/[pcre]/code/trunk/pcre_exec.c
ViewVC logotype

Diff of /code/trunk/pcre_exec.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 609 by ph10, Wed Jun 15 18:09:23 2011 UTC revision 623 by ph10, Tue Jul 19 09:58:42 2011 UTC
# Line 76  negative to avoid the external error cod Line 76  negative to avoid the external error cod
76  #define MATCH_ACCEPT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_COMMIT       (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_KETRPOS      (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_PRUNE        (-996)  #define MATCH_ONCE         (-996)
80  #define MATCH_SKIP         (-995)  #define MATCH_PRUNE        (-995)
81  #define MATCH_SKIP_ARG     (-994)  #define MATCH_SKIP         (-994)
82  #define MATCH_THEN         (-993)  #define MATCH_SKIP_ARG     (-993)
83    #define MATCH_THEN         (-992)
84    
85  /* This is a convenience macro for code that occurs many times. */  /* This is a convenience macro for code that occurs many times. */
86    
# Line 276  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 277  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
277         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
278         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
279         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280         RM61,  RM62, RM63};         RM61,  RM62, RM63 };
281    
282  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
283  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
# Line 383  typedef struct heapframe { Line 384  typedef struct heapframe {
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
386    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
387    int Xoclength;    int Xoclength;
388    uschar Xocchars[8];    uschar Xocchars[8];
389  #endif  #endif
# Line 550  HEAP_RECURSE: Line 548  HEAP_RECURSE:
548  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
549  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
550  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
 #define prop_category      frame->Xprop_category  
 #define prop_chartype      frame->Xprop_chartype  
 #define prop_script        frame->Xprop_script  
551  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
552  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
553  #endif  #endif
# Line 610  BOOL prev_is_word; Line 605  BOOL prev_is_word;
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
608  int oclength;  int oclength;
609  uschar occhars[8];  uschar occhars[8];
610  #endif  #endif
# Line 808  for (;;) Line 800  for (;;)
800      subject position in the working slot at the top of the vector. We mustn't      subject position in the working slot at the top of the vector. We mustn't
801      change the current values of the data slot, because they may be set from a      change the current values of the data slot, because they may be set from a
802      previous iteration of this group, and be referred to by a reference inside      previous iteration of this group, and be referred to by a reference inside
803      the group. If we fail to match, we need to restore this value and also the      the group. A failure to match might occur after the group has succeeded,
804      values of the final offsets, in case they were set by a previous iteration      if something later on doesn't match. For this reason, we need to restore
805      of the same bracket.      the working value and also the values of the final offsets, in case they
806        were set by a previous iteration of the same bracket.
807    
808      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
809      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 844  for (;;) Line 837  for (;;)
837          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839            eptrb, RM1);            eptrb, RM1);
840            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841          if (rrc != MATCH_NOMATCH &&          if (rrc != MATCH_NOMATCH &&
842              (rrc != MATCH_THEN || md->start_match_ptr != ecode))              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
843            RRETURN(rrc);            RRETURN(rrc);
# Line 853  for (;;) Line 847  for (;;)
847          }          }
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854          /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855          MATCH_THEN. */
856    
857        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
858        RRETURN(MATCH_NOMATCH);        RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
859        }        }
860    
861      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 873  for (;;) Line 869  for (;;)
869      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
871    
872      /* Non-capturing bracket, except for possessive with unlimited repeat. Loop      /* Non-capturing or atomic group, except for possessive with unlimited
873      for all the alternatives. When we get to the final alternative within the      repeat. Loop for all the alternatives. When we get to the final alternative
874      brackets, we used to return the result of a recursive call to match()      within the brackets, we used to return the result of a recursive call to
875      whatever happened so it was possible to reduce stack usage by turning this      match() whatever happened so it was possible to reduce stack usage by
876      into a tail recursion, except in the case of a possibly empty group.      turning this into a tail recursion, except in the case of a possibly empty
877      However, now that there is the possiblity of (*THEN) occurring in the final      group. However, now that there is the possiblity of (*THEN) occurring in
878      alternative, this optimization is no longer possible. */      the final alternative, this optimization is no longer possible.
