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

Diff of /code/trunk/pcre_exec.c

Parent Directory | Revision Log | View Patch Patch

-revision 165 by ph10,
Wed May  9 10:50:57 2007 UTC
+revision 384 by ph10,
Sun Mar  8 16:27:43 2009 UTC
 Line 6
  and semantics are as close as possible to those of the Perl 5 language.
                         Written by Philip Hazel
-            Copyright (c) 1997-2007 University of Cambridge
+            Copyright (c) 1997-2009 University of Cambridge
  -----------------------------------------------------------------------------
  Redistribution and use in source and binary forms, with or without
 Line 42 
 POSSIBILITY OF SUCH DAMAGE.
  pattern matching using an NFA algorithm, trying to mimic Perl as closely as
  possible. There are also some static supporting functions. */
+ #ifdef HAVE_CONFIG_H
+ #include "config.h"
+ #endif
  #define NLBLOCK md             /* Block containing newline information */
  #define PSSTART start_subject  /* Field containing processed string start */
  #define PSEND   end_subject    /* Field containing processed string end */
-Line 53 
 possible. There are also some static sup
+Line 57 
 possible. There are also some static sup
  #undef min
  #undef max
- /* The chain of eptrblocks for tail recursions uses memory in stack workspace,
- obtained at top level, the size of which is defined by EPTR_WORK_SIZE. */
- #define EPTR_WORK_SIZE (1000)
  /* Flag bits for the match() function */
  #define match_condassert     0x01  /* Called to check a condition assertion */
  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */
- #define match_tail_recursed  0x04  /* Tail recursive call */
  /* Non-error returns from the match() function. Error returns are externally
  defined PCRE_ERROR_xxx codes, which are all negative. */
-Line 70 
 defined PCRE_ERROR_xxx codes, which are
+Line 68 
 defined PCRE_ERROR_xxx codes, which are
  #define MATCH_MATCH        1
  #define MATCH_NOMATCH      0
+ /* Special internal returns from the match() function. Make them sufficiently
+ negative to avoid the external error codes. */
+ #define MATCH_COMMIT       (-999)
+ #define MATCH_PRUNE        (-998)
+ #define MATCH_SKIP         (-997)
+ #define MATCH_THEN         (-996)
  /* Maximum number of ints of offset to save on the stack for recursive calls.
  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
  because the offset vector is always a multiple of 3 long. */
-Line 152 
 printf("\n");
+Line 158 
 printf("\n");
  if (length > md->end_subject - eptr) return FALSE;
- /* Separate the caselesss case for speed */
+ /* Separate the caseless case for speed. In UTF-8 mode we can only do this
+ properly if Unicode properties are supported. Otherwise, we can check only
+ ASCII characters. */
  if ((ims & PCRE_CASELESS) != 0)
    {
+ #ifdef SUPPORT_UTF8
+ #ifdef SUPPORT_UCP
+   if (md->utf8)
+     {
+     USPTR endptr = eptr + length;
+     while (eptr < endptr)
+       {
+       int c, d;
+       GETCHARINC(c, eptr);
+       GETCHARINC(d, p);
+       if (c != d && c != UCD_OTHERCASE(d)) return FALSE;
+       }
+     }
+   else
+ #endif
+ #endif
+   /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
+   is no UCP support. */
    while (length-- > 0)
-     if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE;
+     { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }
    }
+ /* In the caseful case, we can just compare the bytes, whether or not we
+ are in UTF-8 mode. */
  else
    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }
-Line 205 
 variable instead of being passed in the
+Line 237 
 variable instead of being passed in the
  ****************************************************************************
  ***************************************************************************/
+ /* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN
- /* Numbers for RMATCH calls */
+ below must be updated in sync.  */
  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,
         RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,
         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
-        RM41,  RM42, RM43, RM44, RM45, RM46, RM47 };
+        RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
+        RM51,  RM52, RM53, RM54 };
  /* These versions of the macros use the stack, as normal. There are debugging
  versions and production versions. Note that the "rw" argument of RMATCH isn't
-Line 226 
 actuall used in this definition. */
+Line 258 
 actuall used in this definition. */
  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
    { \
    printf("match() called in line %d\n", __LINE__); \
-   rrc = match(ra,rb,rc,rd,re,rf,rg,rdepth+1); \
+   rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \
    printf("to line %d\n", __LINE__); \
    }
  #define RRETURN(ra) \
-Line 236 
 actuall used in this definition. */
+Line 268 
 actuall used in this definition. */
    }
  #else
  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
-   rrc = match(ra,rb,rc,rd,re,rf,rg,rdepth+1)
+   rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1)
  #define RRETURN(ra) return ra
  #endif
-Line 255 
 argument of match(), which never changes
+Line 287 
 argument of match(), which never changes
    frame->Xwhere = rw; \
    newframe->Xeptr = ra;\
    newframe->Xecode = rb;\
+   newframe->Xmstart = mstart;\
    newframe->Xoffset_top = rc;\
    newframe->Xims = re;\
    newframe->Xeptrb = rf;\
-Line 291 
 typedef struct heapframe {
+Line 324 
 typedef struct heapframe {
    const uschar *Xeptr;
    const uschar *Xecode;
+   const uschar *Xmstart;
    int Xoffset_top;
    long int Xims;
    eptrblock *Xeptrb;
-Line 371 
 made performance worse.
+Line 405 
 made performance worse.
  Arguments:
     eptr        pointer to current character in subject
     ecode       pointer to current position in compiled code
+    mstart      pointer to the current match start position (can be modified
+                  by encountering \K)
     offset_top  current top pointer
     md          pointer to "static" info for the match
     ims         current /i, /m, and /s options
-Line 380 
 Arguments:
+Line 416 
 Arguments:
                   match_condassert - this is an assertion condition
                   match_cbegroup - this is the start of an unlimited repeat
                     group that can match an empty string
-                  match_tail_recursed - this is a tail_recursed group
     rdepth      the recursion depth
  Returns:       MATCH_MATCH if matched            )  these values are >= 0
-Line 390 
 Returns:       MATCH_MATCH if matched
+Line 425 
 Returns:       MATCH_MATCH if matched
  */
  static int
- match(REGISTER USPTR eptr, REGISTER const uschar *ecode,
+ match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart,
    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,
    int flags, unsigned int rdepth)
  {
-Line 418 
 frame->Xprevframe = NULL;            /*
+Line 453 
 frame->Xprevframe = NULL;            /*
  frame->Xeptr = eptr;
  frame->Xecode = ecode;
+ frame->Xmstart = mstart;
  frame->Xoffset_top = offset_top;
  frame->Xims = ims;
  frame->Xeptrb = eptrb;
-Line 432 
 HEAP_RECURSE:
+Line 468 
 HEAP_RECURSE:
  #define eptr               frame->Xeptr
  #define ecode              frame->Xecode
+ #define mstart             frame->Xmstart
  #define offset_top         frame->Xoffset_top
  #define ims                frame->Xims
  #define eptrb              frame->Xeptrb
-Line 580 
 original_ims = ims;    /* Save for reset
+Line 617 
 original_ims = ims;    /* Save for reset
  string, the match_cbegroup flag is set. When this is the case, add the current
  subject pointer to the chain of such remembered pointers, to be checked when we
  hit the closing ket, in order to break infinite loops that match no characters.
- When match() is called in other circumstances, don't add to the chain. If this
+ When match() is called in other circumstances, don't add to the chain. The
- is a tail recursion, use a block from the workspace, as the one on the stack is
+ match_cbegroup flag must NOT be used with tail recursion, because the memory
- already used. */
+ block that is used is on the stack, so a new one may be required for each
+ match(). */
  if ((flags & match_cbegroup) != 0)
    {
-   eptrblock *p;
+   newptrb.epb_saved_eptr = eptr;
-   if ((flags & match_tail_recursed) != 0)
+   newptrb.epb_prev = eptrb;
-     {
+   eptrb = &newptrb;
-     if (md->eptrn >= EPTR_WORK_SIZE) RRETURN(PCRE_ERROR_NULLWSLIMIT);
-     p = md->eptrchain + md->eptrn++;
-     }
-   else p = &newptrb;
-   p->epb_saved_eptr = eptr;
-   p->epb_prev = eptrb;
-   eptrb = p;
    }
  /* Now start processing the opcodes. */
-Line 610 
 for (;;)
+Line 641 
 for (;;)
    if (md->partial &&
        eptr >= md->end_subject &&
-       eptr > md->start_match)
+       eptr > mstart)
      md->hitend = TRUE;
    switch(op)
      {
+     case OP_FAIL:
+     RRETURN(MATCH_NOMATCH);
+     case OP_PRUNE:
+     RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
+       ims, eptrb, flags, RM51);
+     if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+     RRETURN(MATCH_PRUNE);
+     case OP_COMMIT:
+     RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
+       ims, eptrb, flags, RM52);
+     if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+     RRETURN(MATCH_COMMIT);
+     case OP_SKIP:
+     RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
+       ims, eptrb, flags, RM53);
+     if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+     md->start_match_ptr = eptr;   /* Pass back current position */
+     RRETURN(MATCH_SKIP);
+     case OP_THEN:
+     RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
+       ims, eptrb, flags, RM54);
+     if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+     RRETURN(MATCH_THEN);
      /* Handle a capturing bracket. If there is space in the offset vector, save
      the current subject position in the working slot at the top of the vector.
