/[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 619 by ph10, Sun Jul 17 13:23:14 2011 UTC revision 835 by ph10, Wed Dec 28 16:10:09 2011 UTC
# Line 57  possible. There are also some static sup Line 57  possible. There are also some static sup
57  #undef min  #undef min
58  #undef max  #undef max
59    
60  /* Values for setting in md->match_function_type to indicate two special types  /* Values for setting in md->match_function_type to indicate two special types
61  of call to match(). We do it this way to save on using another stack variable,  of call to match(). We do it this way to save on using another stack variable,
62  as stack usage is to be discouraged. */  as stack usage is to be discouraged. */
63    
64  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
# Line 138  while (length-- > 0) Line 138  while (length-- > 0)
138    
139  /* Normally, if a back reference hasn't been set, the length that is passed is  /* Normally, if a back reference hasn't been set, the length that is passed is
140  negative, so the match always fails. However, in JavaScript compatibility mode,  negative, so the match always fails. However, in JavaScript compatibility mode,
141  the length passed is zero. Note that in caseless UTF-8 mode, the number of  the length passed is zero. Note that in caseless UTF-8 mode, the number of
142  subject bytes matched may be different to the number of reference bytes.  subject bytes matched may be different to the number of reference bytes.
143    
144  Arguments:  Arguments:
# Line 185  if (caseless) Line 185  if (caseless)
185  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
186    if (md->utf8)    if (md->utf8)
187      {      {
188      /* Match characters up to the end of the reference. NOTE: the number of      /* Match characters up to the end of the reference. NOTE: the number of
189      bytes matched may differ, because there are some characters whose upper and      bytes matched may differ, because there are some characters whose upper and
190      lower case versions code as different numbers of bytes. For example, U+023A      lower case versions code as different numbers of bytes. For example, U+023A
191      (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);      (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
192      a sequence of 3 of the former uses 6 bytes, as does a sequence of two of      a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
193      the latter. It is important, therefore, to check the length along the      the latter. It is important, therefore, to check the length along the
194      reference, not along the subject (earlier code did this wrong). */      reference, not along the subject (earlier code did this wrong). */
195    
196      USPTR endptr = p + length;      USPTR endptr = p + length;
197      while (p < endptr)      while (p < endptr)
198        {        {
# Line 210  if (caseless) Line 210  if (caseless)
210    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
211    is no UCP support. */    is no UCP support. */
212      {      {
213      if (eptr + length > md->end_subject) return -1;      if (eptr + length > md->end_subject) return -1;
214      while (length-- > 0)      while (length-- > 0)
215        { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }        { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
216      }      }
217    }    }
218    
219  /* In the caseful case, we can just compare the bytes, whether or not we  /* In the caseful case, we can just compare the bytes, whether or not we
220  are in UTF-8 mode. */  are in UTF-8 mode. */
221    
222  else  else
223    {    {
224    if (eptr + length > md->end_subject) return -1;    if (eptr + length > md->end_subject) return -1;
225    while (length-- > 0) if (*p++ != *eptr++) return -1;    while (length-- > 0) if (*p++ != *eptr++) return -1;
226    }    }
227    
228  return eptr - eptr_start;  return eptr - eptr_start;
# Line 384  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 478  Returns:       MATCH_MATCH if matched Line 475  Returns:       MATCH_MATCH if matched
475    
476  static int  static int
477  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
478    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
479    unsigned int rdepth)    unsigned int rdepth)
480  {  {
481  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
# Line 551  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 591  declarations can be cut out in a block. Line 585  declarations can be cut out in a block.
585  below are for variables that do not have to be preserved over a recursive call  below are for variables that do not have to be preserved over a recursive call
586  to RMATCH(). */  to RMATCH(). */
587    
588  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
589  const uschar *charptr;  const uschar *charptr;
590  #endif  #endif
591  const uschar *callpat;  const uschar *callpat;
592  const uschar *data;  const uschar *data;
593  const uschar *next;  const uschar *next;
594  USPTR         pp;  USPTR         pp;
595  const uschar *prev;  const uschar *prev;
596  USPTR         saved_eptr;  USPTR         saved_eptr;
597    
598  recursion_info new_recursive;  recursion_info new_recursive;
599    
600  BOOL cur_is_word;  BOOL cur_is_word;
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
# Line 611  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 633  int stacksave[REC_STACK_SAVE_MAX]; Line 624  int stacksave[REC_STACK_SAVE_MAX];
624  eptrblock newptrb;  eptrblock newptrb;
625  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
626    
627  /* To save space on the stack and in the heap frame, I have doubled up on some  /* To save space on the stack and in the heap frame, I have doubled up on some
628  of the local variables that are used only in localised parts of the code, but  of the local variables that are used only in localised parts of the code, but
629  still need to be preserved over recursive calls of match(). These macros define  still need to be preserved over recursive calls of match(). These macros define
630  the alternative names that are used. */  the alternative names that are used. */
631    
632  #define allow_zero    cur_is_word  #define allow_zero    cur_is_word
# Line 681  if (md->match_call_count++ >= md->match_ Line 672  if (md->match_call_count++ >= md->match_
672  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
673    
674  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
675  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
676  done this way to save having to use another function argument, which would take  done this way to save having to use another function argument, which would take
677  up space on the stack. See also MATCH_CONDASSERT below.  up space on the stack. See also MATCH_CONDASSERT below.
678    
679  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
# Line 706  for (;;) Line 697  for (;;)
697    {    {
698    minimize = possessive = FALSE;    minimize = possessive = FALSE;
699    op = *ecode;    op = *ecode;
700    
701    switch(op)    switch(op)
702      {      {
703      case OP_MARK:      case OP_MARK:
# Line 784  for (;;) Line 775  for (;;)
775      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
776      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
777    
778      /* For THEN (and THEN_ARG) we pass back the address of the bracket or      /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
779      the alt that is at the start of the current branch. This makes it possible      the branch in which it occurs can be determined. Overload the start of
780      to skip back past alternatives that precede the THEN within the current      match pointer to do this. */
     branch. */  
781    
782      case OP_THEN:      case OP_THEN:
783      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
784        eptrb, RM54);        eptrb, RM54);
785      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
786      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
787      MRRETURN(MATCH_THEN);      MRRETURN(MATCH_THEN);
788    
789      case OP_THEN_ARG:      case OP_THEN_ARG:
790      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top,
791        offset_top, md, eptrb, RM58);        md, eptrb, RM58);
792      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
793      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
794      md->mark = ecode + LINK_SIZE + 2;      md->mark = ecode + 2;
795      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
796    
797        /* Handle an atomic group that does not contain any capturing parentheses.
798        This can be handled like an assertion. Prior to 8.13, all atomic groups
799        were handled this way. In 8.13, the code was changed as below for ONCE, so
800        that backups pass through the group and thereby reset captured values.
801        However, this uses a lot more stack, so in 8.20, atomic groups that do not
802        contain any captures generate OP_ONCE_NC, which can be handled in the old,
803        less stack intensive way.
804    
805        Check the alternative branches in turn - the matching won't pass the KET
806        for this kind of subpattern. If any one branch matches, we carry on as at
807        the end of a normal bracket, leaving the subject pointer, but resetting
808        the start-of-match value in case it was changed by \K. */
809    
810        case OP_ONCE_NC:
811        prev = ecode;
812        saved_eptr = eptr;
813        do
814          {
815          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
816          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
817            {
818            mstart = md->start_match_ptr;
819            markptr = md->mark;
820            break;
821            }
822          if (rrc == MATCH_THEN)
823            {
824            next = ecode + GET(ecode,1);
825            if (md->start_match_ptr < next &&
826                (*ecode == OP_ALT || *next == OP_ALT))
827              rrc = MATCH_NOMATCH;
828            }
829    
830          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
831          ecode += GET(ecode,1);
832          }
833        while (*ecode == OP_ALT);
834    
835        /* If hit the end of the group (which could be repeated), fail */
836    
837        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
838    
839        /* Continue as from after the group, updating the offsets high water
840        mark, since extracts may have been taken. */
841    
842        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
843    
844        offset_top = md->end_offset_top;
845        eptr = md->end_match_ptr;
846    
847        /* For a non-repeating ket, just continue at this level. This also
848        happens for a repeating ket if no characters were matched in the group.