879    
880        MATCH_ONCE is returned when the end of an atomic group is successfully
881        reached, but subsequent matching fails. It passes back up the tree (causing
882        captured values to be reset) until the original atomic group level is
883        reached. This is tested by comparing md->once_target with the start of the
884        group. At this point, the return is converted into MATCH_NOMATCH so that
885        previous backup points can be taken. */
886    
887        case OP_ONCE:
888      case OP_BRA:      case OP_BRA:
889      case OP_SBRA:      case OP_SBRA:
890      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
891    
892      for (;;)      for (;;)
893        {        {
894        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
895        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
896          RM2);          RM2);
897        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
898            (rrc != MATCH_THEN || md->start_match_ptr != ecode))            (rrc != MATCH_THEN || md->start_match_ptr != ecode))
899            {
900            if (rrc == MATCH_ONCE)
901              {
902              const uschar *scode = ecode;
903              if (*scode != OP_ONCE)           /* If not at start, find it */
904                {
905                while (*scode == OP_ALT) scode += GET(scode, 1);
906                scode -= GET(scode, 1);
907                }
908              if (md->once_target == scode) rrc = MATCH_NOMATCH;
909              }
910          RRETURN(rrc);          RRETURN(rrc);
911            }
912        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
913        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
914        }        }
   
915      if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;      if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
916      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
917    
# Line 966  for (;;) Line 982  for (;;)
982          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
983          if (*ecode != OP_ALT) break;          if (*ecode != OP_ALT) break;
984          }          }
985    
986        if (!matched_once)        if (!matched_once)
987          {          {
988          md->offset_vector[offset] = save_offset1;          md->offset_vector[offset] = save_offset1;
# Line 1015  for (;;) Line 1031  for (;;)
1031          eptrb, RM48);          eptrb, RM48);
1032        if (rrc == MATCH_KETRPOS)        if (rrc == MATCH_KETRPOS)
1033          {          {
1034            offset_top = md->end_offset_top;
1035          eptr = md->end_match_ptr;          eptr = md->end_match_ptr;
1036          ecode = md->start_code + code_offset;          ecode = md->start_code + code_offset;
1037          matched_once = TRUE;          matched_once = TRUE;
# Line 1026  for (;;) Line 1043  for (;;)
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1045        }        }
1046    
1047      if (matched_once || allow_zero)      if (matched_once || allow_zero)
1048        {        {
1049        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
# Line 1227  for (;;) Line 1244  for (;;)
1244        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1245        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1246          {          {
1247            if (md->end_offset_top > offset_top)
1248              offset_top = md->end_offset_top;  /* Captures may have happened */
1249          condition = TRUE;          condition = TRUE;
1250          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1251          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
# Line 1290  for (;;) Line 1309  for (;;)
1309      break;      break;
1310    
1311    
1312      /* End of the pattern, either real or forced. If we are in a recursion, we      /* End of the pattern, either real or forced. */
     should restore the offsets appropriately, and if it's a top-level  
     recursion, continue from after the call. */  
1313    
     case OP_ACCEPT:  
1314      case OP_END:      case OP_END:
1315      if (md->recursive != NULL)      case OP_ACCEPT:
1316        {      case OP_ASSERT_ACCEPT:
1317        recursion_info *rec = md->recursive;  
1318        md->recursive = rec->prevrec;      /* If we have matched an empty string, fail if not in an assertion and not
1319        memmove(md->offset_vector, rec->offset_save,      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1320          rec->saved_max * sizeof(int));      is set and we have matched at the start of the subject. In both cases,
1321        offset_top = rec->save_offset_top;      backtracking will then try other alternatives, if any. */
1322        if (rec->group_num == 0)  
1323          {      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1324          ecode = rec->after_call;           md->recursive == NULL &&
1325          break;           (md->notempty ||
1326          }             (md->notempty_atstart &&
1327        }               mstart == md->start_subject + md->start_offset)))
   
     /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is  
     set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of  
     the subject. In both cases, backtracking will then try other alternatives,  
     if any. */  
   