      We mustn't change the current values of the data slot, because they may be
-Line 656 
 for (;;)
+Line 715 
 for (;;)
          {
          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
            ims, eptrb, flags, RM1);
-         if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+         if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
          md->capture_last = save_capture_last;
          ecode += GET(ecode, 1);
          }
-Line 671 
 for (;;)
+Line 730 
 for (;;)
        RRETURN(MATCH_NOMATCH);
        }
-     /* Insufficient room for saving captured contents. Treat as a non-capturing
+     /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
-     bracket. */
+     as a non-capturing bracket. */
+     /* VVVVVVVVVVVVVVVVVVVVVVVVV */
+     /* VVVVVVVVVVVVVVVVVVVVVVVVV */
      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
+     /* VVVVVVVVVVVVVVVVVVVVVVVVV */
+     /* VVVVVVVVVVVVVVVVVVVVVVVVV */
      /* Non-capturing bracket. Loop for all the alternatives. When we get to the
      final alternative within the brackets, we would return the result of a
      recursive call to match() whatever happened. We can reduce stack usage by
-     turning this into a tail recursion. */
+     turning this into a tail recursion, except in the case when match_cbegroup
+     is set.*/
      case OP_BRA:
      case OP_SBRA:
-Line 687 
 for (;;)
+Line 753 
 for (;;)
      flags = (op >= OP_SBRA)? match_cbegroup : 0;
      for (;;)
        {
-       if (ecode[GET(ecode, 1)] != OP_ALT)
+       if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */
          {
-         ecode += _pcre_OP_lengths[*ecode];
+         if (flags == 0)    /* Not a possibly empty group */
-         flags |= match_tail_recursed;
+           {
-         DPRINTF(("bracket 0 tail recursion\n"));
+           ecode += _pcre_OP_lengths[*ecode];
-         goto TAIL_RECURSE;
+           DPRINTF(("bracket 0 tail recursion\n"));
+           goto TAIL_RECURSE;
+           }
+         /* Possibly empty group; can't use tail recursion. */
+         RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,
+           eptrb, flags, RM48);
+         RRETURN(rrc);
          }
        /* For non-final alternatives, continue the loop for a NOMATCH result;
-Line 700 
 for (;;)
+Line 774 
 for (;;)
        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,
          eptrb, flags, RM2);
-       if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
        ecode += GET(ecode, 1);
        }
      /* Control never reaches here. */
-Line 713 
 for (;;)
+Line 787 
 for (;;)
      case OP_COND:
      case OP_SCOND:
+     /* Because of the way auto-callout works during compile, a callout item is
+     inserted between OP_COND and an assertion condition. */
+     if (ecode[LINK_SIZE+1] == OP_CALLOUT)
+       {
+       if (pcre_callout != NULL)
+         {
+         pcre_callout_block cb;
+         cb.version          = 1;   /* Version 1 of the callout block */
+         cb.callout_number   = ecode[LINK_SIZE+2];
+         cb.offset_vector    = md->offset_vector;
+         cb.subject          = (PCRE_SPTR)md->start_subject;
+         cb.subject_length   = md->end_subject - md->start_subject;
+         cb.start_match      = mstart - md->start_subject;
+         cb.current_position = eptr - md->start_subject;
+         cb.pattern_position = GET(ecode, LINK_SIZE + 3);
+         cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
+         cb.capture_top      = offset_top/2;
+         cb.capture_last     = md->capture_last;
+         cb.callout_data     = md->callout_data;
+         if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);
+         if (rrc < 0) RRETURN(rrc);
+         }
+       ecode += _pcre_OP_lengths[OP_CALLOUT];
+       }
+     /* Now see what the actual condition is */
      if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */
        {
        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/
-Line 748 
 for (;;)
+Line 850 
 for (;;)
          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
          }
-       else if (rrc != MATCH_NOMATCH)
+       else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
          {
          RRETURN(rrc);         /* Need braces because of following else */
          }
-Line 760 
 for (;;)
+Line 862 
 for (;;)
        }
      /* We are now at the branch that is to be obeyed. As there is only one,
-     we can use tail recursion to avoid using another stack frame. If the second
+     we can use tail recursion to avoid using another stack frame, except when
-     alternative doesn't exist, we can just plough on. */
+     match_cbegroup is required for an unlimited repeat of a possibly empty
+     group. If the second alternative doesn't exist, we can just plough on. */
      if (condition || *ecode == OP_ALT)
        {
        ecode += 1 + LINK_SIZE;
-       flags = match_tail_recursed | ((op == OP_SCOND)? match_cbegroup : 0);
+       if (op == OP_SCOND)        /* Possibly empty group */
-       goto TAIL_RECURSE;
+         {
+         RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);
+         RRETURN(rrc);
+         }
+       else                       /* Group must match something */
+         {
+         flags = 0;
+         goto TAIL_RECURSE;
+         }
        }
-     else
+     else                         /* Condition false & no 2nd alternative */
        {
        ecode += 1 + LINK_SIZE;
        }
      break;
-     /* End of the pattern. If we are in a top-level recursion, we should
+     /* End of the pattern, either real or forced. If we are in a top-level
-     restore the offsets appropriately and continue from after the call. */
+     recursion, we should restore the offsets appropriately and continue from
+     after the call. */
+     case OP_ACCEPT:
      case OP_END:
      if (md->recursive != NULL && md->recursive->group_num == 0)
        {
-Line 787 
 for (;;)
+Line 900 
 for (;;)
        md->recursive = rec->prevrec;
        memmove(md->offset_vector, rec->offset_save,
          rec->saved_max * sizeof(int));
-       md->start_match = rec->save_start;
+       mstart = rec->save_start;
        ims = original_ims;
        ecode = rec->after_call;
        break;
-Line 796 
 for (;;)
+Line 909 
 for (;;)
      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty
      string - backtracking will then try other alternatives, if any. */
-     if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH);
+     if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH);
-     md->end_match_ptr = eptr;          /* Record where we ended */
+     md->end_match_ptr = eptr;           /* Record where we ended */
-     md->end_offset_top = offset_top;   /* and how many extracts were taken */
+     md->end_offset_top = offset_top;    /* and how many extracts were taken */
+     md->start_match_ptr = mstart;       /* and the start (\K can modify) */
      RRETURN(MATCH_MATCH);
      /* Change option settings */
-Line 822 
 for (;;)
+Line 936 
 for (;;)
        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
          RM4);
        if (rrc == MATCH_MATCH) break;
-       if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
        ecode += GET(ecode, 1);
        }
      while (*ecode == OP_ALT);
-Line 849 
 for (;;)
+Line 963 
 for (;;)
        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
          RM5);
        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);
-       if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
        ecode += GET(ecode,1);
        }
      while (*ecode == OP_ALT);
-Line 873 
 for (;;)
+Line 987 
 for (;;)
          {
          eptr--;
          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
-         BACKCHAR(eptr)
+         BACKCHAR(eptr);
          }
        }
      else
-Line 904 
 for (;;)
+Line 1018 
 for (;;)
        cb.offset_vector    = md->offset_vector;
        cb.subject          = (PCRE_SPTR)md->start_subject;
        cb.subject_length   = md->end_subject - md->start_subject;
-       cb.start_match      = md->start_match - md->start_subject;
+       cb.start_match      = mstart - md->start_subject;
        cb.current_position = eptr - md->start_subject;
        cb.pattern_position = GET(ecode, 2);
        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
-Line 966 
 for (;;)
+Line 1080 
 for (;;)
        memcpy(new_recursive.offset_save, md->offset_vector,
              new_recursive.saved_max * sizeof(int));
-       new_recursive.save_start = md->start_match;
+       new_recursive.save_start = mstart;
-       md->start_match = eptr;
+       mstart = eptr;
        /* OK, now we can do the recursion. For each top-level alternative we
        restore the offset and recursion data. */
-Line 986 
 for (;;)
+Line 1100 
 for (;;)
              (pcre_free)(new_recursive.offset_save);
            RRETURN(MATCH_MATCH);
            }
-         else if (rrc != MATCH_NOMATCH)
+         else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
            {
            DPRINTF(("Recursion gave error %d\n", rrc));
            RRETURN(rrc);
-Line 1020 
 for (;;)
+Line 1134 
 for (;;)
      do
        {
-       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,
+       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);
-         eptrb, 0, RM7);
        if (rrc == MATCH_MATCH) break;
-       if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
        ecode += GET(ecode,1);
        }
      while (*ecode == OP_ALT);
-Line 1066 
 for (;;)
+Line 1179 
 for (;;)
      if (*ecode == OP_KETRMIN)
        {
-       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0,
+       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);
-         RM8);
        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
        ecode = prev;
-       flags = match_tail_recursed;
+       flags = 0;
        goto TAIL_RECURSE;
        }
      else  /* OP_KETRMAX */
-Line 1078 
 for (;;)
+Line 1190 
 for (;;)
        RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);
        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
        ecode += 1 + LINK_SIZE;
-       flags = match_tail_recursed;
+       flags = 0;
        goto TAIL_RECURSE;
        }
