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

Diff of /code/trunk/pcre_exec.c

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

revision 534 by ph10, Thu Jun 3 18:26:05 2010 UTC revision 608 by ph10, Sun Jun 12 16:25:55 2011 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2010 University of Cambridge             Copyright (c) 1997-2011 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 57  possible. There are also some static sup Line 57  possible. There are also some static sup
57  #undef min  #undef min
58  #undef max  #undef max
59    
60  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
61    of call to match(). We do it this way to save on using another stack variable,
62    as stack usage is to be discouraged. */
63    
64  #define match_condassert     0x01  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
65  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
66    
67  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
68  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 73  negative to avoid the external error cod Line 75  negative to avoid the external error cod
75    
76  #define MATCH_ACCEPT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_COMMIT       (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_PRUNE        (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_SKIP         (-996)  #define MATCH_PRUNE        (-996)
80  #define MATCH_SKIP_ARG     (-995)  #define MATCH_SKIP         (-995)
81  #define MATCH_THEN         (-994)  #define MATCH_SKIP_ARG     (-994)
82    #define MATCH_THEN         (-993)
83    
84  /* This is a convenience macro for code that occurs many times. */  /* This is a convenience macro for code that occurs many times. */
85    
# Line 132  while (length-- > 0) Line 135  while (length-- > 0)
135  *          Match a back-reference                *  *          Match a back-reference                *
136  *************************************************/  *************************************************/
137    
138  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
139  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
140    the length passed is zero. Note that in caseless UTF-8 mode, the number of
141    subject bytes matched may be different to the number of reference bytes.
142    
143  Arguments:  Arguments:
144    offset      index into the offset vector    offset      index into the offset vector
145    eptr        points into the subject    eptr        pointer into the subject
146    length      length to be matched    length      length of reference to be matched (number of bytes)
147    md          points to match data block    md          points to match data block
148    ims         the ims flags    caseless    TRUE if caseless
149    
150  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
151  */  */
152    
153  static BOOL  static int
154  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
155    unsigned long int ims)    BOOL caseless)
156  {  {
157  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
158    register USPTR p = md->start_subject + md->offset_vector[offset];
159    
160  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
161  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 164  pchars(p, length, FALSE, md); Line 170  pchars(p, length, FALSE, md);
170  printf("\n");  printf("\n");
171  #endif  #endif
172    
173  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
174    
175  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
176    
177  /* Separate the caseless case for speed. In UTF-8 mode we can only do this  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
178  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
179  ASCII characters. */  ASCII characters. */
180    
181  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
182    {    {
183  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
184  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
185    if (md->utf8)    if (md->utf8)
186      {      {
187      USPTR endptr = eptr + length;      /* Match characters up to the end of the reference. NOTE: the number of
188      while (eptr < endptr)      bytes matched may differ, because there are some characters whose upper and
189        lower case versions code as different numbers of bytes. For example, U+023A
190        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
191        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
192        the latter. It is important, therefore, to check the length along the
193        reference, not along the subject (earlier code did this wrong). */
194    
195        USPTR endptr = p + length;
196        while (p < endptr)
197        {        {
198        int c, d;        int c, d;
199          if (eptr >= md->end_subject) return -1;
200        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
201        GETCHARINC(d, p);        GETCHARINC(d, p);
202        if (c != d && c != UCD_OTHERCASE(d)) return FALSE;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
203        }        }
204      }      }
205    else    else
# Line 193  if ((ims & PCRE_CASELESS) != 0) Line 208  if ((ims & PCRE_CASELESS) != 0)
208    
209    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
210    is no UCP support. */    is no UCP support. */
211        {
212    while (length-- > 0)      if (eptr + length > md->end_subject) return -1;
213      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }      while (length-- > 0)
214          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
215        }
216    }    }
217    
218  /* In the caseful case, we can just compare the bytes, whether or not we  /* In the caseful case, we can just compare the bytes, whether or not we
219  are in UTF-8 mode. */  are in UTF-8 mode. */
220    
221  else  else
222    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
223      if (eptr + length > md->end_subject) return -1;
224      while (length-- > 0) if (*p++ != *eptr++) return -1;
225      }
226    
227  return TRUE;  return eptr - eptr_start;
228  }  }
229    
230    
# Line 256  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 276  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
276         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
277         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
278         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
279         RM61,  RM62 };         RM61,  RM62, RM63, RM64 };
280    
281  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
282  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
# Line 266  actually used in this definition. */ Line 286  actually used in this definition. */
286  #define REGISTER register  #define REGISTER register
287    
288  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
289  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
290    { \    { \
291    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
292    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
293    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
294    }    }
295  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 278  actually used in this definition. */ Line 298  actually used in this definition. */
298    return ra; \    return ra; \
299    }    }
300  #else  #else
301  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
302    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
303  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
304  #endif  #endif
305    
# Line 292  argument of match(), which never changes Line 312  argument of match(), which never changes
312    
313  #define REGISTER  #define REGISTER
314    
315  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
316    {\    {\
317    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
318    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
319    frame->Xwhere = rw; \    frame->Xwhere = rw; \
320    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
# Line 302  argument of match(), which never changes Line 322  argument of match(), which never changes
322    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
323    newframe->Xmarkptr = markptr;\    newframe->Xmarkptr = markptr;\
324    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
325    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
326    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
327    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
328    frame = newframe;\    frame = newframe;\
# Line 340  typedef struct heapframe { Line 358  typedef struct heapframe {
358    USPTR Xmstart;    USPTR Xmstart;
359    USPTR Xmarkptr;    USPTR Xmarkptr;
360    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
361    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
362    unsigned int Xrdepth;    unsigned int Xrdepth;
363    
364    /* Function local variables */    /* Function local variables */
# Line 363  typedef struct heapframe { Line 379  typedef struct heapframe {
379    BOOL Xcondition;    BOOL Xcondition;
380    BOOL Xprev_is_word;    BOOL Xprev_is_word;
381    
   unsigned long int Xoriginal_ims;  
   
382  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
383    int Xprop_type;    int Xprop_type;
384    int Xprop_value;    int Xprop_value;
# Line 422  immediately. The second one is used when Line 436  immediately. The second one is used when
436  the subject. */  the subject. */
437    
438  #define CHECK_PARTIAL()\  #define CHECK_PARTIAL()\
439    if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\    if (md->partial != 0 && eptr >= md->end_subject && \
440      {\        eptr > md->start_used_ptr) \
441      md->hitend = TRUE;\      { \
442      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\      md->hitend = TRUE; \
443        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
444      }      }
445    
446  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
447    if (md->partial != 0 && eptr > mstart)\    if (md->partial != 0 && eptr > md->start_used_ptr) \
448      {\      { \
449      md->hitend = TRUE;\      md->hitend = TRUE; \
450      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
451      }      }
452    
453    
# Line 449  Arguments: Line 464  Arguments:
464     markptr     pointer to the most recent MARK name, or NULL     markptr     pointer to the most recent MARK name, or NULL
465     offset_top  current top pointer     offset_top  current top pointer
466     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
467     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
468                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
469     rdepth      the recursion depth     rdepth      the recursion depth
470    
471  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
# Line 467  Returns:       MATCH_MATCH if matched Line 477  Returns:       MATCH_MATCH if matched
477    
478  static int  static int
479  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
480    const uschar *markptr, int offset_top, match_data *md, unsigned long int ims,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
481    eptrblock *eptrb, int flags, unsigned int rdepth)    unsigned int rdepth)
482  {  {
483  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
484  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 480  register unsigned int c;   /* Character Line 490  register unsigned int c;   /* Character
490  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
491    
492  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
493    BOOL caseless;
494  int condcode;  int condcode;
495    
496  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
# Line 488  heap storage. Set up the top-level frame Line 499  heap storage. Set up the top-level frame
499  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
500    
501  #ifdef NO_RECURSE  #ifdef NO_RECURSE
502  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
503  if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);  if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
504  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
505    
# Line 499  frame->Xecode = ecode; Line 510  frame->Xecode = ecode;
510  frame->Xmstart = mstart;  frame->Xmstart = mstart;
511  frame->Xmarkptr = markptr;  frame->Xmarkptr = markptr;
512  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
513  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
514  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
515    
516  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 515  HEAP_RECURSE: Line 524  HEAP_RECURSE:
524  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
525  #define markptr            frame->Xmarkptr  #define markptr            frame->Xmarkptr
526  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
527  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
528  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
529    
530  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 539  HEAP_RECURSE: Line 546  HEAP_RECURSE:
546  #define condition          frame->Xcondition  #define condition          frame->Xcondition
547  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
548    
 #define original_ims       frame->Xoriginal_ims  
   
