/[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 623 by ph10, Tue Jul 19 09:58:42 2011 UTC revision 625 by ph10, Wed Jul 20 16:46:19 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 475  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 585  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 624  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 672  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 697  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 800  for (;;) Line 800  for (;;)
800      subject position in the working slot at the top of the vector. We mustn't      subject position in the working slot at the top of the vector. We mustn't
801      change the current values of the data slot, because they may be set from a      change the current values of the data slot, because they may be set from a
802      previous iteration of this group, and be referred to by a reference inside      previous iteration of this group, and be referred to by a reference inside
803      the group. A failure to match might occur after the group has succeeded,      the group. A failure to match might occur after the group has succeeded,
804      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
805      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
806      were set by a previous iteration of the same bracket.      were set by a previous iteration of the same bracket.
# Line 813  for (;;) Line 813  for (;;)
813      case OP_SCBRA:      case OP_SCBRA:
814      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
815      offset = number << 1;      offset = number << 1;
816    
817  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
818      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
819      printf("subject=");      printf("subject=");
# Line 834  for (;;) Line 834  for (;;)
834    
835        for (;;)        for (;;)
836          {          {
837          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839            eptrb, RM1);            eptrb, RM1);
840          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841          if (rrc != MATCH_NOMATCH &&          if (rrc != MATCH_NOMATCH &&
# Line 843  for (;;) Line 843  for (;;)
843            RRETURN(rrc);            RRETURN(rrc);
844          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
845          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
846          if (*ecode != OP_ALT) break;          if (*ecode != OP_ALT) break;
847          }          }
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855        MATCH_THEN. */        MATCH_THEN. */
856    
857        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
# Line 875  for (;;) Line 875  for (;;)
875      match() whatever happened so it was possible to reduce stack usage by      match() whatever happened so it was possible to reduce stack usage by
876      turning this into a tail recursion, except in the case of a possibly empty      turning this into a tail recursion, except in the case of a possibly empty
877      group. However, now that there is the possiblity of (*THEN) occurring in      group. However, now that there is the possiblity of (*THEN) occurring in
878      the final alternative, this optimization is no longer possible.      the final alternative, this optimization is no longer possible.
879    
880      MATCH_ONCE is returned when the end of an atomic group is successfully      MATCH_ONCE is returned when the end of an atomic group is successfully
881      reached, but subsequent matching fails. It passes back up the tree (causing      reached, but subsequent matching fails. It passes back up the tree (causing
882      captured values to be reset) until the original atomic group level is      captured values to be reset) until the original atomic group level is
883      reached. This is tested by comparing md->once_target with the start of the      reached. This is tested by comparing md->once_target with the start of the
884      group. At this point, the return is converted into MATCH_NOMATCH so that      group. At this point, the return is converted into MATCH_NOMATCH so that
885      previous backup points can be taken. */      previous backup points can be taken. */
# Line 892  for (;;) Line 892  for (;;)
892      for (;;)      for (;;)
893        {        {
894        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;
895        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
896          RM2);          RM2);
897        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
898            (rrc != MATCH_THEN || md->start_match_ptr != ecode))            (rrc != MATCH_THEN || md->start_match_ptr != ecode))
899          {          {
900          if (rrc == MATCH_ONCE)          if (rrc == MATCH_ONCE)
901            {            {
902            const uschar *scode = ecode;            const uschar *scode = ecode;
# Line 904  for (;;) Line 904  for (;;)
904              {              {
905              while (*scode == OP_ALT) scode += GET(scode, 1);              while (*scode == OP_ALT) scode += GET(scode, 1);
906              scode -= GET(scode, 1);              scode -= GET(scode, 1);
907              }              }
908            if (md->once_target == scode) rrc = MATCH_NOMATCH;            if (md->once_target == scode) rrc = MATCH_NOMATCH;
909            }            }
910          RRETURN(rrc);          RRETURN(rrc);
911          }          }
912        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
913        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
914        }        }
915      if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;      if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