849        This is the forcible breaking of infinite loops as implemented in Perl
850        5.005. */
851    
852        if (*ecode == OP_KET || eptr == saved_eptr)
853          {
854          ecode += 1+LINK_SIZE;
855          break;
856          }
857    
858        /* The repeating kets try the rest of the pattern or restart from the
859        preceding bracket, in the appropriate order. The second "call" of match()
860        uses tail recursion, to avoid using another stack frame. */
861    
862        if (*ecode == OP_KETRMIN)
863          {
864          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
865          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
866          ecode = prev;
867          goto TAIL_RECURSE;
868          }
869        else  /* OP_KETRMAX */
870          {
871          md->match_function_type = MATCH_CBEGROUP;
872          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
873          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
874          ecode += 1 + LINK_SIZE;
875          goto TAIL_RECURSE;
876          }
877        /* Control never gets here */
878    
879      /* Handle a capturing bracket, other than those that are possessive with an      /* Handle a capturing bracket, other than those that are possessive with an
880      unlimited repeat. If there is space in the offset vector, save the current      unlimited repeat. If there is space in the offset vector, save the current
881      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
882      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
883      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
884      the group. A failure to match might occur after the group has succeeded,      the group. A failure to match might occur after the group has succeeded,
885      if something later on doesn't match. For this reason, we need to restore      if something later on doesn't match. For this reason, we need to restore
886      the working value and also the values of the final offsets, in case they      the working value and also the values of the final offsets, in case they
887      were set by a previous iteration of the same bracket.      were set by a previous iteration of the same bracket.
# Line 822  for (;;) Line 894  for (;;)
894      case OP_SCBRA:      case OP_SCBRA:
895      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
896      offset = number << 1;      offset = number << 1;
897    
898  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
899      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
900      printf("subject=");      printf("subject=");
# Line 843  for (;;) Line 915  for (;;)
915    
916        for (;;)        for (;;)
917          {          {
918          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
919          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
920            eptrb, RM1);            eptrb, RM1);
921          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
922          if (rrc != MATCH_NOMATCH &&  
923              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* If we backed up to a THEN, check whether it is within the current
924            RRETURN(rrc);          branch by comparing the address of the THEN that is passed back with
925            the end of the branch. If it is within the current branch, and the
926            branch is one of two or more alternatives (it either starts or ends
927            with OP_ALT), we have reached the limit of THEN's action, so convert
928            the return code to NOMATCH, which will cause normal backtracking to
929            happen from now on. Otherwise, THEN is passed back to an outer
930            alternative. This implements Perl's treatment of parenthesized groups,
931            where a group not containing | does not affect the current alternative,
932            that is, (X) is NOT the same as (X|(*F)). */
933    
934            if (rrc == MATCH_THEN)
935              {
936              next = ecode + GET(ecode,1);
937              if (md->start_match_ptr < next &&
938                  (*ecode == OP_ALT || *next == OP_ALT))
939                rrc = MATCH_NOMATCH;
940              }
941    
942            /* Anything other than NOMATCH is passed back. */
943    
944            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
945          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
946          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
947          if (*ecode != OP_ALT) break;          if (*ecode != OP_ALT) break;
948          }          }
949    
950        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
951        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
952        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
953        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
   
       /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or  
       MATCH_THEN. */  
954    
955        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
956        RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));  
957          if (md->mark == NULL) md->mark = markptr;
958          RRETURN(rrc);
959        }        }
960    
961      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 879  for (;;) Line 970  for (;;)
970      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
971    
972      /* Non-capturing or atomic group, except for possessive with unlimited      /* Non-capturing or atomic group, except for possessive with unlimited
973      repeat. Loop for all the alternatives. When we get to the final alternative      repeat and ONCE group with no captures. Loop for all the alternatives.
974      within the brackets, we used to return the result of a recursive call to  
975      match() whatever happened so it was possible to reduce stack usage by      When we get to the final alternative within the brackets, we used to return
976      turning this into a tail recursion, except in the case of a possibly empty      the result of a recursive call to match() whatever happened so it was
977      group. However, now that there is the possiblity of (*THEN) occurring in      possible to reduce stack usage by turning this into a tail recursion,
978      the final alternative, this optimization is no longer possible.      except in the case of a possibly empty group. However, now that there is
979        the possiblity of (*THEN) occurring in the final alternative, this
980      MATCH_ONCE is returned when the end of an atomic group is successfully      optimization is no longer always possible.
981      reached, but subsequent matching fails. It passes back up the tree (causing  
982      captured values to be reset) until the original atomic group level is      We can optimize if we know there are no (*THEN)s in the pattern; at present
983        this is the best that can be done.
984    
985        MATCH_ONCE is returned when the end of an atomic group is successfully
986        reached, but subsequent matching fails. It passes back up the tree (causing
987        captured values to be reset) until the original atomic group level is
988      reached. This is tested by comparing md->once_target with the start of the      reached. This is tested by comparing md->once_target with the start of the
989      group. At this point, the return is converted into MATCH_NOMATCH so that      group. At this point, the return is converted into MATCH_NOMATCH so that
990      previous backup points can be taken. */      previous backup points can be taken. */
# Line 901  for (;;) Line 997  for (;;)
997      for (;;)      for (;;)
998        {        {
999        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
1000        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,  
1001          /* If this is not a possibly empty group, and there are no (*THEN)s in
1002          the pattern, and this is the final alternative, optimize as described
1003          above. */
1004    
1005          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1006            {
1007            ecode += _pcre_OP_lengths[*ecode];
1008            goto TAIL_RECURSE;
1009            }
1010    
1011          /* In all other cases, we have to make another call to match(). */
1012    
1013          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
1014          RM2);          RM2);
1015        if (rrc != MATCH_NOMATCH &&  
1016            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* See comment in the code for capturing groups above about handling
1017          {        THEN. */
1018    
1019          if (rrc == MATCH_THEN)
1020            {
1021            next = ecode + GET(ecode,1);
1022            if (md->start_match_ptr < next &&
1023                (*ecode == OP_ALT || *next == OP_ALT))
1024              rrc = MATCH_NOMATCH;
1025            }
1026    
1027          if (rrc != MATCH_NOMATCH)
1028            {
1029          if (rrc == MATCH_ONCE)          if (rrc == MATCH_ONCE)
1030            {            {
1031            const uschar *scode = ecode;            const uschar *scode = ecode;
# Line 913  for (;;) Line 1033  for (;;)
1033              {              {
1034              while (*scode == OP_ALT) scode += GET(scode, 1);              while (*scode == OP_ALT) scode += GET(scode, 1);
1035              scode -= GET(scode, 1);              scode -= GET(scode, 1);
1036              }              }
1037            if (md->once_target == scode) rrc = MATCH_NOMATCH;            if (md->once_target == scode) rrc = MATCH_NOMATCH;
1038            }            }
1039          RRETURN(rrc);          RRETURN(rrc);
1040          }          }
1041        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1042        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1043        }        }
1044      if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;  
1045        if (md->mark == NULL) md->mark = markptr;
1046      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1047    
1048      /* Handle possessive capturing brackets with an unlimited repeat. We come      /* Handle possessive capturing brackets with an unlimited repeat. We come
1049      here from BRAZERO with allow_zero set TRUE. The offset_vector values are      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1050      handled similarly to the normal case above. However, the matching is      handled similarly to the normal case above. However, the matching is
1051      different. The end of these brackets will always be OP_KETRPOS, which      different. The end of these brackets will always be OP_KETRPOS, which
