise inspect the subject character, but have an
987        argument that is not a data character. It is one of these opcodes:        argument that is not a data character. It is one of these opcodes:
988        OP_ANY, OP_DIGIT, OP_NOT_DIGIT, OP_WHITESPACE, OP_NOT_SPACE, OP_WORDCHAR,        OP_ANY, OP_ALLANY, OP_DIGIT, OP_NOT_DIGIT, OP_WHITESPACE, OP_NOT_SPACE,
989        OP_NOT_WORDCHAR. The value is loaded into d. */        OP_WORDCHAR, OP_NOT_WORDCHAR. The value is loaded into d. */
990    
991        case OP_TYPEPLUS:        case OP_TYPEPLUS:
992        case OP_TYPEMINPLUS:        case OP_TYPEMINPLUS:
# Line 864  for (;;) Line 997  for (;;)
997          {          {
998          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
999              (c < 256 &&              (c < 256 &&
1000                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1001                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1002            {            {
1003            if (count > 0 && codevalue == OP_TYPEPOSPLUS)            if (count > 0 && codevalue == OP_TYPEPOSPLUS)
# Line 890  for (;;) Line 1020  for (;;)
1020          {          {
1021          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1022              (c < 256 &&              (c < 256 &&
1023                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1024                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1025            {            {
1026            if (codevalue == OP_TYPEPOSQUERY)            if (codevalue == OP_TYPEPOSQUERY)
# Line 915  for (;;) Line 1042  for (;;)
1042          {          {
1043          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1044              (c < 256 &&              (c < 256 &&
1045                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1046                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1047            {            {
1048            if (codevalue == OP_TYPEPOSSTAR)            if (codevalue == OP_TYPEPOSSTAR)
# Line 938  for (;;) Line 1062  for (;;)
1062          {          {
1063          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1064              (c < 256 &&              (c < 256 &&
1065                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1066                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1067            {            {
1068            if (++count >= GET2(code, 1))            if (++count >= GET2(code, 1))
# Line 962  for (;;) Line 1083  for (;;)
1083          {          {
1084          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1085              (c < 256 &&              (c < 256 &&
1086                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1087                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1088            {            {
1089            if (codevalue == OP_TYPEPOSUPTO)            if (codevalue == OP_TYPEPOSUPTO)
# Line 996  for (;;) Line 1114  for (;;)
1114        if (clen > 0)        if (clen > 0)
1115          {          {
1116          BOOL OK;          BOOL OK;
1117          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1118          switch(code[2])          switch(code[2])
1119            {            {
1120            case PT_ANY:            case PT_ANY:
# Line 1004  for (;;) Line 1122  for (;;)
1122            break;            break;
1123    
1124            case PT_LAMP:            case PT_LAMP:
1125            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1126            break;            break;
1127    
1128            case PT_GC:            case PT_GC:
1129            OK = category == code[3];            OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1130            break;            break;
1131    
1132            case PT_PC:            case PT_PC:
1133            OK = chartype == code[3];            OK = prop->chartype == code[3];
1134            break;            break;
1135    
1136            case PT_SC:            case PT_SC:
1137            OK = script == code[3];            OK = prop->script == code[3];
1138            break;            break;
1139    
1140            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1045  for (;;) Line 1163  for (;;)
1163        case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:        case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:
1164        count = current_state->count;  /* Already matched */        count = current_state->count;  /* Already matched */
1165        if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }        if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1166        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1167          {          {
1168          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1169          int ncount = 0;          int ncount = 0;
# Line 1059  for (;;) Line 1177  for (;;)
1177            int nd;            int nd;
1178            int ndlen = 1;            int ndlen = 1;
1179            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1180            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1181            ncount++;            ncount++;
1182            nptr += ndlen;            nptr += ndlen;
1183            }            }
# Line 1218  for (;;) Line 1336  for (;;)
1336        if (clen > 0)        if (clen > 0)
1337          {          {
1338          BOOL OK;          BOOL OK;
1339          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1340          switch(code[2])          switch(code[2])
1341            {            {
1342            case PT_ANY:            case PT_ANY:
# Line 1226  for (;;) Line 1344  for (;;)
1344            break;            break;