      /* Control never gets here */
-Line 1090 
 for (;;)
+Line 1202 
 for (;;)
      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
      break;
-     /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating
+     /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
-     that it may occur zero times. It may repeat infinitely, or not at all -
+     indicating that it may occur zero times. It may repeat infinitely, or not
-     i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper
+     at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
-     repeat limits are compiled as a number of copies, with the optional ones
+     with fixed upper repeat limits are compiled as a number of copies, with the
-     preceded by BRAZERO or BRAMINZERO. */
+     optional ones preceded by BRAZERO or BRAMINZERO. */
      case OP_BRAZERO:
        {
-Line 1116 
 for (;;)
+Line 1228 
 for (;;)
        }
      break;
+     case OP_SKIPZERO:
+       {
+       next = ecode+1;
+       do next += GET(next,1); while (*next == OP_ALT);
+       ecode = next + 1 + LINK_SIZE;
+       }
+     break;
      /* End of a group, repeated or non-repeating. */
      case OP_KET:
-Line 1180 
 for (;;)
+Line 1300 
 for (;;)
          recursion_info *rec = md->recursive;
          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
          md->recursive = rec->prevrec;
-         md->start_match = rec->save_start;
+         mstart = rec->save_start;
          memcpy(md->offset_vector, rec->offset_save,
            rec->saved_max * sizeof(int));
          ecode = rec->after_call;
-Line 1209 
 for (;;)
+Line 1329 
 for (;;)
      /* The repeating kets try the rest of the pattern or restart from the
      preceding bracket, in the appropriate order. In the second case, we can use
-     tail recursion to avoid using another stack frame. */
+     tail recursion to avoid using another stack frame, unless we have an
+     unlimited repeat of a group that can match an empty string. */
      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;
      if (*ecode == OP_KETRMIN)
        {
-       RMATCH(eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0,
+       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);
-         RM12);
        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+       if (flags != 0)    /* Could match an empty string */
+         {
+         RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);
+         RRETURN(rrc);
+         }
        ecode = prev;
-       flags |= match_tail_recursed;
        goto TAIL_RECURSE;
        }
      else  /* OP_KETRMAX */
-Line 1227 
 for (;;)
+Line 1351 
 for (;;)
        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);
        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
        ecode += 1 + LINK_SIZE;
-       flags = match_tail_recursed;
+       flags = 0;
        goto TAIL_RECURSE;
        }
      /* Control never gets here */
-Line 1260 
 for (;;)
+Line 1384 
 for (;;)
      ecode++;
      break;
+     /* Reset the start of match point */
+     case OP_SET_SOM:
+     mstart = eptr;
+     ecode++;
+     break;
      /* Assert before internal newline if multiline, or before a terminating
      newline unless endonly is set, else end of subject unless noteol is set. */
-Line 1352 
 for (;;)
+Line 1483 
 for (;;)
      /* Match a single character type; inline for speed */
      case OP_ANY:
-     if ((ims & PCRE_DOTALL) == 0)
+     if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
-       {
+     /* Fall through */
-       if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
-       }
+     case OP_ALLANY:
      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);
-     if (utf8)
+     if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
-       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
      ecode++;
      break;
-Line 1457 
 for (;;)
+Line 1587 
 for (;;)
        case 0x000d:
        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
        break;
        case 0x000a:
+       break;
        case 0x000b:
        case 0x000c:
        case 0x0085:
        case 0x2028:
        case 0x2029:
+       if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
+       break;
+       }
+     ecode++;
+     break;
+     case OP_NOT_HSPACE:
+     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     GETCHARINCTEST(c, eptr);
+     switch(c)
+       {
+       default: break;
+       case 0x09:      /* HT */
+       case 0x20:      /* SPACE */
+       case 0xa0:      /* NBSP */
+       case 0x1680:    /* OGHAM SPACE MARK */
+       case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
+       case 0x2000:    /* EN QUAD */
+       case 0x2001:    /* EM QUAD */
+       case 0x2002:    /* EN SPACE */
+       case 0x2003:    /* EM SPACE */
+       case 0x2004:    /* THREE-PER-EM SPACE */
+       case 0x2005:    /* FOUR-PER-EM SPACE */
+       case 0x2006:    /* SIX-PER-EM SPACE */
+       case 0x2007:    /* FIGURE SPACE */
+       case 0x2008:    /* PUNCTUATION SPACE */
+       case 0x2009:    /* THIN SPACE */
+       case 0x200A:    /* HAIR SPACE */
+       case 0x202f:    /* NARROW NO-BREAK SPACE */
+       case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
+       case 0x3000:    /* IDEOGRAPHIC SPACE */
+       RRETURN(MATCH_NOMATCH);
+       }
+     ecode++;
+     break;
+     case OP_HSPACE:
+     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     GETCHARINCTEST(c, eptr);
+     switch(c)
+       {
+       default: RRETURN(MATCH_NOMATCH);
+       case 0x09:      /* HT */
+       case 0x20:      /* SPACE */
+       case 0xa0:      /* NBSP */
+       case 0x1680:    /* OGHAM SPACE MARK */
+       case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
+       case 0x2000:    /* EN QUAD */
+       case 0x2001:    /* EM QUAD */
+       case 0x2002:    /* EN SPACE */
+       case 0x2003:    /* EM SPACE */
+       case 0x2004:    /* THREE-PER-EM SPACE */
+       case 0x2005:    /* FOUR-PER-EM SPACE */
+       case 0x2006:    /* SIX-PER-EM SPACE */
+       case 0x2007:    /* FIGURE SPACE */
+       case 0x2008:    /* PUNCTUATION SPACE */
+       case 0x2009:    /* THIN SPACE */
+       case 0x200A:    /* HAIR SPACE */
+       case 0x202f:    /* NARROW NO-BREAK SPACE */
+       case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
+       case 0x3000:    /* IDEOGRAPHIC SPACE */
+       break;
+       }
+     ecode++;
+     break;
+     case OP_NOT_VSPACE:
+     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     GETCHARINCTEST(c, eptr);
+     switch(c)
+       {
+       default: break;
+       case 0x0a:      /* LF */
+       case 0x0b:      /* VT */
+       case 0x0c:      /* FF */
+       case 0x0d:      /* CR */
+       case 0x85:      /* NEL */
+       case 0x2028:    /* LINE SEPARATOR */
+       case 0x2029:    /* PARAGRAPH SEPARATOR */
+       RRETURN(MATCH_NOMATCH);
+       }
+     ecode++;
+     break;
+     case OP_VSPACE:
+     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     GETCHARINCTEST(c, eptr);
+     switch(c)
+       {
+       default: RRETURN(MATCH_NOMATCH);
+       case 0x0a:      /* LF */
+       case 0x0b:      /* VT */
+       case 0x0c:      /* FF */
+       case 0x0d:      /* CR */
+       case 0x85:      /* NEL */
+       case 0x2028:    /* LINE SEPARATOR */
+       case 0x2029:    /* PARAGRAPH SEPARATOR */
        break;
        }
      ecode++;
-Line 1477 
 for (;;)
+Line 1707 
 for (;;)
      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
      GETCHARINCTEST(c, eptr);
        {
-       int chartype, script;
+       const ucd_record *prop = GET_UCD(c);
-       int category = _pcre_ucp_findprop(c, &chartype, &script);
        switch(ecode[1])
          {
-Line 1487 
 for (;;)
+Line 1716 
 for (;;)
          break;
          case PT_LAMP:
-         if ((chartype == ucp_Lu ||
+         if ((prop->chartype == ucp_Lu ||
-              chartype == ucp_Ll ||
+              prop->chartype == ucp_Ll ||
-              chartype == ucp_Lt) == (op == OP_NOTPROP))
+              prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
            RRETURN(MATCH_NOMATCH);
           break;
          case PT_GC:
-         if ((ecode[2] != category) == (op == OP_PROP))
+         if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
            RRETURN(MATCH_NOMATCH);
          break;
          case PT_PC:
-         if ((ecode[2] != chartype) == (op == OP_PROP))
+         if ((ecode[2] != prop->chartype) == (op == OP_PROP))
            RRETURN(MATCH_NOMATCH);
          break;
          case PT_SC:
-         if ((ecode[2] != script) == (op == OP_PROP))
+         if ((ecode[2] != prop->script) == (op == OP_PROP))
            RRETURN(MATCH_NOMATCH);
          break;
-Line 1523 
 for (;;)
+Line 1752 
 for (;;)
      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
      GETCHARINCTEST(c, eptr);
        {
-       int chartype, script;
+       int category = UCD_CATEGORY(c);
-       int category = _pcre_ucp_findprop(c, &chartype, &script);
        if (category == ucp_M) RRETURN(MATCH_NOMATCH);
        while (eptr < md->end_subject)
          {
-Line 1533 
 for (;;)
+Line 1761 
 for (;;)
            {
            GETCHARLEN(c, eptr, len);
            }
-         category = _pcre_ucp_findprop(c, &chartype, &script);
+         category = UCD_CATEGORY(c);
          if (category != ucp_M) break;
          eptr += len;
          }
-Line 1554 
 for (;;)
+Line 1782 
 for (;;)
      case OP_REF:
        {
        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
-       ecode += 3;                                 /* Advance past item */
+       ecode += 3;
+       /* If the reference is unset, there are two possibilities:
-       /* If the reference is unset, set the length to be longer than the amount
+       (a) In the default, Perl-compatible state, set the length to be longer
-       of subject left; this ensures that every attempt at a match fails. We
+       than the amount of subject left; this ensures that every attempt at a
-       can't just fail here, because of the possibility of quantifiers with zero
+       match fails. We can't just fail here, because of the possibility of
-       minima. */
+       quantifiers with zero minima.