1052      returns MATCH_KETRPOS without going further in the pattern. By this means      returns MATCH_KETRPOS without going further in the pattern. By this means
1053      we can handle the group by iteration rather than recursion, thereby      we can handle the group by iteration rather than recursion, thereby
1054      reducing the amount of stack needed. */      reducing the amount of stack needed. */
1055    
1056      case OP_CBRAPOS:      case OP_CBRAPOS:
1057      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1058      allow_zero = FALSE;      allow_zero = FALSE;
1059    
1060      POSSESSIVE_CAPTURE:      POSSESSIVE_CAPTURE:
1061      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
1062      offset = number << 1;      offset = number << 1;
# Line 950  for (;;) Line 1071  for (;;)
1071      if (offset < md->offset_max)      if (offset < md->offset_max)
1072        {        {
1073        matched_once = FALSE;        matched_once = FALSE;
1074        code_offset = ecode - md->start_code;        code_offset = ecode - md->start_code;
1075    
1076        save_offset1 = md->offset_vector[offset];        save_offset1 = md->offset_vector[offset];
1077        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
# Line 958  for (;;) Line 1079  for (;;)
1079        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
1080    
1081        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1082    
1083        /* Each time round the loop, save the current subject position for use        /* Each time round the loop, save the current subject position for use
1084        when the group matches. For MATCH_MATCH, the group has matched, so we        when the group matches. For MATCH_MATCH, the group has matched, so we
1085        restart it with a new subject starting position, remembering that we had        restart it with a new subject starting position, remembering that we had
1086        at least one match. For MATCH_NOMATCH, carry on with the alternatives, as        at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1087        usual. If we haven't matched any alternatives in any iteration, check to        usual. If we haven't matched any alternatives in any iteration, check to
1088        see if a previous iteration matched. If so, the group has matched;        see if a previous iteration matched. If so, the group has matched;
1089        continue from afterwards. Otherwise it has failed; restore the previous        continue from afterwards. Otherwise it has failed; restore the previous
1090        capture values before returning NOMATCH. */        capture values before returning NOMATCH. */
1091    
1092        for (;;)        for (;;)
1093          {          {
1094          md->offset_vector[md->offset_end - number] =          md->offset_vector[md->offset_end - number] =
1095            (int)(eptr - md->start_subject);            (int)(eptr - md->start_subject);
1096          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1097          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1098            eptrb, RM63);            eptrb, RM63);
1099          if (rrc == MATCH_KETRPOS)          if (rrc == MATCH_KETRPOS)
1100            {            {
1101            offset_top = md->end_offset_top;            offset_top = md->end_offset_top;
1102            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
1103            ecode = md->start_code + code_offset;            ecode = md->start_code + code_offset;
1104            save_capture_last = md->capture_last;            save_capture_last = md->capture_last;
1105            matched_once = TRUE;            matched_once = TRUE;
1106            continue;            continue;
1107            }            }
1108          if (rrc != MATCH_NOMATCH &&  
1109              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* See comment in the code for capturing groups above about handling
1110            RRETURN(rrc);          THEN. */
1111    
1112            if (rrc == MATCH_THEN)
1113              {
1114              next = ecode + GET(ecode,1);
1115              if (md->start_match_ptr < next &&
1116                  (*ecode == OP_ALT || *next == OP_ALT))
1117                rrc = MATCH_NOMATCH;
1118              }
1119    
1120            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1121          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
1122          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
1123          if (*ecode != OP_ALT) break;          if (*ecode != OP_ALT) break;
1124          }          }
1125    
1126        if (!matched_once)        if (!matched_once)
1127          {          {
1128          md->offset_vector[offset] = save_offset1;          md->offset_vector[offset] = save_offset1;
1129          md->offset_vector[offset+1] = save_offset2;          md->offset_vector[offset+1] = save_offset2;
1130          md->offset_vector[md->offset_end - number] = save_offset3;          md->offset_vector[md->offset_end - number] = save_offset3;
1131          }          }
1132    
1133        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;        if (md->mark == NULL) md->mark = markptr;
1134        if (allow_zero || matched_once)        if (allow_zero || matched_once)
1135          {          {
1136          ecode += 1 + LINK_SIZE;          ecode += 1 + LINK_SIZE;
1137          break;          break;
1138          }          }
1139    
1140        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1141        }        }
1142    
1143      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1144      as a non-capturing bracket. */      as a non-capturing bracket. */
1145    
# Line 1020  for (;;) Line 1151  for (;;)
1151      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1152      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1153    
1154      /* Non-capturing possessive bracket with unlimited repeat. We come here      /* Non-capturing possessive bracket with unlimited repeat. We come here
1155      from BRAZERO with allow_zero = TRUE. The code is similar to the above,      from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1156      without the capturing complication. It is written out separately for speed      without the capturing complication. It is written out separately for speed
1157      and cleanliness. */      and cleanliness. */
1158    
1159      case OP_BRAPOS:      case OP_BRAPOS:
1160      case OP_SBRAPOS:      case OP_SBRAPOS:
1161      allow_zero = FALSE;      allow_zero = FALSE;
1162    
1163      POSSESSIVE_NON_CAPTURE:      POSSESSIVE_NON_CAPTURE:
1164      matched_once = FALSE;      matched_once = FALSE;
1165      code_offset = ecode - md->start_code;      code_offset = ecode - md->start_code;
1166    
1167      for (;;)      for (;;)
1168        {        {
1169        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1170        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1171          eptrb, RM48);          eptrb, RM48);
1172        if (rrc == MATCH_KETRPOS)        if (rrc == MATCH_KETRPOS)
1173          {          {
1174          offset_top = md->end_offset_top;          offset_top = md->end_offset_top;
1175          eptr = md->end_match_ptr;          eptr = md->end_match_ptr;
1176          ecode = md->start_code + code_offset;          ecode = md->start_code + code_offset;
1177          matched_once = TRUE;          matched_once = TRUE;
1178          continue;          continue;
1179          }          }
1180        if (rrc != MATCH_NOMATCH &&  
1181            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* See comment in the code for capturing groups above about handling
1182          RRETURN(rrc);        THEN. */
1183    
1184          if (rrc == MATCH_THEN)
1185            {
1186            next = ecode + GET(ecode,1);
1187            if (md->start_match_ptr < next &&
1188                (*ecode == OP_ALT || *next == OP_ALT))
1189              rrc = MATCH_NOMATCH;
1190            }
1191    
1192          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1193        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1194        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1195        }        }
1196    
1197      if (matched_once || allow_zero)      if (matched_once || allow_zero)
1198        {        {
1199        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1200        break;        break;
1201        }        }
1202      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1203    
1204      /* Control never reaches here. */      /* Control never reaches here. */
# Line 1079  for (;;) Line 1220  for (;;)
1220        if (pcre_callout != NULL)        if (pcre_callout != NULL)
1221          {          {
1222          pcre_callout_block cb;          pcre_callout_block cb;
1223          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1224          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1225          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1226          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 1091  for (;;) Line 1232  for (;;)
1232          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1233          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1234          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1235            cb.mark             = markptr;
1236          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1237          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1238          }          }
# Line 1111  for (;;) Line 1253  for (;;)
1253        else        else
1254          {          {
1255          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1256          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1257    
1258          /* If the test is for recursion into a specific subpattern, and it is          /* If the test is for recursion into a specific subpattern, and it is
1259          false, but the test was set up by name, scan the table to see if the          false, but the test was set up by name, scan the table to see if the
1260          name refers to any other numbers, and test them. The condition is true          name refers to any other numbers, and test them. The condition is true
1261          if any one is set. */          if any one is set. */
1262    
1263          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1264            {            {
1265            uschar *slotA = md->name_table;            uschar *slotA = md->name_table;
