/[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 439 by ph10, Tue Sep 8 17:27:24 2009 UTC revision 518 by ph10, Tue May 18 15:47:01 2010 UTC
# Line 7  and semantics are as close as possible t Line 7  and semantics are as close as possible t
7  below for why this module is different).  below for why this module is different).
8    
9                         Written by Philip Hazel                         Written by Philip Hazel
10             Copyright (c) 1997-2009 University of Cambridge             Copyright (c) 1997-2010 University of Cambridge
11    
12  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
13  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 45  FSM). This is NOT Perl- compatible, but Line 45  FSM). This is NOT Perl- compatible, but
45  applications. */  applications. */
46    
47    
48  /* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved  /* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved
49  the performance of his patterns greatly. I could not use it as it stood, as it  the performance of his patterns greatly. I could not use it as it stood, as it
50  was not thread safe, and made assumptions about pattern sizes. Also, it caused  was not thread safe, and made assumptions about pattern sizes. Also, it caused
51  test 7 to loop, and test 9 to crash with a segfault.  test 7 to loop, and test 9 to crash with a segfault.
52    
53  The issue is the check for duplicate states, which is done by a simple linear  The issue is the check for duplicate states, which is done by a simple linear
# Line 68  was the extra time to initialize the ind Line 68  was the extra time to initialize the ind
68  of internal_dfa_exec(). (The supplied patch used a static vector, initialized  of internal_dfa_exec(). (The supplied patch used a static vector, initialized
69  only once - I suspect this was the cause of the problems with the tests.)  only once - I suspect this was the cause of the problems with the tests.)
70    
71  Overall, I concluded that the gains in some cases did not outweigh the losses  Overall, I concluded that the gains in some cases did not outweigh the losses
72  in others, so I abandoned this code. */  in others, so I abandoned this code. */
73    
74    
# Line 106  never stored, so we push them well clear Line 106  never stored, so we push them well clear
106    
107    
108  /* This table identifies those opcodes that are followed immediately by a  /* This table identifies those opcodes that are followed immediately by a
109  character that is to be tested in some way. This makes is possible to  character that is to be tested in some way. This makes it possible to
110  centralize the loading of these characters. In the case of Type * etc, the  centralize the loading of these characters. In the case of Type * etc, the
111  "character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a  "character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a
112  small value. ***NOTE*** If the start of this table is modified, the two tables  small value. Non-zero values in the table are the offsets from the opcode where
113  that follow must also be modified. */  the character is to be found. ***NOTE*** If the start of this table is
114    modified, the three tables that follow must also be modified. */
115    
116  static const uschar coptable[] = {  static const uschar coptable[] = {
117    0,                             /* End                                    */    0,                             /* End                                    */
118    0, 0, 0, 0, 0,                 /* \A, \G, \K, \B, \b                     */    0, 0, 0, 0, 0,                 /* \A, \G, \K, \B, \b                     */
119    0, 0, 0, 0, 0, 0,              /* \D, \d, \S, \s, \W, \w                 */    0, 0, 0, 0, 0, 0,              /* \D, \d, \S, \s, \W, \w                 */
120    0, 0, 0,                       /* Any, AllAny, Anybyte                   */    0, 0, 0,                       /* Any, AllAny, Anybyte                   */
121    0, 0, 0,                       /* NOTPROP, PROP, EXTUNI                  */    0, 0,                          /* \P, \p                                 */
122    0, 0, 0, 0, 0,                 /* \R, \H, \h, \V, \v                     */    0, 0, 0, 0, 0,                 /* \R, \H, \h, \V, \v                     */
123      0,                             /* \X                                     */
124    0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */    0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */
125    1,                             /* Char                                   */    1,                             /* Char                                   */
