/[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 199 by ph10, Tue Jul 31 14:39:09 2007 UTC revision 473 by ph10, Sat Jan 2 12:40:07 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  #ifdef HAVE_CONFIG_H  /* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved
49  #include <config.h>  the performance of his patterns greatly. I could not use it as it stood, as it
50  #endif  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
77    #include "config.h"
78    #endif
79    
80  #define NLBLOCK md             /* Block containing newline information */  #define NLBLOCK md             /* Block containing newline information */
81  #define PSSTART start_subject  /* Field containing processed string start */  #define PSSTART start_subject  /* Field containing processed string start */
# 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, 0,                       /* NOTPROP, PROP, EXTUNI                  */
122    0, 0, 0, 0, 0,                 /* \R, \H, \h, \V, \v                     */    0, 0, 0, 0, 0,                 /* \R, \H, \h, \V, \v                     */
123    0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */    0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */
# Line 130  static uschar coptable[] = { Line 159  static uschar coptable[] = {
159    0,                             /* CREF                                   */    0,                             /* CREF                                   */
160    0,                             /* RREF                                   */    0,                             /* RREF                                   */
161    0,                             /* DEF                                    */    0,                             /* DEF                                    */
162    0, 0                           /* BRAZERO, BRAMINZERO                    */    0, 0,                          /* BRAZERO, BRAMINZERO                    */
163      0, 0, 0, 0,                    /* PRUNE, SKIP, THEN, COMMIT              */
164      0, 0, 0, 0                     /* FAIL, ACCEPT, CLOSE, SKIPZERO          */
165    };
166    
167    /* This table identifies those opcodes that inspect a character. It is used to
168    remember the fact that a character could have been inspected when the end of
169    the subject is reached. ***NOTE*** If the start of this table is modified, the
170    two tables that follow must also be modified. */
171    
172    static const uschar poptable[] = {
173      0,                             /* End                                    */
174      0, 0, 0, 1, 1,                 /* \A, \G, \K, \B, \b                     */
175      1, 1, 1, 1, 1, 1,              /* \D, \d, \S, \s, \W, \w                 */
176      1, 1, 1,                       /* Any, AllAny, Anybyte                   */
177      1, 1, 1,                       /* NOTPROP, PROP, EXTUNI                  */
178      1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */
179      0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */
180      1,                             /* Char                                   */
181      1,                             /* Charnc                                 */
182      1,                             /* not                                    */
183      /* Positive single-char repeats                                          */
184      1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */
185      1, 1, 1,                       /* upto, minupto, exact                   */
186      1, 1, 1, 1,                    /* *+, ++, ?+, upto+                      */
187      /* Negative single-char repeats - only for chars < 256                   */
188      1, 1, 1, 1, 1, 1,              /* NOT *, *?, +, +?, ?, ??                */
189      1, 1, 1,                       /* NOT upto, minupto, exact               */
190      1, 1, 1, 1,                    /* NOT *+, ++, ?+, upto+                  */
191      /* Positive type repeats                                                 */
192      1, 1, 1, 1, 1, 1,              /* Type *, *?, +, +?, ?, ??               */
193      1, 1, 1,                       /* Type upto, minupto, exact              */
194      1, 1, 1, 1,                    /* Type *+, ++, ?+, upto+                 */
195      /* Character class & ref repeats                                         */
196      1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */
197      1, 1,                          /* CRRANGE, CRMINRANGE                    */
198      1,                             /* CLASS                                  */
199      1,                             /* NCLASS                                 */
200      1,                             /* XCLASS - variable length               */
201      0,                             /* REF                                    */
202      0,                             /* RECURSE                                */
203      0,                             /* CALLOUT                                */
204      0,                             /* Alt                                    */
205      0,                             /* Ket                                    */
206      0,                             /* KetRmax                                */
207      0,                             /* KetRmin                                */
208      0,                             /* Assert                                 */
209      0,                             /* Assert not                             */
210      0,                             /* Assert behind                          */
211      0,                             /* Assert behind not                      */
212      0,                             /* Reverse                                */
213      0, 0, 0, 0,                    /* ONCE, BRA, CBRA, COND                  */
214      0, 0, 0,                       /* SBRA, SCBRA, SCOND                     */
215      0,                             /* CREF                                   */
216      0,                             /* RREF                                   */
217      0,                             /* DEF                                    */
218      0, 0,                          /* BRAZERO, BRAMINZERO                    */
219      0, 0, 0, 0,                    /* PRUNE, SKIP, THEN, COMMIT              */
220      0, 0, 0, 0                     /* FAIL, ACCEPT, CLOSE, SKIPZERO          */
221  };  };
222    
223  /* 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,
224  and \w */  and \w */
225    
226  static uschar toptable1[] = {  static const uschar toptable1[] = {
227    0, 0, 0, 0, 0, 0,    0, 0, 0, 0, 0, 0,
228    ctype_digit, ctype_digit,    ctype_digit, ctype_digit,
229    ctype_space, ctype_space,    ctype_space, ctype_space,
230    ctype_word,  ctype_word,    ctype_word,  ctype_word,
231    0                               /* OP_ANY */    0, 0                            /* OP_ANY, OP_ALLANY */
232  };  };
233    
234  static uschar toptable2[] = {  static const uschar toptable2[] = {
235    0, 0, 0, 0, 0, 0,    0, 0, 0, 0, 0, 0,
236    ctype_digit, 0,    ctype_digit, 0,
237    ctype_space, 0,    ctype_space, 0,
238    ctype_word,  0,    ctype_word,  0,
239    1                               /* OP_ANY */    1, 1                            /* OP_ANY, OP_ALLANY */
240  };  };
241    
242    
# Line 221  Arguments: Line 308  Arguments:
308    rlevel            function call recursion level    rlevel            function call recursion level
309    recursing         regex recursive call level    recursing         regex recursive call level
310    
311  Returns:            > 0 =>  Returns:            > 0 => number of match offset pairs placed in offsets
312                      = 0 =>                      = 0 => offsets overflowed; longest matches are present
313                       -1 => failed to match                       -1 => failed to match
314                     < -1 => some kind of unexpected problem                     < -1 => some kind of unexpected problem
315    
# Line 388  if (*first_op == OP_REVERSE) Line 475  if (*first_op == OP_REVERSE)
475      current_subject -= gone_back;      current_subject -= gone_back;
476      }      }
477    
478      /* Save the earliest consulted character */
479    
480      if (current_subject < md->start_used_ptr)
481        md->start_used_ptr = current_subject;
482    
483    /* Now we can process the individual branches. */    /* Now we can process the individual branches. */
484    
485    end_code = this_start_code;    end_code = this_start_code;
# Line 452  for (;;) Line 544  for (;;)
544    int i, j;    int i, j;
545    int clen, dlen;    int clen, dlen;
546    unsigned int c, d;    unsigned int c, d;
547      int forced_fail = 0;
548      BOOL could_continue = FALSE;
549    
550    /* 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
551    new state list. */    new state list. */
# Line 509  for (;;) Line 603  for (;;)
603      stateblock *current_state = active_states + i;      stateblock *current_state = active_states + i;
604      const uschar *code;      const uschar *code;
605      int state_offset = current_state->offset;      int state_offset = current_state->offset;
606      int count, codevalue;      int count, codevalue, rrc;
 #ifdef SUPPORT_UCP  
     int chartype, script;  
 #endif  
607    
608  #ifdef DEBUG  #ifdef DEBUG
609      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);
# Line 544  for (;;) Line 635  for (;;)
635          }          }
636        }        }
637    
638      /* 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.
