/[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 501 by ph10, Sun Mar 7 11:49:54 2010 UTC revision 771 by ph10, Tue Nov 29 15:34:12 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 71  defined PCRE_ERROR_xxx codes, which are Line 73  defined PCRE_ERROR_xxx codes, which are
73  /* Special internal returns from the match() function. Make them sufficiently  /* Special internal returns from the match() function. Make them sufficiently
74  negative to avoid the external error codes. */  negative to avoid the external error codes. */
75    
76  #define MATCH_COMMIT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_PRUNE        (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_SKIP         (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_THEN         (-996)  #define MATCH_ONCE         (-996)
80    #define MATCH_PRUNE        (-995)
81    #define MATCH_SKIP         (-994)
82    #define MATCH_SKIP_ARG     (-993)
83    #define MATCH_THEN         (-992)
84    
85  /* Maximum number of ints of offset to save on the stack for recursive calls.  /* Maximum number of ints of offset to save on the stack for recursive calls.
86  If the offset vector is bigger, malloc is used. This should be a multiple of 3,  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
# Line 122  while (length-- > 0) Line 128  while (length-- > 0)
128  *          Match a back-reference                *  *          Match a back-reference                *
129  *************************************************/  *************************************************/
130    
131  /* 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
132  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
133    the length passed is zero. Note that in caseless UTF-8 mode, the number of
134    subject bytes matched may be different to the number of reference bytes.
135    
136  Arguments:  Arguments:
137    offset      index into the offset vector    offset      index into the offset vector
138    eptr        points into the subject    eptr        pointer into the subject
139    length      length to be matched    length      length of reference to be matched (number of bytes)
140    md          points to match data block    md          points to match data block
141    ims         the ims flags    caseless    TRUE if caseless
142    
143  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
144  */  */
145    
146  static BOOL  static int
147  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
148    unsigned long int ims)    BOOL caseless)
149  {  {
150  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
151    register USPTR p = md->start_subject + md->offset_vector[offset];
152    
153  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
154  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 154  pchars(p, length, FALSE, md); Line 163  pchars(p, length, FALSE, md);
163  printf("\n");  printf("\n");
164  #endif  #endif
165    
166  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
167    
168  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
169    
170  /* 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
171  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
172  ASCII characters. */  ASCII characters. */
173    
174  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
175    {    {
176  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
177  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
178    if (md->utf8)    if (md->utf8)
179      {      {
180      USPTR endptr = eptr + length;      /* Match characters up to the end of the reference. NOTE: the number of
181      while (eptr < endptr)      bytes matched may differ, because there are some characters whose upper and
182        lower case versions code as different numbers of bytes. For example, U+023A
183        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
184        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
185        the latter. It is important, therefore, to check the length along the
186        reference, not along the subject (earlier code did this wrong). */
187    
188        USPTR endptr = p + length;
189        while (p < endptr)
190        {        {
191        int c, d;        int c, d;
192          if (eptr >= md->end_subject) return -1;
193        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
194        GETCHARINC(d, p);        GETCHARINC(d, p);
195        if (c != d && c != UCD_OTHERCASE(d)) return FALSE;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
196        }        }
197      }      }
198    else    else
# Line 183  if ((ims & PCRE_CASELESS) != 0) Line 201  if ((ims & PCRE_CASELESS) != 0)
201    
202    /* 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
203    is no UCP support. */    is no UCP support. */
204        {
205    while (length-- > 0)      if (eptr + length > md->end_subject) return -1;
206      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }      while (length-- > 0)
207          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
208        }
209    }    }
210    
211  /* 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
212  are in UTF-8 mode. */  are in UTF-8 mode. */
213    
214  else  else
215    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
216      if (eptr + length > md->end_subject) return -1;
217      while (length-- > 0) if (*p++ != *eptr++) return -1;
218      }
219    
220  return TRUE;  return eptr - eptr_start;
221  }  }
222    
223    
# Line 245  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 268  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
268         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
269         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
270         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
271         RM51,  RM52, RM53, RM54 };         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
272           RM61,  RM62, RM63, RM64, RM65, RM66 };
273    
274  /* 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
275  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 255  actually used in this definition. */ Line 279  actually used in this definition. */
279  #define REGISTER register  #define REGISTER register
280    
281  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
282  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
283    { \    { \
284    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
285    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \
286    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
287    }    }
288  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 267  actually used in this definition. */ Line 291  actually used in this definition. */
291    return ra; \    return ra; \
292    }    }
293  #else  #else
294  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
295    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1)
296  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
297  #endif  #endif
298    
# Line 281  argument of match(), which never changes Line 305  argument of match(), which never changes
305    
306  #define REGISTER  #define REGISTER
307    
308  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
309    {\    {\
310    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
311      if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
312    frame->Xwhere = rw; \    frame->Xwhere = rw; \
313    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
314    newframe->Xecode = rb;\    newframe->Xecode = rb;\
315    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
   newframe->Xmarkptr = markptr;\  
316    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
317    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
318    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
319    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
320    frame = newframe;\    frame = newframe;\
# Line 304  argument of match(), which never changes Line 326  argument of match(), which never changes
326    
327  #define RRETURN(ra)\  #define RRETURN(ra)\
328    {\    {\
329    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
330    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
331    (pcre_stack_free)(newframe);\    (pcre_stack_free)(oldframe);\
332    if (frame != NULL)\    if (frame != NULL)\
333      {\      {\
334      rrc = ra;\      rrc = ra;\
# Line 326  typedef struct heapframe { Line 348  typedef struct heapframe {
348    USPTR Xeptr;    USPTR Xeptr;
349    const uschar *Xecode;    const uschar *Xecode;
350    USPTR Xmstart;    USPTR Xmstart;
   USPTR Xmarkptr;  
351    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
352    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
353    unsigned int Xrdepth;    unsigned int Xrdepth;
354    
355    /* Function local variables */    /* Function local variables */
# Line 351  typedef struct heapframe { Line 370  typedef struct heapframe {
370    BOOL Xcondition;    BOOL Xcondition;
371    BOOL Xprev_is_word;    BOOL Xprev_is_word;
372    
   unsigned long int Xoriginal_ims;  
   
373  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
374    int Xprop_type;    int Xprop_type;
375    int Xprop_value;    int Xprop_value;
376    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
377    int Xoclength;    int Xoclength;
378    uschar Xocchars[8];    uschar Xocchars[8];
379  #endif  #endif
# Line 410  immediately. The second one is used when Line 424  immediately. The second one is used when
424  the subject. */  the subject. */
425    
426  #define CHECK_PARTIAL()\  #define CHECK_PARTIAL()\
427    if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\    if (md->partial != 0 && eptr >= md->end_subject && \
428      {\        eptr > md->start_used_ptr) \
429      md->hitend = TRUE;\      { \
430      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\      md->hitend = TRUE; \
431        if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
432      }      }
433    
434  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
435    if (md->partial != 0 && eptr > mstart)\    if (md->partial != 0 && eptr > md->start_used_ptr) \
436      {\      { \
437      md->hitend = TRUE;\      md->hitend = TRUE; \
438      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
439      }      }
440    
441    
# Line 434  Arguments: Line 449  Arguments:
449     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
450     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
451                   by encountering \K)                   by encountering \K)
    markptr     pointer to the most recent MARK name, or NULL  
452     offset_top  current top pointer     offset_top  current top pointer
453     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
454     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
455                   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  
456     rdepth      the recursion depth     rdepth      the recursion depth
457    
458  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
459                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
460                   a negative MATCH_xxx value for PRUNE, SKIP, etc
461                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
462                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
463  */  */
464    
465  static int  static int
466  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, USPTR  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
467    markptr, int offset_top, match_data *md, unsigned long int ims,    int offset_top, match_data *md, eptrblock *eptrb, unsigned int rdepth)
   eptrblock *eptrb, int flags, unsigned int rdepth)  
468  {  {
469  /* 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,
470  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 467  register unsigned int c;   /* Character Line 476  register unsigned int c;   /* Character
476  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
477    
478  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
479    BOOL caseless;
480  int condcode;  int condcode;
481    
482  /* 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 475  heap storage. Set up the top-level frame Line 485  heap storage. Set up the top-level frame
485  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
486    
487  #ifdef NO_RECURSE  #ifdef NO_RECURSE
488  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
489    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
490  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
491    
492  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 483  frame->Xprevframe = NULL;            /* Line 494  frame->Xprevframe = NULL;            /*
494  frame->Xeptr = eptr;  frame->Xeptr = eptr;
495  frame->Xecode = ecode;  frame->Xecode = ecode;
496  frame->Xmstart = mstart;  frame->Xmstart = mstart;
 frame->Xmarkptr = markptr;  
497  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
498  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
499  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
500    
501  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 499  HEAP_RECURSE: Line 507  HEAP_RECURSE:
507  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
508  #define ecode              frame->Xecode  #define ecode              frame->Xecode
509  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
 #define markptr            frame->Xmarkptr  
510  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
511  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
512  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
513    
514  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 525  HEAP_RECURSE: Line 530  HEAP_RECURSE:
530  #define condition          frame->Xcondition  #define condition          frame->Xcondition
531  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
532    
 #define original_ims       frame->Xoriginal_ims  
   
533  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
534  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
535  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
536  #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  
537  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
538  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
539  #endif  #endif
# Line 563  i, and fc and c, can be the same variabl Line 563  i, and fc and c, can be the same variabl
563  #define fi i  #define fi i
564  #define fc c  #define fc c
565    
566    /* Many of the following variables are used only in small blocks of the code.
567    My normal style of coding would have declared them within each of those blocks.
568    However, in order to accommodate the version of this code that uses an external
569    "stack" implemented on the heap, it is easier to declare them all here, so the
570    declarations can be cut out in a block. The only declarations within blocks
571    below are for variables that do not have to be preserved over a recursive call
572    to RMATCH(). */
573    
574    #ifdef SUPPORT_UTF8
575    const uschar *charptr;
576    #endif
577    const uschar *callpat;
578    const uschar *data;
579    const uschar *next;
580    USPTR         pp;
581    const uschar *prev;
582    USPTR         saved_eptr;
583    
584    recursion_info new_recursive;
585    
586  #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().          */  
587  BOOL condition;  BOOL condition;
588  BOOL prev_is_word;  BOOL prev_is_word;
589    
 unsigned long int original_ims;  
   
590  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
591  int prop_type;  int prop_type;
592  int prop_value;  int prop_value;
593  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
594  int oclength;  int oclength;
595  uschar occhars[8];  uschar occhars[8];
596  #endif  #endif
# Line 608  int stacksave[REC_STACK_SAVE_MAX]; Line 610  int stacksave[REC_STACK_SAVE_MAX];
610  eptrblock newptrb;  eptrblock newptrb;
611  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
612    
613    /* To save space on the stack and in the heap frame, I have doubled up on some
614    of the local variables that are used only in localised parts of the code, but
615    still need to be preserved over recursive calls of match(). These macros define
616    the alternative names that are used. */
617    
618    #define allow_zero    cur_is_word
619    #define cbegroup      condition
620    #define code_offset   codelink
621    #define condassert    condition
622    #define matched_once  prev_is_word
623    
624  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
625  variables. */  variables. */
626    
# Line 644  haven't exceeded the recursive call limi Line 657  haven't exceeded the recursive call limi
657  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
658  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
659    
 original_ims = ims;    /* Save for resetting on ')' */  
   