916      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
917    
918      /* Handle possessive capturing brackets with an unlimited repeat. We come      /* Handle possessive capturing brackets with an unlimited repeat. We come
919      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
920      handled similarly to the normal case above. However, the matching is      handled similarly to the normal case above. However, the matching is
921      different. The end of these brackets will always be OP_KETRPOS, which      different. The end of these brackets will always be OP_KETRPOS, which
922      returns MATCH_KETRPOS without going further in the pattern. By this means      returns MATCH_KETRPOS without going further in the pattern. By this means
923      we can handle the group by iteration rather than recursion, thereby      we can handle the group by iteration rather than recursion, thereby
924      reducing the amount of stack needed. */      reducing the amount of stack needed. */
925    
926      case OP_CBRAPOS:      case OP_CBRAPOS:
927      case OP_SCBRAPOS:      case OP_SCBRAPOS:
928      allow_zero = FALSE;      allow_zero = FALSE;
929    
930      POSSESSIVE_CAPTURE:      POSSESSIVE_CAPTURE:
931      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
932      offset = number << 1;      offset = number << 1;
# Line 941  for (;;) Line 941  for (;;)
941      if (offset < md->offset_max)      if (offset < md->offset_max)
942        {        {
943        matched_once = FALSE;        matched_once = FALSE;
944        code_offset = ecode - md->start_code;        code_offset = ecode - md->start_code;
945    
946        save_offset1 = md->offset_vector[offset];        save_offset1 = md->offset_vector[offset];
947        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
# Line 949  for (;;) Line 949  for (;;)
949        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
950    
951        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
952    
953        /* Each time round the loop, save the current subject position for use        /* Each time round the loop, save the current subject position for use
954        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
955        restart it with a new subject starting position, remembering that we had        restart it with a new subject starting position, remembering that we had
956        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
957        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
958        see if a previous iteration matched. If so, the group has matched;        see if a previous iteration matched. If so, the group has matched;
959        continue from afterwards. Otherwise it has failed; restore the previous        continue from afterwards. Otherwise it has failed; restore the previous
960        capture values before returning NOMATCH. */        capture values before returning NOMATCH. */
961    
962        for (;;)        for (;;)
963          {          {
964          md->offset_vector[md->offset_end - number] =          md->offset_vector[md->offset_end - number] =
965            (int)(eptr - md->start_subject);            (int)(eptr - md->start_subject);
966          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
967          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
968            eptrb, RM63);            eptrb, RM63);
969          if (rrc == MATCH_KETRPOS)          if (rrc == MATCH_KETRPOS)
970            {            {
971            offset_top = md->end_offset_top;            offset_top = md->end_offset_top;
972            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
973            ecode = md->start_code + code_offset;            ecode = md->start_code + code_offset;
974            save_capture_last = md->capture_last;            save_capture_last = md->capture_last;
975            matched_once = TRUE;            matched_once = TRUE;
976            continue;            continue;
977            }            }
978          if (rrc != MATCH_NOMATCH &&          if (rrc != MATCH_NOMATCH &&
979              (rrc != MATCH_THEN || md->start_match_ptr != ecode))              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
980            RRETURN(rrc);            RRETURN(rrc);
981          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
982          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
983          if (*ecode != OP_ALT) break;          if (*ecode != OP_ALT) break;
984          }          }
985    
986        if (!matched_once)        if (!matched_once)
987          {          {
988          md->offset_vector[offset] = save_offset1;          md->offset_vector[offset] = save_offset1;
989          md->offset_vector[offset+1] = save_offset2;          md->offset_vector[offset+1] = save_offset2;
990          md->offset_vector[md->offset_end - number] = save_offset3;          md->offset_vector[md->offset_end - number] = save_offset3;
991          }          }
992    
993        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
994        if (allow_zero || matched_once)        if (allow_zero || matched_once)
995          {          {
996          ecode += 1 + LINK_SIZE;          ecode += 1 + LINK_SIZE;
997          break;          break;
998          }          }
999    
1000        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1001        }        }
1002    
1003      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1004      as a non-capturing bracket. */      as a non-capturing bracket. */
1005    
# Line 1011  for (;;) Line 1011  for (;;)
1011      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1012      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1013    