1345    
1346            case PT_LAMP:            case PT_LAMP:
1347            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1348            break;            break;
1349    
1350            case PT_GC:            case PT_GC:
1351            OK = category == code[3];            OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1352            break;            break;
1353    
1354            case PT_PC:            case PT_PC:
1355            OK = chartype == code[3];            OK = prop->chartype == code[3];
1356            break;            break;
1357    
1358            case PT_SC:            case PT_SC:
1359            OK = script == code[3];            OK = prop->script == code[3];
1360            break;            break;
1361    
1362            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1276  for (;;) Line 1394  for (;;)
1394        QS2:        QS2:
1395    
1396        ADD_ACTIVE(state_offset + 2, 0);        ADD_ACTIVE(state_offset + 2, 0);
1397        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1398          {          {
1399          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1400          int ncount = 0;          int ncount = 0;
# Line 1291  for (;;) Line 1409  for (;;)
1409            int nd;            int nd;
1410            int ndlen = 1;            int ndlen = 1;
1411            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1412            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1413            ncount++;            ncount++;
1414            nptr += ndlen;            nptr += ndlen;
1415            }            }
# Line 1465  for (;;) Line 1583  for (;;)
1583        if (clen > 0)        if (clen > 0)
1584          {          {
1585          BOOL OK;          BOOL OK;
1586          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1587          switch(code[4])          switch(code[4])
1588            {            {
1589            case PT_ANY:            case PT_ANY:
# Line 1473  for (;;) Line 1591  for (;;)
1591            break;            break;
1592    
1593            case PT_LAMP:            case PT_LAMP:
1594            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1595            break;            break;
1596    
1597            case PT_GC:            case PT_GC:
1598            OK = category == code[5];            OK = _pcre_ucp_gentype[prop->chartype] == code[5];
1599            break;            break;
1600    
1601            case PT_PC:            case PT_PC:
1602            OK = chartype == code[5];            OK = prop->chartype == code[5];
1603            break;            break;
1604    
1605            case PT_SC:            case PT_SC:
1606            OK = script == code[5];            OK = prop->script == code[5];
1607            break;            break;
1608    
1609            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1518  for (;;) Line 1636  for (;;)
1636        if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)        if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)
1637          { ADD_ACTIVE(state_offset + 4, 0); }          { ADD_ACTIVE(state_offset + 4, 0); }
1638        count = current_state->count;  /* Number already matched */        count = current_state->count;  /* Number already matched */
1639        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1640          {          {
1641          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1642          int ncount = 0;          int ncount = 0;
# Line 1532  for (;;) Line 1650  for (;;)
1650            int nd;            int nd;
1651            int ndlen = 1;            int ndlen = 1;
1652            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1653            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1654            ncount++;            ncount++;
1655            nptr += ndlen;            nptr += ndlen;
1656            }            }
# Line 1712  for (;;) Line 1830  for (;;)
1830            other case of the character. */            other case of the character. */
1831    
1832  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1833            othercase = _pcre_ucp_othercase(c);            othercase = UCD_OTHERCASE(c);
1834  #else  #else
1835            othercase = NOTACHAR;            othercase = NOTACHAR;
1836  #endif  #endif
# Line 1737  for (;;) Line 1855  for (;;)
1855        to wait for them to pass before continuing. */        to wait for them to pass before continuing. */
1856    
1857        case OP_EXTUNI:        case OP_EXTUNI:
1858        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1859          {          {
1860          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1861          int ncount = 0;          int ncount = 0;
# Line 1745  for (;;) Line 1863  for (;;)
1863            {            {
1864            int nclen = 1;            int nclen = 1;
1865            GETCHARLEN(c, nptr, nclen);            GETCHARLEN(c, nptr, nclen);
1866            if (_pcre_ucp_findprop(c, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(c) != ucp_M) break;
1867            ncount++;            ncount++;
1868            nptr += nclen;            nptr += nclen;
1869            }            }
# Line 1913  for (;;) Line 2031  for (;;)
2031            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2032              {              {
2033  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2034              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2035  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2036              }              }
2037            else            else
