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

Diff of /code/trunk/pcre_exec.c

Parent Directory | Revision Log | View Patch Patch

-revision 227 by ph10,
Tue Aug 21 15:00:15 2007 UTC
+revision 500 by ph10,
Sat Mar  6 19:00:29 2010 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) 1997-2010 University of Cambridge
  -----------------------------------------------------------------------------
  Redistribution and use in source and binary forms, with or without
 Line 43 
 pattern matching using an NFA algorithm,
  possible. There are also some static supporting functions. */
  #ifdef HAVE_CONFIG_H
- #include <config.h>
+ #include "config.h"
  #endif
  #define NLBLOCK md             /* Block containing newline information */
 Line 89 
 static const char rep_max[] = { 0, 0, 0,
- #ifdef DEBUG
+ #ifdef PCRE_DEBUG
  /*************************************************
  *        Debugging function to print chars       *
  *************************************************/
 Line 141 
 match_ref(int offset, register USPTR ept
  {
  USPTR p = md->start_subject + md->offset_vector[offset];
- #ifdef DEBUG
+ #ifdef PCRE_DEBUG
  if (eptr >= md->end_subject)
    printf("matching subject <null>");
  else
 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 228 
 actuall used in this definition. */
+Line 254 
 actuall used in this definition. */
  #ifndef NO_RECURSE
  #define REGISTER register
- #ifdef DEBUG
+ #ifdef PCRE_DEBUG
  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
    { \
    printf("match() called in line %d\n", __LINE__); \
-Line 296 
 typedef struct heapframe {
+Line 322 
 typedef struct heapframe {
    /* Function arguments that may change */
-   const uschar *Xeptr;
+   USPTR Xeptr;
    const uschar *Xecode;
-   const uschar *Xmstart;
+   USPTR Xmstart;
    int Xoffset_top;
    long int Xims;
    eptrblock *Xeptrb;
-Line 307 
 typedef struct heapframe {
+Line 333 
 typedef struct heapframe {
    /* Function local variables */
-   const uschar *Xcallpat;
+   USPTR Xcallpat;
-   const uschar *Xcharptr;
+ #ifdef SUPPORT_UTF8
-   const uschar *Xdata;
+   USPTR Xcharptr;
-   const uschar *Xnext;
+ #endif
-   const uschar *Xpp;
+   USPTR Xdata;
-   const uschar *Xprev;
+   USPTR Xnext;
-   const uschar *Xsaved_eptr;
+   USPTR Xpp;
+   USPTR Xprev;
+   USPTR Xsaved_eptr;
    recursion_info Xnew_recursive;
-Line 334 
 typedef struct heapframe {
+Line 362 
 typedef struct heapframe {
    uschar Xocchars[8];
  #endif
+   int Xcodelink;
    int Xctype;
    unsigned int Xfc;
    int Xfi;
-Line 369 
 typedef struct heapframe {
+Line 398 
 typedef struct heapframe {
  /* This function is called recursively in many circumstances. Whenever it
  returns a negative (error) response, the outer incarnation must also return the
- same response.
+ same response. */
+ /* These macros pack up tests that are used for partial matching, and which
+ appears several times in the code. We set the "hit end" flag if the pointer is
+ at the end of the subject and also past the start of the subject (i.e.
+ something has been matched). For hard partial matching, we then return
+ immediately. The second one is used when we already know we are past the end of
+ the subject. */
- Performance note: It might be tempting to extract commonly used fields from the
+ #define CHECK_PARTIAL()\
- md structure (e.g. utf8, end_subject) into individual variables to improve
+   if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\
+     {\
+     md->hitend = TRUE;\
+     if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\
+     }
+ #define SCHECK_PARTIAL()\
+   if (md->partial != 0 && eptr > mstart)\
+     {\
+     md->hitend = TRUE;\
+     if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\
+     }
+ /* Performance note: It might be tempting to extract commonly used fields from
+ the md structure (e.g. utf8, end_subject) into individual variables to improve
  performance. Tests using gcc on a SPARC disproved this; in the first case, it
  made performance worse.
-Line 399 
 Returns:       MATCH_MATCH if matched
+Line 450 
 Returns:       MATCH_MATCH if matched
  */
  static int
- match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart,
+ match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,
    int flags, unsigned int rdepth)
  {
-Line 413 
 register unsigned int c;   /* Character
+Line 464 
 register unsigned int c;   /* Character
  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
  BOOL minimize, possessive; /* Quantifier options */
+ int condcode;
  /* When recursion is not being used, all "local" variables that have to be
  preserved over calls to RMATCH() are part of a "frame" which is obtained from
-Line 455 
 HEAP_RECURSE:
+Line 507 
 HEAP_RECURSE:
  #define charptr            frame->Xcharptr
  #endif
  #define callpat            frame->Xcallpat
+ #define codelink           frame->Xcodelink
  #define data               frame->Xdata
  #define next               frame->Xnext
  #define pp                 frame->Xpp
-Line 535 
 int oclength;
+Line 588 
 int oclength;
  uschar occhars[8];
  #endif
+ int codelink;
  int ctype;
  int length;
  int max;
-Line 568 
 TAIL_RECURSE:
+Line 622 
 TAIL_RECURSE:
  /* OK, now we can get on with the real code of the function. Recursive calls
  are specified by the macro RMATCH and RRETURN is used to return. When
  NO_RECURSE is *not* defined, these just turn into a recursive call to match()
- and a "return", respectively (possibly with some debugging if DEBUG is
+ and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
  defined). However, RMATCH isn't like a function call because it's quite a
  complicated macro. It has to be used in one particular way. This shouldn't,
  however, impact performance when true recursion is being used. */
-Line 610 
 for (;;)
+Line 664 
 for (;;)
    minimize = possessive = FALSE;
    op = *ecode;
-   /* For partial matching, remember if we ever hit the end of the subject after
-   matching at least one subject character. */
-   if (md->partial &&
-       eptr >= md->end_subject &&
-       eptr > mstart)
-     md->hitend = TRUE;
    switch(op)
      {
      case OP_FAIL:
-Line 667 
 for (;;)
+Line 713 
 for (;;)
      number = GET2(ecode, 1+LINK_SIZE);
      offset = number << 1;
- #ifdef DEBUG
+ #ifdef PCRE_DEBUG
      printf("start bracket %d\n", number);
      printf("subject=");
      pchars(eptr, 16, TRUE, md);
-Line 761 
 for (;;)
+Line 807 
 for (;;)
      case OP_COND:
      case OP_SCOND:
-     if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */
+     codelink= GET(ecode, 1);
+     /* 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)
        {
-       offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/
+       if (pcre_callout != NULL)
-       condition = md->recursive != NULL &&
+         {
-         (offset == RREF_ANY || offset == md->recursive->group_num);
+         pcre_callout_block cb;
-       ecode += condition? 3 : GET(ecode, 1);
+         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];
+       }
+     condcode = ecode[LINK_SIZE+1];
+     /* Now see what the actual condition is */
+     if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
+       {
+       if (md->recursive == NULL)                /* Not recursing => FALSE */
+         {
+         condition = FALSE;
+         ecode += GET(ecode, 1);
+         }
+       else
+         {
+         int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
+         condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
+         /* If the test is for recursion into a specific subpattern, and it is
+         false, but the test was set up by name, scan the table to see if the
+         name refers to any other numbers, and test them. The condition is true
+         if any one is set. */
+         if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
+           {
+           uschar *slotA = md->name_table;
+           for (i = 0; i < md->name_count; i++)
+             {
+             if (GET2(slotA, 0) == recno) break;
+             slotA += md->name_entry_size;
+             }
+           /* Found a name for the number - there can be only one; duplicate
+           names for different numbers are allowed, but not vice versa. First
+           scan down for duplicates. */
+           if (i < md->name_count)
+             {
+             uschar *slotB = slotA;
+             while (slotB > md->name_table)
+               {
+               slotB -= md->name_entry_size;
+               if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
+                 {
+                 condition = GET2(slotB, 0) == md->recursive->group_num;
+                 if (condition) break;
+                 }
+               else break;
+               }
+             /* Scan up for duplicates */
+             if (!condition)
+               {
+               slotB = slotA;
+               for (i++; i < md->name_count; i++)
+                 {
+                 slotB += md->name_entry_size;
+                 if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
+                   {
+                   condition = GET2(slotB, 0) == md->recursive->group_num;
+                   if (condition) break;
+                   }
+                 else break;
+                 }
+               }
+             }
+           }
+         /* Chose branch according to the condition */
+         ecode += condition? 3 : GET(ecode, 1);
+         }
        }
-     else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */
+     else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
        {
        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
        condition = offset < offset_top && md->offset_vector[offset] >= 0;
+       /* If the numbered capture is unset, but the reference was by name,