1014      /* Non-capturing possessive bracket with unlimited repeat. We come here      /* Non-capturing possessive bracket with unlimited repeat. We come here
1015      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,
1016      without the capturing complication. It is written out separately for speed      without the capturing complication. It is written out separately for speed
1017      and cleanliness. */      and cleanliness. */
1018    
1019      case OP_BRAPOS:      case OP_BRAPOS:
1020      case OP_SBRAPOS:      case OP_SBRAPOS:
1021      allow_zero = FALSE;      allow_zero = FALSE;
1022    
1023      POSSESSIVE_NON_CAPTURE:      POSSESSIVE_NON_CAPTURE:
1024      matched_once = FALSE;      matched_once = FALSE;
1025      code_offset = ecode - md->start_code;      code_offset = ecode - md->start_code;
1026    
1027      for (;;)      for (;;)
1028        {        {
1029        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1030        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1031          eptrb, RM48);          eptrb, RM48);
1032        if (rrc == MATCH_KETRPOS)        if (rrc == MATCH_KETRPOS)
1033          {          {
1034          offset_top = md->end_offset_top;          offset_top = md->end_offset_top;
1035          eptr = md->end_match_ptr;          eptr = md->end_match_ptr;
1036          ecode = md->start_code + code_offset;          ecode = md->start_code + code_offset;
1037          matched_once = TRUE;          matched_once = TRUE;
1038          continue;          continue;
1039          }          }
1040        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
1041            (rrc != MATCH_THEN || md->start_match_ptr != ecode))            (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1042          RRETURN(rrc);          RRETURN(rrc);
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1045        }        }
1046    
1047      if (matched_once || allow_zero)      if (matched_once || allow_zero)
1048        {        {
1049        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1050        break;        break;
1051        }        }
1052      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1053    
1054      /* Control never reaches here. */      /* Control never reaches here. */
# Line 1240  for (;;) Line 1240  for (;;)
1240    
1241      else      else
1242        {        {
1243        md->match_function_type = MATCH_CONDASSERT;        md->match_function_type = MATCH_CONDASSERT;
1244        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1245        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1246          {          {
# Line 1264  for (;;) Line 1264  for (;;)
1264    
1265      /* 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,
1266      we used to use tail recursion to avoid using another stack frame, except      we used to use tail recursion to avoid using another stack frame, except
1267      when there was unlimited repeat of a possibly empty group. However, that      when there was unlimited repeat of a possibly empty group. However, that
1268      strategy no longer works because of the possibilty of (*THEN) being      strategy no longer works because of the possibilty of (*THEN) being
1269      encountered in the branch. A recursive call to match() is always required,      encountered in the branch. A recursive call to match() is always required,
1270      unless the second alternative doesn't exist, in which case we can just      unless the second alternative doesn't exist, in which case we can just
1271      plough on. */      plough on. */
1272    
1273      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1274        {        {
1275        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1276        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1277        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1278          rrc = MATCH_NOMATCH;          rrc = MATCH_NOMATCH;
1279        RRETURN(rrc);        RRETURN(rrc);
1280        }        }
# Line 1313  for (;;) Line 1313  for (;;)
1313    
1314      case OP_END:      case OP_END:
1315      case OP_ACCEPT:      case OP_ACCEPT:
1316      case OP_ASSERT_ACCEPT:      case OP_ASSERT_ACCEPT:
1317    
1318      /* 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
1319      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
1320      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 1328  for (;;) Line 1328  for (;;)
1328        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1329    
1330      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1331    
1332      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1333      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1334      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
# Line 1343  for (;;) Line 1343  for (;;)
1343      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,
1344      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
1345      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
1346      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
1347      of a condition, we want to return immediately afterwards. The caller of      of a condition, we want to return immediately afterwards. The caller of
1348      this incarnation of the match() function will have set MATCH_CONDASSERT in      this incarnation of the match() function will have set MATCH_CONDASSERT in
1349      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
1350      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
1351      match() to remember this case. */      match() to remember this case. */
1352    
1353      case OP_ASSERT:      case OP_ASSERT:
# Line 1357  for (;;) Line 1357  for (;;)
1357        condassert = TRUE;        condassert = TRUE;