-       length = (offset >= offset_top || md->offset_vector[offset] < 0)?
+       (b) If the JavaScript compatibility flag is set, set the length to zero
-         md->end_subject - eptr + 1 :
+       so that the back reference matches an empty string.
-         md->offset_vector[offset+1] - md->offset_vector[offset];
+       Otherwise, set the length to the length of what was matched by the
+       referenced subpattern. */
+       if (offset >= offset_top || md->offset_vector[offset] < 0)
+         length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;
+       else
+         length = md->offset_vector[offset+1] - md->offset_vector[offset];
        /* Set up for repetition, or handle the non-repeated case */
-Line 1838 
 for (;;)
+Line 2075 
 for (;;)
      /* Match an extended character class. This opcode is encountered only
-     in UTF-8 mode, because that's the only time it is compiled. */
+     when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8
+     mode, because Unicode properties are supported in non-UTF-8 mode. */
  #ifdef SUPPORT_UTF8
      case OP_XCLASS:
-Line 1880 
 for (;;)
+Line 2118 
 for (;;)
        for (i = 1; i <= min; i++)
          {
          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
-         GETCHARINC(c, eptr);
+         GETCHARINCTEST(c, eptr);
          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);
          }
-Line 1899 
 for (;;)
+Line 2137 
 for (;;)
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
-           GETCHARINC(c, eptr);
+           GETCHARINCTEST(c, eptr);
            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);
            }
          /* Control never gets here */
-Line 1914 
 for (;;)
+Line 2152 
 for (;;)
            {
            int len = 1;
            if (eptr >= md->end_subject) break;
-           GETCHARLEN(c, eptr, len);
+           GETCHARLENTEST(c, eptr, len);
            if (!_pcre_xclass(c, data)) break;
            eptr += len;
            }
-Line 1923 
 for (;;)
+Line 2161 
 for (;;)
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
            if (eptr-- == pp) break;        /* Stop if tried at original pos */
-           BACKCHAR(eptr)
+           if (utf8) BACKCHAR(eptr);
            }
          RRETURN(MATCH_NOMATCH);
          }
-Line 1989 
 for (;;)
+Line 2227 
 for (;;)
          if (fc != dc)
            {
  #ifdef SUPPORT_UCP
-           if (dc != _pcre_ucp_othercase(fc))
+           if (dc != UCD_OTHERCASE(fc))
  #endif
              RRETURN(MATCH_NOMATCH);
            }
-Line 2080 
 for (;;)
+Line 2318 
 for (;;)
  #ifdef SUPPORT_UCP
          unsigned int othercase;
          if ((ims & PCRE_CASELESS) != 0 &&
-             (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)
+             (othercase = UCD_OTHERCASE(fc)) != fc)
            oclength = _pcre_ord2utf8(othercase, occhars);
          else oclength = 0;
  #endif  /* SUPPORT_UCP */
-Line 2400 
 for (;;)
+Line 2638 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
              GETCHARINC(d, eptr);
              if (d < 256) d = md->lcc[d];
-             if (fi >= max || eptr >= md->end_subject || fc == d)
+             if (fc == d) RRETURN(MATCH_NOMATCH);
-               RRETURN(MATCH_NOMATCH);
              }
            }
          else
-Line 2509 
 for (;;)
+Line 2748 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
              GETCHARINC(d, eptr);
-             if (fi >= max || eptr >= md->end_subject || fc == d)
+             if (fc == d) RRETURN(MATCH_NOMATCH);
-               RRETURN(MATCH_NOMATCH);
              }
            }
          else
-Line 2676 
 for (;;)
+Line 2915 
 for (;;)
            for (i = 1; i <= min; i++)
              {
              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
-             GETCHARINC(c, eptr);
+             GETCHARINCTEST(c, eptr);
              }
            break;
-Line 2684 
 for (;;)
+Line 2923 
 for (;;)
            for (i = 1; i <= min; i++)
              {
              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
-             GETCHARINC(c, eptr);
+             GETCHARINCTEST(c, eptr);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_chartype = UCD_CHARTYPE(c);
              if ((prop_chartype == ucp_Lu ||
                   prop_chartype == ucp_Ll ||
                   prop_chartype == ucp_Lt) == prop_fail_result)
-Line 2697 
 for (;;)
+Line 2936 
 for (;;)
            for (i = 1; i <= min; i++)
              {
              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
-             GETCHARINC(c, eptr);
+             GETCHARINCTEST(c, eptr);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_category = UCD_CATEGORY(c);
              if ((prop_category == prop_value) == prop_fail_result)
                RRETURN(MATCH_NOMATCH);
              }
-Line 2708 
 for (;;)
+Line 2947 
 for (;;)
            for (i = 1; i <= min; i++)
              {
              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
-             GETCHARINC(c, eptr);
+             GETCHARINCTEST(c, eptr);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_chartype = UCD_CHARTYPE(c);
              if ((prop_chartype == prop_value) == prop_fail_result)
                RRETURN(MATCH_NOMATCH);
              }
-Line 2719 
 for (;;)
+Line 2958 
 for (;;)
            for (i = 1; i <= min; i++)
              {
              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
-             GETCHARINC(c, eptr);
+             GETCHARINCTEST(c, eptr);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_script = UCD_SCRIPT(c);
              if ((prop_script == prop_value) == prop_fail_result)
                RRETURN(MATCH_NOMATCH);
              }
-Line 2739 
 for (;;)
+Line 2978 
 for (;;)
          for (i = 1; i <= min; i++)
            {
            GETCHARINCTEST(c, eptr);
-           prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+           prop_category = UCD_CATEGORY(c);
            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);
            while (eptr < md->end_subject)
              {
-Line 2748 
 for (;;)
+Line 2987 
 for (;;)
                {
                GETCHARLEN(c, eptr, len);
                }
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_category = UCD_CATEGORY(c);
              if (prop_category != ucp_M) break;
              eptr += len;
              }
-Line 2766 
 for (;;)
+Line 3005 
 for (;;)
          case OP_ANY:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject ||
+           if (eptr >= md->end_subject || IS_NEWLINE(eptr))
-                ((ims & PCRE_DOTALL) == 0 && IS_NEWLINE(eptr)))
              RRETURN(MATCH_NOMATCH);
            eptr++;
            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
            }
          break;
+         case OP_ALLANY:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           eptr++;
+           while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
+           }
+         break;
          case OP_ANYBYTE:
          eptr += min;
          break;
-Line 2789 
 for (;;)
+Line 3036 
 for (;;)
              case 0x000d:
              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
              break;
              case 0x000a:
+             break;
              case 0x000b:
              case 0x000c:
              case 0x0085:
              case 0x2028:
              case 0x2029:
+             if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
+             break;
+             }
+           }
+         break;
+         case OP_NOT_HSPACE:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           GETCHARINC(c, eptr);
+           switch(c)
+             {
+             default: break;
+             case 0x09:      /* HT */
+             case 0x20:      /* SPACE */
+             case 0xa0:      /* NBSP */
+             case 0x1680:    /* OGHAM SPACE MARK */
+             case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
+             case 0x2000:    /* EN QUAD */
+             case 0x2001:    /* EM QUAD */
+             case 0x2002:    /* EN SPACE */
+             case 0x2003:    /* EM SPACE */
+             case 0x2004:    /* THREE-PER-EM SPACE */
+             case 0x2005:    /* FOUR-PER-EM SPACE */
+             case 0x2006:    /* SIX-PER-EM SPACE */
+             case 0x2007:    /* FIGURE SPACE */
+             case 0x2008:    /* PUNCTUATION SPACE */
+             case 0x2009:    /* THIN SPACE */
+             case 0x200A:    /* HAIR SPACE */
+             case 0x202f:    /* NARROW NO-BREAK SPACE */
+             case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
+             case 0x3000:    /* IDEOGRAPHIC SPACE */
+             RRETURN(MATCH_NOMATCH);
+             }
+           }
+         break;
+         case OP_HSPACE:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           GETCHARINC(c, eptr);
+           switch(c)
+             {
+             default: RRETURN(MATCH_NOMATCH);
+             case 0x09:      /* HT */
+             case 0x20:      /* SPACE */
+             case 0xa0:      /* NBSP */
+             case 0x1680:    /* OGHAM SPACE MARK */
+             case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
+             case 0x2000:    /* EN QUAD */
+             case 0x2001:    /* EM QUAD */
+             case 0x2002:    /* EN SPACE */
+             case 0x2003:    /* EM SPACE */