1266            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
# Line 1249  for (;;) Line 1391  for (;;)
1391    
1392      else      else
1393        {        {
1394        md->match_function_type = MATCH_CONDASSERT;        md->match_function_type = MATCH_CONDASSERT;
1395        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1396        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1397          {          {
# Line 1259  for (;;) Line 1401  for (;;)
1401          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1402          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1403          }          }
1404        else if (rrc != MATCH_NOMATCH &&  
1405                (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1406          assertion; it is therefore treated as NOMATCH. */
1407    
1408          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1409          {          {
1410          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1411          }          }
# Line 1271  for (;;) Line 1416  for (;;)
1416          }          }
1417        }        }
1418    
1419      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one, can
1420      we used to use tail recursion to avoid using another stack frame, except      use tail recursion to avoid using another stack frame, except when there is
1421      when there was unlimited repeat of a possibly empty group. However, that      unlimited repeat of a possibly empty group. In the latter case, a recursive
1422      strategy no longer works because of the possibilty of (*THEN) being      call to match() is always required, unless the second alternative doesn't
1423      encountered in the branch. A recursive call to match() is always required,      exist, in which case we can just plough on. Note that, for compatibility
1424      unless the second alternative doesn't exist, in which case we can just      with Perl, the | in a conditional group is NOT treated as creating two
1425      plough on. */      alternatives. If a THEN is encountered in the branch, it propagates out to
1426        the enclosing alternative (unless nested in a deeper set of alternatives,
1427        of course). */
1428    
1429      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1430        {        {
1431        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;        if (op != OP_SCOND)
1432            {
1433            ecode += 1 + LINK_SIZE;
1434            goto TAIL_RECURSE;
1435            }
1436    
1437          md->match_function_type = MATCH_CBEGROUP;
1438        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
       if (rrc == MATCH_THEN && md->start_match_ptr == ecode)  
         rrc = MATCH_NOMATCH;  
1439        RRETURN(rrc);        RRETURN(rrc);
1440        }        }
1441      else                         /* Condition false & no alternative */  
1442         /* Condition false & no alternative; continue after the group. */
1443    
1444        else
1445        {        {
1446        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1447        }        }
# Line 1322  for (;;) Line 1476  for (;;)
1476    
1477      case OP_END:      case OP_END:
1478      case OP_ACCEPT:      case OP_ACCEPT:
1479      case OP_ASSERT_ACCEPT:      case OP_ASSERT_ACCEPT:
1480    
1481      /* If we have matched an empty string, fail if not in an assertion and not      /* If we have matched an empty string, fail if not in an assertion and not
1482      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1483      is set and we have matched at the start of the subject. In both cases,      is set and we have matched at the start of the subject. In both cases,
# Line 1337  for (;;) Line 1491  for (;;)
1491        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1492    
1493      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1494    
1495      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1496      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1497      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
# Line 1352  for (;;) Line 1506  for (;;)
1506      matching won't pass the KET for an assertion. If any one branch matches,      matching won't pass the KET for an assertion. If any one branch matches,
1507      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
1508      start of each branch to move the current point backwards, so the code at      start of each branch to move the current point backwards, so the code at
1509      this level is identical to the lookahead case. When the assertion is part      this level is identical to the lookahead case. When the assertion is part
1510      of a condition, we want to return immediately afterwards. The caller of      of a condition, we want to return immediately afterwards. The caller of
1511      this incarnation of the match() function will have set MATCH_CONDASSERT in      this incarnation of the match() function will have set MATCH_CONDASSERT in
1512      md->match_function type, and one of these opcodes will be the first opcode      md->match_function type, and one of these opcodes will be the first opcode
1513      that is processed. We use a local variable that is preserved over calls to      that is processed. We use a local variable that is preserved over calls to
1514      match() to remember this case. */      match() to remember this case. */
1515    
1516      case OP_ASSERT:      case OP_ASSERT:
# Line 1366  for (;;) Line 1520  for (;;)
1520        condassert = TRUE;        condassert = TRUE;
1521        md->match_function_type = 0;        md->match_function_type = 0;
1522        }        }
1523      else condassert = FALSE;      else condassert = FALSE;
1524    
1525      do      do
1526        {        {
1527        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1528        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1529          {          {
1530          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1531            markptr = md->mark;
1532          break;          break;
1533          }          }
1534        if (rrc != MATCH_NOMATCH &&  
1535            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1536          RRETURN(rrc);        as NOMATCH. */
1537    
1538          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1539        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1540        }        }
1541      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1542    
1543      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1544    
1545      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
# Line 1408  for (;;) Line 1565  for (;;)
1565        condassert = TRUE;        condassert = TRUE;
1566        md->match_function_type = 0;        md->match_function_type = 0;
1567        }        }
1568      else condassert = FALSE;      else condassert = FALSE;
1569    
1570      do      do
1571        {        {
# Line 1419  for (;;) Line 1576  for (;;)
1576          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1577          break;          break;
1578          }          }
1579        if (rrc != MATCH_NOMATCH &&  
1580            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1581          RRETURN(rrc);        as NOMATCH. */
1582    
1583          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1584        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1585        }        }
1586      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1587    
1588      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1589    
1590      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1591      continue;      continue;
1592    
# Line 1472  for (;;) Line 1631  for (;;)
1631      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1632        {        {
1633        pcre_callout_block cb;        pcre_callout_block cb;
1634        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1635        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1636        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1637        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 1484  for (;;) Line 1643  for (;;)
1643        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1644        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1645        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1646          cb.mark             = markptr;
1647        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1648        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1649        }        }
# Line 1493  for (;;) Line 1653  for (;;)
1653      /* Recursion either matches the current regex, or some subexpression. The      /* Recursion either matches the current regex, or some subexpression. The
1654      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
1655      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1656    
1657      The state of the capturing groups is preserved over recursion, and      The state of the capturing groups is preserved over recursion, and
1658      re-instated afterwards. We don't know how many are started and not yet      re-instated afterwards. We don't know how many are started and not yet
1659      finished (offset_top records the completed total) so we just have to save      finished (offset_top records the completed total) so we just have to save
1660      all the potential data. There may be up to 65535 such values, which is too      all the potential data. There may be up to 65535 such values, which is too
1661      large to put on the stack, but using malloc for small numbers seems      large to put on the stack, but using malloc for small numbers seems
# Line 1504  for (;;) Line 1664  for (;;)
1664    
1665      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
1666      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
1667      for the original version of this logic. It has, however, been hacked around      for the original version of this logic. It has, however, been hacked around
1668      a lot, so he is not to blame for the current way it works. */      a lot, so he is not to blame for the current way it works. */
1669    
1670      case OP_RECURSE:      case OP_RECURSE:
1671        {        {
1672          recursion_info *ri;
1673          int recno;
1674    
1675        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1676        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1677          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1678    
1679          /* Check for repeating a recursion without advancing the subject pointer.
