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

Diff of /code/trunk/pcre_exec.c

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

revision 409 by ph10, Sat Mar 28 17:10:56 2009 UTC revision 852 by zherczeg, Thu Jan 5 19:18:12 2012 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) 1997-2012 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          {
208          if (TABLE_GET(*p, md->lcc, *p) != TABLE_GET(*eptr, md->lcc, *eptr)) return -1;
209          p++;
210          eptr++;
211          }
212        }
213    }    }
214    
215  /* 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
216  are in UTF-8 mode. */  are in UTF-8 mode. */
217    
218  else  else
219    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
220      if (eptr + length > md->end_subject) return -1;
221      while (length-- > 0) if (*p++ != *eptr++) return -1;
222      }
223    
224  return TRUE;  return (int)(eptr - eptr_start);
225  }  }
226    
227    
# Line 245  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 272  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
272         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
273         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
274         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
275         RM51,  RM52, RM53, RM54 };         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
276           RM61,  RM62, RM63, RM64, RM65, RM66 };
277    
278  /* 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
279  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
280  actuall used in this definition. */  actually used in this definition. */
281    
282  #ifndef NO_RECURSE  #ifndef NO_RECURSE
283  #define REGISTER register  #define REGISTER register
284    
285  #ifdef DEBUG  #ifdef PCRE_DEBUG
286  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
287    { \    { \
288    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
289    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \
290    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
291    }    }
292  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 267  actuall used in this definition. */ Line 295  actuall used in this definition. */
295    return ra; \    return ra; \
296    }    }
297  #else  #else
298  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
299    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1)
300  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
301  #endif  #endif
302    
# Line 281  argument of match(), which never changes Line 309  argument of match(), which never changes
309    
310  #define REGISTER  #define REGISTER
311    
312  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
313    {\    {\
314    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\
315      if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
316    frame->Xwhere = rw; \    frame->Xwhere = rw; \
317    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
318    newframe->Xecode = rb;\    newframe->Xecode = rb;\
319    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
320    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
321    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
322    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
323    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
324    frame = newframe;\    frame = newframe;\
# Line 303  argument of match(), which never changes Line 330  argument of match(), which never changes
330    
331  #define RRETURN(ra)\  #define RRETURN(ra)\
332    {\    {\
333    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
334    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
335    (pcre_stack_free)(newframe);\    (PUBL(stack_free))(oldframe);\
336    if (frame != NULL)\    if (frame != NULL)\
337      {\      {\
338      rrc = ra;\      rrc = ra;\
# Line 322  typedef struct heapframe { Line 349  typedef struct heapframe {
349    
350    /* Function arguments that may change */    /* Function arguments that may change */
351    
352    USPTR Xeptr;    PCRE_PUCHAR Xeptr;
353    const uschar *Xecode;    const pcre_uchar *Xecode;
354    USPTR Xmstart;    PCRE_PUCHAR Xmstart;
355    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
356    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
357    unsigned int Xrdepth;    unsigned int Xrdepth;
358    
359    /* Function local variables */    /* Function local variables */
360    
361    USPTR Xcallpat;    PCRE_PUCHAR Xcallpat;
362  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
363    USPTR Xcharptr;    PCRE_PUCHAR Xcharptr;
364  #endif  #endif
365    USPTR Xdata;    PCRE_PUCHAR Xdata;
366    USPTR Xnext;    PCRE_PUCHAR Xnext;
367    USPTR Xpp;    PCRE_PUCHAR Xpp;
368    USPTR Xprev;    PCRE_PUCHAR Xprev;
369    USPTR Xsaved_eptr;    PCRE_PUCHAR Xsaved_eptr;
370    
371    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
372    
# Line 349  typedef struct heapframe { Line 374  typedef struct heapframe {
374    BOOL Xcondition;    BOOL Xcondition;
375    BOOL Xprev_is_word;    BOOL Xprev_is_word;
376    
   unsigned long int Xoriginal_ims;  
   
377  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
378    int Xprop_type;    int Xprop_type;
379    int Xprop_value;    int Xprop_value;
380    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
381    int Xoclength;    int Xoclength;
382    uschar Xocchars[8];    pcre_uchar Xocchars[6];
383  #endif  #endif
384    
385    int Xcodelink;    int Xcodelink;
# Line 398  typedef struct heapframe { Line 418  typedef struct heapframe {
418    
419  /* This function is called recursively in many circumstances. Whenever it  /* This function is called recursively in many circumstances. Whenever it
420  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
421  same response.  same response. */
422    
423    /* These macros pack up tests that are used for partial matching, and which
424    appear several times in the code. We set the "hit end" flag if the pointer is
425    at the end of the subject and also past the start of the subject (i.e.
426    something has been matched). For hard partial matching, we then return
427    immediately. The second one is used when we already know we are past the end of
428    the subject. */
429    
430    #define CHECK_PARTIAL()\
431      if (md->partial != 0 && eptr >= md->end_subject && \
432          eptr > md->start_used_ptr) \
433        { \
434        md->hitend = TRUE; \
435        if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
436        }
437    
438  Performance note: It might be tempting to extract commonly used fields from the  #define SCHECK_PARTIAL()\
439  md structure (e.g. utf8, end_subject) into individual variables to improve    if (md->partial != 0 && eptr > md->start_used_ptr) \
440        { \
441        md->hitend = TRUE; \
442        if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
443        }
444    
445    
446    /* Performance note: It might be tempting to extract commonly used fields from
447    the md structure (e.g. utf, end_subject) into individual variables to improve
448  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
449  made performance worse.  made performance worse.
450    
# Line 412  Arguments: Line 455  Arguments:
455                   by encountering \K)                   by encountering \K)
456     offset_top  current top pointer     offset_top  current top pointer
457     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
458     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
459                   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  
460     rdepth      the recursion depth     rdepth      the recursion depth
461    
462  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
463                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
464                   a negative MATCH_xxx value for PRUNE, SKIP, etc
465                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
466                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
467  */  */
468    
469  static int  static int
470  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode,
471    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
472    int flags, unsigned int rdepth)    unsigned int rdepth)
473  {  {
474  /* 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,
475  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 439  so they can be ordinary variables in all Line 478  so they can be ordinary variables in all
478  register int  rrc;         /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
479  register int  i;           /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
480  register unsigned int c;   /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
481  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf;         /* Local copy of UTF flag for speed */
482    
483  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
484    BOOL caseless;
485  int condcode;  int condcode;
486    
487  /* 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 450  heap storage. Set up the top-level frame Line 490  heap storage. Set up the top-level frame
490  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
491    
492  #ifdef NO_RECURSE  #ifdef NO_RECURSE
493  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));
494    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
495  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
496    
497  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 459  frame->Xeptr = eptr; Line 500  frame->Xeptr = eptr;
500  frame->Xecode = ecode;  frame->Xecode = ecode;
501  frame->Xmstart = mstart;  frame->Xmstart = mstart;
502  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
503  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
504  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
505    
506  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 474  HEAP_RECURSE: Line 513  HEAP_RECURSE:
513  #define ecode              frame->Xecode  #define ecode              frame->Xecode
514  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
515  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
516  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
517  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
518    
519  /* Ditto for the local variables */  /* Ditto for the local variables */
520    
521  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
522  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
523  #endif  #endif
524  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 498  HEAP_RECURSE: Line 535  HEAP_RECURSE:
535  #define condition          frame->Xcondition  #define condition          frame->Xcondition
536  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
537    
 #define original_ims       frame->Xoriginal_ims  
   
538  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
539  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
540  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
541  #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  
542  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
543  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
544  #endif  #endif
# Line 536  i, and fc and c, can be the same variabl Line 568  i, and fc and c, can be the same variabl
568  #define fi i  #define fi i
569  #define fc c  #define fc c
570    
571    /* Many of the following variables are used only in small blocks of the code.