     else if (eptr == mstart &&  
         (md->notempty ||  
           (md->notempty_atstart &&  
             mstart == md->start_subject + md->start_offset)))  
1328        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1329    
1330      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1478  for (;;) Line 1484  for (;;)
1484      /* Recursion either matches the current regex, or some subexpression. The      /* Recursion either matches the current regex, or some subexpression. The
1485      offset data is the offset to the starting bracket from the start of the      offset data is the offset to the starting bracket from the start of the
1486      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1487    
1488      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1489      save their starting points and reinstate them after the recursion. However,      re-instated afterwards. We don't know how many are started and not yet
1490      we don't know how many such there are (offset_top records the completed      finished (offset_top records the completed total) so we just have to save
1491      total) so we just have to save all the potential data. There may be up to      all the potential data. There may be up to 65535 such values, which is too
1492      65535 such values, which is too large to put on the stack, but using malloc      large to put on the stack, but using malloc for small numbers seems
1493      for small numbers seems expensive. As a compromise, the stack is used when      expensive. As a compromise, the stack is used when there are no more than
1494      there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc      REC_STACK_SAVE_MAX values to store; otherwise malloc is used.
     is used. A problem is what to do if the malloc fails ... there is no way of  
     returning to the top level with an error. Save the top REC_STACK_SAVE_MAX  
     values on the stack, and accept that the rest may be wrong.  
1495    
1496      There are also other values that have to be saved. We use a chained      There are also other values that have to be saved. We use a chained
1497      sequence of blocks that actually live on the stack. Thanks to Robin Houston      sequence of blocks that actually live on the stack. Thanks to Robin Houston
1498      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1499        a lot, so he is not to blame for the current way it works. */
1500    
1501      case OP_RECURSE:      case OP_RECURSE:
1502        {        {
# Line 1505  for (;;) Line 1509  for (;;)
1509        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1510        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1511    
1512        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1513    
1514        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1515    
1516        /* Now save the offset data. */        /* Now save the offset data */
1517    
1518        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1519        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1521  for (;;) Line 1524  for (;;)
1524            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1525          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1526          }          }
   
1527        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1528              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1529    
1530        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1531        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1532          might be changed, so reset it before looping. */
1533    
1534        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1535        cbegroup = (*callpat >= OP_SBRA);        cbegroup = (*callpat >= OP_SBRA);
# Line 1536  for (;;) Line 1538  for (;;)
1538          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1539          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1540            md, eptrb, RM6);            md, eptrb, RM6);
1541            memcpy(md->offset_vector, new_recursive.offset_save,
1542                new_recursive.saved_max * sizeof(int));
1543          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1544            {            {
1545            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1546            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1547            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1548              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1549            MRRETURN(MATCH_MATCH);  
1550              /* Set where we got to in the subject, and reset the start in case
1551              it was changed by \K. This *is* propagated back out of a recursion,
1552              for Perl compatibility. */
1553    
1554              eptr = md->end_match_ptr;
1555              mstart = md->start_match_ptr;
1556              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1557            }            }
1558          else if (rrc != MATCH_NOMATCH &&          else if (rrc != MATCH_NOMATCH &&
1559                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
# Line 1554  for (;;) Line 1565  for (;;)
1565            }            }
1566    