660  /* 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
661  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
662  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
663  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
664  When match() is called in other circumstances, don't add to the chain. The  
665  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
666  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
667  match(). */  to break infinite loops that match no characters. When match() is called in
668    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
669    NOT be used with tail recursion, because the memory block that is used is on
670    the stack, so a new one may be required for each match(). */
671    
672  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
673    {    {
674    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
675    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
676    eptrb = &newptrb;    eptrb = &newptrb;
677      md->match_function_type = 0;
678    }    }
679    
680  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 671  for (;;) Line 686  for (;;)
686    
687    switch(op)    switch(op)
688      {      {
689        case OP_MARK:
690        md->nomatch_mark = ecode + 2;
691        md->mark = NULL;    /* In case previously set by assertion */
692        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
693          eptrb, RM55);
694        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
695             md->mark == NULL) md->mark = ecode + 2;
696    
697        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
698        argument, and we must check whether that argument matches this MARK's
699        argument. It is passed back in md->start_match_ptr (an overloading of that
700        variable). If it does match, we reset that variable to the current subject
701        position and return MATCH_SKIP. Otherwise, pass back the return code
702        unaltered. */
703    
704        else if (rrc == MATCH_SKIP_ARG &&
705            strcmp((char *)(ecode + 2), (char *)(md->start_match_ptr)) == 0)
706          {
707          md->start_match_ptr = eptr;
708          RRETURN(MATCH_SKIP);
709          }
710        RRETURN(rrc);
711    
712      case OP_FAIL:      case OP_FAIL:
713      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
714    
715      case OP_PRUNE:      /* COMMIT overrides PRUNE, SKIP, and THEN */
     RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,  
       ims, eptrb, flags, RM51);  
     if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
     RRETURN(MATCH_PRUNE);  
716    
717      case OP_COMMIT:      case OP_COMMIT:
718      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
719        ims, eptrb, flags, RM52);        eptrb, RM52);
720      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
721            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
722            rrc != MATCH_THEN)
723          RRETURN(rrc);
724      RRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
725    
726        /* PRUNE overrides THEN */
727    
728        case OP_PRUNE:
729        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
730          eptrb, RM51);
731        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
732        RRETURN(MATCH_PRUNE);
733    
734        case OP_PRUNE_ARG:
735        md->nomatch_mark = ecode + 2;
736        md->mark = NULL;    /* In case previously set by assertion */
737        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
738          eptrb, RM56);
739        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
740             md->mark == NULL) md->mark = ecode + 2;
741        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
742        RRETURN(MATCH_PRUNE);
743    
744        /* SKIP overrides PRUNE and THEN */
745    
746      case OP_SKIP:      case OP_SKIP:
747      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
748        ims, eptrb, flags, RM53);        eptrb, RM53);
749      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
750          RRETURN(rrc);
751      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
752      RRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
753    
754        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
755        nomatch_mark. There is a flag that disables this opcode when re-matching a
756        pattern that ended with a SKIP for which there was not a matching MARK. */
757    
758        case OP_SKIP_ARG:
759        if (md->ignore_skip_arg)
760          {
761          ecode += _pcre_OP_lengths[*ecode] + ecode[1];
762          break;
763          }
764        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
765          eptrb, RM57);
766        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
767          RRETURN(rrc);
768    
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
771        caught by a matching MARK, or get to the top, where it causes a rematch
772        with the md->ignore_skip_arg flag set. */
773    
774        md->start_match_ptr = ecode + 2;
775        RRETURN(MATCH_SKIP_ARG);
776    
777        /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
778        the branch in which it occurs can be determined. Overload the start of
779        match pointer to do this. */
780    
781      case OP_THEN:      case OP_THEN:
782      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
783        ims, eptrb, flags, RM54);        eptrb, RM54);
784      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
785        md->start_match_ptr = ecode;
786      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
787    
788      /* Handle a capturing bracket. If there is space in the offset vector, save      case OP_THEN_ARG:
789      the current subject position in the working slot at the top of the vector.      md->nomatch_mark = ecode + 2;
790      We mustn't change the current values of the data slot, because they may be      md->mark = NULL;    /* In case previously set by assertion */
791      set from a previous iteration of this group, and be referred to by a      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top,
792      reference inside the group.        md, eptrb, RM58);
793        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
794      If the bracket fails to match, we need to restore this value and also the           md->mark == NULL) md->mark = ecode + 2;
795      values of the final offsets, in case they were set by a previous iteration      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
796      of the same bracket.      md->start_match_ptr = ecode;
797        RRETURN(MATCH_THEN);
798    
799        /* Handle an atomic group that does not contain any capturing parentheses.
800        This can be handled like an assertion. Prior to 8.13, all atomic groups
801        were handled this way. In 8.13, the code was changed as below for ONCE, so
802        that backups pass through the group and thereby reset captured values.
803        However, this uses a lot more stack, so in 8.20, atomic groups that do not
804        contain any captures generate OP_ONCE_NC, which can be handled in the old,
805        less stack intensive way.
806    
807        Check the alternative branches in turn - the matching won't pass the KET
808        for this kind of subpattern. If any one branch matches, we carry on as at
809        the end of a normal bracket, leaving the subject pointer, but resetting
810        the start-of-match value in case it was changed by \K. */
811    
812        case OP_ONCE_NC:
813        prev = ecode;
814        saved_eptr = eptr;
815        do
816          {
817          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
818          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
819            {
820            mstart = md->start_match_ptr;
821            break;
822            }
823          if (rrc == MATCH_THEN)
824            {
825            next = ecode + GET(ecode,1);
826            if (md->start_match_ptr < next &&
827                (*ecode == OP_ALT || *next == OP_ALT))
828              rrc = MATCH_NOMATCH;
829            }
830    
831          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
832          ecode += GET(ecode,1);
833          }
834        while (*ecode == OP_ALT);
835    
836        /* If hit the end of the group (which could be repeated), fail */
837    
838        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
839    
840        /* Continue as from after the group, updating the offsets high water
841        mark, since extracts may have been taken. */
842    
843        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
844    
845        offset_top = md->end_offset_top;
846        eptr = md->end_match_ptr;
847    
848        /* For a non-repeating ket, just continue at this level. This also
849        happens for a repeating ket if no characters were matched in the group.
850        This is the forcible breaking of infinite loops as implemented in Perl
851        5.005. */
852    
853        if (*ecode == OP_KET || eptr == saved_eptr)
854          {
855          ecode += 1+LINK_SIZE;
856          break;
857          }
858    
859        /* The repeating kets try the rest of the pattern or restart from the
860        preceding bracket, in the appropriate order. The second "call" of match()
861        uses tail recursion, to avoid using another stack frame. */
862    
863        if (*ecode == OP_KETRMIN)
864          {
865          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
866          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
867          ecode = prev;
868          goto TAIL_RECURSE;
869          }
870        else  /* OP_KETRMAX */
871          {
872          md->match_function_type = MATCH_CBEGROUP;
873          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
874          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
875          ecode += 1 + LINK_SIZE;
876          goto TAIL_RECURSE;
877          }
878        /* Control never gets here */
879    
880        /* Handle a capturing bracket, other than those that are possessive with an
881        unlimited repeat. If there is space in the offset vector, save the current
882        subject position in the working slot at the top of the vector. We mustn't
883        change the current values of the data slot, because they may be set from a
884        previous iteration of this group, and be referred to by a reference inside
885        the group. A failure to match might occur after the group has succeeded,
886        if something later on doesn't match. For this reason, we need to restore
887        the working value and also the values of the final offsets, in case they
888        were set by a previous iteration of the same bracket.
889    
890      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
891      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 733  for (;;) Line 911  for (;;)
911        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
912    
913        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
914        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
915            (int)(eptr - md->start_subject);
916    
917        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
918          {          {
919            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
920          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
921            ims, eptrb, flags, RM1);            eptrb, RM1);
922          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
923    
924            /* If we backed up to a THEN, check whether it is within the current
925            branch by comparing the address of the THEN that is passed back with
926            the end of the branch. If it is within the current branch, and the
927            branch is one of two or more alternatives (it either starts or ends
928            with OP_ALT), we have reached the limit of THEN's action, so convert
929            the return code to NOMATCH, which will cause normal backtracking to
930            happen from now on. Otherwise, THEN is passed back to an outer
931            alternative. This implements Perl's treatment of parenthesized groups,
932            where a group not containing | does not affect the current alternative,
933            that is, (X) is NOT the same as (X|(*F)). */
934    
935            if (rrc == MATCH_THEN)
936              {
937              next = ecode + GET(ecode,1);
938              if (md->start_match_ptr < next &&
939                  (*ecode == OP_ALT || *next == OP_ALT))
940                rrc = MATCH_NOMATCH;
941              }
942    
943            /* Anything other than NOMATCH is passed back. */
944    
945            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
946          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
947          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
948            if (*ecode != OP_ALT) break;
949          }          }
       while (*ecode == OP_ALT);  
950    
951        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
952        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
953        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
954        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
955    
956        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
957    
958          RRETURN(rrc);
959        }        }
960    
961      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 766  for (;;) Line 969  for (;;)
969      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
970      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
971    
972      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
973      final alternative within the brackets, we would return the result of a      repeat and ONCE group with no captures. Loop for all the alternatives.
     recursive call to match() whatever happened. We can reduce stack usage by  
     turning this into a tail recursion, except in the case when match_cbegroup  
     is set.*/  
974    
975        When we get to the final alternative within the brackets, we used to return
976        the result of a recursive call to match() whatever happened so it was
977        possible to reduce stack usage by turning this into a tail recursion,
978        except in the case of a possibly empty group. However, now that there is
979        the possiblity of (*THEN) occurring in the final alternative, this
980        optimization is no longer always possible.
981    
982        We can optimize if we know there are no (*THEN)s in the pattern; at present
983        this is the best that can be done.
984    
985        MATCH_ONCE is returned when the end of an atomic group is successfully
986        reached, but subsequent matching fails. It passes back up the tree (causing
987        captured values to be reset) until the original atomic group level is
988        reached. This is tested by comparing md->once_target with the start of the
989        group. At this point, the return is converted into MATCH_NOMATCH so that
990        previous backup points can be taken. */
991    
992        case OP_ONCE:
993      case OP_BRA:      case OP_BRA:
994      case OP_SBRA:      case OP_SBRA:
995      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
996      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
997      for (;;)      for (;;)
998        {        {
999        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
1000    
1001          /* If this is not a possibly empty group, and there are no (*THEN)s in
1002          the pattern, and this is the final alternative, optimize as described
1003          above. */
1004    
1005          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1006            {
1007            ecode += _pcre_OP_lengths[*ecode];
1008            goto TAIL_RECURSE;
1009            }
1010    
1011          /* In all other cases, we have to make another call to match(). */
1012    
1013          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
1014            RM2);
1015    
1016          /* See comment in the code for capturing groups above about handling
1017          THEN. */
1018    
1019          if (rrc == MATCH_THEN)
1020          {          {
1021          if (flags == 0)    /* Not a possibly empty group */          next = ecode + GET(ecode,1);
1022            if (md->start_match_ptr < next &&
1023                (*ecode == OP_ALT || *next == OP_ALT))
1024              rrc = MATCH_NOMATCH;
1025            }
1026    
1027          if (rrc != MATCH_NOMATCH)
1028            {
1029            if (rrc == MATCH_ONCE)
1030            {            {
1031            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
1032            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1033            goto TAIL_RECURSE;              {
1034                while (*scode == OP_ALT) scode += GET(scode, 1);
1035                scode -= GET(scode, 1);
1036                }
1037              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1038            }            }
1039            RRETURN(rrc);
1040            }
1041          ecode += GET(ecode, 1);
1042          if (*ecode != OP_ALT) break;
1043          }
1044    
1045          /* Possibly empty group; can't use tail recursion. */      RRETURN(MATCH_NOMATCH);
1046    
1047          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* Handle possessive capturing brackets with an unlimited repeat. We come
1048            eptrb, flags, RM48);      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1049          RRETURN(rrc);      handled similarly to the normal case above. However, the matching is
1050        different. The end of these brackets will always be OP_KETRPOS, which
1051        returns MATCH_KETRPOS without going further in the pattern. By this means
1052        we can handle the group by iteration rather than recursion, thereby
1053        reducing the amount of stack needed. */
1054    
1055        case OP_CBRAPOS:
1056        case OP_SCBRAPOS:
1057        allow_zero = FALSE;
1058    
1059        POSSESSIVE_CAPTURE:
1060        number = GET2(ecode, 1+LINK_SIZE);
1061        offset = number << 1;
1062    
1063    #ifdef PCRE_DEBUG
1064        printf("start possessive bracket %d\n", number);
1065        printf("subject=");
1066        pchars(eptr, 16, TRUE, md);
1067        printf("\n");
1068    #endif
1069    
1070        if (offset < md->offset_max)
1071          {
1072          matched_once = FALSE;
1073          code_offset = ecode - md->start_code;
1074    
1075          save_offset1 = md->offset_vector[offset];
1076          save_offset2 = md->offset_vector[offset+1];
1077          save_offset3 = md->offset_vector[md->offset_end - number];
1078          save_capture_last = md->capture_last;
1079    
1080          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1081    
1082          /* Each time round the loop, save the current subject position for use
1083          when the group matches. For MATCH_MATCH, the group has matched, so we
1084          restart it with a new subject starting position, remembering that we had
1085          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1086          usual. If we haven't matched any alternatives in any iteration, check to
1087          see if a previous iteration matched. If so, the group has matched;
1088          continue from afterwards. Otherwise it has failed; restore the previous
1089          capture values before returning NOMATCH. */
1090    
1091          for (;;)
1092            {
1093            md->offset_vector[md->offset_end - number] =
1094              (int)(eptr - md->start_subject);
1095            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1096            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1097              eptrb, RM63);
1098            if (rrc == MATCH_KETRPOS)
1099              {
1100              offset_top = md->end_offset_top;
1101              eptr = md->end_match_ptr;
1102              ecode = md->start_code + code_offset;
1103              save_capture_last = md->capture_last;
1104              matched_once = TRUE;
1105              continue;
1106              }
1107    
1108            /* See comment in the code for capturing groups above about handling
1109            THEN. */
1110    
1111            if (rrc == MATCH_THEN)
1112              {
1113              next = ecode + GET(ecode,1);
1114              if (md->start_match_ptr < next &&
1115                  (*ecode == OP_ALT || *next == OP_ALT))
1116                rrc = MATCH_NOMATCH;
1117              }
1118    
1119            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1120            md->capture_last = save_capture_last;
1121            ecode += GET(ecode, 1);
1122            if (*ecode != OP_ALT) break;
1123          }          }
1124    
1125        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1126        otherwise return. */          {
1127            md->offset_vector[offset] = save_offset1;
1128            md->offset_vector[offset+1] = save_offset2;
1129            md->offset_vector[md->offset_end - number] = save_offset3;
1130            }
1131    
1132        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (allow_zero || matched_once)
1133          eptrb, flags, RM2);          {
1134        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          ecode += 1 + LINK_SIZE;
1135            break;
1136            }
1137    
1138          RRETURN(MATCH_NOMATCH);
1139          }
1140    
1141        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1142        as a non-capturing bracket. */
1143    
1144        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1145        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1146    
1147        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1148    
1149        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1150        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1151    
1152        /* Non-capturing possessive bracket with unlimited repeat. We come here
1153        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1154        without the capturing complication. It is written out separately for speed
1155        and cleanliness. */
1156    
1157        case OP_BRAPOS:
1158        case OP_SBRAPOS:
1159        allow_zero = FALSE;
1160    
1161        POSSESSIVE_NON_CAPTURE:
1162        matched_once = FALSE;
1163        code_offset = ecode - md->start_code;
1164    
1165        for (;;)
1166          {
1167          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1168          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1169            eptrb, RM48);
1170          if (rrc == MATCH_KETRPOS)
1171            {
1172            offset_top = md->end_offset_top;
1173            eptr = md->end_match_ptr;
1174            ecode = md->start_code + code_offset;
1175            matched_once = TRUE;
1176            continue;
1177            }
1178    
1179          /* See comment in the code for capturing groups above about handling
1180          THEN. */
1181    
1182          if (rrc == MATCH_THEN)
1183            {
1184            next = ecode + GET(ecode,1);
1185            if (md->start_match_ptr < next &&
1186                (*ecode == OP_ALT || *next == OP_ALT))
1187              rrc = MATCH_NOMATCH;
1188            }
1189    
1190          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1191        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1192          if (*ecode != OP_ALT) break;
1193          }
1194    
1195        if (matched_once || allow_zero)
1196          {
1197          ecode += 1 + LINK_SIZE;
1198          break;
1199        }        }
1200        RRETURN(MATCH_NOMATCH);
1201    
1202      /* Control never reaches here. */      /* Control never reaches here. */
1203    
1204      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1205      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
1206      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
1207      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. */  
1208    
1209      case OP_COND:      case OP_COND:
1210      case OP_SCOND:      case OP_SCOND:
1211      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1212    
1213      /* 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
1214      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 822  for (;;) Line 1218  for (;;)
1218        if (pcre_callout != NULL)        if (pcre_callout != NULL)
1219          {          {
1220          pcre_callout_block cb;          pcre_callout_block cb;
1221          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1222          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1223          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1224          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1225          cb.subject_length   = md->end_subject - md->start_subject;          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1226          cb.start_match      = mstart - md->start_subject;          cb.start_match      = (int)(mstart - md->start_subject);
1227          cb.current_position = eptr - md->start_subject;          cb.current_position = (int)(eptr - md->start_subject);
1228          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1229          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1230          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1231          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1232          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1233            cb.mark             = md->nomatch_mark;
1234          if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);          if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1235          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1236          }          }
# Line 854  for (;;) Line 1251  for (;;)
1251        else        else
1252          {          {
1253          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1254          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1255    
1256          /* If the test is for recursion into a specific subpattern, and it is          /* If the test is for recursion into a specific subpattern, and it is
1257          false, but the test was set up by name, scan the table to see if the          false, but the test was set up by name, scan the table to see if the
1258          name refers to any other numbers, and test them. The condition is true          name refers to any other numbers, and test them. The condition is true
1259          if any one is set. */          if any one is set. */
1260    
1261          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1262            {            {
1263            uschar *slotA = md->name_table;            uschar *slotA = md->name_table;
1264            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
# Line 987  for (;;) Line 1384  for (;;)
1384        }        }
1385    
1386      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1387      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
1388      assertion. */      an assertion. */
1389    
1390      else      else
1391        {        {
1392        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1393            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1394        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1395          {          {
1396            if (md->end_offset_top > offset_top)
1397              offset_top = md->end_offset_top;  /* Captures may have happened */
1398          condition = TRUE;          condition = TRUE;
1399          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1400          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1401          }          }
1402    
1403          /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1404          assertion; it is therefore treated as NOMATCH. */
1405    
1406        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1407          {          {
1408          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
# Line 1011  for (;;) Line 1414  for (;;)
1414          }          }
1415        }        }
1416    
1417      /* 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, can
1418      we can use tail recursion to avoid using another stack frame, except when      use tail recursion to avoid using another stack frame, except when there is
1419      match_cbegroup is required for an unlimited repeat of a possibly empty      unlimited repeat of a possibly empty group. In the latter case, a recursive
1420      group. If the second alternative doesn't exist, we can just plough on. */      call to match() is always required, unless the second alternative doesn't
1421        exist, in which case we can just plough on. Note that, for compatibility
1422        with Perl, the | in a conditional group is NOT treated as creating two
1423        alternatives. If a THEN is encountered in the branch, it propagates out to
1424        the enclosing alternative (unless nested in a deeper set of alternatives,
1425        of course). */
1426    
1427      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1428        {        {
1429        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
1430          {          {
1431          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          ecode += 1 + LINK_SIZE;
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
1432          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1433          }          }
1434    
1435          md->match_function_type = MATCH_CBEGROUP;
1436          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1437          RRETURN(rrc);
1438        }        }
1439      else                         /* Condition false & no alternative */  
1440         /* Condition false & no alternative; continue after the group. */
1441    
1442        else
1443        {        {
1444        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1445        }        }
# Line 1054  for (;;) Line 1463  for (;;)
1463        {        {
1464        md->offset_vector[offset] =        md->offset_vector[offset] =
1465          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1466        md->offset_vector[offset+1] = eptr - md->start_subject;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1467        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1468        }        }
1469      ecode += 3;      ecode += 3;
1470      break;      break;
1471    
1472    
1473      /* 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. */  
1474    
     case OP_ACCEPT:  
1475      case OP_END:      case OP_END:
1476      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1477        {      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;  
       }  
1478    
1479      /* 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
1480      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
1481      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,
1482      if any. */      backtracking will then try other alternatives, if any. */
1483    
1484      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1485          (md->notempty ||           md->recursive == NULL &&
1486            (md->notempty_atstart &&           (md->notempty ||
1487              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1488                 mstart == md->start_subject + md->start_offset)))
1489        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1490    
1491      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1096  for (;;) Line 1493  for (;;)
1493      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1494      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1495      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
     RRETURN(MATCH_MATCH);  
1496    
1497      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1498        given as the argument to RRETURN when the heap is in use. */
1499    
1500      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1501      ims = ecode[1];      RRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1502    
1503      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1504      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,
1505      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
1506      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
1507      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1508        of a condition, we want to return immediately afterwards. The caller of
1509        this incarnation of the match() function will have set MATCH_CONDASSERT in
1510        md->match_function type, and one of these opcodes will be the first opcode
1511        that is processed. We use a local variable that is preserved over calls to
1512        match() to remember this case. */
1513    
1514      case OP_ASSERT:      case OP_ASSERT:
1515      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1516        if (md->match_function_type == MATCH_CONDASSERT)
1517          {
1518          condassert = TRUE;
1519          md->match_function_type = 0;
1520          }
1521        else condassert = FALSE;
1522    
1523      do      do
1524        {        {
1525        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1526          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
       if (rrc == MATCH_MATCH)  
1527          {          {
1528          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1529          break;          break;
1530          }          }
1531    
1532          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1533          as NOMATCH. */
1534    
1535        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1536        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1537        }        }
1538      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1539    
1540      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1541    
1542      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1543    
1544      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1545    
1546      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1547      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1147  for (;;) Line 1557  for (;;)
1557    
1558      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1559      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1560        if (md->match_function_type == MATCH_CONDASSERT)
1561          {
1562          condassert = TRUE;
1563          md->match_function_type = 0;
1564          }
1565        else condassert = FALSE;
1566    
1567      do      do
1568        {        {
1569        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1570          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
       if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);  
1571        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1572          {          {
1573          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1574          break;          break;
1575          }          }
1576    
1577          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1578          as NOMATCH. */
1579    
1580        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1581        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1582        }        }
1583      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1584    
1585      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1586    
1587      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1588      continue;      continue;
# Line 1208  for (;;) Line 1628  for (;;)
1628      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1629        {        {
1630        pcre_callout_block cb;        pcre_callout_block cb;
1631        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1632        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1633        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1634        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1635        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1636        cb.start_match      = mstart - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1637        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1638        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1639        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1640        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1641        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1642        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1643          cb.mark             = md->nomatch_mark;
1644        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1645        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1646        }        }
# Line 1230  for (;;) Line 1651  for (;;)
1651      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
1652      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1653    
1654      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1655      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
1656      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
1657      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
1658      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
1659      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
1660      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.  
1661    
1662      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
1663      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
1664      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1665        a lot, so he is not to blame for the current way it works. */
1666    
1667      case OP_RECURSE:      case OP_RECURSE:
1668        {        {
1669          recursion_info *ri;
1670          int recno;
1671    
1672        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1673        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1674          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1675    
1676          /* Check for repeating a recursion without advancing the subject pointer.
1677          This should catch convoluted mutual recursions. (Some simple cases are
1678          caught at compile time.) */
1679    
1680          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1681            if (recno == ri->group_num && eptr == ri->subject_position)
1682              RRETURN(PCRE_ERROR_RECURSELOOP);
1683    
1684        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1685    
1686          new_recursive.group_num = recno;
1687          new_recursive.subject_position = eptr;
1688        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1689        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1690    
1691        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1692    
1693        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1694    
1695        /* Now save the offset data. */        /* Now save the offset data */
1696    
1697        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1698        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1272  for (;;) Line 1703  for (;;)
1703            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1704          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1705          }          }
   