+       scan the table to see if the name refers to any other numbers, and test
+       them. The condition is true if any one is set. This is tediously similar
+       to the code above, but not close enough to try to amalgamate. */
+       if (!condition && condcode == OP_NCREF)
+         {
+         int refno = offset >> 1;
+         uschar *slotA = md->name_table;
+         for (i = 0; i < md->name_count; i++)
+           {
+           if (GET2(slotA, 0) == refno) break;
+           slotA += md->name_entry_size;
+           }
+         /* Found a name for the number - there can be only one; duplicate names
+         for different numbers are allowed, but not vice versa. First scan down
+         for duplicates. */
+         if (i < md->name_count)
+           {
+           uschar *slotB = slotA;
+           while (slotB > md->name_table)
+             {
+             slotB -= md->name_entry_size;
+             if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
+               {
+               offset = GET2(slotB, 0) << 1;
+               condition = offset < offset_top &&
+                 md->offset_vector[offset] >= 0;
+               if (condition) break;
+               }
+             else break;
+             }
+           /* Scan up for duplicates */
+           if (!condition)
+             {
+             slotB = slotA;
+             for (i++; i < md->name_count; i++)
+               {
+               slotB += md->name_entry_size;
+               if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
+                 {
+                 offset = GET2(slotB, 0) << 1;
+                 condition = offset < offset_top &&
+                   md->offset_vector[offset] >= 0;
+                 if (condition) break;
+                 }
+               else break;
+               }
+             }
+           }
+         }
+       /* Chose branch according to the condition */
        ecode += condition? 3 : GET(ecode, 1);
        }
-     else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */
+     else if (condcode == OP_DEF)     /* DEFINE - always false */
        {
        condition = FALSE;
        ecode += GET(ecode, 1);
-Line 803 
 for (;;)
+Line 1002 
 for (;;)
        else
          {
          condition = FALSE;
-         ecode += GET(ecode, 1);
+         ecode += codelink;
          }
        }
-Line 826 
 for (;;)
+Line 1025 
 for (;;)
          goto TAIL_RECURSE;
          }
        }
-     else                         /* Condition false & no 2nd alternative */
+     else                         /* Condition false & no alternative */
        {
        ecode += 1 + LINK_SIZE;
        }
      break;
+     /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
+     to close any currently open capturing brackets. */
+     case OP_CLOSE:
+     number = GET2(ecode, 1);
+     offset = number << 1;
+ #ifdef PCRE_DEBUG
+       printf("end bracket %d at *ACCEPT", number);
+       printf("\n");
+ #endif
+     md->capture_last = number;
+     if (offset >= md->offset_max) md->offset_overflow = TRUE; else
+       {
+       md->offset_vector[offset] =
+         md->offset_vector[md->offset_end - number];
+       md->offset_vector[offset+1] = eptr - md->start_subject;
+       if (offset_top <= offset) offset_top = offset + 2;
+       }
+     ecode += 3;
+     break;
      /* End of the pattern, either real or forced. If we are in a top-level
      recursion, we should restore the offsets appropriately and continue from
      after the call. */
-Line 846 
 for (;;)
+Line 1069 
 for (;;)
        md->recursive = rec->prevrec;
        memmove(md->offset_vector, rec->offset_save,
          rec->saved_max * sizeof(int));
-       mstart = rec->save_start;
+       offset_top = rec->save_offset_top;
        ims = original_ims;
        ecode = rec->after_call;
        break;
        }
-     /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty
+     /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is
-     string - backtracking will then try other alternatives, if any. */
+     set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of
+     the subject. In both cases, backtracking will then try other alternatives,
+     if any. */
+     if (eptr == mstart &&
+         (md->notempty ||
+           (md->notempty_atstart &&
+             mstart == md->start_subject + md->start_offset)))
+       RRETURN(MATCH_NOMATCH);
+     /* Otherwise, we have a match. */
-     if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH);
      md->end_match_ptr = eptr;           /* Record where we ended */
      md->end_offset_top = offset_top;    /* and how many extracts were taken */
      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
-Line 881 
 for (;;)
+Line 1113 
 for (;;)
        {
        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
          RM4);
-       if (rrc == MATCH_MATCH) break;
+       if (rrc == MATCH_MATCH)
+         {
+         mstart = md->start_match_ptr;   /* In case \K reset it */
+         break;
+         }
        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
        ecode += GET(ecode, 1);
        }
-Line 900 
 for (;;)
+Line 1136 
 for (;;)
      offset_top = md->end_offset_top;
      continue;
-     /* Negative assertion: all branches must fail to match */
+     /* Negative assertion: all branches must fail to match. Encountering SKIP,
+     PRUNE, or COMMIT means we must assume failure without checking subsequent
+     branches. */
      case OP_ASSERT_NOT:
      case OP_ASSERTBACK_NOT:
-Line 909 
 for (;;)
+Line 1147 
 for (;;)
        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
          RM5);
        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);
+       if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
+         {
+         do ecode += GET(ecode,1); while (*ecode == OP_ALT);
+         break;
+         }
        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
        ecode += GET(ecode,1);
        }
-Line 946 
 for (;;)
+Line 1189 
 for (;;)
        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
        }
-     /* Skip to next op code */
+     /* Save the earliest consulted character, then skip to next op code */
+     if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
      ecode += 1 + LINK_SIZE;
      break;
-Line 1026 
 for (;;)
+Line 1270 
 for (;;)
        memcpy(new_recursive.offset_save, md->offset_vector,
              new_recursive.saved_max * sizeof(int));
-       new_recursive.save_start = mstart;
+       new_recursive.save_offset_top = offset_top;
-       mstart = eptr;
        /* OK, now we can do the recursion. For each top-level alternative we
        restore the offset and recursion data. */
-Line 1049 
 for (;;)
+Line 1292 
 for (;;)
          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
            {
            DPRINTF(("Recursion gave error %d\n", rrc));
+           if (new_recursive.offset_save != stacksave)
+             (pcre_free)(new_recursive.offset_save);
            RRETURN(rrc);
            }
-Line 1072 
 for (;;)
+Line 1317 
 for (;;)
      a move back into the brackets. Friedl calls these "atomic" subpatterns.
      Check the alternative branches in turn - the matching won't pass the KET
      for this kind of subpattern. If any one branch matches, we carry on as at
-     the end of a normal bracket, leaving the subject pointer. */
+     the end of a normal bracket, leaving the subject pointer, but resetting
+     the start-of-match value in case it was changed by \K. */
      case OP_ONCE:
      prev = ecode;
-Line 1081 
 for (;;)
+Line 1327 
 for (;;)
      do
        {
        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);
-       if (rrc == MATCH_MATCH) break;
+       if (rrc == MATCH_MATCH)
+         {
+         mstart = md->start_match_ptr;
+         break;
+         }
        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
        ecode += GET(ecode,1);
        }
-Line 1148 
 for (;;)
+Line 1398 
 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 1174 
 for (;;)
+Line 1424 
 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 1192 
 for (;;)
+Line 1450 
 for (;;)
        }
      else saved_eptr = NULL;
-     /* If we are at the end of an assertion group, stop matching and return
+     /* If we are at the end of an assertion group or an atomic group, stop
-     MATCH_MATCH, but record the current high water mark for use by positive
+     matching and return MATCH_MATCH, but record the current high water mark for
-     assertions. Do this also for the "once" (atomic) groups. */
+     use by positive assertions. We also need to record the match start in case
+     it was changed by \K. */
      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||
-Line 1202 
 for (;;)
+Line 1461 
 for (;;)
        {
        md->end_match_ptr = eptr;      /* For ONCE */
        md->end_offset_top = offset_top;
+       md->start_match_ptr = mstart;
        RRETURN(MATCH_MATCH);
        }
-Line 1216 
 for (;;)
+Line 1476 
 for (;;)
        number = GET2(prev, 1+LINK_SIZE);
        offset = number << 1;
- #ifdef DEBUG
+ #ifdef PCRE_DEBUG
        printf("end bracket %d", number);
        printf("\n");
  #endif
-Line 1238 
 for (;;)
+Line 1498 
 for (;;)
          recursion_info *rec = md->recursive;
          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
          md->recursive = rec->prevrec;
-         mstart = rec->save_start;
          memcpy(md->offset_vector, rec->offset_save,
            rec->saved_max * sizeof(int));
+         offset_top = rec->save_offset_top;
          ecode = rec->after_call;
          ims = original_ims;
          break;
-Line 1380 
 for (;;)
+Line 1640 
 for (;;)
        /* Find out if the previous and current characters are "word" characters.
