/[pcre]/code/branches/pcre16/pcre_exec.c
ViewVC logotype

Diff of /code/branches/pcre16/pcre_exec.c

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

code/trunk/pcre_exec.c revision 459 by ph10, Sun Oct 4 09:21:39 2009 UTC code/branches/pcre16/pcre_exec.c revision 804 by zherczeg, Wed Dec 14 11:18:01 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) 1997-2009 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 89  static const char rep_max[] = { 0, 0, 0, Line 95  static const char rep_max[] = { 0, 0, 0,
95    
96    
97    
98  #ifdef DEBUG  #ifdef PCRE_DEBUG
99  /*************************************************  /*************************************************
100  *        Debugging function to print chars       *  *        Debugging function to print chars       *
101  *************************************************/  *************************************************/
# Line 107  Returns:     nothing Line 113  Returns:     nothing
113  */  */
114    
115  static void  static void
116  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)  pchars(const pcre_uchar *p, int length, BOOL is_subject, match_data *md)
117  {  {
118  unsigned int c;  unsigned int c;
119  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;
# 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 PCRE_PUCHAR eptr, int length, match_data *md,
148    unsigned long int ims)    BOOL caseless)
149  {  {
150  USPTR p = md->start_subject + md->offset_vector[offset];  PCRE_PUCHAR eptr_start = eptr;
151    register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
152    
153  #ifdef DEBUG  #ifdef PCRE_DEBUG
154  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
155    printf("matching subject <null>");    printf("matching subject <null>");
156  else  else
# 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_UTF
177  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
178    if (md->utf8)    if (md->utf)
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        PCRE_PUCHAR 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 (int)(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
276  actuall used in this definition. */  actually used in this definition. */
277    
278  #ifndef NO_RECURSE  #ifndef NO_RECURSE
279  #define REGISTER register  #define REGISTER register
280    
281  #ifdef 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,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  actuall used in this definition. */ Line 291  actuall 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,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 *)(PUBL(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;\
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 303  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);\    (PUBL(stack_free))(oldframe);\
332    if (frame != NULL)\    if (frame != NULL)\
333      {\      {\
334      rrc = ra;\      rrc = ra;\
# Line 322  typedef struct heapframe { Line 345  typedef struct heapframe {
345    
346    /* Function arguments that may change */    /* Function arguments that may change */
347    
348    USPTR Xeptr;    PCRE_PUCHAR Xeptr;
349    const uschar *Xecode;    const pcre_uchar *Xecode;
350    USPTR Xmstart;    PCRE_PUCHAR Xmstart;
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 */
356    
357    USPTR Xcallpat;    PCRE_PUCHAR Xcallpat;
358  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
359    USPTR Xcharptr;    PCRE_PUCHAR Xcharptr;
360  #endif  #endif
361    USPTR Xdata;    PCRE_PUCHAR Xdata;
362    USPTR Xnext;    PCRE_PUCHAR Xnext;
363    USPTR Xpp;    PCRE_PUCHAR Xpp;
364    USPTR Xprev;    PCRE_PUCHAR Xprev;
365    USPTR Xsaved_eptr;    PCRE_PUCHAR Xsaved_eptr;
366    
367    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
368    
# Line 349  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];    pcre_uchar Xocchars[6];
379  #endif  #endif
380    
381    int Xcodelink;    int Xcodelink;
# Line 401  returns a negative (error) response, the Line 417  returns a negative (error) response, the
417  same response. */  same response. */
418    
419  /* These macros pack up tests that are used for partial matching, and which  /* These macros pack up tests that are used for partial matching, and which
420  appears several times in the code. We set the "hit end" flag if the pointer is  appear several times in the code. We set the "hit end" flag if the pointer is
421  at the end of the subject and also past the start of the subject (i.e.  at the end of the subject and also past the start of the subject (i.e.
422  something has been matched). For hard partial matching, we then return  something has been matched). For hard partial matching, we then return
423  immediately. The second one is used when we already know we are past the end of  immediately. The second one is used when we already know we are past the end of
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 && 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    
442  /* Performance note: It might be tempting to extract commonly used fields from  /* Performance note: It might be tempting to extract commonly used fields from
443  the md structure (e.g. utf8, end_subject) into individual variables to improve  the md structure (e.g. utf, end_subject) into individual variables to improve
444  performance. Tests using gcc on a SPARC disproved this; in the first case, it  performance. Tests using gcc on a SPARC disproved this; in the first case, it
445  made performance worse.  made performance worse.
446    
# Line 434  Arguments: Line 451  Arguments:
451                   by encountering \K)                   by encountering \K)
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,  match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode,
467    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
468    int flags, unsigned int rdepth)    unsigned int rdepth)
469  {  {
470  /* 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,
471  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 461  so they can be ordinary variables in all Line 474  so they can be ordinary variables in all
474  register int  rrc;         /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
475  register int  i;           /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
476  register unsigned int c;   /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
477  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf;         /* Local copy of UTF flag for speed */
478    
479  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
480    BOOL caseless;
481  int condcode;  int condcode;
482    
483  /* 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 472  heap storage. Set up the top-level frame Line 486  heap storage. Set up the top-level frame
486  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
487    
488  #ifdef NO_RECURSE  #ifdef NO_RECURSE
489  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));
490    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
491  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
492    
493  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 481  frame->Xeptr = eptr; Line 496  frame->Xeptr = eptr;
496  frame->Xecode = ecode;  frame->Xecode = ecode;
497  frame->Xmstart = mstart;  frame->Xmstart = mstart;
498  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
499  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
500  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
501    
502  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 496  HEAP_RECURSE: Line 509  HEAP_RECURSE:
509  #define ecode              frame->Xecode  #define ecode              frame->Xecode
510  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
511  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
512  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
513  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
514    
515  /* Ditto for the local variables */  /* Ditto for the local variables */
516    
517  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
518  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
519  #endif  #endif
520  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 520  HEAP_RECURSE: Line 531  HEAP_RECURSE:
531  #define condition          frame->Xcondition  #define condition          frame->Xcondition
532  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
533    
 #define original_ims       frame->Xoriginal_ims  
   
534  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
535  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
536  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
537  #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  
538  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
539  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
540  #endif  #endif
# Line 558  i, and fc and c, can be the same variabl Line 564  i, and fc and c, can be the same variabl
564  #define fi i  #define fi i
565  #define fc c  #define fc c
566    
567    /* Many of the following variables are used only in small blocks of the code.
568    My normal style of coding would have declared them within each of those blocks.
569    However, in order to accommodate the version of this code that uses an external
570    "stack" implemented on the heap, it is easier to declare them all here, so the
571    declarations can be cut out in a block. The only declarations within blocks
572    below are for variables that do not have to be preserved over a recursive call
573    to RMATCH(). */
574    
575    #ifdef SUPPORT_UTF
576    const pcre_uchar *charptr;
577    #endif
578    const pcre_uchar *callpat;
579    const pcre_uchar *data;
580    const pcre_uchar *next;
581    PCRE_PUCHAR       pp;
582    const pcre_uchar *prev;
583    PCRE_PUCHAR       saved_eptr;
584    
585  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  recursion_info new_recursive;
586  const uschar *charptr;             /* in small blocks of the code. My normal */  
587  #endif                             /* style of coding would have declared    */  BOOL cur_is_word;
 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().          */  
588  BOOL condition;  BOOL condition;
589  BOOL prev_is_word;  BOOL prev_is_word;
590    
 unsigned long int original_ims;  
   
591  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
592  int prop_type;  int prop_type;
593  int prop_value;  int prop_value;
594  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
595  int oclength;  int oclength;
596  uschar occhars[8];  pcre_uchar occhars[6];
597  #endif  #endif
598    
599  int codelink;  int codelink;
# Line 603  int stacksave[REC_STACK_SAVE_MAX]; Line 611  int stacksave[REC_STACK_SAVE_MAX];
611  eptrblock newptrb;  eptrblock newptrb;
612  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
613    
614    /* To save space on the stack and in the heap frame, I have doubled up on some
615    of the local variables that are used only in localised parts of the code, but
616    still need to be preserved over recursive calls of match(). These macros define
617    the alternative names that are used. */
618    
619    #define allow_zero    cur_is_word
620    #define cbegroup      condition
621    #define code_offset   codelink
622    #define condassert    condition
623    #define matched_once  prev_is_word
624    #define foc           number
625    
626  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
627  variables. */  variables. */
628    
# Line 622  TAIL_RECURSE: Line 642  TAIL_RECURSE:
642  /* OK, now we can get on with the real code of the function. Recursive calls  /* OK, now we can get on with the real code of the function. Recursive calls
643  are specified by the macro RMATCH and RRETURN is used to return. When  are specified by the macro RMATCH and RRETURN is used to return. When
644  NO_RECURSE is *not* defined, these just turn into a recursive call to match()  NO_RECURSE is *not* defined, these just turn into a recursive call to match()
645  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
646  defined). However, RMATCH isn't like a function call because it's quite a  defined). However, RMATCH isn't like a function call because it's quite a
647  complicated macro. It has to be used in one particular way. This shouldn't,  complicated macro. It has to be used in one particular way. This shouldn't,
648  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
649    
650  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
651  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
652  #else  #else
653  utf8 = FALSE;  utf = FALSE;
654  #endif  #endif
655    
656  /* First check that we haven't called match() too many times, or that we  /* First check that we haven't called match() too many times, or that we
# Line 639  haven't exceeded the recursive call limi Line 659  haven't exceeded the recursive call limi
659  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
660  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
661    
 original_ims = ims;    /* Save for resetting on ')' */  
   
662  /* 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
663  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
664  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
665  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
666  When match() is called in other circumstances, don't add to the chain. The  
667  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
668  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
669  match(). */  to break infinite loops that match no characters. When match() is called in
670    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
671    NOT be used with tail recursion, because the memory block that is used is on
672    the stack, so a new one may be required for each match(). */
673    
674  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
675    {    {
676    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
677    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
678    eptrb = &newptrb;    eptrb = &newptrb;
679      md->match_function_type = 0;
680    }    }
681    
682  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 666  for (;;) Line 688  for (;;)
688    
689    switch(op)    switch(op)
690      {      {
691        case OP_MARK:
692        md->nomatch_mark = ecode + 2;
693        md->mark = NULL;    /* In case previously set by assertion */
694        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
695          eptrb, RM55);
696        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
697             md->mark == NULL) md->mark = ecode + 2;
698    
699        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
700        argument, and we must check whether that argument matches this MARK's
701        argument. It is passed back in md->start_match_ptr (an overloading of that
702        variable). If it does match, we reset that variable to the current subject
703        position and return MATCH_SKIP. Otherwise, pass back the return code
704        unaltered. */
705    
706        else if (rrc == MATCH_SKIP_ARG &&
707            STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0)
708          {
709          md->start_match_ptr = eptr;
710          RRETURN(MATCH_SKIP);
711          }
712        RRETURN(rrc);
713    
714      case OP_FAIL:      case OP_FAIL:
715      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
716    
717      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);  
718    
719      case OP_COMMIT:      case OP_COMMIT:
720      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
721        ims, eptrb, flags, RM52);        eptrb, RM52);
722      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
723            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
724            rrc != MATCH_THEN)
725          RRETURN(rrc);
726      RRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
727    
728        /* PRUNE overrides THEN */
729    
730        case OP_PRUNE:
731        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
732          eptrb, RM51);
733        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
734        RRETURN(MATCH_PRUNE);
735    
736        case OP_PRUNE_ARG:
737        md->nomatch_mark = ecode + 2;
738        md->mark = NULL;    /* In case previously set by assertion */
739        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
740          eptrb, RM56);
741        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
742             md->mark == NULL) md->mark = ecode + 2;
743        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
744        RRETURN(MATCH_PRUNE);
745    
746        /* SKIP overrides PRUNE and THEN */
747    
748      case OP_SKIP:      case OP_SKIP:
749      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
750        ims, eptrb, flags, RM53);        eptrb, RM53);
751      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
752          RRETURN(rrc);
753      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
754      RRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
755    
756        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
757        nomatch_mark. There is a flag that disables this opcode when re-matching a
758        pattern that ended with a SKIP for which there was not a matching MARK. */
759    
760        case OP_SKIP_ARG:
761        if (md->ignore_skip_arg)
762          {
763          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
764          break;
765          }
766        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
767          eptrb, RM57);
768        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
769          RRETURN(rrc);
770    
771        /* Pass back the current skip name by overloading md->start_match_ptr and
772        returning the special MATCH_SKIP_ARG return code. This will either be
773        caught by a matching MARK, or get to the top, where it causes a rematch
774        with the md->ignore_skip_arg flag set. */
775    
776        md->start_match_ptr = ecode + 2;
777        RRETURN(MATCH_SKIP_ARG);
778    
779        /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
780        the branch in which it occurs can be determined. Overload the start of
781        match pointer to do this. */
782    
783      case OP_THEN:      case OP_THEN:
784      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
785        ims, eptrb, flags, RM54);        eptrb, RM54);
786        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
787        md->start_match_ptr = ecode;
788        RRETURN(MATCH_THEN);
789    
790        case OP_THEN_ARG:
791        md->nomatch_mark = ecode + 2;
792        md->mark = NULL;    /* In case previously set by assertion */
793        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
794          md, eptrb, RM58);
795        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
796             md->mark == NULL) md->mark = ecode + 2;
797      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
798        md->start_match_ptr = ecode;
799      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
800    
801      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle an atomic group that does not contain any capturing parentheses.