1706        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1707              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1708    
1709        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1710        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1711          might be changed, so reset it before looping. */
1712    
1713        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1714        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1715        do        do
1716          {          {
1717            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1718          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1719            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1720          if (rrc == MATCH_MATCH)          memcpy(md->offset_vector, new_recursive.offset_save,
1721                new_recursive.saved_max * sizeof(int));
1722            md->recursive = new_recursive.prevrec;
1723            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1724            {            {
1725            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1726            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1727              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1728            RRETURN(MATCH_MATCH);  
1729              /* Set where we got to in the subject, and reset the start in case
1730              it was changed by \K. This *is* propagated back out of a recursion,
1731              for Perl compatibility. */
1732    
1733              eptr = md->end_match_ptr;
1734              mstart = md->start_match_ptr;
1735              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1736            }            }
1737    
1738            /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1739            as NOMATCH. */
1740    
1741          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1742            {            {
1743            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
# Line 1303  for (;;) Line 1747  for (;;)
1747            }            }
1748    
1749          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1750          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1751          }          }
1752        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1315  for (;;) Line 1757  for (;;)
1757          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1758        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1759        }        }
     /* 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)  
         {  
         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);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
1760    
1761      offset_top = md->end_offset_top;      RECURSION_MATCHED:
1762      eptr = md->end_match_ptr;      break;
   