549  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
550  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
551  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
# Line 577  i, and fc and c, can be the same variabl Line 582  i, and fc and c, can be the same variabl
582  #define fi i  #define fi i
583  #define fc c  #define fc c
584    
585    /* Many of the following variables are used only in small blocks of the code.
586  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  My normal style of coding would have declared them within each of those blocks.
587  const uschar *charptr;             /* in small blocks of the code. My normal */  However, in order to accommodate the version of this code that uses an external
588  #endif                             /* style of coding would have declared    */  "stack" implemented on the heap, it is easier to declare them all here, so the
589  const uschar *callpat;             /* them within each of those blocks.      */  declarations can be cut out in a block. The only declarations within blocks
590  const uschar *data;                /* However, in order to accommodate the   */  below are for variables that do not have to be preserved over a recursive call
591  const uschar *next;                /* version of this code that uses an      */  to RMATCH(). */
592  USPTR         pp;                  /* external "stack" implemented on the    */  
593  const uschar *prev;                /* heap, it is easier to declare them all */  #ifdef SUPPORT_UTF8
594  USPTR         saved_eptr;          /* here, so the declarations can be cut   */  const uschar *charptr;
595                                     /* out in a block. The only declarations  */  #endif
596  recursion_info new_recursive;      /* within blocks below are for variables  */  const uschar *callpat;
597                                     /* that do not have to be preserved over  */  const uschar *data;
598  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  const uschar *next;
599    USPTR         pp;
600    const uschar *prev;
601    USPTR         saved_eptr;
602    
603    recursion_info new_recursive;
604    
605    BOOL cur_is_word;
606  BOOL condition;  BOOL condition;
607  BOOL prev_is_word;  BOOL prev_is_word;
608    
 unsigned long int original_ims;  
   
609  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
610  int prop_type;  int prop_type;
611  int prop_value;  int prop_value;
# Line 622  int stacksave[REC_STACK_SAVE_MAX]; Line 632  int stacksave[REC_STACK_SAVE_MAX];
632  eptrblock newptrb;  eptrblock newptrb;
633  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
634    
635    /* To save space on the stack and in the heap frame, I have doubled up on some
636    of the local variables that are used only in localised parts of the code, but
637    still need to be preserved over recursive calls of match(). These macros define
638    the alternative names that are used. */
639    
640    #define allow_zero    cur_is_word
641    #define cbegroup      condition
642    #define code_offset   codelink
643    #define condassert    condition
644    #define matched_once  prev_is_word
645    
646  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
647  variables. */  variables. */
648    
# Line 658  haven't exceeded the recursive call limi Line 679  haven't exceeded the recursive call limi
679  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
680  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
681    
 original_ims = ims;    /* Save for resetting on ')' */  
   