639        See the note at the head of this module about the possibility of improving
640        performance here. */
641    
642      for (j = 0; j < i; j++)      for (j = 0; j < i; j++)
643        {        {
# Line 561  for (;;) Line 654  for (;;)
654      code = start_code + state_offset;      code = start_code + state_offset;
655      codevalue = *code;      codevalue = *code;
656    
657        /* If this opcode inspects a character, but we are at the end of the
658        subject, remember the fact for use when testing for a partial match. */
659    
660        if (clen == 0 && poptable[codevalue] != 0)
661          could_continue = TRUE;
662    
663      /* 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
664      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
665      is wrong, because sometimes zero repetitions of the subject are      is wrong, because sometimes zero repetitions of the subject are
# Line 611  for (;;) Line 710  for (;;)
710  /* ========================================================================== */  /* ========================================================================== */
711        /* 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
712        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
713        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
714          start of the subject, save the match data, shifting up all previous
715        matches so we always have the longest first. */        matches so we always have the longest first. */
716    
717        case OP_KET:        case OP_KET:
# Line 625  for (;;) Line 725  for (;;)
725            ADD_ACTIVE(state_offset - GET(code, 1), 0);            ADD_ACTIVE(state_offset - GET(code, 1), 0);
726            }            }
727          }          }
728        else if (ptr > current_subject || (md->moptions & PCRE_NOTEMPTY) == 0)        else
729          {          {
730          if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;          if (ptr > current_subject ||
731            else if (match_count > 0 && ++match_count * 2 >= offsetcount)              ((md->moptions & PCRE_NOTEMPTY) == 0 &&
732              match_count = 0;                ((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 ||
733          count = ((match_count == 0)? offsetcount : match_count * 2) - 2;                  current_subject > start_subject + md->start_offset)))
734          if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));            {
735          if (offsetcount >= 2)            if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
736            {              else if (match_count > 0 && ++match_count * 2 >= offsetcount)
737            offsets[0] = current_subject - start_subject;                match_count = 0;
738            offsets[1] = ptr - start_subject;            count = ((match_count == 0)? offsetcount : match_count * 2) - 2;
739            DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,            if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));
740              offsets[1] - offsets[0], current_subject));            if (offsetcount >= 2)
741            }              {
742          if ((md->moptions & PCRE_DFA_SHORTEST) != 0)              offsets[0] = current_subject - start_subject;
743            {              offsets[1] = ptr - start_subject;
744            DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"              DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,
745              "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,                offsets[1] - offsets[0], current_subject));
746              match_count, rlevel*2-2, SP));              }
747            return match_count;            if ((md->moptions & PCRE_DFA_SHORTEST) != 0)
748                {
749                DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"
750                  "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,
751                  match_count, rlevel*2-2, SP));
752                return match_count;
753                }
754            }            }
755          }          }
756        break;        break;
# Line 692  for (;;) Line 798  for (;;)
798        break;        break;
799    
800        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
801          case OP_SKIPZERO:
802          code += 1 + GET(code, 2);
803          while (*code == OP_ALT) code += GET(code, 1);
804          ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0);
805          break;
806    
807          /*-----------------------------------------------------------------*/
808        case OP_CIRC:        case OP_CIRC:
809        if ((ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0) ||        if ((ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0) ||
810            ((ims & PCRE_MULTILINE) != 0 &&            ((ims & PCRE_MULTILINE) != 0 &&
# Line 730  for (;;) Line 843  for (;;)
843    
844        /*-----------------------------------------------------------------*/        /*-----------------------------------------------------------------*/
845        case OP_ANY:        case OP_ANY:
846        if (clen > 0 && ((ims & PCRE_DOTALL) != 0 || !IS_NEWLINE(ptr)))        if (clen > 0 && !IS_NEWLINE(ptr))
847            { ADD_NEW(state_offset + 1, 0); }
848          break;
849    
850          /*-----------------------------------------------------------------*/
851          case OP_ALLANY:
852          if (clen > 0)
853          { ADD_NEW(state_offset + 1, 0); }          { ADD_NEW(state_offset + 1, 0); }
854        break;        break;
855    
# Line 745  for (;;) Line 864  for (;;)
864        if ((md->moptions & PCRE_NOTEOL) == 0)        if ((md->moptions & PCRE_NOTEOL) == 0)
865          {          {
866          if (clen == 0 ||          if (clen == 0 ||
867              (IS_NEWLINE(ptr) &&              ((md->poptions & PCRE_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr) &&
868                 ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen)                 ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen)
869              ))              ))
870            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 782  for (;;) Line 901  for (;;)
901          if (ptr > start_subject)          if (ptr > start_subject)
902            {            {
903            const uschar *temp = ptr - 1;            const uschar *temp = ptr - 1;
904              if (temp < md->start_used_ptr) md->start_used_ptr = temp;
905  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
906            if (utf8) BACKCHAR(temp);            if (utf8) BACKCHAR(temp);
907  #endif  #endif
# Line 790  for (;;) Line 910  for (;;)
910            }            }
911          else left_word = 0;          else left_word = 0;
912    
913          if (clen > 0) right_word = c < 256 && (ctypes[c] & ctype_word) != 0;          if (clen > 0)
914            else right_word = 0;            right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
915            else right_word = 0;
916    
917          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))
918            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 810  for (;;) Line 931  for (;;)
931        if (clen > 0)        if (clen > 0)
932          {          {
933          BOOL OK;          BOOL OK;
934          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
935          switch(code[1])          switch(code[1])
936            {            {
937            case PT_ANY:            case PT_ANY:
# Line 818  for (;;) Line 939  for (;;)
939            break;            break;
940    
941            case PT_LAMP:            case PT_LAMP:
942            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
943            break;            break;
944    
945            case PT_GC:            case PT_GC:
946            OK = category == code[2];            OK = _pcre_ucp_gentype[prop->chartype] == code[2];
947            break;            break;
948    
949            case PT_PC:            case PT_PC:
950            OK = chartype == code[2];            OK = prop->chartype == code[2];
951            break;            break;
952    
953            case PT_SC:            case PT_SC:
954            OK = script == code[2];            OK = prop->script == code[2];
955            break;            break;
956    
957            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 850  for (;;) Line 971  for (;;)
971  /* ========================================================================== */  /* ========================================================================== */
972        /* These opcodes likewise inspect the subject character, but have an        /* These opcodes likewise inspect the subject character, but have an
973        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:
974        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,
975        OP_NOT_WORDCHAR. The value is loaded into d. */        OP_WORDCHAR, OP_NOT_WORDCHAR. The value is loaded into d. */
976    
977        case OP_TYPEPLUS:        case OP_TYPEPLUS:
978        case OP_TYPEMINPLUS:        case OP_TYPEMINPLUS:
# Line 862  for (;;) Line 983  for (;;)
983          {          {
984          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
985              (c < 256 &&              (c < 256 &&
986                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
987                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
988            {            {
989            if (count > 0 && codevalue == OP_TYPEPOSPLUS)            if (count > 0 && codevalue == OP_TYPEPOSPLUS)
# Line 888  for (;;) Line 1006  for (;;)
1006          {          {
1007          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1008              (c < 256 &&              (c < 256 &&
1009                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1010                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1011            {            {
1012            if (codevalue == OP_TYPEPOSQUERY)            if (codevalue == OP_TYPEPOSQUERY)
# Line 913  for (;;) Line 1028  for (;;)
1028          {          {
1029          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1030              (c < 256 &&              (c < 256 &&
1031                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1032                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1033            {            {
1034            if (codevalue == OP_TYPEPOSSTAR)            if (codevalue == OP_TYPEPOSSTAR)
# Line 936  for (;;) Line 1048  for (;;)
1048          {          {
1049          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1050              (c < 256 &&              (c < 256 &&
1051                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1052                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1053            {            {
1054            if (++count >= GET2(code, 1))            if (++count >= GET2(code, 1))
# Line 960  for (;;) Line 1069  for (;;)
1069          {          {
1070          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||          if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1071              (c < 256 &&              (c < 256 &&
1072                (d != OP_ANY ||                (d != OP_ANY || !IS_NEWLINE(ptr)) &&
                (ims & PCRE_DOTALL) != 0 ||  
                !IS_NEWLINE(ptr)  
               ) &&  
1073                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))                ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1074            {            {
1075            if (codevalue == OP_TYPEPOSUPTO)            if (codevalue == OP_TYPEPOSUPTO)
# Line 994  for (;;) Line 1100  for (;;)
1100        if (clen > 0)        if (clen > 0)
1101          {          {
1102          BOOL OK;          BOOL OK;
1103          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1104          switch(code[2])          switch(code[2])
1105            {            {
1106            case PT_ANY:            case PT_ANY:
# Line 1002  for (;;) Line 1108  for (;;)
1108            break;            break;
1109    
1110            case PT_LAMP:            case PT_LAMP:
1111            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1112            break;            break;
1113    
1114            case PT_GC:            case PT_GC:
1115            OK = category == code[3];            OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1116            break;            break;
1117    
1118            case PT_PC:            case PT_PC:
1119            OK = chartype == code[3];            OK = prop->chartype == code[3];
1120            break;            break;
1121    
1122            case PT_SC:            case PT_SC:
1123            OK = script == code[3];            OK = prop->script == code[3];
1124            break;            break;
1125    
1126            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1043  for (;;) Line 1149  for (;;)
1149        case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:        case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:
1150        count = current_state->count;  /* Already matched */        count = current_state->count;  /* Already matched */
1151        if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }        if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1152        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1153          {          {
1154          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1155          int ncount = 0;          int ncount = 0;
# Line 1057  for (;;) Line 1163  for (;;)
1163            int nd;            int nd;
1164            int ndlen = 1;            int ndlen = 1;
1165            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1166            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1167            ncount++;            ncount++;
1168            nptr += ndlen;            nptr += ndlen;
1169            }            }
# Line 1078  for (;;) Line 1184  for (;;)
1184          int ncount = 0;          int ncount = 0;
1185          switch (c)          switch (c)
1186            {            {
           case 0x000d:  
           if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;  
           /* Fall through */  
           case 0x000a:  
1187            case 0x000b:            case 0x000b:
1188            case 0x000c:            case 0x000c:
1189            case 0x0085:            case 0x0085:
1190            case 0x2028:            case 0x2028:
1191            case 0x2029:            case 0x2029:
1192              if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break;
1193              goto ANYNL01;
1194    
1195              case 0x000d:
1196              if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;
1197              /* Fall through */
1198    
1199              ANYNL01:
1200              case 0x000a:
1201            if (count > 0 && codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSPLUS)            if (count > 0 && codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSPLUS)
1202              {              {
1203              active_count--;           /* Remove non-match possibility */              active_count--;           /* Remove non-match possibility */
# Line 1095  for (;;) Line 1206  for (;;)
1206            count++;            count++;
1207            ADD_NEW_DATA(-state_offset, count, ncount);            ADD_NEW_DATA(-state_offset, count, ncount);
1208            break;            break;
1209    
1210            default:            default:
1211            break;            break;
1212            }            }
# Line 1210  for (;;) Line 1322  for (;;)
1322        if (clen > 0)        if (clen > 0)
1323          {          {
1324          BOOL OK;          BOOL OK;
1325          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1326          switch(code[2])          switch(code[2])
1327            {            {
1328            case PT_ANY:            case PT_ANY:
# Line 1218  for (;;) Line 1330  for (;;)
1330            break;            break;
1331    
1332            case PT_LAMP:            case PT_LAMP:
1333            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1334            break;            break;
1335    
1336            case PT_GC:            case PT_GC:
1337            OK = category == code[3];            OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1338            break;            break;
1339    
1340            case PT_PC:            case PT_PC:
1341            OK = chartype == code[3];            OK = prop->chartype == code[3];
1342            break;            break;
1343    
1344            case PT_SC:            case PT_SC:
1345            OK = script == code[3];            OK = prop->script == code[3];
1346            break;            break;
1347    
1348            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1268  for (;;) Line 1380  for (;;)
1380        QS2:        QS2:
1381    
1382        ADD_ACTIVE(state_offset + 2, 0);        ADD_ACTIVE(state_offset + 2, 0);
1383        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1384          {          {
1385          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1386          int ncount = 0;          int ncount = 0;
# Line 1283  for (;;) Line 1395  for (;;)
1395            int nd;            int nd;
1396            int ndlen = 1;            int ndlen = 1;
1397            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1398            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1399            ncount++;            ncount++;
1400            nptr += ndlen;            nptr += ndlen;
1401            }            }
# Line 1311  for (;;) Line 1423  for (;;)
1423          int ncount = 0;          int ncount = 0;
1424          switch (c)          switch (c)
1425            {            {
           case 0x000d:  
           if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;  
           /* Fall through */  
           case 0x000a:  
1426            case 0x000b:            case 0x000b:
1427            case 0x000c:            case 0x000c:
1428            case 0x0085:            case 0x0085:
1429            case 0x2028:            case 0x2028:
1430            case 0x2029:            case 0x2029:
1431              if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break;
1432              goto ANYNL02;
1433    
1434              case 0x000d:
1435              if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;
1436              /* Fall through */
1437    
1438              ANYNL02:
1439              case 0x000a:
1440            if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSSTAR ||            if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSSTAR ||
1441                codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSQUERY)                codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSQUERY)
1442              {              {
# Line 1328  for (;;) Line 1445  for (;;)
1445              }              }
1446            ADD_NEW_DATA(-(state_offset + count), 0, ncount);            ADD_NEW_DATA(-(state_offset + count), 0, ncount);
1447            break;            break;
1448    
1449            default:            default:
1450            break;            break;
1451            }            }
# Line 1451  for (;;) Line 1569  for (;;)
1569        if (clen > 0)        if (clen > 0)
1570          {          {
1571          BOOL OK;          BOOL OK;
1572          int category = _pcre_ucp_findprop(c, &chartype, &script);          const ucd_record * prop = GET_UCD(c);
1573          switch(code[4])          switch(code[4])
1574            {            {
1575            case PT_ANY:            case PT_ANY:
# Line 1459  for (;;) Line 1577  for (;;)
1577            break;            break;
1578    
1579            case PT_LAMP:            case PT_LAMP:
1580            OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;
1581            break;            break;
1582    
1583            case PT_GC:            case PT_GC:
1584            OK = category == code[5];            OK = _pcre_ucp_gentype[prop->chartype] == code[5];
1585            break;            break;
1586    
1587            case PT_PC:            case PT_PC:
1588            OK = chartype == code[5];            OK = prop->chartype == code[5];
1589            break;            break;
1590    
1591            case PT_SC:            case PT_SC:
1592            OK = script == code[5];            OK = prop->script == code[5];
1593            break;            break;
1594    
1595            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
# Line 1504  for (;;) Line 1622  for (;;)
1622        if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)        if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)
1623          { ADD_ACTIVE(state_offset + 4, 0); }          { ADD_ACTIVE(state_offset + 4, 0); }
1624        count = current_state->count;  /* Number already matched */        count = current_state->count;  /* Number already matched */
1625        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1626          {          {
1627          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1628          int ncount = 0;          int ncount = 0;
# Line 1518  for (;;) Line 1636  for (;;)
1636            int nd;            int nd;
1637            int ndlen = 1;            int ndlen = 1;
1638            GETCHARLEN(nd, nptr, ndlen);            GETCHARLEN(nd, nptr, ndlen);
1639            if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(nd) != ucp_M) break;
1640            ncount++;            ncount++;
1641            nptr += ndlen;            nptr += ndlen;
1642            }            }
# Line 1543  for (;;) Line 1661  for (;;)
1661          int ncount = 0;          int ncount = 0;
1662          switch (c)          switch (c)
1663            {            {
           case 0x000d:  
           if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;  
           /* Fall through */  
           case 0x000a:  
1664            case 0x000b:            case 0x000b:
1665            case 0x000c:            case 0x000c:
1666            case 0x0085:            case 0x0085:
1667            case 0x2028:            case 0x2028:
1668            case 0x2029:            case 0x2029:
1669              if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break;
1670              goto ANYNL03;
1671    
1672              case 0x000d:
1673              if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;
1674              /* Fall through */
1675    
1676              ANYNL03:
1677              case 0x000a:
1678            if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSUPTO)            if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSUPTO)
1679              {              {
1680              active_count--;           /* Remove non-match possibility */              active_count--;           /* Remove non-match possibility */
# Line 1562  for (;;) Line 1685  for (;;)
1685            else            else
1686              { ADD_NEW_DATA(-state_offset, count, ncount); }              { ADD_NEW_DATA(-state_offset, count, ncount); }
1687            break;            break;
1688    
1689            default:            default:
1690            break;            break;
1691            }            }
# Line 1692  for (;;) Line 1816  for (;;)
1816            other case of the character. */            other case of the character. */
1817    
1818  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1819            othercase = _pcre_ucp_othercase(c);            othercase = UCD_OTHERCASE(c);
1820  #else  #else
1821            othercase = NOTACHAR;            othercase = NOTACHAR;
1822  #endif  #endif
# Line 1717  for (;;) Line 1841  for (;;)
1841        to wait for them to pass before continuing. */        to wait for them to pass before continuing. */
1842    
1843        case OP_EXTUNI:        case OP_EXTUNI:
1844        if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M)        if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1845          {          {
1846          const uschar *nptr = ptr + clen;          const uschar *nptr = ptr + clen;
1847          int ncount = 0;          int ncount = 0;
# Line 1725  for (;;) Line 1849  for (;;)
1849            {            {
1850            int nclen = 1;            int nclen = 1;
1851            GETCHARLEN(c, nptr, nclen);            GETCHARLEN(c, nptr, nclen);