572    My normal style of coding would have declared them within each of those blocks.
573    However, in order to accommodate the version of this code that uses an external
574    "stack" implemented on the heap, it is easier to declare them all here, so the
575    declarations can be cut out in a block. The only declarations within blocks
576    below are for variables that do not have to be preserved over a recursive call
577    to RMATCH(). */
578    
579    #ifdef SUPPORT_UTF
580    const pcre_uchar *charptr;
581    #endif
582    const pcre_uchar *callpat;
583    const pcre_uchar *data;
584    const pcre_uchar *next;
585    PCRE_PUCHAR       pp;
586    const pcre_uchar *prev;
587    PCRE_PUCHAR       saved_eptr;
588    
589  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  recursion_info new_recursive;
590  const uschar *charptr;             /* in small blocks of the code. My normal */  
591  #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().          */  
592  BOOL condition;  BOOL condition;
593  BOOL prev_is_word;  BOOL prev_is_word;
594    
 unsigned long int original_ims;  
   
595  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
596  int prop_type;  int prop_type;
597  int prop_value;  int prop_value;
598  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
599  int oclength;  int oclength;
600  uschar occhars[8];  pcre_uchar occhars[6];
601  #endif  #endif
602    
603  int codelink;  int codelink;
# Line 581  int stacksave[REC_STACK_SAVE_MAX]; Line 615  int stacksave[REC_STACK_SAVE_MAX];
615  eptrblock newptrb;  eptrblock newptrb;
616  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
617    
618    /* To save space on the stack and in the heap frame, I have doubled up on some
619    of the local variables that are used only in localised parts of the code, but
620    still need to be preserved over recursive calls of match(). These macros define
621    the alternative names that are used. */
622    
623    #define allow_zero    cur_is_word
624    #define cbegroup      condition
625    #define code_offset   codelink
626    #define condassert    condition
627    #define matched_once  prev_is_word
628    #define foc           number
629    
630  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
631  variables. */  variables. */
632    
# Line 600  TAIL_RECURSE: Line 646  TAIL_RECURSE:
646  /* 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
647  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
648  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()
649  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
650  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
651  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,
652  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
653    
654  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
655  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
656  #else  #else
657  utf8 = FALSE;  utf = FALSE;
658  #endif  #endif
659    
660  /* 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 617  haven't exceeded the recursive call limi Line 663  haven't exceeded the recursive call limi
663  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
664  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
665    
 original_ims = ims;    /* Save for resetting on ')' */  
   
666  /* 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
667  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
668  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
669  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
670  When match() is called in other circumstances, don't add to the chain. The  
671  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
672  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
673  match(). */  to break infinite loops that match no characters. When match() is called in
674    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
675    NOT be used with tail recursion, because the memory block that is used is on
676    the stack, so a new one may be required for each match(). */
677    
678  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
679    {    {
680    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
681    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
682    eptrb = &newptrb;    eptrb = &newptrb;
683      md->match_function_type = 0;
684    }    }
685    
686  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 642  for (;;) Line 690  for (;;)
690    minimize = possessive = FALSE;    minimize = possessive = FALSE;
691    op = *ecode;    op = *ecode;
692    
   /* For partial matching, remember if we ever hit the end of the subject after  
   matching at least one subject character. */  
   
   if (md->partial &&  
       eptr >= md->end_subject &&  
       eptr > mstart)  
     md->hitend = TRUE;  
   
693    switch(op)    switch(op)
694      {      {
695        case OP_MARK:
696        md->nomatch_mark = ecode + 2;
697        md->mark = NULL;    /* In case previously set by assertion */
698        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
699          eptrb, RM55);
700        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
701             md->mark == NULL) md->mark = ecode + 2;
702    
703        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
704        argument, and we must check whether that argument matches this MARK's
705        argument. It is passed back in md->start_match_ptr (an overloading of that
706        variable). If it does match, we reset that variable to the current subject
707        position and return MATCH_SKIP. Otherwise, pass back the return code
708        unaltered. */
709    
710        else if (rrc == MATCH_SKIP_ARG &&
711            STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0)
712          {
713          md->start_match_ptr = eptr;
714          RRETURN(MATCH_SKIP);
715          }
716        RRETURN(rrc);
717    
718      case OP_FAIL:      case OP_FAIL:
719      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
720    
721      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);  
722    
723      case OP_COMMIT:      case OP_COMMIT:
724      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
725        ims, eptrb, flags, RM52);        eptrb, RM52);
726      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
727            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
728            rrc != MATCH_THEN)
729          RRETURN(rrc);
730      RRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
731    
732        /* PRUNE overrides THEN */
733    
734        case OP_PRUNE:
735        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
736          eptrb, RM51);
737        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
738        RRETURN(MATCH_PRUNE);
739    
740        case OP_PRUNE_ARG:
741        md->nomatch_mark = ecode + 2;
742        md->mark = NULL;    /* In case previously set by assertion */
743        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
744          eptrb, RM56);
745        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
746             md->mark == NULL) md->mark = ecode + 2;
747        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
748        RRETURN(MATCH_PRUNE);
749    
750        /* SKIP overrides PRUNE and THEN */
751    
752      case OP_SKIP:      case OP_SKIP:
753      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
754        ims, eptrb, flags, RM53);        eptrb, RM53);
755      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
756          RRETURN(rrc);
757      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
758      RRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
759    
760        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
761        nomatch_mark. There is a flag that disables this opcode when re-matching a
762        pattern that ended with a SKIP for which there was not a matching MARK. */
763    
764        case OP_SKIP_ARG:
765        if (md->ignore_skip_arg)
766          {
767          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
768          break;
769          }
770        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
771          eptrb, RM57);
772        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
773          RRETURN(rrc);
774    
775        /* Pass back the current skip name by overloading md->start_match_ptr and
776        returning the special MATCH_SKIP_ARG return code. This will either be
777        caught by a matching MARK, or get to the top, where it causes a rematch
778        with the md->ignore_skip_arg flag set. */
779    
780        md->start_match_ptr = ecode + 2;
781        RRETURN(MATCH_SKIP_ARG);
782    
783        /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
784        the branch in which it occurs can be determined. Overload the start of
785        match pointer to do this. */
786    
787      case OP_THEN:      case OP_THEN:
788      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
789        ims, eptrb, flags, RM54);        eptrb, RM54);
790        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
791        md->start_match_ptr = ecode;
792        RRETURN(MATCH_THEN);
793    
794        case OP_THEN_ARG:
795        md->nomatch_mark = ecode + 2;
796        md->mark = NULL;    /* In case previously set by assertion */
797        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
798          md, eptrb, RM58);
799        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
800             md->mark == NULL) md->mark = ecode + 2;
801      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
802        md->start_match_ptr = ecode;
803      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
804    
805      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle an atomic group that does not contain any capturing parentheses.