# Line 1951  for (;;) Line 2069  for (;;)
2069            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2070              {              {
2071  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2072              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2073  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2074              }              }
2075            else            else
# Line 1987  for (;;) Line 2105  for (;;)
2105            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2106              {              {
2107  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2108              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2109  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2110              }              }
2111            else            else
# Line 2019  for (;;) Line 2137  for (;;)
2137            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2138              {              {
2139  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2140              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2141  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2142              }              }
2143            else            else
# Line 2054  for (;;) Line 2172  for (;;)
2172            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2173              {              {
2174  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2175              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2176  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2177              }              }
2178            else            else
# Line 2162  for (;;) Line 2280  for (;;)
2280    
2281  /* ========================================================================== */  /* ========================================================================== */
2282        /* These are the opcodes for fancy brackets of various kinds. We have        /* These are the opcodes for fancy brackets of various kinds. We have
2283        to use recursion in order to handle them. */        to use recursion in order to handle them. The "always failing" assertion
2284          (?!) is optimised to OP_FAIL when compiling, so we have to support that,
2285          though the other "backtracking verbs" are not supported. */
2286    
2287          case OP_FAIL:
2288          forced_fail++;    /* Count FAILs for multiple states */
2289          break;
2290    
2291        case OP_ASSERT:        case OP_ASSERT:
2292        case OP_ASSERT_NOT:        case OP_ASSERT_NOT:
# Line 2189  for (;;) Line 2313  for (;;)
2313            rlevel,                               /* function recursion level */            rlevel,                               /* function recursion level */
2314            recursing);                           /* pass on regex recursion */            recursing);                           /* pass on regex recursion */
2315    
2316            if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2317          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))
2318              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
2319          }          }
# Line 2200  for (;;) Line 2325  for (;;)
2325          {          {
2326          int local_offsets[1000];          int local_offsets[1000];
2327          int local_workspace[1000];          int local_workspace[1000];
2328          int condcode = code[LINK_SIZE+1];          int codelink = GET(code, 1);
2329            int condcode;
2330    
2331            /* Because of the way auto-callout works during compile, a callout item
2332            is inserted between OP_COND and an assertion condition. This does not
2333            happen for the other conditions. */
2334    
2335            if (code[LINK_SIZE+1] == OP_CALLOUT)
2336              {
2337              rrc = 0;
2338              if (pcre_callout != NULL)
2339                {
2340                pcre_callout_block cb;
2341                cb.version          = 1;   /* Version 1 of the callout block */
2342                cb.callout_number   = code[LINK_SIZE+2];
2343                cb.offset_vector    = offsets;
2344                cb.subject          = (PCRE_SPTR)start_subject;
2345                cb.subject_length   = end_subject - start_subject;
2346                cb.start_match      = current_subject - start_subject;
2347                cb.current_position = ptr - start_subject;
2348                cb.pattern_position = GET(code, LINK_SIZE + 3);
2349                cb.next_item_length = GET(code, 3 + 2*LINK_SIZE);
2350                cb.capture_top      = 1;
2351                cb.capture_last     = -1;
2352                cb.callout_data     = md->callout_data;
2353                if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc;   /* Abandon */
2354                }
2355              if (rrc > 0) break;                      /* Fail this thread */
2356              code += _pcre_OP_lengths[OP_CALLOUT];    /* Skip callout data */
2357              }
2358    
2359            condcode = code[LINK_SIZE+1];
2360    
2361          /* Back reference conditions are not supported */          /* Back reference conditions are not supported */
2362    
2363          if (condcode == OP_CREF) return PCRE_ERROR_DFA_UCOND;          if (condcode == OP_CREF || condcode == OP_NCREF)
2364              return PCRE_ERROR_DFA_UCOND;
2365    
2366          /* The DEFINE condition is always false */          /* The DEFINE condition is always false */
2367    
2368          if (condcode == OP_DEF)          if (condcode == OP_DEF)
2369            {            { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
           ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0);  
           }  
2370    
2371          /* 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,
2372          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
2373          recursed groups. */          recursed groups. */
2374    
2375          else if (condcode == OP_RREF)          else if (condcode == OP_RREF || condcode == OP_NRREF)
2376            {            {
2377            int value = GET2(code, LINK_SIZE+2);            int value = GET2(code, LINK_SIZE+2);
2378            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;
2379            if (recursing > 0) { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }            if (recursing > 0)
2380              else { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); }              { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }
2381              else { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2382            }            }
2383    
2384          /* Otherwise, the condition is an assertion */          /* Otherwise, the condition is an assertion */
# Line 2248  for (;;) Line 2404  for (;;)
2404              rlevel,                               /* function recursion level */              rlevel,                               /* function recursion level */
2405              recursing);                           /* pass on regex recursion */              recursing);                           /* pass on regex recursion */
2406    
2407              if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2408            if ((rc >= 0) ==            if ((rc >= 0) ==
2409                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))
2410              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
2411            else            else
2412              { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); }              { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2413            }            }
2414          }          }
2415        break;        break;
# Line 2404  for (;;) Line 2561  for (;;)
2561        /* Handle callouts */        /* Handle callouts */
2562    
2563        case OP_CALLOUT:        case OP_CALLOUT:
2564          rrc = 0;
2565        if (pcre_callout != NULL)        if (pcre_callout != NULL)
2566          {          {
         int rrc;  
2567          pcre_callout_block cb;          pcre_callout_block cb;
2568          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 1;   /* Version 1 of the callout block */
2569          cb.callout_number   = code[1];          cb.callout_number   = code[1];
# Line 2421  for (;;) Line 2578  for (;;)
2578          cb.capture_last     = -1;          cb.capture_last     = -1;
2579          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
2580          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); }  
2581          }          }
2582          if (rrc == 0)
2583            { ADD_ACTIVE(state_offset + _pcre_OP_lengths[OP_CALLOUT], 0); }
2584        break;        break;
2585    
2586    
# Line 2438  for (;;) Line 2596  for (;;)
2596    /* We have finished the processing at the current subject character. If no    /* We have finished the processing at the current subject character. If no
2597    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
2598    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
2599    matching has been requested, check for appropriate conditions. */    matching has been requested, check for appropriate conditions.
2600    
2601      The "forced_ fail" variable counts the number of (*F) encountered for the
2602      character. If it is equal to the original active_count (saved in
2603      workspace[1]) it means that (*F) was found on every active state. In this
2604      case we don't want to give a partial match.
2605    
2606      The "could_continue" variable is true if a state could have continued but
2607      for the fact that the end of the subject was reached. */
2608    
2609    if (new_count <= 0)    if (new_count <= 0)
2610      {      {
2611      if (match_count < 0 &&                     /* No matches found */      if (rlevel == 1 &&                               /* Top level, and */
2612          rlevel == 1 &&                         /* Top level match function */          could_continue &&                            /* Some could go on */
2613          (md->moptions & PCRE_PARTIAL) != 0 &&  /* Want partial matching */          forced_fail != workspace[1] &&               /* Not all forced fail & */
2614          ptr >= end_subject &&                  /* Reached end of subject */          (                                            /* either... */
2615          ptr > current_subject)                 /* Matched non-empty string */          (md->moptions & PCRE_PARTIAL_HARD) != 0      /* Hard partial */
2616            ||                                           /* or... */
2617            ((md->moptions & PCRE_PARTIAL_SOFT) != 0 &&  /* Soft partial and */
2618             match_count < 0)                            /* no matches */
2619            ) &&                                         /* And... */
2620            ptr >= end_subject &&                     /* Reached end of subject */
2621            ptr > current_subject)                    /* Matched non-empty string */
2622        {        {
2623        if (offsetcount >= 2)        if (offsetcount >= 2)
2624          {          {
2625          offsets[0] = current_subject - start_subject;          offsets[0] = md->start_used_ptr - start_subject;
2626          offsets[1] = end_subject - start_subject;          offsets[1] = end_subject - start_subject;