     /* 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 */  
1763    
1764      /* 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
1765      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1409  for (;;) Line 1774  for (;;)
1774      with fixed upper repeat limits are compiled as a number of copies, with the      with fixed upper repeat limits are compiled as a number of copies, with the
1775      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1776    
1777      case OP_BRAZERO:      case OP_BRAZERO:
1778        {      next = ecode + 1;
1779        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1780        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1781        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1782        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1783      break;      break;
1784    
1785      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1786        {      next = ecode + 1;
1787        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1788        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1789        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1790        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1791      break;      break;
1792    
1793      case OP_SKIPZERO:      case OP_SKIPZERO:
1794        {      next = ecode+1;
1795        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1796        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1797      break;      break;
1798    
1799        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1800        here; just jump to the group, with allow_zero set TRUE. */
1801    
1802        case OP_BRAPOSZERO:
1803        op = *(++ecode);
1804        allow_zero = TRUE;
1805        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1806          goto POSSESSIVE_NON_CAPTURE;
1807    
1808      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1809    
1810      case OP_KET:      case OP_KET:
1811      case OP_KETRMIN:      case OP_KETRMIN:
1812      case OP_KETRMAX:      case OP_KETRMAX:
1813        case OP_KETRPOS:
1814      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1815    
1816      /* 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
1817      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1818      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1819    
1820      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1821        {        {
1822        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1823        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1824        }        }
1825      else saved_eptr = NULL;      else saved_eptr = NULL;
1826    
1827      /* 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 or a non-capturing atomic
1828      matching and return MATCH_MATCH, but record the current high water mark for      group, stop matching and return MATCH_MATCH, but record the current high
1829      use by positive assertions. We also need to record the match start in case      water mark for use by positive assertions. We also need to record the match
1830      it was changed by \K. */      start in case it was changed by \K. */
1831    
1832      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1833          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||           *prev == OP_ONCE_NC)
         *prev == OP_ONCE)  
1834        {        {
1835        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1836        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1837        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1838        RRETURN(MATCH_MATCH);        RRETURN(MATCH_MATCH);         /* Sets md->mark */
1839        }        }
1840    
1841      /* 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
1842      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1843      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
1844      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
1845      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
1846        the current subject position and start match pointer and give a MATCH
1847        return. */
1848    
1849      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1850            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1851        {        {
1852        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1853        offset = number << 1;        offset = number << 1;
# Line 1486  for (;;) Line 1857  for (;;)
1857        printf("\n");        printf("\n");
1858  #endif  #endif
1859    
1860          /* Handle a recursively called group. */
1861    
1862          if (md->recursive != NULL && md->recursive->group_num == number)
1863            {
1864            md->end_match_ptr = eptr;
1865            md->start_match_ptr = mstart;
1866            RRETURN(MATCH_MATCH);
1867            }
1868    
1869          /* Deal with capturing */
1870    
1871        md->capture_last = number;        md->capture_last = number;
1872        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1873          {          {
1874            /* If offset is greater than offset_top, it means that we are
1875            "skipping" a capturing group, and that group's offsets must be marked
1876            unset. In earlier versions of PCRE, all the offsets were unset at the
1877            start of matching, but this doesn't work because atomic groups and
1878            assertions can cause a value to be set that should later be unset.
1879            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1880            part of the atomic group, but this is not on the final matching path,
1881            so must be unset when 2 is set. (If there is no group 2, there is no
1882            problem, because offset_top will then be 2, indicating no capture.) */
1883    
1884            if (offset > offset_top)
1885              {
1886              register int *iptr = md->offset_vector + offset_top;
1887              register int *iend = md->offset_vector + offset;
1888              while (iptr < iend) *iptr++ = -1;
1889              }
1890    
1891            /* Now make the extraction */
1892    
1893          md->offset_vector[offset] =          md->offset_vector[offset] =
1894            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1895          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1896          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1897          }          }
1898          }
1899    
1900        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1901        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1902        group. This is the forcible breaking of infinite loops as implemented in
1903        Perl 5.005. For a non-repeating atomic group that includes captures,
1904        establish a backup point by processing the rest of the pattern at a lower
1905        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1906        original OP_ONCE level, thereby bypassing intermediate backup points, but
1907        resetting any captures that happened along the way. */
1908    
1909        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1910          {
1911          if (*prev == OP_ONCE)
1912          {          {
1913          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1914          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1915          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1916          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;  
1917          }          }
1918          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1919          break;
1920        }        }
1921    
1922      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1923      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1924        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
   