802      the current subject position in the working slot at the top of the vector.      This can be handled like an assertion. Prior to 8.13, all atomic groups
803      We mustn't change the current values of the data slot, because they may be      were handled this way. In 8.13, the code was changed as below for ONCE, so
804      set from a previous iteration of this group, and be referred to by a      that backups pass through the group and thereby reset captured values.
805      reference inside the group.      However, this uses a lot more stack, so in 8.20, atomic groups that do not
806        contain any captures generate OP_ONCE_NC, which can be handled in the old,
807      If the bracket fails to match, we need to restore this value and also the      less stack intensive way.
808      values of the final offsets, in case they were set by a previous iteration  
809      of the same bracket.      Check the alternative branches in turn - the matching won't pass the KET
810        for this kind of subpattern. If any one branch matches, we carry on as at
811        the end of a normal bracket, leaving the subject pointer, but resetting
812        the start-of-match value in case it was changed by \K. */
813    
814        case OP_ONCE_NC:
815        prev = ecode;
816        saved_eptr = eptr;
817        do
818          {
819          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
820          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
821            {
822            mstart = md->start_match_ptr;
823            break;
824            }
825          if (rrc == MATCH_THEN)
826            {
827            next = ecode + GET(ecode,1);
828            if (md->start_match_ptr < next &&
829                (*ecode == OP_ALT || *next == OP_ALT))
830              rrc = MATCH_NOMATCH;
831            }
832    
833          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
834          ecode += GET(ecode,1);
835          }
836        while (*ecode == OP_ALT);
837    
838        /* If hit the end of the group (which could be repeated), fail */
839    
840        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
841    
842        /* Continue as from after the group, updating the offsets high water
843        mark, since extracts may have been taken. */
844    
845        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
846    
847        offset_top = md->end_offset_top;
848        eptr = md->end_match_ptr;
849    
850        /* For a non-repeating ket, just continue at this level. This also
851        happens for a repeating ket if no characters were matched in the group.
852        This is the forcible breaking of infinite loops as implemented in Perl
853        5.005. */
854    
855        if (*ecode == OP_KET || eptr == saved_eptr)
856          {
857          ecode += 1+LINK_SIZE;
858          break;
859          }
860    
861        /* The repeating kets try the rest of the pattern or restart from the
862        preceding bracket, in the appropriate order. The second "call" of match()
863        uses tail recursion, to avoid using another stack frame. */
864    
865        if (*ecode == OP_KETRMIN)
866          {
867          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
868          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
869          ecode = prev;
870          goto TAIL_RECURSE;
871          }
872        else  /* OP_KETRMAX */
873          {
874          md->match_function_type = MATCH_CBEGROUP;
875          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
876          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
877          ecode += 1 + LINK_SIZE;
878          goto TAIL_RECURSE;
879          }
880        /* Control never gets here */
881    
882        /* Handle a capturing bracket, other than those that are possessive with an
883        unlimited repeat. If there is space in the offset vector, save the current
884        subject position in the working slot at the top of the vector. We mustn't
885        change the current values of the data slot, because they may be set from a
886        previous iteration of this group, and be referred to by a reference inside
887        the group. A failure to match might occur after the group has succeeded,
888        if something later on doesn't match. For this reason, we need to restore
889        the working value and also the values of the final offsets, in case they
890        were set by a previous iteration of the same bracket.
891    
892      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
893      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 713  for (;;) Line 898  for (;;)
898      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
899      offset = number << 1;      offset = number << 1;
900    
901  #ifdef DEBUG  #ifdef PCRE_DEBUG
902      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
903      printf("subject=");      printf("subject=");
904      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 728  for (;;) Line 913  for (;;)
913        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
914    
915        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
916        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
917            (int)(eptr - md->start_subject);
918    
919        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
920          {          {
921          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
922            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
923          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);            eptrb, RM1);
924            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
925    
926            /* If we backed up to a THEN, check whether it is within the current
927            branch by comparing the address of the THEN that is passed back with
928            the end of the branch. If it is within the current branch, and the
929            branch is one of two or more alternatives (it either starts or ends
930            with OP_ALT), we have reached the limit of THEN's action, so convert
931            the return code to NOMATCH, which will cause normal backtracking to
932            happen from now on. Otherwise, THEN is passed back to an outer
933            alternative. This implements Perl's treatment of parenthesized groups,
934            where a group not containing | does not affect the current alternative,
935            that is, (X) is NOT the same as (X|(*F)). */
936    
937            if (rrc == MATCH_THEN)
938              {
939              next = ecode + GET(ecode,1);
940              if (md->start_match_ptr < next &&
941                  (*ecode == OP_ALT || *next == OP_ALT))
942                rrc = MATCH_NOMATCH;
943              }
944    
945            /* Anything other than NOMATCH is passed back. */
946    
947            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
948          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
949          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
950            if (*ecode != OP_ALT) break;
951          }          }
       while (*ecode == OP_ALT);  
952    
953        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
954        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
955        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
956        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
957    
958        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
959    
960          RRETURN(rrc);
961        }        }
962    
963      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 761  for (;;) Line 971  for (;;)
971      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
972      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
973    
974      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
975      final alternative within the brackets, we would return the result of a      repeat and ONCE group with no captures. Loop for all the alternatives.
976      recursive call to match() whatever happened. We can reduce stack usage by  
977      turning this into a tail recursion, except in the case when match_cbegroup      When we get to the final alternative within the brackets, we used to return
978      is set.*/      the result of a recursive call to match() whatever happened so it was
979        possible to reduce stack usage by turning this into a tail recursion,
980        except in the case of a possibly empty group. However, now that there is
981        the possiblity of (*THEN) occurring in the final alternative, this
982        optimization is no longer always possible.
983    
984        We can optimize if we know there are no (*THEN)s in the pattern; at present
985        this is the best that can be done.