        It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to
-       be "non-word" characters. */
+       be "non-word" characters. Remember the earliest consulted character for
+       partial matching. */
  #ifdef SUPPORT_UTF8
        if (utf8)
          {
          if (eptr == md->start_subject) prev_is_word = FALSE; else
            {
-           const uschar *lastptr = eptr - 1;
+           USPTR lastptr = eptr - 1;
            while((*lastptr & 0xc0) == 0x80) lastptr--;
+           if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
            GETCHAR(c, lastptr);
            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
            }
-         if (eptr >= md->end_subject) cur_is_word = FALSE; else
+         if (eptr >= md->end_subject)
+           {
+           SCHECK_PARTIAL();
+           cur_is_word = FALSE;
+           }
+         else
            {
            GETCHAR(c, eptr);
            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
-Line 1401 
 for (;;)
+Line 1668 
 for (;;)
        else
  #endif
-       /* More streamlined when not in UTF-8 mode */
+       /* Not in UTF-8 mode */
          {
-         prev_is_word = (eptr != md->start_subject) &&
+         if (eptr == md->start_subject) prev_is_word = FALSE; else
-           ((md->ctypes[eptr[-1]] & ctype_word) != 0);
+           {
-         cur_is_word = (eptr < md->end_subject) &&
+           if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
-           ((md->ctypes[*eptr] & ctype_word) != 0);
+           prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
+           }
+         if (eptr >= md->end_subject)
+           {
+           SCHECK_PARTIAL();
+           cur_is_word = FALSE;
+           }
+         else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
          }
        /* Now see if the situation is what we want */
-Line 1421 
 for (;;)
+Line 1695 
 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 */
+     case OP_ALLANY:
+     if (eptr++ >= md->end_subject)
        {
-       if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
        }
-     if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
-     if (utf8)
-       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
      ecode++;
      break;
-Line 1435 
 for (;;)
+Line 1712 
 for (;;)
      any byte, even newline, independent of the setting of PCRE_DOTALL. */
      case OP_ANYBYTE:
-     if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr++ >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      ecode++;
      break;
      case OP_NOT_DIGIT:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      if (
  #ifdef SUPPORT_UTF8
-Line 1453 
 for (;;)
+Line 1738 
 for (;;)
      break;
      case OP_DIGIT:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      if (
  #ifdef SUPPORT_UTF8
-Line 1466 
 for (;;)
+Line 1755 
 for (;;)
      break;
      case OP_NOT_WHITESPACE:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      if (
  #ifdef SUPPORT_UTF8
-Line 1479 
 for (;;)
+Line 1772 
 for (;;)
      break;
      case OP_WHITESPACE:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      if (
  #ifdef SUPPORT_UTF8
-Line 1492 
 for (;;)
+Line 1789 
 for (;;)
      break;
      case OP_NOT_WORDCHAR:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      if (
  #ifdef SUPPORT_UTF8
-Line 1505 
 for (;;)
+Line 1806 
 for (;;)
      break;
      case OP_WORDCHAR:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      if (
  #ifdef SUPPORT_UTF8
-Line 1518 
 for (;;)
+Line 1823 
 for (;;)
      break;
      case OP_ANYNL:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      switch(c)
        {
-Line 1526 
 for (;;)
+Line 1835 
 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);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      switch(c)
        {
-Line 1568 
 for (;;)
+Line 1885 
 for (;;)
      break;
      case OP_HSPACE:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      switch(c)
        {
-Line 1598 
 for (;;)
+Line 1919 
 for (;;)
      break;
      case OP_NOT_VSPACE:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      switch(c)
        {
-Line 1616 
 for (;;)
+Line 1941 
 for (;;)
      break;
      case OP_VSPACE:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      GETCHARINCTEST(c, eptr);
      switch(c)
        {
-Line 1639 
 for (;;)
+Line 1968 
 for (;;)
      case OP_PROP:
      case OP_NOTPROP:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       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 1652 
 for (;;)
+Line 1984 
 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 1685 
 for (;;)
+Line 2017 
 for (;;)
      is in the binary; otherwise a compile-time error occurs. */
      case OP_EXTUNI:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       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 1698 
 for (;;)
+Line 2033 
 for (;;)
            {
            GETCHARLEN(c, eptr, len);
            }
-         category = _pcre_ucp_findprop(c, &chartype, &script);
+         category = UCD_CATEGORY(c);
          if (category != ucp_M) break;
          eptr += len;
          }
-Line 1719 
 for (;;)
+Line 2054 
 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 1757 
 for (;;)
+Line 2101 
 for (;;)
          break;
          default:               /* No repeat follows */
-         if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);
+         if (!match_ref(offset, eptr, length, md, ims))
+           {
+           CHECK_PARTIAL();
+           RRETURN(MATCH_NOMATCH);
+           }
          eptr += length;
          continue;              /* With the main loop */
          }
-Line 1773 
 for (;;)
+Line 2121 
 for (;;)
        for (i = 1; i <= min; i++)
          {
-         if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);
+         if (!match_ref(offset, eptr, length, md, ims))
+           {
+           CHECK_PARTIAL();
+           RRETURN(MATCH_NOMATCH);
+           }
          eptr += length;
          }
-Line 1790 
 for (;;)
+Line 2142 
 for (;;)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-           if (fi >= max || !match_ref(offset, eptr, length, md, ims))
+           if (fi >= max) RRETURN(MATCH_NOMATCH);
+           if (!match_ref(offset, eptr, length, md, ims))
+             {
+             CHECK_PARTIAL();
              RRETURN(MATCH_NOMATCH);
+             }
            eptr += length;
            }
          /* Control never gets here */
-Line 1804 
 for (;;)
+Line 2160 
 for (;;)
          pp = eptr;
          for (i = min; i < max; i++)
            {
-           if (!match_ref(offset, eptr, length, md, ims)) break;
+           if (!match_ref(offset, eptr, length, md, ims))
+             {
+             CHECK_PARTIAL();
+             break;
+             }
            eptr += length;
            }
          while (eptr >= pp)
-Line 1818 
 for (;;)
+Line 2178 
 for (;;)
        }
      /* Control never gets here */
      /* Match a bit-mapped character class, possibly repeatedly. This op code is
      used when all the characters in the class have values in the range 0-255,
      and either the matching is caseful, or the characters are in the range
-Line 1874 
 for (;;)
+Line 2232 
 for (;;)
          {
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(c, eptr);
            if (c > 255)
              {
-Line 1892 
 for (;;)
+Line 2254 
 for (;;)
          {
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            c = *eptr++;
            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
            }
-Line 1916 
 for (;;)
+Line 2282 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               RRETURN(MATCH_NOMATCH);
+               }
              GETCHARINC(c, eptr);
              if (c > 255)
                {
-Line 1936 
 for (;;)
+Line 2307 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               RRETURN(MATCH_NOMATCH);
+               }
              c = *eptr++;
              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
              }
-Line 1957 
 for (;;)
+Line 2333 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (c > 255)
                {
-Line 1983 
 for (;;)
+Line 2363 
 for (;;)
            {
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              c = *eptr;
              if ((data[c/8] & (1 << (c&7))) == 0) break;
              eptr++;
-Line 2003 
 for (;;)
+Line 2387 
 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 2044 
 for (;;)
+Line 2429 
 for (;;)
        for (i = 1; i <= min; i++)
          {
-         if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+         if (eptr >= md->end_subject)
-         GETCHARINC(c, eptr);
+           {
+           SCHECK_PARTIAL();
+           RRETURN(MATCH_NOMATCH);
+           }
+         GETCHARINCTEST(c, eptr);
          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);
          }
-Line 2063 
 for (;;)
+Line 2452 
 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);
+           if (fi >= max) RRETURN(MATCH_NOMATCH);
-           GETCHARINC(c, eptr);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
+           GETCHARINCTEST(c, eptr);
            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);
            }
          /* Control never gets here */
-Line 2078 
 for (;;)
+Line 2472 
 for (;;)
          for (i = min; i < max; i++)
            {
            int len = 1;
-           if (eptr >= md->end_subject) break;
+           if (eptr >= md->end_subject)
-           GETCHARLEN(c, eptr, len);
+             {
+             SCHECK_PARTIAL();
+             break;
+             }
+           GETCHARLENTEST(c, eptr, len);
            if (!_pcre_xclass(c, data)) break;
            eptr += len;
            }
-Line 2106 
 for (;;)
+Line 2504 
 for (;;)
        length = 1;
        ecode++;
        GETCHARLEN(fc, ecode, length);
-       if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);
+       if (length > md->end_subject - eptr)
+         {
+         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
+         RRETURN(MATCH_NOMATCH);
+         }
        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);
        }
      else
-Line 2114 
 for (;;)
+Line 2516 
 for (;;)
      /* Non-UTF-8 mode */
        {
-       if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);
+       if (md->end_subject - eptr < 1)
