/[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 511 by ph10, Mon Mar 29 09:25:38 2010 UTC revision 645 by ph10, Sun Jul 31 17:02:18 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_ONCE         (-996)
80  #define MATCH_SKIP_ARG     (-995)  #define MATCH_PRUNE        (-995)
81  #define MATCH_THEN         (-994)  #define MATCH_SKIP         (-994)
82    #define MATCH_SKIP_ARG     (-993)
83    #define MATCH_THEN         (-992)
84    
85  /* This is a convenience macro for code that occurs many times. */  /* This is a convenience macro for code that occurs many times. */
86    
# Line 132  while (length-- > 0) Line 136  while (length-- > 0)
136  *          Match a back-reference                *  *          Match a back-reference                *
137  *************************************************/  *************************************************/
138    
139  /* 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
140  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
141    the length passed is zero. Note that in caseless UTF-8 mode, the number of
142    subject bytes matched may be different to the number of reference bytes.
143    
144  Arguments:  Arguments:
145    offset      index into the offset vector    offset      index into the offset vector
146    eptr        points into the subject    eptr        pointer into the subject
147    length      length to be matched    length      length of reference to be matched (number of bytes)
148    md          points to match data block    md          points to match data block
149    ims         the ims flags    caseless    TRUE if caseless
150    
151  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
152  */  */
153    
154  static BOOL  static int
155  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
156    unsigned long int ims)    BOOL caseless)
157  {  {
158  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
159    register USPTR p = md->start_subject + md->offset_vector[offset];
160    
161  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
162  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 164  pchars(p, length, FALSE, md); Line 171  pchars(p, length, FALSE, md);
171  printf("\n");  printf("\n");
172  #endif  #endif
173    
174  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
175    
176  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
177    
178  /* 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
179  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
180  ASCII characters. */  ASCII characters. */
181    
182  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
183    {    {
184  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
185  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
186    if (md->utf8)    if (md->utf8)
187      {      {
188      USPTR endptr = eptr + length;      /* Match characters up to the end of the reference. NOTE: the number of
189      while (eptr < endptr)      bytes matched may differ, because there are some characters whose upper and
190        lower case versions code as different numbers of bytes. For example, U+023A
191        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
192        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
193        the latter. It is important, therefore, to check the length along the
194        reference, not along the subject (earlier code did this wrong). */
195    
196        USPTR endptr = p + length;
197        while (p < endptr)
198        {        {
199        int c, d;        int c, d;
200          if (eptr >= md->end_subject) return -1;
201        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
202        GETCHARINC(d, p);        GETCHARINC(d, p);
203        if (c != d && c != UCD_OTHERCASE(d)) return FALSE;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
204        }        }
205      }      }
206    else    else
# Line 193  if ((ims & PCRE_CASELESS) != 0) Line 209  if ((ims & PCRE_CASELESS) != 0)
209    
210    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
211    is no UCP support. */    is no UCP support. */
212        {
213    while (length-- > 0)      if (eptr + length > md->end_subject) return -1;
214      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }      while (length-- > 0)
215          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
216        }
217    }    }
218    
219  /* In the caseful case, we can just compare the bytes, whether or not we  /* In the caseful case, we can just compare the bytes, whether or not we
220  are in UTF-8 mode. */  are in UTF-8 mode. */
221    
222  else  else
223    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
224      if (eptr + length > md->end_subject) return -1;
225      while (length-- > 0) if (*p++ != *eptr++) return -1;
226      }
227    
228  return TRUE;  return eptr - eptr_start;
229  }  }
230    
231    
# Line 255  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 276  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
276         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
277         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
278         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
279         RM51,  RM52, RM53, RM54 };         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280           RM61,  RM62, RM63 };
281    
282  /* 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
283  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 265  actually used in this definition. */ Line 287  actually used in this definition. */
287  #define REGISTER register  #define REGISTER register
288    
289  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
290  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
291    { \    { \
292    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
293    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
294    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
295    }    }
296  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 277  actually used in this definition. */ Line 299  actually used in this definition. */
299    return ra; \    return ra; \
300    }    }
301  #else  #else
302  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
303    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
304  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
305  #endif  #endif
306    
# Line 291  argument of match(), which never changes Line 313  argument of match(), which never changes
313    
314  #define REGISTER  #define REGISTER
315    
316  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
317    {\    {\
318    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
319      if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
320    frame->Xwhere = rw; \    frame->Xwhere = rw; \
321    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
322    newframe->Xecode = rb;\    newframe->Xecode = rb;\
323    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
324    newframe->Xmarkptr = markptr;\    newframe->Xmarkptr = markptr;\
325    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
326    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
327    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
328    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
329    frame = newframe;\    frame = newframe;\
# Line 314  argument of match(), which never changes Line 335  argument of match(), which never changes
335    
336  #define RRETURN(ra)\  #define RRETURN(ra)\
337    {\    {\
338    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
339    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
340    (pcre_stack_free)(newframe);\    (pcre_stack_free)(oldframe);\
341    if (frame != NULL)\    if (frame != NULL)\
342      {\      {\
343      rrc = ra;\      rrc = ra;\
# Line 338  typedef struct heapframe { Line 359  typedef struct heapframe {
359    USPTR Xmstart;    USPTR Xmstart;
360    USPTR Xmarkptr;    USPTR Xmarkptr;
361    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
362    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
363    unsigned int Xrdepth;    unsigned int Xrdepth;
364    
365    /* Function local variables */    /* Function local variables */
# Line 361  typedef struct heapframe { Line 380  typedef struct heapframe {
380    BOOL Xcondition;    BOOL Xcondition;
381    BOOL Xprev_is_word;    BOOL Xprev_is_word;
382    
   unsigned long int Xoriginal_ims;  
   
383  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
386    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
387    int Xoclength;    int Xoclength;
388    uschar Xocchars[8];    uschar Xocchars[8];
389  #endif  #endif
# Line 420  immediately. The second one is used when Line 434  immediately. The second one is used when
434  the subject. */  the subject. */
435    
436  #define CHECK_PARTIAL()\  #define CHECK_PARTIAL()\
437    if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\    if (md->partial != 0 && eptr >= md->end_subject && \
438      {\        eptr > md->start_used_ptr) \
439      md->hitend = TRUE;\      { \
440      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\      md->hitend = TRUE; \
441        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
442      }      }
443    
444  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
445    if (md->partial != 0 && eptr > mstart)\    if (md->partial != 0 && eptr > md->start_used_ptr) \
446      {\      { \
447      md->hitend = TRUE;\      md->hitend = TRUE; \
448      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
449      }      }
450    
451    
# Line 447  Arguments: Line 462  Arguments:
462     markptr     pointer to the most recent MARK name, or NULL     markptr     pointer to the most recent MARK name, or NULL
463     offset_top  current top pointer     offset_top  current top pointer
464     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
465     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
466                   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  
467     rdepth      the recursion depth     rdepth      the recursion depth
468    
469  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
# Line 465  Returns:       MATCH_MATCH if matched Line 475  Returns:       MATCH_MATCH if matched
475    
476  static int  static int
477  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
478    const uschar *markptr, int offset_top, match_data *md, unsigned long int ims,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
479    eptrblock *eptrb, int flags, unsigned int rdepth)    unsigned int rdepth)
480  {  {
481  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
482  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 478  register unsigned int c;   /* Character Line 488  register unsigned int c;   /* Character
488  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
489    
490  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
491    BOOL caseless;
492  int condcode;  int condcode;
493    
494  /* 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 486  heap storage. Set up the top-level frame Line 497  heap storage. Set up the top-level frame
497  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
498    
499  #ifdef NO_RECURSE  #ifdef NO_RECURSE
500  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
501    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
502  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
503    
504  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 496  frame->Xecode = ecode; Line 508  frame->Xecode = ecode;
508  frame->Xmstart = mstart;  frame->Xmstart = mstart;
509  frame->Xmarkptr = markptr;  frame->Xmarkptr = markptr;
510  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
511  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
512  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
513    
514  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 512  HEAP_RECURSE: Line 522  HEAP_RECURSE:
522  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
523  #define markptr            frame->Xmarkptr  #define markptr            frame->Xmarkptr
524  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
525  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
526  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
527    
528  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 536  HEAP_RECURSE: Line 544  HEAP_RECURSE:
544  #define condition          frame->Xcondition  #define condition          frame->Xcondition
545  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
546    
 #define original_ims       frame->Xoriginal_ims  
   
547  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
548  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
549  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
550  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
 #define prop_category      frame->Xprop_category  
 #define prop_chartype      frame->Xprop_chartype  
 #define prop_script        frame->Xprop_script  
551  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
552  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
553  #endif  #endif
# Line 574  i, and fc and c, can be the same variabl Line 577  i, and fc and c, can be the same variabl
577  #define fi i  #define fi i
578  #define fc c  #define fc c
579    
580    /* Many of the following variables are used only in small blocks of the code.
581    My normal style of coding would have declared them within each of those blocks.
582    However, in order to accommodate the version of this code that uses an external
583    "stack" implemented on the heap, it is easier to declare them all here, so the
584    declarations can be cut out in a block. The only declarations within blocks
585    below are for variables that do not have to be preserved over a recursive call
586    to RMATCH(). */
587    
588    #ifdef SUPPORT_UTF8
589    const uschar *charptr;
590    #endif
591    const uschar *callpat;
592    const uschar *data;
593    const uschar *next;
594    USPTR         pp;
595    const uschar *prev;
596    USPTR         saved_eptr;
597    
598    recursion_info new_recursive;
599    
600  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  BOOL cur_is_word;
 const uschar *charptr;             /* in small blocks of the code. My normal */  
 #endif                             /* style of coding would have declared    */  
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
 unsigned long int original_ims;  
   
604  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
608  int oclength;  int oclength;
609  uschar occhars[8];  uschar occhars[8];
610  #endif  #endif
# Line 619  int stacksave[REC_STACK_SAVE_MAX]; Line 624  int stacksave[REC_STACK_SAVE_MAX];
624  eptrblock newptrb;  eptrblock newptrb;
625  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
626    
627    /* To save space on the stack and in the heap frame, I have doubled up on some
628    of the local variables that are used only in localised parts of the code, but
629    still need to be preserved over recursive calls of match(). These macros define
630    the alternative names that are used. */
631    
632    #define allow_zero    cur_is_word
633    #define cbegroup      condition
634    #define code_offset   codelink
635    #define condassert    condition
636    #define matched_once  prev_is_word
637    
638  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
639  variables. */  variables. */
640    
# Line 655  haven't exceeded the recursive call limi Line 671  haven't exceeded the recursive call limi
671  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
672  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
673    
 original_ims = ims;    /* Save for resetting on ')' */  
   