1358        md->match_function_type = 0;        md->match_function_type = 0;
1359        }        }
1360      else condassert = FALSE;      else condassert = FALSE;
1361    
1362      do      do
1363        {        {
1364        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
# Line 1373  for (;;) Line 1373  for (;;)
1373        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1374        }        }
1375      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1376    
1377      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1378    
1379      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
# Line 1399  for (;;) Line 1399  for (;;)
1399        condassert = TRUE;        condassert = TRUE;
1400        md->match_function_type = 0;        md->match_function_type = 0;
1401        }        }
1402      else condassert = FALSE;      else condassert = FALSE;
1403    
1404      do      do
1405        {        {
# Line 1418  for (;;) Line 1418  for (;;)
1418      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1419    
1420      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1421    
1422      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1423      continue;      continue;
1424    
# Line 1484  for (;;) Line 1484  for (;;)
1484      /* Recursion either matches the current regex, or some subexpression. The      /* Recursion either matches the current regex, or some subexpression. The
1485      offset data is the offset to the starting bracket from the start of the      offset data is the offset to the starting bracket from the start of the
1486      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1487    
1488      The state of the capturing groups is preserved over recursion, and      The state of the capturing groups is preserved over recursion, and
1489      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
1490      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
1491      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
1492      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 1495  for (;;) Line 1495  for (;;)
1495    
1496      There are also other values that have to be saved. We use a chained      There are also other values that have to be saved. We use a chained
1497      sequence of blocks that actually live on the stack. Thanks to Robin Houston      sequence of blocks that actually live on the stack. Thanks to Robin Houston
1498      for the original version of this logic. It has, however, been hacked around      for the original version of this logic. It has, however, been hacked around
1499      a lot, so he is not to blame for the current way it works. */      a lot, so he is not to blame for the current way it works. */
1500    
1501      case OP_RECURSE:      case OP_RECURSE:
# Line 1526  for (;;) Line 1526  for (;;)
1526          }          }
1527        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1528              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1529    
1530        /* OK, now we can do the recursion. After processing each alternative,        /* OK, now we can do the recursion. After processing each alternative,
1531        restore the offset data. If there were nested recursions, md->recursive        restore the offset data. If there were nested recursions, md->recursive
1532        might be changed, so reset it before looping. */        might be changed, so reset it before looping. */
1533    
1534        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
# Line 1548  for (;;) Line 1548  for (;;)
1548              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1549    
1550            /* 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
1551            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,
1552            for Perl compatibility. */            for Perl compatibility. */
1553    
1554            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
1555            mstart = md->start_match_ptr;            mstart = md->start_match_ptr;
1556            goto RECURSION_MATCHED;        /* Exit loop; end processing */            goto RECURSION_MATCHED;        /* Exit loop; end processing */
# Line 1575  for (;;) Line 1575  for (;;)
1575          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1576        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1577        }        }
1578    
1579      RECURSION_MATCHED:      RECURSION_MATCHED:
1580      break;      break;
1581    
# Line 1591  for (;;) Line 1591  for (;;)
1591      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
1592      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
1593      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1594    
1595      case OP_BRAZERO:      case OP_BRAZERO:
1596      next = ecode + 1;      next = ecode + 1;
1597      RMATCH(eptr, next, offset_top, md, eptrb, RM10);      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
# Line 1599  for (;;) Line 1599  for (;;)
1599      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
1600      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1601      break;      break;
1602    
1603      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1604      next = ecode + 1;      next = ecode + 1;
1605      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
# Line 1613  for (;;) Line 1613  for (;;)
1613      do next += GET(next,1); while (*next == OP_ALT);      do next += GET(next,1); while (*next == OP_ALT);
1614      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1615      break;      break;
1616    
1617      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1618      here; just jump to the group, with allow_zero set TRUE. */      here; just jump to the group, with allow_zero set TRUE. */