806      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
807      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
808      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.
809      reference inside the group.      However, this uses a lot more stack, so in 8.20, atomic groups that do not
810        contain any captures generate OP_ONCE_NC, which can be handled in the old,
811      If the bracket fails to match, we need to restore this value and also the      less stack intensive way.
812      values of the final offsets, in case they were set by a previous iteration  
813      of the same bracket.      Check the alternative branches in turn - the matching won't pass the KET
814        for this kind of subpattern. If any one branch matches, we carry on as at
815        the end of a normal bracket, leaving the subject pointer, but resetting
816        the start-of-match value in case it was changed by \K. */
817    
818        case OP_ONCE_NC:
819        prev = ecode;
820        saved_eptr = eptr;
821        do
822          {
823          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
824          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
825            {
826            mstart = md->start_match_ptr;
827            break;
828            }
829          if (rrc == MATCH_THEN)
830            {
831            next = ecode + GET(ecode,1);
832            if (md->start_match_ptr < next &&
833                (*ecode == OP_ALT || *next == OP_ALT))
834              rrc = MATCH_NOMATCH;
835            }
836    
837          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
838          ecode += GET(ecode,1);
839          }
840        while (*ecode == OP_ALT);
841    
842        /* If hit the end of the group (which could be repeated), fail */
843    
844        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
845    
846        /* Continue as from after the group, updating the offsets high water
847        mark, since extracts may have been taken. */
848    
849        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
850    
851        offset_top = md->end_offset_top;
852        eptr = md->end_match_ptr;
853    
854        /* For a non-repeating ket, just continue at this level. This also
855        happens for a repeating ket if no characters were matched in the group.
856        This is the forcible breaking of infinite loops as implemented in Perl
857        5.005. */
858    
859        if (*ecode == OP_KET || eptr == saved_eptr)
860          {
861          ecode += 1+LINK_SIZE;
862          break;
863          }
864    
865        /* The repeating kets try the rest of the pattern or restart from the
866        preceding bracket, in the appropriate order. The second "call" of match()
867        uses tail recursion, to avoid using another stack frame. */
868    
869        if (*ecode == OP_KETRMIN)
870          {
871          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
872          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
873          ecode = prev;
874          goto TAIL_RECURSE;
875          }
876        else  /* OP_KETRMAX */
877          {
878          md->match_function_type = MATCH_CBEGROUP;
879          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
880          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
881          ecode += 1 + LINK_SIZE;
882          goto TAIL_RECURSE;
883          }
884        /* Control never gets here */
885    
886        /* Handle a capturing bracket, other than those that are possessive with an
887        unlimited repeat. If there is space in the offset vector, save the current
888        subject position in the working slot at the top of the vector. We mustn't
889        change the current values of the data slot, because they may be set from a
890        previous iteration of this group, and be referred to by a reference inside
891        the group. A failure to match might occur after the group has succeeded,
892        if something later on doesn't match. For this reason, we need to restore
893        the working value and also the values of the final offsets, in case they
894        were set by a previous iteration of the same bracket.
895    
896      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
897      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 699  for (;;) Line 902  for (;;)
902      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
903      offset = number << 1;      offset = number << 1;
904    
905  #ifdef DEBUG  #ifdef PCRE_DEBUG
906      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
907      printf("subject=");      printf("subject=");
908      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 714  for (;;) Line 917  for (;;)
917        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
918    
919        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
920        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
921            (int)(eptr - md->start_subject);
922    
923        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
924          {          {
925          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
926            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
927          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);            eptrb, RM1);
928            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
929    
930            /* If we backed up to a THEN, check whether it is within the current
931            branch by comparing the address of the THEN that is passed back with
932            the end of the branch. If it is within the current branch, and the
933            branch is one of two or more alternatives (it either starts or ends
934            with OP_ALT), we have reached the limit of THEN's action, so convert
935            the return code to NOMATCH, which will cause normal backtracking to
936            happen from now on. Otherwise, THEN is passed back to an outer
937            alternative. This implements Perl's treatment of parenthesized groups,
938            where a group not containing | does not affect the current alternative,
939            that is, (X) is NOT the same as (X|(*F)). */
940    
941            if (rrc == MATCH_THEN)
942              {
943              next = ecode + GET(ecode,1);
944              if (md->start_match_ptr < next &&
945                  (*ecode == OP_ALT || *next == OP_ALT))
946                rrc = MATCH_NOMATCH;
947              }
948    
949            /* Anything other than NOMATCH is passed back. */
950    
951            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
952          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
953          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
954            if (*ecode != OP_ALT) break;
955          }          }
       while (*ecode == OP_ALT);  
956    
957        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
958        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
959        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
960        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
961    
962        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
963    
964          RRETURN(rrc);
965        }        }
966    
967      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 747  for (;;) Line 975  for (;;)
975      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
976      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
977    
978      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
979      final alternative within the brackets, we would return the result of a      repeat and ONCE group with no captures. Loop for all the alternatives.
980      recursive call to match() whatever happened. We can reduce stack usage by  
981      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
982      is set.*/      the result of a recursive call to match() whatever happened so it was
983        possible to reduce stack usage by turning this into a tail recursion,
984        except in the case of a possibly empty group. However, now that there is
985        the possiblity of (*THEN) occurring in the final alternative, this
986        optimization is no longer always possible.
987    
988        We can optimize if we know there are no (*THEN)s in the pattern; at present
989        this is the best that can be done.