2627          }          }
2628        match_count = PCRE_ERROR_PARTIAL;        match_count = PCRE_ERROR_PARTIAL;
# Line 2505  Returns:          > 0 => number of match Line 2677  Returns:          > 0 => number of match
2677                   < -1 => some kind of unexpected problem                   < -1 => some kind of unexpected problem
2678  */  */
2679    
2680  PCRE_EXP_DEFN int  PCRE_EXP_DEFN int PCRE_CALL_CONVENTION
2681  pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,  pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,
2682    const char *subject, int length, int start_offset, int options, int *offsets,    const char *subject, int length, int start_offset, int options, int *offsets,
2683    int offsetcount, int *workspace, int wscount)    int offsetcount, int *workspace, int wscount)
# Line 2592  md->start_code = (const uschar *)argumen Line 2764  md->start_code = (const uschar *)argumen
2764      re->name_table_offset + re->name_count * re->name_entry_size;      re->name_table_offset + re->name_count * re->name_entry_size;
2765  md->start_subject = (const unsigned char *)subject;  md->start_subject = (const unsigned char *)subject;
2766  md->end_subject = end_subject;  md->end_subject = end_subject;
2767    md->start_offset = start_offset;
2768  md->moptions = options;  md->moptions = options;
2769  md->poptions = re->options;  md->poptions = re->options;
2770    
# Line 2604  if ((md->moptions & (PCRE_BSR_ANYCRLF|PC Line 2777  if ((md->moptions & (PCRE_BSR_ANYCRLF|PC
2777      md->moptions |= re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE);      md->moptions |= re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE);
2778  #ifdef BSR_ANYCRLF  #ifdef BSR_ANYCRLF
2779    else md->moptions |= PCRE_BSR_ANYCRLF;    else md->moptions |= PCRE_BSR_ANYCRLF;
2780  #endif  #endif
2781    }    }
2782    
2783  /* Handle different types of newline. The three bits give eight cases. If  /* Handle different types of newline. The three bits give eight cases. If
2784  nothing is set at run time, whatever was used at compile time applies. */  nothing is set at run time, whatever was used at compile time applies. */
# Line 2614  switch ((((options & PCRE_NEWLINE_BITS) Line 2787  switch ((((options & PCRE_NEWLINE_BITS)
2787           PCRE_NEWLINE_BITS)           PCRE_NEWLINE_BITS)
2788    {    {
2789    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Compile-time default */
2790    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = CHAR_CR; break;
2791    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = CHAR_NL; break;
2792    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
2793         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break;
2794    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
2795    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
2796    default: return PCRE_ERROR_BADNEWLINE;    default: return PCRE_ERROR_BADNEWLINE;
# Line 2696  if (!anchored) Line 2869  if (!anchored)
2869      }      }
2870    else    else
2871      {      {
2872      if (startline && study != NULL &&      if (!startline && study != NULL &&
2873           (study->options & PCRE_STUDY_MAPPED) != 0)           (study->flags & PCRE_STUDY_MAPPED) != 0)
2874        start_bits = study->start_bits;        start_bits = study->start_bits;
2875      }      }
2876    }    }
# Line 2713  if ((re->flags & PCRE_REQCHSET) != 0) Line 2886  if ((re->flags & PCRE_REQCHSET) != 0)
2886    }    }
2887    
2888  /* 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
2889  failed match. Unless restarting, optimize by moving to the first match  failed match. If not restarting, perform certain optimizations at the start of
2890  character if possible, when not anchored. Then unless wanting a partial match,  a match. */
 check for a required later character. */  
2891    
2892  for (;;)  for (;;)
2893    {    {
# Line 2725  for (;;) Line 2897  for (;;)
2897      {      {
2898      const uschar *save_end_subject = end_subject;      const uschar *save_end_subject = end_subject;
2899    
2900      /* 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
2901      start of the match is constrained to the first line of a multiline string.      line of a multiline string. Implement this by temporarily adjusting
2902      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
2903      scanning at a newline. If the match fails at the newline, later code breaks      the newline, later code breaks this loop. */
     this loop. */  
2904    
2905      if (firstline)      if (firstline)
2906        {        {
2907        const uschar *t = current_subject;        USPTR t = current_subject;
2908    #ifdef SUPPORT_UTF8
2909          if (utf8)
2910            {
2911            while (t < md->end_subject && !IS_NEWLINE(t))
2912              {
2913              t++;
2914              while (t < end_subject && (*t & 0xc0) == 0x80) t++;
2915              }
2916            }
2917          else
2918    #endif
2919        while (t < md->end_subject && !IS_NEWLINE(t)) t++;        while (t < md->end_subject && !IS_NEWLINE(t)) t++;
2920        end_subject = t;        end_subject = t;
2921        }        }
2922    
2923      if (first_byte >= 0)      /* There are some optimizations that avoid running the match if a known
2924        starting point is not found. However, there is an option that disables
2925        these, for testing and for ensuring that all callouts do actually occur. */
2926    
2927        if ((options & PCRE_NO_START_OPTIMIZE) == 0)
2928        {        {
2929        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++;  
       }  
2930    