986    
987        MATCH_ONCE is returned when the end of an atomic group is successfully
988        reached, but subsequent matching fails. It passes back up the tree (causing
989        captured values to be reset) until the original atomic group level is
990        reached. This is tested by comparing md->once_target with the start of the
991        group. At this point, the return is converted into MATCH_NOMATCH so that
992        previous backup points can be taken. */
993    
994        case OP_ONCE:
995      case OP_BRA:      case OP_BRA:
996      case OP_SBRA:      case OP_SBRA:
997      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
998      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
999      for (;;)      for (;;)
1000        {        {
1001        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
1002    
1003          /* If this is not a possibly empty group, and there are no (*THEN)s in
1004          the pattern, and this is the final alternative, optimize as described
1005          above. */
1006    
1007          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1008            {
1009            ecode += PRIV(OP_lengths)[*ecode];
1010            goto TAIL_RECURSE;
1011            }
1012    
1013          /* In all other cases, we have to make another call to match(). */
1014    
1015          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1016            RM2);
1017    
1018          /* See comment in the code for capturing groups above about handling
1019          THEN. */
1020    
1021          if (rrc == MATCH_THEN)
1022            {
1023            next = ecode + GET(ecode,1);
1024            if (md->start_match_ptr < next &&
1025                (*ecode == OP_ALT || *next == OP_ALT))
1026              rrc = MATCH_NOMATCH;
1027            }
1028    
1029          if (rrc != MATCH_NOMATCH)
1030          {          {
1031          if (flags == 0)    /* Not a possibly empty group */          if (rrc == MATCH_ONCE)
1032            {            {
1033            ecode += _pcre_OP_lengths[*ecode];            const pcre_uchar *scode = ecode;
1034            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1035            goto TAIL_RECURSE;              {
1036                while (*scode == OP_ALT) scode += GET(scode, 1);
1037                scode -= GET(scode, 1);
1038                }
1039              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1040            }            }
1041            RRETURN(rrc);
1042            }
1043          ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045          }
1046    
1047          /* Possibly empty group; can't use tail recursion. */      RRETURN(MATCH_NOMATCH);
1048    
1049          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* Handle possessive capturing brackets with an unlimited repeat. We come
1050            eptrb, flags, RM48);      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1051          RRETURN(rrc);      handled similarly to the normal case above. However, the matching is
1052        different. The end of these brackets will always be OP_KETRPOS, which
1053        returns MATCH_KETRPOS without going further in the pattern. By this means
1054        we can handle the group by iteration rather than recursion, thereby
1055        reducing the amount of stack needed. */
1056    
1057        case OP_CBRAPOS:
1058        case OP_SCBRAPOS:
1059        allow_zero = FALSE;
1060    
1061        POSSESSIVE_CAPTURE:
1062        number = GET2(ecode, 1+LINK_SIZE);
1063        offset = number << 1;
1064    
1065    #ifdef PCRE_DEBUG
1066        printf("start possessive bracket %d\n", number);
1067        printf("subject=");
1068        pchars(eptr, 16, TRUE, md);
1069        printf("\n");
1070    #endif
1071    
1072        if (offset < md->offset_max)
1073          {
1074          matched_once = FALSE;
1075          code_offset = (int)(ecode - md->start_code);
1076    
1077          save_offset1 = md->offset_vector[offset];
1078          save_offset2 = md->offset_vector[offset+1];
1079          save_offset3 = md->offset_vector[md->offset_end - number];
1080          save_capture_last = md->capture_last;
1081    
1082          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1083    
1084          /* Each time round the loop, save the current subject position for use
1085          when the group matches. For MATCH_MATCH, the group has matched, so we
1086          restart it with a new subject starting position, remembering that we had
1087          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1088          usual. If we haven't matched any alternatives in any iteration, check to
1089          see if a previous iteration matched. If so, the group has matched;
1090          continue from afterwards. Otherwise it has failed; restore the previous
1091          capture values before returning NOMATCH. */
1092    
1093          for (;;)
1094            {
1095            md->offset_vector[md->offset_end - number] =
1096              (int)(eptr - md->start_subject);
1097            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1098            RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1099              eptrb, RM63);
1100            if (rrc == MATCH_KETRPOS)
1101              {
1102              offset_top = md->end_offset_top;
1103              eptr = md->end_match_ptr;
1104              ecode = md->start_code + code_offset;
1105              save_capture_last = md->capture_last;
1106              matched_once = TRUE;
1107              continue;
1108              }
1109    
1110            /* See comment in the code for capturing groups above about handling
1111            THEN. */
1112    
1113            if (rrc == MATCH_THEN)
1114              {
1115              next = ecode + GET(ecode,1);
1116              if (md->start_match_ptr < next &&
1117                  (*ecode == OP_ALT || *next == OP_ALT))
1118                rrc = MATCH_NOMATCH;
1119              }
1120    
1121            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1122            md->capture_last = save_capture_last;
1123            ecode += GET(ecode, 1);
1124            if (*ecode != OP_ALT) break;
1125          }          }
1126    
1127        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1128        otherwise return. */          {
1129            md->offset_vector[offset] = save_offset1;
1130            md->offset_vector[offset+1] = save_offset2;
1131            md->offset_vector[md->offset_end - number] = save_offset3;
1132            }
1133    
1134        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (allow_zero || matched_once)
1135          eptrb, flags, RM2);          {
1136        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          ecode += 1 + LINK_SIZE;
1137            break;
1138            }
1139    
1140          RRETURN(MATCH_NOMATCH);
1141          }
1142    
1143        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1144        as a non-capturing bracket. */
1145    
1146        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1147        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1148    
1149        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1150    
1151        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1152        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1153    
1154        /* Non-capturing possessive bracket with unlimited repeat. We come here
1155        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1156        without the capturing complication. It is written out separately for speed
1157        and cleanliness. */
1158    
1159        case OP_BRAPOS:
1160        case OP_SBRAPOS:
1161        allow_zero = FALSE;
1162    
1163        POSSESSIVE_NON_CAPTURE:
1164        matched_once = FALSE;
1165        code_offset = (int)(ecode - md->start_code);
1166    
1167        for (;;)
1168          {
1169          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1170          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1171            eptrb, RM48);
1172          if (rrc == MATCH_KETRPOS)
1173            {
1174            offset_top = md->end_offset_top;
1175            eptr = md->end_match_ptr;
1176            ecode = md->start_code + code_offset;
1177            matched_once = TRUE;
1178            continue;
1179            }
1180    
1181          /* See comment in the code for capturing groups above about handling
1182          THEN. */
1183    
1184          if (rrc == MATCH_THEN)
1185            {
1186            next = ecode + GET(ecode,1);
1187            if (md->start_match_ptr < next &&
1188                (*ecode == OP_ALT || *next == OP_ALT))
1189              rrc = MATCH_NOMATCH;
1190            }
1191    
1192          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1193        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1194          if (*ecode != OP_ALT) break;
1195          }
1196    
1197        if (matched_once || allow_zero)
1198          {
1199          ecode += 1 + LINK_SIZE;
1200          break;
1201        }        }
1202        RRETURN(MATCH_NOMATCH);
1203    
1204      /* Control never reaches here. */      /* Control never reaches here. */
1205    
1206      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1207      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
1208      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
1209      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. */  
1210    
1211      case OP_COND:      case OP_COND:
1212      case OP_SCOND:      case OP_SCOND:
1213      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1214    
1215      /* 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
1216      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
1217    
1218      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1219        {        {
1220        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1221          {          {
1222          pcre_callout_block cb;          pcre_callout_block cb;
1223          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1224          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1225          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1226          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1227          cb.subject_length   = md->end_subject - md->start_subject;          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1228          cb.start_match      = mstart - md->start_subject;          cb.start_match      = (int)(mstart - md->start_subject);
1229          cb.current_position = eptr - md->start_subject;          cb.current_position = (int)(eptr - md->start_subject);
1230          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1231          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1232          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1233          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1234          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1235          if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1236            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1237          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1238          }          }
1239        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1240        }        }
1241    
1242      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
# Line 843  for (;;) Line 1247  for (;;)
1247        {        {
1248        if (md->recursive == NULL)                /* Not recursing => FALSE */        if (md->recursive == NULL)                /* Not recursing => FALSE */
1249          {          {
1250          condition = FALSE;          condition = FALSE;
1251          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
1252          }          }
1253        else        else
1254          {          {
1255          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1256          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1257    
1258          /* 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
1259          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
1260          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
1261          if any one is set. */          if any one is set. */
1262    
1263          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1264            {            {
1265            uschar *slotA = md->name_table;            pcre_uchar *slotA = md->name_table;
1266            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
1267              {              {
1268              if (GET2(slotA, 0) == recno) break;              if (GET2(slotA, 0) == recno) break;
1269              slotA += md->name_entry_size;              slotA += md->name_entry_size;
1270              }              }
1271    
1272            /* Found a name for the number - there can be only one; duplicate            /* Found a name for the number - there can be only one; duplicate
1273            names for different numbers are allowed, but not vice versa. First            names for different numbers are allowed, but not vice versa. First
1274            scan down for duplicates. */            scan down for duplicates. */
1275    
1276            if (i < md->name_count)            if (i < md->name_count)
1277              {              {
1278              uschar *slotB = slotA;              pcre_uchar *slotB = slotA;
1279              while (slotB > md->name_table)              while (slotB > md->name_table)
1280                {                {
1281                slotB -= md->name_entry_size;                slotB -= md->name_entry_size;
1282                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1283                  {                  {
1284                  condition = GET2(slotB, 0) == md->recursive->group_num;                  condition = GET2(slotB, 0) == md->recursive->group_num;
1285                  if (condition) break;                  if (condition) break;
1286                  }                  }
1287                else break;                else break;
1288                }                }
1289    
1290              /* Scan up for duplicates */              /* Scan up for duplicates */
1291    
1292              if (!condition)              if (!condition)
1293                {                {
1294                slotB = slotA;                slotB = slotA;
1295                for (i++; i < md->name_count; i++)                for (i++; i < md->name_count; i++)
1296                  {                  {
1297                  slotB += md->name_entry_size;                  slotB += md->name_entry_size;
1298                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1299                    {                    {
1300                    condition = GET2(slotB, 0) == md->recursive->group_num;                    condition = GET2(slotB, 0) == md->recursive->group_num;
1301                    if (condition) break;                    if (condition) break;
1302                    }                    }
1303                  else break;                  else break;
1304                  }                  }
1305                }                }
1306              }              }
1307            }            }
1308    
1309          /* Chose branch according to the condition */          /* Chose branch according to the condition */
1310    
1311          ecode += condition? 3 : GET(ecode, 1);          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1312          }          }
1313        }        }
1314    
1315      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1316        {        {
1317        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1318        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1319    
1320        /* If the numbered capture is unset, but the reference was by name,        /* If the numbered capture is unset, but the reference was by name,
1321        scan the table to see if the name refers to any other numbers, and test        scan the table to see if the name refers to any other numbers, and test
1322        them. The condition is true if any one is set. This is tediously similar        them. The condition is true if any one is set. This is tediously similar