1567          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1568          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1569          }          }
1570        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1566  for (;;) Line 1575  for (;;)
1575          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1576        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1577        }        }
1578      /* Control never reaches here */  
1579        RECURSION_MATCHED:
1580      /* "Once" brackets are like assertion brackets except that after a match,      break;
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer, but resetting  
     the start-of-match value in case it was changed by \K. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);  
       if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */  
         {  
         mstart = md->start_match_ptr;  
         break;  
         }  
       if (rrc != MATCH_NOMATCH &&  
           (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
         RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. */  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       md->match_function_type = MATCH_CBEGROUP;  
       RMATCH(eptr, prev, offset_top, md, eptrb, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
1581    
1582      /* An alternation is the end of a branch; scan along to find the end of the      /* An alternation is the end of a branch; scan along to find the end of the
1583      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1691  for (;;) Line 1630  for (;;)
1630      case OP_KETRMAX:      case OP_KETRMAX:
1631      case OP_KETRPOS:      case OP_KETRPOS:
1632      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1633    
1634      /* If this was a group that remembered the subject start, in order to break      /* If this was a group that remembered the subject start, in order to break
1635      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1636      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1637    
1638      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1639        {        {
1640        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1641        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1642        }        }
1643      else saved_eptr = NULL;      else saved_eptr = NULL;
1644    
1645      /* If we are at the end of an assertion group or an atomic group, stop      /* If we are at the end of an assertion group, stop matching and return
1646      matching and return MATCH_MATCH, but record the current high water mark for      MATCH_MATCH, but record the current high water mark for use by positive
1647      use by positive assertions. We also need to record the match start in case      assertions. We also need to record the match start in case it was changed
1648      it was changed by \K. */      by \K. */
1649    
1650      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1651          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1652        {        {
1653        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1654        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
# Line 1720  for (;;) Line 1658  for (;;)
1658    
1659      /* For capturing groups we have to check the group number back at the start      /* For capturing groups we have to check the group number back at the start
1660      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1661      bumping the high water mark. Note that whole-pattern recursion is coded as      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1662      a recurse into group 0, so it won't be picked up here. Instead, we catch it      into group 0, so it won't be picked up here. Instead, we catch it when the
1663      when the OP_END is reached. Other recursion is handled here. */      OP_END is reached. Other recursion is handled here. We just have to record
1664        the current subject position and start match pointer and give a MATCH
1665        return. */
1666    
1667      if (*prev == OP_CBRA || *prev == OP_SCBRA ||      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1668          *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)          *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
# Line 1735  for (;;) Line 1675  for (;;)
1675        printf("\n");        printf("\n");
1676  #endif  #endif
1677    
1678          /* Handle a recursively called group. */
1679    
1680          if (md->recursive != NULL && md->recursive->group_num == number)
1681            {
1682            md->end_match_ptr = eptr;
1683            md->start_match_ptr = mstart;
1684            RRETURN(MATCH_MATCH);
1685            }
1686    
1687          /* Deal with capturing */
1688    
1689        md->capture_last = number;        md->capture_last = number;
1690        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1691          {          {
1692            /* If offset is greater than offset_top, it means that we are
1693            "skipping" a capturing group, and that group's offsets must be marked
1694            unset. In earlier versions of PCRE, all the offsets were unset at the
1695            start of matching, but this doesn't work because atomic groups and
1696            assertions can cause a value to be set that should later be unset.