1680          This should catch convoluted mutual recursions. (Some simple cases are
1681          caught at compile time.) */
1682    
1683          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1684            if (recno == ri->group_num && eptr == ri->subject_position)
1685              RRETURN(PCRE_ERROR_RECURSELOOP);
1686    
1687        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1688    
1689          new_recursive.group_num = recno;
1690          new_recursive.subject_position = eptr;
1691        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1692        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1693    
# Line 1535  for (;;) Line 1708  for (;;)
1708          }          }
1709        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1710              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1711    
1712        /* OK, now we can do the recursion. After processing each alternative,        /* OK, now we can do the recursion. After processing each alternative,
1713        restore the offset data. If there were nested recursions, md->recursive        restore the offset data. If there were nested recursions, md->recursive
1714        might be changed, so reset it before looping. */        might be changed, so reset it before looping. */
1715    
1716        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
# Line 1549  for (;;) Line 1722  for (;;)
1722            md, eptrb, RM6);            md, eptrb, RM6);
1723          memcpy(md->offset_vector, new_recursive.offset_save,          memcpy(md->offset_vector, new_recursive.offset_save,
1724              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1725            md->recursive = new_recursive.prevrec;
1726          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1727            {            {
1728            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1729            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1730              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1731    
1732            /* Set where we got to in the subject, and reset the start in case            /* Set where we got to in the subject, and reset the start in case
1733            it was changed by \K. This *is* propagated back out of a recursion,            it was changed by \K. This *is* propagated back out of a recursion,
1734            for Perl compatibility. */            for Perl compatibility. */
1735    
1736            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
1737            mstart = md->start_match_ptr;            mstart = md->start_match_ptr;
1738            goto RECURSION_MATCHED;        /* Exit loop; end processing */            goto RECURSION_MATCHED;        /* Exit loop; end processing */
1739            }            }
1740          else if (rrc != MATCH_NOMATCH &&  
1741                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1742            as NOMATCH. */
1743    
1744            else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1745            {            {
1746            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1747            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
# Line 1584  for (;;) Line 1760  for (;;)
1760          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1761        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1762        }        }
1763    
1764      RECURSION_MATCHED:      RECURSION_MATCHED:
1765      break;      break;
1766    
# Line 1600  for (;;) Line 1776  for (;;)
1776      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1777      with fixed upper repeat limits are compiled as a number of copies, with the      with fixed upper repeat limits are compiled as a number of copies, with the
1778      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1779    
1780      case OP_BRAZERO:      case OP_BRAZERO:
1781      next = ecode + 1;      next = ecode + 1;
1782      RMATCH(eptr, next, offset_top, md, eptrb, RM10);      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
# Line 1608  for (;;) Line 1784  for (;;)
1784      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
1785      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1786      break;      break;
1787    
1788      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1789      next = ecode + 1;      next = ecode + 1;
1790      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
# Line 1622  for (;;) Line 1798  for (;;)
1798      do next += GET(next,1); while (*next == OP_ALT);      do next += GET(next,1); while (*next == OP_ALT);
1799      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1800      break;      break;
1801    
1802      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1803      here; just jump to the group, with allow_zero set TRUE. */      here; just jump to the group, with allow_zero set TRUE. */
1804    
1805      case OP_BRAPOSZERO:      case OP_BRAPOSZERO:
1806      op = *(++ecode);      op = *(++ecode);
1807      allow_zero = TRUE;      allow_zero = TRUE;
1808      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1809        goto POSSESSIVE_NON_CAPTURE;        goto POSSESSIVE_NON_CAPTURE;
# Line 1637  for (;;) Line 1813  for (;;)
1813      case OP_KET:      case OP_KET:
1814      case OP_KETRMIN:      case OP_KETRMIN:
1815      case OP_KETRMAX:      case OP_KETRMAX:
1816      case OP_KETRPOS:      case OP_KETRPOS:
1817      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1818    
1819      /* 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
1820      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1821      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
# Line 1651  for (;;) Line 1827  for (;;)
1827        }        }
1828      else saved_eptr = NULL;      else saved_eptr = NULL;
1829    
1830      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group or a non-capturing atomic
1831      MATCH_MATCH, but record the current high water mark for use by positive      group, stop matching and return MATCH_MATCH, but record the current high
1832      assertions. We also need to record the match start in case it was changed      water mark for use by positive assertions. We also need to record the match
1833      by \K. */      start in case it was changed by \K. */
1834    
1835      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1836          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)           *prev == OP_ONCE_NC)
1837        {        {
1838        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1839        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1840        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1841        MRRETURN(MATCH_MATCH);        MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1842        }        }
1843    
1844      /* 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
# Line 1698  for (;;) Line 1874  for (;;)
1874        md->capture_last = number;        md->capture_last = number;
1875        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1876          {          {
1877          /* If offset is greater than offset_top, it means that we are          /* If offset is greater than offset_top, it means that we are
1878          "skipping" a capturing group, and that group's offsets must be marked          "skipping" a capturing group, and that group's offsets must be marked
1879          unset. In earlier versions of PCRE, all the offsets were unset at the          unset. In earlier versions of PCRE, all the offsets were unset at the
1880          start of matching, but this doesn't work because atomic groups and          start of matching, but this doesn't work because atomic groups and
1881          assertions can cause a value to be set that should later be unset.          assertions can cause a value to be set that should later be unset.
1882          Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as          Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1883          part of the atomic group, but this is not on the final matching path,          part of the atomic group, but this is not on the final matching path,
1884          so must be unset when 2 is set. (If there is no group 2, there is no          so must be unset when 2 is set. (If there is no group 2, there is no
1885          problem, because offset_top will then be 2, indicating no capture.) */          problem, because offset_top will then be 2, indicating no capture.) */
1886    
1887          if (offset > offset_top)          if (offset > offset_top)
1888            {            {
1889            register int *iptr = md->offset_vector + offset_top;            register int *iptr = md->offset_vector + offset_top;
1890            register int *iend = md->offset_vector + offset;            register int *iend = md->offset_vector + offset;
1891            while (iptr < iend) *iptr++ = -1;            while (iptr < iend) *iptr++ = -1;
1892            }            }
1893    
1894          /* Now make the extraction */          /* Now make the extraction */
1895    
1896          md->offset_vector[offset] =          md->offset_vector[offset] =
# Line 1727  for (;;) Line 1903  for (;;)
1903      /* For an ordinary non-repeating ket, just continue at this level. This      /* For an ordinary non-repeating ket, just continue at this level. This
1904      also happens for a repeating ket if no characters were matched in the      also happens for a repeating ket if no characters were matched in the
1905      group. This is the forcible breaking of infinite loops as implemented in      group. This is the forcible breaking of infinite loops as implemented in
1906      Perl 5.005. For a non-repeating atomic group, establish a backup point by      Perl 5.005. For a non-repeating atomic group that includes captures,
1907      processing the rest of the pattern at a lower level. If this results in a      establish a backup point by processing the rest of the pattern at a lower
1908      NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby      level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1909      bypassing intermediate backup points, but resetting any captures that      original OP_ONCE level, thereby bypassing intermediate backup points, but
1910      happened along the way. */      resetting any captures that happened along the way. */
1911    
1912      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1913        {        {
# Line 1740  for (;;) Line 1916  for (;;)
1916          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1917          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1918          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1919          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
1920          }          }
1921        ecode += 1 + LINK_SIZE;    /* Carry on at this level */        ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1922        break;        break;
1923        }        }
1924    
1925      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1926      and return the MATCH_KETRPOS. This makes it possible to do the repeats one      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1927      at a time from the outer level, thus saving stack. */      at a time from the outer level, thus saving stack. */
1928    
1929      if (*ecode == OP_KETRPOS)      if (*ecode == OP_KETRPOS)
1930        {        {
1931        md->end_match_ptr = eptr;        md->end_match_ptr = eptr;
1932        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1933        RRETURN(MATCH_KETRPOS);        RRETURN(MATCH_KETRPOS);
1934        }        }
1935    
1936      /* 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
1937      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
# Line 1765  for (;;) Line 1941  for (;;)
1941    
1942      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1943        {        {
1944        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1945        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1946        if (*prev == OP_ONCE)        if (*prev == OP_ONCE)
1947          {          {
1948          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1949          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1950          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1951          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
1952          }          }
1953        if (*prev >= OP_SBRA)    /* Could match an empty string */        if (*prev >= OP_SBRA)    /* Could match an empty string */
1954          {          {
1955          md->match_function_type = MATCH_CBEGROUP;          md->match_function_type = MATCH_CBEGROUP;
1956          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1957          RRETURN(rrc);          RRETURN(rrc);
1958          }          }
# Line 1785  for (;;) Line 1961  for (;;)
1961        }        }
1962      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1963        {        {