1323        to the code above, but not close enough to try to amalgamate. */        to the code above, but not close enough to try to amalgamate. */
1324    
1325        if (!condition && condcode == OP_NCREF)        if (!condition && condcode == OP_NCREF)
1326          {          {
1327          int refno = offset >> 1;          int refno = offset >> 1;
1328          uschar *slotA = md->name_table;          pcre_uchar *slotA = md->name_table;
1329    
1330          for (i = 0; i < md->name_count; i++)          for (i = 0; i < md->name_count; i++)
1331            {            {
1332            if (GET2(slotA, 0) == refno) break;            if (GET2(slotA, 0) == refno) break;
1333            slotA += md->name_entry_size;            slotA += md->name_entry_size;
1334            }            }
1335    
1336          /* Found a name for the number - there can be only one; duplicate names          /* Found a name for the number - there can be only one; duplicate names
1337          for different numbers are allowed, but not vice versa. First scan down          for different numbers are allowed, but not vice versa. First scan down
1338          for duplicates. */          for duplicates. */
1339    
1340          if (i < md->name_count)          if (i < md->name_count)
1341            {            {
1342            uschar *slotB = slotA;            pcre_uchar *slotB = slotA;
1343            while (slotB > md->name_table)            while (slotB > md->name_table)
1344              {              {
1345              slotB -= md->name_entry_size;              slotB -= md->name_entry_size;
1346              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)              if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1347                {                {
1348                offset = GET2(slotB, 0) << 1;                offset = GET2(slotB, 0) << 1;
1349                condition = offset < offset_top &&                condition = offset < offset_top &&
1350                  md->offset_vector[offset] >= 0;                  md->offset_vector[offset] >= 0;
1351                if (condition) break;                if (condition) break;
1352                }                }
1353              else break;              else break;
1354              }              }
1355    
1356            /* Scan up for duplicates */            /* Scan up for duplicates */
1357    
1358            if (!condition)            if (!condition)
1359              {              {
1360              slotB = slotA;              slotB = slotA;
1361              for (i++; i < md->name_count; i++)              for (i++; i < md->name_count; i++)
1362                {                {
1363                slotB += md->name_entry_size;                slotB += md->name_entry_size;
1364                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1365                  {                  {
1366                  offset = GET2(slotB, 0) << 1;                  offset = GET2(slotB, 0) << 1;
1367                  condition = offset < offset_top &&                  condition = offset < offset_top &&
1368                    md->offset_vector[offset] >= 0;                    md->offset_vector[offset] >= 0;
1369                  if (condition) break;                  if (condition) break;
1370                  }                  }
1371                else break;                else break;
1372                }                }
1373              }              }
1374            }            }
1375          }          }
1376    
1377        /* Chose branch according to the condition */        /* Chose branch according to the condition */
1378    
1379        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1380        }        }
1381    
1382      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 982  for (;;) Line 1386  for (;;)
1386        }        }
1387    
1388      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1389      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
1390      assertion. */      an assertion. */
1391    
1392      else      else
1393        {        {
1394        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1395            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1396        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1397          {          {
1398            if (md->end_offset_top > offset_top)
1399              offset_top = md->end_offset_top;  /* Captures may have happened */
1400          condition = TRUE;          condition = TRUE;
1401          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1402          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1403          }          }
1404    
1405          /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1406          assertion; it is therefore treated as NOMATCH. */
1407    
1408        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1409          {          {
1410          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
# Line 1006  for (;;) Line 1416  for (;;)
1416          }          }
1417        }        }
1418    
1419      /* 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
1420      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
1421      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
1422      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
1423        exist, in which case we can just plough on. Note that, for compatibility
1424        with Perl, the | in a conditional group is NOT treated as creating two
1425        alternatives. If a THEN is encountered in the branch, it propagates out to
1426        the enclosing alternative (unless nested in a deeper set of alternatives,
1427        of course). */
1428    
1429      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1430        {        {
1431        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
         {  
         RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);  
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
1432          {          {
1433          flags = 0;          ecode += 1 + LINK_SIZE;
1434          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1435          }          }
1436    
1437          md->match_function_type = MATCH_CBEGROUP;
1438          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1439          RRETURN(rrc);
1440        }        }
1441      else                         /* Condition false & no alternative */  
1442         /* Condition false & no alternative; continue after the group. */
1443    
1444        else
1445        {        {
1446        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1447        }        }
1448      break;      break;
1449    
1450    
1451      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1452      to close any currently open capturing brackets. */      to close any currently open capturing brackets. */
1453    
1454      case OP_CLOSE:      case OP_CLOSE:
1455      number = GET2(ecode, 1);      number = GET2(ecode, 1);
1456      offset = number << 1;      offset = number << 1;
1457    
1458  #ifdef DEBUG  #ifdef PCRE_DEBUG
1459        printf("end bracket %d at *ACCEPT", number);        printf("end bracket %d at *ACCEPT", number);
1460        printf("\n");        printf("\n");
1461  #endif  #endif
# Line 1049  for (;;) Line 1465  for (;;)
1465        {        {
1466        md->offset_vector[offset] =        md->offset_vector[offset] =
1467          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1468        md->offset_vector[offset+1] = eptr - md->start_subject;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1469        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1470        }        }
1471      ecode += 3;      ecode += 1 + IMM2_SIZE;
1472      break;      break;
1473    
1474    
1475      /* 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. */  
1476    
     case OP_ACCEPT:  
1477      case OP_END:      case OP_END:
1478      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1479        {      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->offset_top;  
       mstart = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1480    
1481      /* 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
1482      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
1483      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,
1484      if any. */      backtracking will then try other alternatives, if any. */
1485    
1486      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1487          (md->notempty ||           md->recursive == NULL &&
1488            (md->notempty_atstart &&           (md->notempty ||
1489              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1490                 mstart == md->start_subject + md->start_offset)))
1491        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1492    
1493      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1092  for (;;) Line 1495  for (;;)
1495      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1496      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1497      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);  
1498    
1499      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1500        given as the argument to RRETURN when the heap is in use. */
1501    
1502      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1503      ims = ecode[1];      RRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1504    
1505      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1506      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,
1507      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
1508      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
1509      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1510        of a condition, we want to return immediately afterwards. The caller of
1511        this incarnation of the match() function will have set MATCH_CONDASSERT in
1512        md->match_function type, and one of these opcodes will be the first opcode
1513        that is processed. We use a local variable that is preserved over calls to
1514        match() to remember this case. */
1515    
1516      case OP_ASSERT:      case OP_ASSERT:
1517      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1518        if (md->match_function_type == MATCH_CONDASSERT)
1519          {
1520          condassert = TRUE;
1521          md->match_function_type = 0;
1522          }
1523        else condassert = FALSE;
1524    
1525      do      do
1526        {        {
1527        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1528          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1529        if (rrc == MATCH_MATCH) break;          {
1530            mstart = md->start_match_ptr;   /* In case \K reset it */
1531            break;
1532            }
1533    
1534          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1535          as NOMATCH. */
1536    
1537        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1538        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1539        }        }
1540      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1541    
1542      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1543    
1544      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1545    
1546      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1547    
1548      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1549      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1133  for (;;) Line 1553  for (;;)
1553      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1554      continue;      continue;
1555    
1556      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1557        PRUNE, or COMMIT means we must assume failure without checking subsequent
1558        branches. */
1559    
1560      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1561      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1562        if (md->match_function_type == MATCH_CONDASSERT)
1563          {
1564          condassert = TRUE;
1565          md->match_function_type = 0;
1566          }
1567        else condassert = FALSE;
1568    
1569      do      do
1570        {        {
1571        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1572          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1573        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1574            {
1575            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1576            break;
1577            }
1578    
1579          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1580          as NOMATCH. */
1581    
1582        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1583        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1584        }        }
1585      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1586    
1587      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1588    
1589      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1590      continue;      continue;
# Line 1158  for (;;) Line 1595  for (;;)
1595      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1596    
1597      case OP_REVERSE:      case OP_REVERSE:
1598  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1599      if (utf8)      if (utf)
1600        {        {
1601        i = GET(ecode, 1);        i = GET(ecode, 1);
1602        while (i-- > 0)        while (i-- > 0)
# Line 1190  for (;;) Line 1627  for (;;)
1627      function is able to force a failure. */      function is able to force a failure. */
1628    
1629      case OP_CALLOUT:      case OP_CALLOUT:
1630      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1631        {        {
1632        pcre_callout_block cb;        pcre_callout_block cb;
1633        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1634        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1635        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1636        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1637        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1638        cb.start_match      = mstart - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1639        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1640        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1641        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1642        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1643        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1644        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1645        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1646          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1647        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1648        }        }
1649      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1215  for (;;) Line 1653  for (;;)
1653      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
1654      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1655    
1656      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1657      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
1658      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
1659      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
1660      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
1661      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
1662      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.  
1663    
1664      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
1665      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
1666      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1667        a lot, so he is not to blame for the current way it works. */
1668    
1669      case OP_RECURSE:      case OP_RECURSE:
1670        {        {
1671          recursion_info *ri;
1672          int recno;
1673    
1674        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1675        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1676          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1677    
1678          /* Check for repeating a recursion without advancing the subject pointer.