682  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
683  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
684  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
685  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
686  When match() is called in other circumstances, don't add to the chain. The  
687  match_cbegroup flag must NOT be used with tail recursion, because the memory  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
688  block that is used is on the stack, so a new one may be required for each  such remembered pointers, to be checked when we hit the closing ket, in order
689  match(). */  to break infinite loops that match no characters. When match() is called in
690    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
691    NOT be used with tail recursion, because the memory block that is used is on
692    the stack, so a new one may be required for each match(). */
693    
694  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
695    {    {
696    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
697    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
698    eptrb = &newptrb;    eptrb = &newptrb;
699      md->match_function_type = 0;
700    }    }
701    
702  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 682  for (;;) Line 705  for (;;)
705    {    {
706    minimize = possessive = FALSE;    minimize = possessive = FALSE;
707    op = *ecode;    op = *ecode;
708    
709    switch(op)    switch(op)
710      {      {
711      case OP_MARK:      case OP_MARK:
712      markptr = ecode + 2;      markptr = ecode + 2;
713      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
714        ims, eptrb, flags, RM55);        eptrb, RM55);
715    
716      /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an      /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
717      argument, and we must check whether that argument matches this MARK's      argument, and we must check whether that argument matches this MARK's
# Line 710  for (;;) Line 733  for (;;)
733      case OP_FAIL:      case OP_FAIL:
734      MRRETURN(MATCH_NOMATCH);      MRRETURN(MATCH_NOMATCH);
735    
736        /* COMMIT overrides PRUNE, SKIP, and THEN */
737    
738      case OP_COMMIT:      case OP_COMMIT:
739      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
740        ims, eptrb, flags, RM52);        eptrb, RM52);
741      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
742            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
743            rrc != MATCH_THEN)
744          RRETURN(rrc);
745      MRRETURN(MATCH_COMMIT);      MRRETURN(MATCH_COMMIT);
746    
747        /* PRUNE overrides THEN */
748    
749      case OP_PRUNE:      case OP_PRUNE:
750      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
751        ims, eptrb, flags, RM51);        eptrb, RM51);
752      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
753      MRRETURN(MATCH_PRUNE);      MRRETURN(MATCH_PRUNE);
754    
755      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
756      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
757        ims, eptrb, flags, RM56);        eptrb, RM56);
758      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
759      md->mark = ecode + 2;      md->mark = ecode + 2;
760      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
761    
762        /* SKIP overrides PRUNE and THEN */
763    
764      case OP_SKIP:      case OP_SKIP:
765      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
766        ims, eptrb, flags, RM53);        eptrb, RM53);
767      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
770      MRRETURN(MATCH_SKIP);      MRRETURN(MATCH_SKIP);
771    
772      case OP_SKIP_ARG:      case OP_SKIP_ARG:
773      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
774        ims, eptrb, flags, RM57);        eptrb, RM57);
775      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
776          RRETURN(rrc);
777    
778      /* Pass back the current skip name by overloading md->start_match_ptr and      /* Pass back the current skip name by overloading md->start_match_ptr and
779      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
# Line 749  for (;;) Line 783  for (;;)
783      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
784      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
785    
786        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
787        the alt that is at the start of the current branch. This makes it possible
788        to skip back past alternatives that precede the THEN within the current
789        branch. */
790    
791      case OP_THEN:      case OP_THEN:
792      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
793        ims, eptrb, flags, RM54);        eptrb, RM54);
794      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
795        md->start_match_ptr = ecode - GET(ecode, 1);
796      MRRETURN(MATCH_THEN);      MRRETURN(MATCH_THEN);
797    
798      case OP_THEN_ARG:      case OP_THEN_ARG:
799      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
800        ims, eptrb, flags, RM58);        offset_top, md, eptrb, RM58);
801      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
802      md->mark = ecode + 2;      md->start_match_ptr = ecode - GET(ecode, 1);
803        md->mark = ecode + LINK_SIZE + 2;
804      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
805    
806      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle a capturing bracket, other than those that are possessive with an
807      the current subject position in the working slot at the top of the vector.      unlimited repeat. If there is space in the offset vector, save the current
808      We mustn't change the current values of the data slot, because they may be      subject position in the working slot at the top of the vector. We mustn't
809      set from a previous iteration of this group, and be referred to by a      change the current values of the data slot, because they may be set from a
810      reference inside the group.      previous iteration of this group, and be referred to by a reference inside
811        the group. If we fail to match, we need to restore this value and also the
     If the bracket fails to match, we need to restore this value and also the  
812      values of the final offsets, in case they were set by a previous iteration      values of the final offsets, in case they were set by a previous iteration
813      of the same bracket.      of the same bracket.
814    
# Line 780  for (;;) Line 820  for (;;)
820      case OP_SCBRA:      case OP_SCBRA:
821      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
822      offset = number << 1;      offset = number << 1;
823    
824  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
825      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
826      printf("subject=");      printf("subject=");
# Line 799  for (;;) Line 839  for (;;)
839        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
840          (int)(eptr - md->start_subject);          (int)(eptr - md->start_subject);
841    
842        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
843          {          {
844          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
845            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
846          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);            eptrb, RM1);
847            if (rrc != MATCH_NOMATCH &&
848                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
849              RRETURN(rrc);
850          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
851          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
852            if (*ecode != OP_ALT) break;
853          }          }
       while (*ecode == OP_ALT);  
854    
855        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
856    
# Line 831  for (;;) Line 873  for (;;)
873      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
874      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
875    
876      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing bracket, except for possessive with unlimited repeat. Loop
877      final alternative within the brackets, we would return the result of a      for all the alternatives. When we get to the final alternative within the
878      recursive call to match() whatever happened. We can reduce stack usage by      brackets, we would return the result of a recursive call to match()
879      turning this into a tail recursion, except in the case when match_cbegroup      whatever happened. We can reduce stack usage by turning this into a tail
880      is set.*/      recursion, except in the case of a possibly empty group.*/
881    
882      case OP_BRA:      case OP_BRA:
883      case OP_SBRA:      case OP_SBRA:
884      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
     flags = (op >= OP_SBRA)? match_cbegroup : 0;  
885      for (;;)      for (;;)
886        {        {
887        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */
888          {          {
889          if (flags == 0)    /* Not a possibly empty group */          if (op >= OP_SBRA)   /* Possibly empty group */
890            {            {
891            ecode += _pcre_OP_lengths[*ecode];            md->match_function_type = MATCH_CBEGROUP;
892            DPRINTF(("bracket 0 tail recursion\n"));            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
893            goto TAIL_RECURSE;              RM48);
894            }            if (rrc == MATCH_NOMATCH) md->mark = markptr;
895              RRETURN(rrc);
896              }
897            /* Not a possibly empty group; use tail recursion */
898            ecode += _pcre_OP_lengths[*ecode];
899            DPRINTF(("bracket 0 tail recursion\n"));
900            goto TAIL_RECURSE;
901            }
902    
903          /* Possibly empty group; can't use tail recursion. */        /* For non-final alternatives, continue the loop for a NOMATCH result;
904          otherwise return. */
905    
906          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
907            eptrb, flags, RM48);        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
908          if (rrc == MATCH_NOMATCH) md->mark = markptr;          RM2);
909          if (rrc != MATCH_NOMATCH &&
910              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
911          RRETURN(rrc);          RRETURN(rrc);
912          ecode += GET(ecode, 1);
913          }
914        /* Control never reaches here. */
915    
916        /* Handle possessive capturing brackets with an unlimited repeat. We come
917        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
918        handled similarly to the normal case above. However, the matching is
919        different. The end of these brackets will always be OP_KETRPOS, which
920        returns MATCH_KETRPOS without going further in the pattern. By this means
921        we can handle the group by iteration rather than recursion, thereby
922        reducing the amount of stack needed. */
923    
924        case OP_CBRAPOS:
925        case OP_SCBRAPOS:
926        allow_zero = FALSE;
927    
928        POSSESSIVE_CAPTURE:
929        number = GET2(ecode, 1+LINK_SIZE);
930        offset = number << 1;
931    
932    #ifdef PCRE_DEBUG
933        printf("start possessive bracket %d\n", number);
934        printf("subject=");
935        pchars(eptr, 16, TRUE, md);
936        printf("\n");
937    #endif
938    
939        if (offset < md->offset_max)
940          {
941          matched_once = FALSE;
942          code_offset = ecode - md->start_code;
943    
944          save_offset1 = md->offset_vector[offset];
945          save_offset2 = md->offset_vector[offset+1];
946          save_offset3 = md->offset_vector[md->offset_end - number];
947          save_capture_last = md->capture_last;
948    
949          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
950    
951          /* Each time round the loop, save the current subject position for use
952          when the group matches. For MATCH_MATCH, the group has matched, so we
953          restart it with a new subject starting position, remembering that we had
954          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
955          usual. If we haven't matched any alternatives in any iteration, check to
956          see if a previous iteration matched. If so, the group has matched;
957          continue from afterwards. Otherwise it has failed; restore the previous
958          capture values before returning NOMATCH. */
959    
960          for (;;)
961            {
962            md->offset_vector[md->offset_end - number] =
963              (int)(eptr - md->start_subject);
964            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
965            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
966              eptrb, RM63);
967            if (rrc == MATCH_KETRPOS)
968              {
969              offset_top = md->end_offset_top;
970              eptr = md->end_match_ptr;
971              ecode = md->start_code + code_offset;
972              save_capture_last = md->capture_last;
973              matched_once = TRUE;
974              continue;
975              }
976            if (rrc != MATCH_NOMATCH &&
977                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
978              RRETURN(rrc);
979            md->capture_last = save_capture_last;
980            ecode += GET(ecode, 1);
981            if (*ecode != OP_ALT) break;
982          }          }
983    
984          if (!matched_once)
985            {
986            md->offset_vector[offset] = save_offset1;
987            md->offset_vector[offset+1] = save_offset2;
988            md->offset_vector[md->offset_end - number] = save_offset3;
989            }
990    
991          if (rrc != MATCH_THEN) md->mark = markptr;
992          if (allow_zero || matched_once)
993            {
994            ecode += 1 + LINK_SIZE;
995            break;
996            }
997    
998          RRETURN(MATCH_NOMATCH);
999          }
1000    
1001        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1002        as a non-capturing bracket. */
1003    
1004        /* For non-final alternatives, continue the loop for a NOMATCH result;      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1005        otherwise return. */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1006    
1007        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1008    
1009        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1010        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1011    
1012        /* Non-capturing possessive bracket with unlimited repeat. We come here
1013        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1014        without the capturing complication. It is written out separately for speed
1015        and cleanliness. */
1016    
1017        case OP_BRAPOS:
1018        case OP_SBRAPOS:
1019        allow_zero = FALSE;
1020    
1021        POSSESSIVE_NON_CAPTURE:
1022        matched_once = FALSE;
1023        code_offset = ecode - md->start_code;
1024    
1025        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      for (;;)
1026          eptrb, flags, RM2);        {
1027        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1028          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1029            eptrb, RM64);
1030          if (rrc == MATCH_KETRPOS)
1031            {
1032            eptr = md->end_match_ptr;
1033            ecode = md->start_code + code_offset;
1034            matched_once = TRUE;
1035            continue;
1036            }
1037          if (rrc != MATCH_NOMATCH &&
1038              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1039            RRETURN(rrc);
1040        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1041          if (*ecode != OP_ALT) break;
1042        }        }
1043    
1044        if (matched_once || allow_zero)
1045          {
1046          ecode += 1 + LINK_SIZE;
1047          break;
1048          }
1049        RRETURN(MATCH_NOMATCH);
1050    
1051      /* Control never reaches here. */      /* Control never reaches here. */
1052    
1053      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
# Line 878  for (;;) Line 1058  for (;;)
1058    
1059      case OP_COND:      case OP_COND:
1060      case OP_SCOND:      case OP_SCOND:
1061      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1062    
1063      /* Because of the way auto-callout works during compile, a callout item is      /* Because of the way auto-callout works during compile, a callout item is
1064      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 1053  for (;;) Line 1233  for (;;)
1233        }        }
1234    
1235      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1236      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1237      assertion. */      an assertion. */
1238    
1239      else      else
1240        {        {
1241        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1242            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1243        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1244          {          {
1245          condition = TRUE;          condition = TRUE;
1246          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1247          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1248          }          }
1249        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH &&
1250                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1251          {          {
1252          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1253          }          }
# Line 1079  for (;;) Line 1260  for (;;)
1260    
1261      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one,
1262      we can use tail recursion to avoid using another stack frame, except when      we can use tail recursion to avoid using another stack frame, except when
1263      match_cbegroup is required for an unlimited repeat of a possibly empty      we have an unlimited repeat of a possibly empty group. If the second
1264      group. If the second alternative doesn't exist, we can just plough on. */      alternative doesn't exist, we can just plough on. */
1265    
1266      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1267        {        {
1268        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1269        if (op == OP_SCOND)        /* Possibly empty group */        if (op == OP_SCOND)        /* Possibly empty group */
1270          {          {
1271          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          md->match_function_type = MATCH_CBEGROUP;
1272            RMATCH(eptr, ecode, offset_top, md, eptrb, RM49);
1273          RRETURN(rrc);          RRETURN(rrc);
1274          }          }
1275        else                       /* Group must match something */        else goto TAIL_RECURSE;
         {  
         flags = 0;  
         goto TAIL_RECURSE;  
         }  
1276        }        }
1277      else                         /* Condition false & no alternative */      else                         /* Condition false & no alternative */
1278        {        {
# Line 1127  for (;;) Line 1305  for (;;)
1305      break;      break;
1306    
1307    
1308      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. If we are in a recursion, we
1309      recursion, we should restore the offsets appropriately and continue from      should restore the offsets appropriately, and if it's a top-level
1310      after the call. */      recursion, continue from after the call. */
1311    
1312      case OP_ACCEPT:      case OP_ACCEPT:
1313      case OP_END:      case OP_END:
1314      if (md->recursive != NULL && md->recursive->group_num == 0)      if (md->recursive != NULL)
1315        {        {
1316        recursion_info *rec = md->recursive;        recursion_info *rec = md->recursive;
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
1317        md->recursive = rec->prevrec;        md->recursive = rec->prevrec;
1318        memmove(md->offset_vector, rec->offset_save,        memmove(md->offset_vector, rec->offset_save,
1319          rec->saved_max * sizeof(int));          rec->saved_max * sizeof(int));
1320        offset_top = rec->save_offset_top;        offset_top = rec->save_offset_top;
1321        ims = original_ims;        if (rec->group_num == 0)
1322        ecode = rec->after_call;          {
1323        break;          ecode = rec->after_call;
1324            break;
1325            }
1326        }        }
1327    
1328      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is
# Line 1151  for (;;) Line 1330  for (;;)
1330      the subject. In both cases, backtracking will then try other alternatives,      the subject. In both cases, backtracking will then try other alternatives,
1331      if any. */      if any. */
1332    
1333      if (eptr == mstart &&      else if (eptr == mstart &&
1334          (md->notempty ||          (md->notempty ||
1335            (md->notempty_atstart &&            (md->notempty_atstart &&
1336              mstart == md->start_subject + md->start_offset)))              mstart == md->start_subject + md->start_offset)))
1337        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1338    
1339      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1340    
1341      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1342      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1343      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
# Line 1169  for (;;) Line 1348  for (;;)
1348      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1349      MRRETURN(rrc);      MRRETURN(rrc);
1350    
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
   