     /* 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. */  
1925    
1926      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRPOS)
1927        {        {
1928        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1929        break;        md->end_offset_top = offset_top;
1930          RRETURN(MATCH_KETRPOS);
1931        }        }
1932    
1933      /* 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
1934      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
1935      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
1936      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
1937        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1938    
1939      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1940        {        {
1941        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1942        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1943        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1944            {
1945            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1946            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1947            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1948            RRETURN(MATCH_ONCE);
1949            }
1950          if (*prev >= OP_SBRA)    /* Could match an empty string */
1951          {          {
1952          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1953            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1954          RRETURN(rrc);          RRETURN(rrc);
1955          }          }
1956        ecode = prev;        ecode = prev;
# Line 1551  for (;;) Line 1958  for (;;)
1958        }        }
1959      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1960        {        {
1961        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1962          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1963          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1964        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1965          if (*prev == OP_ONCE)
1966            {
1967            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1968            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1969            md->once_target = prev;
1970            RRETURN(MATCH_ONCE);
1971            }
1972        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1973        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1974        }        }
1975      /* Control never gets here */      /* Control never gets here */
1976    
1977      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1978    
1979      case OP_CIRC:      case OP_CIRC:
1980      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
     if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         RRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
1981    
1982      /* Start of subject assertion */      /* Start of subject assertion */
1983    
# Line 1580  for (;;) Line 1986  for (;;)
1986      ecode++;      ecode++;
1987      break;      break;
1988    
1989        /* Multiline mode: start of subject unless notbol, or after any newline. */
1990    
1991        case OP_CIRCM:
1992        if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
1993        if (eptr != md->start_subject &&
1994            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1995          RRETURN(MATCH_NOMATCH);
1996        ecode++;
1997        break;
1998    
1999      /* Start of match assertion */      /* Start of match assertion */
2000    
2001      case OP_SOM:      case OP_SOM:
# Line 1594  for (;;) Line 2010  for (;;)
2010      ecode++;      ecode++;
2011      break;      break;
2012    
2013      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2014      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2015    
2016      case OP_DOLL:      case OP_DOLLM:
2017      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2018        {        { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }
       if (eptr < md->end_subject)  
         { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }  
       else  
         { if (md->noteol) RRETURN(MATCH_NOMATCH); }  
       ecode++;  
       break;  
       }  
2019      else      else
2020        {        {
2021        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2022        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
2023        }        }
2024        ecode++;
2025        break;
2026    
2027        /* Not multiline mode: assert before a terminating newline or before end of
2028        subject unless noteol is set. */
2029    
2030        case OP_DOLL:
2031        if (md->noteol) RRETURN(MATCH_NOMATCH);
2032        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2033    
2034      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2035    
2036      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2037    
2038      case OP_EOD:      case OP_EOD:
2039      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2040        SCHECK_PARTIAL();
2041      ecode++;      ecode++;
2042      break;      break;
2043    
2044      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
2045    
2046      case OP_EODN:      case OP_EODN:
2047      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
2048        if (eptr < md->end_subject &&
2049          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2050        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2051    
2052        /* Either at end of string or \n before end. */
2053    
2054        SCHECK_PARTIAL();
2055      ecode++;      ecode++;
2056      break;      break;
2057    
# Line 1651  for (;;) Line 2069  for (;;)
2069  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2070        if (utf8)        if (utf8)
2071          {          {
2072            /* Get status of previous character */
2073    
2074          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2075            {            {
2076            USPTR lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
2077            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
2078            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2079            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2080    #ifdef SUPPORT_UCP
2081              if (md->use_ucp)
2082                {
2083                if (c == '_') prev_is_word = TRUE; else
2084                  {
2085                  int cat = UCD_CATEGORY(c);
2086                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2087                  }
2088                }
2089              else
2090    #endif
2091            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2092            }            }
2093    
2094            /* Get status of next character */
2095    
2096          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2097            {            {
2098            SCHECK_PARTIAL();            SCHECK_PARTIAL();
# Line 1667  for (;;) Line 2101  for (;;)
2101          else          else
2102            {            {
2103            GETCHAR(c, eptr);            GETCHAR(c, eptr);
2104    #ifdef SUPPORT_UCP
2105              if (md->use_ucp)
2106                {
2107                if (c == '_') cur_is_word = TRUE; else
2108                  {
2109                  int cat = UCD_CATEGORY(c);
2110                  cur_is_word = (cat == ucp_L || cat == ucp_N);
2111                  }
2112                }
2113              else
2114    #endif
2115            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2116            }            }
2117          }          }
2118        else        else
2119  #endif  #endif
2120    
2121        /* Not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
2122          consistency with the behaviour of \w we do use it in this case. */
2123    
2124          {          {
2125            /* Get status of previous character */
2126    
2127          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2128            {            {
2129            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
2130    #ifdef SUPPORT_UCP
2131              if (md->use_ucp)
2132                {
2133                c = eptr[-1];
2134                if (c == '_') prev_is_word = TRUE; else
2135                  {
2136                  int cat = UCD_CATEGORY(c);
2137                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2138                  }
2139                }
2140              else
2141    #endif
2142            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2143            }            }
2144    
2145            /* Get status of next character */
2146    
2147          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2148            {            {
2149            SCHECK_PARTIAL();            SCHECK_PARTIAL();
2150            cur_is_word = FALSE;            cur_is_word = FALSE;
2151            }            }
2152          else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          else
2153    #ifdef SUPPORT_UCP
2154            if (md->use_ucp)
2155              {
2156              c = *eptr;
2157              if (c == '_') cur_is_word = TRUE; else
2158                {
2159                int cat = UCD_CATEGORY(c);
2160                cur_is_word = (cat == ucp_L || cat == ucp_N);
2161                }
2162              }
2163            else
2164    #endif
2165            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2166          }          }
2167    
2168        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
# Line 1704  for (;;) Line 2180  for (;;)
2180      /* Fall through */      /* Fall through */
2181    
2182      case OP_ALLANY:      case OP_ALLANY:
2183      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2184        {        {                            /* not be updated before SCHECK_PARTIAL. */
2185        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2186        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2187        }        }
2188        eptr++;
2189      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2190      ecode++;      ecode++;
2191      break;      break;
# Line 1717  for (;;) Line 2194  for (;;)
2194      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2195    
2196      case OP_ANYBYTE:      case OP_ANYBYTE:
2197      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2198        {        {                            /* not be updated before SCHECK_PARTIAL. */
2199        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2200        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2201        }        }
2202        eptr++;
2203      ecode++;      ecode++;
2204      break;      break;
2205    
# Line 1837  for (;;) Line 2315  for (;;)
2315      switch(c)      switch(c)
2316        {        {
2317        default: RRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2318    
2319        case 0x000d:        case 0x000d:
2320        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2321        break;        break;
# Line 1993  for (;;) Line 2472  for (;;)
2472               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2473               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2474            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2475           break;          break;
2476    
2477          case PT_GC:          case PT_GC:
2478          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
# Line 2010  for (;;) Line 2489  for (;;)
2489            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2490          break;          break;
2491    
2492            /* These are specials */
2493    
2494            case PT_ALNUM:
2495            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2496                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2497              RRETURN(MATCH_NOMATCH);
2498            break;
2499    
2500            case PT_SPACE:    /* Perl space */
2501            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2502                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2503                   == (op == OP_NOTPROP))
2504              RRETURN(MATCH_NOMATCH);
2505            break;
2506    
2507            case PT_PXSPACE:  /* POSIX space */
2508            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2509                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2510                 c == CHAR_FF || c == CHAR_CR)
2511                   == (op == OP_NOTPROP))
2512              RRETURN(MATCH_NOMATCH);
2513            break;
2514    
2515            case PT_WORD:
2516            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2517                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2518                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2519              RRETURN(MATCH_NOMATCH);
2520            break;
2521    
2522            /* This should never occur */
2523    
2524          default:          default:
2525          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2526          }          }
# Line 2028  for (;;) Line 2539  for (;;)
2539        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2540        }        }
2541      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2542        if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH);
2543        while (eptr < md->end_subject)
2544        {        {
2545        int category = UCD_CATEGORY(c);        int len = 1;
2546        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2547        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2548          {        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;  
         }  
2549        }        }
2550      ecode++;      ecode++;
2551      break;      break;
# Line 2057  for (;;) Line 2561  for (;;)
2561      loops). */      loops). */
2562    
2563      case OP_REF:      case OP_REF:
2564        {      case OP_REFI:
2565        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2566        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2567        ecode += 3;
2568    
2569        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2570    
2571        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2572        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
2573        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.  
2574    
2575        (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
2576        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2577    
2578        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
2579        referenced subpattern. */      referenced subpattern. */
2580    
2581        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2582          length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;        length = (md->jscript_compat)? 0 : -1;
2583        else      else
2584          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2585    
2586        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2587    
2588        switch (*ecode)      switch (*ecode)
2589          {        {
2590          case OP_CRSTAR:        case OP_CRSTAR:
2591          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2592          case OP_CRPLUS:        case OP_CRPLUS:
2593          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2594          case OP_CRQUERY:        case OP_CRQUERY:
2595          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2596          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2597          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2598          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2599          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2600          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2601          break;        break;
2602    
2603          case OP_CRRANGE:        case OP_CRRANGE:
2604          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2605          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2606          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2607          max = GET2(ecode, 3);        max = GET2(ecode, 3);
2608          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2609          ecode += 5;        ecode += 5;
2610          break;        break;
2611    
2612          default:               /* No repeat follows */        default:               /* No repeat follows */
2613          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2614            {          {
2615            CHECK_PARTIAL();          CHECK_PARTIAL();
2616            RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
           }  
         eptr += length;  
         continue;              /* With the main loop */  
2617          }          }
2618          eptr += length;
2619          continue;              /* With the main loop */
2620          }
2621    
2622        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2623        main loop. */      zero, just continue with the main loop. */
2624    
2625        if (length == 0) continue;      if (length == 0) continue;
2626    
2627        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2628        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2629        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2630    
2631        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2632          {
2633          int slength;
2634          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2635          {          {
2636          if (!match_ref(offset, eptr, length, md, ims))          CHECK_PARTIAL();
2637            {          RRETURN(MATCH_NOMATCH);
           CHECK_PARTIAL();  
           RRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2638          }          }
2639          eptr += slength;
2640          }
2641    
2642        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2643        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2644    
2645        if (min == max) continue;      if (min == max) continue;
2646    
2647        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2648    
2649        if (minimize)      if (minimize)
2650          {
2651          for (fi = min;; fi++)
2652          {          {
2653          for (fi = min;; fi++)          int slength;
2654            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2655            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2656            if (fi >= max) RRETURN(MATCH_NOMATCH);
2657            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2658            {            {
2659            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2660            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            RRETURN(MATCH_NOMATCH);
           if (fi >= max) RRETURN(MATCH_NOMATCH);  
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             RRETURN(MATCH_NOMATCH);  
             }  
           eptr += length;  
2661            }            }
2662          /* Control never gets here */          eptr += slength;
2663          }          }
2664          /* Control never gets here */
2665          }
2666    
2667        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2668    
2669        else      else
2670          {
2671          pp = eptr;
2672          for (i = min; i < max; i++)
2673          {          {
2674          pp = eptr;          int slength;
2675          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
           {  
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             break;  
             }  
           eptr += length;  
           }  
         while (eptr >= pp)  
2676            {            {
2677            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2678            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2679            }            }
2680          RRETURN(MATCH_NOMATCH);          eptr += slength;
2681            }
2682          while (eptr >= pp)
2683            {
2684            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2685            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2686            eptr -= length;
2687          }          }
2688          RRETURN(MATCH_NOMATCH);
2689        }        }
2690      /* Control never gets here */      /* Control never gets here */
2691    
# Line 2285  for (;;) Line 2791  for (;;)
2791            {            {
2792            for (fi = min;; fi++)            for (fi = min;; fi++)
2793              {              {
2794              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2795              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2796              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2797              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2310  for (;;) Line 2816  for (;;)
2816            {            {
2817            for (fi = min;; fi++)            for (fi = min;; fi++)
2818              {              {
2819              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2820              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2821              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2822              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2356  for (;;) Line 2862  for (;;)
2862              }              }
2863            for (;;)            for (;;)
2864              {              {
2865              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2866              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2867              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2868              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2379  for (;;) Line 2885  for (;;)
2885              }              }
2886            while (eptr >= pp)            while (eptr >= pp)
2887              {              {
2888              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2889              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2890              eptr--;              eptr--;
2891              }              }
# Line 2455  for (;;) Line 2961  for (;;)
2961          {          {
2962          for (fi = min;; fi++)          for (fi = min;; fi++)
2963            {            {
2964            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2965            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2966            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
2967            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2488  for (;;) Line 2994  for (;;)
2994            }            }
2995          for(;;)          for(;;)
2996            {            {
2997            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2998            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2999            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3000            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2533  for (;;) Line 3039  for (;;)
3039    
3040      /* Match a single character, caselessly */      /* Match a single character, caselessly */
3041    
3042      case OP_CHARNC:      case OP_CHARI:
3043  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3044      if (utf8)      if (utf8)
3045        {        {
# Line 2593  for (;;) Line 3099  for (;;)
3099      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3100    
3101      case OP_EXACT:      case OP_EXACT:
3102        case OP_EXACTI:
3103      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3104      ecode += 3;      ecode += 3;
3105      goto REPEATCHAR;      goto REPEATCHAR;
3106    
3107      case OP_POSUPTO:      case OP_POSUPTO:
3108        case OP_POSUPTOI:
3109      possessive = TRUE;      possessive = TRUE;
3110      /* Fall through */      /* Fall through */
3111    
3112      case OP_UPTO:      case OP_UPTO:
3113        case OP_UPTOI:
3114      case OP_MINUPTO:      case OP_MINUPTO:
3115        case OP_MINUPTOI:
3116      min = 0;      min = 0;
3117      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3118      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3119      ecode += 3;      ecode += 3;
3120      goto REPEATCHAR;      goto REPEATCHAR;
3121    
3122      case OP_POSSTAR:      case OP_POSSTAR:
3123        case OP_POSSTARI:
3124      possessive = TRUE;      possessive = TRUE;
3125      min = 0;      min = 0;
3126      max = INT_MAX;      max = INT_MAX;
# Line 2617  for (;;) Line 3128  for (;;)
3128      goto REPEATCHAR;      goto REPEATCHAR;
3129    
3130      case OP_POSPLUS:      case OP_POSPLUS:
3131        case OP_POSPLUSI:
3132      possessive = TRUE;      possessive = TRUE;
3133      min = 1;      min = 1;
3134      max = INT_MAX;      max = INT_MAX;
# Line 2624  for (;;) Line 3136  for (;;)
3136      goto REPEATCHAR;      goto REPEATCHAR;
3137    
3138      case OP_POSQUERY:      case OP_POSQUERY:
3139        case OP_POSQUERYI:
3140      possessive = TRUE;      possessive = TRUE;
3141      min = 0;      min = 0;
3142      max = 1;      max = 1;
# Line 2631  for (;;) Line 3144  for (;;)
3144      goto REPEATCHAR;      goto REPEATCHAR;
3145    
3146      case OP_STAR:      case OP_STAR:
3147        case OP_STARI:
3148      case OP_MINSTAR:      case OP_MINSTAR:
3149        case OP_MINSTARI:
3150      case OP_PLUS:      case OP_PLUS:
3151        case OP_PLUSI:
3152      case OP_MINPLUS:      case OP_MINPLUS:
3153        case OP_MINPLUSI:
3154      case OP_QUERY:      case OP_QUERY:
3155        case OP_QUERYI:
3156      case OP_MINQUERY:      case OP_MINQUERY:
3157      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3158        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3159      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3160      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3161      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3162      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2661  for (;;) Line 3179  for (;;)
3179          {          {
3180  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3181          unsigned int othercase;          unsigned int othercase;
3182          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3183              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3184            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3185          else oclength = 0;          else oclength = 0;
# Line 2689  for (;;) Line 3207  for (;;)
3207            {            {
3208            for (fi = min;; fi++)            for (fi = min;; fi++)
3209              {              {
3210              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3211              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3212              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3213              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2731  for (;;) Line 3249  for (;;)
3249    
3250            for(;;)            for(;;)
3251              {              {
3252              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3253              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3254              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3255  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2768  for (;;) Line 3286  for (;;)
3286      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3287        max, eptr));        max, eptr));
3288    
3289      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3290        {        {
3291        fc = md->lcc[fc];        fc = md->lcc[fc];
3292        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2785  for (;;) Line 3303  for (;;)
3303          {          {
3304          for (fi = min;; fi++)          for (fi = min;; fi++)
3305            {            {
3306            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3307            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3308            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3309            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2815  for (;;) Line 3333  for (;;)
3333    
3334          while (eptr >= pp)          while (eptr >= pp)
3335            {            {
3336            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3337            eptr--;            eptr--;
3338            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3339            }            }
# Line 2844  for (;;) Line 3362  for (;;)
3362          {          {
3363          for (fi = min;; fi++)          for (fi = min;; fi++)
3364            {            {
3365            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3366            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3367            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3368            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2873  for (;;) Line 3391  for (;;)
3391    
3392          while (eptr >= pp)          while (eptr >= pp)
3393            {            {
3394            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3395            eptr--;            eptr--;
3396            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3397            }            }
# Line 2886  for (;;) Line 3404  for (;;)
3404      checking can be multibyte. */      checking can be multibyte. */
3405    
3406      case OP_NOT:      case OP_NOT:
3407        case OP_NOTI:
3408      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3409        {        {
3410        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 2893  for (;;) Line 3412  for (;;)
3412        }        }
3413      ecode++;      ecode++;
3414      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3415      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3416        {        {
3417  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3418        if (c < 256)        if (c < 256)
# Line 2901  for (;;) Line 3420  for (;;)
3420        c = md->lcc[c];        c = md->lcc[c];
3421        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);
3422        }        }
3423      else      else    /* Caseful */
3424        {        {
3425        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);
3426        }        }
# Line 2915  for (;;) Line 3434  for (;;)
3434      about... */      about... */
3435    
3436      case OP_NOTEXACT:      case OP_NOTEXACT:
3437        case OP_NOTEXACTI:
3438      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3439      ecode += 3;      ecode += 3;
3440      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3441    
3442      case OP_NOTUPTO:      case OP_NOTUPTO:
3443        case OP_NOTUPTOI:
3444      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3445        case OP_NOTMINUPTOI:
3446      min = 0;      min = 0;
3447      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3448      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3449      ecode += 3;      ecode += 3;
3450      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3451    
3452      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3453        case OP_NOTPOSSTARI:
3454      possessive = TRUE;      possessive = TRUE;
3455      min = 0;      min = 0;
3456      max = INT_MAX;      max = INT_MAX;
# Line 2935  for (;;) Line 3458  for (;;)
3458      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3459    
3460      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3461        case OP_NOTPOSPLUSI:
3462      possessive = TRUE;      possessive = TRUE;
3463      min = 1;      min = 1;
3464      max = INT_MAX;      max = INT_MAX;
# Line 2942  for (;;) Line 3466  for (;;)
3466      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3467    
3468      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3469        case OP_NOTPOSQUERYI:
3470      possessive = TRUE;      possessive = TRUE;
3471      min = 0;      min = 0;
3472      max = 1;      max = 1;
# Line 2949  for (;;) Line 3474  for (;;)
3474      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3475    
3476      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3477        case OP_NOTPOSUPTOI:
3478      possessive = TRUE;      possessive = TRUE;
3479      min = 0;      min = 0;
3480      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 2956  for (;;) Line 3482  for (;;)
3482      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3483    
3484      case OP_NOTSTAR:      case OP_NOTSTAR:
3485        case OP_NOTSTARI:
3486      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3487        case OP_NOTMINSTARI:
3488      case OP_NOTPLUS:      case OP_NOTPLUS:
3489        case OP_NOTPLUSI:
3490      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3491        case OP_NOTMINPLUSI:
3492      case OP_NOTQUERY:      case OP_NOTQUERY:
3493        case OP_NOTQUERYI:
3494      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3495      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3496        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3497      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3498      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3499      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 2983  for (;;) Line 3515  for (;;)
3515      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,
3516        max, eptr));        max, eptr));
3517    
3518      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3519        {        {
3520        fc = md->lcc[fc];        fc = md->lcc[fc];
3521    
# Line 3031  for (;;) Line 3563  for (;;)
3563            register unsigned int d;            register unsigned int d;
3564            for (fi = min;; fi++)            for (fi = min;; fi++)
3565              {              {
3566              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3567              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3568              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3569              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3050  for (;;) Line 3582  for (;;)
3582            {            {
3583            for (fi = min;; fi++)            for (fi = min;; fi++)
3584              {              {
3585              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3586              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3587              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3588              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3091  for (;;) Line 3623  for (;;)
3623          if (possessive) continue;          if (possessive) continue;
3624          for(;;)          for(;;)
3625              {              {
3626              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3627              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3628              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3629              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3114  for (;;) Line 3646  for (;;)
3646            if (possessive) continue;            if (possessive) continue;
3647            while (eptr >= pp)            while (eptr >= pp)
3648              {              {
3649              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3650              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3651              eptr--;              eptr--;
3652              }              }
# Line 3171  for (;;) Line 3703  for (;;)
3703            register unsigned int d;            register unsigned int d;
3704            for (fi = min;; fi++)            for (fi = min;; fi++)
3705              {              {
3706              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3707              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3708              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3709              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3189  for (;;) Line 3721  for (;;)
3721            {            {
3722            for (fi = min;; fi++)            for (fi = min;; fi++)
3723              {              {
3724              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3725              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3726              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3727              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3229  for (;;) Line 3761  for (;;)
3761            if (possessive) continue;            if (possessive) continue;
3762            for(;;)            for(;;)
3763              {              {
3764              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3765              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3766              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3767              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3252  for (;;) Line 3784  for (;;)
3784            if (possessive) continue;            if (possessive) continue;
3785            while (eptr >= pp)            while (eptr >= pp)
3786              {              {
3787              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3788              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3789              eptr--;              eptr--;
3790              }              }
# Line 3344  for (;;) Line 3876  for (;;)
3876      is tidier. Also separate the UCP code, which can be the same for both UTF-8      is tidier. Also separate the UCP code, which can be the same for both UTF-8
3877      and single-bytes. */      and single-bytes. */
3878    
3879      if (min > 0)      if (min > 0)
3880        {        {
3881  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3882        if (prop_type >= 0)        if (prop_type >= 0)
3883          {          {
3884          switch(prop_type)          switch(prop_type)
3885            {            {
3886            case PT_ANY:            case PT_ANY:
3887            if (prop_fail_result) RRETURN(MATCH_NOMATCH);            if (prop_fail_result) RRETURN(MATCH_NOMATCH);
3888              for (i = 1; i <= min; i++)
3889                {
3890                if (eptr >= md->end_subject)
3891                  {
3892                  SCHECK_PARTIAL();
3893                  RRETURN(MATCH_NOMATCH);
3894                  }
3895                GETCHARINCTEST(c, eptr);
3896                }
3897              break;
3898    
3899              case PT_LAMP:
3900              for (i = 1; i <= min; i++)
3901                {
3902                int chartype;
3903                if (eptr >= md->end_subject)
3904                  {
3905                  SCHECK_PARTIAL();
3906                  RRETURN(MATCH_NOMATCH);
3907                  }
3908                GETCHARINCTEST(c, eptr);
3909                chartype = UCD_CHARTYPE(c);
3910                if ((chartype == ucp_Lu ||
3911                     chartype == ucp_Ll ||
3912                     chartype == ucp_Lt) == prop_fail_result)
3913                  RRETURN(MATCH_NOMATCH);
3914                }
3915              break;
3916    
3917              case PT_GC:
3918              for (i = 1; i <= min; i++)
3919                {
3920                if (eptr >= md->end_subject)
3921                  {
3922                  SCHECK_PARTIAL();
3923                  RRETURN(MATCH_NOMATCH);
3924                  }
3925                GETCHARINCTEST(c, eptr);
3926                if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
3927                  RRETURN(MATCH_NOMATCH);
3928                }
3929              break;
3930    
3931              case PT_PC:
3932            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3933              {              {
3934              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3361  for (;;) Line 3937  for (;;)
3937                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3938                }                }
3939              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3940                if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
3941                  RRETURN(MATCH_NOMATCH);
3942              }              }
3943            break;            break;
3944    
3945            case PT_LAMP:            case PT_SC:
3946            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3947              {              {
3948              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3373  for (;;) Line 3951  for (;;)
3951                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3952                }                }
3953              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3954              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == ucp_Lu ||  
                  prop_chartype == ucp_Ll ||  
                  prop_chartype == ucp_Lt) == prop_fail_result)  
3955                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3956              }              }
3957            break;            break;
3958    
3959            case PT_GC:            case PT_ALNUM:
3960            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3961              {              {
3962                int category;
3963              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3964                {                {
3965                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3966                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3967                }                }
3968              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3969              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3970              if ((prop_category == prop_value) == prop_fail_result)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
3971                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3972              }              }
3973            break;            break;
3974    
3975            case PT_PC:            case PT_SPACE:    /* Perl space */
3976            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3977              {              {
3978              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3405  for (;;) Line 3981  for (;;)
3981                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3982                }                }
3983              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3984              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3985              if ((prop_chartype == prop_value) == prop_fail_result)                   c == CHAR_FF || c == CHAR_CR)
3986                       == prop_fail_result)
3987                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3988              }              }
3989            break;            break;
3990    
3991            case PT_SC:            case PT_PXSPACE:  /* POSIX space */
3992              for (i = 1; i <= min; i++)
3993                {
3994                if (eptr >= md->end_subject)
3995                  {
3996                  SCHECK_PARTIAL();
3997                  RRETURN(MATCH_NOMATCH);
3998                  }
3999                GETCHARINCTEST(c, eptr);
4000                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4001                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4002                       == prop_fail_result)
4003                  RRETURN(MATCH_NOMATCH);
4004                }
4005              break;
4006    
4007              case PT_WORD:
4008            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4009              {              {
4010                int category;
4011              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4012                {                {
4013                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4014                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4015                }                }
4016              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4017              prop_script = UCD_SCRIPT(c);              category = UCD_CATEGORY(c);
4018              if ((prop_script == prop_value) == prop_fail_result)              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
4019                       == prop_fail_result)
4020                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4021              }              }
4022            break;            break;
4023    
4024              /* This should not occur */
4025    
4026            default:            default:
4027            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
4028            }            }
# Line 3444  for (;;) Line 4041  for (;;)
4041              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
4042              }              }
4043            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4044            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);  
4045            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4046              {              {
4047              int len = 1;              int len = 1;
4048              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4049                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4050              eptr += len;              eptr += len;
4051              }              }
4052            }            }
# Line 3510  for (;;) Line 4104  for (;;)
4104            switch(c)            switch(c)
4105              {              {
4106              default: RRETURN(MATCH_NOMATCH);              default: RRETURN(MATCH_NOMATCH);
4107    
4108              case 0x000d:              case 0x000d:
4109              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4110              break;              break;
# Line 3786  for (;;) Line 4381  for (;;)
4381            switch(*eptr++)            switch(*eptr++)
4382              {              {
4383              default: RRETURN(MATCH_NOMATCH);              default: RRETURN(MATCH_NOMATCH);
4384    
4385              case 0x000d:              case 0x000d:
4386              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4387              break;              break;
4388    
4389              case 0x000a:              case 0x000a:
4390              break;              break;
4391    
# Line 3978  for (;;) Line 4575  for (;;)
4575            case PT_ANY:            case PT_ANY:
4576            for (fi = min;; fi++)            for (fi = min;; fi++)
4577              {              {
4578              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4579              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4580              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
4581              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3986  for (;;) Line 4583  for (;;)
4583                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4584                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4585                }                }
4586              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4587              if (prop_fail_result) RRETURN(MATCH_NOMATCH);              if (prop_fail_result) RRETURN(MATCH_NOMATCH);
4588              }              }
4589            /* Control never gets here */            /* Control never gets here */
# Line 3994  for (;;) Line 4591  for (;;)
4591            case PT_LAMP:            case PT_LAMP:
4592            for (fi = min;; fi++)            for (fi = min;; fi++)
4593              {              {
4594              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              int chartype;
4595                RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4596              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4597              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
4598              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4002  for (;;) Line 4600  for (;;)
4600                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4601                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4602                }                }
4603              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4604              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4605              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4606                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4607                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4608                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4609              }              }
4610            /* Control never gets here */            /* Control never gets here */
# Line 4014  for (;;) Line 4612  for (;;)
4612            case PT_GC:            case PT_GC:
4613            for (fi = min;; fi++)            for (fi = min;; fi++)
4614              {              {
4615              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4616              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4617              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
4618              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4022  for (;;) Line 4620  for (;;)
4620                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4621                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4622                }                }
4623              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4624              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4625                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4626              }              }
4627            /* Control never gets here */            /* Control never gets here */
# Line 4032  for (;;) Line 4629  for (;;)
4629            case PT_PC:            case PT_PC:
4630            for (fi = min;; fi++)            for (fi = min;; fi++)
4631              {              {
4632              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
4633              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4634              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
4635              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4040  for (;;) Line 4637  for (;;)
4637                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4638                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4639                }                }
4640              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4641              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4642                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4643              }              }
4644            /* Control never gets here */            /* Control never gets here */
# Line 4050  for (;;) Line 4646  for (;;)
4646            case PT_SC:            case PT_SC:
4647            for (fi = min;; fi++)            for (fi = min;; fi++)
4648              {              {
4649              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
4650              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4651              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
4652              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4058  for (;;) Line 4654  for (;;)
4654                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4655                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4656                }                }
4657              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4658              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
4659              if ((prop_script == prop_value) == prop_fail_result)                RRETURN(MATCH_NOMATCH);
4660                }
4661              /* Control never gets here */
4662    
4663              case PT_ALNUM:
4664              for (fi = min;; fi++)
4665                {
4666                int category;
4667                RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4668                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4669                if (fi >= max) RRETURN(MATCH_NOMATCH);
4670                if (eptr >= md->end_subject)
4671                  {
4672                  SCHECK_PARTIAL();
4673                  RRETURN(MATCH_NOMATCH);
4674                  }
4675                GETCHARINCTEST(c, eptr);
4676                category = UCD_CATEGORY(c);
4677                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
4678                  RRETURN(MATCH_NOMATCH);
4679                }
4680              /* Control never gets here */
4681    
4682              case PT_SPACE:    /* Perl space */
4683              for (fi = min;; fi++)
4684                {
4685                RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4686                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4687                if (fi >= max) RRETURN(MATCH_NOMATCH);
4688                if (eptr >= md->end_subject)
4689                  {
4690                  SCHECK_PARTIAL();
4691                  RRETURN(MATCH_NOMATCH);
4692                  }
4693                GETCHARINCTEST(c, eptr);
4694                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4695                     c == CHAR_FF || c == CHAR_CR)
4696                       == prop_fail_result)
4697                  RRETURN(MATCH_NOMATCH);
4698                }
4699              /* Control never gets here */
4700    
4701              case PT_PXSPACE:  /* POSIX space */
4702              for (fi = min;; fi++)
4703                {
4704                RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4705                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4706                if (fi >= max) RRETURN(MATCH_NOMATCH);
4707                if (eptr >= md->end_subject)
4708                  {
4709                  SCHECK_PARTIAL();
4710                  RRETURN(MATCH_NOMATCH);
4711                  }
4712                GETCHARINCTEST(c, eptr);
4713                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4714                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4715                       == prop_fail_result)
4716                  RRETURN(MATCH_NOMATCH);
4717                }
4718              /* Control never gets here */
4719    
4720              case PT_WORD:
4721              for (fi = min;; fi++)
4722                {
4723                int category;
4724                RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4725                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4726                if (fi >= max) RRETURN(MATCH_NOMATCH);
4727                if (eptr >= md->end_subject)
4728                  {
4729                  SCHECK_PARTIAL();
4730                  RRETURN(MATCH_NOMATCH);
4731                  }
4732                GETCHARINCTEST(c, eptr);
4733                category = UCD_CATEGORY(c);
4734                if ((category == ucp_L ||
4735                     category == ucp_N ||
4736                     c == CHAR_UNDERSCORE)
4737                       == prop_fail_result)
4738                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4739              }              }
4740            /* Control never gets here */            /* Control never gets here */
4741    
4742              /* This should never occur */
4743    
4744            default:            default:
4745            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
4746            }            }
# Line 4077  for (;;) Line 4753  for (;;)
4753          {          {
4754          for (fi = min;; fi++)          for (fi = min;; fi++)
4755            {            {
4756            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4757            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4758            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
4759            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4086  for (;;) Line 4762  for (;;)
4762              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
4763              }              }
4764            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4765            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);  
4766            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4767              {              {
4768              int len = 1;              int len = 1;
4769              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4770                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4771              eptr += len;              eptr += len;
4772              }              }
4773            }            }
4774          }          }
   