126    1,                             /* Charnc                                 */    1,                             /* Charnc                                 */
# Line 155  static const uschar coptable[] = { Line 157  static const uschar coptable[] = {
157    0,                             /* Reverse                                */    0,                             /* Reverse                                */
158    0, 0, 0, 0,                    /* ONCE, BRA, CBRA, COND                  */    0, 0, 0, 0,                    /* ONCE, BRA, CBRA, COND                  */
159    0, 0, 0,                       /* SBRA, SCBRA, SCOND                     */    0, 0, 0,                       /* SBRA, SCBRA, SCOND                     */
160    0,                             /* CREF                                   */    0, 0,                          /* CREF, NCREF                            */
161    0,                             /* RREF                                   */    0, 0,                          /* RREF, NRREF                            */
162    0,                             /* DEF                                    */    0,                             /* DEF                                    */
163    0, 0,                          /* BRAZERO, BRAMINZERO                    */    0, 0,                          /* BRAZERO, BRAMINZERO                    */
164    0, 0, 0, 0,                    /* PRUNE, SKIP, THEN, COMMIT              */    0, 0, 0,                       /* MARK, PRUNE, PRUNE_ARG,                */
165    0, 0, 0                        /* FAIL, ACCEPT, SKIPZERO                 */    0, 0, 0, 0,                    /* SKIP, SKIP_ARG, THEN, THEN_ARG,        */
166      0, 0, 0, 0, 0                  /* COMMIT, FAIL, ACCEPT, CLOSE, SKIPZERO  */
167    };
168    
169    /* This table identifies those opcodes that inspect a character. It is used to
170    remember the fact that a character could have been inspected when the end of
171    the subject is reached. ***NOTE*** If the start of this table is modified, the
172    two tables that follow must also be modified. */
173    
174    static const uschar poptable[] = {
175      0,                             /* End                                    */
176      0, 0, 0, 1, 1,                 /* \A, \G, \K, \B, \b                     */
177      1, 1, 1, 1, 1, 1,              /* \D, \d, \S, \s, \W, \w                 */
178      1, 1, 1,                       /* Any, AllAny, Anybyte                   */
179      1, 1,                          /* \P, \p                                 */
180      1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */
181      1,                             /* \X                                     */
182      0, 0, 0, 0, 0,                 /* \Z, \z, Opt, ^, $                      */
183      1,                             /* Char                                   */
184      1,                             /* Charnc                                 */
185      1,                             /* not                                    */
186      /* Positive single-char repeats                                          */
187      1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */
188      1, 1, 1,                       /* upto, minupto, exact                   */
189      1, 1, 1, 1,                    /* *+, ++, ?+, upto+                      */
190      /* Negative single-char repeats - only for chars < 256                   */
191      1, 1, 1, 1, 1, 1,              /* NOT *, *?, +, +?, ?, ??                */
192      1, 1, 1,                       /* NOT upto, minupto, exact               */
193      1, 1, 1, 1,                    /* NOT *+, ++, ?+, upto+                  */
194      /* Positive type repeats                                                 */
195      1, 1, 1, 1, 1, 1,              /* Type *, *?, +, +?, ?, ??               */
196      1, 1, 1,                       /* Type upto, minupto, exact              */
197      1, 1, 1, 1,                    /* Type *+, ++, ?+, upto+                 */
198      /* Character class & ref repeats                                         */
199      1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */
200      1, 1,                          /* CRRANGE, CRMINRANGE                    */
201      1,                             /* CLASS                                  */
202      1,                             /* NCLASS                                 */
203      1,                             /* XCLASS - variable length               */
204      0,                             /* REF                                    */
205      0,                             /* RECURSE                                */
206      0,                             /* CALLOUT                                */