+         {
+         SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
+         RRETURN(MATCH_NOMATCH);
+         }
        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
        ecode += 2;
        }
-Line 2130 
 for (;;)
+Line 2536 
 for (;;)
        ecode++;
        GETCHARLEN(fc, ecode, length);
-       if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);
+       if (length > md->end_subject - eptr)
+         {
+         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
+         RRETURN(MATCH_NOMATCH);
+         }
        /* If the pattern character's value is < 128, we have only one byte, and
        can use the fast lookup table. */
-Line 2154 
 for (;;)
+Line 2564 
 for (;;)
          if (fc != dc)
            {
  #ifdef SUPPORT_UCP
-           if (dc != _pcre_ucp_othercase(fc))
+           if (dc != UCD_OTHERCASE(fc))
  #endif
              RRETURN(MATCH_NOMATCH);
            }
-Line 2165 
 for (;;)
+Line 2575 
 for (;;)
      /* Non-UTF-8 mode */
        {
-       if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);
+       if (md->end_subject - eptr < 1)
+         {
+         SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
+         RRETURN(MATCH_NOMATCH);
+         }
        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
        ecode += 2;
        }
-Line 2219 
 for (;;)
+Line 2633 
 for (;;)
      case OP_MINQUERY:
      c = *ecode++ - OP_STAR;
      minimize = (c & 1) != 0;
      min = rep_min[c];                 /* Pick up values from tables; */
      max = rep_max[c];                 /* zero for max => infinity */
      if (max == 0) max = INT_MAX;
-     /* Common code for all repeated single-character matches. We can give
+     /* Common code for all repeated single-character matches. */
-     up quickly if there are fewer than the minimum number of characters left in
-     the subject. */
      REPEATCHAR:
  #ifdef SUPPORT_UTF8
-Line 2234 
 for (;;)
+Line 2647 
 for (;;)
        length = 1;
        charptr = ecode;
        GETCHARLEN(fc, ecode, length);
-       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);
        ecode += length;
        /* Handle multibyte character matching specially here. There is
-Line 2245 
 for (;;)
+Line 2657 
 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 */
          for (i = 1; i <= min; i++)
            {
-           if (memcmp(eptr, charptr, length) == 0) eptr += length;
+           if (eptr <= md->end_subject - length &&
+             memcmp(eptr, charptr, length) == 0) eptr += length;
  #ifdef SUPPORT_UCP
-           /* Need braces because of following else */
+           else if (oclength > 0 &&
-           else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }
+                    eptr <= md->end_subject - oclength &&
+                    memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
+ #endif  /* SUPPORT_UCP */
            else
              {
-             if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);
+             CHECK_PARTIAL();
-             eptr += oclength;
+             RRETURN(MATCH_NOMATCH);
              }
- #else   /* without SUPPORT_UCP */
-           else { RRETURN(MATCH_NOMATCH); }
- #endif  /* SUPPORT_UCP */
            }
          if (min == max) continue;
-Line 2274 
 for (;;)
+Line 2686 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
-             if (memcmp(eptr, charptr, length) == 0) eptr += length;
+             if (eptr <= md->end_subject - length &&
+               memcmp(eptr, charptr, length) == 0) eptr += length;
  #ifdef SUPPORT_UCP
-             /* Need braces because of following else */
+             else if (oclength > 0 &&
-             else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }
+                      eptr <= md->end_subject - oclength &&
+                      memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
+ #endif  /* SUPPORT_UCP */
              else
                {
-               if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);
+               CHECK_PARTIAL();
-               eptr += oclength;
+               RRETURN(MATCH_NOMATCH);
                }
- #else   /* without SUPPORT_UCP */
-             else { RRETURN (MATCH_NOMATCH); }
- #endif  /* SUPPORT_UCP */
              }
            /* Control never gets here */
            }
-Line 2296 
 for (;;)
+Line 2708 
 for (;;)
            pp = eptr;
            for (i = min; i < max; i++)
              {
-             if (eptr > md->end_subject - length) break;
+             if (eptr <= md->end_subject - length &&
-             if (memcmp(eptr, charptr, length) == 0) eptr += length;
+                 memcmp(eptr, charptr, length) == 0) eptr += length;
  #ifdef SUPPORT_UCP
-             else if (oclength == 0) break;
+             else if (oclength > 0 &&
+                      eptr <= md->end_subject - oclength &&
+                      memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
+ #endif  /* SUPPORT_UCP */
              else
                {
-               if (memcmp(eptr, occhars, oclength) != 0) break;
+               CHECK_PARTIAL();
-               eptr += oclength;
+               break;
                }
- #else   /* without SUPPORT_UCP */
-             else break;
- #endif  /* SUPPORT_UCP */
              }
            if (possessive) continue;
            for(;;)
-            {
+             {
-            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);
+             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);
-            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+             if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-            if (eptr == pp) RRETURN(MATCH_NOMATCH);
+             if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
  #ifdef SUPPORT_UCP
-            eptr--;
+             eptr--;
-            BACKCHAR(eptr);
+             BACKCHAR(eptr);
  #else   /* without SUPPORT_UCP */
-            eptr -= length;
+             eptr -= length;
  #endif  /* SUPPORT_UCP */
-            }
+             }
            }
          /* Control never gets here */
          }
-Line 2335 
 for (;;)
+Line 2748 
 for (;;)
  #endif  /* SUPPORT_UTF8 */
      /* When not in UTF-8 mode, load a single-byte character. */
-       {
-       if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);
+     fc = *ecode++;
-       fc = *ecode++;
-       }
      /* The value of fc at this point is always less than 256, though we may or
      may not be in UTF-8 mode. The code is duplicated for the caseless and
-Line 2356 
 for (;;)
+Line 2767 
 for (;;)
        {
        fc = md->lcc[fc];
        for (i = 1; i <= min; i++)
+         {
+         if (eptr >= md->end_subject)
+           {
+           SCHECK_PARTIAL();
+           RRETURN(MATCH_NOMATCH);
+           }
          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
+         }
        if (min == max) continue;
        if (minimize)
          {
-Line 2364 
 for (;;)
+Line 2782 
 for (;;)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-           if (fi >= max || eptr >= md->end_subject ||
+           if (fi >= max) RRETURN(MATCH_NOMATCH);
-               fc != md->lcc[*eptr++])
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
              RRETURN(MATCH_NOMATCH);
+             }
+           if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
            }
          /* Control never gets here */
          }
-Line 2375 
 for (;;)
+Line 2797 
 for (;;)
          pp = eptr;
          for (i = min; i < max; i++)
            {
-           if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             break;
+             }
+           if (fc != md->lcc[*eptr]) break;
            eptr++;
            }
          if (possessive) continue;
          while (eptr >= pp)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);
-Line 2394 
 for (;;)
+Line 2823 
 for (;;)
      else
        {
-       for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
+       for (i = 1; i <= min; i++)
+         {
+         if (eptr >= md->end_subject)
+           {
+           SCHECK_PARTIAL();
+           RRETURN(MATCH_NOMATCH);
+           }
+         if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
+         }
        if (min == max) continue;
        if (minimize)
          {
          for (fi = min;; fi++)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-           if (fi >= max || eptr >= md->end_subject || fc != *eptr++)
+           if (fi >= max) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
              RRETURN(MATCH_NOMATCH);
+             }
+           if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
            }
          /* Control never gets here */
          }
-Line 2412 
 for (;;)
+Line 2856 
 for (;;)
          pp = eptr;
          for (i = min; i < max; i++)
            {
-           if (eptr >= md->end_subject || fc != *eptr) break;
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             break;
+             }
+           if (fc != *eptr) break;
            eptr++;
            }
          if (possessive) continue;
          while (eptr >= pp)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);
-Line 2431 
 for (;;)
+Line 2881 
 for (;;)
      checking can be multibyte. */
      case OP_NOT:
-     if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+     if (eptr >= md->end_subject)
+       {
+       SCHECK_PARTIAL();
+       RRETURN(MATCH_NOMATCH);
+       }
      ecode++;
      GETCHARINCTEST(c, eptr);
      if ((ims & PCRE_CASELESS) != 0)
-Line 2508 
 for (;;)
+Line 2962 
 for (;;)
      max = rep_max[c];                 /* zero for max => infinity */
      if (max == 0) max = INT_MAX;
-     /* Common code for all repeated single-byte matches. We can give up quickly
+     /* Common code for all repeated single-byte matches. */
-     if there are fewer than the minimum number of bytes left in the
-     subject. */
      REPEATNOTCHAR:
-     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);
      fc = *ecode++;
      /* The code is duplicated for the caseless and caseful cases, for speed,
-Line 2538 
 for (;;)
+Line 2989 
 for (;;)
          register unsigned int d;
          for (i = 1; i <= min; i++)
            {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(d, eptr);
            if (d < 256) d = md->lcc[d];