4775        else        else
4776  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4777    
# Line 4109  for (;;) Line 4781  for (;;)
4781          {          {
4782          for (fi = min;; fi++)          for (fi = min;; fi++)
4783            {            {
4784            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4785            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4786            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
4787            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4272  for (;;) Line 4944  for (;;)
4944          {          {
4945          for (fi = min;; fi++)          for (fi = min;; fi++)
4946            {            {
4947            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4948            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4949            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
4950            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4411  for (;;) Line 5083  for (;;)
5083                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5084                break;                break;
5085                }                }
5086              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5087              if (prop_fail_result) break;              if (prop_fail_result) break;
5088              eptr+= len;              eptr+= len;
5089              }              }
# Line 4420  for (;;) Line 5092  for (;;)
5092            case PT_LAMP:            case PT_LAMP:
5093            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5094              {              {
5095                int chartype;
5096              int len = 1;              int len = 1;
5097              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5098                {                {
5099                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5100                break;                break;
5101                }                }
5102              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5103              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
5104              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
5105                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
5106                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
5107                break;                break;
5108              eptr+= len;              eptr+= len;
5109              }              }
# Line 4445  for (;;) Line 5118  for (;;)
5118                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5119                break;                break;
5120                }                }
5121              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5122              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
5123              eptr+= len;              eptr+= len;
5124              }              }
5125            break;            break;
# Line 4462  for (;;) Line 5133  for (;;)
5133                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5134                break;                break;
5135                }                }
5136              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5137              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
5138              eptr+= len;              eptr+= len;
5139              }              }
5140            break;            break;
# Line 4479  for (;;) Line 5148  for (;;)
5148                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5149                break;                break;
5150                }                }
5151              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5152              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
5153              if ((prop_script == prop_value) == prop_fail_result)              eptr+= len;
5154                }
5155              break;
5156    
5157              case PT_ALNUM:
5158              for (i = min; i < max; i++)
5159                {
5160                int category;
5161                int len = 1;
5162                if (eptr >= md->end_subject)
5163                  {
5164                  SCHECK_PARTIAL();
5165                  break;
5166                  }
5167                GETCHARLENTEST(c, eptr, len);
5168                category = UCD_CATEGORY(c);
5169                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
5170                  break;
5171                eptr+= len;
5172                }
5173              break;
5174    
5175              case PT_SPACE:    /* Perl space */
5176              for (i = min; i < max; i++)
5177                {
5178                int len = 1;
5179                if (eptr >= md->end_subject)
5180                  {
5181                  SCHECK_PARTIAL();
5182                  break;
5183                  }
5184                GETCHARLENTEST(c, eptr, len);
5185                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
5186                     c == CHAR_FF || c == CHAR_CR)
5187                     == prop_fail_result)
5188                  break;
5189                eptr+= len;
5190                }
5191              break;
5192    
5193              case PT_PXSPACE:  /* POSIX space */
5194              for (i = min; i < max; i++)
5195                {
5196                int len = 1;
5197                if (eptr >= md->end_subject)
5198                  {
5199                  SCHECK_PARTIAL();
5200                  break;
5201                  }
5202                GETCHARLENTEST(c, eptr, len);
5203                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
5204                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5205                     == prop_fail_result)
5206                  break;
5207                eptr+= len;
5208                }
5209              break;
5210    
5211              case PT_WORD:
5212              for (i = min; i < max; i++)
5213                {
5214                int category;
5215                int len = 1;
5216                if (eptr >= md->end_subject)
5217                  {
5218                  SCHECK_PARTIAL();
5219                  break;
5220                  }
5221                GETCHARLENTEST(c, eptr, len);
5222                category = UCD_CATEGORY(c);
5223                if ((category == ucp_L || category == ucp_N ||
5224                     c == CHAR_UNDERSCORE) == prop_fail_result)
5225                break;                break;
5226              eptr+= len;              eptr+= len;
5227              }              }
5228            break;            break;
5229    
5230              default:
5231              RRETURN(PCRE_ERROR_INTERNAL);
5232            }            }
5233    
5234          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
# Line 4493  for (;;) Line 5236  for (;;)
5236          if (possessive) continue;          if (possessive) continue;
5237          for(;;)          for(;;)
5238            {            {
5239            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5240            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5241            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5242            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4507  for (;;) Line 5250  for (;;)
5250          {          {
5251          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5252            {            {
5253              int len = 1;
5254            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5255              {              {
5256              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5257              break;              break;
5258              }              }
5259            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5260            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5261            if (prop_category == ucp_M) break;            eptr += len;
5262            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5263              {              {
5264              int len = 1;              len = 1;
5265              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5266                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5267              eptr += len;              eptr += len;
5268              }              }
5269            }            }
# Line 4534  for (;;) Line 5274  for (;;)
5274    
5275          for(;;)          for(;;)
5276            {            {
5277            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5278            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5279            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5280            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5281              {              {
             int len = 1;  
5282              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5283                {                {
5284                BACKCHAR(eptr);                BACKCHAR(eptr);
5285                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5286                }                }
5287              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5288              eptr--;              eptr--;
5289              }              }
5290            }            }
# Line 4610  for (;;) Line 5348  for (;;)
5348                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
5349                }                }
5350              }              }
5351            else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */            else
5352                {
5353                eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
5354                SCHECK_PARTIAL();
5355                }
5356            break;            break;
5357    
5358            /* The byte case is the same as non-UTF8 */            /* The byte case is the same as non-UTF8 */
# Line 4818  for (;;) Line 5560  for (;;)
5560            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5561            }            }
5562    
5563          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5564            done (no backing up). Otherwise, match at this position; anything other
5565            than no match is immediately returned. For nomatch, back up one
5566            character, unless we are matching \R and the last thing matched was
5567            \r\n, in which case, back up two bytes. */
5568    
5569          if (possessive) continue;          if (possessive) continue;
5570          for(;;)          for(;;)
5571            {            {
5572            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5573            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5574            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5575            BACKCHAR(eptr);            BACKCHAR(eptr);
5576              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5577                  eptr[-1] == '\r') eptr--;
5578            }            }
5579          }          }
5580        else        else
# Line 5025  for (;;) Line 5773  for (;;)
5773            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5774            }            }
5775    
5776          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5777            done (no backing up). Otherwise, match at this position; anything other
5778            than no match is immediately returned. For nomatch, back up one
5779            character (byte), unless we are matching \R and the last thing matched
5780            was \r\n, in which case, back up two bytes. */
5781    
5782          if (possessive) continue;          if (possessive) continue;
5783          while (eptr >= pp)          while (eptr >= pp)
5784            {            {
5785            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5786            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5787              eptr--;
5788              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5789                  eptr[-1] == '\r') eptr--;
5790            }            }
5791          }          }
5792    
# Line 5071  switch (frame->Xwhere) Line 5825  switch (frame->Xwhere)
5825    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)
5826    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)
5827    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)
5828    LBL(53) LBL(54)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64)
5829      LBL(65) LBL(66)
5830  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5831    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)
5832    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
5833  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5834    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)
5835      LBL(59) LBL(60) LBL(61) LBL(62)
5836  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
5837  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
5838    default:    default:
# Line 5099  Undefine all the macros that were define Line 5855  Undefine all the macros that were define
5855  #undef ecode  #undef ecode
5856  #undef mstart  #undef mstart
5857  #undef offset_top  #undef offset_top
 #undef ims  
5858  #undef eptrb  #undef eptrb
5859  #undef flags  #undef flags
5860    
# Line 5117  Undefine all the macros that were define Line 5872  Undefine all the macros that were define
5872  #undef condition  #undef condition
5873  #undef prev_is_word  #undef prev_is_word
5874    
 #undef original_ims  
   