207      0,                             /* Alt                                    */
208      0,                             /* Ket                                    */
209      0,                             /* KetRmax                                */
210      0,                             /* KetRmin                                */
211      0,                             /* Assert                                 */
212      0,                             /* Assert not                             */
213      0,                             /* Assert behind                          */
214      0,                             /* Assert behind not                      */
215      0,                             /* Reverse                                */
216      0, 0, 0, 0,                    /* ONCE, BRA, CBRA, COND                  */
217      0, 0, 0,                       /* SBRA, SCBRA, SCOND                     */
218      0, 0,                          /* CREF, NCREF                            */
219      0, 0,                          /* RREF, NRREF                            */
220      0,                             /* DEF                                    */
221      0, 0,                          /* BRAZERO, BRAMINZERO                    */
222      0, 0, 0,                       /* MARK, PRUNE, PRUNE_ARG,                */
223      0, 0, 0, 0,                    /* SKIP, SKIP_ARG, THEN, THEN_ARG,        */
224      0, 0, 0, 0, 0                  /* COMMIT, FAIL, ACCEPT, CLOSE, SKIPZERO  */
225  };  };
226    
227  /* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W,  /* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W,
# Line 198  typedef struct stateblock { Line 259  typedef struct stateblock {
259  #define INTS_PER_STATEBLOCK  (sizeof(stateblock)/sizeof(int))  #define INTS_PER_STATEBLOCK  (sizeof(stateblock)/sizeof(int))
260    
261    
262  #ifdef DEBUG  #ifdef PCRE_DEBUG
263  /*************************************************  /*************************************************
264  *             Print character string             *  *             Print character string             *
265  *************************************************/  *************************************************/
# Line 417  if (*first_op == OP_REVERSE) Line 478  if (*first_op == OP_REVERSE)
478        current_subject - start_subject : max_back;        current_subject - start_subject : max_back;
479      current_subject -= gone_back;      current_subject -= gone_back;
480      }      }
481    
482    /* Save the earliest consulted character */    /* Save the earliest consulted character */
483    
484    if (current_subject < md->start_used_ptr)    if (current_subject < md->start_used_ptr)
485      md->start_used_ptr = current_subject;      md->start_used_ptr = current_subject;
486    
487    /* Now we can process the individual branches. */    /* Now we can process the individual branches. */
488    
# Line 488  for (;;) Line 549  for (;;)
549    int clen, dlen;    int clen, dlen;
550    unsigned int c, d;    unsigned int c, d;
551    int forced_fail = 0;    int forced_fail = 0;
552    int reached_end = 0;    BOOL could_continue = FALSE;
553    
554    /* Make the new state list into the active state list and empty the    /* Make the new state list into the active state list and empty the
555    new state list. */    new state list. */
# Line 502  for (;;) Line 563  for (;;)
563    workspace[0] ^= 1;              /* Remember for the restarting feature */    workspace[0] ^= 1;              /* Remember for the restarting feature */
564    workspace[1] = active_count;    workspace[1] = active_count;
565    
566  #ifdef DEBUG  #ifdef PCRE_DEBUG
567    printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP);    printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP);
568    pchars((uschar *)ptr, strlen((char *)ptr), stdout);    pchars((uschar *)ptr, strlen((char *)ptr), stdout);
569    printf("\"\n");    printf("\"\n");
# Line 548  for (;;) Line 609  for (;;)
609      int state_offset = current_state->offset;      int state_offset = current_state->offset;
610      int count, codevalue, rrc;      int count, codevalue, rrc;
611    
612  #ifdef DEBUG  #ifdef PCRE_DEBUG
613      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);      printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);
614      if (clen == 0) printf("EOL\n");      if (clen == 0) printf("EOL\n");
615        else if (c > 32 && c < 127) printf("'%c'\n", c);        else if (c > 32 && c < 127) printf("'%c'\n", c);
# Line 578  for (;;) Line 639  for (;;)
639          }          }
640        }        }
641    
642      /* Check for a duplicate state with the same count, and skip if found.      /* Check for a duplicate state with the same count, and skip if found.