1964        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1965        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1966        if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;        if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1967        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1968        if (*prev == OP_ONCE)        if (*prev == OP_ONCE)
1969          {          {
1970          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1971          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1972          md->once_target = prev;          md->once_target = prev;
1973          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
1974          }          }
1975        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1976        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1977        }        }
# Line 1805  for (;;) Line 1981  for (;;)
1981    
1982      case OP_CIRC:      case OP_CIRC:
1983      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1984    
1985      /* Start of subject assertion */      /* Start of subject assertion */
1986    
1987      case OP_SOD:      case OP_SOD:
1988      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1989      ecode++;      ecode++;
1990      break;      break;
1991    
1992      /* Multiline mode: start of subject unless notbol, or after any newline. */      /* Multiline mode: start of subject unless notbol, or after any newline. */
1993    
1994      case OP_CIRCM:      case OP_CIRCM:
# Line 1851  for (;;) Line 2027  for (;;)
2027      ecode++;      ecode++;
2028      break;      break;
2029    
2030      /* Not multiline mode: assert before a terminating newline or before end of      /* Not multiline mode: assert before a terminating newline or before end of
2031      subject unless noteol is set. */      subject unless noteol is set. */
2032    
2033      case OP_DOLL:      case OP_DOLL:
# Line 2007  for (;;) Line 2183  for (;;)
2183      /* Fall through */      /* Fall through */
2184    
2185      case OP_ALLANY:      case OP_ALLANY:
2186      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2187        {        {                            /* not be updated before SCHECK_PARTIAL. */
2188        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2189        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2190        }        }
2191        eptr++;
2192      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2193      ecode++;      ecode++;
2194      break;      break;
# Line 2020  for (;;) Line 2197  for (;;)
2197      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2198    
2199      case OP_ANYBYTE:      case OP_ANYBYTE:
2200      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2201        {        {                            /* not be updated before SCHECK_PARTIAL. */
2202        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2203        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2204        }        }
2205        eptr++;
2206      ecode++;      ecode++;
2207      break;      break;
2208    
# Line 2140  for (;;) Line 2318  for (;;)
2318      switch(c)      switch(c)
2319        {        {
2320        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2321    
2322        case 0x000d:        case 0x000d:
2323        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2324        break;        break;
# Line 2364  for (;;) Line 2542  for (;;)
2542        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2543        }        }
2544      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2545        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2546        while (eptr < md->end_subject)
2547        {        {
2548        int category = UCD_CATEGORY(c);        int len = 1;
2549        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2550        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2551          {        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;  
         }  
2552        }        }
2553      ecode++;      ecode++;
2554      break;      break;
# Line 2393  for (;;) Line 2564  for (;;)
2564      loops). */      loops). */
2565    
2566      case OP_REF:      case OP_REF:
2567      case OP_REFI:      case OP_REFI:
2568      caseless = op == OP_REFI;      caseless = op == OP_REFI;
2569      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2570      ecode += 3;      ecode += 3;
2571    
# Line 2462  for (;;) Line 2633  for (;;)
2633    
2634      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2635        {        {
2636        int slength;        int slength;
2637        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2638          {          {
2639          CHECK_PARTIAL();          CHECK_PARTIAL();
# Line 2482  for (;;) Line 2653  for (;;)
2653        {        {
2654        for (fi = min;; fi++)        for (fi = min;; fi++)
2655          {          {
2656          int slength;          int slength;
2657          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2658          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2659          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 2503  for (;;) Line 2674  for (;;)
2674        pp = eptr;        pp = eptr;
2675        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2676          {          {
2677          int slength;          int slength;
2678          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2679            {            {
2680            CHECK_PARTIAL();            CHECK_PARTIAL();
# Line 3236  for (;;) Line 3407  for (;;)
3407      checking can be multibyte. */      checking can be multibyte. */
3408    
3409      case OP_NOT:      case OP_NOT:
3410      case OP_NOTI:      case OP_NOTI:
3411      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3412        {        {
3413        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 3731  for (;;) Line 3902  for (;;)
3902            case PT_LAMP:            case PT_LAMP:
3903            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3904              {              {
3905                int chartype;
3906              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3907                {                {
3908                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3909                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3910                }                }
3911              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3912              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
3913              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
3914                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
3915                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
3916                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3917              }              }
3918            break;            break;
# Line 3754  for (;;) Line 3926  for (;;)
3926                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3927                }                }
3928              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3929              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
3930                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3931              }              }
3932            break;            break;
# Line 3769  for (;;) Line 3940  for (;;)
3940                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3941                }                }
3942              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3943              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
3944                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3945              }              }
3946            break;            break;
# Line 3784  for (;;) Line 3954  for (;;)
3954                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3955                }                }
3956              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3957              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
3958                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3959              }              }
3960            break;            break;
# Line 3793  for (;;) Line 3962  for (;;)
3962            case PT_ALNUM:            case PT_ALNUM:
3963            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3964              {              {
3965                int category;
3966              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3967                {                {
3968                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3969                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3970                }                }
3971              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3972              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3973              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
3974                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3975              }              }
3976            break;            break;
# Line 3815  for (;;) Line 3984  for (;;)
3984                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3985                }                }
3986              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3987              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 ||  
3988                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
3989                     == prop_fail_result)                     == prop_fail_result)
3990                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3832  for (;;) Line 4000  for (;;)
4000                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4001                }                }
4002              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4003              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 ||  
4004                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4005                     == prop_fail_result)                     == prop_fail_result)
4006                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3843  for (;;) Line 4010  for (;;)
4010            case PT_WORD:            case PT_WORD:
4011            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4012              {              {
4013                int category;
4014              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4015                {                {
4016                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4017                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4018                }                }
4019              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4020              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4021              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
                  c == CHAR_UNDERSCORE)  
4022                     == prop_fail_result)                     == prop_fail_result)
4023                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4024              }              }
# Line 3877  for (;;) Line 4044  for (;;)
4044              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4045              }              }
4046            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4047            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4048            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4049              {              {
4050              int len = 1;              int len = 1;
4051              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4052                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4053              eptr += len;              eptr += len;
4054              }              }
4055            }            }
# Line 3943  for (;;) Line 4107  for (;;)
4107            switch(c)            switch(c)
4108              {              {
4109              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4110    
4111              case 0x000d:              case 0x000d:
4112              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4113              break;              break;
# Line 4220  for (;;) Line 4384  for (;;)
4384            switch(*eptr++)            switch(*eptr++)
4385              {              {
4386              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4387    
4388              case 0x000d:              case 0x000d:
4389              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4390              break;              break;
4391    
4392              case 0x000a:              case 0x000a:
4393              break;              break;
4394    
# Line 4430  for (;;) Line 4594  for (;;)
4594            case PT_LAMP:            case PT_LAMP:
4595            for (fi = min;; fi++)            for (fi = min;; fi++)
4596              {              {
4597                int chartype;
4598              RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4599              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4600              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4439  for (;;) Line 4604  for (;;)