5875  #undef ctype  #undef ctype
5876  #undef length  #undef length
5877  #undef max  #undef max
# Line 5175  pcre_exec(const pcre *argument_re, const Line 5928  pcre_exec(const pcre *argument_re, const
5928    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5929    int offsetcount)    int offsetcount)
5930  {  {
5931  int rc, resetcount, ocount;  int rc, ocount, arg_offset_max;
5932  int first_byte = -1;  int first_byte = -1;
5933  int req_byte = -1;  int req_byte = -1;
5934  int req_byte2 = -1;  int req_byte2 = -1;
5935  int newline;  int newline;
 unsigned long int ims;  
5936  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5937  BOOL anchored;  BOOL anchored;
5938  BOOL startline;  BOOL startline;
# Line 5210  if ((options & ~PUBLIC_EXEC_OPTIONS) != Line 5962  if ((options & ~PUBLIC_EXEC_OPTIONS) !=
5962  if (re == NULL || subject == NULL ||  if (re == NULL || subject == NULL ||
5963     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;
5964  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
5965    if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
5966    
5967    /* These two settings are used in the code for checking a UTF-8 string that
5968    follows immediately afterwards. Other values in the md block are used only
5969    during "normal" pcre_exec() processing, not when the JIT support is in use,
5970    so they are set up later. */
5971    
5972    utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;
5973    md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5974                  ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5975    
5976    /* Check a UTF-8 string if required. Pass back the character offset and error
5977    code for an invalid string if a results vector is available. */
5978    
5979    #ifdef SUPPORT_UTF8
5980    if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5981      {
5982      int erroroffset;
5983      int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5984      if (errorcode != 0)
5985        {
5986        if (offsetcount >= 2)
5987          {
5988          offsets[0] = erroroffset;
5989          offsets[1] = errorcode;
5990          }
5991        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5992          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5993        }
5994    
5995      /* Check that a start_offset points to the start of a UTF-8 character. */
5996      if (start_offset > 0 && start_offset < length &&
5997          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
5998        return PCRE_ERROR_BADUTF8_OFFSET;
5999      }
6000    #endif
6001    
6002  /* This information is for finding all the numbers associated with a given  /* If the pattern was successfully studied with JIT support, run the JIT
6003  name, for condition testing. */  executable instead of the rest of this function. Most options must be set at
6004    compile time for the JIT code to be usable. Fallback to the normal code path if
6005    an unsupported flag is set. In particular, JIT does not support partial
6006    matching. */
6007    
6008    #ifdef SUPPORT_JIT
6009    if (extra_data != NULL
6010        && (extra_data->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0
6011        && extra_data->executable_jit != NULL
6012        && (extra_data->flags & PCRE_EXTRA_TABLES) == 0
6013        && (options & ~(PCRE_NO_UTF8_CHECK | PCRE_NOTBOL | PCRE_NOTEOL |
6014                        PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART)) == 0)
6015      return _pcre_jit_exec(re, extra_data->executable_jit, subject, length,
6016        start_offset, options, ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0)
6017        ? MATCH_LIMIT : extra_data->match_limit, offsets, offsetcount);
6018    #endif
6019    
6020    /* Carry on with non-JIT matching. This information is for finding all the
6021    numbers associated with a given name, for condition testing. */
6022    
6023  md->name_table = (uschar *)re + re->name_table_offset;  md->name_table = (uschar *)re + re->name_table_offset;
6024  md->name_count = re->name_count;  md->name_count = re->name_count;
# Line 5279  md->end_subject = md->start_subject + le Line 6085  md->end_subject = md->start_subject + le
6085  end_subject = md->end_subject;  end_subject = md->end_subject;
6086    
6087  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
6088  utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
6089  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
6090    md->ignore_skip_arg = FALSE;
6091    
6092    /* Some options are unpacked into BOOL variables in the hope that testing
6093    them will be faster than individual option bits. */
6094    
6095  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
6096  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
6097  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
6098  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
6099  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  
               ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;  
6100  md->hitend = FALSE;  md->hitend = FALSE;
6101    md->mark = md->nomatch_mark = NULL;     /* In case never set */
6102    
6103  md->recursive = NULL;                   /* No recursion at top level */  md->recursive = NULL;                   /* No recursion at top level */
6104    md->hasthen = (re->flags & PCRE_HASTHEN) != 0;
6105    
6106  md->lcc = tables + lcc_offset;  md->lcc = tables + lcc_offset;
6107  md->ctypes = tables + ctypes_offset;  md->ctypes = tables + ctypes_offset;
# Line 5368  defined (though never set). So there's n Line 6179  defined (though never set). So there's n
6179  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
6180    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
6181    
 /* Check a UTF-8 string if required. Unfortunately there's no way of passing  
 back the character offset. */  
   