1351      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1352      matching won't pass the KET for an assertion. If any one branch matches,      matching won't pass the KET for an assertion. If any one branch matches,
1353      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
1354      start of each branch to move the current point backwards, so the code at      start of each branch to move the current point backwards, so the code at
1355      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1356        of a condition, we want to return immediately afterwards. The caller of
1357        this incarnation of the match() function will have set MATCH_CONDASSERT in
1358        md->match_function type, and one of these opcodes will be the first opcode
1359        that is processed. We use a local variable that is preserved over calls to
1360        match() to remember this case. */
1361    
1362      case OP_ASSERT:      case OP_ASSERT:
1363      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1364        if (md->match_function_type == MATCH_CONDASSERT)
1365          {
1366          condassert = TRUE;
1367          md->match_function_type = 0;
1368          }
1369        else condassert = FALSE;
1370    
1371      do      do
1372        {        {
1373        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1374        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1375          {          {
1376          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1377          break;          break;
1378          }          }
1379        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH &&
1380              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1381            RRETURN(rrc);
1382        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1383        }        }
1384      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1385    
1386      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1387    
1388      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1389    
1390      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1391    
1392      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1393      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1218  for (;;) Line 1403  for (;;)
1403    
1404      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1405      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1406        if (md->match_function_type == MATCH_CONDASSERT)
1407          {
1408          condassert = TRUE;
1409          md->match_function_type = 0;
1410          }
1411        else condassert = FALSE;
1412    
1413      do      do
1414        {        {
1415        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
         RM5);  
1416        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1417        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1418          {          {
1419          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1420          break;          break;
1421          }          }
1422        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH &&
1423              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1424            RRETURN(rrc);
1425        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1426        }        }
1427      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1428    
1429      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1430    
1431      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1432      continue;      continue;
1433    
# Line 1347  for (;;) Line 1540  for (;;)
1540        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1541              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1542        new_recursive.save_offset_top = offset_top;        new_recursive.save_offset_top = offset_top;
1543    
1544        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. For each top-level alternative we
1545        restore the offset and recursion data. */        restore the offset and recursion data. */
1546    
1547        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1548        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1549        do        do
1550          {          {
1551            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1552          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1553            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1554          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1555            {            {
1556            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
# Line 1365  for (;;) Line 1559  for (;;)
1559              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1560            MRRETURN(MATCH_MATCH);            MRRETURN(MATCH_MATCH);
1561            }            }
1562          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH &&
1563                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1564            {            {
1565            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1566            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
# Line 1402  for (;;) Line 1597  for (;;)
1597    
1598      do      do
1599        {        {
1600        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1601        if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */        if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
1602          {          {
1603          mstart = md->start_match_ptr;          mstart = md->start_match_ptr;
1604          break;          break;
1605          }          }
1606        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH &&
1607              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1608            RRETURN(rrc);
1609        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1610        }        }
1611      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
# Line 1439  for (;;) Line 1636  for (;;)
1636    
1637      /* The repeating kets try the rest of the pattern or restart from the      /* The repeating kets try the rest of the pattern or restart from the
1638      preceding bracket, in the appropriate order. The second "call" of match()      preceding bracket, in the appropriate order. The second "call" of match()
1639      uses tail recursion, to avoid using another stack frame. We need to reset      uses tail recursion, to avoid using another stack frame. */
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
1640    
1641      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1642        {        {
1643        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM8);
1644        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1645        ecode = prev;        ecode = prev;
       flags = 0;  
1646        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1647        }        }
1648      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1649        {        {
1650        RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);        md->match_function_type = MATCH_CBEGROUP;
1651          RMATCH(eptr, prev, offset_top, md, eptrb, RM9);
1652        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1653        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1654        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1655        }        }
1656      /* Control never gets here */      /* Control never gets here */
# Line 1479  for (;;) Line 1667  for (;;)
1667      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1668      with fixed upper repeat limits are compiled as a number of copies, with the      with fixed upper repeat limits are compiled as a number of copies, with the
1669      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1670    
1671      case OP_BRAZERO:      case OP_BRAZERO:
1672        {      next = ecode + 1;
1673        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1674        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1675        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1676        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1677      break;      break;
1678    
1679      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1680        {      next = ecode + 1;
1681        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1682        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1683        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1684        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1685      break;      break;
1686    
1687      case OP_SKIPZERO:      case OP_SKIPZERO:
1688        {      next = ecode+1;
1689        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1690        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1691      break;      break;
1692    
1693        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1694        here; just jump to the group, with allow_zero set TRUE. */
1695    
1696        case OP_BRAPOSZERO:
1697        op = *(++ecode);
1698        allow_zero = TRUE;
1699        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1700          goto POSSESSIVE_NON_CAPTURE;
1701    
1702      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1703    
1704      case OP_KET:      case OP_KET:
1705      case OP_KETRMIN:      case OP_KETRMIN:
1706      case OP_KETRMAX:      case OP_KETRMAX:
1707        case OP_KETRPOS:
1708      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1709    
1710      /* If this was a group that remembered the subject start, in order to break      /* If this was a group that remembered the subject start, in order to break
# Line 1547  for (;;) Line 1739  for (;;)
1739      a recurse into group 0, so it won't be picked up here. Instead, we catch it      a recurse into group 0, so it won't be picked up here. Instead, we catch it
1740      when the OP_END is reached. Other recursion is handled here. */      when the OP_END is reached. Other recursion is handled here. */
1741    
1742      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1743            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1744        {        {
1745        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1746        offset = number << 1;        offset = number << 1;
# Line 1578  for (;;) Line 1771  for (;;)
1771            rec->saved_max * sizeof(int));            rec->saved_max * sizeof(int));
1772          offset_top = rec->save_offset_top;          offset_top = rec->save_offset_top;
1773          ecode = rec->after_call;          ecode = rec->after_call;
         ims = original_ims;  
1774          break;          break;
1775          }          }
1776        }        }
1777    
     /* For both capturing and non-capturing groups, reset the value of the ims  
     flags, in case they got changed during the group. */  
   