1697            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1698            part of the atomic group, but this is not on the final matching path,
1699            so must be unset when 2 is set. (If there is no group 2, there is no
1700            problem, because offset_top will then be 2, indicating no capture.) */
1701    
1702            if (offset > offset_top)
1703              {
1704              register int *iptr = md->offset_vector + offset_top;
1705              register int *iend = md->offset_vector + offset;
1706              while (iptr < iend) *iptr++ = -1;
1707              }
1708    
1709            /* Now make the extraction */
1710    
1711          md->offset_vector[offset] =          md->offset_vector[offset] =
1712            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1713          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1714          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1715          }          }
   
       /* Handle a recursively called group. Restore the offsets  
       appropriately and continue from after the call. */  
   
       if (md->recursive != NULL && md->recursive->group_num == number)  
         {  
         recursion_info *rec = md->recursive;  
         DPRINTF(("Recursion (%d) succeeded - continuing\n", number));  
         md->recursive = rec->prevrec;  
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         offset_top = rec->save_offset_top;  
         ecode = rec->after_call;  
         break;  
         }  
1716        }        }
1717    
1718      /* For a non-repeating ket, just continue at this level. This also      /* For an ordinary non-repeating ket, just continue at this level. This
1719      happens for a repeating ket if no characters were matched in the group.      also happens for a repeating ket if no characters were matched in the
1720      This is the forcible breaking of infinite loops as implemented in Perl      group. This is the forcible breaking of infinite loops as implemented in
1721      5.005. If there is an options reset, it will get obeyed in the normal      Perl 5.005. For a non-repeating atomic group, establish a backup point by
1722      course of events. */      processing the rest of the pattern at a lower level. If this results in a
1723        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1724        bypassing intermediate backup points, but resetting any captures that
1725        happened along the way. */
1726    
1727      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1728        {        {
1729        ecode += 1 + LINK_SIZE;        if (*prev == OP_ONCE)
1730            {
1731            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1732            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1733            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1734            RRETURN(MATCH_ONCE);
1735            }
1736          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1737        break;        break;
1738        }        }
1739    
# Line 1786  for (;;) Line 1751  for (;;)
1751      /* The normal repeating kets try the rest of the pattern or restart from      /* The normal repeating kets try the rest of the pattern or restart from
1752      the preceding bracket, in the appropriate order. In the second case, we can      the preceding bracket, in the appropriate order. In the second case, we can
1753      use tail recursion to avoid using another stack frame, unless we have an      use tail recursion to avoid using another stack frame, unless we have an
1754      unlimited repeat of a group that can match an empty string. */      an atomic group or an unlimited repeat of a group that can match an empty
1755        string. */
1756    
1757      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1758        {        {
1759        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1760        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1761          if (*prev == OP_ONCE)
1762            {
1763            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1764            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1765            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1766            RRETURN(MATCH_ONCE);
1767            }
1768        if (*prev >= OP_SBRA)    /* Could match an empty string */        if (*prev >= OP_SBRA)    /* Could match an empty string */
1769          {          {
1770          md->match_function_type = MATCH_CBEGROUP;          md->match_function_type = MATCH_CBEGROUP;
# Line 1805  for (;;) Line 1778  for (;;)
1778        {        {
1779        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1780        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1781          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1782        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1783          if (*prev == OP_ONCE)
1784            {
1785            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1786            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1787            md->once_target = prev;
1788            RRETURN(MATCH_ONCE);
1789            }
1790        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1791        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1792        }        }
# Line 2374  for (;;) Line 2355  for (;;)
2355        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2356        }        }
2357      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2358        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2359        while (eptr < md->end_subject)
2360        {        {
2361        int category = UCD_CATEGORY(c);        int len = 1;
2362        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2363        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2364          {        eptr += len;
         int len = 1;  
         if (!utf8) c = *eptr; else  
           {  
           GETCHARLEN(c, eptr, len);  
           }  
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
         eptr += len;  
         }  
2365        }        }
2366      ecode++;      ecode++;
2367      break;      break;
# Line 3741  for (;;) Line 3715  for (;;)
3715            case PT_LAMP:            case PT_LAMP:
3716            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3717              {              {
3718                int chartype;
3719              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3720                {                {
3721                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3722                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3723                }                }
3724              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3725              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
3726              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
3727                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
3728                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
3729                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3730              }              }
3731            break;            break;
# Line 3764  for (;;) Line 3739  for (;;)
3739                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3740                }                }
3741              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3742              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
3743                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3744              }              }
3745            break;            break;
# Line 3779  for (;;) Line 3753  for (;;)
3753                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3754                }                }