1679          This should catch convoluted mutual recursions. (Some simple cases are
1680          caught at compile time.) */
1681    
1682          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1683            if (recno == ri->group_num && eptr == ri->subject_position)
1684              RRETURN(PCRE_ERROR_RECURSELOOP);
1685    
1686        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1687    
1688          new_recursive.group_num = recno;
1689          new_recursive.subject_position = eptr;
1690        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1691        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1692    
1693        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1694    
1695        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1696    
1697        /* Now save the offset data. */        /* Now save the offset data */
1698    
1699        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1700        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1254  for (;;) Line 1702  for (;;)
1702        else        else
1703          {          {
1704          new_recursive.offset_save =          new_recursive.offset_save =
1705            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1706          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1707          }          }
   
1708        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1709              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = mstart;  
       new_recursive.offset_top = offset_top;  
       mstart = eptr;  
1710    
1711        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1712        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1713          might be changed, so reset it before looping. */
1714    
1715        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1716        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1717        do        do
1718          {          {
1719          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1720            md, ims, eptrb, flags, RM6);          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1721          if (rrc == MATCH_MATCH)            md, eptrb, RM6);
1722            memcpy(md->offset_vector, new_recursive.offset_save,
1723                new_recursive.saved_max * sizeof(int));
1724            md->recursive = new_recursive.prevrec;
1725            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1726            {            {
1727            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1728            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1729              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1730            RRETURN(MATCH_MATCH);  
1731              /* Set where we got to in the subject, and reset the start in case
1732              it was changed by \K. This *is* propagated back out of a recursion,
1733              for Perl compatibility. */
1734    
1735              eptr = md->end_match_ptr;
1736              mstart = md->start_match_ptr;
1737              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1738            }            }
1739    
1740            /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1741            as NOMATCH. */
1742    
1743          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1744            {            {
1745            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1746            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1747              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1748            RRETURN(rrc);            RRETURN(rrc);
1749            }            }
1750    
1751          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1752          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1753          }          }
1754        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1299  for (;;) Line 1756  for (;;)
1756        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1757        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1758        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1759          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1760        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1761        }        }
     /* 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. */  
   
     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) 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);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
1762    
1763      if (ecode[1+LINK_SIZE] == OP_OPT)      RECURSION_MATCHED:
1764        {      break;
       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 */  
1765    
1766      /* 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
1767      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1392  for (;;) Line 1777  for (;;)
1777      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1778    
1779      case OP_BRAZERO:      case OP_BRAZERO:
1780        {      next = ecode + 1;
1781        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1782        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1783        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1784        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1785      break;      break;
1786    
1787      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1788        {      next = ecode + 1;
1789        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1790        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1791        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1792        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1793      break;      break;
1794    
1795      case OP_SKIPZERO:      case OP_SKIPZERO:
1796        {      next = ecode+1;
1797        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1798        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1799      break;      break;
1800    
1801        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1802        here; just jump to the group, with allow_zero set TRUE. */
1803    
1804        case OP_BRAPOSZERO:
1805        op = *(++ecode);
1806        allow_zero = TRUE;
1807        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1808          goto POSSESSIVE_NON_CAPTURE;
1809    
1810      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1811    
1812      case OP_KET:      case OP_KET:
1813      case OP_KETRMIN:      case OP_KETRMIN:
1814      case OP_KETRMAX:      case OP_KETRMAX:
1815        case OP_KETRPOS:
1816      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1817    
1818      /* 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
1819      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1820      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1821    
1822      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1823        {        {
1824        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1825        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1826        }        }
1827      else saved_eptr = NULL;      else saved_eptr = NULL;
1828    
1829      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group or a non-capturing atomic
1830      MATCH_MATCH, but record the current high water mark for use by positive      group, stop matching and return MATCH_MATCH, but record the current high
1831      assertions. Do this also for the "once" (atomic) groups. */      water mark for use by positive assertions. We also need to record the match
1832        start in case it was changed by \K. */
1833      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||  
1834          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1835          *prev == OP_ONCE)           *prev == OP_ONCE_NC)
1836        {        {
1837        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1838        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1839        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1840          RRETURN(MATCH_MATCH);         /* Sets md->mark */
1841        }        }
1842    
1843      /* 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
1844      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1845      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
1846      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
1847      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
1848        the current subject position and start match pointer and give a MATCH
1849        return. */
1850    
1851      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1852            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1853        {        {
1854        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1855        offset = number << 1;        offset = number << 1;
1856    
1857  #ifdef DEBUG  #ifdef PCRE_DEBUG
1858        printf("end bracket %d", number);        printf("end bracket %d", number);
1859        printf("\n");        printf("\n");
1860  #endif  #endif
1861    
1862          /* Handle a recursively called group. */
1863    
1864          if (md->recursive != NULL && md->recursive->group_num == number)
1865            {
1866            md->end_match_ptr = eptr;
1867            md->start_match_ptr = mstart;
1868            RRETURN(MATCH_MATCH);
1869            }
1870    
1871          /* Deal with capturing */
1872    
1873        md->capture_last = number;        md->capture_last = number;
1874        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1875          {          {
1876            /* If offset is greater than offset_top, it means that we are
1877            "skipping" a capturing group, and that group's offsets must be marked
1878            unset. In earlier versions of PCRE, all the offsets were unset at the
1879            start of matching, but this doesn't work because atomic groups and
1880            assertions can cause a value to be set that should later be unset.
1881            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1882            part of the atomic group, but this is not on the final matching path,
1883            so must be unset when 2 is set. (If there is no group 2, there is no
1884            problem, because offset_top will then be 2, indicating no capture.) */
1885    
1886            if (offset > offset_top)
1887              {
1888              register int *iptr = md->offset_vector + offset_top;
1889              register int *iend = md->offset_vector + offset;
1890              while (iptr < iend) *iptr++ = -1;
1891              }
1892    
1893            /* Now make the extraction */
1894    
1895          md->offset_vector[offset] =          md->offset_vector[offset] =
1896            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1897          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1898          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1899          }          }
1900          }
1901    
1902        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1903        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1904        group. This is the forcible breaking of infinite loops as implemented in
1905        Perl 5.005. For a non-repeating atomic group that includes captures,
1906        establish a backup point by processing the rest of the pattern at a lower
1907        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1908        original OP_ONCE level, thereby bypassing intermediate backup points, but
1909        resetting any captures that happened along the way. */
1910    
1911        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1912          {
1913          if (*prev == OP_ONCE)
1914          {          {
1915          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1916          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1917          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1918          mstart = rec->save_start;          RRETURN(MATCH_ONCE);
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         offset_top = rec->offset_top;  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1919          }          }
1920          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1921          break;
1922        }        }
1923    
1924      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1925      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1926        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. */  
1927    
1928      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRPOS)
1929        {        {
1930        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1931        break;        md->end_offset_top = offset_top;
1932          RRETURN(MATCH_KETRPOS);
1933        }        }
1934    
1935      /* 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
1936      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
1937      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
1938      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
1939        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1940    
1941      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1942        {        {
1943        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1944        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1945        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1946            {
1947            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1948            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1949            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1950            RRETURN(MATCH_ONCE);
1951            }
1952          if (*prev >= OP_SBRA)    /* Could match an empty string */
1953          {          {
1954          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1955            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1956          RRETURN(rrc);          RRETURN(rrc);
1957          }          }
1958        ecode = prev;        ecode = prev;
# Line 1532  for (;;) Line 1960  for (;;)
1960        }        }
1961      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1962        {        {
1963        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1964          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1965          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1966        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1967          if (*prev == OP_ONCE)
1968            {
1969            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1970            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1971            md->once_target = prev;
1972            RRETURN(MATCH_ONCE);
1973            }
1974        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1975        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1976        }        }
1977      /* Control never gets here */      /* Control never gets here */
1978    
1979      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1980    
1981      case OP_CIRC:      case OP_CIRC:
1982      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 */  
1983    
1984      /* Start of subject assertion */      /* Start of subject assertion */
1985    
# Line 1561  for (;;) Line 1988  for (;;)
1988      ecode++;      ecode++;
1989      break;      break;
1990    
1991        /* Multiline mode: start of subject unless notbol, or after any newline. */
1992    
1993        case OP_CIRCM:
1994        if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
1995        if (eptr != md->start_subject &&
1996            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1997          RRETURN(MATCH_NOMATCH);
1998        ecode++;
1999        break;
2000    
2001      /* Start of match assertion */      /* Start of match assertion */
2002    
2003      case OP_SOM:      case OP_SOM:
# Line 1575  for (;;) Line 2012  for (;;)
2012      ecode++;      ecode++;
2013      break;      break;
2014    
2015      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2016      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2017    
2018      case OP_DOLL:      case OP_DOLLM:
2019      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2020        {        { 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;  
       }  
2021      else      else
2022        {        {
2023        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2024        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
2025        }        }
2026        ecode++;
2027        break;
2028    
2029        /* Not multiline mode: assert before a terminating newline or before end of
2030        subject unless noteol is set. */
2031    
2032        case OP_DOLL:
2033        if (md->noteol) RRETURN(MATCH_NOMATCH);
2034        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2035    
2036      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2037    
2038      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2039    
2040      case OP_EOD:      case OP_EOD:
2041      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2042        SCHECK_PARTIAL();
2043      ecode++;      ecode++;
2044      break;      break;
2045    
2046      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
2047    
2048      case OP_EODN:      case OP_EODN:
2049      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
2050        if (eptr < md->end_subject &&
2051          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2052        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2053    
2054        /* Either at end of string or \n before end. */
2055    
2056        SCHECK_PARTIAL();
2057      ecode++;      ecode++;
2058      break;      break;
2059    
# Line 1629  for (;;) Line 2068  for (;;)
2068        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2069        partial matching. */        partial matching. */
2070    
2071  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2072        if (utf8)        if (utf)
2073          {          {
2074            /* Get status of previous character */
2075    
2076          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2077            {            {
2078            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2079            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2080            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2081            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2082    #ifdef SUPPORT_UCP