674  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
675  string, the 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
676  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
677  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
678  When match() is called in other circumstances, don't add to the chain. The  
679  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
680  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
681  match(). */  to break infinite loops that match no characters. When match() is called in
682    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
683    NOT be used with tail recursion, because the memory block that is used is on
684    the stack, so a new one may be required for each match(). */
685    
686  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
687    {    {
688    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
689    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
690    eptrb = &newptrb;    eptrb = &newptrb;
691      md->match_function_type = 0;
692    }    }
693    
694  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 685  for (;;) Line 703  for (;;)
703      case OP_MARK:      case OP_MARK:
704      markptr = ecode + 2;      markptr = ecode + 2;
705      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
706        ims, eptrb, flags, RM51);        eptrb, RM55);
707    
708      /* 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
709      argument, and we must check whether that argument matches this MARK's      argument, and we must check whether that argument matches this MARK's
710      argument. It is passed back in md->start_match_ptr (an overloading of that      argument. It is passed back in md->start_match_ptr (an overloading of that
711      variable). If it does match, we reset that variable to the current subject      variable). If it does match, we reset that variable to the current subject
712      position and return MATCH_SKIP. Otherwise, pass back the return code      position and return MATCH_SKIP. Otherwise, pass back the return code
713      unaltered. */      unaltered. */
714    
715      if (rrc == MATCH_SKIP_ARG &&      if (rrc == MATCH_SKIP_ARG &&
716          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
717        {        {
718        md->start_match_ptr = eptr;        md->start_match_ptr = eptr;
719        RRETURN(MATCH_SKIP);        RRETURN(MATCH_SKIP);
720        }        }
721    
722      if (md->mark == NULL) md->mark = markptr;      if (md->mark == NULL) md->mark = markptr;
723      RRETURN(rrc);      RRETURN(rrc);
724    
725      case OP_FAIL:      case OP_FAIL:
726      MRRETURN(MATCH_NOMATCH);      MRRETURN(MATCH_NOMATCH);
727    
728        /* COMMIT overrides PRUNE, SKIP, and THEN */
729    
730      case OP_COMMIT:      case OP_COMMIT:
731      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
732        ims, eptrb, flags, RM52);        eptrb, RM52);
733      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
734            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
735            rrc != MATCH_THEN)
736          RRETURN(rrc);
737      MRRETURN(MATCH_COMMIT);      MRRETURN(MATCH_COMMIT);
738    
739        /* PRUNE overrides THEN */
740    
741      case OP_PRUNE:      case OP_PRUNE:
742      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
743        ims, eptrb, flags, RM51);        eptrb, RM51);
744      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
745      MRRETURN(MATCH_PRUNE);      MRRETURN(MATCH_PRUNE);
746    
747      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
748      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
749        ims, eptrb, flags, RM51);        eptrb, RM56);
750      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
751      md->mark = ecode + 2;      md->mark = ecode + 2;
752      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
753    
754        /* SKIP overrides PRUNE and THEN */
755    
756      case OP_SKIP:      case OP_SKIP:
757      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
758        ims, eptrb, flags, RM53);        eptrb, RM53);
759      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
760          RRETURN(rrc);
761      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
762      MRRETURN(MATCH_SKIP);      MRRETURN(MATCH_SKIP);
763    
764      case OP_SKIP_ARG:      case OP_SKIP_ARG:
765      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
766        ims, eptrb, flags, RM53);        eptrb, RM57);
767      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769      /* Pass back the current skip name by overloading md->start_match_ptr and  
770      returning the special MATCH_SKIP_ARG return code. This will either be      /* Pass back the current skip name by overloading md->start_match_ptr and
771      caught by a matching MARK, or get to the top, where it is treated the same      returning the special MATCH_SKIP_ARG return code. This will either be
772        caught by a matching MARK, or get to the top, where it is treated the same
773      as PRUNE. */      as PRUNE. */
774    
775      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
776      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
777    
778        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
779        the alt that is at the start of the current branch. This makes it possible
780        to skip back past alternatives that precede the THEN within the current
781        branch. */
782    
783      case OP_THEN:      case OP_THEN:
784      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
785        ims, eptrb, flags, RM54);        eptrb, RM54);
786      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
787        md->start_match_ptr = ecode - GET(ecode, 1);
788      MRRETURN(MATCH_THEN);      MRRETURN(MATCH_THEN);
789    
790      case OP_THEN_ARG:      case OP_THEN_ARG:
791      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
792        ims, eptrb, flags, RM54);        offset_top, md, eptrb, RM58);
793      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
794      md->mark = ecode + 2;      md->start_match_ptr = ecode - GET(ecode, 1);
795        md->mark = ecode + LINK_SIZE + 2;
796      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
797    
798      /* 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
799      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
800      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
801      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
802      reference inside the group.      previous iteration of this group, and be referred to by a reference inside
803        the group. A failure to match might occur after the group has succeeded,
804      If the bracket fails to match, we need to restore this value and also the      if something later on doesn't match. For this reason, we need to restore
805      values of the final offsets, in case they were set by a previous iteration      the working value and also the values of the final offsets, in case they
806      of the same bracket.      were set by a previous iteration of the same bracket.
807    
808      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
809      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 793  for (;;) Line 829  for (;;)
829        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
830    
831        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
832        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
833            (int)(eptr - md->start_subject);
834    
835        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
836          {          {
837            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839            ims, eptrb, flags, RM1);            eptrb, RM1);
840          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841            if (rrc != MATCH_NOMATCH &&
842                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
843              RRETURN(rrc);
844          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
845          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
846            if (*ecode != OP_ALT) break;
847          }          }
       while (*ecode == OP_ALT);  
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854        if (rrc != MATCH_THEN) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855        RRETURN(MATCH_NOMATCH);        MATCH_THEN. */
856    
857          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
858          RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
859        }        }
860    
861      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 827  for (;;) Line 869  for (;;)
869      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
871    
872      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
873      final alternative within the brackets, we would return the result of a      repeat. Loop for all the alternatives. When we get to the final alternative
874      recursive call to match() whatever happened. We can reduce stack usage by      within the brackets, we used to return the result of a recursive call to
875      turning this into a tail recursion, except in the case when match_cbegroup      match() whatever happened so it was possible to reduce stack usage by
876      is set.*/      turning this into a tail recursion, except in the case of a possibly empty
877        group. However, now that there is the possiblity of (*THEN) occurring in
878        the final alternative, this optimization is no longer possible.
879    
880        MATCH_ONCE is returned when the end of an atomic group is successfully
881        reached, but subsequent matching fails. It passes back up the tree (causing
882        captured values to be reset) until the original atomic group level is
883        reached. This is tested by comparing md->once_target with the start of the
884        group. At this point, the return is converted into MATCH_NOMATCH so that
885        previous backup points can be taken. */
886    
887        case OP_ONCE:
888      case OP_BRA:      case OP_BRA:
889      case OP_SBRA:      case OP_SBRA:
890      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
891      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
892      for (;;)      for (;;)
893        {        {
894        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
895          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
896            RM2);
897          if (rrc != MATCH_NOMATCH &&
898              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
899          {          {
900          if (flags == 0)    /* Not a possibly empty group */          if (rrc == MATCH_ONCE)
901            {            {
902            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
903            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
904            goto TAIL_RECURSE;              {
905                while (*scode == OP_ALT) scode += GET(scode, 1);
906                scode -= GET(scode, 1);
907                }
908              if (md->once_target == scode) rrc = MATCH_NOMATCH;
909            }            }
910            RRETURN(rrc);
911            }
912          ecode += GET(ecode, 1);
913          if (*ecode != OP_ALT) break;
914          }
915        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
916        RRETURN(MATCH_NOMATCH);
917    
918          /* Possibly empty group; can't use tail recursion. */      /* Handle possessive capturing brackets with an unlimited repeat. We come
919        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
920        handled similarly to the normal case above. However, the matching is
921        different. The end of these brackets will always be OP_KETRPOS, which
922        returns MATCH_KETRPOS without going further in the pattern. By this means
923        we can handle the group by iteration rather than recursion, thereby
924        reducing the amount of stack needed. */
925    
926        case OP_CBRAPOS:
927        case OP_SCBRAPOS:
928        allow_zero = FALSE;
929    
930          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      POSSESSIVE_CAPTURE:
931            eptrb, flags, RM48);      number = GET2(ecode, 1+LINK_SIZE);
932          if (rrc == MATCH_NOMATCH) md->mark = markptr;      offset = number << 1;
933          RRETURN(rrc);  
934    #ifdef PCRE_DEBUG
935        printf("start possessive bracket %d\n", number);
936        printf("subject=");
937        pchars(eptr, 16, TRUE, md);
938        printf("\n");
939    #endif
940    
941        if (offset < md->offset_max)
942          {
943          matched_once = FALSE;
944          code_offset = ecode - md->start_code;
945    
946          save_offset1 = md->offset_vector[offset];
947          save_offset2 = md->offset_vector[offset+1];
948          save_offset3 = md->offset_vector[md->offset_end - number];
949          save_capture_last = md->capture_last;
950    
951          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
952    
953          /* Each time round the loop, save the current subject position for use
954          when the group matches. For MATCH_MATCH, the group has matched, so we
955          restart it with a new subject starting position, remembering that we had
956          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
957          usual. If we haven't matched any alternatives in any iteration, check to
958          see if a previous iteration matched. If so, the group has matched;
959          continue from afterwards. Otherwise it has failed; restore the previous
960          capture values before returning NOMATCH. */
961    
962          for (;;)
963            {
964            md->offset_vector[md->offset_end - number] =
965              (int)(eptr - md->start_subject);
966            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
967            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
968              eptrb, RM63);
969            if (rrc == MATCH_KETRPOS)
970              {
971              offset_top = md->end_offset_top;
972              eptr = md->end_match_ptr;
973              ecode = md->start_code + code_offset;
974              save_capture_last = md->capture_last;
975              matched_once = TRUE;
976              continue;
977              }
978            if (rrc != MATCH_NOMATCH &&
979                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
980              RRETURN(rrc);
981            md->capture_last = save_capture_last;
982            ecode += GET(ecode, 1);
983            if (*ecode != OP_ALT) break;
984            }
985    
986          if (!matched_once)
987            {
988            md->offset_vector[offset] = save_offset1;
989            md->offset_vector[offset+1] = save_offset2;
990            md->offset_vector[md->offset_end - number] = save_offset3;
991            }
992    
993          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
994          if (allow_zero || matched_once)
995            {
996            ecode += 1 + LINK_SIZE;
997            break;
998          }          }
999    
1000        /* For non-final alternatives, continue the loop for a NOMATCH result;        RRETURN(MATCH_NOMATCH);
1001        otherwise return. */        }
1002    
1003        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1004        as a non-capturing bracket. */
1005    
1006        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1007        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008    
1009        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1010    
1011        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1012        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1013    
1014        /* Non-capturing possessive bracket with unlimited repeat. We come here
1015        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1016        without the capturing complication. It is written out separately for speed
1017        and cleanliness. */
1018    
1019        case OP_BRAPOS:
1020        case OP_SBRAPOS:
1021        allow_zero = FALSE;
1022    
1023        POSSESSIVE_NON_CAPTURE:
1024        matched_once = FALSE;
1025        code_offset = ecode - md->start_code;
1026    
1027        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      for (;;)
1028          eptrb, flags, RM2);        {
1029        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1030          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1031            eptrb, RM48);
1032          if (rrc == MATCH_KETRPOS)
1033            {
1034            offset_top = md->end_offset_top;
1035            eptr = md->end_match_ptr;
1036            ecode = md->start_code + code_offset;
1037            matched_once = TRUE;
1038            continue;
1039            }
1040          if (rrc != MATCH_NOMATCH &&
1041              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1042            RRETURN(rrc);
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045          }
1046    
1047        if (matched_once || allow_zero)
1048          {
1049          ecode += 1 + LINK_SIZE;
1050          break;
1051        }        }
1052        RRETURN(MATCH_NOMATCH);
1053    
1054      /* Control never reaches here. */      /* Control never reaches here. */
1055    
1056      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1057      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1058      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1059      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1060    
1061      case OP_COND:      case OP_COND:
1062      case OP_SCOND:      case OP_SCOND:
1063      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1064    
1065      /* 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
1066      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 884  for (;;) Line 1070  for (;;)
1070        if (pcre_callout != NULL)        if (pcre_callout != NULL)
1071          {          {
1072          pcre_callout_block cb;          pcre_callout_block cb;
1073          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1074          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1075          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1076          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1077          cb.subject_length   = md->end_subject - md->start_subject;          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1078          cb.start_match      = mstart - md->start_subject;          cb.start_match      = (int)(mstart - md->start_subject);
1079          cb.current_position = eptr - md->start_subject;          cb.current_position = (int)(eptr - md->start_subject);
1080          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1081          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1082          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1083          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1084          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1085            cb.mark             = markptr;
1086          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1087          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1088          }          }
# Line 1049  for (;;) Line 1236  for (;;)
1236        }        }
1237    
1238      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1239      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
1240      assertion. */      an assertion. */
1241    
1242      else      else
1243        {        {
1244        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1245            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1246        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1247          {          {
1248            if (md->end_offset_top > offset_top)
1249              offset_top = md->end_offset_top;  /* Captures may have happened */
1250          condition = TRUE;          condition = TRUE;
1251          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1252          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1253          }          }
1254        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH &&
1255                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1256          {          {
1257          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1258          }          }
# Line 1074  for (;;) Line 1264  for (;;)
1264        }        }
1265    
1266      /* 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,
1267      we can use tail recursion to avoid using another stack frame, except when      we used to use tail recursion to avoid using another stack frame, except
1268      match_cbegroup is required for an unlimited repeat of a possibly empty      when there was unlimited repeat of a possibly empty group. However, that
1269      group. If the second alternative doesn't exist, we can just plough on. */      strategy no longer works because of the possibilty of (*THEN) being
1270        encountered in the branch. A recursive call to match() is always required,
1271        unless the second alternative doesn't exist, in which case we can just
1272        plough on. */
1273    
1274      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1275        {        {
1276        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1277        if (op == OP_SCOND)        /* Possibly empty group */        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1278          {        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1279          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          rrc = MATCH_NOMATCH;
1280          RRETURN(rrc);        RRETURN(rrc);
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
         goto TAIL_RECURSE;  
         }  
1281        }        }
1282      else                         /* Condition false & no alternative */      else                         /* Condition false & no alternative */
1283        {        {
# Line 1116  for (;;) Line 1303  for (;;)
1303        {        {
1304        md->offset_vector[offset] =        md->offset_vector[offset] =
1305          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1306        md->offset_vector[offset+1] = eptr - md->start_subject;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1307        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1308        }        }
1309      ecode += 3;      ecode += 3;
1310      break;      break;
1311    
1312    
1313      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. */
     recursion, we should restore the offsets appropriately and continue from  
     after the call. */  
1314    
     case OP_ACCEPT:  
1315      case OP_END:      case OP_END:
1316      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1317        {      case OP_ASSERT_ACCEPT:
       recursion_info *rec = md->recursive;  
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
       md->recursive = rec->prevrec;  
       memmove(md->offset_vector, rec->offset_save,  
         rec->saved_max * sizeof(int));  
       offset_top = rec->save_offset_top;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1318    
1319      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is      /* If we have matched an empty string, fail if not in an assertion and not
1320      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1321      the subject. In both cases, backtracking will then try other alternatives,      is set and we have matched at the start of the subject. In both cases,
1322      if any. */      backtracking will then try other alternatives, if any. */
1323    
1324      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1325          (md->notempty ||           md->recursive == NULL &&
1326            (md->notempty_atstart &&           (md->notempty ||
1327              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1328                 mstart == md->start_subject + md->start_offset)))
1329        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1330    
1331      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1158  for (;;) Line 1333  for (;;)
1333      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1334      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1335      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
     MRRETURN(((op == OP_END)? MATCH_MATCH : MATCH_ACCEPT));  
1336    
1337      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1338        given as the argument to MRRETURN when the heap is in use. */
1339    
1340      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1341      ims = ecode[1];      MRRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1342    
1343      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1344      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,
1345      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
1346      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
1347      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1348        of a condition, we want to return immediately afterwards. The caller of
1349        this incarnation of the match() function will have set MATCH_CONDASSERT in
1350        md->match_function type, and one of these opcodes will be the first opcode
1351        that is processed. We use a local variable that is preserved over calls to
1352        match() to remember this case. */
1353    
1354      case OP_ASSERT:      case OP_ASSERT:
1355      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1356        if (md->match_function_type == MATCH_CONDASSERT)
1357          {
1358          condassert = TRUE;
1359          md->match_function_type = 0;
1360          }
1361        else condassert = FALSE;
1362    
1363      do      do
1364        {        {
1365        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1366        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1367          {          {
1368          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1369            markptr = md->mark;
1370          break;          break;
1371          }          }
1372        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH &&
1373              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1374            RRETURN(rrc);
1375        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1376        }        }
1377      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1378    
1379      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1380    
1381      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1382    
1383      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1384    
1385      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1386      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1209  for (;;) Line 1396  for (;;)
1396    
1397      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1398      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1399        if (md->match_function_type == MATCH_CONDASSERT)
1400          {
1401          condassert = TRUE;
1402          md->match_function_type = 0;
1403          }
1404        else condassert = FALSE;
1405    
1406      do      do
1407        {        {
1408        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
         RM5);  
1409        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1410        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1411          {          {
1412          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1413          break;          break;
1414          }          }
1415        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH &&
1416              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1417            RRETURN(rrc);
1418        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1419        }        }
1420      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1421    
1422      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1423    
1424      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1425      continue;      continue;
# Line 1270  for (;;) Line 1465  for (;;)
1465      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1466        {        {
1467        pcre_callout_block cb;        pcre_callout_block cb;
1468        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1469        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1470        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1471        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1472        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1473        cb.start_match      = mstart - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1474        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1475        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1476        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1477        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1478        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1479        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1480          cb.mark             = markptr;
1481        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1482        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1483        }        }
# Line 1292  for (;;) Line 1488  for (;;)
1488      offset data is the offset to the starting bracket from the start of the      offset data is the offset to the starting bracket from the start of the
1489      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1490    
1491      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1492      save their starting points and reinstate them after the recursion. However,      re-instated afterwards. We don't know how many are started and not yet
1493      we don't know how many such there are (offset_top records the completed      finished (offset_top records the completed total) so we just have to save
1494      total) so we just have to save all the potential data. There may be up to      all the potential data. There may be up to 65535 such values, which is too
1495      65535 such values, which is too large to put on the stack, but using malloc      large to put on the stack, but using malloc for small numbers seems
1496      for small numbers seems expensive. As a compromise, the stack is used when      expensive. As a compromise, the stack is used when there are no more than
1497      there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc      REC_STACK_SAVE_MAX values to store; otherwise malloc is used.
     is used. A problem is what to do if the malloc fails ... there is no way of  
     returning to the top level with an error. Save the top REC_STACK_SAVE_MAX  
     values on the stack, and accept that the rest may be wrong.  
1498    
1499      There are also other values that have to be saved. We use a chained      There are also other values that have to be saved. We use a chained
1500      sequence of blocks that actually live on the stack. Thanks to Robin Houston      sequence of blocks that actually live on the stack. Thanks to Robin Houston
1501      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1502        a lot, so he is not to blame for the current way it works. */
1503    
1504      case OP_RECURSE:      case OP_RECURSE:
1505        {        {
1506          recursion_info *ri;
1507          int recno;
1508    
1509        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1510        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1511          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1512    
1513          /* Check for repeating a recursion without advancing the subject pointer.
1514          This should catch convoluted mutual recursions. (Some simple cases are
1515          caught at compile time.) */
1516    
1517          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1518            if (recno == ri->group_num && eptr == ri->subject_position)
1519              RRETURN(PCRE_ERROR_RECURSELOOP);
1520    
1521        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1522    
1523          new_recursive.group_num = recno;
1524          new_recursive.subject_position = eptr;
1525        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1526        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1527    
1528        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1529    
1530        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1531    
1532        /* Now save the offset data. */        /* Now save the offset data */
1533    
1534        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1535        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1334  for (;;) Line 1540  for (;;)
1540            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1541          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1542          }          }
   