1619    
1620      case OP_BRAPOSZERO:      case OP_BRAPOSZERO:
1621      op = *(++ecode);      op = *(++ecode);
1622      allow_zero = TRUE;      allow_zero = TRUE;
1623      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1624        goto POSSESSIVE_NON_CAPTURE;        goto POSSESSIVE_NON_CAPTURE;
# Line 1628  for (;;) Line 1628  for (;;)
1628      case OP_KET:      case OP_KET:
1629      case OP_KETRMIN:      case OP_KETRMIN:
1630      case OP_KETRMAX:      case OP_KETRMAX:
1631      case OP_KETRPOS:      case OP_KETRPOS:
1632      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1633    
1634      /* If this was a group that remembered the subject start, in order to break      /* If this was a group that remembered the subject start, in order to break
1635      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1636      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
# Line 1689  for (;;) Line 1689  for (;;)
1689        md->capture_last = number;        md->capture_last = number;
1690        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1691          {          {
1692          /* If offset is greater than offset_top, it means that we are          /* If offset is greater than offset_top, it means that we are
1693          "skipping" a capturing group, and that group's offsets must be marked          "skipping" a capturing group, and that group's offsets must be marked
1694          unset. In earlier versions of PCRE, all the offsets were unset at the          unset. In earlier versions of PCRE, all the offsets were unset at the
1695          start of matching, but this doesn't work because atomic groups and          start of matching, but this doesn't work because atomic groups and
1696          assertions can cause a value to be set that should later be unset.          assertions can cause a value to be set that should later be unset.
1697          Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as          Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1698          part of the atomic group, but this is not on the final matching path,          part of the atomic group, but this is not on the final matching path,
1699          so must be unset when 2 is set. (If there is no group 2, there is no          so must be unset when 2 is set. (If there is no group 2, there is no
1700          problem, because offset_top will then be 2, indicating no capture.) */          problem, because offset_top will then be 2, indicating no capture.) */
1701    
1702          if (offset > offset_top)          if (offset > offset_top)
1703            {            {
1704            register int *iptr = md->offset_vector + offset_top;            register int *iptr = md->offset_vector + offset_top;
1705            register int *iend = md->offset_vector + offset;            register int *iend = md->offset_vector + offset;
1706            while (iptr < iend) *iptr++ = -1;            while (iptr < iend) *iptr++ = -1;
1707            }            }
1708    
1709          /* Now make the extraction */          /* Now make the extraction */
1710    
1711          md->offset_vector[offset] =          md->offset_vector[offset] =
# Line 1718  for (;;) Line 1718  for (;;)
1718      /* For an ordinary non-repeating ket, just continue at this level. This      /* For an ordinary non-repeating ket, just continue at this level. This
1719      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
1720      group. This is the forcible breaking of infinite loops as implemented in      group. This is the forcible breaking of infinite loops as implemented in
1721      Perl 5.005. For a non-repeating atomic group, establish a backup point by      Perl 5.005. For a non-repeating atomic group, establish a backup point by
1722      processing the rest of the pattern at a lower level. If this results in a      processing the rest of the pattern at a lower level. If this results in a
1723      NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby      NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1724      bypassing intermediate backup points, but resetting any captures that      bypassing intermediate backup points, but resetting any captures that
1725      happened along the way. */      happened along the way. */
1726    
1727      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
# Line 1731  for (;;) Line 1731  for (;;)
1731          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1732          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1733          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1734          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
1735          }          }
1736        ecode += 1 + LINK_SIZE;    /* Carry on at this level */        ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1737        break;        break;
1738        }        }
1739    
1740      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1741      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
1742      at a time from the outer level, thus saving stack. */      at a time from the outer level, thus saving stack. */
1743    
1744      if (*ecode == OP_KETRPOS)      if (*ecode == OP_KETRPOS)
1745        {        {
1746        md->end_match_ptr = eptr;        md->end_match_ptr = eptr;
1747        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1748        RRETURN(MATCH_KETRPOS);        RRETURN(MATCH_KETRPOS);
1749        }        }
1750    
1751      /* The normal repeating kets try the rest of the pattern or restart from      /* The normal repeating kets try the rest of the pattern or restart from