2931      /* Or to just after a linebreak for a multiline match if possible */        if (first_byte >= 0)
2932            {
2933            if (first_byte_caseless)
2934              while (current_subject < end_subject &&
2935                     lcc[*current_subject] != first_byte)
2936                current_subject++;
2937            else
2938              while (current_subject < end_subject &&
2939                     *current_subject != first_byte)
2940                current_subject++;
2941            }
2942    
2943      else if (startline)        /* Or to just after a linebreak for a multiline match if possible */
2944        {  
2945        if (current_subject > md->start_subject + start_offset)        else if (startline)
2946          {          {
2947          while (current_subject <= end_subject && !WAS_NEWLINE(current_subject))          if (current_subject > md->start_subject + start_offset)
2948            current_subject++;            {
2949    #ifdef SUPPORT_UTF8
2950              if (utf8)
2951                {
2952                while (current_subject < end_subject &&
2953                       !WAS_NEWLINE(current_subject))
2954                  {
2955                  current_subject++;
2956                  while(current_subject < end_subject &&
2957                        (*current_subject & 0xc0) == 0x80)
2958                    current_subject++;
2959                  }
2960                }
2961              else
2962    #endif
2963              while (current_subject < end_subject && !WAS_NEWLINE(current_subject))
2964                current_subject++;
2965    
2966              /* If we have just passed a CR and the newline option is ANY or
2967              ANYCRLF, and we are now at a LF, advance the match position by one
2968              more character. */
2969    
2970          /* If we have just passed a CR and the newline option is ANY or            if (current_subject[-1] == CHAR_CR &&
2971          ANYCRLF, and we are now at a LF, advance the match position by one more                 (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
2972          character. */                 current_subject < end_subject &&
2973                   *current_subject == CHAR_NL)
2974          if (current_subject[-1] == '\r' &&              current_subject++;
2975               (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&            }
              current_subject < end_subject &&  
              *current_subject == '\n')  
           current_subject++;  
2976          }          }
       }  
2977    
2978      /* Or to a non-unique first char after study */        /* Or to a non-unique first char after study */
2979    
2980      else if (start_bits != NULL)        else if (start_bits != NULL)
       {  
       while (current_subject < end_subject)  
2981          {          {
2982          register unsigned int c = *current_subject;          while (current_subject < end_subject)
2983          if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++;            {
2984            else break;            register unsigned int c = *current_subject;
2985              if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++;
2986                else break;
2987              }
2988          }          }
2989        }        }
2990    
2991      /* Restore fudged end_subject */      /* Restore fudged end_subject */
2992    
2993      end_subject = save_end_subject;      end_subject = save_end_subject;
     }  
2994    
2995    /* If req_byte is set, we know that that character must appear in the subject      /* The following two optimizations are disabled for partial matching or if
2996    for the match to succeed. If the first character is set, req_byte must be      disabling is explicitly requested (and of course, by the test above, this
2997    later in the subject; otherwise the test starts at the match point. This      code is not obeyed when restarting after a partial match). */
   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);  
   
     /* We don't need to repeat the search if we haven't yet reached the  
     place we found it at last time. */  
2998    
2999      if (p > req_byte_ptr)      if ((options & PCRE_NO_START_OPTIMIZE) == 0 &&
3000            (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT)) == 0)
3001        {        {
3002        if (req_byte_caseless)        /* If the pattern was studied, a minimum subject length may be set. This
3003          {        is a lower bound; no actual string of that length may actually match the
3004          while (p < end_subject)        pattern. Although the value is, strictly, in characters, we treat it as
3005            {        bytes to avoid spending too much time in this optimization. */
3006            register int pp = *p++;  
3007            if (pp == req_byte || pp == req_byte2) { p--; break; }        if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 &&
3008            }            (pcre_uint32)(end_subject - current_subject) < study->minlength)
3009          }          return PCRE_ERROR_NOMATCH;
3010        else  
3011          /* If req_byte is set, we know that that character must appear in the
3012          subject for the match to succeed. If the first character is set, req_byte
3013          must be later in the subject; otherwise the test starts at the match
3014          point. This optimization can save a huge amount of work in patterns with
3015          nested unlimited repeats that aren't going to match. Writing separate
3016          code for cased/caseless versions makes it go faster, as does using an
3017          autoincrement and backing off on a match.