990    
991        MATCH_ONCE is returned when the end of an atomic group is successfully
992        reached, but subsequent matching fails. It passes back up the tree (causing
993        captured values to be reset) until the original atomic group level is
994        reached. This is tested by comparing md->once_target with the start of the
995        group. At this point, the return is converted into MATCH_NOMATCH so that
996        previous backup points can be taken. */
997    
998        case OP_ONCE:
999      case OP_BRA:      case OP_BRA:
1000      case OP_SBRA:      case OP_SBRA:
1001      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
1002      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
1003      for (;;)      for (;;)
1004        {        {
1005        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
1006    
1007          /* If this is not a possibly empty group, and there are no (*THEN)s in
1008          the pattern, and this is the final alternative, optimize as described
1009          above. */
1010    
1011          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1012            {
1013            ecode += PRIV(OP_lengths)[*ecode];
1014            goto TAIL_RECURSE;
1015            }
1016    
1017          /* In all other cases, we have to make another call to match(). */
1018    
1019          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1020            RM2);
1021    
1022          /* See comment in the code for capturing groups above about handling
1023          THEN. */
1024    
1025          if (rrc == MATCH_THEN)
1026            {
1027            next = ecode + GET(ecode,1);
1028            if (md->start_match_ptr < next &&
1029                (*ecode == OP_ALT || *next == OP_ALT))
1030              rrc = MATCH_NOMATCH;
1031            }
1032    
1033          if (rrc != MATCH_NOMATCH)
1034          {          {
1035          if (flags == 0)    /* Not a possibly empty group */          if (rrc == MATCH_ONCE)
1036            {            {
1037            ecode += _pcre_OP_lengths[*ecode];            const pcre_uchar *scode = ecode;
1038            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1039            goto TAIL_RECURSE;              {
1040                while (*scode == OP_ALT) scode += GET(scode, 1);
1041                scode -= GET(scode, 1);
1042                }
1043              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1044              }
1045            RRETURN(rrc);
1046            }
1047          ecode += GET(ecode, 1);
1048          if (*ecode != OP_ALT) break;
1049          }
1050    
1051        RRETURN(MATCH_NOMATCH);
1052    
1053        /* Handle possessive capturing brackets with an unlimited repeat. We come
1054        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1055        handled similarly to the normal case above. However, the matching is
1056        different. The end of these brackets will always be OP_KETRPOS, which
1057        returns MATCH_KETRPOS without going further in the pattern. By this means
1058        we can handle the group by iteration rather than recursion, thereby
1059        reducing the amount of stack needed. */
1060    
1061        case OP_CBRAPOS:
1062        case OP_SCBRAPOS:
1063        allow_zero = FALSE;
1064    
1065        POSSESSIVE_CAPTURE:
1066        number = GET2(ecode, 1+LINK_SIZE);
1067        offset = number << 1;
1068    
1069    #ifdef PCRE_DEBUG
1070        printf("start possessive bracket %d\n", number);
1071        printf("subject=");
1072        pchars(eptr, 16, TRUE, md);
1073        printf("\n");
1074    #endif
1075    
1076        if (offset < md->offset_max)
1077          {
1078          matched_once = FALSE;
1079          code_offset = (int)(ecode - md->start_code);
1080    
1081          save_offset1 = md->offset_vector[offset];
1082          save_offset2 = md->offset_vector[offset+1];
1083          save_offset3 = md->offset_vector[md->offset_end - number];
1084          save_capture_last = md->capture_last;
1085    
1086          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1087    
1088          /* Each time round the loop, save the current subject position for use
1089          when the group matches. For MATCH_MATCH, the group has matched, so we
1090          restart it with a new subject starting position, remembering that we had
1091          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1092          usual. If we haven't matched any alternatives in any iteration, check to
1093          see if a previous iteration matched. If so, the group has matched;
1094          continue from afterwards. Otherwise it has failed; restore the previous
1095          capture values before returning NOMATCH. */
1096    
1097          for (;;)
1098            {
1099            md->offset_vector[md->offset_end - number] =
1100              (int)(eptr - md->start_subject);
1101            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1102            RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1103              eptrb, RM63);
1104            if (rrc == MATCH_KETRPOS)
1105              {
1106              offset_top = md->end_offset_top;
1107              eptr = md->end_match_ptr;
1108              ecode = md->start_code + code_offset;
1109              save_capture_last = md->capture_last;
1110              matched_once = TRUE;
1111              continue;
1112              }
1113    
1114            /* See comment in the code for capturing groups above about handling
1115            THEN. */
1116    
1117            if (rrc == MATCH_THEN)
1118              {
1119              next = ecode + GET(ecode,1);
1120              if (md->start_match_ptr < next &&
1121                  (*ecode == OP_ALT || *next == OP_ALT))
1122                rrc = MATCH_NOMATCH;
1123            }            }
1124    
1125          /* Possibly empty group; can't use tail recursion. */          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1126            md->capture_last = save_capture_last;
1127            ecode += GET(ecode, 1);
1128            if (*ecode != OP_ALT) break;
1129            }
1130    
1131          if (!matched_once)
1132            {
1133            md->offset_vector[offset] = save_offset1;
1134            md->offset_vector[offset+1] = save_offset2;
1135            md->offset_vector[md->offset_end - number] = save_offset3;
1136            }
1137    
1138          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (allow_zero || matched_once)
1139            eptrb, flags, RM48);          {
1140          RRETURN(rrc);          ecode += 1 + LINK_SIZE;
1141            break;
1142          }          }
1143    
1144        /* For non-final alternatives, continue the loop for a NOMATCH result;        RRETURN(MATCH_NOMATCH);
1145        otherwise return. */        }
1146    
1147        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1148        as a non-capturing bracket. */
1149    
1150        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1151          eptrb, flags, RM2);      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1152        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);  
1153        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1154    
1155        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1156        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1157    
1158        /* Non-capturing possessive bracket with unlimited repeat. We come here
1159        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1160        without the capturing complication. It is written out separately for speed
1161        and cleanliness. */
1162    
1163        case OP_BRAPOS:
1164        case OP_SBRAPOS:
1165        allow_zero = FALSE;
1166    
1167        POSSESSIVE_NON_CAPTURE:
1168        matched_once = FALSE;
1169        code_offset = (int)(ecode - md->start_code);
1170    
1171        for (;;)
1172          {
1173          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1174          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1175            eptrb, RM48);
1176          if (rrc == MATCH_KETRPOS)
1177            {
1178            offset_top = md->end_offset_top;
1179            eptr = md->end_match_ptr;
1180            ecode = md->start_code + code_offset;
1181            matched_once = TRUE;
1182            continue;
1183            }
1184    
1185          /* See comment in the code for capturing groups above about handling
1186          THEN. */
1187    
1188          if (rrc == MATCH_THEN)
1189            {
1190            next = ecode + GET(ecode,1);
1191            if (md->start_match_ptr < next &&
1192                (*ecode == OP_ALT || *next == OP_ALT))
1193              rrc = MATCH_NOMATCH;
1194            }
1195    
1196          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1197        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1198          if (*ecode != OP_ALT) break;
1199          }
1200    
1201        if (matched_once || allow_zero)
1202          {
1203          ecode += 1 + LINK_SIZE;
1204          break;
1205        }        }
1206        RRETURN(MATCH_NOMATCH);
1207    
1208      /* Control never reaches here. */      /* Control never reaches here. */
1209    
1210      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1211      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
1212      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
1213      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. */  
1214    
1215      case OP_COND:      case OP_COND:
1216      case OP_SCOND:      case OP_SCOND:
1217      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1218    
1219      /* 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
1220      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
1221    
1222      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1223        {        {
1224        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1225          {          {
1226          pcre_callout_block cb;          PUBL(callout_block) cb;
1227          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1228          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1229          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1230    #ifdef COMPILE_PCRE8
1231          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1232          cb.subject_length   = md->end_subject - md->start_subject;  #else
1233          cb.start_match      = mstart - md->start_subject;          cb.subject          = (PCRE_SPTR16)md->start_subject;
1234          cb.current_position = eptr - md->start_subject;  #endif
1235            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1236            cb.start_match      = (int)(mstart - md->start_subject);
1237            cb.current_position = (int)(eptr - md->start_subject);
1238          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1239          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1240          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1241          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1242          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1243          if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1244            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1245          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1246          }          }
1247        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1248        }        }
1249    
1250      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
1251    
1252      /* Now see what the actual condition is */      /* Now see what the actual condition is */
1253    
1254      if (condcode == OP_RREF)         /* Recursion test */      if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1255        {        {
1256        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (md->recursive == NULL)                /* Not recursing => FALSE */
1257        condition = md->recursive != NULL &&          {
1258          (offset == RREF_ANY || offset == md->recursive->group_num);          condition = FALSE;
1259        ecode += condition? 3 : GET(ecode, 1);          ecode += GET(ecode, 1);
1260            }
1261          else
1262            {
1263            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1264            condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1265    
1266            /* If the test is for recursion into a specific subpattern, and it is
1267            false, but the test was set up by name, scan the table to see if the
1268            name refers to any other numbers, and test them. The condition is true
1269            if any one is set. */
1270    
1271            if (!condition && condcode == OP_NRREF)
1272              {
1273              pcre_uchar *slotA = md->name_table;
1274              for (i = 0; i < md->name_count; i++)
1275                {
1276                if (GET2(slotA, 0) == recno) break;
1277                slotA += md->name_entry_size;
1278                }
1279    
1280              /* Found a name for the number - there can be only one; duplicate
1281              names for different numbers are allowed, but not vice versa. First
1282              scan down for duplicates. */
1283    
1284              if (i < md->name_count)
1285                {