643      See the note at the head of this module about the possibility of improving      See the note at the head of this module about the possibility of improving
644      performance here. */      performance here. */
645    
# Line 597  for (;;) Line 658  for (;;)
658      code = start_code + state_offset;      code = start_code + state_offset;
659      codevalue = *code;      codevalue = *code;
660    
661        /* If this opcode inspects a character, but we are at the end of the
662        subject, remember the fact for use when testing for a partial match. */
663    
664        if (clen == 0 && poptable[codevalue] != 0)
665          could_continue = TRUE;
666    
667      /* If this opcode is followed by an inline character, load it. It is      /* If this opcode is followed by an inline character, load it. It is
668      tempting to test for the presence of a subject character here, but that      tempting to test for the presence of a subject character here, but that
669      is wrong, because sometimes zero repetitions of the subject are      is wrong, because sometimes zero repetitions of the subject are
# Line 643  for (;;) Line 710  for (;;)
710    
711      switch (codevalue)      switch (codevalue)
712        {        {
713    /* ========================================================================== */
714          /* These cases are never obeyed. This is a fudge that causes a compile-
715          time error if the vectors coptable or poptable, which are indexed by
716          opcode, are not the correct length. It seems to be the only way to do
717          such a check at compile time, as the sizeof() operator does not work
718          in the C preprocessor. */
719    
720          case OP_TABLE_LENGTH:
721          case OP_TABLE_LENGTH +
722            ((sizeof(coptable) == OP_TABLE_LENGTH) &&
723             (sizeof(poptable) == OP_TABLE_LENGTH)):
724          break;
725    
726  /* ========================================================================== */  /* ========================================================================== */
727        /* Reached a closing bracket. If not at the end of the pattern, carry        /* Reached a closing bracket. If not at the end of the pattern, carry
728        on with the next opcode. Otherwise, unless we have an empty string and        on with the next opcode. Otherwise, unless we have an empty string and
729        PCRE_NOTEMPTY is set, save the match data, shifting up all previous        PCRE_NOTEMPTY is set, or PCRE_NOTEMPTY_ATSTART is set and we are at the
730          start of the subject, save the match data, shifting up all previous
731        matches so we always have the longest first. */        matches so we always have the longest first. */
732    
733        case OP_KET:        case OP_KET:
# Line 661  for (;;) Line 741  for (;;)
741            ADD_ACTIVE(state_offset - GET(code, 1), 0);            ADD_ACTIVE(state_offset - GET(code, 1), 0);
742            }            }
743          }          }
744        else        else
745          {          {
746          reached_end++;    /* Count branches that reach the end */          if (ptr > current_subject ||
747          if (ptr > current_subject || (md->moptions & PCRE_NOTEMPTY) == 0)              ((md->moptions & PCRE_NOTEMPTY) == 0 &&
748                  ((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 ||
749                    current_subject > start_subject + md->start_offset)))
750            {            {
751            if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;            if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
752              else if (match_count > 0 && ++match_count * 2 >= offsetcount)              else if (match_count > 0 && ++match_count * 2 >= offsetcount)
# Line 685  for (;;) Line 767  for (;;)
767                match_count, rlevel*2-2, SP));                match_count, rlevel*2-2, SP));
768              return match_count;              return match_count;
769              }              }
770            }            }
771          }          }
772        break;        break;
773    
# Line 835  for (;;) Line 917  for (;;)
917          if (ptr > start_subject)          if (ptr > start_subject)
918            {            {
919            const uschar *temp = ptr - 1;            const uschar *temp = ptr - 1;
920            if (temp < md->start_used_ptr) md->start_used_ptr = temp;            if (temp < md->start_used_ptr) md->start_used_ptr = temp;
921  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
922            if (utf8) BACKCHAR(temp);            if (utf8) BACKCHAR(temp);
923  #endif  #endif
924            GETCHARTEST(d, temp);            GETCHARTEST(d, temp);
925    #ifdef SUPPORT_UCP
926              if ((md->poptions & PCRE_UCP) != 0)
927                {
928                if (d == '_') left_word = TRUE; else
929                  {
930                  int cat = UCD_CATEGORY(d);
931                  left_word = (cat == ucp_L || cat == ucp_N);
932                  }
933                }
934              else
935    #endif
936            left_word = d < 256 && (ctypes[d] & ctype_word) != 0;            left_word = d < 256 && (ctypes[d] & ctype_word) != 0;
937            }            }
938          else left_word = 0;          else left_word = FALSE;
939    
940          if (clen > 0)          if (clen > 0)
941              {
942    #ifdef SUPPORT_UCP
943              if ((md->poptions & PCRE_UCP) != 0)
944                {
945                if (c == '_') right_word = TRUE; else