            if (fc == d) RRETURN(MATCH_NOMATCH);
-Line 2549 
 for (;;)
+Line 3005 
 for (;;)
        /* Not UTF-8 mode */
          {
          for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
+           }
          }
        if (min == max) continue;
-Line 2565 
 for (;;)
+Line 3028 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 2579 
 for (;;)
+Line 3047 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                RRETURN(MATCH_NOMATCH);
+               }
+             if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
              }
            }
          /* Control never gets here */
-Line 2600 
 for (;;)
+Line 3073 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(d, eptr, len);
              if (d < 256) d = md->lcc[d];
              if (fc == d) break;
-Line 2621 
 for (;;)
+Line 3098 
 for (;;)
            {
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
+             if (fc == md->lcc[*eptr]) break;
              eptr++;
              }
            if (possessive) continue;
-Line 2649 
 for (;;)
+Line 3131 
 for (;;)
          register unsigned int d;
          for (i = 1; i <= min; i++)
            {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(d, eptr);
            if (fc == d) RRETURN(MATCH_NOMATCH);
            }
-Line 2658 
 for (;;)
+Line 3145 
 for (;;)
        /* Not UTF-8 mode */
          {
          for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);
+           }
          }
        if (min == max) continue;
-Line 2674 
 for (;;)
+Line 3168 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             GETCHARINC(d, eptr);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
-             if (fi >= max || eptr >= md->end_subject || fc == d)
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                RRETURN(MATCH_NOMATCH);
+               }
+             GETCHARINC(d, eptr);
+             if (fc == d) RRETURN(MATCH_NOMATCH);
              }
            }
          else
-Line 2687 
 for (;;)
+Line 3186 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject || fc == *eptr++)
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                RRETURN(MATCH_NOMATCH);
+               }
+             if (fc == *eptr++) RRETURN(MATCH_NOMATCH);
              }
            }
          /* Control never gets here */
-Line 2708 
 for (;;)
+Line 3212 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(d, eptr, len);
              if (fc == d) break;
              eptr += len;
-Line 2728 
 for (;;)
+Line 3236 
 for (;;)
            {
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject || fc == *eptr) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
+             if (fc == *eptr) break;
              eptr++;
              }
            if (possessive) continue;
-Line 2822 
 for (;;)
+Line 3335 
 for (;;)
      /* First, ensure the minimum number of matches are present. Use inline
      code for maximizing the speed, and do the type test once at the start
-     (i.e. keep it out of the loop). Also we can test that there are at least
+     (i.e. keep it out of the loop). Separate the UTF-8 code completely as that
-     the minimum number of bytes before we start. This isn't as effective in
-     UTF-8 mode, but it does no harm. Separate the UTF-8 code completely as that
      is tidier. Also separate the UCP code, which can be the same for both UTF-8
      and single-bytes. */
-     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);
      if (min > 0)
        {
  #ifdef SUPPORT_UCP
-Line 2840 
 for (;;)
+Line 3350 
 for (;;)
            if (prop_fail_result) RRETURN(MATCH_NOMATCH);
            for (i = 1; i <= min; i++)
              {
-             if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               RRETURN(MATCH_NOMATCH);
+               }
              GETCHARINCTEST(c, eptr);
              }
            break;
-Line 2848 
 for (;;)
+Line 3362 
 for (;;)
            case PT_LAMP:
            for (i = 1; i <= min; i++)
              {
-             if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               RRETURN(MATCH_NOMATCH);
+               }
              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 2861 
 for (;;)
+Line 3379 
 for (;;)
            case PT_GC:
            for (i = 1; i <= min; i++)
              {
-             if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               RRETURN(MATCH_NOMATCH);
+               }
              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 2872 
 for (;;)
+Line 3394 
 for (;;)
            case PT_PC:
            for (i = 1; i <= min; i++)
              {
-             if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               RRETURN(MATCH_NOMATCH);
+               }
              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 2883 
 for (;;)
+Line 3409 
 for (;;)
            case PT_SC:
            for (i = 1; i <= min; i++)
              {
-             if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               RRETURN(MATCH_NOMATCH);
+               }
              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 2903 
 for (;;)
+Line 3433 
 for (;;)
          {
          for (i = 1; i <= min; i++)
            {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             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)
              {
              int len = 1;
-             if (!utf8) c = *eptr; else
+             if (!utf8) c = *eptr;
-               {
+               else { GETCHARLEN(c, eptr, len); }
-               GETCHARLEN(c, eptr, len);
+             prop_category = UCD_CATEGORY(c);
-               }
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
              if (prop_category != ucp_M) break;
              eptr += len;
              }
-Line 2931 
 for (;;)
+Line 3464 
 for (;;)
          case OP_ANY:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject ||
+           if (eptr >= md->end_subject)
-                ((ims & PCRE_DOTALL) == 0 && IS_NEWLINE(eptr)))
+             {
+             SCHECK_PARTIAL();
              RRETURN(MATCH_NOMATCH);
+             }
+           if (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)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            eptr++;
            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
            }
          break;
          case OP_ANYBYTE:
+         if (eptr > md->end_subject - min) RRETURN(MATCH_NOMATCH);
          eptr += min;
          break;
          case OP_ANYNL:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(c, eptr);
            switch(c)
              {
-Line 2954 
 for (;;)
+Line 3508 
 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;
              }
            }
-Line 2968 
 for (;;)
+Line 3526 
 for (;;)
          case OP_NOT_HSPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(c, eptr);
            switch(c)
              {
-Line 3000 
 for (;;)
+Line 3562 
 for (;;)
          case OP_HSPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(c, eptr);
            switch(c)
              {
-Line 3032 
 for (;;)
+Line 3598 
 for (;;)
          case OP_NOT_VSPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(c, eptr);
            switch(c)
              {
-Line 3052 
 for (;;)
+Line 3622 
 for (;;)
          case OP_VSPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(c, eptr);
            switch(c)
              {
-Line 3072 
 for (;;)
+Line 3646 
 for (;;)
          case OP_NOT_DIGIT:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            GETCHARINC(c, eptr);
            if (c < 128 && (md->ctypes[c] & ctype_digit) != 0)
              RRETURN(MATCH_NOMATCH);
-Line 3082 
 for (;;)
+Line 3660 
 for (;;)
          case OP_DIGIT:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject ||
+           if (eptr >= md->end_subject)
-              *eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
+           if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0)
              RRETURN(MATCH_NOMATCH);
            /* No need to skip more bytes - we know it's a 1-byte character */
            }
-Line 3092 
 for (;;)
+Line 3674 
 for (;;)
          case OP_NOT_WHITESPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject ||
+           if (eptr >= md->end_subject)
-              (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0))
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
+           if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0)
              RRETURN(MATCH_NOMATCH);
            while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
            }
-Line 3102 
 for (;;)
+Line 3688 
 for (;;)
          case OP_WHITESPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject ||
+           if (eptr >= md->end_subject)
-              *eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
+           if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0)
              RRETURN(MATCH_NOMATCH);
            /* No need to skip more bytes - we know it's a 1-byte character */
            }
-Line 3112 
 for (;;)
+Line 3702 
 for (;;)
          case OP_NOT_WORDCHAR:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject ||
+           if (eptr >= md->end_subject)
-              (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0))
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
+           if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0)
              RRETURN(MATCH_NOMATCH);
            while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
            }
-Line 3122 
 for (;;)
+Line 3716 
 for (;;)
          case OP_WORDCHAR:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject ||
+           if (eptr >= md->end_subject)
-              *eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
+           if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0)
              RRETURN(MATCH_NOMATCH);
            /* No need to skip more bytes - we know it's a 1-byte character */
            }
-Line 3137 
 for (;;)
+Line 3735 
 for (;;)
  #endif     /* SUPPORT_UTF8 */
        /* Code for the non-UTF-8 case for minimum matching of operators other
-       than OP_PROP and OP_NOTPROP. We can assume that there are the minimum
+       than OP_PROP and OP_NOTPROP. */
-       number of bytes present, as this was tested above. */