1286                pcre_uchar *slotB = slotA;
1287                while (slotB > md->name_table)
1288                  {
1289                  slotB -= md->name_entry_size;
1290                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1291                    {
1292                    condition = GET2(slotB, 0) == md->recursive->group_num;
1293                    if (condition) break;
1294                    }
1295                  else break;
1296                  }
1297    
1298                /* Scan up for duplicates */
1299    
1300                if (!condition)
1301                  {
1302                  slotB = slotA;
1303                  for (i++; i < md->name_count; i++)
1304                    {
1305                    slotB += md->name_entry_size;
1306                    if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1307                      {
1308                      condition = GET2(slotB, 0) == md->recursive->group_num;
1309                      if (condition) break;
1310                      }
1311                    else break;
1312                    }
1313                  }
1314                }
1315              }
1316    
1317            /* Chose branch according to the condition */
1318    
1319            ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1320            }
1321        }        }
1322    
1323      else if (condcode == OP_CREF)    /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1324        {        {
1325        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1326        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1327        ecode += condition? 3 : GET(ecode, 1);  
1328          /* If the numbered capture is unset, but the reference was by name,
1329          scan the table to see if the name refers to any other numbers, and test
1330          them. The condition is true if any one is set. This is tediously similar
1331          to the code above, but not close enough to try to amalgamate. */
1332    
1333          if (!condition && condcode == OP_NCREF)
1334            {
1335            int refno = offset >> 1;
1336            pcre_uchar *slotA = md->name_table;
1337    
1338            for (i = 0; i < md->name_count; i++)
1339              {
1340              if (GET2(slotA, 0) == refno) break;
1341              slotA += md->name_entry_size;
1342              }
1343    
1344            /* Found a name for the number - there can be only one; duplicate names
1345            for different numbers are allowed, but not vice versa. First scan down
1346            for duplicates. */
1347    
1348            if (i < md->name_count)
1349              {
1350              pcre_uchar *slotB = slotA;
1351              while (slotB > md->name_table)
1352                {
1353                slotB -= md->name_entry_size;
1354                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1355                  {
1356                  offset = GET2(slotB, 0) << 1;
1357                  condition = offset < offset_top &&
1358                    md->offset_vector[offset] >= 0;
1359                  if (condition) break;
1360                  }
1361                else break;
1362                }
1363    
1364              /* Scan up for duplicates */
1365    
1366              if (!condition)
1367                {
1368                slotB = slotA;
1369                for (i++; i < md->name_count; i++)
1370                  {
1371                  slotB += md->name_entry_size;
1372                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1373                    {
1374                    offset = GET2(slotB, 0) << 1;
1375                    condition = offset < offset_top &&
1376                      md->offset_vector[offset] >= 0;
1377                    if (condition) break;
1378                    }
1379                  else break;
1380                  }
1381                }
1382              }
1383            }
1384    
1385          /* Chose branch according to the condition */
1386    
1387          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1388        }        }
1389    
1390      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 847  for (;;) Line 1394  for (;;)
1394        }        }
1395    
1396      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1397      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
1398      assertion. */      an assertion. */
1399    
1400      else      else
1401        {        {
1402        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1403            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1404        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1405          {          {
1406            if (md->end_offset_top > offset_top)
1407              offset_top = md->end_offset_top;  /* Captures may have happened */
1408          condition = TRUE;          condition = TRUE;
1409          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1410          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1411          }          }
1412    
1413          /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1414          assertion; it is therefore treated as NOMATCH. */
1415    
1416        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1417          {          {
1418          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
# Line 871  for (;;) Line 1424  for (;;)
1424          }          }
1425        }        }
1426    
1427      /* 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
1428      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
1429      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
1430      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
1431        exist, in which case we can just plough on. Note that, for compatibility
1432        with Perl, the | in a conditional group is NOT treated as creating two
1433        alternatives. If a THEN is encountered in the branch, it propagates out to
1434        the enclosing alternative (unless nested in a deeper set of alternatives,
1435        of course). */
1436    
1437      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1438        {        {
1439        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 */  
1440          {          {
1441          flags = 0;          ecode += 1 + LINK_SIZE;
1442          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1443          }          }
1444    
1445          md->match_function_type = MATCH_CBEGROUP;
1446          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1447          RRETURN(rrc);
1448        }        }
1449      else                         /* Condition false & no alternative */  
1450         /* Condition false & no alternative; continue after the group. */
1451    
1452        else
1453        {        {
1454        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1455        }        }
1456      break;      break;
1457    
1458    
1459      /* End of the pattern, either real or forced. If we are in a top-level      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1460      recursion, we should restore the offsets appropriately and continue from      to close any currently open capturing brackets. */
     after the call. */  
1461    
1462      case OP_ACCEPT:      case OP_CLOSE:
1463      case OP_END:      number = GET2(ecode, 1);
1464      if (md->recursive != NULL && md->recursive->group_num == 0)      offset = number << 1;
       {  
       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));  
       mstart = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1465    
1466      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty  #ifdef PCRE_DEBUG
1467      string - backtracking will then try other alternatives, if any. */        printf("end bracket %d at *ACCEPT", number);
1468          printf("\n");
1469    #endif
1470    
1471        md->capture_last = number;
1472        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1473          {
1474          md->offset_vector[offset] =
1475            md->offset_vector[md->offset_end - number];
1476          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1477          if (offset_top <= offset) offset_top = offset + 2;
1478          }
1479        ecode += 1 + IMM2_SIZE;
1480        break;
1481    
1482    
1483        /* End of the pattern, either real or forced. */
1484    
1485        case OP_END:
1486        case OP_ACCEPT:
1487        case OP_ASSERT_ACCEPT:
1488    
1489        /* If we have matched an empty string, fail if not in an assertion and not
1490        in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1491        is set and we have matched at the start of the subject. In both cases,
1492        backtracking will then try other alternatives, if any. */
1493    
1494        if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1495             md->recursive == NULL &&
1496             (md->notempty ||
1497               (md->notempty_atstart &&
1498                 mstart == md->start_subject + md->start_offset)))
1499          RRETURN(MATCH_NOMATCH);
1500    
1501        /* Otherwise, we have a match. */
1502    
     if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH);  
1503      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1504      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1505      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);  
1506    
1507      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1508        given as the argument to RRETURN when the heap is in use. */
1509    
1510      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1511      ims = ecode[1];      RRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1512    
1513      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1514      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,
1515      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
1516      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
1517      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1518        of a condition, we want to return immediately afterwards. The caller of
1519        this incarnation of the match() function will have set MATCH_CONDASSERT in
1520        md->match_function type, and one of these opcodes will be the first opcode
1521        that is processed. We use a local variable that is preserved over calls to
1522        match() to remember this case. */
1523    
1524      case OP_ASSERT:      case OP_ASSERT:
1525      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1526        if (md->match_function_type == MATCH_CONDASSERT)
1527          {
1528          condassert = TRUE;
1529          md->match_function_type = 0;
1530          }
1531        else condassert = FALSE;
1532    
1533      do      do
1534        {        {
1535        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1536          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1537        if (rrc == MATCH_MATCH) break;          {
1538            mstart = md->start_match_ptr;   /* In case \K reset it */
1539            break;
1540            }
1541    
1542          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1543          as NOMATCH. */
1544    
1545        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1546        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1547        }        }
1548      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1549    
1550      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1551    
1552      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1553    
1554      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1555    
1556      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1557      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 964  for (;;) Line 1561  for (;;)
1561      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1562      continue;      continue;
1563    
1564      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1565        PRUNE, or COMMIT means we must assume failure without checking subsequent
1566        branches. */
1567    
1568      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1569      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1570        if (md->match_function_type == MATCH_CONDASSERT)
1571          {
1572          condassert = TRUE;
1573          md->match_function_type = 0;
1574          }
1575        else condassert = FALSE;
1576    
1577      do      do
1578        {        {
1579        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1580          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1581        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1582            {
1583            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1584            break;
1585            }
1586    
1587          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1588          as NOMATCH. */
1589    
1590        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1591        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1592        }        }
1593      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1594    
1595      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1596    
1597      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1598      continue;      continue;