+             case 0x2004:    /* THREE-PER-EM SPACE */
+             case 0x2005:    /* FOUR-PER-EM SPACE */
+             case 0x2006:    /* SIX-PER-EM SPACE */
+             case 0x2007:    /* FIGURE SPACE */
+             case 0x2008:    /* PUNCTUATION SPACE */
+             case 0x2009:    /* THIN SPACE */
+             case 0x200A:    /* HAIR SPACE */
+             case 0x202f:    /* NARROW NO-BREAK SPACE */
+             case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
+             case 0x3000:    /* IDEOGRAPHIC SPACE */
+             break;
+             }
+           }
+         break;
+         case OP_NOT_VSPACE:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           GETCHARINC(c, eptr);
+           switch(c)
+             {
+             default: break;
+             case 0x0a:      /* LF */
+             case 0x0b:      /* VT */
+             case 0x0c:      /* FF */
+             case 0x0d:      /* CR */
+             case 0x85:      /* NEL */
+             case 0x2028:    /* LINE SEPARATOR */
+             case 0x2029:    /* PARAGRAPH SEPARATOR */
+             RRETURN(MATCH_NOMATCH);
+             }
+           }
+         break;
+         case OP_VSPACE:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           GETCHARINC(c, eptr);
+           switch(c)
+             {
+             default: RRETURN(MATCH_NOMATCH);
+             case 0x0a:      /* LF */
+             case 0x0b:      /* VT */
+             case 0x0c:      /* FF */
+             case 0x0d:      /* CR */
+             case 0x85:      /* NEL */
+             case 0x2028:    /* LINE SEPARATOR */
+             case 0x2029:    /* PARAGRAPH SEPARATOR */
              break;
              }
            }
-Line 2824 
 for (;;)
+Line 3179 
 for (;;)
          for (i = 1; i <= min; i++)
            {
            if (eptr >= md->end_subject ||
-              (*eptr < 128 && (md->ctypes[*eptr++] & ctype_space) != 0))
+              (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0))
              RRETURN(MATCH_NOMATCH);
-           while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
+           while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
            }
          break;
-Line 2844 
 for (;;)
+Line 3199 
 for (;;)
          for (i = 1; i <= min; i++)
            {
            if (eptr >= md->end_subject ||
-              (*eptr < 128 && (md->ctypes[*eptr++] & ctype_word) != 0))
+              (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0))
              RRETURN(MATCH_NOMATCH);
-           while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
+           while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
            }
          break;
-Line 2874 
 for (;;)
+Line 3229 
 for (;;)
        switch(ctype)
          {
          case OP_ANY:
-         if ((ims & PCRE_DOTALL) == 0)
+         for (i = 1; i <= min; i++)
            {
-           for (i = 1; i <= min; i++)
+           if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
-             {
+           eptr++;
-             if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
-             eptr++;
-             }
            }
-         else eptr += min;
+         break;
+         case OP_ALLANY:
+         eptr += min;
          break;
          case OP_ANYBYTE:
-Line 2903 
 for (;;)
+Line 3258 
 for (;;)
              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
              break;
              case 0x000a:
+             break;
              case 0x000b:
              case 0x000c:
              case 0x0085:
+             if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
+             break;
+             }
+           }
+         break;
+         case OP_NOT_HSPACE:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           switch(*eptr++)
+             {
+             default: break;
+             case 0x09:      /* HT */
+             case 0x20:      /* SPACE */
+             case 0xa0:      /* NBSP */
+             RRETURN(MATCH_NOMATCH);
+             }
+           }
+         break;
+         case OP_HSPACE:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           switch(*eptr++)
+             {
+             default: RRETURN(MATCH_NOMATCH);
+             case 0x09:      /* HT */
+             case 0x20:      /* SPACE */
+             case 0xa0:      /* NBSP */
+             break;
+             }
+           }
+         break;
+         case OP_NOT_VSPACE:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           switch(*eptr++)
+             {
+             default: break;
+             case 0x0a:      /* LF */
+             case 0x0b:      /* VT */
+             case 0x0c:      /* FF */
+             case 0x0d:      /* CR */
+             case 0x85:      /* NEL */
+             RRETURN(MATCH_NOMATCH);
+             }
+           }
+         break;
+         case OP_VSPACE:
+         for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           switch(*eptr++)
+             {
+             default: RRETURN(MATCH_NOMATCH);
+             case 0x0a:      /* LF */
+             case 0x0b:      /* VT */
+             case 0x0c:      /* FF */
+             case 0x0d:      /* CR */
+             case 0x85:      /* NEL */
              break;
              }
            }
-Line 2981 
 for (;;)
+Line 3403 
 for (;;)
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
              GETCHARINC(c, eptr);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_chartype = UCD_CHARTYPE(c);
              if ((prop_chartype == ucp_Lu ||
                   prop_chartype == ucp_Ll ||
                   prop_chartype == ucp_Lt) == prop_fail_result)
-Line 2996 
 for (;;)
+Line 3418 
 for (;;)
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
              GETCHARINC(c, eptr);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_category = UCD_CATEGORY(c);
              if ((prop_category == prop_value) == prop_fail_result)
                RRETURN(MATCH_NOMATCH);
              }
-Line 3009 
 for (;;)
+Line 3431 
 for (;;)
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
              GETCHARINC(c, eptr);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_chartype = UCD_CHARTYPE(c);
              if ((prop_chartype == prop_value) == prop_fail_result)
                RRETURN(MATCH_NOMATCH);
              }
-Line 3022 
 for (;;)
+Line 3444 
 for (;;)
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
              GETCHARINC(c, eptr);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_script = UCD_SCRIPT(c);
              if ((prop_script == prop_value) == prop_fail_result)
                RRETURN(MATCH_NOMATCH);
              }
-Line 3044 
 for (;;)
+Line 3466 
 for (;;)
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
            GETCHARINCTEST(c, eptr);
-           prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+           prop_category = UCD_CATEGORY(c);
            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);
            while (eptr < md->end_subject)
              {
-Line 3053 
 for (;;)
+Line 3475 
 for (;;)
                {
                GETCHARLEN(c, eptr, len);
                }
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_category = UCD_CATEGORY(c);
              if (prop_category != ucp_M) break;
              eptr += len;
              }
-Line 3072 
 for (;;)
+Line 3494 
 for (;;)
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
            if (fi >= max || eptr >= md->end_subject ||
-                (ctype == OP_ANY && (ims & PCRE_DOTALL) == 0 &&
+                (ctype == OP_ANY && IS_NEWLINE(eptr)))
-                 IS_NEWLINE(eptr)))
              RRETURN(MATCH_NOMATCH);
            GETCHARINC(c, eptr);
            switch(ctype)
              {
-             case OP_ANY:        /* This is the DOTALL case */
+             case OP_ANY:        /* This is the non-NL case */
-             break;
+             case OP_ALLANY:
              case OP_ANYBYTE:
              break;
-Line 3093 
 for (;;)
+Line 3513 
 for (;;)
                if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
                break;
                case 0x000a:
+               break;
                case 0x000b:
                case 0x000c:
                case 0x0085:
                case 0x2028:
                case 0x2029:
+               if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
+               break;
+               }
+             break;
+             case OP_NOT_HSPACE:
+             switch(c)
+               {
+               default: break;
+               case 0x09:      /* HT */
+               case 0x20:      /* SPACE */
+               case 0xa0:      /* NBSP */
+               case 0x1680:    /* OGHAM SPACE MARK */
+               case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
+               case 0x2000:    /* EN QUAD */
+               case 0x2001:    /* EM QUAD */
+               case 0x2002:    /* EN SPACE */
+               case 0x2003:    /* EM SPACE */
+               case 0x2004:    /* THREE-PER-EM SPACE */
+               case 0x2005:    /* FOUR-PER-EM SPACE */
+               case 0x2006:    /* SIX-PER-EM SPACE */
+               case 0x2007:    /* FIGURE SPACE */
+               case 0x2008:    /* PUNCTUATION SPACE */
+               case 0x2009:    /* THIN SPACE */
+               case 0x200A:    /* HAIR SPACE */