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
   
1778      /* For a non-repeating ket, just continue at this level. This also      /* For a non-repeating ket, just continue at this level. This also
1779      happens for a repeating ket if no characters were matched in the group.      happens for a repeating ket if no characters were matched in the group.
1780      This is the forcible breaking of infinite loops as implemented in Perl      This is the forcible breaking of infinite loops as implemented in Perl
# Line 1600  for (;;) Line 1786  for (;;)
1786        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1787        break;        break;
1788        }        }
1789    
1790      /* The repeating kets try the rest of the pattern or restart from the      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1791      preceding bracket, in the appropriate order. In the second case, we can use      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1792      tail recursion to avoid using another stack frame, unless we have an      at a time from the outer level, thus saving stack. */
1793    
1794        if (*ecode == OP_KETRPOS)
1795          {
1796          md->end_match_ptr = eptr;
1797          md->end_offset_top = offset_top;
1798          RRETURN(MATCH_KETRPOS);
1799          }
1800    
1801        /* The normal repeating kets try the rest of the pattern or restart from
1802        the preceding bracket, in the appropriate order. In the second case, we can
1803        use tail recursion to avoid using another stack frame, unless we have an
1804      unlimited repeat of a group that can match an empty string. */      unlimited repeat of a group that can match an empty string. */
1805    
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
   
1806      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1807        {        {
1808        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1809        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1810        if (flags != 0)    /* Could match an empty string */        if (*prev >= OP_SBRA)    /* Could match an empty string */
1811          {          {
1812          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1813            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1814          RRETURN(rrc);          RRETURN(rrc);
1815          }          }
1816        ecode = prev;        ecode = prev;
# Line 1622  for (;;) Line 1818  for (;;)
1818        }        }
1819      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1820        {        {
1821        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1822          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1823        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1824        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1825        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1826        }        }
1827      /* Control never gets here */      /* Control never gets here */
1828    
1829      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1830    
1831      case OP_CIRC:      case OP_CIRC:
1832      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1833      if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         MRRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
   