3755              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3756              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
3757                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3758              }              }
3759            break;            break;
# Line 3794  for (;;) Line 3767  for (;;)
3767                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3768                }                }
3769              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3770              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
3771                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3772              }              }
3773            break;            break;
# Line 3803  for (;;) Line 3775  for (;;)
3775            case PT_ALNUM:            case PT_ALNUM:
3776            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3777              {              {
3778                int category;
3779              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3780                {                {
3781                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3782                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3783                }                }
3784              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3785              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3786              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
3787                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3788              }              }
3789            break;            break;
# Line 3825  for (;;) Line 3797  for (;;)
3797                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3798                }                }
3799              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3800              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
3801                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
3802                     == prop_fail_result)                     == prop_fail_result)
3803                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3842  for (;;) Line 3813  for (;;)
3813                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3814                }                }
3815              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3816              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
3817                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3818                     == prop_fail_result)                     == prop_fail_result)
3819                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3853  for (;;) Line 3823  for (;;)
3823            case PT_WORD:            case PT_WORD:
3824            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3825              {              {
3826                int category;
3827              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3828                {                {
3829                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3830                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3831                }                }
3832              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3833              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3834              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
                  c == CHAR_UNDERSCORE)  
3835                     == prop_fail_result)                     == prop_fail_result)
3836                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3837              }              }
# Line 3887  for (;;) Line 3857  for (;;)
3857              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
3858              }              }
3859            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3860            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
3861            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3862              {              {
3863              int len = 1;              int len = 1;
3864              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
3865                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
3866              eptr += len;              eptr += len;
3867              }              }
3868            }            }
# Line 4440  for (;;) Line 4407  for (;;)
4407            case PT_LAMP:            case PT_LAMP:
4408            for (fi = min;; fi++)            for (fi = min;; fi++)
4409              {              {
4410                int chartype;
4411              RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4412              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4413              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4449  for (;;) Line 4417  for (;;)
4417                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4418                }                }
4419              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4420              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4421              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4422                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4423                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4424                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4425              }              }
4426            /* Control never gets here */            /* Control never gets here */
# Line 4469  for (;;) Line 4437  for (;;)
4437                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4438                }                }
4439              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4440              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4441                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4442              }              }
4443            /* Control never gets here */            /* Control never gets here */
# Line 4487  for (;;) Line 4454  for (;;)
4454                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4455                }                }
4456              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4457              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4458                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4459              }              }
4460            /* Control never gets here */            /* Control never gets here */
# Line 4505  for (;;) Line 4471  for (;;)
4471                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4472                }                }
4473              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4474              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
4475                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4476              }              }
4477            /* Control never gets here */            /* Control never gets here */
# Line 4514  for (;;) Line 4479  for (;;)
4479            case PT_ALNUM:            case PT_ALNUM:
4480            for (fi = min;; fi++)            for (fi = min;; fi++)
4481              {              {
4482                int category;
4483              RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4484              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4485              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4523  for (;;) Line 4489  for (;;)
4489                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4490                }                }
4491              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4492              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4493              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