1543        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1544              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1545    
1546        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1547        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1548          might be changed, so reset it before looping. */
1549    
1550        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1551        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1552        do        do
1553          {          {
1554            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1555          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1556            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1557            memcpy(md->offset_vector, new_recursive.offset_save,
1558                new_recursive.saved_max * sizeof(int));
1559          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1560            {            {
1561            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1562            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1563            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1564              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1565            MRRETURN(MATCH_MATCH);  
1566              /* Set where we got to in the subject, and reset the start in case
1567              it was changed by \K. This *is* propagated back out of a recursion,
1568              for Perl compatibility. */
1569    
1570              eptr = md->end_match_ptr;
1571              mstart = md->start_match_ptr;
1572              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1573            }            }
1574          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH &&
1575                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1576            {            {
1577            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1578            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
# Line 1365  for (;;) Line 1581  for (;;)
1581            }            }
1582    
1583          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1584          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1585          }          }
1586        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1377  for (;;) Line 1591  for (;;)
1591          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1592        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1593        }        }
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer, but resetting  
     the start-of-match value in case it was changed by \K. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */  
         {  
         mstart = md->start_match_ptr;  
         break;  
         }  
       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
1594    
1595      /* Continue as from after the assertion, updating the offsets high water      RECURSION_MATCHED:
1596      mark, since extracts may have been taken. */      break;
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     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));  
       }  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
1597    
1598      /* An alternation is the end of a branch; scan along to find the end of the      /* An alternation is the end of a branch; scan along to find the end of the
1599      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1472  for (;;) Line 1609  for (;;)
1609      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1610    
1611      case OP_BRAZERO:      case OP_BRAZERO:
1612        {      next = ecode + 1;
1613        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1614        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1615        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1616        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1617      break;      break;
1618    
1619      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1620        {      next = ecode + 1;
1621        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1622        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1623        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1624        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1625      break;      break;
1626    
1627      case OP_SKIPZERO:      case OP_SKIPZERO:
1628        {      next = ecode+1;
1629        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1630        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1631      break;      break;
1632    
1633        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1634        here; just jump to the group, with allow_zero set TRUE. */
1635    
1636        case OP_BRAPOSZERO:
1637        op = *(++ecode);
1638        allow_zero = TRUE;
1639        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1640          goto POSSESSIVE_NON_CAPTURE;
1641    
1642      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1643    
1644      case OP_KET:      case OP_KET:
1645      case OP_KETRMIN:      case OP_KETRMIN:
1646      case OP_KETRMAX:      case OP_KETRMAX:
1647        case OP_KETRPOS:
1648      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1649    
1650      /* 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
1651      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1652      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1653    
1654      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1655        {        {
1656        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1657        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1658        }        }
1659      else saved_eptr = NULL;      else saved_eptr = NULL;
1660    
1661      /* If we are at the end of an assertion group or an atomic group, stop      /* If we are at the end of an assertion group, stop matching and return
1662      matching and return MATCH_MATCH, but record the current high water mark for      MATCH_MATCH, but record the current high water mark for use by positive
1663      use by positive assertions. We also need to record the match start in case      assertions. We also need to record the match start in case it was changed
1664      it was changed by \K. */      by \K. */
1665    
1666      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1667          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1668        {        {
1669        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1670        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1671        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1672        MRRETURN(MATCH_MATCH);        MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1673        }        }
1674    
1675      /* For capturing groups we have to check the group number back at the start      /* For capturing groups we have to check the group number back at the start
1676      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1677      bumping the high water mark. Note that whole-pattern recursion is coded as      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1678      a recurse into group 0, so it won't be picked up here. Instead, we catch it      into group 0, so it won't be picked up here. Instead, we catch it when the
1679      when the OP_END is reached. Other recursion is handled here. */      OP_END is reached. Other recursion is handled here. We just have to record
1680        the current subject position and start match pointer and give a MATCH
1681        return. */
1682    
1683      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1684            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1685        {        {
1686        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1687        offset = number << 1;        offset = number << 1;
# Line 1548  for (;;) Line 1691  for (;;)
1691        printf("\n");        printf("\n");
1692  #endif  #endif
1693    
1694          /* Handle a recursively called group. */
1695    
1696          if (md->recursive != NULL && md->recursive->group_num == number)
1697            {
1698            md->end_match_ptr = eptr;
1699            md->start_match_ptr = mstart;
1700            RRETURN(MATCH_MATCH);
1701            }
1702    
1703          /* Deal with capturing */
1704    
1705        md->capture_last = number;        md->capture_last = number;
1706        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1707          {          {
1708            /* If offset is greater than offset_top, it means that we are
1709            "skipping" a capturing group, and that group's offsets must be marked
1710            unset. In earlier versions of PCRE, all the offsets were unset at the
1711            start of matching, but this doesn't work because atomic groups and
1712            assertions can cause a value to be set that should later be unset.
1713            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1714            part of the atomic group, but this is not on the final matching path,
1715            so must be unset when 2 is set. (If there is no group 2, there is no
1716            problem, because offset_top will then be 2, indicating no capture.) */
1717    
1718            if (offset > offset_top)
1719              {
1720              register int *iptr = md->offset_vector + offset_top;
1721              register int *iend = md->offset_vector + offset;
1722              while (iptr < iend) *iptr++ = -1;
1723              }
1724    
1725            /* Now make the extraction */
1726    
1727          md->offset_vector[offset] =          md->offset_vector[offset] =
1728            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1729          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1730          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1731          }          }
1732          }
1733    
1734        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1735        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1736        group. This is the forcible breaking of infinite loops as implemented in
1737        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1738        processing the rest of the pattern at a lower level. If this results in a
1739        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1740        bypassing intermediate backup points, but resetting any captures that
1741        happened along the way. */
1742    
1743        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1744          {
1745          if (*prev == OP_ONCE)
1746          {          {
1747          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1748          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1749          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1750          memcpy(md->offset_vector, rec->offset_save,          RRETURN(MATCH_ONCE);
           rec->saved_max * sizeof(int));  
         offset_top = rec->save_offset_top;  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1751          }          }
1752          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1753          break;
1754        }        }
1755    
1756      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1757      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1758        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
1759    
1760      /* For a non-repeating ket, just continue at this level. This also      if (*ecode == OP_KETRPOS)
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
1761        {        {
1762        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1763        break;        md->end_offset_top = offset_top;
1764          RRETURN(MATCH_KETRPOS);
1765        }        }
1766    
1767      /* The repeating kets try the rest of the pattern or restart from the      /* The normal repeating kets try the rest of the pattern or restart from
1768      preceding bracket, in the appropriate order. In the second case, we can use      the preceding bracket, in the appropriate order. In the second case, we can
1769      tail recursion to avoid using another stack frame, unless we have an      use tail recursion to avoid using another stack frame, unless we have an
1770      unlimited repeat of a group that can match an empty string. */      an atomic group or an unlimited repeat of a group that can match an empty
1771        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1772    
1773      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1774        {        {
1775        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1776        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1777        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1778            {
1779            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1780            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1781            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1782            RRETURN(MATCH_ONCE);
1783            }
1784          if (*prev >= OP_SBRA)    /* Could match an empty string */
1785          {          {
1786          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1787            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1788          RRETURN(rrc);          RRETURN(rrc);
1789          }          }
1790        ecode = prev;        ecode = prev;
# Line 1613  for (;;) Line 1792  for (;;)
1792        }        }
1793      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1794        {        {
1795        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1796          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1797          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1798        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1799          if (*prev == OP_ONCE)
1800            {
1801            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1802            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1803            md->once_target = prev;
1804            RRETURN(MATCH_ONCE);
1805            }
1806        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1807        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1808        }        }
1809      /* Control never gets here */      /* Control never gets here */
1810    
1811      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1812    
1813      case OP_CIRC:      case OP_CIRC:
1814      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
     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 */  
1815    
1816      /* Start of subject assertion */      /* Start of subject assertion */
1817    
# Line 1642  for (;;) Line 1820  for (;;)
1820      ecode++;      ecode++;
1821      break;      break;
1822    
1823        /* Multiline mode: start of subject unless notbol, or after any newline. */
1824    
1825        case OP_CIRCM:
1826        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1827        if (eptr != md->start_subject &&
1828            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1829          MRRETURN(MATCH_NOMATCH);
1830        ecode++;
1831        break;
1832    
1833      /* Start of match assertion */      /* Start of match assertion */
1834    
1835      case OP_SOM:      case OP_SOM:
# Line 1656  for (;;) Line 1844  for (;;)
1844      ecode++;      ecode++;
1845      break;      break;
1846    
1847      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1848      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1849    
1850      case OP_DOLL:      case OP_DOLLM:
1851      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1852        {        { 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;  
       }  
1853      else      else
1854        {        {
1855        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1856        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           MRRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1857        }        }
1858        ecode++;
1859        break;
1860    
1861        /* Not multiline mode: assert before a terminating newline or before end of
1862        subject unless noteol is set. */
1863    
1864        case OP_DOLL:
1865        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1866        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1867    
1868      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1869    
1870      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1871    
1872      case OP_EOD:      case OP_EOD:
1873      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1874        SCHECK_PARTIAL();
1875      ecode++;      ecode++;
1876      break;      break;
1877    
1878      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1879    
1880      case OP_EODN:      case OP_EODN:
1881      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1882        if (eptr < md->end_subject &&
1883          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1884        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1885    
1886        /* Either at end of string or \n before end. */
1887    
1888        SCHECK_PARTIAL();
1889      ecode++;      ecode++;
1890      break;      break;
1891    
# Line 1713  for (;;) Line 1903  for (;;)
1903  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1904        if (utf8)        if (utf8)
1905          {          {
1906            /* Get status of previous character */
1907    
1908          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1909            {            {
1910            USPTR lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1911            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1912            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1913            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1914    #ifdef SUPPORT_UCP
1915              if (md->use_ucp)
1916                {
1917                if (c == '_') prev_is_word = TRUE; else
1918                  {
1919                  int cat = UCD_CATEGORY(c);
1920                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1921                  }
1922                }
1923              else
1924    #endif
1925            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1926            }            }
1927    
1928            /* Get status of next character */
1929    
1930          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
1931            {            {
1932            SCHECK_PARTIAL();            SCHECK_PARTIAL();
# Line 1729  for (;;) Line 1935  for (;;)
1935          else          else
1936            {            {
1937            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1938    #ifdef SUPPORT_UCP
1939              if (md->use_ucp)
1940                {
1941                if (c == '_') cur_is_word = TRUE; else
1942                  {
1943                  int cat = UCD_CATEGORY(c);
1944                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1945                  }
1946                }
1947              else
1948    #endif
1949            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1950            }            }
1951          }          }
1952        else        else
1953  #endif  #endif
1954    
1955        /* Not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1956          consistency with the behaviour of \w we do use it in this case. */
1957    
1958          {          {
1959            /* Get status of previous character */
1960    
1961          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1962            {            {
1963            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1964    #ifdef SUPPORT_UCP
1965              if (md->use_ucp)
1966                {
1967                c = eptr[-1];
1968                if (c == '_') prev_is_word = TRUE; else
1969                  {
1970                  int cat = UCD_CATEGORY(c);
1971                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1972                  }
1973                }
1974              else
1975    #endif
1976            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1977            }            }
1978    
1979            /* Get status of next character */
1980    
1981          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
1982            {            {
1983            SCHECK_PARTIAL();            SCHECK_PARTIAL();
1984            cur_is_word = FALSE;            cur_is_word = FALSE;
1985            }            }
1986          else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          else
1987    #ifdef SUPPORT_UCP
1988            if (md->use_ucp)
1989              {
1990              c = *eptr;
1991              if (c == '_') cur_is_word = TRUE; else
1992                {
1993                int cat = UCD_CATEGORY(c);
1994                cur_is_word = (cat == ucp_L || cat == ucp_N);
1995                }
1996              }
1997            else
1998    #endif
1999            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2000          }          }
2001    
2002        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
# Line 1899  for (;;) Line 2147  for (;;)
2147      switch(c)      switch(c)
2148        {        {
2149        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2150    
2151        case 0x000d:        case 0x000d:
2152        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2153        break;        break;