+               case 0x202f:    /* NARROW NO-BREAK SPACE */
+               case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
+               case 0x3000:    /* IDEOGRAPHIC SPACE */
+               RRETURN(MATCH_NOMATCH);
+               }
+             break;
+             case OP_HSPACE:
+             switch(c)
+               {
+               default: RRETURN(MATCH_NOMATCH);
+               case 0x09:      /* HT */
+               case 0x20:      /* SPACE */
+               case 0xa0:      /* NBSP */
+               case 0x1680:    /* OGHAM SPACE MARK */
+               case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
+               case 0x2000:    /* EN QUAD */
+               case 0x2001:    /* EM QUAD */
+               case 0x2002:    /* EN SPACE */
+               case 0x2003:    /* EM SPACE */
+               case 0x2004:    /* THREE-PER-EM SPACE */
+               case 0x2005:    /* FOUR-PER-EM SPACE */
+               case 0x2006:    /* SIX-PER-EM SPACE */
+               case 0x2007:    /* FIGURE SPACE */
+               case 0x2008:    /* PUNCTUATION SPACE */
+               case 0x2009:    /* THIN SPACE */
+               case 0x200A:    /* HAIR SPACE */
+               case 0x202f:    /* NARROW NO-BREAK SPACE */
+               case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
+               case 0x3000:    /* IDEOGRAPHIC SPACE */
+               break;
+               }
+             break;
+             case OP_NOT_VSPACE:
+             switch(c)
+               {
+               default: break;
+               case 0x0a:      /* LF */
+               case 0x0b:      /* VT */
+               case 0x0c:      /* FF */
+               case 0x0d:      /* CR */
+               case 0x85:      /* NEL */
+               case 0x2028:    /* LINE SEPARATOR */
+               case 0x2029:    /* PARAGRAPH SEPARATOR */
+               RRETURN(MATCH_NOMATCH);
+               }
+             break;
+             case OP_VSPACE:
+             switch(c)
+               {
+               default: RRETURN(MATCH_NOMATCH);
+               case 0x0a:      /* LF */
+               case 0x0b:      /* VT */
+               case 0x0c:      /* FF */
+               case 0x0d:      /* CR */
+               case 0x85:      /* NEL */
+               case 0x2028:    /* LINE SEPARATOR */
+               case 0x2029:    /* PARAGRAPH SEPARATOR */
                break;
                }
              break;
-Line 3146 
 for (;;)
+Line 3653 
 for (;;)
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
            if (fi >= max || eptr >= md->end_subject ||
-                ((ims & PCRE_DOTALL) == 0 && IS_NEWLINE(eptr)))
+                (ctype == OP_ANY && IS_NEWLINE(eptr)))
              RRETURN(MATCH_NOMATCH);
            c = *eptr++;
            switch(ctype)
              {
-             case OP_ANY:   /* This is the DOTALL case */
+             case OP_ANY:     /* This is the non-NL case */
-             break;
+             case OP_ALLANY:
              case OP_ANYBYTE:
              break;
-Line 3165 
 for (;;)
+Line 3671 
 for (;;)
                case 0x000d:
                if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
                break;
                case 0x000a:
+               break;
                case 0x000b:
                case 0x000c:
                case 0x0085:
+               if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
+               break;
+               }
+             break;
+             case OP_NOT_HSPACE:
+             switch(c)
+               {
+               default: break;
+               case 0x09:      /* HT */
+               case 0x20:      /* SPACE */
+               case 0xa0:      /* NBSP */
+               RRETURN(MATCH_NOMATCH);
+               }
+             break;
+             case OP_HSPACE:
+             switch(c)
+               {
+               default: RRETURN(MATCH_NOMATCH);
+               case 0x09:      /* HT */
+               case 0x20:      /* SPACE */
+               case 0xa0:      /* NBSP */
+               break;
+               }
+             break;
+             case OP_NOT_VSPACE:
+             switch(c)
+               {
+               default: break;
+               case 0x0a:      /* LF */
+               case 0x0b:      /* VT */
+               case 0x0c:      /* FF */
+               case 0x0d:      /* CR */
+               case 0x85:      /* NEL */
+               RRETURN(MATCH_NOMATCH);
+               }
+             break;
+             case OP_VSPACE:
+             switch(c)
+               {
+               default: RRETURN(MATCH_NOMATCH);
+               case 0x0a:      /* LF */
+               case 0x0b:      /* VT */
+               case 0x0c:      /* FF */
+               case 0x0d:      /* CR */
+               case 0x85:      /* NEL */
                break;
                }
              break;
-Line 3235 
 for (;;)
+Line 3793 
 for (;;)
              int len = 1;
              if (eptr >= md->end_subject) break;
              GETCHARLEN(c, eptr, len);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_chartype = UCD_CHARTYPE(c);
              if ((prop_chartype == ucp_Lu ||
                   prop_chartype == ucp_Ll ||
                   prop_chartype == ucp_Lt) == prop_fail_result)
-Line 3250 
 for (;;)
+Line 3808 
 for (;;)
              int len = 1;
              if (eptr >= md->end_subject) break;
              GETCHARLEN(c, eptr, len);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_category = UCD_CATEGORY(c);
              if ((prop_category == prop_value) == prop_fail_result)
                break;
              eptr+= len;
-Line 3263 
 for (;;)
+Line 3821 
 for (;;)
              int len = 1;
              if (eptr >= md->end_subject) break;
              GETCHARLEN(c, eptr, len);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_chartype = UCD_CHARTYPE(c);
              if ((prop_chartype == prop_value) == prop_fail_result)
                break;
              eptr+= len;
-Line 3276 
 for (;;)
+Line 3834 
 for (;;)
              int len = 1;
              if (eptr >= md->end_subject) break;
              GETCHARLEN(c, eptr, len);
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_script = UCD_SCRIPT(c);
              if ((prop_script == prop_value) == prop_fail_result)
                break;
              eptr+= len;
-Line 3292 
 for (;;)
+Line 3850 
 for (;;)
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
            if (eptr-- == pp) break;        /* Stop if tried at original pos */
-           BACKCHAR(eptr);
+           if (utf8) BACKCHAR(eptr);
            }
          }
-Line 3305 
 for (;;)
+Line 3863 
 for (;;)
            {
            if (eptr >= md->end_subject) break;
            GETCHARINCTEST(c, eptr);
-           prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+           prop_category = UCD_CATEGORY(c);
            if (prop_category == ucp_M) break;
            while (eptr < md->end_subject)
              {
-Line 3314 
 for (;;)
+Line 3872 
 for (;;)
                {
                GETCHARLEN(c, eptr, len);
                }
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_category = UCD_CATEGORY(c);
              if (prop_category != ucp_M) break;
              eptr += len;
              }
-Line 3331 
 for (;;)
+Line 3889 
 for (;;)
            for (;;)                        /* Move back over one extended */
              {
              int len = 1;
-             BACKCHAR(eptr);
              if (!utf8) c = *eptr; else
                {
+               BACKCHAR(eptr);
                GETCHARLEN(c, eptr, len);
                }
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
+             prop_category = UCD_CATEGORY(c);
              if (prop_category != ucp_M) break;
              eptr--;
              }
-Line 3354 
 for (;;)
+Line 3912 
 for (;;)
          switch(ctype)
            {
            case OP_ANY:
-           /* Special code is required for UTF8, but when the maximum is
-           unlimited we don't need it, so we repeat the non-UTF8 code. This is
-           probably worth it, because .* is quite a common idiom. */
            if (max < INT_MAX)
              {
-             if ((ims & PCRE_DOTALL) == 0)
+             for (i = min; i < max; i++)
-               {
-               for (i = min; i < max; i++)
-                 {
-                 if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
-                 eptr++;
-                 while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
-                 }
-               }
-             else
                {
-               for (i = min; i < max; i++)
+               if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
-                 {
+               eptr++;
-                 if (eptr >= md->end_subject) break;
+               while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
-                 eptr++;
-                 while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
-                 }
                }
              }
-Line 3385 
 for (;;)
+Line 3926 
 for (;;)
            else
              {
-             if ((ims & PCRE_DOTALL) == 0)
+             for (i = min; i < max; i++)
                {
-               for (i = min; i < max; i++)
+               if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