2083              if (md->use_ucp)
2084                {
2085                if (c == '_') prev_is_word = TRUE; else
2086                  {
2087                  int cat = UCD_CATEGORY(c);
2088                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2089                  }
2090                }
2091              else
2092    #endif
2093            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2094            }            }
2095    
2096            /* Get status of next character */
2097    
2098          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2099            {            {
2100            SCHECK_PARTIAL();            SCHECK_PARTIAL();
# Line 1648  for (;;) Line 2103  for (;;)
2103          else          else
2104            {            {
2105            GETCHAR(c, eptr);            GETCHAR(c, eptr);
2106    #ifdef SUPPORT_UCP
2107              if (md->use_ucp)
2108                {
2109                if (c == '_') cur_is_word = TRUE; else
2110                  {
2111                  int cat = UCD_CATEGORY(c);
2112                  cur_is_word = (cat == ucp_L || cat == ucp_N);
2113                  }
2114                }
2115              else
2116    #endif
2117            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2118            }            }
2119          }          }
2120        else        else
2121  #endif  #endif
2122    
2123        /* Not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
2124          consistency with the behaviour of \w we do use it in this case. */
2125    
2126          {          {
2127            /* Get status of previous character */
2128    
2129          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2130            {            {
2131            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
2132            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);  #ifdef SUPPORT_UCP
2133              if (md->use_ucp)
2134                {
2135                c = eptr[-1];
2136                if (c == '_') prev_is_word = TRUE; else
2137                  {
2138                  int cat = UCD_CATEGORY(c);
2139                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2140                  }
2141                }
2142              else
2143    #endif
2144              prev_is_word = MAX_255(eptr[-1])
2145                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2146            }            }
2147    
2148            /* Get status of next character */
2149    
2150          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2151            {            {
2152            SCHECK_PARTIAL();            SCHECK_PARTIAL();
2153            cur_is_word = FALSE;            cur_is_word = FALSE;
2154            }            }
2155          else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          else
2156    #ifdef SUPPORT_UCP
2157            if (md->use_ucp)
2158              {
2159              c = *eptr;
2160              if (c == '_') cur_is_word = TRUE; else
2161                {
2162                int cat = UCD_CATEGORY(c);
2163                cur_is_word = (cat == ucp_L || cat == ucp_N);
2164                }
2165              }
2166            else
2167    #endif
2168            cur_is_word = MAX_255(*eptr)
2169              && ((md->ctypes[*eptr] & ctype_word) != 0);
2170          }          }
2171    
2172        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
# Line 1685  for (;;) Line 2184  for (;;)
2184      /* Fall through */      /* Fall through */
2185    
2186      case OP_ALLANY:      case OP_ALLANY:
2187      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2188        {        {                            /* not be updated before SCHECK_PARTIAL. */
2189        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2190        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2191        }        }
2192      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2193    #ifdef SUPPORT_UTF
2194        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2195    #endif
2196      ecode++;      ecode++;
2197      break;      break;
2198    
# Line 1698  for (;;) Line 2200  for (;;)
2200      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2201    
2202      case OP_ANYBYTE:      case OP_ANYBYTE:
2203      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2204        {        {                            /* not be updated before SCHECK_PARTIAL. */
2205        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2206        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2207        }        }
2208        eptr++;
2209      ecode++;      ecode++;
2210      break;      break;
2211    
# Line 1714  for (;;) Line 2217  for (;;)
2217        }        }
2218      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2219      if (      if (
2220  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2221         c < 256 &&         c < 256 &&
2222  #endif  #endif
2223         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
# Line 1731  for (;;) Line 2234  for (;;)
2234        }        }
2235      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2236      if (      if (
2237  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2238         c >= 256 ||         c > 255 ||
2239  #endif  #endif
2240         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2241         )         )
# Line 1748  for (;;) Line 2251  for (;;)
2251        }        }
2252      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2253      if (      if (
2254  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2255         c < 256 &&         c < 256 &&
2256  #endif  #endif
2257         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
# Line 1765  for (;;) Line 2268  for (;;)
2268        }        }
2269      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2270      if (      if (
2271  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2272         c >= 256 ||         c > 255 ||
2273  #endif  #endif
2274         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2275         )         )
# Line 1782  for (;;) Line 2285  for (;;)
2285        }        }
2286      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2287      if (      if (
2288  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2289         c < 256 &&         c < 256 &&
2290  #endif  #endif
2291         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
# Line 1799  for (;;) Line 2302  for (;;)
2302        }        }
2303      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2304      if (      if (
2305  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2306         c >= 256 ||         c > 255 ||
2307  #endif  #endif
2308         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2309         )         )
# Line 1818  for (;;) Line 2321  for (;;)
2321      switch(c)      switch(c)
2322        {        {
2323        default: RRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2324    
2325        case 0x000d:        case 0x000d:
2326        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2327        break;        break;
# Line 1974  for (;;) Line 2478  for (;;)
2478               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2479               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2480            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2481           break;          break;
2482    
2483          case PT_GC:          case PT_GC:
2484          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2485            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2486          break;          break;
2487    
# Line 1991  for (;;) Line 2495  for (;;)
2495            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2496          break;          break;
2497    
2498            /* These are specials */
2499    
2500            case PT_ALNUM:
2501            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2502                 PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2503              RRETURN(MATCH_NOMATCH);
2504            break;
2505    
2506            case PT_SPACE:    /* Perl space */
2507            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2508                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2509                   == (op == OP_NOTPROP))
2510              RRETURN(MATCH_NOMATCH);
2511            break;
2512    
2513            case PT_PXSPACE:  /* POSIX space */
2514            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2515                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2516                 c == CHAR_FF || c == CHAR_CR)
2517                   == (op == OP_NOTPROP))
2518              RRETURN(MATCH_NOMATCH);
2519            break;
2520    
2521            case PT_WORD:
2522            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2523                 PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2524                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2525              RRETURN(MATCH_NOMATCH);
2526            break;
2527    
2528            /* This should never occur */
2529    
2530          default:          default:
2531          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2532          }          }
# Line 2009  for (;;) Line 2545  for (;;)
2545        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2546        }        }
2547      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2548        if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH);
2549        while (eptr < md->end_subject)
2550        {        {
2551        int category = UCD_CATEGORY(c);        int len = 1;
2552        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2553        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2554          {        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;  
         }  
2555        }        }
2556      ecode++;      ecode++;
2557      break;      break;
# Line 2038  for (;;) Line 2567  for (;;)
2567      loops). */      loops). */
2568    
2569      case OP_REF:      case OP_REF:
2570        {      case OP_REFI:
2571        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2572        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2573        ecode += 1 + IMM2_SIZE;
2574    
2575        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2576    
2577        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2578        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
2579        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.  
2580    
2581        (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
2582        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2583    
2584        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
2585        referenced subpattern. */      referenced subpattern. */
2586    
2587        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2588          length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;        length = (md->jscript_compat)? 0 : -1;
2589        else      else
2590          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2591    
2592        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2593    
2594        switch (*ecode)      switch (*ecode)
2595          {        {
2596          case OP_CRSTAR:        case OP_CRSTAR:
2597          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2598          case OP_CRPLUS:        case OP_CRPLUS:
2599          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2600          case OP_CRQUERY:        case OP_CRQUERY:
2601          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2602          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2603          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2604          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2605          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2606          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2607          break;        break;
2608    
2609          case OP_CRRANGE:        case OP_CRRANGE:
2610          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2611          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2612          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2613          max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2614          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2615          ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2616          break;        break;
2617    
2618          default:               /* No repeat follows */        default:               /* No repeat follows */
2619          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2620            {          {
2621            CHECK_PARTIAL();          CHECK_PARTIAL();
2622            RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
           }  
         eptr += length;  
         continue;              /* With the main loop */  
2623          }          }
2624          eptr += length;
2625          continue;              /* With the main loop */
2626          }
2627    
2628        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2629        main loop. */      zero, just continue with the main loop. */
2630    
2631        if (length == 0) continue;      if (length == 0) continue;
2632    
2633        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2634        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2635        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2636    
2637        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2638          {
2639          int slength;
2640          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2641          {          {
2642          if (!match_ref(offset, eptr, length, md, ims))          CHECK_PARTIAL();
2643            {          RRETURN(MATCH_NOMATCH);
           CHECK_PARTIAL();  
           RRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2644          }          }
2645          eptr += slength;
2646          }
2647    
2648        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2649        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2650    
2651        if (min == max) continue;      if (min == max) continue;
2652    
2653        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2654    
2655        if (minimize)      if (minimize)
2656          {
2657          for (fi = min;; fi++)
2658          {          {
2659          for (fi = min;; fi++)          int slength;
2660            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2661            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2662            if (fi >= max) RRETURN(MATCH_NOMATCH);
2663            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2664            {            {
2665            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2666            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;  
2667            }            }
2668          /* Control never gets here */          eptr += slength;
2669          }          }
2670          /* Control never gets here */
2671          }
2672    
2673        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2674    
2675        else      else
2676          {
2677          pp = eptr;
2678          for (i = min; i < max; i++)
2679          {          {
2680          pp = eptr;          int slength;
2681          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
           {  
           if (!match_ref(offset, eptr, length, md, ims)) break;  
           eptr += length;  
           }  
         while (eptr >= pp)  
2682            {            {
2683            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2684            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2685            }            }
2686          RRETURN(MATCH_NOMATCH);          eptr += slength;
2687            }
2688          while (eptr >= pp)
2689            {
2690            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2691            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2692            eptr -= length;
2693          }          }
2694          RRETURN(MATCH_NOMATCH);
2695        }        }
2696      /* Control never gets here */      /* Control never gets here */
2697    
# Line 2174  for (;;) Line 2709  for (;;)
2709      case OP_NCLASS:      case OP_NCLASS:
2710      case OP_CLASS:      case OP_CLASS:
2711        {        {
2712          /* The data variable is saved across frames, so the byte map needs to
2713          be stored there. */
2714    #define BYTE_MAP ((pcre_uint8 *)data)
2715        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2716        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2717    
2718        switch (*ecode)        switch (*ecode)
2719          {          {
# Line 2196  for (;;) Line 2734  for (;;)
2734          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2735          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2736          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2737          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2738          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2739          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2740          break;          break;
2741    
2742          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2208  for (;;) Line 2746  for (;;)
2746    
2747        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2748    
2749  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2750        /* UTF-8 mode */        if (utf)
       if (utf8)  
2751          {          {
2752          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2753            {            {
# Line 2225  for (;;) Line 2762  for (;;)
2762              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2763              }              }
2764            else            else
2765              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  
             }  
2766            }            }
2767          }          }
2768        else        else
2769  #endif  #endif
2770        /* Not UTF-8 mode */        /* Not UTF mode */
2771          {          {
2772          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2773            {            {
# Line 2242  for (;;) Line 2777  for (;;)
2777              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2778              }              }