# Line 989  for (;;) Line 1603  for (;;)
1603      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1604    
1605      case OP_REVERSE:      case OP_REVERSE:
1606  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1607      if (utf8)      if (utf)
1608        {        {
1609        i = GET(ecode, 1);        i = GET(ecode, 1);
1610        while (i-- > 0)        while (i-- > 0)
# Line 1010  for (;;) Line 1624  for (;;)
1624        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1625        }        }
1626    
1627      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1628    
1629        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1630      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1631      break;      break;
1632    
# Line 1020  for (;;) Line 1635  for (;;)
1635      function is able to force a failure. */      function is able to force a failure. */
1636    
1637      case OP_CALLOUT:      case OP_CALLOUT:
1638      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1639        {        {
1640        pcre_callout_block cb;        PUBL(callout_block) cb;
1641        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1642        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1643        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1644    #ifdef COMPILE_PCRE8
1645        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1646        cb.subject_length   = md->end_subject - md->start_subject;  #else
1647        cb.start_match      = mstart - md->start_subject;        cb.subject          = (PCRE_SPTR16)md->start_subject;
1648        cb.current_position = eptr - md->start_subject;  #endif
1649          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1650          cb.start_match      = (int)(mstart - md->start_subject);
1651          cb.current_position = (int)(eptr - md->start_subject);
1652        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1653        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1654        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1655        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1656        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1657        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1658          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1659        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1660        }        }
1661      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1045  for (;;) Line 1665  for (;;)
1665      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
1666      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1667    
1668      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1669      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
1670      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
1671      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
1672      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
1673      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
1674      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.  
1675    
1676      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
1677      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
1678      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1679        a lot, so he is not to blame for the current way it works. */
1680    
1681      case OP_RECURSE:      case OP_RECURSE:
1682        {        {
1683          recursion_info *ri;
1684          int recno;
1685    
1686        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1687        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1688          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1689    
1690          /* Check for repeating a recursion without advancing the subject pointer.
1691          This should catch convoluted mutual recursions. (Some simple cases are
1692          caught at compile time.) */
1693    
1694          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1695            if (recno == ri->group_num && eptr == ri->subject_position)
1696              RRETURN(PCRE_ERROR_RECURSELOOP);
1697    
1698        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1699    
1700          new_recursive.group_num = recno;
1701          new_recursive.subject_position = eptr;
1702        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1703        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1704    
1705        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1706    
1707        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1708    
1709        /* Now save the offset data. */        /* Now save the offset data */
1710    
1711        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1712        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1084  for (;;) Line 1714  for (;;)
1714        else        else
1715          {          {
1716          new_recursive.offset_save =          new_recursive.offset_save =
1717            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1718          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1719          }          }
   
1720        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1721              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = mstart;  
       mstart = eptr;  
1722    
1723        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1724        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1725          might be changed, so reset it before looping. */
1726    
1727        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1728        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1729        do        do
1730          {          {
1731          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1732            md, ims, eptrb, flags, RM6);          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1733          if (rrc == MATCH_MATCH)            md, eptrb, RM6);
1734            memcpy(md->offset_vector, new_recursive.offset_save,
1735                new_recursive.saved_max * sizeof(int));
1736            md->recursive = new_recursive.prevrec;
1737            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1738            {            {
1739            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1740            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1741              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1742            RRETURN(MATCH_MATCH);  
1743              /* Set where we got to in the subject, and reset the start in case
1744              it was changed by \K. This *is* propagated back out of a recursion,
1745              for Perl compatibility. */
1746    
1747              eptr = md->end_match_ptr;
1748              mstart = md->start_match_ptr;
1749              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1750            }            }
1751    
1752            /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1753            as NOMATCH. */
1754    
1755          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1756            {            {
1757            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
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(rrc);            RRETURN(rrc);
1761            }            }
1762    
1763          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1764          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1765          }          }
1766        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1128  for (;;) Line 1768  for (;;)
1768        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1769        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1770        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1771          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1772        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1773        }        }
     /* 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. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
1774    
1775      if (*ecode == OP_KETRMIN)      RECURSION_MATCHED:
1776        {      break;
       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 */  
1777    
1778      /* 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
1779      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1221  for (;;) Line 1789  for (;;)
1789      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1790    
1791      case OP_BRAZERO:      case OP_BRAZERO:
1792        {      next = ecode + 1;
1793        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1794        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1795        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1796        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1797      break;      break;
1798    
1799      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1800        {      next = ecode + 1;
1801        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1802        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1803        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1804        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1805      break;      break;
1806    
1807      case OP_SKIPZERO:      case OP_SKIPZERO:
1808        {      next = ecode+1;
1809        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1810        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1811      break;      break;
1812    
1813        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1814        here; just jump to the group, with allow_zero set TRUE. */
1815    
1816        case OP_BRAPOSZERO:
1817        op = *(++ecode);
1818        allow_zero = TRUE;
1819        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1820          goto POSSESSIVE_NON_CAPTURE;
1821    
1822      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1823    
1824      case OP_KET:      case OP_KET:
1825      case OP_KETRMIN:      case OP_KETRMIN:
1826      case OP_KETRMAX:      case OP_KETRMAX:
1827        case OP_KETRPOS:
1828      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1829    
1830      /* 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
1831      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1832      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1833    
1834      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1835        {        {
1836        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1837        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1838        }        }
1839      else saved_eptr = NULL;      else saved_eptr = NULL;
1840    
1841      /* 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
1842      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
1843      assertions. Do this also for the "once" (atomic) groups. */      water mark for use by positive assertions. We also need to record the match
1844        start in case it was changed by \K. */
1845      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||  
1846          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1847          *prev == OP_ONCE)           *prev == OP_ONCE_NC)
1848        {        {
1849        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1850        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1851        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1852          RRETURN(MATCH_MATCH);         /* Sets md->mark */
1853        }        }
1854    
1855      /* 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
1856      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1857      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
1858      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
1859      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
1860        the current subject position and start match pointer and give a MATCH
1861        return. */
1862    
1863      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1864            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1865        {        {
1866        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1867        offset = number << 1;        offset = number << 1;
1868    
1869  #ifdef DEBUG  #ifdef PCRE_DEBUG
1870        printf("end bracket %d", number);        printf("end bracket %d", number);
1871        printf("\n");        printf("\n");
1872  #endif  #endif
1873    
1874          /* Handle a recursively called group. */
1875    
1876          if (md->recursive != NULL && md->recursive->group_num == number)
1877            {
1878            md->end_match_ptr = eptr;
1879            md->start_match_ptr = mstart;
1880            RRETURN(MATCH_MATCH);
1881            }
1882    
1883          /* Deal with capturing */
1884    
1885        md->capture_last = number;        md->capture_last = number;
1886        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1887          {          {
1888            /* If offset is greater than offset_top, it means that we are
1889            "skipping" a capturing group, and that group's offsets must be marked
1890            unset. In earlier versions of PCRE, all the offsets were unset at the
1891            start of matching, but this doesn't work because atomic groups and
1892            assertions can cause a value to be set that should later be unset.