4494                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4495              }              }
4496            /* Control never gets here */            /* Control never gets here */
# Line 4542  for (;;) Line 4507  for (;;)
4507                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4508                }                }
4509              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4510              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
4511                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
4512                     == prop_fail_result)                     == prop_fail_result)
4513                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4562  for (;;) Line 4526  for (;;)
4526                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4527                }                }
4528              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4529              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
4530                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4531                     == prop_fail_result)                     == prop_fail_result)
4532                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4573  for (;;) Line 4536  for (;;)
4536            case PT_WORD:            case PT_WORD:
4537            for (fi = min;; fi++)            for (fi = min;; fi++)
4538              {              {
4539                int category;
4540              RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4541              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4542              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4582  for (;;) Line 4546  for (;;)
4546                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4547                }                }
4548              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4549              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4550              if ((prop_category == ucp_L ||              if ((category == ucp_L ||
4551                   prop_category == ucp_N ||                   category == ucp_N ||
4552                   c == CHAR_UNDERSCORE)                   c == CHAR_UNDERSCORE)
4553                     == prop_fail_result)                     == prop_fail_result)
4554                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4614  for (;;) Line 4578  for (;;)
4578              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4579              }              }
4580            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4581            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4582            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4583              {              {
4584              int len = 1;              int len = 1;
4585              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4586                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4587              eptr += len;              eptr += len;
4588              }              }
4589            }            }
4590          }          }
   
4591        else        else
4592  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4593    
# Line 4948  for (;;) Line 4908  for (;;)
4908            case PT_LAMP:            case PT_LAMP:
4909            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4910              {              {
4911                int chartype;
4912              int len = 1;              int len = 1;
4913              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4914                {                {
# Line 4955  for (;;) Line 4916  for (;;)
4916                break;                break;
4917                }                }
4918              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4919              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4920              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4921                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4922                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4923                break;                break;
4924              eptr+= len;              eptr+= len;
4925              }              }
# Line 4974  for (;;) Line 4935  for (;;)
4935                break;                break;
4936                }                }
4937              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4938              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
4939              eptr+= len;              eptr+= len;
4940              }              }
4941            break;            break;
# Line 4991  for (;;) Line 4950  for (;;)
4950                break;                break;
4951                }                }
4952              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4953              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
4954              eptr+= len;              eptr+= len;
4955              }              }
4956            break;            break;
# Line 5008  for (;;) Line 4965  for (;;)
4965                break;                break;
4966                }                }
4967              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4968              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
             if ((prop_script == prop_value) == prop_fail_result)  
               break;  
4969              eptr+= len;              eptr+= len;
4970              }              }
4971            break;            break;
# Line 5018  for (;;) Line 4973  for (;;)
4973            case PT_ALNUM:            case PT_ALNUM:
4974            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4975              {              {
4976                int category;
4977              int len = 1;              int len = 1;
4978              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4979                {                {
# Line 5025  for (;;) Line 4981  for (;;)
4981                break;                break;
4982                }                }
4983              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4984              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4985              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                  == prop_fail_result)  
4986                break;                break;
4987              eptr+= len;              eptr+= len;
4988              }              }
# Line 5043  for (;;) Line 4998  for (;;)
4998                break;                break;
4999                }                }
5000              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5001              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
5002                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
5003                   == prop_fail_result)                   == prop_fail_result)
5004                break;                break;
# Line 5062  for (;;) Line 5016  for (;;)
5016                break;                break;
5017                }                }
5018              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5019              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
5020                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5021                   == prop_fail_result)                   == prop_fail_result)
5022                break;                break;
# Line 5074  for (;;) Line 5027  for (;;)
5027            case PT_WORD:            case PT_WORD:
5028            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5029              {              {
5030                int category;
5031              int len = 1;              int len = 1;