-                 {
+               eptr++;
-                 if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
+               while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
-                 eptr++;
-                 }
-               break;
                }
-             else
+             }
+           break;
+           case OP_ALLANY:
+           if (max < INT_MAX)
+             {
+             for (i = min; i < max; i++)
                {
-               c = max - min;
+               if (eptr >= md->end_subject) break;
-               if (c > (unsigned int)(md->end_subject - eptr))
+               eptr++;
-                 c = md->end_subject - eptr;
+               while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
-               eptr += c;
                }
              }
+           else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
            break;
            /* The byte case is the same as non-UTF8 */
-Line 3426 
 for (;;)
+Line 3970 
 for (;;)
                }
              else
                {
-               if (c != 0x000a && c != 0x000b && c != 0x000c &&
+               if (c != 0x000a &&
-                   c != 0x0085 && c != 0x2028 && c != 0x2029)
+                   (md->bsr_anycrlf ||
+                    (c != 0x000b && c != 0x000c &&
+                     c != 0x0085 && c != 0x2028 && c != 0x2029)))
                  break;
                eptr += len;
                }
              }
            break;
+           case OP_NOT_HSPACE:
+           case OP_HSPACE:
+           for (i = min; i < max; i++)
+             {
+             BOOL gotspace;
+             int len = 1;
+             if (eptr >= md->end_subject) break;
+             GETCHARLEN(c, eptr, len);
+             switch(c)
+               {
+               default: gotspace = FALSE; break;
+               case 0x09:      /* HT */
+               case 0x20:      /* SPACE */
+               case 0xa0:      /* NBSP */
+               case 0x1680:    /* OGHAM SPACE MARK */
+               case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
+               case 0x2000:    /* EN QUAD */
+               case 0x2001:    /* EM QUAD */
+               case 0x2002:    /* EN SPACE */
+               case 0x2003:    /* EM SPACE */
+               case 0x2004:    /* THREE-PER-EM SPACE */
+               case 0x2005:    /* FOUR-PER-EM SPACE */
+               case 0x2006:    /* SIX-PER-EM SPACE */
+               case 0x2007:    /* FIGURE SPACE */
+               case 0x2008:    /* PUNCTUATION SPACE */
+               case 0x2009:    /* THIN SPACE */
+               case 0x200A:    /* HAIR SPACE */
+               case 0x202f:    /* NARROW NO-BREAK SPACE */
+               case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
+               case 0x3000:    /* IDEOGRAPHIC SPACE */
+               gotspace = TRUE;
+               break;
+               }
+             if (gotspace == (ctype == OP_NOT_HSPACE)) break;
+             eptr += len;
+             }
+           break;
+           case OP_NOT_VSPACE:
+           case OP_VSPACE:
+           for (i = min; i < max; i++)
+             {
+             BOOL gotspace;
+             int len = 1;
+             if (eptr >= md->end_subject) break;
+             GETCHARLEN(c, eptr, len);
+             switch(c)
+               {
+               default: gotspace = FALSE; break;
+               case 0x0a:      /* LF */
+               case 0x0b:      /* VT */
+               case 0x0c:      /* FF */
+               case 0x0d:      /* CR */
+               case 0x85:      /* NEL */
+               case 0x2028:    /* LINE SEPARATOR */
+               case 0x2029:    /* PARAGRAPH SEPARATOR */
+               gotspace = TRUE;
+               break;
+               }
+             if (gotspace == (ctype == OP_NOT_VSPACE)) break;
+             eptr += len;
+             }
+           break;
            case OP_NOT_DIGIT:
            for (i = min; i < max; i++)
              {
-Line 3516 
 for (;;)
+Line 4126 
 for (;;)
            }
          }
        else
- #endif
+ #endif  /* SUPPORT_UTF8 */
        /* Not UTF-8 mode */
          {
          switch(ctype)
            {
            case OP_ANY:
-           if ((ims & PCRE_DOTALL) == 0)
+           for (i = min; i < max; i++)
              {
-             for (i = min; i < max; i++)
+             if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
-               {
+             eptr++;
-               if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
-               eptr++;
-               }
-             break;
              }
-           /* For DOTALL case, fall through and treat as \C */
+           break;
+           case OP_ALLANY:
            case OP_ANYBYTE:
            c = max - min;
            if (c > (unsigned int)(md->end_subject - eptr))
-Line 3553 
 for (;;)
+Line 4160 
 for (;;)
                }
              else
                {
-               if (c != 0x000a && c != 0x000b && c != 0x000c && c != 0x0085)
+               if (c != 0x000a &&
+                   (md->bsr_anycrlf ||
+                     (c != 0x000b && c != 0x000c && c != 0x0085)))
                  break;
                eptr++;
                }
              }
            break;
+           case OP_NOT_HSPACE:
+           for (i = min; i < max; i++)
+             {
+             if (eptr >= md->end_subject) break;
+             c = *eptr;
+             if (c == 0x09 || c == 0x20 || c == 0xa0) break;
+             eptr++;
+             }
+           break;
+           case OP_HSPACE:
+           for (i = min; i < max; i++)
+             {
+             if (eptr >= md->end_subject) break;
+             c = *eptr;
+             if (c != 0x09 && c != 0x20 && c != 0xa0) break;
+             eptr++;
+             }
+           break;
+           case OP_NOT_VSPACE:
+           for (i = min; i < max; i++)
+             {
+             if (eptr >= md->end_subject) break;
+             c = *eptr;
+             if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85)
+               break;
+             eptr++;
+             }
+           break;
+           case OP_VSPACE:
+           for (i = min; i < max; i++)
+             {
+             if (eptr >= md->end_subject) break;
+             c = *eptr;
+             if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85)
+               break;
+             eptr++;
+             }
+           break;
            case OP_NOT_DIGIT:
            for (i = min; i < max; i++)
              {
-Line 3661 
 HEAP_RETURN:
+Line 4312 
 HEAP_RETURN:
  switch (frame->Xwhere)
    {
    LBL( 1) LBL( 2) LBL( 3) LBL( 4) LBL( 5) LBL( 6) LBL( 7) LBL( 8)
-   LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(16)
+   LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)
-   LBL(17) LBL(18) LBL(19) LBL(20) LBL(21) LBL(22) LBL(23) LBL(24)
+   LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)
-   LBL(25) LBL(26) LBL(27) LBL(28) LBL(29) LBL(30) LBL(31) LBL(32)
+   LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)
-   LBL(33) LBL(34) LBL(35) LBL(36) LBL(37) LBL(38) LBL(39) LBL(40)
+   LBL(53) LBL(54)
-   LBL(41) LBL(42) LBL(43) LBL(44) LBL(45) LBL(46) LBL(47)
+ #ifdef SUPPORT_UTF8
+   LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)
+   LBL(32) LBL(34) LBL(42) LBL(46)
+ #ifdef SUPPORT_UCP
+   LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)
+ #endif  /* SUPPORT_UCP */
+ #endif  /* SUPPORT_UTF8 */
    default:
    DPRINTF(("jump error in pcre match: label %d non-existent\n", frame->Xwhere));
    return PCRE_ERROR_INTERNAL;
-Line 3684 
 Undefine all the macros that were define
+Line 4341 
 Undefine all the macros that were define
  #ifdef NO_RECURSE
  #undef eptr
  #undef ecode
+ #undef mstart
  #undef offset_top
  #undef ims
  #undef eptrb
-Line 3756 
 Returns:          > 0 => success; value
+Line 4414 
 Returns:          > 0 => success; value
                   < -1 => some kind of unexpected problem
  */
- PCRE_EXP_DEFN int
+ PCRE_EXP_DEFN int PCRE_CALL_CONVENTION
  pcre_exec(const pcre *argument_re, const pcre_extra *extra_data,
    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
    int offsetcount)
-Line 3781 
 const uschar *start_bits = NULL;
+Line 4439 
 const uschar *start_bits = NULL;
  USPTR start_match = (USPTR)subject + start_offset;
  USPTR end_subject;
  USPTR req_byte_ptr = start_match - 1;
- eptrblock eptrchain[EPTR_WORK_SIZE];
  pcre_study_data internal_study;
  const pcre_study_data *study;
-Line 3844 
 if (re->magic_number != MAGIC_NUMBER)
+Line 4501 
 if (re->magic_number != MAGIC_NUMBER)
  /* Set up other data */
  anchored = ((re->options | options) & PCRE_ANCHORED) != 0;
- startline = (re->options & PCRE_STARTLINE) != 0;
+ startline = (re->flags & PCRE_STARTLINE) != 0;
  firstline = (re->options & PCRE_FIRSTLINE) != 0;
  /* The code starts after the real_pcre block and the capture name table. */
-Line 3859 
 end_subject = md->end_subject;
+Line 4516 
 end_subject = md->end_subject;
  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
  utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;
+ md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
  md->notbol = (options & PCRE_NOTBOL) != 0;
  md->noteol = (options & PCRE_NOTEOL) != 0;
-Line 3867 
 md->partial = (options & PCRE_PARTIAL) !