 #ifdef SUPPORT_UTF8  
 if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  
   {  
   if (_pcre_valid_utf8((USPTR)subject, length) >= 0)  
     return PCRE_ERROR_BADUTF8;  
   if (start_offset > 0 && start_offset < length)  
     {  
     int tb = ((USPTR)subject)[start_offset];  
     if (tb > 127)  
       {  
       tb &= 0xc0;  
       if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET;  
       }  
     }  
   }  
 #endif  
   
 /* The ims options can vary during the matching as a result of the presence  
 of (?ims) items in the pattern. They are kept in a local variable so that  
 restoring at the exit of a group is easy. */  
   
 ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL);  
   
6182  /* 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
6183  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.
6184  Otherwise, we can use the vector supplied, rounding down its size to a multiple  Otherwise, we can use the vector supplied, rounding down its size to a multiple
6185  of 3. */  of 3. */
6186    
6187  ocount = offsetcount - (offsetcount % 3);  ocount = offsetcount - (offsetcount % 3);
6188    arg_offset_max = (2*ocount)/3;
6189    
6190  if (re->top_backref > 0 && re->top_backref >= ocount/3)  if (re->top_backref > 0 && re->top_backref >= ocount/3)
6191    {    {
# Line 5416  md->offset_max = (2*ocount)/3; Line 6202  md->offset_max = (2*ocount)/3;
6202  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
6203  md->capture_last = -1;  md->capture_last = -1;
6204    
 /* 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;  
   
6205  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
6206  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
6207  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
6208    offsets for the matched string. This is really just for tidiness with callouts,
6209    in case they inspect these fields. */
6210    
6211  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
6212    {    {
6213    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6214    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6215      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6216    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6217      md->offset_vector[0] = md->offset_vector[1] = -1;
6218    }    }
6219    
6220  /* 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 5465  if ((re->flags & PCRE_REQCHSET) != 0) Line 6248  if ((re->flags & PCRE_REQCHSET) != 0)
6248    }    }
6249    
6250    
6251    
6252    
6253  /* ==========================================================================*/  /* ==========================================================================*/
6254    
6255  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 5475  for(;;) Line 6260  for(;;)
6260    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6261    USPTR new_start_match;    USPTR new_start_match;
6262    
   /* 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;  
     }  
   
6263    /* 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
6264    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
6265    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 5511  for(;;) Line 6287  for(;;)
6287    /* There are some optimizations that avoid running the match if a known    /* There are some optimizations that avoid running the match if a known
6288    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.
6289    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
6290    that all callouts do actually occur. */    that all callouts do actually occur. The option can be set in the regex by
6291      (*NO_START_OPT) or passed in match-time options. */
6292    
6293    if ((options & PCRE_NO_START_OPTIMIZE) == 0)    if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0)
6294      {      {
6295      /* Advance to a unique first byte if there is one. */      /* Advance to a unique first byte if there is one. */
6296    
# Line 5567  for(;;) Line 6344  for(;;)
6344        while (start_match < end_subject)        while (start_match < end_subject)