4604                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4605                }                }
4606              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4607              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4608              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4609                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4610                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4611                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4612              }              }
4613            /* Control never gets here */            /* Control never gets here */
# Line 4459  for (;;) Line 4624  for (;;)
4624                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4625                }                }
4626              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4627              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4628                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4629              }              }
4630            /* Control never gets here */            /* Control never gets here */
# Line 4477  for (;;) Line 4641  for (;;)
4641                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4642                }                }
4643              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4644              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4645                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4646              }              }
4647            /* Control never gets here */            /* Control never gets here */
# Line 4495  for (;;) Line 4658  for (;;)
4658                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4659                }                }
4660              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4661              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
4662                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4663              }              }
4664            /* Control never gets here */            /* Control never gets here */
# Line 4504  for (;;) Line 4666  for (;;)
4666            case PT_ALNUM:            case PT_ALNUM:
4667            for (fi = min;; fi++)            for (fi = min;; fi++)
4668              {              {
4669                int category;
4670              RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4671              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4672              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4513  for (;;) Line 4676  for (;;)
4676                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4677                }                }
4678              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4679              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4680              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
4681                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4682              }              }
4683            /* Control never gets here */            /* Control never gets here */
# Line 4532  for (;;) Line 4694  for (;;)
4694                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4695                }                }
4696              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4697              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 ||  
4698                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
4699                     == prop_fail_result)                     == prop_fail_result)
4700                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4552  for (;;) Line 4713  for (;;)
4713                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4714                }                }
4715              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4716              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 ||  
4717                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4718                     == prop_fail_result)                     == prop_fail_result)
4719                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4563  for (;;) Line 4723  for (;;)
4723            case PT_WORD:            case PT_WORD:
4724            for (fi = min;; fi++)            for (fi = min;; fi++)
4725              {              {
4726                int category;
4727              RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4728              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4729              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4572  for (;;) Line 4733  for (;;)
4733                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4734                }                }
4735              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4736              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4737              if ((prop_category == ucp_L ||              if ((category == ucp_L ||
4738                   prop_category == ucp_N ||                   category == ucp_N ||
4739                   c == CHAR_UNDERSCORE)                   c == CHAR_UNDERSCORE)
4740                     == prop_fail_result)                     == prop_fail_result)
4741                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4604  for (;;) Line 4765  for (;;)
4765              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4766              }              }
4767            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4768            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4769            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4770              {              {
4771              int len = 1;              int len = 1;
4772              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4773                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4774              eptr += len;              eptr += len;
4775              }              }
4776            }            }
4777          }          }
   
4778        else        else
4779  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4780    
# Line 4938  for (;;) Line 5095  for (;;)
5095            case PT_LAMP:            case PT_LAMP:
5096            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5097              {              {
5098                int chartype;
5099              int len = 1;              int len = 1;
5100              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5101                {                {
# Line 4945  for (;;) Line 5103  for (;;)
5103                break;                break;
5104                }                }
5105              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5106              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
5107              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
5108                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
5109                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
5110                break;                break;
5111              eptr+= len;              eptr+= len;
5112              }              }
# Line 4964  for (;;) Line 5122  for (;;)
5122                break;                break;
5123                }                }
5124              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5125              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
5126              eptr+= len;              eptr+= len;
5127              }              }
5128            break;            break;
# Line 4981  for (;;) Line 5137  for (;;)
5137                break;                break;
5138                }                }
5139              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5140              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
5141              eptr+= len;              eptr+= len;
5142              }              }
5143            break;            break;
# Line 4998  for (;;) Line 5152  for (;;)
5152                break;                break;
5153                }                }
5154              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5155              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
             if ((prop_script == prop_value) == prop_fail_result)  
               break;  
5156              eptr+= len;              eptr+= len;
5157              }              }
5158            break;            break;
# Line 5008  for (;;) Line 5160  for (;;)
5160            case PT_ALNUM:            case PT_ALNUM:
5161            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5162              {              {
5163                int category;
5164              int len = 1;              int len = 1;
5165              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5166                {                {
# Line 5015  for (;;) Line 5168  for (;;)
5168                break;                break;
5169                }                }
5170              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5171              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5172              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                  == prop_fail_result)  
5173                break;                break;
5174              eptr+= len;              eptr+= len;
5175              }              }
# Line 5033  for (;;) Line 5185  for (;;)
5185                break;                break;
5186                }                }
5187              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5188              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 ||  
5189                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
5190                   == prop_fail_result)                   == prop_fail_result)
5191                break;                break;
# Line 5052  for (;;) Line 5203  for (;;)
5203                break;                break;
5204                }                }
5205              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5206              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 ||  
5207                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5208                   == prop_fail_result)                   == prop_fail_result)
5209                break;                break;
# Line 5064  for (;;) Line 5214  for (;;)
5214            case PT_WORD:            case PT_WORD:
5215            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5216              {              {
5217                int category;
5218              int len = 1;              int len = 1;
5219              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5220                {                {
# Line 5071  for (;;) Line 5222  for (;;)
5222                break;                break;
5223                }                }
5224              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5225              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5226              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N ||
5227                   c == CHAR_UNDERSCORE) == prop_fail_result)                   c == CHAR_UNDERSCORE) == prop_fail_result)
5228                break;                break;
5229              eptr+= len;              eptr+= len;
# Line 5102  for (;;) Line 5253  for (;;)
5253          {          {
5254          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5255            {            {
5256              int len = 1;
5257            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5258              {              {
5259              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5260              break;              break;
5261              }              }
5262            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5263            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5264            if (prop_category == ucp_M) break;            eptr += len;
5265            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5266              {              {
5267              int len = 1;              len = 1;
5268              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5269                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5270              eptr += len;              eptr += len;
5271              }              }
5272            }            }
# Line 5134  for (;;) Line 5282  for (;;)
5282            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5283            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5284              {              {
             int len = 1;  
5285              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5286                {                {
5287                BACKCHAR(eptr);                BACKCHAR(eptr);
5288                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5289                }                }
5290              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5291              eptr--;              eptr--;
5292              }              }
5293            }            }