# Line 2055  for (;;) Line 2304  for (;;)
2304               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2305               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2306            MRRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2307           break;          break;
2308    
2309          case PT_GC:          case PT_GC:
2310          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
# Line 2072  for (;;) Line 2321  for (;;)
2321            MRRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2322          break;          break;
2323    
2324            /* These are specials */
2325    
2326            case PT_ALNUM:
2327            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2328                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2329              MRRETURN(MATCH_NOMATCH);
2330            break;
2331    
2332            case PT_SPACE:    /* Perl space */
2333            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2334                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2335                   == (op == OP_NOTPROP))
2336              MRRETURN(MATCH_NOMATCH);
2337            break;
2338    
2339            case PT_PXSPACE:  /* POSIX space */
2340            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2341                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2342                 c == CHAR_FF || c == CHAR_CR)
2343                   == (op == OP_NOTPROP))
2344              MRRETURN(MATCH_NOMATCH);
2345            break;
2346    
2347            case PT_WORD:
2348            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2349                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2350                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2351              MRRETURN(MATCH_NOMATCH);
2352            break;
2353    
2354            /* This should never occur */
2355    
2356          default:          default:
2357          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2358          }          }
# Line 2090  for (;;) Line 2371  for (;;)
2371        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2372        }        }
2373      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2374        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2375        while (eptr < md->end_subject)
2376        {        {
2377        int category = UCD_CATEGORY(c);        int len = 1;
2378        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2379        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2380          {        eptr += len;
         int len = 1;  
         if (!utf8) c = *eptr; else  
           {  
           GETCHARLEN(c, eptr, len);  
           }  
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
         eptr += len;  
         }  
2381        }        }
2382      ecode++;      ecode++;
2383      break;      break;
# Line 2119  for (;;) Line 2393  for (;;)
2393      loops). */      loops). */
2394    
2395      case OP_REF:      case OP_REF:
2396        {      case OP_REFI:
2397        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2398        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2399        ecode += 3;
2400    
2401        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2402    
2403        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2404        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
2405        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.  
2406    
2407        (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
2408        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2409    
2410        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
2411        referenced subpattern. */      referenced subpattern. */
2412    
2413        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2414          length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;        length = (md->jscript_compat)? 0 : -1;
2415        else      else
2416          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2417    
2418        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2419    
2420        switch (*ecode)      switch (*ecode)
2421          {        {
2422          case OP_CRSTAR:        case OP_CRSTAR:
2423          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2424          case OP_CRPLUS:        case OP_CRPLUS:
2425          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2426          case OP_CRQUERY:        case OP_CRQUERY:
2427          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2428          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2429          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2430          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2431          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2432          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2433          break;        break;
2434    
2435          case OP_CRRANGE:        case OP_CRRANGE:
2436          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2437          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2438          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2439          max = GET2(ecode, 3);        max = GET2(ecode, 3);
2440          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2441          ecode += 5;        ecode += 5;
2442          break;        break;
2443    
2444          default:               /* No repeat follows */        default:               /* No repeat follows */
2445          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2446            {          {
2447            CHECK_PARTIAL();          CHECK_PARTIAL();
2448            MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
           }  
         eptr += length;  
         continue;              /* With the main loop */  
2449          }          }
2450          eptr += length;
2451          continue;              /* With the main loop */
2452          }
2453    
2454        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2455        main loop. */      zero, just continue with the main loop. */
2456    
2457        if (length == 0) continue;      if (length == 0) continue;
2458    
2459        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2460        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2461        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2462    
2463        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2464          {
2465          int slength;
2466          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2467          {          {
2468          if (!match_ref(offset, eptr, length, md, ims))          CHECK_PARTIAL();
2469            {          MRRETURN(MATCH_NOMATCH);
           CHECK_PARTIAL();  
           MRRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2470          }          }
2471          eptr += slength;
2472          }
2473    
2474        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2475        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2476    
2477        if (min == max) continue;      if (min == max) continue;
2478    
2479        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2480    
2481        if (minimize)      if (minimize)
2482          {
2483          for (fi = min;; fi++)
2484          {          {
2485          for (fi = min;; fi++)          int slength;
2486            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2487            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2488            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2489            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2490            {            {
2491            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2492            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;  
2493            }            }
2494          /* Control never gets here */          eptr += slength;
2495          }          }
2496          /* Control never gets here */
2497          }
2498    
2499        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2500    
2501        else      else
2502          {        {
2503          pp = eptr;        pp = eptr;
2504          for (i = min; i < max; i++)        for (i = min; i < max; i++)
2505            {          {
2506            if (!match_ref(offset, eptr, length, md, ims))          int slength;
2507              {          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
             CHECK_PARTIAL();  
             break;  
             }  
           eptr += length;  
           }  
         while (eptr >= pp)  
2508            {            {
2509            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2510            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2511            }            }
2512          MRRETURN(MATCH_NOMATCH);          eptr += slength;
2513            }
2514          while (eptr >= pp)
2515            {
2516            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2517            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2518            eptr -= length;
2519          }          }
2520          MRRETURN(MATCH_NOMATCH);
2521        }        }
2522      /* Control never gets here */      /* Control never gets here */
2523    
# Line 2347  for (;;) Line 2623  for (;;)
2623            {            {
2624            for (fi = min;; fi++)            for (fi = min;; fi++)
2625              {              {
2626              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2627              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2628              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2629              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2372  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, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
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 2418  for (;;) Line 2694  for (;;)
2694              }              }
2695            for (;;)            for (;;)
2696              {              {
2697              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2698              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2699              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2700              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2441  for (;;) Line 2717  for (;;)
2717              }              }
2718            while (eptr >= pp)            while (eptr >= pp)
2719              {              {
2720              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2721              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2722              eptr--;              eptr--;
2723              }              }
# Line 2517  for (;;) Line 2793  for (;;)
2793          {          {
2794          for (fi = min;; fi++)          for (fi = min;; fi++)
2795            {            {
2796            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2797            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2798            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2799            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2550  for (;;) Line 2826  for (;;)
2826            }            }
2827          for(;;)          for(;;)
2828            {            {
2829            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2830            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2831            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2832            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2595  for (;;) Line 2871  for (;;)
2871    
2872      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2873    
2874      case OP_CHARNC:      case OP_CHARI:
2875  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2876      if (utf8)      if (utf8)
2877        {        {
# Line 2655  for (;;) Line 2931  for (;;)
2931      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2932    
2933      case OP_EXACT:      case OP_EXACT:
2934        case OP_EXACTI:
2935      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2936      ecode += 3;      ecode += 3;
2937      goto REPEATCHAR;      goto REPEATCHAR;
2938    
2939      case OP_POSUPTO:      case OP_POSUPTO:
2940        case OP_POSUPTOI:
2941      possessive = TRUE;      possessive = TRUE;
2942      /* Fall through */      /* Fall through */
2943    
2944      case OP_UPTO:      case OP_UPTO:
2945        case OP_UPTOI:
2946      case OP_MINUPTO:      case OP_MINUPTO:
2947        case OP_MINUPTOI:
2948      min = 0;      min = 0;
2949      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2950      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2951      ecode += 3;      ecode += 3;
2952      goto REPEATCHAR;      goto REPEATCHAR;
2953    
2954      case OP_POSSTAR:      case OP_POSSTAR:
2955        case OP_POSSTARI:
2956      possessive = TRUE;      possessive = TRUE;
2957      min = 0;      min = 0;
2958      max = INT_MAX;      max = INT_MAX;
# Line 2679  for (;;) Line 2960  for (;;)
2960      goto REPEATCHAR;      goto REPEATCHAR;
2961    
2962      case OP_POSPLUS:      case OP_POSPLUS:
2963        case OP_POSPLUSI:
2964      possessive = TRUE;      possessive = TRUE;
2965      min = 1;      min = 1;
2966      max = INT_MAX;      max = INT_MAX;
# Line 2686  for (;;) Line 2968  for (;;)
2968      goto REPEATCHAR;      goto REPEATCHAR;
2969    
2970      case OP_POSQUERY:      case OP_POSQUERY:
2971        case OP_POSQUERYI:
2972      possessive = TRUE;      possessive = TRUE;
2973      min = 0;      min = 0;
2974      max = 1;      max = 1;
# Line 2693  for (;;) Line 2976  for (;;)
2976      goto REPEATCHAR;      goto REPEATCHAR;
2977    
2978      case OP_STAR:      case OP_STAR:
2979        case OP_STARI:
2980      case OP_MINSTAR:      case OP_MINSTAR:
2981        case OP_MINSTARI:
2982      case OP_PLUS:      case OP_PLUS:
2983        case OP_PLUSI:
2984      case OP_MINPLUS:      case OP_MINPLUS:
2985        case OP_MINPLUSI:
2986      case OP_QUERY:      case OP_QUERY:
2987        case OP_QUERYI:
2988      case OP_MINQUERY:      case OP_MINQUERY:
2989      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
2990        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
2991      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
2992      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2993      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2994      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2723  for (;;) Line 3011  for (;;)
3011          {          {
3012  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3013          unsigned int othercase;          unsigned int othercase;
3014          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3015              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3016            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3017          else oclength = 0;          else oclength = 0;
# Line 2751  for (;;) Line 3039  for (;;)
3039            {            {
3040            for (fi = min;; fi++)            for (fi = min;; fi++)
3041              {              {
3042              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3043              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3044              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3045              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2793  for (;;) Line 3081  for (;;)
3081    
3082            for(;;)            for(;;)
3083              {              {
3084              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3085              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3086              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3087  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2830  for (;;) Line 3118  for (;;)
3118      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3119        max, eptr));        max, eptr));
3120    
3121      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3122        {        {
3123        fc = md->lcc[fc];        fc = md->lcc[fc];
3124        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2847  for (;;) Line 3135  for (;;)
3135          {          {
3136          for (fi = min;; fi++)          for (fi = min;; fi++)
3137            {            {
3138            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3139            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3140            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3141            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2877  for (;;) Line 3165  for (;;)
3165    
3166          while (eptr >= pp)          while (eptr >= pp)
3167            {            {
3168            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3169            eptr--;            eptr--;
3170            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3171            }            }
# Line 2906  for (;;) Line 3194  for (;;)
3194          {          {
3195          for (fi = min;; fi++)          for (fi = min;; fi++)
3196            {            {
3197            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3198            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3199            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3200            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2935  for (;;) Line 3223  for (;;)
3223    
3224          while (eptr >= pp)          while (eptr >= pp)
3225            {            {
3226            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3227            eptr--;            eptr--;
3228            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3229            }            }
# Line 2948  for (;;) Line 3236  for (;;)
3236      checking can be multibyte. */      checking can be multibyte. */
3237    
3238      case OP_NOT:      case OP_NOT:
3239        case OP_NOTI:
3240      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3241        {        {
3242        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 2955  for (;;) Line 3244  for (;;)
3244        }        }
3245      ecode++;      ecode++;
3246      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3247      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3248        {        {
3249  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3250        if (c < 256)        if (c < 256)
# Line 2963  for (;;) Line 3252  for (;;)
3252        c = md->lcc[c];        c = md->lcc[c];
3253        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3254        }        }
3255      else      else    /* Caseful */
3256        {        {
3257        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3258        }        }
# Line 2977  for (;;) Line 3266  for (;;)
3266      about... */      about... */
3267    
3268      case OP_NOTEXACT:      case OP_NOTEXACT:
3269        case OP_NOTEXACTI:
3270      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3271      ecode += 3;      ecode += 3;
3272      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3273    
3274      case OP_NOTUPTO:      case OP_NOTUPTO:
3275        case OP_NOTUPTOI:
3276      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3277        case OP_NOTMINUPTOI:
3278      min = 0;      min = 0;
3279      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3280      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3281      ecode += 3;      ecode += 3;
3282      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3283    
3284      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3285        case OP_NOTPOSSTARI:
3286      possessive = TRUE;      possessive = TRUE;
3287      min = 0;      min = 0;
3288      max = INT_MAX;      max = INT_MAX;
# Line 2997  for (;;) Line 3290  for (;;)
3290      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3291    
3292      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3293        case OP_NOTPOSPLUSI:
3294      possessive = TRUE;      possessive = TRUE;
3295      min = 1;      min = 1;
3296      max = INT_MAX;      max = INT_MAX;
# Line 3004  for (;;) Line 3298  for (;;)
3298      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3299    
3300      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3301        case OP_NOTPOSQUERYI:
3302      possessive = TRUE;      possessive = TRUE;
3303      min = 0;      min = 0;
3304      max = 1;      max = 1;
# Line 3011  for (;;) Line 3306  for (;;)
3306      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3307    
3308      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3309        case OP_NOTPOSUPTOI:
3310      possessive = TRUE;      possessive = TRUE;
3311      min = 0;      min = 0;
3312      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 3018  for (;;) Line 3314  for (;;)
3314      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3315    
3316      case OP_NOTSTAR:      case OP_NOTSTAR:
3317        case OP_NOTSTARI:
3318      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3319        case OP_NOTMINSTARI:
3320      case OP_NOTPLUS:      case OP_NOTPLUS:
3321        case OP_NOTPLUSI:
3322      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3323        case OP_NOTMINPLUSI:
3324      case OP_NOTQUERY:      case OP_NOTQUERY:
3325        case OP_NOTQUERYI:
3326      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3327      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3328        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3329      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3330      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3331      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3045  for (;;) Line 3347  for (;;)
3347      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,
3348        max, eptr));        max, eptr));
3349    
3350      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3351        {        {
3352        fc = md->lcc[fc];        fc = md->lcc[fc];
3353    
# Line 3093  for (;;) Line 3395  for (;;)
3395            register unsigned int d;            register unsigned int d;
3396            for (fi = min;; fi++)            for (fi = min;; fi++)
3397              {              {
3398              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3399              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3400              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3401              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3112  for (;;) Line 3414  for (;;)
3414            {            {
3415            for (fi = min;; fi++)            for (fi = min;; fi++)
3416              {              {
3417              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3418              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3419              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3420              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3153  for (;;) Line 3455  for (;;)
3455          if (possessive) continue;          if (possessive) continue;
3456          for(;;)          for(;;)
3457              {              {
3458              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3459              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3460              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3461              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3176  for (;;) Line 3478  for (;;)
3478            if (possessive) continue;            if (possessive) continue;
3479            while (eptr >= pp)            while (eptr >= pp)
3480              {              {
3481              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3482              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3483              eptr--;              eptr--;
3484              }              }
# Line 3233  for (;;) Line 3535  for (;;)
3535            register unsigned int d;            register unsigned int d;
3536            for (fi = min;; fi++)            for (fi = min;; fi++)
3537              {              {
3538              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3539              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3540              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3541              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3251  for (;;) Line 3553  for (;;)
3553            {            {
3554            for (fi = min;; fi++)            for (fi = min;; fi++)
3555              {              {
3556              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3557              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3558              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3559              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3291  for (;;) Line 3593  for (;;)
3593            if (possessive) continue;            if (possessive) continue;
3594            for(;;)            for(;;)
3595              {              {
3596              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3597              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3598              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3599              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3314  for (;;) Line 3616  for (;;)
3616            if (possessive) continue;            if (possessive) continue;
3617            while (eptr >= pp)            while (eptr >= pp)
3618              {              {
3619              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3620              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3621              eptr--;              eptr--;
3622              }              }
# Line 3429  for (;;) Line 3731  for (;;)
3731            case PT_LAMP:            case PT_LAMP:
3732            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3733              {              {
3734                int chartype;
3735              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3736                {                {
3737                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3738                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3739                }                }
3740              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3741              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
3742              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
3743                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
3744                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
3745                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3746              }              }
3747            break;            break;
# Line 3452  for (;;) Line 3755  for (;;)
3755                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3756                }                }
3757              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3758              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
3759                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3760              }              }
3761            break;            break;
# Line 3467  for (;;) Line 3769  for (;;)
3769                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3770                }                }
3771              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3772              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
3773                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3774              }              }
3775            break;            break;
# Line 3482  for (;;) Line 3783  for (;;)
3783                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3784                }                }
3785              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3786              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
3787              if ((prop_script == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
3788                }
3789              break;
3790    
3791              case PT_ALNUM:
3792              for (i = 1; i <= min; i++)
3793                {
3794                int category;
3795                if (eptr >= md->end_subject)
3796                  {
3797                  SCHECK_PARTIAL();
3798                  MRRETURN(MATCH_NOMATCH);
3799                  }
3800                GETCHARINCTEST(c, eptr);
3801                category = UCD_CATEGORY(c);
3802                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
3803                  MRRETURN(MATCH_NOMATCH);
3804                }
3805              break;
3806    
3807              case PT_SPACE:    /* Perl space */
3808              for (i = 1; i <= min; i++)
3809                {
3810                if (eptr >= md->end_subject)
3811                  {
3812                  SCHECK_PARTIAL();
3813                  MRRETURN(MATCH_NOMATCH);
3814                  }
3815                GETCHARINCTEST(c, eptr);
3816                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3817                     c == CHAR_FF || c == CHAR_CR)
3818                       == prop_fail_result)
3819                  MRRETURN(MATCH_NOMATCH);
3820                }
3821              break;
3822    
3823              case PT_PXSPACE:  /* POSIX space */
3824              for (i = 1; i <= min; i++)
3825                {
3826                if (eptr >= md->end_subject)
3827                  {
3828                  SCHECK_PARTIAL();
3829                  MRRETURN(MATCH_NOMATCH);
3830                  }
3831                GETCHARINCTEST(c, eptr);
3832                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3833                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3834                       == prop_fail_result)
3835                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3836              }              }
3837            break;            break;
3838    
3839              case PT_WORD:
3840              for (i = 1; i <= min; i++)
3841                {
3842                int category;
3843                if (eptr >= md->end_subject)
3844                  {
3845                  SCHECK_PARTIAL();
3846                  MRRETURN(MATCH_NOMATCH);
3847                  }
3848                GETCHARINCTEST(c, eptr);
3849                category = UCD_CATEGORY(c);
3850                if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
3851                       == prop_fail_result)
3852                  MRRETURN(MATCH_NOMATCH);
3853                }
3854              break;
3855    
3856              /* This should not occur */
3857    
3858            default:            default:
3859            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
3860            }            }
# Line 3506  for (;;) Line 3873  for (;;)
3873              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
3874              }              }
3875            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3876            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
3877            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3878              {              {
3879              int len = 1;              int len = 1;
3880              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
3881                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
3882              eptr += len;              eptr += len;
3883              }              }
3884            }            }
# Line 3572  for (;;) Line 3936  for (;;)
3936            switch(c)            switch(c)
3937              {              {
3938              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3939    
3940              case 0x000d:              case 0x000d:
3941              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3942              break;              break;
# Line 3848  for (;;) Line 4213  for (;;)
4213            switch(*eptr++)            switch(*eptr++)
4214              {              {
4215              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4216    
4217              case 0x000d:              case 0x000d:
4218              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4219              break;              break;
4220    
4221              case 0x000a:              case 0x000a:
4222              break;              break;
4223    
# Line 4040  for (;;) Line 4407  for (;;)
4407            case PT_ANY:            case PT_ANY:
4408            for (fi = min;; fi++)            for (fi = min;; fi++)
4409              {              {
4410              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4411              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4412              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4413              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4048  for (;;) Line 4415  for (;;)
4415                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4416                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4417                }                }
4418              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4419              if (prop_fail_result) MRRETURN(MATCH_NOMATCH);              if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
4420              }              }
4421            /* Control never gets here */            /* Control never gets here */
# Line 4056  for (;;) Line 4423  for (;;)
4423            case PT_LAMP:            case PT_LAMP:
4424            for (fi = min;; fi++)            for (fi = min;; fi++)
4425              {              {
4426              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              int chartype;
4427                RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4428              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4429              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4430              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4064  for (;;) Line 4432  for (;;)
4432                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4433                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4434                }                }
4435              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4436              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4437              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4438                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4439                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4440                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4441              }              }
4442            /* Control never gets here */            /* Control never gets here */
# Line 4076  for (;;) Line 4444  for (;;)
4444            case PT_GC:            case PT_GC:
4445            for (fi = min;; fi++)            for (fi = min;; fi++)
4446              {              {
4447              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4448              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4449              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4450              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4084  for (;;) Line 4452  for (;;)
4452                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4453                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4454                }                }
4455              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4456              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4457                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4458              }              }
4459            /* Control never gets here */            /* Control never gets here */
# Line 4094  for (;;) Line 4461  for (;;)
4461            case PT_PC:            case PT_PC:
4462            for (fi = min;; fi++)            for (fi = min;; fi++)
4463              {              {
4464              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
4465              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4466              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4467              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4102  for (;;) Line 4469  for (;;)
4469                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4470                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4471                }                }
4472              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4473              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4474                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4475              }              }
4476            /* Control never gets here */            /* Control never gets here */
# Line 4112  for (;;) Line 4478  for (;;)
4478            case PT_SC:            case PT_SC:
4479            for (fi = min;; fi++)            for (fi = min;; fi++)
4480              {              {
4481              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
4482              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4483              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4484              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4120  for (;;) Line 4486  for (;;)
4486                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4487                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4488                }                }
4489              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4490              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
4491              if ((prop_script == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
4492                }
4493              /* Control never gets here */
4494    
4495              case PT_ALNUM:
4496              for (fi = min;; fi++)
4497                {
4498                int category;
4499                RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4500                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4501                if (fi >= max) MRRETURN(MATCH_NOMATCH);
4502                if (eptr >= md->end_subject)
4503                  {
4504                  SCHECK_PARTIAL();
4505                  MRRETURN(MATCH_NOMATCH);
4506                  }
4507                GETCHARINCTEST(c, eptr);
4508                category = UCD_CATEGORY(c);
4509                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
4510                  MRRETURN(MATCH_NOMATCH);
4511                }
4512              /* Control never gets here */
4513    
4514              case PT_SPACE:    /* Perl space */
4515              for (fi = min;; fi++)
4516                {
4517                RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4518                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4519                if (fi >= max) MRRETURN(MATCH_NOMATCH);
4520                if (eptr >= md->end_subject)
4521                  {
4522                  SCHECK_PARTIAL();
4523                  MRRETURN(MATCH_NOMATCH);
4524                  }
4525                GETCHARINCTEST(c, eptr);
4526                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4527                     c == CHAR_FF || c == CHAR_CR)
4528                       == prop_fail_result)
4529                  MRRETURN(MATCH_NOMATCH);
4530                }
4531              /* Control never gets here */
4532    
4533              case PT_PXSPACE:  /* POSIX space */
4534              for (fi = min;; fi++)
4535                {
4536                RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4537                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4538                if (fi >= max) MRRETURN(MATCH_NOMATCH);
4539                if (eptr >= md->end_subject)
4540                  {
4541                  SCHECK_PARTIAL();
4542                  MRRETURN(MATCH_NOMATCH);
4543                  }
4544                GETCHARINCTEST(c, eptr);
4545                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4546                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4547                       == prop_fail_result)
4548                  MRRETURN(MATCH_NOMATCH);
4549                }
4550              /* Control never gets here */
4551    
4552              case PT_WORD:
4553              for (fi = min;; fi++)
4554                {
4555                int category;
4556                RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4557                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4558                if (fi >= max) MRRETURN(MATCH_NOMATCH);
4559                if (eptr >= md->end_subject)
4560                  {
4561                  SCHECK_PARTIAL();
4562                  MRRETURN(MATCH_NOMATCH);
4563                  }
4564                GETCHARINCTEST(c, eptr);
4565                category = UCD_CATEGORY(c);
4566                if ((category == ucp_L ||
4567                     category == ucp_N ||
4568                     c == CHAR_UNDERSCORE)
4569                       == prop_fail_result)
4570                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4571              }              }
4572            /* Control never gets here */            /* Control never gets here */
4573    
4574              /* This should never occur */
4575    
4576            default:            default:
4577            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
4578            }            }
# Line 4139  for (;;) Line 4585  for (;;)
4585          {          {
4586          for (fi = min;; fi++)          for (fi = min;; fi++)
4587            {            {
4588            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4589            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4590            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4591            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4148  for (;;) Line 4594  for (;;)
4594              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4595              }              }
4596            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4597            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4598            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4599              {              {
4600              int len = 1;              int len = 1;
4601              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4602                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4603              eptr += len;              eptr += len;
4604              }              }
4605            }            }
4606          }          }
   