        switch(ctype)
          {
          case OP_ANY:
-         if ((ims & PCRE_DOTALL) == 0)
+         for (i = 1; i <= min; i++)
            {
-           for (i = 1; i <= min; i++)
+           if (eptr >= md->end_subject)
              {
-             if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
+             SCHECK_PARTIAL();
-             eptr++;
+             RRETURN(MATCH_NOMATCH);
              }
+           if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
+           eptr++;
            }
-         else eptr += min;
          break;
-         case OP_ANYBYTE:
+         case OP_ALLANY:
+         if (eptr > md->end_subject - min)
+           {
+           SCHECK_PARTIAL();
+           RRETURN(MATCH_NOMATCH);
+           }
          eptr += min;
          break;
-         /* Because of the CRLF case, we can't assume the minimum number of
+         case OP_ANYBYTE:
-         bytes are present in this case. */
+         if (eptr > md->end_subject - min)
+           {
+           SCHECK_PARTIAL();
+           RRETURN(MATCH_NOMATCH);
+           }
+         eptr += min;
+         break;
          case OP_ANYNL:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            switch(*eptr++)
              {
              default: RRETURN(MATCH_NOMATCH);
-Line 3172 
 for (;;)
+Line 3785 
 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;
              }
            }
-Line 3183 
 for (;;)
+Line 3799 
 for (;;)
          case OP_NOT_HSPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            switch(*eptr++)
              {
              default: break;
-Line 3198 
 for (;;)
+Line 3818 
 for (;;)
          case OP_HSPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            switch(*eptr++)
              {
              default: RRETURN(MATCH_NOMATCH);
-Line 3213 
 for (;;)
+Line 3837 
 for (;;)
          case OP_NOT_VSPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            switch(*eptr++)
              {
              default: break;
-Line 3230 
 for (;;)
+Line 3858 
 for (;;)
          case OP_VSPACE:
          for (i = 1; i <= min; i++)
            {
-           if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            switch(*eptr++)
              {
              default: RRETURN(MATCH_NOMATCH);
-Line 3246 
 for (;;)
+Line 3878 
 for (;;)
          case OP_NOT_DIGIT:
          for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH);
+           }
          break;
          case OP_DIGIT:
          for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH);
+           }
          break;
          case OP_NOT_WHITESPACE:
          for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            if ((md->ctypes[*eptr++] & ctype_space) != 0) RRETURN(MATCH_NOMATCH);
+           }
          break;
          case OP_WHITESPACE:
          for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            if ((md->ctypes[*eptr++] & ctype_space) == 0) RRETURN(MATCH_NOMATCH);
+           }
          break;
          case OP_NOT_WORDCHAR:
          for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            if ((md->ctypes[*eptr++] & ctype_word) != 0)
              RRETURN(MATCH_NOMATCH);
+           }
          break;
          case OP_WORDCHAR:
          for (i = 1; i <= min; i++)
+           {
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
            if ((md->ctypes[*eptr++] & ctype_word) == 0)
              RRETURN(MATCH_NOMATCH);
+           }
          break;
          default:
-Line 3301 
 for (;;)
+Line 3975 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               RRETURN(MATCH_NOMATCH);
+               }
              GETCHARINC(c, eptr);
              if (prop_fail_result) RRETURN(MATCH_NOMATCH);
              }
-Line 3312 
 for (;;)
+Line 3991 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 3327 
 for (;;)
+Line 4011 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 3340 
 for (;;)
+Line 4029 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 3353 
 for (;;)
+Line 4047 
 for (;;)
              {
              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);
              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-             if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+             if (fi >= max) RRETURN(MATCH_NOMATCH);
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 3375 
 for (;;)
+Line 4074 
 for (;;)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-           if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
+           if (fi >= max) RRETURN(MATCH_NOMATCH);
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             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)
              {
              int len = 1;
-             if (!utf8) c = *eptr; else
+             if (!utf8) c = *eptr;
-               {
+               else { GETCHARLEN(c, eptr, len); }
-               GETCHARLEN(c, eptr, len);
+             prop_category = UCD_CATEGORY(c);
-               }
-             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
              if (prop_category != ucp_M) break;
              eptr += len;
              }
-Line 3404 
 for (;;)
+Line 4106 
 for (;;)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-           if (fi >= max || eptr >= md->end_subject ||
+           if (fi >= max) RRETURN(MATCH_NOMATCH);
-                (ctype == OP_ANY && (ims & PCRE_DOTALL) == 0 &&
+           if (eptr >= md->end_subject)
-                 IS_NEWLINE(eptr)))
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
+           if (ctype == OP_ANY && 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 3426 
 for (;;)
+Line 4130 
 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;
-Line 3562 
 for (;;)
+Line 4269 
 for (;;)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);
            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
-           if (fi >= max || eptr >= md->end_subject ||
+           if (fi >= max) RRETURN(MATCH_NOMATCH);
-                ((ims & PCRE_DOTALL) == 0 && IS_NEWLINE(eptr)))
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             RRETURN(MATCH_NOMATCH);
+             }
+           if (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 3582 
 for (;;)
+Line 4292 
 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;
-Line 3687 
 for (;;)
+Line 4401 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (prop_fail_result) break;
              eptr+= len;
-Line 3698 
 for (;;)
+Line 4416 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 3713 
 for (;;)
+Line 4435 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 3726 
 for (;;)
+Line 4452 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 3739 
 for (;;)
+Line 4469 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               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 3768 
 for (;;)
+Line 4502 
 for (;;)
          {
          for (i = min; i < max; i++)
            {
-           if (eptr >= md->end_subject) break;
+           if (eptr >= md->end_subject)
+             {
+             SCHECK_PARTIAL();
+             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 3779 
 for (;;)
+Line 4517 
 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 3788 
 for (;;)
+Line 4526 
 for (;;)
          /* eptr is now past the end of the maximum run */
          if (possessive) continue;
          for(;;)
            {
            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);
-Line 3801 
 for (;;)
+Line 4540 
 for (;;)
                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 3821 
 for (;;)
+Line 4560 
 for (;;)
            case OP_ANY:
            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)
                  {
-                 if (eptr >= md->end_subject) break;
+                 SCHECK_PARTIAL();
-                 eptr++;
+                 break;
-                 while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                  }
+               if (IS_NEWLINE(eptr)) break;
+               eptr++;
+               while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                }
              }
-Line 3845 
 for (;;)
+Line 4577 
 for (;;)
            else
              {
-             if ((ims & PCRE_DOTALL) == 0)
+             for (i = min; i < max; i++)
                {
-               for (i = min; i < max; i++)
+               if (eptr >= md->end_subject)
                  {
-                 if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
+                 SCHECK_PARTIAL();
-                 eptr++;
+                 break;
-                 while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                  }
+               if (IS_NEWLINE(eptr)) break;
+               eptr++;
+               while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                }
-             else
+             }
+           break;
+           case OP_ALLANY:
+           if (max < INT_MAX)
+             {
+             for (i = min; i < max; i++)
                {
-               eptr = md->end_subject;
+               if (eptr >= md->end_subject)
+                 {
+                 SCHECK_PARTIAL();
+                 break;
+                 }
+               eptr++;
+               while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                }
              }
+           else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
            break;
            /* The byte case is the same as non-UTF8 */
-Line 3866 
 for (;;)
+Line 4613 
 for (;;)
            case OP_ANYBYTE:
            c = max - min;
            if (c > (unsigned int)(md->end_subject - eptr))
-             c = md->end_subject - eptr;
+             {
-           eptr += c;
+             eptr = md->end_subject;
+             SCHECK_PARTIAL();
+             }
+           else eptr += c;
            break;
            case OP_ANYNL:
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (c == 0x000d)
                {
-Line 3883 
 for (;;)
+Line 4637 
 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;
                }
-Line 3897 
 for (;;)
+Line 4653 
 for (;;)
              {
              BOOL gotspace;
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              switch(c)
                {
-Line 3935 