# Line 5205  for (;;) Line 5351  for (;;)
5351                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
5352                }                }
5353              }              }
5354            else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */            else
5355                {
5356                eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
5357                SCHECK_PARTIAL();
5358                }
5359            break;            break;
5360    
5361            /* The byte case is the same as non-UTF8 */            /* The byte case is the same as non-UTF8 */
# Line 5426  for (;;) Line 5576  for (;;)
5576            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5577            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5578            BACKCHAR(eptr);            BACKCHAR(eptr);
5579            if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&            if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5580                eptr[-1] == '\r') eptr--;                eptr[-1] == '\r') eptr--;
5581            }            }
5582          }          }
# Line 5638  for (;;) Line 5788  for (;;)
5788            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
5789            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5790            eptr--;            eptr--;
5791            if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&            if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5792                eptr[-1] == '\r') eptr--;                eptr[-1] == '\r') eptr--;
5793            }            }
5794          }          }
# Line 5679  switch (frame->Xwhere) Line 5829  switch (frame->Xwhere)
5829    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)
5830    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)
5831    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64)
5832    LBL(65) LBL(66)    LBL(65) LBL(66)
5833  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5834    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)
5835    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
# Line 5781  pcre_exec(const pcre *argument_re, const Line 5931  pcre_exec(const pcre *argument_re, const
5931    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5932    int offsetcount)    int offsetcount)
5933  {  {
5934  int rc, ocount;  int rc, ocount, arg_offset_max;
5935  int first_byte = -1;  int first_byte = -1;
5936  int req_byte = -1;  int req_byte = -1;
5937  int req_byte2 = -1;  int req_byte2 = -1;
# Line 5817  if (re == NULL || subject == NULL || Line 5967  if (re == NULL || subject == NULL ||
5967  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
5968  if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;  if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
5969    
5970  /* This information is for finding all the numbers associated with a given  /* These two settings are used in the code for checking a UTF-8 string that
5971  name, for condition testing. */  follows immediately afterwards. Other values in the md block are used only
5972    during "normal" pcre_exec() processing, not when the JIT support is in use,
5973    so they are set up later. */
5974    
5975    utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;
5976    md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5977                  ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5978    
5979    /* Check a UTF-8 string if required. Pass back the character offset and error
5980    code for an invalid string if a results vector is available. */
5981    
5982    #ifdef SUPPORT_UTF8
5983    if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5984      {
5985      int erroroffset;
5986      int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5987      if (errorcode != 0)
5988        {
5989        if (offsetcount >= 2)
5990          {
5991          offsets[0] = erroroffset;
5992          offsets[1] = errorcode;
5993          }
5994        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5995          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5996        }
5997    
5998      /* Check that a start_offset points to the start of a UTF-8 character. */
5999      if (start_offset > 0 && start_offset < length &&
6000          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
6001        return PCRE_ERROR_BADUTF8_OFFSET;
6002      }
6003    #endif
6004    
6005    /* If the pattern was successfully studied with JIT support, run the JIT
6006    executable instead of the rest of this function. Most options must be set at
6007    compile time for the JIT code to be usable. Fallback to the normal code path if
6008    an unsupported flag is set. In particular, JIT does not support partial
6009    matching. */
6010    
6011    #ifdef SUPPORT_JIT
6012    if (extra_data != NULL
6013        && (extra_data->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0
6014        && extra_data->executable_jit != NULL
6015        && (extra_data->flags & PCRE_EXTRA_TABLES) == 0
6016        && (options & ~(PCRE_NO_UTF8_CHECK | PCRE_NOTBOL | PCRE_NOTEOL |
6017                        PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART)) == 0)
6018      return _pcre_jit_exec(re, extra_data->executable_jit, subject, length,
6019        start_offset, options, ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0)
6020        ? MATCH_LIMIT : extra_data->match_limit, offsets, offsetcount);
6021    #endif
6022    
6023    /* Carry on with non-JIT matching. This information is for finding all the
6024    numbers associated with a given name, for condition testing. */
6025    
6026  md->name_table = (uschar *)re + re->name_table_offset;  md->name_table = (uschar *)re + re->name_table_offset;
6027  md->name_count = re->name_count;  md->name_count = re->name_count;
# Line 5885  md->end_subject = md->start_subject + le Line 6088  md->end_subject = md->start_subject + le
6088  end_subject = md->end_subject;  end_subject = md->end_subject;
6089    
6090  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
 utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;  
6091  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
6092  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
6093    
# Line 5896  md->notbol = (options & PCRE_NOTBOL) != Line 6098  md->notbol = (options & PCRE_NOTBOL) !=
6098  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
6099  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
6100  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
6101  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  
               ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;  
   
   
6102  md->hitend = FALSE;  md->hitend = FALSE;
6103  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
6104    
6105  md->recursive = NULL;                   /* No recursion at top level */  md->recursive = NULL;                   /* No recursion at top level */
6106    md->hasthen = (re->flags & PCRE_HASTHEN) != 0;
6107    
6108  md->lcc = tables + lcc_offset;  md->lcc = tables + lcc_offset;
6109  md->ctypes = tables + ctypes_offset;  md->ctypes = tables + ctypes_offset;
# Line 5981  defined (though never set). So there's n Line 6181  defined (though never set). So there's n
6181  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
6182    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
6183    
 /* Check a UTF-8 string if required. Pass back the character offset and error  
 code for an invalid string if a results vector is available. */  
   
 #ifdef SUPPORT_UTF8  
 if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  
   {  
   int erroroffset;  
   int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);  
   if (errorcode != 0)  
     {  
     if (offsetcount >= 2)  
       {  
       offsets[0] = erroroffset;  
       offsets[1] = errorcode;  
       }  
     return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?  
       PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;  
     }  
   
   /* Check that a start_offset points to the start of a UTF-8 character. */  
   
   if (start_offset > 0 && start_offset < length &&  
       (((USPTR)subject)[start_offset] & 0xc0) == 0x80)  
     return PCRE_ERROR_BADUTF8_OFFSET;  
   }  
 #endif  
   
6184  /* If the expression has got more back references than the offsets supplied can  /* If the expression has got more back references than the offsets supplied can
6185  hold, we get a temporary chunk of working store to use during the matching.  hold, we get a temporary chunk of working store to use during the matching.
6186  Otherwise, we can use the vector supplied, rounding down its size to a multiple  Otherwise, we can use the vector supplied, rounding down its size to a multiple
6187  of 3. */  of 3. */
6188    
6189  ocount = offsetcount - (offsetcount % 3);  ocount = offsetcount - (offsetcount % 3);
6190    arg_offset_max = (2*ocount)/3;
6191    
6192  if (re->top_backref > 0 && re->top_backref >= ocount/3)  if (re->top_backref > 0 && re->top_backref >= ocount/3)
6193    {    {
# Line 6193  for(;;) Line 6367  for(;;)
6367    /* The following two optimizations are disabled for partial matching or if    /* The following two optimizations are disabled for partial matching or if
6368    disabling is explicitly requested. */    disabling is explicitly requested. */
6369    
6370    if ((options & PCRE_NO_START_OPTIMIZE) == 0 && !md->partial)    if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0 && !md->partial)
6371      {      {
6372      /* If the pattern was studied, a minimum subject length may be set. This is      /* If the pattern was studied, a minimum subject length may be set. This is
6373      a lower bound; no actual string of that length may actually match the      a lower bound; no actual string of that length may actually match the
# Line 6276  for(;;) Line 6450  for(;;)
6450    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6451    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6452    md->match_call_count = 0;    md->match_call_count = 0;
6453    md->match_function_type = 0;    md->match_function_type = 0;
6454    md->end_offset_top = 0;    md->end_offset_top = 0;
6455    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);
6456    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
# Line 6388  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6562  if (rc == MATCH_MATCH || rc == MATCH_ACC
6562    {    {
6563    if (using_temporary_offsets)    if (using_temporary_offsets)
6564      {      {
6565      if (offsetcount >= 4)      if (arg_offset_max >= 4)
6566        {        {
6567        memcpy(offsets + 2, md->offset_vector + 2,        memcpy(offsets + 2, md->offset_vector + 2,
6568          (offsetcount - 2) * sizeof(int));          (arg_offset_max - 2) * sizeof(int));
6569        DPRINTF(("Copied offsets from temporary memory\n"));        DPRINTF(("Copied offsets from temporary memory\n"));
6570        }        }
6571      if (md->end_offset_top > offsetcount) md->offset_overflow = TRUE;      if (md->end_offset_top > arg_offset_max) md->offset_overflow = TRUE;
6572      DPRINTF(("Freeing temporary memory\n"));      DPRINTF(("Freeing temporary memory\n"));
6573      (pcre_free)(md->offset_vector);      (pcre_free)(md->offset_vector);
6574      }      }
6575    
6576    /* Set the return code to the number of captured strings, or 0 if there are    /* Set the return code to the number of captured strings, or 0 if there were
6577    too many to fit into the vector. */    too many to fit into the vector. */
6578    
6579    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = (md->offset_overflow && md->end_offset_top >= arg_offset_max)?
6580        0 : md->end_offset_top/2;
6581    
6582      /* If there is space in the offset vector, set any unused pairs at the end of
6583      the pattern to -1 for backwards compatibility. It is documented that this
6584      happens. In earlier versions, the whole set of potential capturing offsets
6585      was set to -1 each time round the loop, but this is handled differently now.
6586      "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only
6587      those at the end that need unsetting here. We can't just unset them all at
6588      the start of the whole thing because they may get set in one branch that is
6589      not the final matching branch. */
6590    
6591      if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL)
6592        {
6593        register int *iptr, *iend;
6594        int resetcount = 2 + re->top_bracket * 2;
6595        if (resetcount > offsetcount) resetcount = ocount;
6596        iptr = offsets + md->end_offset_top;
6597        iend = offsets + resetcount;
6598        while (iptr < iend) *iptr++ = -1;
6599        }
6600    
6601    /* If there is space, set up the whole thing as substring 0. The value of    /* If there is space, set up the whole thing as substring 0. The value of
6602    md->start_match_ptr might be modified if \K was encountered on the success    md->start_match_ptr might be modified if \K was encountered on the success

Legend:
Removed from v.619  
changed lines
  Added in v.835

  ViewVC Help
Powered by ViewVC 1.1.5