3018    
3019          HOWEVER: when the subject string is very, very long, searching to its end
3020          can take a long time, and give bad performance on quite ordinary
3021          patterns. This showed up when somebody was matching /^C/ on a 32-megabyte
3022          string... so we don't do this when the string is sufficiently long. */
3023    
3024          if (req_byte >= 0 && end_subject - current_subject < REQ_BYTE_MAX)
3025          {          {
3026          while (p < end_subject)          register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);
3027    
3028            /* We don't need to repeat the search if we haven't yet reached the
3029            place we found it at last time. */
3030    
3031            if (p > req_byte_ptr)
3032            {            {
3033            if (*p++ == req_byte) { p--; break; }            if (req_byte_caseless)
3034            }              {
3035          }              while (p < end_subject)
3036                  {
3037                  register int pp = *p++;
3038                  if (pp == req_byte || pp == req_byte2) { p--; break; }
3039                  }
3040                }
3041              else
3042                {
3043                while (p < end_subject)
3044                  {
3045                  if (*p++ == req_byte) { p--; break; }
3046                  }
3047                }
3048    
3049        /* If we can't find the required character, break the matching loop,            /* If we can't find the required character, break the matching loop,
3050        which will cause a return or PCRE_ERROR_NOMATCH. */            which will cause a return or PCRE_ERROR_NOMATCH. */
3051    
3052        if (p >= end_subject) break;            if (p >= end_subject) break;
3053    
3054        /* If we have found the required character, save the point where we            /* If we have found the required character, save the point where we
3055        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
3056        the start hasn't passed this character yet. */            the start hasn't passed this character yet. */
3057    
3058        req_byte_ptr = p;            req_byte_ptr = p;
3059              }
3060            }
3061        }        }
3062      }      }   /* End of optimizations that are done when not restarting */
3063    
3064    /* OK, now we can do the business */    /* OK, now we can do the business */
3065    
3066      md->start_used_ptr = current_subject;
3067    
3068    rc = internal_dfa_exec(    rc = internal_dfa_exec(
3069      md,                                /* fixed match data */      md,                                /* fixed match data */
3070      md->start_code,                    /* this subexpression's code */      md->start_code,                    /* this subexpression's code */
# Line 2879  for (;;) Line 3099  for (;;)
3099    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
3100    or ANY or ANYCRLF, advance the match position by one more character. */    or ANY or ANYCRLF, advance the match position by one more character. */
3101    
3102    if (current_subject[-1] == '\r' &&    if (current_subject[-1] == CHAR_CR &&
3103        current_subject < end_subject &&        current_subject < end_subject &&
3104        *current_subject == '\n' &&        *current_subject == CHAR_NL &&
3105        (re->flags & PCRE_HASCRORLF) == 0 &&        (re->flags & PCRE_HASCRORLF) == 0 &&
3106          (md->nltype == NLTYPE_ANY ||          (md->nltype == NLTYPE_ANY ||
3107           md->nltype == NLTYPE_ANYCRLF ||           md->nltype == NLTYPE_ANYCRLF ||

Legend:
Removed from v.236  
changed lines
  Added in v.498

  ViewVC Help
Powered by ViewVC 1.1.5