+Line 4525 
 md->partial = (options & PCRE_PARTIAL) !
  md->hitend = FALSE;
  md->recursive = NULL;                   /* No recursion at top level */
- md->eptrchain = eptrchain;              /* Make workspace generally available */
  md->lcc = tables + lcc_offset;
  md->ctypes = tables + ctypes_offset;
+ /* Handle different \R options. */
+ switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
+   {
+   case 0:
+   if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0)
+     md->bsr_anycrlf = (re->options & PCRE_BSR_ANYCRLF) != 0;
+   else
+ #ifdef BSR_ANYCRLF
+   md->bsr_anycrlf = TRUE;
+ #else
+   md->bsr_anycrlf = FALSE;
+ #endif
+   break;
+   case PCRE_BSR_ANYCRLF:
+   md->bsr_anycrlf = TRUE;
+   break;
+   case PCRE_BSR_UNICODE:
+   md->bsr_anycrlf = FALSE;
+   break;
+   default: return PCRE_ERROR_BADNEWLINE;
+   }
  /* Handle different types of newline. The three bits give eight cases. If
  nothing is set at run time, whatever was used at compile time applies. */
- switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options : (pcre_uint32)options) &
+ switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options :
-        PCRE_NEWLINE_BITS)
+         (pcre_uint32)options) & PCRE_NEWLINE_BITS)
    {
    case 0: newline = NEWLINE; break;   /* Compile-time default */
    case PCRE_NEWLINE_CR: newline = '\r'; break;
-Line 3915 
 else
+Line 4598 
 else
  /* Partial matching is supported only for a restricted set of regexes at the
  moment. */
- if (md->partial && (re->options & PCRE_NOPARTIAL) != 0)
+ if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
    return PCRE_ERROR_BADPARTIAL;
  /* Check a UTF-8 string if required. Unfortunately there's no way of passing
-Line 3992 
 studied, there may be a bitmap of possib
+Line 4675 
 studied, there may be a bitmap of possib
  if (!anchored)
    {
-   if ((re->options & PCRE_FIRSTSET) != 0)
+   if ((re->flags & PCRE_FIRSTSET) != 0)
      {
      first_byte = re->first_byte & 255;
      if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE)
-Line 4007 
 if (!anchored)
+Line 4690 
 if (!anchored)
  /* For anchored or unanchored matches, there may be a "last known required
  character" set. */
- if ((re->options & PCRE_REQCHSET) != 0)
+ if ((re->flags & PCRE_REQCHSET) != 0)
    {
    req_byte = re->req_byte & 255;
    req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;
-Line 4023 
 the loop runs just once. */
+Line 4706 
 the loop runs just once. */
  for(;;)
    {
    USPTR save_end_subject = end_subject;
+   USPTR new_start_match;
    /* Reset the maximum number of extractions we might see. */
-Line 4042 
 for(;;)
+Line 4726 
 for(;;)
    if (firstline)
      {
      USPTR t = start_match;
+ #ifdef SUPPORT_UTF8
+     if (utf8)
+       {
+       while (t < md->end_subject && !IS_NEWLINE(t))
+         {
+         t++;
+         while (t < end_subject && (*t & 0xc0) == 0x80) t++;
+         }
+       }
+     else
+ #endif
      while (t < md->end_subject && !IS_NEWLINE(t)) t++;
      end_subject = t;
      }
-   /* Now test for a unique first byte */
+   /* Now advance to a unique first byte if there is one. */
    if (first_byte >= 0)
      {
      if (first_byte_caseless)
-       while (start_match < end_subject &&
+       while (start_match < end_subject && md->lcc[*start_match] != first_byte)
-              md->lcc[*start_match] != first_byte)
          start_match++;
      else
        while (start_match < end_subject && *start_match != first_byte)
          start_match++;
      }
-   /* Or to just after a linebreak for a multiline match if possible */
+   /* Or to just after a linebreak for a multiline match */
    else if (startline)
      {
      if (start_match > md->start_subject + start_offset)
        {
-       while (start_match <= end_subject && !WAS_NEWLINE(start_match))
+ #ifdef SUPPORT_UTF8
+       if (utf8)
+         {
+         while (start_match < end_subject && !WAS_NEWLINE(start_match))
+           {
+           start_match++;
+           while(start_match < end_subject && (*start_match & 0xc0) == 0x80)
+             start_match++;
+           }
+         }
+       else
+ #endif
+       while (start_match < end_subject && !WAS_NEWLINE(start_match))
          start_match++;
        /* If we have just passed a CR and the newline option is ANY or ANYCRLF,
-Line 4080 
 for(;;)
+Line 4786 
 for(;;)
        }
      }
-   /* Or to a non-unique first char after study */
+   /* Or to a non-unique first byte after study */
    else if (start_bits != NULL)
      {
      while (start_match < end_subject)
        {
        register unsigned int c = *start_match;
-       if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++; else break;
+       if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++;
+         else break;
        }
      }
-Line 4163 
 for(;;)
+Line 4870 
 for(;;)
    /* OK, we can now run the match. */
-   md->start_match = start_match;
+   md->start_match_ptr = start_match;
    md->match_call_count = 0;
-   md->eptrn = 0;                          /* Next free eptrchain slot */
+   rc = match(start_match, md->start_code, start_match, 2, md, ims, NULL, 0, 0);
-   rc = match(start_match, md->start_code, 2, md, ims, NULL, 0, 0);
+   switch(rc)
+     {
+     /* NOMATCH and PRUNE advance by one character. THEN at this level acts
+     exactly like PRUNE. */
-   /* Any return other than MATCH_NOMATCH breaks the loop. */
+     case MATCH_NOMATCH:
+     case MATCH_PRUNE:
+     case MATCH_THEN:
+     new_start_match = start_match + 1;
+ #ifdef SUPPORT_UTF8
+     if (utf8)
+       while(new_start_match < end_subject && (*new_start_match & 0xc0) == 0x80)
+         new_start_match++;
+ #endif
+     break;
-   if (rc != MATCH_NOMATCH) break;
+     /* SKIP passes back the next starting point explicitly. */
+     case MATCH_SKIP:
+     new_start_match = md->start_match_ptr;
+     break;
+     /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */
+     case MATCH_COMMIT:
+     rc = MATCH_NOMATCH;
+     goto ENDLOOP;
+     /* Any other return is some kind of error. */
+     default:
+     goto ENDLOOP;
+     }
+   /* Control reaches here for the various types of "no match at this point"
+   result. Reset the code to MATCH_NOMATCH for subsequent checking. */
+   rc = MATCH_NOMATCH;
    /* If PCRE_FIRSTLINE is set, the match must happen before or at the first
    newline in the subject (though it may continue over the newline). Therefore,
-Line 4178 
 for(;;)
+Line 4919 
 for(;;)
    if (firstline && IS_NEWLINE(start_match)) break;
-   /* Advance the match position by one character. */
+   /* Advance to new matching position */
-   start_match++;
+   start_match = new_start_match;
- #ifdef SUPPORT_UTF8
-   if (utf8)
-     while(start_match < end_subject && (*start_match & 0xc0) == 0x80)
-       start_match++;
- #endif
    /* Break the loop if the pattern is anchored or if we have passed the end of
    the subject. */
    if (anchored || start_match > end_subject) break;
-   /* If we have just passed a CR and the newline option is CRLF or ANY or
+   /* If we have just passed a CR and we are now at a LF, and the pattern does
-   ANYCRLF, and we are now at a LF, advance the match position by one more
+   not contain any explicit matches for \r or \n, and the newline option is CRLF
-   character. */
+   or ANY or ANYCRLF, advance the match position by one more character. */
    if (start_match[-1] == '\r' &&
-        (md->nltype == NLTYPE_ANY ||
+       start_match < end_subject &&
-         md->nltype == NLTYPE_ANYCRLF ||
+       *start_match == '\n' &&
-         md->nllen == 2) &&
+       (re->flags & PCRE_HASCRORLF) == 0 &&
-        start_match < end_subject &&
+         (md->nltype == NLTYPE_ANY ||
-        *start_match == '\n')
+          md->nltype == NLTYPE_ANYCRLF ||
+          md->nllen == 2))
      start_match++;
    }   /* End of for(;;) "bumpalong" loop */
-Line 4211 
 for(;;)
+Line 4948 
 for(;;)
  /* We reach here when rc is not MATCH_NOMATCH, or if one of the stopping
  conditions is true:
- (1) The pattern is anchored;
+ (1) The pattern is anchored or the match was failed by (*COMMIT);
  (2) We are past the end of the subject;
-Line 4226 
 processing, copy those that we can. In t
+Line 4963 
 processing, copy those that we can. In t
  certain parts of the pattern were not used, even though there are more
  capturing parentheses than vector slots. */
+ ENDLOOP:
  if (rc == MATCH_MATCH)
    {
    if (using_temporary_offsets)
-Line 4246 
 if (rc == MATCH_MATCH)
+Line 4985 
 if (rc == MATCH_MATCH)
    rc = md->offset_overflow? 0 : md->end_offset_top/2;
-   /* If there is space, set up the whole thing as substring 0. */
+   /* If there is space, set up the whole thing as substring 0. The value of
+   md->start_match_ptr might be modified if \K was encountered on the success
+   matching path. */
    if (offsetcount < 2) rc = 0; else
      {
-     offsets[0] = start_match - md->start_subject;
+     offsets[0] = md->start_match_ptr - md->start_subject;
      offsets[1] = md->end_match_ptr - md->start_subject;
      }

 Legend:



Removed from v.165
 


changed lines


 
Added in v.384
 Legend:



Removed from v.165
 


changed lines


 
Added in v.384
-Removed from v.165
+Added in v.384

webmaster@exim.org	ViewVC Help
Powered by ViewVC 1.1.12