 for (;;)
+Line 4695 
 for (;;)
              {
              BOOL gotspace;
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              switch(c)
                {
-Line 3959 
 for (;;)
+Line 4723 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) break;
              eptr+= len;
-Line 3970 
 for (;;)
+Line 4738 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (c >= 256 ||(md->ctypes[c] & ctype_digit) == 0) break;
              eptr+= len;
-Line 3981 
 for (;;)
+Line 4753 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (c < 256 && (md->ctypes[c] & ctype_space) != 0) break;
              eptr+= len;
-Line 3992 
 for (;;)
+Line 4768 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (c >= 256 ||(md->ctypes[c] & ctype_space) == 0) break;
              eptr+= len;
-Line 4003 
 for (;;)
+Line 4783 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (c < 256 && (md->ctypes[c] & ctype_word) != 0) break;
              eptr+= len;
-Line 4014 
 for (;;)
+Line 4798 
 for (;;)
            for (i = min; i < max; i++)
              {
              int len = 1;
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              GETCHARLEN(c, eptr, len);
              if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) break;
              eptr+= len;
-Line 4044 
 for (;;)
+Line 4832 
 for (;;)
          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)
                {
-               if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
+               SCHECK_PARTIAL();
-               eptr++;
+               break;
                }
-             break;
+             if (IS_NEWLINE(eptr)) break;
+             eptr++;
              }
-           /* 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))
-             c = md->end_subject - eptr;
+             {
-           eptr += c;
+             eptr = md->end_subject;
+             SCHECK_PARTIAL();
+             }
+           else eptr += c;
            break;
            case OP_ANYNL:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              c = *eptr;
              if (c == 0x000d)
                {
-Line 4074 
 for (;;)
+Line 4871 
 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++;
                }
-Line 4084 
 for (;;)
+Line 4883 
 for (;;)
            case OP_NOT_HSPACE:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              c = *eptr;
              if (c == 0x09 || c == 0x20 || c == 0xa0) break;
              eptr++;
-Line 4094 
 for (;;)
+Line 4897 
 for (;;)
            case OP_HSPACE:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              c = *eptr;
              if (c != 0x09 && c != 0x20 && c != 0xa0) break;
              eptr++;
-Line 4104 
 for (;;)
+Line 4911 
 for (;;)
            case OP_NOT_VSPACE:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              c = *eptr;
              if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85)
                break;
-Line 4115 
 for (;;)
+Line 4926 
 for (;;)
            case OP_VSPACE:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject) break;
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
+               break;
+               }
              c = *eptr;
              if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85)
                break;
-Line 4126 
 for (;;)
+Line 4941 
 for (;;)
            case OP_NOT_DIGIT:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_digit) != 0)
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                break;
+               }
+             if ((md->ctypes[*eptr] & ctype_digit) != 0) break;
              eptr++;
              }
            break;
-Line 4135 
 for (;;)
+Line 4954 
 for (;;)
            case OP_DIGIT:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_digit) == 0)
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                break;
+               }
+             if ((md->ctypes[*eptr] & ctype_digit) == 0) break;
              eptr++;
              }
            break;
-Line 4144 
 for (;;)
+Line 4967 
 for (;;)
            case OP_NOT_WHITESPACE:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_space) != 0)
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                break;
+               }
+             if ((md->ctypes[*eptr] & ctype_space) != 0) break;
              eptr++;
              }
            break;
-Line 4153 
 for (;;)
+Line 4980 
 for (;;)
            case OP_WHITESPACE:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_space) == 0)
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                break;
+               }
+             if ((md->ctypes[*eptr] & ctype_space) == 0) break;
              eptr++;
              }
            break;
-Line 4162 
 for (;;)
+Line 4993 
 for (;;)
            case OP_NOT_WORDCHAR:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_word) != 0)
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                break;
+               }
+             if ((md->ctypes[*eptr] & ctype_word) != 0) break;
              eptr++;
              }
            break;
-Line 4171 
 for (;;)
+Line 5006 
 for (;;)
            case OP_WORDCHAR:
            for (i = min; i < max; i++)
              {
-             if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_word) == 0)
+             if (eptr >= md->end_subject)
+               {
+               SCHECK_PARTIAL();
                break;
+               }
+             if ((md->ctypes[*eptr] & ctype_word) == 0) break;
              eptr++;
              }
            break;
-Line 4224 
 HEAP_RETURN:
+Line 5063 
 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) LBL(48)
+ #ifdef SUPPORT_UTF8
-   LBL(49) LBL(50) LBL(51) LBL(52) LBL(53) LBL(54)
+   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 4321 
 Returns:          > 0 => success; value
+Line 5165 
 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 4345 
 const uschar *tables;
+Line 5189 
 const uschar *tables;
  const uschar *start_bits = NULL;
  USPTR start_match = (USPTR)subject + start_offset;
  USPTR end_subject;
+ USPTR start_partial = NULL;
  USPTR req_byte_ptr = start_match - 1;
  pcre_study_data internal_study;
-Line 4361 
 if (re == NULL || subject == NULL ||
+Line 5206 
 if (re == NULL || subject == NULL ||
     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;
  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
+ /* This information is for finding all the numbers associated with a given
+ name, for condition testing. */
+ md->name_table = (uschar *)re + re->name_table_offset;
+ md->name_count = re->name_count;
+ md->name_entry_size = re->name_entry_size;
  /* Fish out the optional data from the extra_data structure, first setting
  the default values. */
-Line 4408 
 if (re->magic_number != MAGIC_NUMBER)
+Line 5260 
 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 4423 
 end_subject = md->end_subject;
+Line 5275 
 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;
  md->notempty = (options & PCRE_NOTEMPTY) != 0;
- md->partial = (options & PCRE_PARTIAL) != 0;
+ md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
+ md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
+               ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
  md->hitend = FALSE;
  md->recursive = NULL;                   /* No recursion at top level */
-Line 4435 
 md->recursive = NULL;
+Line 5290 
 md->recursive = NULL;
  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;
+   case PCRE_NEWLINE_CR: newline = CHAR_CR; break;
-   case PCRE_NEWLINE_LF: newline = '\n'; break;
+   case PCRE_NEWLINE_LF: newline = CHAR_NL; break;
    case PCRE_NEWLINE_CR+
-        PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;
+        PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break;
    case PCRE_NEWLINE_ANY: newline = -1; break;
    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
    default: return PCRE_ERROR_BADNEWLINE;
-Line 4475 
 else
+Line 5356 
 else
      }
    }
- /* Partial matching is supported only for a restricted set of regexes at the
+ /* Partial matching was originally supported only for a restricted set of
- moment. */
+ regexes; from release 8.00 there are no restrictions, but the bits are still
+ defined (though never set). So there's no harm in leaving this code. */
- 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 4487 
 back the character offset. */
+Line 5369 
 back the character offset. */
  #ifdef SUPPORT_UTF8
  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
    {
-   if (_pcre_valid_utf8((uschar *)subject, length) >= 0)
+   if (_pcre_valid_utf8((USPTR)subject, length) >= 0)
      return PCRE_ERROR_BADUTF8;
    if (start_offset > 0 && start_offset < length)
      {
-     int tb = ((uschar *)subject)[start_offset];
+     int tb = ((USPTR)subject)[start_offset];
      if (tb > 127)
        {
        tb &= 0xc0;
-Line 4555 
 studied, there may be a bitmap of possib
+Line 5437 
 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 4563 
 if (!anchored)
+Line 5445 
 if (!anchored)
      }
    else
      if (!startline && study != NULL &&
-       (study->options & PCRE_STUDY_MAPPED) != 0)
+       (study->flags & PCRE_STUDY_MAPPED) != 0)
          start_bits = study->start_bits;
    }
  /* 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 4597 
 for(;;)
+Line 5479 
 for(;;)
      while (iptr < iend) *iptr++ = -1;
      }
-   /* Advance to a unique first char if possible. If firstline is TRUE, the
+   /* If firstline is TRUE, the start of the match is constrained to the first
-   start of the match is constrained to the first line of a multiline string.
+   line of a multiline string. That is, the match must be before or at the first
-   That is, the match must be before or at the first newline. Implement this by
+   newline. Implement this by temporarily adjusting end_subject so that we stop
-   temporarily adjusting end_subject so that we stop scanning at a newline. If
+   scanning at a newline. If the match fails at the newline, later code breaks
-   the match fails at the newline, later code breaks this loop. */
+   this loop. */
    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 */
+   /* There are some optimizations that avoid running the match if a known
+   starting point is not found, or if a known later character is not present.