1752      the preceding bracket, in the appropriate order. In the second case, we can      the preceding bracket, in the appropriate order. In the second case, we can
# Line 1763  for (;;) Line 1763  for (;;)
1763          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1764          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1765          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1766          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
1767          }          }
1768        if (*prev >= OP_SBRA)    /* Could match an empty string */        if (*prev >= OP_SBRA)    /* Could match an empty string */
1769          {          {
1770          md->match_function_type = MATCH_CBEGROUP;          md->match_function_type = MATCH_CBEGROUP;
1771          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1772          RRETURN(rrc);          RRETURN(rrc);
1773          }          }
# Line 1776  for (;;) Line 1776  for (;;)
1776        }        }
1777      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1778        {        {
1779        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1780        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1781        if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;        if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1782        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
# Line 1785  for (;;) Line 1785  for (;;)
1785          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1786          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1787          md->once_target = prev;          md->once_target = prev;
1788          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
1789          }          }
1790        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1791        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1792        }        }
# Line 1796  for (;;) Line 1796  for (;;)
1796    
1797      case OP_CIRC:      case OP_CIRC:
1798      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1799    
1800      /* Start of subject assertion */      /* Start of subject assertion */
1801    
1802      case OP_SOD:      case OP_SOD:
1803      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1804      ecode++;      ecode++;
1805      break;      break;
1806    
1807      /* Multiline mode: start of subject unless notbol, or after any newline. */      /* Multiline mode: start of subject unless notbol, or after any newline. */
1808    
1809      case OP_CIRCM:      case OP_CIRCM:
# Line 1842  for (;;) Line 1842  for (;;)
1842      ecode++;      ecode++;
1843      break;      break;
1844    
1845      /* Not multiline mode: assert before a terminating newline or before end of      /* Not multiline mode: assert before a terminating newline or before end of
1846      subject unless noteol is set. */      subject unless noteol is set. */
1847    
1848      case OP_DOLL:      case OP_DOLL:
# Line 2131  for (;;) Line 2131  for (;;)
2131      switch(c)      switch(c)
2132        {        {
2133        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2134    
2135        case 0x000d:        case 0x000d:
2136        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2137        break;        break;
# Line 2377  for (;;) Line 2377  for (;;)
2377      loops). */      loops). */
2378    
2379      case OP_REF:      case OP_REF:
2380      case OP_REFI:      case OP_REFI:
2381      caseless = op == OP_REFI;      caseless = op == OP_REFI;
2382      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2383      ecode += 3;      ecode += 3;
2384    
# Line 2446  for (;;) Line 2446  for (;;)
2446    
2447      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2448        {        {
2449        int slength;        int slength;
2450        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2451          {          {
2452          CHECK_PARTIAL();          CHECK_PARTIAL();
# Line 2466  for (;;) Line 2466  for (;;)
2466        {        {
2467        for (fi = min;; fi++)        for (fi = min;; fi++)
2468          {          {
2469          int slength;          int slength;
2470          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2471          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2472          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 2487  for (;;) Line 2487  for (;;)
2487        pp = eptr;        pp = eptr;
2488        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2489          {          {
2490          int slength;          int slength;
2491          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2492            {            {
2493            CHECK_PARTIAL();            CHECK_PARTIAL();
# Line 3220  for (;;) Line 3220  for (;;)
3220      checking can be multibyte. */      checking can be multibyte. */
3221    
3222      case OP_NOT:      case OP_NOT:
3223      case OP_NOTI:      case OP_NOTI:
3224      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3225        {        {
3226        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 3715  for (;;) Line 3715  for (;;)
3715            case PT_LAMP:            case PT_LAMP:
3716            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3717              {              {
3718              int chartype;              int chartype;
3719              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3720                {                {
3721                SCHECK_PARTIAL();                SCHECK_PARTIAL();
# Line 3775  for (;;) Line 3775  for (;;)
3775            case PT_ALNUM:            case PT_ALNUM:
3776            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3777              {              {
3778              int category;              int category;