2779            c = *eptr++;            c = *eptr++;
2780            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2781              if (c > 255)
2782                {
2783                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2784                }
2785              else
2786    #endif
2787                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2788            }            }
2789          }          }
2790    
# Line 2256  for (;;) Line 2798  for (;;)
2798    
2799        if (minimize)        if (minimize)
2800          {          {
2801  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2802          /* UTF-8 mode */          if (utf)
         if (utf8)  
2803            {            {
2804            for (fi = min;; fi++)            for (fi = min;; fi++)
2805              {              {
2806              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2807              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2808              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2809              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2276  for (;;) Line 2817  for (;;)
2817                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2818                }                }
2819              else              else
2820                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  
               }  
2821              }              }
2822            }            }
2823          else          else
2824  #endif  #endif
2825          /* Not UTF-8 mode */          /* Not UTF mode */
2826            {            {
2827            for (fi = min;; fi++)            for (fi = min;; fi++)
2828              {              {
2829              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2830              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2831              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2832              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2296  for (;;) Line 2835  for (;;)
2835                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2836                }                }
2837              c = *eptr++;              c = *eptr++;
2838              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2839                if (c > 255)
2840                  {
2841                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2842                  }
2843                else
2844    #endif
2845                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2846              }              }
2847            }            }
2848          /* Control never gets here */          /* Control never gets here */
# Line 2308  for (;;) Line 2854  for (;;)
2854          {          {
2855          pp = eptr;          pp = eptr;
2856    
2857  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2858          /* UTF-8 mode */          if (utf)
         if (utf8)  
2859            {            {
2860            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2861              {              {
2862              int len = 1;              int len = 1;
2863              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2864                  {
2865                  SCHECK_PARTIAL();
2866                  break;
2867                  }
2868              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2869              if (c > 255)              if (c > 255)
2870                {                {
2871                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
2872                }                }
2873              else              else
2874                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
2875              eptr += len;              eptr += len;
2876              }              }
2877            for (;;)            for (;;)
2878              {              {
2879              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2880              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2881              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2882              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2337  for (;;) Line 2884  for (;;)
2884            }            }
2885          else          else
2886  #endif  #endif
2887            /* Not UTF-8 mode */            /* Not UTF mode */
2888            {            {
2889            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2890              {              {
2891              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2892                  {
2893                  SCHECK_PARTIAL();
2894                  break;
2895                  }
2896              c = *eptr;              c = *eptr;
2897              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
2898                if (c > 255)
2899                  {
2900                  if (op == OP_CLASS) break;
2901                  }
2902                else
2903    #endif
2904                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
2905              eptr++;              eptr++;
2906              }              }
2907            while (eptr >= pp)            while (eptr >= pp)
2908              {              {
2909              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2910              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2911              eptr--;              eptr--;
2912              }              }
# Line 2356  for (;;) Line 2914  for (;;)
2914    
2915          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2916          }          }
2917    #undef BYTE_MAP
2918        }        }
2919      /* Control never gets here */      /* Control never gets here */
2920    
# Line 2364  for (;;) Line 2923  for (;;)
2923      when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8      when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8
2924      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
2925    
2926  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2927      case OP_XCLASS:      case OP_XCLASS:
2928        {        {
2929        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2389  for (;;) Line 2948  for (;;)
2948          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2949          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2950          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2951          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2952          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2953          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2954          break;          break;
2955    
2956          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2409  for (;;) Line 2968  for (;;)
2968            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2969            }            }
2970          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
2971          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
2972          }          }
2973    
2974        /* If max == min we can continue with the main loop without the        /* If max == min we can continue with the main loop without the
# Line 2424  for (;;) Line 2983  for (;;)
2983          {          {
2984          for (fi = min;; fi++)          for (fi = min;; fi++)
2985            {            {
2986            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2987            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2988            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
2989            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2433  for (;;) Line 2992  for (;;)
2992              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2993              }              }
2994            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
2995            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
2996            }            }
2997          /* Control never gets here */          /* Control never gets here */
2998          }          }
# Line 2446  for (;;) Line 3005  for (;;)
3005          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3006            {            {
3007            int len = 1;            int len = 1;
3008            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
3009                {
3010                SCHECK_PARTIAL();
3011                break;
3012                }
3013    #ifdef SUPPORT_UTF
3014            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3015            if (!_pcre_xclass(c, data)) break;  #else
3016              c = *eptr;
3017    #endif
3018              if (!PRIV(xclass)(c, data, utf)) break;
3019            eptr += len;            eptr += len;
3020            }            }
3021          for(;;)          for(;;)
3022            {            {
3023            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3024            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3025            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3026            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3027              if (utf) BACKCHAR(eptr);
3028    #endif
3029            }            }
3030          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3031          }          }
# Line 2468  for (;;) Line 3037  for (;;)
3037      /* Match a single character, casefully */      /* Match a single character, casefully */
3038    
3039      case OP_CHAR:      case OP_CHAR:
3040  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3041      if (utf8)      if (utf)
3042        {        {
3043        length = 1;        length = 1;
3044        ecode++;        ecode++;
# Line 2483  for (;;) Line 3052  for (;;)
3052        }        }
3053      else      else
3054  #endif  #endif
3055        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3056        {        {
3057        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3058          {          {
# Line 2496  for (;;) Line 3064  for (;;)
3064        }        }
3065      break;      break;
3066    
3067      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3068        subject, give up immediately. */
3069    
3070        case OP_CHARI:
3071        if (eptr >= md->end_subject)
3072          {
3073          SCHECK_PARTIAL();
3074          RRETURN(MATCH_NOMATCH);
3075          }
3076    
3077      case OP_CHARNC:  #ifdef SUPPORT_UTF
3078  #ifdef SUPPORT_UTF8      if (utf)
     if (utf8)  
3079        {        {
3080        length = 1;        length = 1;
3081        ecode++;        ecode++;
3082        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3083    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         RRETURN(MATCH_NOMATCH);  
         }  
   
3084        /* If the pattern character's value is < 128, we have only one byte, and        /* If the pattern character's value is < 128, we have only one byte, and
3085        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3086          fast lookup table. We know that there is at least one byte left in the
3087          subject. */
3088    
3089        if (fc < 128)        if (fc < 128)
3090          {          {
3091          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[fc]
3092                != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3093            ecode++;
3094            eptr++;
3095          }          }
3096    
3097        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3098          use the value of "length" to check for sufficient bytes left, because the
3099          other case of the character may have more or fewer bytes.  */
3100    
3101        else        else
3102          {          {
# Line 2541  for (;;) Line 3117  for (;;)
3117          }          }
3118        }        }
3119      else      else
3120  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3121    
3122      /* Non-UTF-8 mode */      /* Not UTF mode */
3123        {        {
3124        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3125          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3126          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         RRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);  
3127        ecode += 2;        ecode += 2;
3128        }        }
3129      break;      break;
# Line 2558  for (;;) Line 3131  for (;;)
3131      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3132    
3133      case OP_EXACT:      case OP_EXACT:
3134        case OP_EXACTI:
3135      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3136      ecode += 3;      ecode += 1 + IMM2_SIZE;
3137      goto REPEATCHAR;      goto REPEATCHAR;
3138    
3139      case OP_POSUPTO:      case OP_POSUPTO:
3140        case OP_POSUPTOI:
3141      possessive = TRUE;      possessive = TRUE;
3142      /* Fall through */      /* Fall through */
3143    
3144      case OP_UPTO:      case OP_UPTO:
3145        case OP_UPTOI:
3146      case OP_MINUPTO:      case OP_MINUPTO:
3147        case OP_MINUPTOI:
3148      min = 0;      min = 0;
3149      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3150      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3151      ecode += 3;      ecode += 1 + IMM2_SIZE;
3152      goto REPEATCHAR;      goto REPEATCHAR;
3153    
3154      case OP_POSSTAR:      case OP_POSSTAR:
3155        case OP_POSSTARI:
3156      possessive = TRUE;      possessive = TRUE;
3157      min = 0;      min = 0;
3158      max = INT_MAX;      max = INT_MAX;
# Line 2582  for (;;) Line 3160  for (;;)
3160      goto REPEATCHAR;      goto REPEATCHAR;
3161    
3162      case OP_POSPLUS:      case OP_POSPLUS:
3163        case OP_POSPLUSI:
3164      possessive = TRUE;      possessive = TRUE;
3165      min = 1;      min = 1;
3166      max = INT_MAX;      max = INT_MAX;
# Line 2589  for (;;) Line 3168  for (;;)
3168      goto REPEATCHAR;      goto REPEATCHAR;
3169    
3170      case OP_POSQUERY:      case OP_POSQUERY:
3171        case OP_POSQUERYI:
3172      possessive = TRUE;      possessive = TRUE;
3173      min = 0;      min = 0;
3174      max = 1;      max = 1;
# Line 2596  for (;;) Line 3176  for (;;)
3176      goto REPEATCHAR;      goto REPEATCHAR;
3177    
3178      case OP_STAR:      case OP_STAR:
3179        case OP_STARI:
3180      case OP_MINSTAR:      case OP_MINSTAR:
3181        case OP_MINSTARI:
3182      case OP_PLUS:      case OP_PLUS:
3183        case OP_PLUSI:
3184      case OP_MINPLUS:      case OP_MINPLUS:
3185        case OP_MINPLUSI:
3186      case OP_QUERY:      case OP_QUERY:
3187        case OP_QUERYI:
3188      case OP_MINQUERY:      case OP_MINQUERY:
3189      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3190        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3191      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3192      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3193      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3194      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2611  for (;;) Line 3196  for (;;)
3196      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3197    
3198      REPEATCHAR:      REPEATCHAR:
3199  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3200      if (utf8)      if (utf)
3201        {        {
3202        length = 1;        length = 1;
3203        charptr = ecode;        charptr = ecode;
# Line 2626  for (;;) Line 3211  for (;;)
3211          {          {
3212  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3213          unsigned int othercase;          unsigned int othercase;
3214          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3215              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3216            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3217          else oclength = 0;          else oclength = 0;
3218  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3219    
3220          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3221            {            {
3222            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3223              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3224  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3225            else if (oclength > 0 &&            else if (oclength > 0 &&
3226                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3227                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3228  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3229            else            else
3230              {              {
# Line 2654  for (;;) Line 3239  for (;;)
3239            {            {
3240            for (fi = min;; fi++)            for (fi = min;; fi++)
3241              {              {
3242              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3243              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3244              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3245              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3246                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3247  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3248              else if (oclength > 0 &&              else if (oclength > 0 &&
3249                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3250                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3251  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3252              else              else
3253                {                {
# Line 2679  for (;;) Line 3264  for (;;)
3264            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3265              {              {
3266              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3267                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3268  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3269              else if (oclength > 0 &&              else if (oclength > 0 &&
3270                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3271                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3272  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3273              else break;              else
3274                  {
3275                  CHECK_PARTIAL();
3276                  break;
3277                  }
3278              }              }
3279    
3280            if (possessive) continue;            if (possessive) continue;
3281    
3282            for(;;)            for(;;)
3283              {              {