1834      /* Start of subject assertion */      /* Start of subject assertion */
1835    
1836      case OP_SOD:      case OP_SOD:
1837      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1838      ecode++;      ecode++;
1839      break;      break;
1840    
1841        /* Multiline mode: start of subject unless notbol, or after any newline. */
1842    
1843        case OP_CIRCM:
1844        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1845        if (eptr != md->start_subject &&
1846            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1847          MRRETURN(MATCH_NOMATCH);
1848        ecode++;
1849        break;
1850    
1851      /* Start of match assertion */      /* Start of match assertion */
1852    
# Line 1665  for (;;) Line 1862  for (;;)
1862      ecode++;      ecode++;
1863      break;      break;
1864    
1865      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1866      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1867    
1868      case OP_DOLL:      case OP_DOLLM:
1869      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1870        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
       if (eptr < md->end_subject)  
         { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }  
       else  
         { if (md->noteol) MRRETURN(MATCH_NOMATCH); }  
       ecode++;  
       break;  
       }  
1871      else      else
1872        {        {
1873        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1874        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           MRRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1875        }        }
1876        ecode++;
1877        break;
1878    
1879        /* Not multiline mode: assert before a terminating newline or before end of
1880        subject unless noteol is set. */
1881    
1882        case OP_DOLL:
1883        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1884        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1885    
1886      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1887    
1888      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1889    
1890      case OP_EOD:      case OP_EOD:
1891      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1892        SCHECK_PARTIAL();
1893      ecode++;      ecode++;
1894      break;      break;
1895    
1896      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1897    
1898      case OP_EODN:      case OP_EODN:
1899      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1900        if (eptr < md->end_subject &&
1901          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1902        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1903    
1904        /* Either at end of string or \n before end. */
1905    
1906        SCHECK_PARTIAL();
1907      ecode++;      ecode++;
1908      break;      break;
1909    
# Line 1966  for (;;) Line 2165  for (;;)
2165      switch(c)      switch(c)
2166        {        {
2167        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2168    
2169        case 0x000d:        case 0x000d:
2170        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2171        break;        break;
# Line 2218  for (;;) Line 2418  for (;;)
2418      loops). */      loops). */
2419    
2420      case OP_REF:      case OP_REF:
2421        {      case OP_REFI:
2422        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2423        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2424        ecode += 3;
2425    
2426        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2427    
2428        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2429        than the amount of subject left; this ensures that every attempt at a      this ensures that every attempt at a match fails. We can't just fail
2430        match fails. We can't just fail here, because of the possibility of      here, because of the possibility of quantifiers with zero minima.
       quantifiers with zero minima.  
2431    
2432        (b) If the JavaScript compatibility flag is set, set the length to zero      (b) If the JavaScript compatibility flag is set, set the length to zero
2433        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2434    
2435        Otherwise, set the length to the length of what was matched by the      Otherwise, set the length to the length of what was matched by the
2436        referenced subpattern. */      referenced subpattern. */
2437    
2438        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2439          length = (md->jscript_compat)? 0 : (int)(md->end_subject - eptr + 1);        length = (md->jscript_compat)? 0 : -1;
2440        else      else
2441          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2442    
2443        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2444    
2445        switch (*ecode)      switch (*ecode)
2446          {        {
2447          case OP_CRSTAR:        case OP_CRSTAR:
2448          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2449          case OP_CRPLUS:        case OP_CRPLUS:
2450          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2451          case OP_CRQUERY:        case OP_CRQUERY:
2452          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2453          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2454          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2455          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2456          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2457          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2458          break;        break;
2459    
2460          case OP_CRRANGE:        case OP_CRRANGE:
2461          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2462          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2463          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2464          max = GET2(ecode, 3);        max = GET2(ecode, 3);
2465          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2466          ecode += 5;        ecode += 5;
2467          break;        break;
2468    
2469          default:               /* No repeat follows */        default:               /* No repeat follows */
2470          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2471            {          {
2472            CHECK_PARTIAL();          CHECK_PARTIAL();
2473            MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
           }  
         eptr += length;  
         continue;              /* With the main loop */  
2474          }          }
2475          eptr += length;
2476          continue;              /* With the main loop */
2477          }
2478    
2479        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2480        main loop. */      zero, just continue with the main loop. */
2481    
2482        if (length == 0) continue;      if (length == 0) continue;
2483    
2484        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2485        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2486        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2487    
2488        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2489          {
2490          int slength;
2491          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2492          {          {
2493          if (!match_ref(offset, eptr, length, md, ims))          CHECK_PARTIAL();
2494            {          MRRETURN(MATCH_NOMATCH);
           CHECK_PARTIAL();  
           MRRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2495          }          }
2496          eptr += slength;
2497          }
2498    
2499        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2500        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2501    
2502        if (min == max) continue;      if (min == max) continue;
2503    
2504        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2505    
2506        if (minimize)      if (minimize)
2507          {
2508          for (fi = min;; fi++)
2509          {          {
2510          for (fi = min;; fi++)          int slength;
2511            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2512            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2513            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2514            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2515            {            {
2516            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2517            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max) MRRETURN(MATCH_NOMATCH);  
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             MRRETURN(MATCH_NOMATCH);  
             }  
           eptr += length;  
2518            }            }
2519          /* Control never gets here */          eptr += slength;
2520          }          }
2521          /* Control never gets here */
2522          }
2523    
2524        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2525    
2526        else      else
2527          {
2528          pp = eptr;
2529          for (i = min; i < max; i++)
2530          {          {
2531          pp = eptr;          int slength;
2532          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
           {  
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             break;  
             }  
           eptr += length;  
           }  
         while (eptr >= pp)  
2533            {            {
2534            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2535            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2536            }            }
2537          MRRETURN(MATCH_NOMATCH);          eptr += slength;
2538          }          }
2539          while (eptr >= pp)
2540            {
2541            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2542            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2543            eptr -= length;
2544            }
2545          MRRETURN(MATCH_NOMATCH);
2546        }        }
2547      /* Control never gets here */      /* Control never gets here */
2548    
# Line 2446  for (;;) Line 2648  for (;;)
2648            {            {
2649            for (fi = min;; fi++)            for (fi = min;; fi++)
2650              {              {
2651              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2652              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2653              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2654              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2471  for (;;) Line 2673  for (;;)
2673            {            {
2674            for (fi = min;; fi++)            for (fi = min;; fi++)
2675              {              {
2676              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2677              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2678              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2679              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2517  for (;;) Line 2719  for (;;)
2719              }              }
2720            for (;;)            for (;;)
2721              {              {
2722              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2723              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2724              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2725              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2540  for (;;) Line 2742  for (;;)
2742              }              }
2743            while (eptr >= pp)            while (eptr >= pp)
2744              {              {
2745              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2746              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2747              eptr--;              eptr--;
2748              }              }
# Line 2616  for (;;) Line 2818  for (;;)
2818          {          {
2819          for (fi = min;; fi++)          for (fi = min;; fi++)
2820            {            {
2821            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2822            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2823            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2824            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2649  for (;;) Line 2851  for (;;)
2851            }            }
2852          for(;;)          for(;;)
2853            {            {
2854            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2855            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2856            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2857            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2694  for (;;) Line 2896  for (;;)
2896    
2897      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2898    
2899      case OP_CHARNC:      case OP_CHARI:
2900  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2901      if (utf8)      if (utf8)
2902        {        {
# Line 2754  for (;;) Line 2956  for (;;)
2956      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2957    
2958      case OP_EXACT:      case OP_EXACT:
2959        case OP_EXACTI:
2960      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2961      ecode += 3;      ecode += 3;
2962      goto REPEATCHAR;      goto REPEATCHAR;
2963    
2964      case OP_POSUPTO:      case OP_POSUPTO:
2965        case OP_POSUPTOI:
2966      possessive = TRUE;      possessive = TRUE;
2967      /* Fall through */      /* Fall through */
2968    
2969      case OP_UPTO:      case OP_UPTO:
2970        case OP_UPTOI:
2971      case OP_MINUPTO:      case OP_MINUPTO:
2972        case OP_MINUPTOI:
2973      min = 0;      min = 0;
2974      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2975      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2976      ecode += 3;      ecode += 3;
2977      goto REPEATCHAR;      goto REPEATCHAR;
2978    
2979      case OP_POSSTAR:      case OP_POSSTAR:
2980        case OP_POSSTARI:
2981      possessive = TRUE;      possessive = TRUE;
2982      min = 0;      min = 0;
2983      max = INT_MAX;      max = INT_MAX;
# Line 2778  for (;;) Line 2985  for (;;)
2985      goto REPEATCHAR;      goto REPEATCHAR;
2986    
2987      case OP_POSPLUS:      case OP_POSPLUS:
2988        case OP_POSPLUSI:
2989      possessive = TRUE;      possessive = TRUE;
2990      min = 1;      min = 1;
2991      max = INT_MAX;      max = INT_MAX;
# Line 2785  for (;;) Line 2993  for (;;)
2993      goto REPEATCHAR;      goto REPEATCHAR;
2994    
2995      case OP_POSQUERY:      case OP_POSQUERY:
2996        case OP_POSQUERYI:
2997      possessive = TRUE;      possessive = TRUE;
2998      min = 0;      min = 0;
2999      max = 1;      max = 1;
# Line 2792  for (;;) Line 3001  for (;;)
3001      goto REPEATCHAR;      goto REPEATCHAR;
3002    
3003      case OP_STAR:      case OP_STAR:
3004        case OP_STARI:
3005      case OP_MINSTAR:      case OP_MINSTAR:
3006        case OP_MINSTARI:
3007      case OP_PLUS:      case OP_PLUS:
3008        case OP_PLUSI:
3009      case OP_MINPLUS:      case OP_MINPLUS:
3010        case OP_MINPLUSI:
3011      case OP_QUERY:      case OP_QUERY:
3012        case OP_QUERYI:
3013      case OP_MINQUERY:      case OP_MINQUERY:
3014      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3015        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3016      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3017      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3018      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3019      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2822  for (;;) Line 3036  for (;;)
3036          {          {
3037  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3038          unsigned int othercase;          unsigned int othercase;
3039          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3040              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3041            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3042          else oclength = 0;          else oclength = 0;
# Line 2850  for (;;) Line 3064  for (;;)
3064            {            {
3065            for (fi = min;; fi++)            for (fi = min;; fi++)
3066              {              {
3067              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3068              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3069              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3070              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2892  for (;;) Line 3106  for (;;)
3106    
3107            for(;;)            for(;;)
3108              {              {
3109              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3110              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3111              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3112  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2929  for (;;) Line 3143  for (;;)
3143      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3144        max, eptr));        max, eptr));
3145    
3146      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3147        {        {
3148        fc = md->lcc[fc];        fc = md->lcc[fc];
3149        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2946  for (;;) Line 3160  for (;;)
3160          {          {
3161          for (fi = min;; fi++)          for (fi = min;; fi++)
3162            {            {
3163            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3164            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3165            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3166            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2976  for (;;) Line 3190  for (;;)
3190    
3191          while (eptr >= pp)          while (eptr >= pp)
3192            {            {
3193            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3194            eptr--;            eptr--;
3195            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3196            }            }
# Line 3005  for (;;) Line 3219  for (;;)
3219          {          {
3220          for (fi = min;; fi++)          for (fi = min;; fi++)
3221            {            {
3222            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3223            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3224            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3225            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3034  for (;;) Line 3248  for (;;)
3248    
3249          while (eptr >= pp)          while (eptr >= pp)
3250            {            {
3251            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3252            eptr--;            eptr--;
3253            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3254            }            }
# Line 3047  for (;;) Line 3261  for (;;)
3261      checking can be multibyte. */      checking can be multibyte. */
3262    
3263      case OP_NOT:      case OP_NOT:
3264        case OP_NOTI:
3265      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3266        {        {
3267        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 3054  for (;;) Line 3269  for (;;)
3269        }        }
3270      ecode++;      ecode++;
3271      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3272      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3273        {        {
3274  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3275        if (c < 256)        if (c < 256)
# Line 3062  for (;;) Line 3277  for (;;)
3277        c = md->lcc[c];        c = md->lcc[c];
3278        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3279        }        }
3280      else      else    /* Caseful */
3281        {        {
3282        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3283        }        }
# Line 3076  for (;;) Line 3291  for (;;)
3291      about... */      about... */
3292    
3293      case OP_NOTEXACT:      case OP_NOTEXACT:
3294        case OP_NOTEXACTI:
3295      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3296      ecode += 3;      ecode += 3;
3297      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3298    
3299      case OP_NOTUPTO:      case OP_NOTUPTO:
3300        case OP_NOTUPTOI:
3301      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3302        case OP_NOTMINUPTOI:
3303      min = 0;      min = 0;
3304      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3305      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3306      ecode += 3;      ecode += 3;
3307      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3308    
3309      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3310        case OP_NOTPOSSTARI:
3311      possessive = TRUE;      possessive = TRUE;
3312      min = 0;      min = 0;
3313      max = INT_MAX;      max = INT_MAX;
# Line 3096  for (;;) Line 3315  for (;;)
3315      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3316    
3317      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3318        case OP_NOTPOSPLUSI:
3319      possessive = TRUE;      possessive = TRUE;
3320      min = 1;      min = 1;
3321      max = INT_MAX;      max = INT_MAX;
# Line 3103  for (;;) Line 3323  for (;;)
3323      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3324    
3325      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3326        case OP_NOTPOSQUERYI:
3327      possessive = TRUE;      possessive = TRUE;
3328      min = 0;      min = 0;
3329      max = 1;      max = 1;
# Line 3110  for (;;) Line 3331  for (;;)
3331      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3332    
3333      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3334        case OP_NOTPOSUPTOI:
3335      possessive = TRUE;      possessive = TRUE;
3336      min = 0;      min = 0;
3337      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 3117  for (;;) Line 3339  for (;;)
3339      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3340    
3341      case OP_NOTSTAR:      case OP_NOTSTAR:
3342        case OP_NOTSTARI:
3343      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3344        case OP_NOTMINSTARI:
3345      case OP_NOTPLUS:      case OP_NOTPLUS:
3346        case OP_NOTPLUSI:
3347      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3348        case OP_NOTMINPLUSI:
3349      case OP_NOTQUERY:      case OP_NOTQUERY:
3350        case OP_NOTQUERYI:
3351      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3352      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3353        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3354      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3355      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3356      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3144  for (;;) Line 3372  for (;;)
3372      DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3373        max, eptr));        max, eptr));
3374    
3375      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3376        {        {
3377        fc = md->lcc[fc];        fc = md->lcc[fc];
3378    
# Line 3192  for (;;) Line 3420  for (;;)
3420            register unsigned int d;            register unsigned int d;
3421            for (fi = min;; fi++)            for (fi = min;; fi++)
3422              {              {
3423              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3424              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3425              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3426              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3211  for (;;) Line 3439  for (;;)
3439            {            {
3440            for (fi = min;; fi++)            for (fi = min;; fi++)
3441              {              {
3442              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3443              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3444              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3445              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3252  for (;;) Line 3480  for (;;)
3480          if (possessive) continue;          if (possessive) continue;
3481          for(;;)          for(;;)
3482              {              {
3483              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3484              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3485              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3486              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3275  for (;;) Line 3503  for (;;)
3503            if (possessive) continue;            if (possessive) continue;
3504            while (eptr >= pp)            while (eptr >= pp)
3505              {              {
3506              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3507              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3508              eptr--;              eptr--;
3509              }              }
# Line 3332  for (;;) Line 3560  for (;;)
3560            register unsigned int d;            register unsigned int d;
3561            for (fi = min;; fi++)            for (fi = min;; fi++)
3562              {              {
3563              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3564              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3565              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3566              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3350  for (;;) Line 3578  for (;;)
3578            {            {
3579            for (fi = min;; fi++)            for (fi = min;; fi++)
3580              {              {
3581              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3582              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3583              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3584              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3390  for (;;) Line 3618  for (;;)
3618            if (possessive) continue;            if (possessive) continue;
3619            for(;;)            for(;;)
3620              {              {
3621              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3622              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3623              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3624              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3413  for (;;) Line 3641  for (;;)
3641            if (possessive) continue;            if (possessive) continue;
3642            while (eptr >= pp)            while (eptr >= pp)
3643              {              {
3644              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3645              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3646              eptr--;              eptr--;
3647              }              }
# Line 3740  for (;;) Line 3968  for (;;)
3968            switch(c)            switch(c)
3969              {              {
3970              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3971    
3972              case 0x000d:              case 0x000d:
3973              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3974              break;              break;
# Line 4016  for (;;) Line 4245  for (;;)
4245            switch(*eptr++)            switch(*eptr++)
4246              {              {
4247              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4248    
4249              case 0x000d:              case 0x000d:
4250              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4251              break;              break;
4252    
4253              case 0x000a:              case 0x000a:
4254              break;              break;
4255    
# Line 4208  for (;;) Line 4439  for (;;)
4439            case PT_ANY:            case PT_ANY:
4440            for (fi = min;; fi++)            for (fi = min;; fi++)
4441              {              {
4442              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4443              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4444              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4445              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4224  for (;;) Line 4455  for (;;)
4455            case PT_LAMP:            case PT_LAMP:
4456            for (fi = min;; fi++)            for (fi = min;; fi++)
4457              {              {
4458              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4459              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4460              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4461              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4244  for (;;) Line 4475  for (;;)
4475            case PT_GC:            case PT_GC:
4476            for (fi = min;; fi++)            for (fi = min;; fi++)
4477              {              {
4478              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4479              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4480              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4481              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4262  for (;;) Line 4493  for (;;)
4493            case PT_PC:            case PT_PC:
4494            for (fi = min;; fi++)            for (fi = min;; fi++)
4495              {              {
4496              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
4497              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4498              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4499              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4280  for (;;) Line 4511  for (;;)
4511            case PT_SC:            case PT_SC:
4512            for (fi = min;; fi++)            for (fi = min;; fi++)
4513              {              {
4514              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
4515              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4516              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4517              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4298  for (;;) Line 4529  for (;;)
4529            case PT_ALNUM:            case PT_ALNUM:
4530            for (fi = min;; fi++)            for (fi = min;; fi++)
4531              {              {
4532              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4533              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4534              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4535              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4317  for (;;) Line 4548  for (;;)
4548            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4549            for (fi = min;; fi++)            for (fi = min;; fi++)
4550              {              {
4551              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4552              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4553              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4554              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4337  for (;;) Line 4568  for (;;)
4568            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4569            for (fi = min;; fi++)            for (fi = min;; fi++)
4570              {              {
4571              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4572              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4573              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4574              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4357  for (;;) Line 4588  for (;;)
4588            case PT_WORD:            case PT_WORD:
4589            for (fi = min;; fi++)            for (fi = min;; fi++)
4590              {              {
4591              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4592              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4593              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4594              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4389  for (;;) Line 4620  for (;;)
4620          {          {
4621          for (fi = min;; fi++)          for (fi = min;; fi++)
4622            {            {
4623            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4624            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4625            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4626            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4421  for (;;) Line 4652  for (;;)
4652          {          {
4653          for (fi = min;; fi++)          for (fi = min;; fi++)
4654            {            {
4655            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4656            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4657            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4658            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4584  for (;;) Line 4815  for (;;)
4815          {          {
4816          for (fi = min;; fi++)          for (fi = min;; fi++)
4817            {            {
4818            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4819            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4820            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4821            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4882  for (;;) Line 5113  for (;;)
5113          if (possessive) continue;          if (possessive) continue;
5114          for(;;)          for(;;)
5115            {            {
5116            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5117            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5118            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5119            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4923  for (;;) Line 5154  for (;;)
5154    
5155          for(;;)          for(;;)
5156            {            {
5157            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5158            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5159            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5160            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
# Line 5207  for (;;) Line 5438  for (;;)
5438            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5439            }            }
5440    
5441          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5442            done (no backing up). Otherwise, match at this position; anything other
5443            than no match is immediately returned. For nomatch, back up one
5444            character, unless we are matching \R and the last thing matched was
5445            \r\n, in which case, back up two bytes. */
5446    
5447          if (possessive) continue;          if (possessive) continue;
5448          for(;;)          for(;;)
5449            {            {
5450            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5451            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5452            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5453            BACKCHAR(eptr);            BACKCHAR(eptr);
5454              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5455                  eptr[-1] == '\r') eptr--;
5456            }            }
5457          }          }
5458        else        else
# Line 5414  for (;;) Line 5651  for (;;)
5651            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5652            }            }
5653    
5654          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5655            done (no backing up). Otherwise, match at this position; anything other
5656            than no match is immediately returned. For nomatch, back up one
5657            character (byte), unless we are matching \R and the last thing matched
5658            was \r\n, in which case, back up two bytes. */
5659    
5660          if (possessive) continue;          if (possessive) continue;
5661          while (eptr >= pp)          while (eptr >= pp)
5662            {            {
5663            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5664            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5665              eptr--;
5666              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5667                  eptr[-1] == '\r') eptr--;
5668            }            }
5669          }          }
5670    
# Line 5460  switch (frame->Xwhere) Line 5703  switch (frame->Xwhere)
5703    LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)    LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)
5704    LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)    LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)
5705    LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)    LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)
5706    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64)
5707  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5708    LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)    LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)
5709    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
# Line 5489  Undefine all the macros that were define Line 5732  Undefine all the macros that were define
5732  #undef ecode  #undef ecode
5733  #undef mstart  #undef mstart
5734  #undef offset_top  #undef offset_top
 #undef ims  
5735  #undef eptrb  #undef eptrb
5736  #undef flags  #undef flags
5737    
# Line 5507  Undefine all the macros that were define Line 5749  Undefine all the macros that were define
5749  #undef condition  #undef condition
5750  #undef prev_is_word  #undef prev_is_word
5751    
 #undef original_ims  
   