3779              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3780                {                {
3781                SCHECK_PARTIAL();                SCHECK_PARTIAL();
# Line 3823  for (;;) Line 3823  for (;;)
3823            case PT_WORD:            case PT_WORD:
3824            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3825              {              {
3826              int category;              int category;
3827              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3828                {                {
3829                SCHECK_PARTIAL();                SCHECK_PARTIAL();
# Line 3920  for (;;) Line 3920  for (;;)
3920            switch(c)            switch(c)
3921              {              {
3922              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3923    
3924              case 0x000d:              case 0x000d:
3925              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3926              break;              break;
# Line 4197  for (;;) Line 4197  for (;;)
4197            switch(*eptr++)            switch(*eptr++)
4198              {              {
4199              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4200    
4201              case 0x000d:              case 0x000d:
4202              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4203              break;              break;
4204    
4205              case 0x000a:              case 0x000a:
4206              break;              break;
4207    
# Line 4407  for (;;) Line 4407  for (;;)
4407            case PT_LAMP:            case PT_LAMP:
4408            for (fi = min;; fi++)            for (fi = min;; fi++)
4409              {              {
4410              int chartype;              int chartype;
4411              RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4412              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4413              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4479  for (;;) Line 4479  for (;;)
4479            case PT_ALNUM:            case PT_ALNUM:
4480            for (fi = min;; fi++)            for (fi = min;; fi++)
4481              {              {
4482              int category;              int category;
4483              RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4484              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4485              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4536  for (;;) Line 4536  for (;;)
4536            case PT_WORD:            case PT_WORD:
4537            for (fi = min;; fi++)            for (fi = min;; fi++)
4538              {              {
4539              int category;              int category;
4540              RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4541              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4542              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
# Line 4908  for (;;) Line 4908  for (;;)
4908            case PT_LAMP:            case PT_LAMP:
4909            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4910              {              {
4911              int chartype;              int chartype;
4912              int len = 1;              int len = 1;
4913              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4914                {                {
# Line 4973  for (;;) Line 4973  for (;;)
4973            case PT_ALNUM:            case PT_ALNUM:
4974            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4975              {              {
4976              int category;              int category;
4977              int len = 1;              int len = 1;
4978              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4979                {                {
# Line 5027  for (;;) Line 5027  for (;;)
5027            case PT_WORD:            case PT_WORD:
5028            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5029              {              {
5030              int category;              int category;
5031              int len = 1;              int len = 1;
5032              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5033                {                {
# Line 5066  for (;;) Line 5066  for (;;)
5066          {          {
5067          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5068            {            {
5069            int len = 1;            int len = 1;
5070            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5071              {              {
5072              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 5074  for (;;) Line 5074  for (;;)
5074              }              }
5075            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5076            if (UCD_CATEGORY(c) == ucp_M) break;            if (UCD_CATEGORY(c) == ucp_M) break;
5077            eptr += len;            eptr += len;
5078            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5079              {              {
5080              len = 1;              len = 1;
# Line 5385  for (;;) Line 5385  for (;;)
5385            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5386            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5387            BACKCHAR(eptr);            BACKCHAR(eptr);
5388            if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&            if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5389                eptr[-1] == '\r') eptr--;                eptr[-1] == '\r') eptr--;
5390            }            }
5391          }          }
# Line 5597  for (;;) Line 5597  for (;;)
5597            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
5598            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5599            eptr--;            eptr--;