4607        else        else
4608  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4609    
# Line 4171  for (;;) Line 4613  for (;;)
4613          {          {
4614          for (fi = min;; fi++)          for (fi = min;; fi++)
4615            {            {
4616            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4617            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4618            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4619            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4334  for (;;) Line 4776  for (;;)
4776          {          {
4777          for (fi = min;; fi++)          for (fi = min;; fi++)
4778            {            {
4779            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4780            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4781            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4782            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4473  for (;;) Line 4915  for (;;)
4915                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4916                break;                break;
4917                }                }
4918              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4919              if (prop_fail_result) break;              if (prop_fail_result) break;
4920              eptr+= len;              eptr+= len;
4921              }              }
# Line 4482  for (;;) Line 4924  for (;;)
4924            case PT_LAMP:            case PT_LAMP:
4925            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4926              {              {
4927                int chartype;
4928              int len = 1;              int len = 1;
4929              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4930                {                {
4931                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4932                break;                break;
4933                }                }
4934              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4935              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4936              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4937                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4938                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4939                break;                break;
4940              eptr+= len;              eptr+= len;
4941              }              }
# Line 4507  for (;;) Line 4950  for (;;)
4950                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4951                break;                break;
4952                }                }
4953              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4954              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
4955              eptr+= len;              eptr+= len;
4956              }              }
4957            break;            break;
# Line 4524  for (;;) Line 4965  for (;;)
4965                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4966                break;                break;
4967                }                }
4968              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4969              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
4970              eptr+= len;              eptr+= len;
4971              }              }
4972            break;            break;
# Line 4541  for (;;) Line 4980  for (;;)
4980                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4981                break;                break;
4982                }                }
4983              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4984              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
4985              if ((prop_script == prop_value) == prop_fail_result)              eptr+= len;
4986                }
4987              break;
4988    
4989              case PT_ALNUM:
4990              for (i = min; i < max; i++)
4991                {
4992                int category;
4993                int len = 1;
4994                if (eptr >= md->end_subject)
4995                  {
4996                  SCHECK_PARTIAL();
4997                  break;
4998                  }
4999                GETCHARLENTEST(c, eptr, len);
5000                category = UCD_CATEGORY(c);
5001                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
5002                  break;
5003                eptr+= len;
5004                }
5005              break;
5006    
5007              case PT_SPACE:    /* Perl space */
5008              for (i = min; i < max; i++)
5009                {
5010                int len = 1;
5011                if (eptr >= md->end_subject)
5012                  {
5013                  SCHECK_PARTIAL();
5014                  break;
5015                  }
5016                GETCHARLENTEST(c, eptr, len);
5017                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
5018                     c == CHAR_FF || c == CHAR_CR)
5019                     == prop_fail_result)
5020                  break;
5021                eptr+= len;
5022                }
5023              break;
5024    
5025              case PT_PXSPACE:  /* POSIX space */
5026              for (i = min; i < max; i++)
5027                {
5028                int len = 1;
5029                if (eptr >= md->end_subject)
5030                  {
5031                  SCHECK_PARTIAL();
5032                  break;
5033                  }
5034                GETCHARLENTEST(c, eptr, len);
5035                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
5036                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5037                     == prop_fail_result)
5038                  break;
5039                eptr+= len;
5040                }
5041              break;
5042    
5043              case PT_WORD:
5044              for (i = min; i < max; i++)
5045                {
5046                int category;
5047                int len = 1;
5048                if (eptr >= md->end_subject)
5049                  {
5050                  SCHECK_PARTIAL();
5051                  break;
5052                  }
5053                GETCHARLENTEST(c, eptr, len);
5054                category = UCD_CATEGORY(c);
5055                if ((category == ucp_L || category == ucp_N ||
5056                     c == CHAR_UNDERSCORE) == prop_fail_result)
5057                break;                break;
5058              eptr+= len;              eptr+= len;
5059              }              }
5060            break;            break;
5061    
5062              default:
5063              RRETURN(PCRE_ERROR_INTERNAL);
5064            }            }
5065    
5066          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
# Line 4555  for (;;) Line 5068  for (;;)
5068          if (possessive) continue;          if (possessive) continue;
5069          for(;;)          for(;;)
5070            {            {
5071            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5072            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5073            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5074            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4569  for (;;) Line 5082  for (;;)
5082          {          {
5083          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5084            {            {
5085              int len = 1;
5086            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5087              {              {
5088              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5089              break;              break;
5090              }              }
5091            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5092            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5093            if (prop_category == ucp_M) break;            eptr += len;
5094            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5095              {              {
5096              int len = 1;              len = 1;
5097              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5098                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5099              eptr += len;              eptr += len;
5100              }              }
5101            }            }
# Line 4596  for (;;) Line 5106  for (;;)
5106    
5107          for(;;)          for(;;)
5108            {            {
5109            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5110            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5111            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5112            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5113              {              {
             int len = 1;  
5114              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5115                {                {
5116                BACKCHAR(eptr);                BACKCHAR(eptr);
5117                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5118                }                }
5119              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5120              eptr--;              eptr--;
5121              }              }
5122            }            }
# Line 4880  for (;;) Line 5388  for (;;)
5388            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5389            }            }
5390    
5391          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5392            done (no backing up). Otherwise, match at this position; anything other
5393            than no match is immediately returned. For nomatch, back up one
5394            character, unless we are matching \R and the last thing matched was
5395            \r\n, in which case, back up two bytes. */
5396    
5397          if (possessive) continue;          if (possessive) continue;
5398          for(;;)          for(;;)
5399            {            {
5400            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5401            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5402            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5403            BACKCHAR(eptr);            BACKCHAR(eptr);
5404              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5405                  eptr[-1] == '\r') eptr--;
5406            }            }
5407          }          }
5408        else        else
# Line 5087  for (;;) Line 5601  for (;;)
5601            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5602            }            }
5603    
5604          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5605            done (no backing up). Otherwise, match at this position; anything other
5606            than no match is immediately returned. For nomatch, back up one
5607            character (byte), unless we are matching \R and the last thing matched
5608            was \r\n, in which case, back up two bytes. */
5609    
5610          if (possessive) continue;          if (possessive) continue;
5611          while (eptr >= pp)          while (eptr >= pp)
5612            {            {
5613            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5614            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5615              eptr--;
5616              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5617                  eptr[-1] == '\r') eptr--;
5618            }            }
5619          }          }
5620    
# Line 5133  switch (frame->Xwhere) Line 5653  switch (frame->Xwhere)
5653    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)
5654    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)
5655    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)
5656    LBL(53) LBL(54)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63)
5657  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5658    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)
5659    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
5660  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5661    LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)    LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)
5662      LBL(59) LBL(60) LBL(61) LBL(62)
5663  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
5664  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
5665    default:    default:
# Line 5161  Undefine all the macros that were define Line 5682  Undefine all the macros that were define
5682  #undef ecode  #undef ecode
5683  #undef mstart  #undef mstart
5684  #undef offset_top  #undef offset_top
 #undef ims  
5685  #undef eptrb  #undef eptrb
5686  #undef flags  #undef flags
5687    
# Line 5179  Undefine all the macros that were define Line 5699  Undefine all the macros that were define
5699  #undef condition  #undef condition
5700  #undef prev_is_word  #undef prev_is_word
5701    
 #undef original_ims  
   