946                  {
947                  int cat = UCD_CATEGORY(c);
948                  right_word = (cat == ucp_L || cat == ucp_N);
949                  }
950                }
951              else
952    #endif
953            right_word = c < 256 && (ctypes[c] & ctype_word) != 0;            right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
         else              /* This is a fudge to ensure that if this is the */  
           {               /* last item in the pattern, we don't count it as */  
           reached_end--;  /* reached, thus disabling a partial match. */  
           right_word = 0;  
954            }            }
955            else right_word = FALSE;
956    
957          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))          if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))
958            { ADD_ACTIVE(state_offset + 1, 0); }            { ADD_ACTIVE(state_offset + 1, 0); }
# Line 877  for (;;) Line 979  for (;;)
979            break;            break;
980    
981            case PT_LAMP:            case PT_LAMP:
982            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
983                   prop->chartype == ucp_Lt;
984            break;            break;
985    
986            case PT_GC:            case PT_GC:
# Line 891  for (;;) Line 994  for (;;)
994            case PT_SC:            case PT_SC:
995            OK = prop->script == code[2];            OK = prop->script == code[2];
996            break;            break;
997    
998              /* These are specials for combination cases. */
999    
1000              case PT_ALNUM:
1001              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1002                   _pcre_ucp_gentype[prop->chartype] == ucp_N;
1003              break;
1004    
1005              case PT_SPACE:    /* Perl space */
1006              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1007                   c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1008              break;
1009    
1010              case PT_PXSPACE:  /* POSIX space */
1011              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1012                   c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1013                   c == CHAR_FF || c == CHAR_CR;
1014              break;
1015    
1016              case PT_WORD:
1017              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1018                   _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1019                   c == CHAR_UNDERSCORE;
1020              break;
1021    
1022            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
1023    
# Line 1046  for (;;) Line 1173  for (;;)
1173            break;            break;
1174    
1175            case PT_LAMP:            case PT_LAMP:
1176            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1177                prop->chartype == ucp_Lt;
1178            break;            break;
1179    
1180            case PT_GC:            case PT_GC:
# Line 1061  for (;;) Line 1189  for (;;)
1189            OK = prop->script == code[3];            OK = prop->script == code[3];
1190            break;            break;
1191    
1192              /* These are specials for combination cases. */
1193    
1194              case PT_ALNUM:
1195              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1196                   _pcre_ucp_gentype[prop->chartype] == ucp_N;
1197              break;
1198    
1199              case PT_SPACE:    /* Perl space */
1200              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1201                   c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1202              break;
1203    
1204              case PT_PXSPACE:  /* POSIX space */
1205              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1206                   c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1207                   c == CHAR_FF || c == CHAR_CR;
1208              break;
1209    
1210              case PT_WORD:
1211              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1212                   _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1213                   c == CHAR_UNDERSCORE;
1214              break;
1215    
1216            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
1217    
1218            default:            default:
# Line 1268  for (;;) Line 1420  for (;;)
1420            break;            break;
1421    
1422            case PT_LAMP:            case PT_LAMP:
1423            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1424                prop->chartype == ucp_Lt;
1425            break;            break;
1426    
1427            case PT_GC:            case PT_GC:
# Line 1282  for (;;) Line 1435  for (;;)
1435            case PT_SC:            case PT_SC:
1436            OK = prop->script == code[3];            OK = prop->script == code[3];
1437            break;            break;
1438    
1439              /* These are specials for combination cases. */
1440    
1441              case PT_ALNUM:
1442              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1443                   _pcre_ucp_gentype[prop->chartype] == ucp_N;
1444              break;
1445    
1446              case PT_SPACE:    /* Perl space */
1447              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1448                   c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1449              break;
1450    
1451              case PT_PXSPACE:  /* POSIX space */