1893            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1894            part of the atomic group, but this is not on the final matching path,
1895            so must be unset when 2 is set. (If there is no group 2, there is no
1896            problem, because offset_top will then be 2, indicating no capture.) */
1897    
1898            if (offset > offset_top)
1899              {
1900              register int *iptr = md->offset_vector + offset_top;
1901              register int *iend = md->offset_vector + offset;
1902              while (iptr < iend) *iptr++ = -1;
1903              }
1904    
1905            /* Now make the extraction */
1906    
1907          md->offset_vector[offset] =          md->offset_vector[offset] =
1908            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1909          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1910          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1911          }          }
1912          }
1913    
1914        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1915        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1916        group. This is the forcible breaking of infinite loops as implemented in
1917        Perl 5.005. For a non-repeating atomic group that includes captures,
1918        establish a backup point by processing the rest of the pattern at a lower
1919        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1920        original OP_ONCE level, thereby bypassing intermediate backup points, but
1921        resetting any captures that happened along the way. */
1922    
1923        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1924          {
1925          if (*prev == OP_ONCE)
1926          {          {
1927          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1928          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1929          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1930          mstart = rec->save_start;          RRETURN(MATCH_ONCE);
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1931          }          }
1932          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1933          break;
1934        }        }
1935    
1936      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1937      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1938        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. */  
1939    
1940      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRPOS)
1941        {        {
1942        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1943        break;        md->end_offset_top = offset_top;
1944          RRETURN(MATCH_KETRPOS);
1945        }        }
1946    
1947      /* 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
1948      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
1949      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
1950      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
1951        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1952    
1953      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1954        {        {
1955        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1956        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1957        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1958          {          {
1959          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1960            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1961            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1962            RRETURN(MATCH_ONCE);
1963            }
1964          if (*prev >= OP_SBRA)    /* Could match an empty string */
1965            {
1966            md->match_function_type = MATCH_CBEGROUP;
1967            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1968          RRETURN(rrc);          RRETURN(rrc);
1969          }          }
1970        ecode = prev;        ecode = prev;
# Line 1360  for (;;) Line 1972  for (;;)
1972        }        }
1973      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1974        {        {
1975        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1976          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1977          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1978        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1979          if (*prev == OP_ONCE)
1980            {
1981            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1982            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1983            md->once_target = prev;
1984            RRETURN(MATCH_ONCE);
1985            }
1986        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1987        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1988        }        }
1989      /* Control never gets here */      /* Control never gets here */
1990    
1991      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1992    
1993      case OP_CIRC:      case OP_CIRC:
1994      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 */  
1995    
1996      /* Start of subject assertion */      /* Start of subject assertion */
1997    
# Line 1389  for (;;) Line 2000  for (;;)
2000      ecode++;      ecode++;
2001      break;      break;
2002    
2003        /* Multiline mode: start of subject unless notbol, or after any newline. */
2004    
2005        case OP_CIRCM:
2006        if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2007        if (eptr != md->start_subject &&
2008            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
2009          RRETURN(MATCH_NOMATCH);
2010        ecode++;
2011        break;
2012    
2013      /* Start of match assertion */      /* Start of match assertion */
2014    
2015      case OP_SOM:      case OP_SOM:
# Line 1403  for (;;) Line 2024  for (;;)
2024      ecode++;      ecode++;
2025      break;      break;
2026    
2027      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2028      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2029    
2030      case OP_DOLL:      case OP_DOLLM:
2031      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2032        {        { 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;  
       }  
2033      else      else
2034        {        {
2035        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2036        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
2037        }        }
2038        ecode++;
2039        break;
2040    
2041        /* Not multiline mode: assert before a terminating newline or before end of
2042        subject unless noteol is set. */
2043    
2044        case OP_DOLL:
2045        if (md->noteol) RRETURN(MATCH_NOMATCH);
2046        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2047    
2048      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2049    
2050      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2051    
2052      case OP_EOD:      case OP_EOD:
2053      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2054        SCHECK_PARTIAL();
2055      ecode++;      ecode++;
2056      break;      break;
2057    
2058      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
2059    
2060      case OP_EODN:      case OP_EODN:
2061      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
2062        if (eptr < md->end_subject &&
2063          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2064        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2065    
2066        /* Either at end of string or \n before end. */
2067    
2068        SCHECK_PARTIAL();
2069      ecode++;      ecode++;
2070      break;      break;
2071    
# Line 1454  for (;;) Line 2077  for (;;)
2077    
2078        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
2079        It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to        It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to
2080        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
2081          partial matching. */
2082    
2083  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2084        if (utf8)        if (utf)
2085          {          {
2086            /* Get status of previous character */
2087    
2088          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2089            {            {
2090            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2091            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2092              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2093            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2094    #ifdef SUPPORT_UCP
2095              if (md->use_ucp)
2096                {
2097                if (c == '_') prev_is_word = TRUE; else
2098                  {
2099                  int cat = UCD_CATEGORY(c);
2100                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2101                  }
2102                }
2103              else
2104    #endif
2105            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2106            }            }
2107          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
2108            /* Get status of next character */
2109    
2110            if (eptr >= md->end_subject)
2111              {
2112              SCHECK_PARTIAL();
2113              cur_is_word = FALSE;