3284              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3285              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3286              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3287  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2711  for (;;) Line 3300  for (;;)
3300        value of fc will always be < 128. */        value of fc will always be < 128. */
3301        }        }
3302      else      else
3303  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3304          /* When not in UTF-8 mode, load a single-byte character. */
3305          fc = *ecode++;
3306    
3307      /* When not in UTF-8 mode, load a single-byte character. */      /* The value of fc at this point is always one character, though we may
3308        or may not be in UTF mode. The code is duplicated for the caseless and
     fc = *ecode++;  
   
     /* The value of fc at this point is always less than 256, though we may or  
     may not be in UTF-8 mode. The code is duplicated for the caseless and  
3309      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3310      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3311      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 2729  for (;;) Line 3316  for (;;)
3316      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3317        max, eptr));        max, eptr));
3318    
3319      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3320        {        {
3321        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3322          /* fc must be < 128 if UTF is enabled. */
3323          foc = md->fcc[fc];
3324    #else
3325    #ifdef SUPPORT_UTF
3326    #ifdef SUPPORT_UCP
3327          if (utf && fc > 127)
3328            foc = UCD_OTHERCASE(fc);
3329    #else
3330          if (utf && fc > 127)
3331            foc = fc;
3332    #endif /* SUPPORT_UCP */
3333          else
3334    #endif /* SUPPORT_UTF */
3335            foc = TABLE_GET(fc, md->fcc, fc);
3336    #endif /* COMPILE_PCRE8 */
3337    
3338        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3339          {          {
3340          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
# Line 2739  for (;;) Line 3342  for (;;)
3342            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3343            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3344            }            }
3345          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3346            eptr++;
3347          }          }
3348        if (min == max) continue;        if (min == max) continue;
3349        if (minimize)        if (minimize)
3350          {          {
3351          for (fi = min;; fi++)          for (fi = min;; fi++)
3352            {            {
3353            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3354            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3355            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3356            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2754  for (;;) Line 3358  for (;;)
3358              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3359              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3360              }              }
3361            if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3362              eptr++;
3363            }            }
3364          /* Control never gets here */          /* Control never gets here */
3365          }          }
# Line 2763  for (;;) Line 3368  for (;;)
3368          pp = eptr;          pp = eptr;
3369          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3370            {            {
3371            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3372                {
3373                SCHECK_PARTIAL();
3374                break;
3375                }
3376              if (fc != *eptr && foc != *eptr) break;
3377            eptr++;            eptr++;
3378            }            }
3379    
# Line 2771  for (;;) Line 3381  for (;;)
3381    
3382          while (eptr >= pp)          while (eptr >= pp)
3383            {            {
3384            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3385            eptr--;            eptr--;
3386            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3387            }            }
# Line 2800  for (;;) Line 3410  for (;;)
3410          {          {
3411          for (fi = min;; fi++)          for (fi = min;; fi++)
3412            {            {
3413            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3414            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3415            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3416            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2817  for (;;) Line 3427  for (;;)
3427          pp = eptr;          pp = eptr;
3428          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3429            {            {
3430            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3431                {
3432                SCHECK_PARTIAL();
3433                break;
3434                }
3435              if (fc != *eptr) break;
3436            eptr++;            eptr++;
3437            }            }
3438          if (possessive) continue;          if (possessive) continue;
3439    
3440          while (eptr >= pp)          while (eptr >= pp)
3441            {            {
3442            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3443            eptr--;            eptr--;
3444            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3445            }            }
# Line 2837  for (;;) Line 3452  for (;;)
3452      checking can be multibyte. */      checking can be multibyte. */
3453    
3454      case OP_NOT:      case OP_NOT:
3455        case OP_NOTI:
3456      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3457        {        {
3458        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 2844  for (;;) Line 3460  for (;;)
3460        }        }
3461      ecode++;      ecode++;
3462      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3463      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3464        {        {
3465  #ifdef SUPPORT_UTF8        register int ch, och;
3466        if (c < 256)        ch = *ecode++;
3467  #endif  #ifdef COMPILE_PCRE8
3468        c = md->lcc[c];        /* ch must be < 128 if UTF is enabled. */
3469        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        och = md->fcc[ch];
3470    #else
3471    #ifdef SUPPORT_UTF
3472    #ifdef SUPPORT_UCP
3473          if (utf && ch > 127)
3474            och = UCD_OTHERCASE(ch);
3475    #else
3476          if (utf && ch > 127)
3477            och = ch;
3478    #endif /* SUPPORT_UCP */
3479          else
3480    #endif /* SUPPORT_UTF */
3481            och = TABLE_GET(ch, md->fcc, ch);
3482    #endif /* COMPILE_PCRE8 */
3483          if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3484        }        }
3485      else      else    /* Caseful */
3486        {        {
3487        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);
3488        }        }
# Line 2866  for (;;) Line 3496  for (;;)
3496      about... */      about... */
3497    
3498      case OP_NOTEXACT:      case OP_NOTEXACT:
3499        case OP_NOTEXACTI:
3500      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3501      ecode += 3;      ecode += 1 + IMM2_SIZE;
3502      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3503    
3504      case OP_NOTUPTO:      case OP_NOTUPTO:
3505        case OP_NOTUPTOI:
3506      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3507        case OP_NOTMINUPTOI:
3508      min = 0;      min = 0;
3509      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3510      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3511      ecode += 3;      ecode += 1 + IMM2_SIZE;
3512      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3513    
3514      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3515        case OP_NOTPOSSTARI:
3516      possessive = TRUE;      possessive = TRUE;
3517      min = 0;      min = 0;
3518      max = INT_MAX;      max = INT_MAX;
# Line 2886  for (;;) Line 3520  for (;;)
3520      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3521    
3522      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3523        case OP_NOTPOSPLUSI:
3524      possessive = TRUE;      possessive = TRUE;
3525      min = 1;      min = 1;
3526      max = INT_MAX;      max = INT_MAX;
# Line 2893  for (;;) Line 3528  for (;;)
3528      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3529    
3530      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3531        case OP_NOTPOSQUERYI:
3532      possessive = TRUE;      possessive = TRUE;
3533      min = 0;      min = 0;
3534      max = 1;      max = 1;
# Line 2900  for (;;) Line 3536  for (;;)
3536      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3537    
3538      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3539        case OP_NOTPOSUPTOI:
3540      possessive = TRUE;      possessive = TRUE;
3541      min = 0;      min = 0;
3542      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3543      ecode += 3;      ecode += 1 + IMM2_SIZE;
3544      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3545    
3546      case OP_NOTSTAR:      case OP_NOTSTAR:
3547        case OP_NOTSTARI:
3548      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3549        case OP_NOTMINSTARI:
3550      case OP_NOTPLUS:      case OP_NOTPLUS:
3551        case OP_NOTPLUSI:
3552      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3553        case OP_NOTMINPLUSI:
3554      case OP_NOTQUERY:      case OP_NOTQUERY:
3555        case OP_NOTQUERYI:
3556      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3557      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3558        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3559      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3560      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3561      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 2934  for (;;) Line 3577  for (;;)
3577      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,
3578        max, eptr));        max, eptr));
3579    
3580      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3581        {        {
3582        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3583          /* fc must be < 128 if UTF is enabled. */
3584          foc = md->fcc[fc];
3585    #else
3586    #ifdef SUPPORT_UTF
3587    #ifdef SUPPORT_UCP
3588          if (utf && fc > 127)
3589            foc = UCD_OTHERCASE(fc);
3590    #else
3591          if (utf && fc > 127)
3592            foc = fc;
3593    #endif /* SUPPORT_UCP */
3594          else
3595    #endif /* SUPPORT_UTF */
3596            foc = TABLE_GET(fc, md->fcc, fc);
3597    #endif /* COMPILE_PCRE8 */
3598    
3599  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3600        /* UTF-8 mode */        if (utf)
       if (utf8)  
3601          {          {
3602          register unsigned int d;          register unsigned int d;
3603          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 2951  for (;;) Line 3608  for (;;)
3608              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3609              }              }
3610            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3611            if (d < 256) d = md->lcc[d];            if (fc == d || foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) RRETURN(MATCH_NOMATCH);  
3612            }            }
3613          }          }
3614        else        else
3615  #endif  #endif
3616          /* Not UTF mode */
       /* Not UTF-8 mode */  
3617          {          {
3618          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3619            {            {
# Line 2967  for (;;) Line 3622  for (;;)
3622              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3623              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3624              }              }
3625            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3626              eptr++;
3627            }            }
3628          }          }
3629    
# Line 2975  for (;;) Line 3631  for (;;)
3631    
3632        if (minimize)        if (minimize)
3633          {          {
3634  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3635          /* UTF-8 mode */          if (utf)
         if (utf8)  
3636            {            {
3637            register unsigned int d;            register unsigned int d;
3638            for (fi = min;; fi++)            for (fi = min;; fi++)
3639              {              {
3640              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3641              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3642              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3643              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2991  for (;;) Line 3646  for (;;)
3646                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3647                }                }
3648              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3649              if (d < 256) d = md->lcc[d];              if (fc == d || foc == d) RRETURN(MATCH_NOMATCH);
             if (fc == d) RRETURN(MATCH_NOMATCH);  
3650              }              }
3651            }            }
3652          else          else
3653  #endif  #endif
3654          /* Not UTF-8 mode */          /* Not UTF mode */
3655            {            {
3656            for (fi = min;; fi++)            for (fi = min;; fi++)
3657              {              {
3658              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3659              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3660              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3661              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3009  for (;;) Line 3663  for (;;)
3663                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3664                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3665                }                }
3666              if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);              if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3667                eptr++;
3668              }              }
3669            }            }
3670          /* Control never gets here */          /* Control never gets here */
# Line 3021  for (;;) Line 3676  for (;;)
3676          {          {
3677          pp = eptr;          pp = eptr;
3678    
3679  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3680          /* UTF-8 mode */          if (utf)
         if (utf8)  
3681            {            {
3682            register unsigned int d;            register unsigned int d;
3683            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3684              {              {
3685              int len = 1;              int len = 1;
3686              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3687                  {
3688                  SCHECK_PARTIAL();
3689                  break;
3690                  }
3691              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3692              if (d < 256) d = md->lcc[d];              if (fc == d || foc == d) break;
             if (fc == d) break;  
3693              eptr += len;              eptr += len;
3694              }              }
3695          if (possessive) continue;          if (possessive) continue;
3696          for(;;)          for(;;)
3697              {              {
3698              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3699              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3700              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3701              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3046  for (;;) Line 3703  for (;;)
3703            }            }
3704          else          else
3705  #endif  #endif
3706          /* Not UTF-8 mode */          /* Not UTF mode */
3707            {            {
3708            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3709              {              {
3710              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3711                  {
3712                  SCHECK_PARTIAL();
3713                  break;
3714                  }
3715                if (fc == *eptr || foc == *eptr) break;
3716              eptr++;              eptr++;
3717              }              }
3718            if (possessive) continue;            if (possessive) continue;
3719            while (eptr >= pp)            while (eptr >= pp)
3720              {              {
3721              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3722              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3723              eptr--;              eptr--;
3724              }              }
# Line 3071  for (;;) Line 3733  for (;;)
3733    
3734      else      else
3735        {        {
3736  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3737        /* UTF-8 mode */        if (utf)
       if (utf8)  
3738          {          {
3739          register unsigned int d;          register unsigned int d;
3740          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3089  for (;;) Line 3750  for (;;)
3750          }          }
3751        else        else
3752  #endif  #endif
3753        /* Not UTF-8 mode */        /* Not UTF mode */
3754          {          {
3755          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3756            {            {
# Line 3106  for (;;) Line 3767  for (;;)
3767    
3768        if (minimize)        if (minimize)
3769          {          {
3770  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3771          /* UTF-8 mode */          if (utf)
         if (utf8)  
3772            {            {
3773            register unsigned int d;            register unsigned int d;
3774            for (fi = min;; fi++)            for (fi = min;; fi++)
3775              {              {
3776              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3777              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3778              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3779              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3127  for (;;) Line 3787  for (;;)
3787            }            }
3788          else          else
3789  #endif  #endif
3790   &nb