1452              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1453                   c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1454                   c == CHAR_FF || c == CHAR_CR;
1455              break;
1456    
1457              case PT_WORD:
1458              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1459                   _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1460                   c == CHAR_UNDERSCORE;
1461              break;
1462    
1463            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
1464    
# Line 1515  for (;;) Line 1692  for (;;)
1692            break;            break;
1693    
1694            case PT_LAMP:            case PT_LAMP:
1695            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt;            OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1696                prop->chartype == ucp_Lt;
1697            break;            break;
1698    
1699            case PT_GC:            case PT_GC:
# Line 1529  for (;;) Line 1707  for (;;)
1707            case PT_SC:            case PT_SC:
1708            OK = prop->script == code[5];            OK = prop->script == code[5];
1709            break;            break;
1710    
1711              /* These are specials for combination cases. */
1712    
1713              case PT_ALNUM:
1714              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1715                   _pcre_ucp_gentype[prop->chartype] == ucp_N;
1716              break;
1717    
1718              case PT_SPACE:    /* Perl space */
1719              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1720                   c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1721              break;
1722    
1723              case PT_PXSPACE:  /* POSIX space */
1724              OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1725                   c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1726                   c == CHAR_FF || c == CHAR_CR;
1727              break;
1728    
1729              case PT_WORD:
1730              OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1731                   _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1732                   c == CHAR_UNDERSCORE;
1733              break;
1734    
1735            /* Should never occur, but keep compilers from grumbling. */            /* Should never occur, but keep compilers from grumbling. */
1736    
# Line 2237  for (;;) Line 2439  for (;;)
2439            rlevel,                               /* function recursion level */            rlevel,                               /* function recursion level */
2440            recursing);                           /* pass on regex recursion */            recursing);                           /* pass on regex recursion */
2441    
2442            if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2443          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))          if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))
2444              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
2445          }          }
# Line 2283  for (;;) Line 2486  for (;;)
2486    
2487          /* Back reference conditions are not supported */          /* Back reference conditions are not supported */
2488    
2489          if (condcode == OP_CREF) return PCRE_ERROR_DFA_UCOND;          if (condcode == OP_CREF || condcode == OP_NCREF)
2490              return PCRE_ERROR_DFA_UCOND;
2491    
2492          /* The DEFINE condition is always false */          /* The DEFINE condition is always false */
2493    
# Line 2294  for (;;) Line 2498  for (;;)
2498          which means "test if in any recursion". We can't test for specifically          which means "test if in any recursion". We can't test for specifically
2499          recursed groups. */          recursed groups. */
2500    
2501          else if (condcode == OP_RREF)          else if (condcode == OP_RREF || condcode == OP_NRREF)
2502            {            {
2503            int value = GET2(code, LINK_SIZE+2);            int value = GET2(code, LINK_SIZE+2);
2504            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;            if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;
# Line 2326  for (;;) Line 2530  for (;;)
2530              rlevel,                               /* function recursion level */              rlevel,                               /* function recursion level */
2531              recursing);                           /* pass on regex recursion */              recursing);                           /* pass on regex recursion */
2532    
2533              if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2534            if ((rc >= 0) ==            if ((rc >= 0) ==
2535                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))                  (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))
2536              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }              { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); }
# Line 2517  for (;;) Line 2722  for (;;)
2722    /* We have finished the processing at the current subject character. If no    /* We have finished the processing at the current subject character. If no
2723    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
2724    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
2725    matching has been requested, check for appropriate conditions. The "forced_    matching has been requested, check for appropriate conditions.