5032              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5033                {                {
# Line 5081  for (;;) Line 5035  for (;;)
5035                break;                break;
5036                }                }
5037              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5038              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5039              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N ||
5040                   c == CHAR_UNDERSCORE) == prop_fail_result)                   c == CHAR_UNDERSCORE) == prop_fail_result)
5041                break;                break;
5042              eptr+= len;              eptr+= len;
# Line 5112  for (;;) Line 5066  for (;;)
5066          {          {
5067          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5068            {            {
5069              int len = 1;
5070            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5071              {              {
5072              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5073              break;              break;
5074              }              }
5075            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5076            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5077            if (prop_category == ucp_M) break;            eptr += len;
5078            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5079              {              {
5080              int len = 1;              len = 1;
5081              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5082                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5083              eptr += len;              eptr += len;
5084              }              }
5085            }            }
# Line 5144  for (;;) Line 5095  for (;;)
5095            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5096            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5097              {              {
             int len = 1;  
5098              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5099                {                {
5100                BACKCHAR(eptr);                BACKCHAR(eptr);
5101                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5102                }                }
5103              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5104              eptr--;              eptr--;
5105              }              }
5106            }            }
# Line 5688  switch (frame->Xwhere) Line 5637  switch (frame->Xwhere)
5637    LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)    LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)
5638    LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)    LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)
5639    LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)    LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)
5640    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63)
5641  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5642    LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)    LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)
5643    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
# Line 5790  pcre_exec(const pcre *argument_re, const Line 5739  pcre_exec(const pcre *argument_re, const
5739    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5740    int offsetcount)    int offsetcount)
5741  {  {
5742  int rc, resetcount, ocount;  int rc, ocount;
5743  int first_byte = -1;  int first_byte = -1;
5744  int req_byte = -1;  int req_byte = -1;
5745  int req_byte2 = -1;  int req_byte2 = -1;
# Line 5898  utf8 = md->utf8 = (re->options & PCRE_UT Line 5847  utf8 = md->utf8 = (re->options & PCRE_UT
5847  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
5848  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
5849    
5850    /* Some options are unpacked into BOOL variables in the hope that testing
5851    them will be faster than individual option bits. */
5852    
5853  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
5854  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
5855  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
5856  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
5857  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5858                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5859    
5860    
5861  md->hitend = FALSE;  md->hitend = FALSE;
5862  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
5863    
# Line 6034  md->offset_max = (2*ocount)/3; Line 5988  md->offset_max = (2*ocount)/3;
5988  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
5989  md->capture_last = -1;  md->capture_last = -1;
5990    
 /* Compute the minimum number of offsets that we need to reset each time. Doing  
 this makes a huge difference to execution time when there aren't many brackets  
 in the pattern. */  
   
 resetcount = 2 + re->top_bracket * 2;  
 if (resetcount > offsetcount) resetcount = ocount;  
   
5991  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
5992  never be used unless previously set, but they get saved and restored, and so we  never be used unless previously set, but they get saved and restored, and so we
5993  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
5994    offsets for the matched string. This is really just for tidiness with callouts,
5995    in case they inspect these fields. */
5996    
5997  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
5998    {    {
5999    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6000    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6001      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6002    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6003      md->offset_vector[0] = md->offset_vector[1] = -1;
6004    }    }
6005    
6006  /* 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 6083  if ((re->flags & PCRE_REQCHSET) != 0) Line 6034  if ((re->flags & PCRE_REQCHSET) != 0)
6034    }    }
6035    
6036    
6037    
6038    
6039  /* ==========================================================================*/  /* ==========================================================================*/
6040    
6041  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 6093  for(;;) Line 6046  for(;;)
6046    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6047    USPTR new_start_match;    USPTR new_start_match;
6048    
   /* Reset the maximum number of extractions we might see. */  
   
   if (md->offset_vector != NULL)  
     {  
     register int *iptr = md->offset_vector;  
     register int *iend = iptr + resetcount;  
     while (iptr < iend) *iptr++ = -1;  
     }  
   
6049    /* If firstline is TRUE, the start of the match is constrained to the first    /* If firstline is TRUE, the start of the match is constrained to the first
6050    line of a multiline string. That is, the match must be before or at the first    line of a multiline string. That is, the match must be before or at the first
6051    newline. Implement this by temporarily adjusting end_subject so that we stop    newline. Implement this by temporarily adjusting end_subject so that we stop
# Line 6291  for(;;) Line 6235  for(;;)
6235    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6236    md->match_call_count = 0;    md->match_call_count = 0;
6237    md->match_function_type = 0;    md->match_function_type = 0;
6238      md->end_offset_top = 0;
6239    rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);    rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6240    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
6241    

Legend:
Removed from v.609  
changed lines
  Added in v.623

  ViewVC Help
Powered by ViewVC 1.1.5