+   However, there is an option that disables these, for testing and for ensuring
+   that all callouts do actually occur. */
-   if (first_byte >= 0)
+   if ((options & PCRE_NO_START_OPTIMIZE) == 0)
      {
-     if (first_byte_caseless)
+     /* Advance to a unique first byte if there is one. */
-       while (start_match < end_subject &&
-              md->lcc[*start_match] != first_byte)
+     if (first_byte >= 0)
-         start_match++;
+       {
-     else
+       if (first_byte_caseless)
-       while (start_match < end_subject && *start_match != first_byte)
+         while (start_match < end_subject && md->lcc[*start_match] != first_byte)
-         start_match++;
+           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)
+     else if (startline)
-     {
-     if (start_match > md->start_subject + start_offset)
        {
-       while (start_match <= end_subject && !WAS_NEWLINE(start_match))
+       if (start_match > md->start_subject + start_offset)
-         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,
+         and we are now at a LF, advance the match position by one more character.
+         */
-       /* If we have just passed a CR and the newline option is ANY or ANYCRLF,
+         if (start_match[-1] == CHAR_CR &&
-       and we are now at a LF, advance the match position by one more character.
+              (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
-       */
+              start_match < end_subject &&
+              *start_match == CHAR_NL)
-       if (start_match[-1] == '\r' &&
+           start_match++;
-            (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
+         }
-            start_match < end_subject &&
-            *start_match == '\n')
-         start_match++;
        }
-     }
-   /* Or to a non-unique first char after study */
+     /* Or to a non-unique first byte after study */
-   else if (start_bits != NULL)
+     else if (start_bits != NULL)
-     {
-     while (start_match < end_subject)
        {
-       register unsigned int c = *start_match;
+       while (start_match < end_subject)
-       if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++; else break;
+         {
+         register unsigned int c = *start_match;
+         if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++;
+           else break;
+         }
        }
-     }
+     }   /* Starting optimizations */
    /* Restore fudged end_subject */
    end_subject = save_end_subject;
- #ifdef DEBUG  /* Sigh. Some compilers never learn. */
+   /* The following two optimizations are disabled for partial matching or if
-   printf(">>>> Match against: ");
+   disabling is explicitly requested. */
-   pchars(start_match, end_subject - start_match, TRUE, md);
-   printf("\n");
- #endif
-   /* If req_byte is set, we know that that character must appear in the subject
+   if ((options & PCRE_NO_START_OPTIMIZE) == 0 && !md->partial)
-   for the match to succeed. If the first character is set, req_byte must be
-   later in the subject; otherwise the test starts at the match point. This
-   optimization can save a huge amount of backtracking in patterns with nested
-   unlimited repeats that aren't going to match. Writing separate code for
-   cased/caseless versions makes it go faster, as does using an autoincrement
-   and backing off on a match.
-   HOWEVER: when the subject string is very, very long, searching to its end can
-   take a long time, and give bad performance on quite ordinary patterns. This
-   showed up when somebody was matching something like /^\d+C/ on a 32-megabyte
-   string... so we don't do this when the string is sufficiently long.
-   ALSO: this processing is disabled when partial matching is requested.
-   */
-   if (req_byte >= 0 &&
-       end_subject - start_match < REQ_BYTE_MAX &&
-       !md->partial)
      {
-     register USPTR p = start_match + ((first_byte >= 0)? 1 : 0);
+     /* If the pattern was studied, a minimum subject length may be set. This is
+     a lower bound; no actual string of that length may actually match the
+     pattern. Although the value is, strictly, in characters, we treat it as
+     bytes to avoid spending too much time in this optimization. */
+     if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 &&
+         (pcre_uint32)(end_subject - start_match) < study->minlength)
+       {
+       rc = MATCH_NOMATCH;
+       break;
+       }
+     /* If req_byte is set, we know that that character must appear in the
+     subject for the match to succeed. If the first character is set, req_byte
+     must be later in the subject; otherwise the test starts at the match point.
+     This optimization can save a huge amount of backtracking in patterns with
+     nested unlimited repeats that aren't going to match. Writing separate code
+     for cased/caseless versions makes it go faster, as does using an
+     autoincrement and backing off on a match.
-     /* We don't need to repeat the search if we haven't yet reached the
+     HOWEVER: when the subject string is very, very long, searching to its end
-     place we found it at last time. */
+     can take a long time, and give bad performance on quite ordinary patterns.
+     This showed up when somebody was matching something like /^\d+C/ on a
+-megabyte string... so we don't do this when the string is sufficiently
+     long. */
-     if (p > req_byte_ptr)
+     if (req_byte >= 0 && end_subject - start_match < REQ_BYTE_MAX)
        {
-       if (req_byte_caseless)
+       register USPTR p = start_match + ((first_byte >= 0)? 1 : 0);
+       /* We don't need to repeat the search if we haven't yet reached the
+       place we found it at last time. */
+       if (p > req_byte_ptr)
          {
-         while (p < end_subject)
+         if (req_byte_caseless)
            {
-           register int pp = *p++;
+           while (p < end_subject)
-           if (pp == req_byte || pp == req_byte2) { p--; break; }
+             {
+             register int pp = *p++;
+             if (pp == req_byte || pp == req_byte2) { p--; break; }
+             }
            }
-         }
+         else
-       else
-         {
-         while (p < end_subject)
            {
-           if (*p++ == req_byte) { p--; break; }
+           while (p < end_subject)
+             {
+             if (*p++ == req_byte) { p--; break; }
+             }
            }
-         }
-       /* If we can't find the required character, break the matching loop,
+         /* If we can't find the required character, break the matching loop,
-       forcing a match failure. */
+         forcing a match failure. */
-       if (p >= end_subject)
+         if (p >= end_subject)
-         {
+           {
-         rc = MATCH_NOMATCH;
+           rc = MATCH_NOMATCH;
-         break;
+           break;
-         }
+           }
-       /* If we have found the required character, save the point where we
+         /* If we have found the required character, save the point where we
-       found it, so that we don't search again next time round the loop if
+         found it, so that we don't search again next time round the loop if
-       the start hasn't passed this character yet. */
+         the start hasn't passed this character yet. */
-       req_byte_ptr = p;
+         req_byte_ptr = p;
+         }
        }
      }
-   /* OK, we can now run the match. */
+ #ifdef PCRE_DEBUG  /* Sigh. Some compilers never learn. */
+   printf(">>>> Match against: ");
+   pchars(start_match, end_subject - start_match, TRUE, md);
+   printf("\n");
+ #endif
+   /* OK, we can now run the match. If "hitend" is set afterwards, remember the
+   first starting point for which a partial match was found. */
    md->start_match_ptr = start_match;
+   md->start_used_ptr = start_match;
    md->match_call_count = 0;
    rc = match(start_match, md->start_code, start_match, 2, md, ims, NULL, 0, 0);
+   if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
    switch(rc)
      {
-Line 4759 
 for(;;)
+Line 5689 
 for(;;)
      rc = MATCH_NOMATCH;
      goto ENDLOOP;
-     /* Any other return is some kind of error. */
+     /* Any other return is either a match, or some kind of error. */
      default:
      goto ENDLOOP;
-Line 4789 
 for(;;)
+Line 5719 
 for(;;)
    not contain any explicit matches for \r or \n, and the newline option is CRLF
    or ANY or ANYCRLF, advance the match position by one more character. */
-   if (start_match[-1] == '\r' &&
+   if (start_match[-1] == CHAR_CR &&
        start_match < end_subject &&
-       *start_match == '\n' &&
+       *start_match == CHAR_NL &&
-       (re->options & PCRE_HASCRORLF) == 0 &&
+       (re->flags & PCRE_HASCRORLF) == 0 &&
          (md->nltype == NLTYPE_ANY ||
           md->nltype == NLTYPE_ANYCRLF ||
           md->nllen == 2))
-Line 4865 
 if (using_temporary_offsets)
+Line 5795 
 if (using_temporary_offsets)
    (pcre_free)(md->offset_vector);
    }
- if (rc != MATCH_NOMATCH)
+ if (rc != MATCH_NOMATCH && rc != PCRE_ERROR_PARTIAL)
    {
    DPRINTF((">>>> error: returning %d\n", rc));
    return rc;
    }
- else if (md->partial && md->hitend)
+ else if (start_partial != NULL)
    {
    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));
+   if (offsetcount > 1)
+     {
+     offsets[0] = start_partial - (USPTR)subject;
+     offsets[1] = end_subject - (USPTR)subject;
+     }
    return PCRE_ERROR_PARTIAL;
    }
  else

 Legend:



Removed from v.227
 


changed lines


 
Added in v.500
 Legend:



Removed from v.227
 


changed lines


 
Added in v.500
-Removed from v.227
+Added in v.500

webmaster@exim.org	ViewVC Help
Powered by ViewVC 1.1.12