2726    fail" variable counts the number of (*F) encountered for the character. If it  
2727    is equal to the original active_count (saved in workspace[1]) it means that    The "forced_ fail" variable counts the number of (*F) encountered for the
2728    (*F) was found on every active state. In this case we don't want to give a    character. If it is equal to the original active_count (saved in
2729    partial match. */    workspace[1]) it means that (*F) was found on every active state. In this
2730      case we don't want to give a partial match.
2731    
2732      The "could_continue" variable is true if a state could have continued but
2733      for the fact that the end of the subject was reached. */
2734    
2735    if (new_count <= 0)    if (new_count <= 0)
2736      {      {
2737      if (rlevel == 1 &&                               /* Top level, and */      if (rlevel == 1 &&                               /* Top level, and */
2738          reached_end != workspace[1] &&               /* Not all reached end */          could_continue &&                            /* Some could go on */
2739          forced_fail != workspace[1] &&               /* Not all forced fail & */          forced_fail != workspace[1] &&               /* Not all forced fail & */
2740          (                                            /* either... */          (                                            /* either... */
2741          (md->moptions & PCRE_PARTIAL_HARD) != 0      /* Hard partial */          (md->moptions & PCRE_PARTIAL_HARD) != 0      /* Hard partial */
# Line 2681  md->start_code = (const uschar *)argumen Line 2890  md->start_code = (const uschar *)argumen
2890      re->name_table_offset + re->name_count * re->name_entry_size;      re->name_table_offset + re->name_count * re->name_entry_size;
2891  md->start_subject = (const unsigned char *)subject;  md->start_subject = (const unsigned char *)subject;
2892  md->end_subject = end_subject;  md->end_subject = end_subject;
2893    md->start_offset = start_offset;
2894  md->moptions = options;  md->moptions = options;
2895  md->poptions = re->options;  md->poptions = re->options;
2896    
# Line 2785  if (!anchored) Line 2995  if (!anchored)
2995      }      }
2996    else    else
2997      {      {
2998      if (startline && study != NULL &&      if (!startline && study != NULL &&
2999           (study->options & PCRE_STUDY_MAPPED) != 0)           (study->flags & PCRE_STUDY_MAPPED) != 0)
3000        start_bits = study->start_bits;        start_bits = study->start_bits;
3001      }      }
3002    }    }
# Line 2837  for (;;) Line 3047  for (;;)
3047        }        }
3048    
3049      /* There are some optimizations that avoid running the match if a known      /* There are some optimizations that avoid running the match if a known
3050      starting point is not found, or if a known later character is not present.      starting point is not found. However, there is an option that disables
3051      However, there is an option that disables these, for testing and for      these, for testing and for ensuring that all callouts do actually occur. */
     ensuring that all callouts do actually occur. */  
3052    
3053      if ((options & PCRE_NO_START_OPTIMIZE) == 0)      if ((options & PCRE_NO_START_OPTIMIZE) == 0)
3054        {        {
   
3055        /* Advance to a known first byte. */        /* Advance to a known first byte. */
3056    
3057        if (first_byte >= 0)        if (first_byte >= 0)
# Line 2909  for (;;) Line 3117  for (;;)
3117      /* Restore fudged end_subject */      /* Restore fudged end_subject */
3118    
3119      end_subject = save_end_subject;      end_subject = save_end_subject;
     }  
3120    
3121    /* 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
3122    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
3123    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, and can  
   also be explicitly deactivated. Furthermore, we have to disable when  
   restarting after a partial match, because the required character may have  
   already been matched. */  
   
   if ((options & PCRE_NO_START_OPTIMIZE) == 0 &&  
       req_byte >= 0 &&  
       end_subject - current_subject < REQ_BYTE_MAX &&  
       (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_DFA_RESTART)) == 0)  
     {  
     register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);  
3124    
3125      /* We don't need to repeat the search if we haven't yet reached the      if ((options & PCRE_NO_START_OPTIMIZE) == 0 &&