5600            if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&            if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5601                eptr[-1] == '\r') eptr--;                eptr[-1] == '\r') eptr--;
5602            }            }
5603          }          }
# Line 5637  switch (frame->Xwhere) Line 5637  switch (frame->Xwhere)
5637    LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)    LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)
5638    LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)    LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)
5639    LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)    LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)
5640    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63)
5641  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5642    LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)    LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)
5643    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
# Line 5856  md->notempty = (options & PCRE_NOTEMPTY) Line 5856  md->notempty = (options & PCRE_NOTEMPTY)
5856  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
5857  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5858                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5859    
5860    
5861  md->hitend = FALSE;  md->hitend = FALSE;
5862  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
5863    
# Line 5939  defined (though never set). So there's n Line 5939  defined (though never set). So there's n
5939  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
5940    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
5941    
5942  /* Check a UTF-8 string if required. Pass back the character offset and error  /* Check a UTF-8 string if required. Pass back the character offset and error
5943  code for an invalid string if a results vector is available. */  code for an invalid string if a results vector is available. */
5944    
5945  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5946  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5947    {    {
5948    int erroroffset;    int erroroffset;
5949    int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);    int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5950    if (errorcode != 0)    if (errorcode != 0)
5951      {      {
# Line 5953  if (utf8 && (options & PCRE_NO_UTF8_CHEC Line 5953  if (utf8 && (options & PCRE_NO_UTF8_CHEC
5953        {        {
5954        offsets[0] = erroroffset;        offsets[0] = erroroffset;
5955        offsets[1] = errorcode;        offsets[1] = errorcode;
5956        }        }
5957      return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?      return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5958        PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;        PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5959      }      }
5960    
5961    /* Check that a start_offset points to the start of a UTF-8 character. */    /* Check that a start_offset points to the start of a UTF-8 character. */
5962    
5963    if (start_offset > 0 && start_offset < length &&    if (start_offset > 0 && start_offset < length &&
5964        (((USPTR)subject)[start_offset] & 0xc0) == 0x80)        (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
5965      return PCRE_ERROR_BADUTF8_OFFSET;      return PCRE_ERROR_BADUTF8_OFFSET;
5966    }    }
5967  #endif  #endif
# Line 6234  for(;;) Line 6234  for(;;)
6234    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6235    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6236    md->match_call_count = 0;    md->match_call_count = 0;
6237    md->match_function_type = 0;    md->match_function_type = 0;
6238    md->end_offset_top = 0;    md->end_offset_top = 0;
6239    rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);    rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6240    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
# Line 6359  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6359  if (rc == MATCH_MATCH || rc == MATCH_ACC
6359    
6360    /* 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 are
6361    too many to fit into the vector. */    too many to fit into the vector. */
6362    
6363    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = md->offset_overflow? 0 : md->end_offset_top/2;
6364    
6365      /* If there is space in the offset vector, set any unused pairs at the end of
6366      the pattern to -1 for backwards compatibility. It is documented that this
6367      happens. In earlier versions, the whole set of potential capturing offsets
6368      was set to -1 each time round the loop, but this is handled differently now.
6369      "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only
6370      those at the end that need unsetting here. We can't just unset them all at
6371      the start of the whole thing because they may get set in one branch that is
6372      not the final matching branch. */
6373    
6374      if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL)
6375        {
6376        register int *iptr, *iend;
6377        int resetcount = 2 + re->top_bracket * 2;
6378        if (resetcount > offsetcount) resetcount = ocount;
6379        iptr = offsets + md->end_offset_top;
6380        iend = offsets + resetcount;
6381        while (iptr < iend) *iptr++ = -1;
6382        }
6383    
6384    /* 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
6385    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
6386    matching path. */    matching path. */

Legend:
Removed from v.623  
changed lines
  Added in v.625

  ViewVC Help
Powered by ViewVC 1.1.5