5702  #undef ctype  #undef ctype
5703  #undef length  #undef length
5704  #undef max  #undef max
# Line 5237  pcre_exec(const pcre *argument_re, const Line 5755  pcre_exec(const pcre *argument_re, const
5755    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5756    int offsetcount)    int offsetcount)
5757  {  {
5758  int rc, resetcount, ocount;  int rc, ocount;
5759  int first_byte = -1;  int first_byte = -1;
5760  int req_byte = -1;  int req_byte = -1;
5761  int req_byte2 = -1;  int req_byte2 = -1;
5762  int newline;  int newline;
 unsigned long int ims;  
5763  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5764  BOOL anchored;  BOOL anchored;
5765  BOOL startline;  BOOL startline;
# Line 5272  if ((options & ~PUBLIC_EXEC_OPTIONS) != Line 5789  if ((options & ~PUBLIC_EXEC_OPTIONS) !=
5789  if (re == NULL || subject == NULL ||  if (re == NULL || subject == NULL ||
5790     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;
5791  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
5792    if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
5793    
5794  /* This information is for finding all the numbers associated with a given  /* This information is for finding all the numbers associated with a given
5795  name, for condition testing. */  name, for condition testing. */
# Line 5342  end_subject = md->end_subject; Line 5860  end_subject = md->end_subject;
5860    
5861  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
5862  utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;  utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;
5863    md->use_ucp = (re->options & PCRE_UCP) != 0;
5864  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
5865    
5866    /* Some options are unpacked into BOOL variables in the hope that testing
5867    them will be faster than individual option bits. */
5868    
5869  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
5870  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
5871  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
5872  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
5873  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5874                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5875    
5876    
5877  md->hitend = FALSE;  md->hitend = FALSE;
5878  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
5879    
# Line 5431  defined (though never set). So there's n Line 5955  defined (though never set). So there's n
5955  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
5956    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
5957    
5958  /* 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
5959  back the character offset. */  code for an invalid string if a results vector is available. */
5960    
5961  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5962  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5963    {    {
5964    if (_pcre_valid_utf8((USPTR)subject, length) >= 0)    int erroroffset;
5965      return PCRE_ERROR_BADUTF8;    int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5966    if (start_offset > 0 && start_offset < length)    if (errorcode != 0)
5967      {      {
5968      int tb = ((USPTR)subject)[start_offset];      if (offsetcount >= 2)
     if (tb > 127)  
5969        {        {
5970        tb &= 0xc0;        offsets[0] = erroroffset;
5971        if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET;        offsets[1] = errorcode;
5972        }        }
5973        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5974          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5975      }      }
   }  
 #endif  
5976    
5977  /* The ims options can vary during the matching as a result of the presence    /* Check that a start_offset points to the start of a UTF-8 character. */
 of (?ims) items in the pattern. They are kept in a local variable so that  
 restoring at the exit of a group is easy. */  
5978    
5979  ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL);    if (start_offset > 0 && start_offset < length &&
5980          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
5981        return PCRE_ERROR_BADUTF8_OFFSET;
5982      }
5983    #endif
5984    
5985  /* 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
5986  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.
# Line 5479  md->offset_max = (2*ocount)/3; Line 6004  md->offset_max = (2*ocount)/3;
6004  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
6005  md->capture_last = -1;  md->capture_last = -1;
6006    
 /* Compute the minimum number of offsets that we need to reset each time. Doing  
 this makes a huge difference to execution time when there aren't many brackets  
 in the pattern. */  
   