1852            if (_pcre_ucp_findprop(c, &chartype, &script) != ucp_M) break;            if (UCD_CATEGORY(c) != ucp_M) break;
1853            ncount++;            ncount++;
1854            nptr += nclen;            nptr += nclen;
1855            }            }
# Line 1742  for (;;) Line 1866  for (;;)
1866        case OP_ANYNL:        case OP_ANYNL:
1867        if (clen > 0) switch(c)        if (clen > 0) switch(c)
1868          {          {
         case 0x000a:  
1869          case 0x000b:          case 0x000b:
1870          case 0x000c:          case 0x000c:
1871          case 0x0085:          case 0x0085:
1872          case 0x2028:          case 0x2028:
1873          case 0x2029:          case 0x2029:
1874            if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break;
1875    
1876            case 0x000a:
1877          ADD_NEW(state_offset + 1, 0);          ADD_NEW(state_offset + 1, 0);
1878          break;          break;
1879    
1880          case 0x000d:          case 0x000d:
1881          if (ptr + 1 < end_subject && ptr[1] == 0x0a)          if (ptr + 1 < end_subject && ptr[1] == 0x0a)
1882            {            {
# Line 1890  for (;;) Line 2017  for (;;)
2017            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2018              {              {
2019  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2020              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2021  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2022              }              }
2023            else            else
# Line 1928  for (;;) Line 2055  for (;;)
2055            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2056              {              {
2057  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2058              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2059  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2060              }              }
2061            else            else
# Line 1964  for (;;) Line 2091  for (;;)
2091            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2092              {              {
2093  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2094              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2095  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2096              }              }
2097            else            else
# Line 1996  for (;;) Line 2123  for (;;)
2123            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2124              {              {
2125  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2126              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2127  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2128              }              }
2129            else            else
# Line 2031  for (;;) Line 2158  for (;;)
2158            if (utf8 && d >= 128)            if (utf8 && d >= 128)
2159              {              {
2160  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2161              otherd = _pcre_ucp_othercase(d);              otherd = UCD_OTHERCASE(d);
2162  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2163              }              }
2164            else            else
# Line 2139  for (;;) Line 2266  for (;;)
2266    
2267  /* ========================================================================== */  /* ========================================================================== */
2268        /* These are the opcodes for fancy brackets of various kinds. We have        /* These are the opcodes for fancy brackets of various kinds. We have
2269        to use recursion in order to handle them. */        to use recursion in order to handle them. The "always failing" assertion
2270          (?!) is optimised to OP_FAIL when compiling, so we have to support that,
2271          though the other "backtracking verbs" are not supported. */
2272    
2273          case OP_FAIL:
2274          forced_fail++;    /* Count FAILs for multiple states */
2275          break;
2276    
2277        case OP_ASSERT:        case OP_ASSERT:
2278        case OP_ASSERT_NOT:        case OP_ASSERT_NOT:
# Line 2165  for (;;) Line 2298  for (;;)
2298            ims,                                  /* the current ims flags */            ims,                                  /* the current ims flags */
2299            rlevel,                               /* function recursion level */            rlevel,                               /* function recursion level */
2300            recursing);                           /* pass on regex recursion */            recursing);                           /* pass on regex recursion */
2301    
2302            if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2303          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))
2304              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
2305          }          }
# Line 2177  for (;;) Line 2311  for (;;)
2311          {          {
2312          int local_offsets[1000];          int local_offsets[1000];
2313          int local_workspace[1000];          int local_workspace[1000];
2314          int condcode = code[LINK_SIZE+1];          int codelink = GET(code, 1);
2315            int condcode;
2316    
2317            /* Because of the way auto-callout works during compile, a callout item
2318            is inserted between OP_COND and an assertion condition. This does not
2319            happen for the other conditions. */
2320    
2321            if (code[LINK_SIZE+1] == OP_CALLOUT)
2322              {
2323              rrc = 0;
2324              if (pcre_callout != NULL)
2325                {
2326                pcre_callout_block cb;
2327                cb.version          = 1;   /* Version 1 of the callout block */
2328                cb.callout_number   = code[LINK_SIZE+2];
2329                cb.offset_vector    = offsets;
2330                cb.subject          = (PCRE_SPTR)start_subject;
2331                cb.subject_length   = end_subject - start_subject;
2332                cb.start_match      = current_subject - start_subject;
2333                cb.current_position = ptr - start_subject;
2334                cb.pattern_position = GET(code, LINK_SIZE + 3);
2335                cb.next_item_length = GET(code, 3 + 2*LINK_SIZE);
2336                cb.capture_top      = 1;
2337                cb.capture_last     = -1;
2338                cb.callout_data     = md->callout_data;
2339                if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc;   /* Abandon */
2340                }
2341              if (rrc > 0) break;                      /* Fail this thread */
2342              code += _pcre_OP_lengths[OP_CALLOUT];    /* Skip callout data */
2343              }
2344    
2345            condcode = code[LINK_SIZE+1];
2346    
2347          /* Back reference conditions are not supported */          /* Back reference conditions are not supported */
2348    
2349          if (condcode == OP_CREF) return PCRE_ERROR_DFA_UCOND;          if (condcode == OP_CREF || condcode == OP_NCREF)
2350              return PCRE_ERROR_DFA_UCOND;
2351    
2352          /* The DEFINE condition is always false */          /* The DEFINE condition is always false */
2353    
2354          if (condcode == OP_DEF)          if (condcode == OP_DEF)
2355            {            { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
           ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0);  
           }  
2356    
2357          /* 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,
2358          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
2359          recursed groups. */          recursed groups. */
2360    
2361          else if (condcode == OP_RREF)          else if (condcode == OP_RREF || condcode == OP_NRREF)
2362            {            {
2363            int value = GET2(code, LINK_SIZE+2);            int value = GET2(code, LINK_SIZE+2);
2364            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;
2365            if (recursing > 0) { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }            if (recursing > 0)
2366              else { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); }              { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }
2367              else { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2368            }            }
2369    
2370          /* Otherwise, the condition is an assertion */          /* Otherwise, the condition is an assertion */
# Line 2225  for (;;) Line 2390  for (;;)
2390              rlevel,                               /* function recursion level */              rlevel,                               /* function recursion level */
2391              recursing);                           /* pass on regex recursion */              recursing);                           /* pass on regex recursion */
2392    
2393              if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2394            if ((rc >= 0) ==            if ((rc >= 0) ==
2395                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))
2396              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
2397            else            else
2398              { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); }              { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2399            }            }
2400          }          }
2401        break;        break;
# Line 2381  for (;;) Line 2547  for (;;)
2547        /* Handle callouts */        /* Handle callouts */
2548    
2549        case OP_CALLOUT:        case OP_CALLOUT:
2550          rrc = 0;
2551        if (pcre_callout != NULL)        if (pcre_callout != NULL)
2552          {          {
         int rrc;  
2553          pcre_callout_block cb;          pcre_callout_block cb;
2554          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 1;   /* Version 1 of the callout block */
2555          cb.callout_number   = code[1];          cb.callout_number   = code[1];
# Line 2398  for (;;) Line 2564  for (;;)
2564          cb.capture_last     = -1;          cb.capture_last     = -1;
2565          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
2566          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); }  
2567          }          }
2568          if (rrc == 0)
2569            { ADD_ACTIVE(state_offset + _pcre_OP_lengths[OP_CALLOUT], 0); }
2570        break;        break;
2571    
2572    
# Line 2415  for (;;) Line 2582  for (;;)
2582    /* We have finished the processing at the current subject character. If no    /* We have finished the processing at the current subject character. If no
2583    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
2584    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
2585    matching has been requested, check for appropriate conditions. */    matching has been requested, check for appropriate conditions.
2586    
2587      The "forced_ fail" variable counts the number of (*F) encountered for the
2588      character. If it is equal to the original active_count (saved in
2589      workspace[1]) it means that (*F) was found on every active state. In this
2590      case we don't want to give a partial match.
2591    
2592      The "could_continue" variable is true if a state could have continued but
2593      for the fact that the end of the subject was reached. */
2594    
2595    if (new_count <= 0)    if (new_count <= 0)
2596      {      {
2597      if (match_count < 0 &&                     /* No matches found */      if (rlevel == 1 &&                               /* Top level, and */
2598          rlevel == 1 &&                         /* Top level match function */          could_continue &&                            /* Some could go on */
2599          (md->moptions & PCRE_PARTIAL) != 0 &&  /* Want partial matching */          forced_fail != workspace[1] &&               /* Not all forced fail & */
2600          ptr >= end_subject &&                  /* Reached end of subject */          (                                            /* either... */
2601          ptr > current_subject)                 /* Matched non-empty string */          (md->moptions & PCRE_PARTIAL_HARD) != 0      /* Hard partial */
2602            ||                                           /* or... */
2603            ((md->moptions & PCRE_PARTIAL_SOFT) != 0 &&  /* Soft partial and */
2604             match_count < 0)                            /* no matches */
2605            ) &&                                         /* And... */
2606            ptr >= end_subject &&                     /* Reached end of subject */
2607            ptr > current_subject)                    /* Matched non-empty string */
2608        {        {
2609        if (offsetcount >= 2)        if (offsetcount >= 2)
2610          {          {
2611          offsets[0] = current_subject - start_subject;          offsets[0] = md->start_used_ptr - start_subject;
2612          offsets[1] = end_subject - start_subject;          offsets[1] = end_subject - start_subject;
2613          }          }
2614        match_count = PCRE_ERROR_PARTIAL;        match_count = PCRE_ERROR_PARTIAL;
# Line 2482  Returns: > 0 => number of match Line 2663  Returns: > 0 => number of match
2663                   < -1 => some kind of unexpected problem                   < -1 => some kind of unexpected problem
2664  */  */
2665    
2666  PCRE_EXP_DEFN int  PCRE_EXP_DEFN int PCRE_CALL_CONVENTION
2667  pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,  pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,
2668    const char *subject, int length, int start_offset, int options, int *offsets,    const char *subject, int length, int start_offset, int options, int *offsets,
2669    int offsetcount, int *workspace, int wscount)    int offsetcount, int *workspace, int wscount)
# Line 2569  md->start_code = (const uschar *)argumen Line 2750  md->start_code = (const uschar *)argumen
2750      re->name_table_offset + re->name_count * re->name_entry_size;      re->name_table_offset + re->name_count * re->name_entry_size;
2751  md->start_subject = (const unsigned char *)subject;  md->start_subject = (const unsigned char *)subject;
2752  md->end_subject = end_subject;  md->end_subject = end_subject;
2753    md->start_offset = start_offset;
2754  md->moptions = options;  md->moptions = options;
2755  md->poptions = re->options;  md->poptions = re->options;
2756    
2757    /* If the BSR option is not set at match time, copy what was set
2758    at compile time. */
2759    
2760    if ((md->moptions & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) == 0)
2761      {
2762      if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0)
2763        md->moptions |= re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE);
2764    #ifdef BSR_ANYCRLF
2765      else md->moptions |= PCRE_BSR_ANYCRLF;
2766    #endif
2767      }
2768    
2769  /* Handle different types of newline. The three bits give eight cases. If  /* Handle different types of newline. The three bits give eight cases. If
2770  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. */
2771    
# Line 2579  switch ((((options & PCRE_NEWLINE_BITS) Line 2773  switch ((((options & PCRE_NEWLINE_BITS)
2773           PCRE_NEWLINE_BITS)           PCRE_NEWLINE_BITS)
2774    {    {
2775    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Compile-time default */
2776    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = CHAR_CR; break;
2777    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = CHAR_NL; break;
2778    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
2779         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break;
2780    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
2781    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
2782    default: return PCRE_ERROR_BADNEWLINE;    default: return PCRE_ERROR_BADNEWLINE;
# Line 2642  if (md->tables == NULL) md->tables = _pc Line 2836  if (md->tables == NULL) md->tables = _pc
2836  used in a loop when finding where to start. */  used in a loop when finding where to start. */
2837    
2838  lcc = md->tables + lcc_offset;  lcc = md->tables + lcc_offset;
2839  startline = (re->options & PCRE_STARTLINE) != 0;  startline = (re->flags & PCRE_STARTLINE) != 0;
2840  firstline = (re->options & PCRE_FIRSTLINE) != 0;  firstline = (re->options & PCRE_FIRSTLINE) != 0;