5752  #undef ctype  #undef ctype
5753  #undef length  #undef length
5754  #undef max  #undef max
# Line 5570  int first_byte = -1; Line 5810  int first_byte = -1;
5810  int req_byte = -1;  int req_byte = -1;
5811  int req_byte2 = -1;  int req_byte2 = -1;
5812  int newline;  int newline;
 unsigned long int ims;  
5813  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5814  BOOL anchored;  BOOL anchored;
5815  BOOL startline;  BOOL startline;
# Line 5600  if ((options & ~PUBLIC_EXEC_OPTIONS) != Line 5839  if ((options & ~PUBLIC_EXEC_OPTIONS) !=
5839  if (re == NULL || subject == NULL ||  if (re == NULL || subject == NULL ||
5840     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;
5841  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
5842    if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
5843    
5844  /* This information is for finding all the numbers associated with a given  /* This information is for finding all the numbers associated with a given
5845  name, for condition testing. */  name, for condition testing. */
# Line 5760  defined (though never set). So there's n Line 6000  defined (though never set). So there's n
6000  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
6001    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
6002    
6003  /* Check a UTF-8 string if required. Unfortunately there's no way of passing  /* Check a UTF-8 string if required. Pass back the character offset and error
6004  back the character offset. */  code for an invalid string if a results vector is available. */
6005    
6006  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
6007  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
6008    {    {
6009    if (_pcre_valid_utf8((USPTR)subject, length) >= 0)    int erroroffset;
6010      return PCRE_ERROR_BADUTF8;    int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
6011    if (start_offset > 0 && start_offset < length)    if (errorcode != 0)
6012      {      {
6013      int tb = ((USPTR)subject)[start_offset];      if (offsetcount >= 2)
     if (tb > 127)  
6014        {        {
6015        tb &= 0xc0;        offsets[0] = erroroffset;
6016        if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET;        offsets[1] = errorcode;
6017        }        }
6018      }      return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
6019          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
6020        }
6021    
6022      /* Check that a start_offset points to the start of a UTF-8 character. */
6023    
6024      if (start_offset > 0 && start_offset < length &&
6025          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
6026        return PCRE_ERROR_BADUTF8_OFFSET;
6027    }    }
6028  #endif  #endif
6029    
 /* The ims options can vary during the matching as a result of the presence  
 of (?ims) items in the pattern. They are kept in a local variable so that  
 restoring at the exit of a group is easy. */  
   
 ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL);  
   
6030  /* If the expression has got more back references than the offsets supplied can  /* If the expression has got more back references than the offsets supplied can
6031  hold, we get a temporary chunk of working store to use during the matching.  hold, we get a temporary chunk of working store to use during the matching.
6032  Otherwise, we can use the vector supplied, rounding down its size to a multiple  Otherwise, we can use the vector supplied, rounding down its size to a multiple
# Line 5903  for(;;) Line 6144  for(;;)
6144    /* There are some optimizations that avoid running the match if a known    /* There are some optimizations that avoid running the match if a known
6145    starting point is not found, or if a known later character is not present.    starting point is not found, or if a known later character is not present.
6146    However, there is an option that disables these, for testing and for ensuring    However, there is an option that disables these, for testing and for ensuring
6147    that all callouts do actually occur. */    that all callouts do actually occur. The option can be set in the regex by
6148      (*NO_START_OPT) or passed in match-time options. */
6149    
6150    if ((options & PCRE_NO_START_OPTIMIZE) == 0)    if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0)
6151      {      {
6152      /* Advance to a unique first byte if there is one. */      /* Advance to a unique first byte if there is one. */
6153    
# Line 5959  for(;;) Line 6201  for(;;)
6201        while (start_match < end_subject)        while (start_match < end_subject)
6202          {          {
6203          register unsigned int c = *start_match;          register unsigned int c = *start_match;
6204          if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++;          if ((start_bits[c/8] & (1 << (c&7))) == 0)
6205            else break;            {
6206              start_match++;
6207    #ifdef SUPPORT_UTF8
6208              if (utf8)
6209                while(start_match < end_subject && (*start_match & 0xc0) == 0x80)
6210                  start_match++;
6211    #endif
6212              }
6213            else break;
6214          }          }
6215        }        }
6216      }   /* Starting optimizations */      }   /* Starting optimizations */
# Line 6055  for(;;) Line 6305  for(;;)
6305    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6306    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6307    md->match_call_count = 0;    md->match_call_count = 0;
6308    rc = match(start_match, md->start_code, start_match, NULL, 2, md, ims, NULL,    md->match_function_type = 0;
6309      0, 0);    rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6310    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
6311    
6312    switch(rc)    switch(rc)
6313      {      {
6314      /* NOMATCH and PRUNE advance by one character. If MATCH_SKIP_ARG reaches      /* SKIP passes back the next starting point explicitly, but if it is the
6315      this level it means that a MARK that matched the SKIP's arg was not found.      same as the match we have just done, treat it as NOMATCH. */
6316      We treat this as NOMATCH. THEN at this level acts exactly like PRUNE. */  
6317        case MATCH_SKIP:
6318        if (md->start_match_ptr != start_match)
6319          {
6320          new_start_match = md->start_match_ptr;
6321          break;
6322          }
6323        /* Fall through */
6324    
6325        /* If MATCH_SKIP_ARG reaches this level it means that a MARK that matched
6326        the SKIP's arg was not found. We also treat this as NOMATCH. */
6327    
6328        case MATCH_SKIP_ARG:
6329        /* Fall through */
6330    
6331        /* NOMATCH and PRUNE advance by one character. THEN at this level acts
6332        exactly like PRUNE. */
6333    
6334      case MATCH_NOMATCH:      case MATCH_NOMATCH:
6335      case MATCH_PRUNE:      case MATCH_PRUNE:
     case MATCH_SKIP_ARG:  
6336      case MATCH_THEN:      case MATCH_THEN:
6337      new_start_match = start_match + 1;      new_start_match = start_match + 1;
6338  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 6077  for(;;) Line 6342  for(;;)
6342  #endif  #endif
6343      break;      break;
6344    
     /* SKIP passes back the next starting point explicitly. */  
   
     case MATCH_SKIP:  
     new_start_match = md->start_match_ptr;  
     break;  
   
6345      /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */      /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */
6346    
6347      case MATCH_COMMIT:      case MATCH_COMMIT:
# Line 6170  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6429  if (rc == MATCH_MATCH || rc == MATCH_ACC
6429    
6430    /* Set the return code to the number of captured strings, or 0 if there are    /* Set the return code to the number of captured strings, or 0 if there are
6431    too many to fit into the vector. */    too many to fit into the vector. */
6432    
6433    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = md->offset_overflow? 0 : md->end_offset_top/2;
6434    
6435    /* If there is space, set up the whole thing as substring 0. The value of    /* If there is space, set up the whole thing as substring 0. The value of

Legend:
Removed from v.534  
changed lines
  Added in v.608

  ViewVC Help
Powered by ViewVC 1.1.5