3126      place we found it at last time. */          (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT)) == 0)
   
     if (p > req_byte_ptr)  
3127        {        {
3128        if (req_byte_caseless)        /* If the pattern was studied, a minimum subject length may be set. This
3129          {        is a lower bound; no actual string of that length may actually match the
3130          while (p < end_subject)        pattern. Although the value is, strictly, in characters, we treat it as
3131            {        bytes to avoid spending too much time in this optimization. */
3132            register int pp = *p++;  
3133            if (pp == req_byte || pp == req_byte2) { p--; break; }        if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 &&
3134            }            (pcre_uint32)(end_subject - current_subject) < study->minlength)
3135          }          return PCRE_ERROR_NOMATCH;
3136        else  
3137          /* If req_byte is set, we know that that character must appear in the
3138          subject for the match to succeed. If the first character is set, req_byte
3139          must be later in the subject; otherwise the test starts at the match
3140          point. This optimization can save a huge amount of work in patterns with
3141          nested unlimited repeats that aren't going to match. Writing separate
3142          code for cased/caseless versions makes it go faster, as does using an
3143          autoincrement and backing off on a match.
3144    
3145          HOWEVER: when the subject string is very, very long, searching to its end
3146          can take a long time, and give bad performance on quite ordinary
3147          patterns. This showed up when somebody was matching /^C/ on a 32-megabyte
3148          string... so we don't do this when the string is sufficiently long. */
3149    
3150          if (req_byte >= 0 && end_subject - current_subject < REQ_BYTE_MAX)
3151          {          {
3152          while (p < end_subject)          register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);
3153    
3154            /* We don't need to repeat the search if we haven't yet reached the
3155            place we found it at last time. */
3156    
3157            if (p > req_byte_ptr)
3158            {            {
3159            if (*p++ == req_byte) { p--; break; }            if (req_byte_caseless)
3160            }              {
3161          }              while (p < end_subject)
3162                  {
3163                  register int pp = *p++;
3164                  if (pp == req_byte || pp == req_byte2) { p--; break; }
3165                  }
3166                }
3167              else
3168                {
3169                while (p < end_subject)
3170                  {
3171                  if (*p++ == req_byte) { p--; break; }
3172                  }
3173                }
3174    
3175        /* If we can't find the required character, break the matching loop,            /* If we can't find the required character, break the matching loop,
3176        which will cause a return or PCRE_ERROR_NOMATCH. */            which will cause a return or PCRE_ERROR_NOMATCH. */
3177    
3178        if (p >= end_subject) break;            if (p >= end_subject) break;
3179    
3180        /* If we have found the required character, save the point where we            /* If we have found the required character, save the point where we
3181        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
3182        the start hasn't passed this character yet. */            the start hasn't passed this character yet. */
3183    
3184        req_byte_ptr = p;            req_byte_ptr = p;
3185              }
3186            }
3187        }        }
3188      }      }   /* End of optimizations that are done when not restarting */
3189    
3190    /* OK, now we can do the business */    /* OK, now we can do the business */
3191    
3192    md->start_used_ptr = current_subject;    md->start_used_ptr = current_subject;
3193    
3194    rc = internal_dfa_exec(    rc = internal_dfa_exec(
3195      md,                                /* fixed match data */      md,                                /* fixed match data */
3196      md->start_code,                    /* this subexpression's code */      md->start_code,                    /* this subexpression's code */

Legend:
Removed from v.439  
changed lines
  Added in v.518

  ViewVC Help
Powered by ViewVC 1.1.5