2841    
2842  /* Set up the first character to match, if available. The first_byte value is  /* Set up the first character to match, if available. The first_byte value is
# Line 2653  studied, there may be a bitmap of possib Line 2847  studied, there may be a bitmap of possib
2847    
2848  if (!anchored)  if (!anchored)
2849    {    {
2850    if ((re->options & PCRE_FIRSTSET) != 0)    if ((re->flags & PCRE_FIRSTSET) != 0)
2851      {      {
2852      first_byte = re->first_byte & 255;      first_byte = re->first_byte & 255;
2853      if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE)      if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE)
# Line 2661  if (!anchored) Line 2855  if (!anchored)
2855      }      }
2856    else    else
2857      {      {
2858      if (startline && study != NULL &&      if (!startline && study != NULL &&
2859           (study->options & PCRE_STUDY_MAPPED) != 0)           (study->flags & PCRE_STUDY_MAPPED) != 0)
2860        start_bits = study->start_bits;        start_bits = study->start_bits;
2861      }      }
2862    }    }
# Line 2670  if (!anchored) Line 2864  if (!anchored)
2864  /* For anchored or unanchored matches, there may be a "last known required  /* For anchored or unanchored matches, there may be a "last known required
2865  character" set. */  character" set. */
2866    
2867  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
2868    {    {
2869    req_byte = re->req_byte & 255;    req_byte = re->req_byte & 255;
2870    req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;    req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;
# Line 2678  if ((re->options & PCRE_REQCHSET) != 0) Line 2872  if ((re->options & PCRE_REQCHSET) != 0)
2872    }    }
2873    
2874  /* 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
2875  failed match. Unless restarting, optimize by moving to the first match  failed match. If not restarting, perform certain optimizations at the start of
2876  character if possible, when not anchored. Then unless wanting a partial match,  a match. */
 check for a required later character. */  
2877    
2878  for (;;)  for (;;)
2879    {    {
# Line 2690  for (;;) Line 2883  for (;;)
2883      {      {
2884      const uschar *save_end_subject = end_subject;      const uschar *save_end_subject = end_subject;
2885    
2886      /* 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
2887      start of the match is constrained to the first line of a multiline string.      line of a multiline string. Implement this by temporarily adjusting
2888      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
2889      scanning at a newline. If the match fails at the newline, later code breaks      the newline, later code breaks this loop. */
     this loop. */  
2890    
2891      if (firstline)      if (firstline)
2892        {        {
2893        const uschar *t = current_subject;        USPTR t = current_subject;
2894    #ifdef SUPPORT_UTF8
2895          if (utf8)
2896            {
2897            while (t < md->end_subject && !IS_NEWLINE(t))
2898              {
2899              t++;
2900              while (t < end_subject && (*t & 0xc0) == 0x80) t++;
2901              }
2902            }
2903          else
2904    #endif
2905        while (t < md->end_subject && !IS_NEWLINE(t)) t++;        while (t < md->end_subject && !IS_NEWLINE(t)) t++;
2906        end_subject = t;        end_subject = t;
2907        }        }
2908    
2909      if (first_byte >= 0)      /* There are some optimizations that avoid running the match if a known
2910        starting point is not found. However, there is an option that disables
2911        these, for testing and for ensuring that all callouts do actually occur. */
2912    
2913        if ((options & PCRE_NO_START_OPTIMIZE) == 0)
2914        {        {
2915        if (first_byte_caseless)        /* Advance to a known first byte. */
2916          while (current_subject < end_subject &&  
2917                 lcc[*current_subject] != first_byte)        if (first_byte >= 0)
2918            current_subject++;          {
2919        else          if (first_byte_caseless)
2920          while (current_subject < end_subject && *current_subject != first_byte)            while (current_subject < end_subject &&
2921            current_subject++;                   lcc[*current_subject] != first_byte)
2922        }              current_subject++;
2923            else
2924              while (current_subject < end_subject &&
2925                     *current_subject != first_byte)
2926                current_subject++;
2927            }
2928    
2929      /* Or to just after a linebreak for a multiline match if possible */        /* Or to just after a linebreak for a multiline match if possible */
2930    
2931      else if (startline)        else if (startline)
       {  
       if (current_subject > md->start_subject + start_offset)  
2932          {          {
2933          while (current_subject <= end_subject && !WAS_NEWLINE(current_subject))          if (current_subject > md->start_subject + start_offset)
2934            current_subject++;            {
2935    #ifdef SUPPORT_UTF8
2936              if (utf8)
2937                {
2938                while (current_subject < end_subject &&
2939                       !WAS_NEWLINE(current_subject))
2940                  {
2941                  current_subject++;
2942                  while(current_subject < end_subject &&
2943                        (*current_subject & 0xc0) == 0x80)
2944                    current_subject++;
2945                  }
2946                }
2947              else
2948    #endif
2949              while (current_subject < end_subject && !WAS_NEWLINE(current_subject))
2950                current_subject++;
2951    
2952              /* If we have just passed a CR and the newline option is ANY or
2953              ANYCRLF, and we are now at a LF, advance the match position by one
2954              more character. */
2955    
2956          /* If we have just passed a CR and the newline option is ANY or            if (current_subject[-1] == CHAR_CR &&
2957          ANYCRLF, and we are now at a LF, advance the match position by one more                 (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
2958          character. */                 current_subject < end_subject &&
2959                   *current_subject == CHAR_NL)
2960          if (current_subject[-1] == '\r' &&              current_subject++;
2961               (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&            }
              current_subject < end_subject &&  
              *current_subject == '\n')  
           current_subject++;  
2962          }          }
       }  
2963    
2964      /* Or to a non-unique first char after study */        /* Or to a non-unique first char after study */
2965    
2966      else if (start_bits != NULL)        else if (start_bits != NULL)
       {  
       while (current_subject < end_subject)  
2967          {          {
2968          register unsigned int c = *current_subject;          while (current_subject < end_subject)
2969          if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++;            {
2970            else break;            register unsigned int c = *current_subject;
2971              if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++;
2972                else break;
2973              }
2974          }          }
2975        }        }
2976    
2977      /* Restore fudged end_subject */      /* Restore fudged end_subject */
2978    
2979      end_subject = save_end_subject;      end_subject = save_end_subject;
     }  
   
   /* If req_byte is set, we know that that character must appear in the subject  
   for the match to succeed. If the first character is set, req_byte must be  
   later in the subject; otherwise the test starts at the match point. This  
   optimization can save a huge amount of work in patterns with nested unlimited  
   repeats that aren't going to match. Writing separate code for cased/caseless  
   versions makes it go faster, as does using an autoincrement and backing off  
   on a match.  