 resetcount = 2 + re->top_bracket * 2;  
 if (resetcount > offsetcount) resetcount = ocount;  
   
6007  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
6008  never be used unless previously set, but they get saved and restored, and so we  never be used unless previously set, but they get saved and restored, and so we
6009  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
6010    offsets for the matched string. This is really just for tidiness with callouts,
6011    in case they inspect these fields. */
6012    
6013  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
6014    {    {
6015    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6016    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6017      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6018    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6019      md->offset_vector[0] = md->offset_vector[1] = -1;
6020    }    }
6021    
6022  /* Set up the first character to match, if available. The first_byte value is  /* Set up the first character to match, if available. The first_byte value is
# Line 5528  if ((re->flags & PCRE_REQCHSET) != 0) Line 6050  if ((re->flags & PCRE_REQCHSET) != 0)
6050    }    }
6051    
6052    
6053    
6054    
6055  /* ==========================================================================*/  /* ==========================================================================*/
6056    
6057  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 5538  for(;;) Line 6062  for(;;)
6062    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6063    USPTR new_start_match;    USPTR new_start_match;
6064    
   /* Reset the maximum number of extractions we might see. */  
   
   if (md->offset_vector != NULL)  
     {  
     register int *iptr = md->offset_vector;  
     register int *iend = iptr + resetcount;  
     while (iptr < iend) *iptr++ = -1;  
     }  
   
6065    /* If firstline is TRUE, the start of the match is constrained to the first    /* If firstline is TRUE, the start of the match is constrained to the first
6066    line of a multiline string. That is, the match must be before or at the first    line of a multiline string. That is, the match must be before or at the first
6067    newline. Implement this by temporarily adjusting end_subject so that we stop    newline. Implement this by temporarily adjusting end_subject so that we stop
# Line 5574  for(;;) Line 6089  for(;;)
6089    /* There are some optimizations that avoid running the match if a known    /* There are some optimizations that avoid running the match if a known
6090    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.
6091    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
6092    that all callouts do actually occur. */    that all callouts do actually occur. The option can be set in the regex by
6093      (*NO_START_OPT) or passed in match-time options. */
6094    
6095    if ((options & PCRE_NO_START_OPTIMIZE) == 0)    if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0)
6096      {      {
6097      /* Advance to a unique first byte if there is one. */      /* Advance to a unique first byte if there is one. */
6098    
# Line 5630  for(;;) Line 6146  for(;;)
6146        while (start_match < end_subject)        while (start_match < end_subject)
6147          {          {
6148          register unsigned int c = *start_match;          register unsigned int c = *start_match;
6149          if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++;          if ((start_bits[c/8] & (1 << (c&7))) == 0)
6150            else break;            {
6151              start_match++;
6152    #ifdef SUPPORT_UTF8
6153              if (utf8)
6154                while(start_match < end_subject && (*start_match & 0xc0) == 0x80)
6155                  start_match++;
6156    #endif
6157              }
6158            else break;
6159          }          }
6160        }        }
6161      }   /* Starting optimizations */      }   /* Starting optimizations */
# Line 5722  for(;;) Line 6246  for(;;)
6246    
6247    /* OK, we can now run the match. If "hitend" is set afterwards, remember the    /* OK, we can now run the match. If "hitend" is set afterwards, remember the
6248    first starting point for which a partial match was found. */    first starting point for which a partial match was found. */
6249    
6250    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6251    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6252    md->match_call_count = 0;    md->match_call_count = 0;
6253    rc = match(start_match, md->start_code, start_match, NULL, 2, md, ims, NULL,    md->match_function_type = 0;
6254      0, 0);    md->end_offset_top = 0;
6255      rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6256    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
6257    
6258    switch(rc)    switch(rc)
6259      {      {
6260      /* 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
6261      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. */
6262      We treat this as NOMATCH. THEN at this level acts exactly like PRUNE. */  
6263        case MATCH_SKIP:
6264        if (md->start_match_ptr != start_match)
6265          {
6266          new_start_match = md->start_match_ptr;
6267          break;
6268          }
6269        /* Fall through */
6270    
6271        /* If MATCH_SKIP_ARG reaches this level it means that a MARK that matched
6272        the SKIP's arg was not found. We also treat this as NOMATCH. */
6273    
6274        case MATCH_SKIP_ARG:
6275        /* Fall through */
6276    
6277        /* NOMATCH and PRUNE advance by one character. THEN at this level acts
6278        exactly like PRUNE. */
6279    
6280      case MATCH_NOMATCH:      case MATCH_NOMATCH:
6281      case MATCH_PRUNE:      case MATCH_PRUNE:
     case MATCH_SKIP_ARG:  
6282      case MATCH_THEN:      case MATCH_THEN:
6283      new_start_match = start_match + 1;      new_start_match = start_match + 1;
6284  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 5748  for(;;) Line 6288  for(;;)
6288  #endif  #endif
6289      break;      break;
6290    
     /* SKIP passes back the next starting point explicitly. */  
   
     case MATCH_SKIP:  
     new_start_match = md->start_match_ptr;  
     break;  
   
6291      /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */      /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */
6292    
6293      case MATCH_COMMIT:      case MATCH_COMMIT:
# Line 5844  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6378  if (rc == MATCH_MATCH || rc == MATCH_ACC
6378    
6379    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = md->offset_overflow? 0 : md->end_offset_top/2;
6380    
6381      /* If there is space in the offset vector, set any unused pairs at the end of
6382      the pattern to -1 for backwards compatibility. It is documented that this
6383      happens. In earlier versions, the whole set of potential capturing offsets
6384      was set to -1 each time round the loop, but this is handled differently now.
6385      "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only
6386      those at the end that need unsetting here. We can't just unset them all at
6387      the start of the whole thing because they may get set in one branch that is
6388      not the final matching branch. */
6389    
6390      if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL)
6391        {
6392        register int *iptr, *iend;
6393        int resetcount = 2 + re->top_bracket * 2;
6394        if (resetcount > offsetcount) resetcount = ocount;
6395        iptr = offsets + md->end_offset_top;
6396        iend = offsets + resetcount;
6397        while (iptr < iend) *iptr++ = -1;
6398        }
6399    
6400    /* 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
6401    md->start_match_ptr might be modified if \K was encountered on the success    md->start_match_ptr might be modified if \K was encountered on the success
6402    matching path. */    matching path. */
6403    
6404    if (offsetcount < 2) rc = 0; else    if (offsetcount < 2) rc = 0; else
6405      {      {
6406      offsets[0] = md->start_match_ptr - md->start_subject;      offsets[0] = (int)(md->start_match_ptr - md->start_subject);
6407      offsets[1] = md->end_match_ptr - md->start_subject;      offsets[1] = (int)(md->end_match_ptr - md->start_subject);
6408      }      }
6409    
6410    DPRINTF((">>>> returning %d\n", rc));    DPRINTF((">>>> returning %d\n", rc));
# Line 5866  if (using_temporary_offsets) Line 6419  if (using_temporary_offsets)
6419    DPRINTF(("Freeing temporary memory\n"));    DPRINTF(("Freeing temporary memory\n"));
6420    (pcre_free)(md->offset_vector);    (pcre_free)(md->offset_vector);
6421    }    }
6422    
6423  /* For anything other than nomatch or partial match, just return the code. */  /* For anything other than nomatch or partial match, just return the code. */
6424    
6425  if (rc != MATCH_NOMATCH && rc != PCRE_ERROR_PARTIAL)  if (rc != MATCH_NOMATCH && rc != PCRE_ERROR_PARTIAL)
6426    {    {
6427    DPRINTF((">>>> error: returning %d\n", rc));    DPRINTF((">>>> error: returning %d\n", rc));
6428    return rc;    return rc;
6429    }    }
6430    
6431  /* Handle partial matches - disable any mark data */  /* Handle partial matches - disable any mark data */
6432    
6433  if (start_partial != NULL)  if (start_partial != NULL)
6434    {    {
6435    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));
6436    md->mark = NULL;    md->mark = NULL;
6437    if (offsetcount > 1)    if (offsetcount > 1)
6438      {      {
6439      offsets[0] = start_partial - (USPTR)subject;      offsets[0] = (int)(start_partial - (USPTR)subject);
6440      offsets[1] = end_subject - (USPTR)subject;      offsets[1] = (int)(end_subject - (USPTR)subject);
6441      }      }
6442    rc = PCRE_ERROR_PARTIAL;    rc = PCRE_ERROR_PARTIAL;
6443    }    }
6444    
6445  /* This is the classic nomatch case */  /* This is the classic nomatch case */
6446    
6447  else  else
6448    {    {
6449    DPRINTF((">>>> returning PCRE_ERROR_NOMATCH\n"));    DPRINTF((">>>> returning PCRE_ERROR_NOMATCH\n"));
6450    rc = PCRE_ERROR_NOMATCH;    rc = PCRE_ERROR_NOMATCH;
6451    }    }
6452    
6453  /* Return the MARK data if it has been requested. */  /* Return the MARK data if it has been requested. */
6454    
6455  RETURN_MARK:  RETURN_MARK:
6456    
6457  if (extra_data != NULL && (extra_data->flags & PCRE_EXTRA_MARK) != 0)  if (extra_data != NULL && (extra_data->flags & PCRE_EXTRA_MARK) != 0)
6458    *(extra_data->mark) = (unsigned char *)(md->mark);    *(extra_data->mark) = (unsigned char *)(md->mark);
6459  return rc;  return rc;
6460  }  }
6461    
6462  /* End of pcre_exec.c */  /* End of pcre_exec.c */

Legend:
Removed from v.511  
changed lines
  Added in v.645

  ViewVC Help
Powered by ViewVC 1.1.5