   
   HOWEVER: when the subject string is very, very long, searching to its end can  
   take a long time, and give bad performance on quite ordinary patterns. This  
   showed up when somebody was matching /^C/ on a 32-megabyte string... so we  
   don't do this when the string is sufficiently long.  
   
   ALSO: this processing is disabled when partial matching is requested.  
   */  
   
   if (req_byte >= 0 &&  
       end_subject - current_subject < REQ_BYTE_MAX &&  
       (options & PCRE_PARTIAL) == 0)  
     {  
     register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);  
2980    
2981      /* We don't need to repeat the search if we haven't yet reached the      /* The following two optimizations are disabled for partial matching or if
2982      place we found it at last time. */      disabling is explicitly requested (and of course, by the test above, this
2983        code is not obeyed when restarting after a partial match). */
2984    
2985      if (p > req_byte_ptr)      if ((options & PCRE_NO_START_OPTIMIZE) == 0 &&
2986            (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT)) == 0)
2987        {        {
2988        if (req_byte_caseless)        /* If the pattern was studied, a minimum subject length may be set. This
2989          {        is a lower bound; no actual string of that length may actually match the
2990          while (p < end_subject)        pattern. Although the value is, strictly, in characters, we treat it as
2991            {        bytes to avoid spending too much time in this optimization. */
2992            register int pp = *p++;  
2993            if (pp == req_byte || pp == req_byte2) { p--; break; }        if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 &&
2994            }            end_subject - current_subject < study->minlength)
2995          }          return PCRE_ERROR_NOMATCH;
2996        else  
2997          /* If req_byte is set, we know that that character must appear in the
2998          subject for the match to succeed. If the first character is set, req_byte
2999          must be later in the subject; otherwise the test starts at the match
3000          point. This optimization can save a huge amount of work in patterns with
3001          nested unlimited repeats that aren't going to match. Writing separate
3002          code for cased/caseless versions makes it go faster, as does using an
3003          autoincrement and backing off on a match.
3004    
3005          HOWEVER: when the subject string is very, very long, searching to its end
3006          can take a long time, and give bad performance on quite ordinary
3007          patterns. This showed up when somebody was matching /^C/ on a 32-megabyte
3008          string... so we don't do this when the string is sufficiently long. */
3009    
3010          if (req_byte >= 0 && end_subject - current_subject < REQ_BYTE_MAX)
3011          {          {
3012          while (p < end_subject)          register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);
3013    
3014            /* We don't need to repeat the search if we haven't yet reached the
3015            place we found it at last time. */
3016    
3017            if (p > req_byte_ptr)
3018            {            {
3019            if (*p++ == req_byte) { p--; break; }            if (req_byte_caseless)
3020            }              {
3021          }              while (p < end_subject)
3022                  {
3023                  register int pp = *p++;
3024                  if (pp == req_byte || pp == req_byte2) { p--; break; }
3025                  }
3026                }
3027              else
3028                {
3029                while (p < end_subject)
3030                  {
3031                  if (*p++ == req_byte) { p--; break; }
3032                  }
3033                }
3034    
3035        /* If we can't find the required character, break the matching loop,            /* If we can't find the required character, break the matching loop,
3036        which will cause a return or PCRE_ERROR_NOMATCH. */            which will cause a return or PCRE_ERROR_NOMATCH. */
3037    
3038        if (p >= end_subject) break;            if (p >= end_subject) break;
3039    
3040        /* If we have found the required character, save the point where we            /* If we have found the required character, save the point where we
3041        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
3042        the start hasn't passed this character yet. */            the start hasn't passed this character yet. */
3043    
3044        req_byte_ptr = p;            req_byte_ptr = p;
3045              }
3046            }
3047        }        }
3048      }      }   /* End of optimizations that are done when not restarting */
3049    
3050    /* OK, now we can do the business */    /* OK, now we can do the business */
3051    
3052      md->start_used_ptr = current_subject;
3053    
3054    rc = internal_dfa_exec(    rc = internal_dfa_exec(
3055      md,                                /* fixed match data */      md,                                /* fixed match data */
3056      md->start_code,                    /* this subexpression's code */      md->start_code,                    /* this subexpression's code */
# Line 2840  for (;;) Line 3081  for (;;)
3081      }      }
3082    if (current_subject > end_subject) break;    if (current_subject > end_subject) break;
3083    
3084    /* If we have just passed a CR and the newline option is CRLF or ANY or    /* If we have just passed a CR and we are now at a LF, and the pattern does
3085    ANYCRLF, and we are now at a LF, advance the match position by one more    not contain any explicit matches for \r or \n, and the newline option is CRLF
3086    character. */    or ANY or ANYCRLF, advance the match position by one more character. */
3087    
3088    if (current_subject[-1] == '\r' &&    if (current_subject[-1] == CHAR_CR &&
3089         (md->nltype == NLTYPE_ANY ||        current_subject < end_subject &&
3090          md->nltype == NLTYPE_ANYCRLF ||        *current_subject == CHAR_NL &&
3091          md->nllen == 2) &&        (re->flags & PCRE_HASCRORLF) == 0 &&
3092         current_subject < end_subject &&          (md->nltype == NLTYPE_ANY ||
3093         *current_subject == '\n')           md->nltype == NLTYPE_ANYCRLF ||
3094             md->nllen == 2))
3095      current_subject++;      current_subject++;
3096    
3097    }   /* "Bumpalong" loop */    }   /* "Bumpalong" loop */

Legend:
Removed from v.199  
changed lines
  Added in v.473

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12