2114              }
2115            else
2116            {            {
2117            GETCHAR(c, eptr);            GETCHAR(c, eptr);
2118    #ifdef SUPPORT_UCP
2119              if (md->use_ucp)
2120                {
2121                if (c == '_') cur_is_word = TRUE; else
2122                  {
2123                  int cat = UCD_CATEGORY(c);
2124                  cur_is_word = (cat == ucp_L || cat == ucp_N);
2125                  }
2126                }
2127              else
2128    #endif
2129            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2130            }            }
2131          }          }
2132        else        else
2133  #endif  #endif
2134    
2135        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
2136          consistency with the behaviour of \w we do use it in this case. */
2137    
2138          {          {
2139          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
2140            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
2141          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
2142            ((md->ctypes[*eptr] & ctype_word) != 0);            {
2143              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
2144    #ifdef SUPPORT_UCP
2145              if (md->use_ucp)
2146                {
2147                c = eptr[-1];
2148                if (c == '_') prev_is_word = TRUE; else
2149                  {
2150                  int cat = UCD_CATEGORY(c);
2151                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2152                  }
2153                }
2154              else
2155    #endif
2156              prev_is_word = MAX_255(eptr[-1])
2157                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2158              }
2159    
2160            /* Get status of next character */
2161    
2162            if (eptr >= md->end_subject)
2163              {
2164              SCHECK_PARTIAL();
2165              cur_is_word = FALSE;
2166              }
2167            else
2168    #ifdef SUPPORT_UCP
2169            if (md->use_ucp)
2170              {
2171              c = *eptr;
2172              if (c == '_') cur_is_word = TRUE; else
2173                {
2174                int cat = UCD_CATEGORY(c);
2175                cur_is_word = (cat == ucp_L || cat == ucp_N);
2176                }
2177              }
2178            else
2179    #endif
2180            cur_is_word = MAX_255(*eptr)
2181              && ((md->ctypes[*eptr] & ctype_word) != 0);
2182          }          }
2183    
2184        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
# Line 1499  for (;;) Line 2196  for (;;)
2196      /* Fall through */      /* Fall through */
2197    
2198      case OP_ALLANY:      case OP_ALLANY:
2199      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2200      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;        {                            /* not be updated before SCHECK_PARTIAL. */
2201          SCHECK_PARTIAL();
2202          RRETURN(MATCH_NOMATCH);
2203          }
2204        eptr++;
2205    #ifdef SUPPORT_UTF
2206        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2207    #endif
2208      ecode++;      ecode++;
2209      break;      break;
2210    
# Line 1508  for (;;) Line 2212  for (;;)
2212      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2213    
2214      case OP_ANYBYTE:      case OP_ANYBYTE:
2215      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2216          {                            /* not be updated before SCHECK_PARTIAL. */
2217          SCHECK_PARTIAL();
2218          RRETURN(MATCH_NOMATCH);
2219          }
2220        eptr++;
2221      ecode++;      ecode++;
2222      break;      break;
2223    
2224      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2225      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2226          {
2227          SCHECK_PARTIAL();
2228          RRETURN(MATCH_NOMATCH);
2229          }
2230      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2231      if (      if (
2232  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2233         c < 256 &&         c < 256 &&
2234  #endif  #endif
2235         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
# Line 1526  for (;;) Line 2239  for (;;)
2239      break;      break;
2240    
2241      case OP_DIGIT:      case OP_DIGIT:
2242      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2243          {
2244          SCHECK_PARTIAL();
2245          RRETURN(MATCH_NOMATCH);
2246          }
2247      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2248      if (      if (
2249  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2250         c >= 256 ||         c > 255 ||
2251  #endif  #endif
2252         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2253         )         )
# Line 1539  for (;;) Line 2256  for (;;)
2256      break;      break;
2257    
2258      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2259      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2260          {
2261          SCHECK_PARTIAL();
2262          RRETURN(MATCH_NOMATCH);
2263          }
2264      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2265      if (      if (
2266  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2267         c < 256 &&         c < 256 &&
2268  #endif  #endif
2269         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
# Line 1552  for (;;) Line 2273  for (;;)
2273      break;      break;
2274    
2275      case OP_WHITESPACE:      case OP_WHITESPACE:
2276      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2277          {
2278          SCHECK_PARTIAL();
2279          RRETURN(MATCH_NOMATCH);
2280          }
2281      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2282      if (      if (
2283  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2284         c >= 256 ||         c > 255 ||
2285  #endif  #endif
2286         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2287         )         )
# Line 1565  for (;;) Line 2290  for (;;)
2290      break;      break;
2291    
2292      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2293      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2294          {
2295          SCHECK_PARTIAL();
2296          RRETURN(MATCH_NOMATCH);
2297          }
2298      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2299      if (      if (
2300  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2301         c < 256 &&         c < 256 &&
2302  #endif  #endif
2303         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
# Line 1578  for (;;) Line 2307  for (;;)
2307      break;      break;
2308    
2309      case OP_WORDCHAR:      case OP_WORDCHAR:
2310      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2311          {
2312          SCHECK_PARTIAL();
2313          RRETURN(MATCH_NOMATCH);
2314          }
2315      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2316      if (      if (
2317  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2318         c >= 256 ||         c > 255 ||
2319  #endif  #endif
2320         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2321         )         )
# Line 1591  for (;;) Line 2324  for (;;)
2324      break;      break;
2325    
2326      case OP_ANYNL:      case OP_ANYNL:
2327      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2328          {
2329          SCHECK_PARTIAL();
2330          RRETURN(MATCH_NOMATCH);
2331          }
2332      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2333      switch(c)      switch(c)
2334        {        {
2335        default: RRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2336    
2337        case 0x000d:        case 0x000d:
2338        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2339        break;        break;
# Line 1615  for (;;) Line 2353  for (;;)
2353      break;      break;
2354    
2355      case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
2356      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2357          {
2358          SCHECK_PARTIAL();
2359          RRETURN(MATCH_NOMATCH);
2360          }
2361      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2362      switch(c)      switch(c)
2363        {        {
# Line 1645  for (;;) Line 2387  for (;;)
2387      break;      break;
2388    
2389      case OP_HSPACE:      case OP_HSPACE:
2390      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2391          {
2392          SCHECK_PARTIAL();
2393          RRETURN(MATCH_NOMATCH);
2394          }
2395      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2396      switch(c)      switch(c)
2397        {        {
# Line 1675  for (;;) Line 2421  for (;;)
2421      break;      break;
2422    
2423      case OP_NOT_VSPACE:      case OP_NOT_VSPACE:
2424      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2425          {
2426          SCHECK_PARTIAL();
2427          RRETURN(MATCH_NOMATCH);
2428          }
2429      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2430      switch(c)      switch(c)
2431        {        {
# Line 1693  for (;;) Line 2443  for (;;)
2443      break;      break;
2444    
2445      case OP_VSPACE:      case OP_VSPACE:
2446      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2447          {
2448          SCHECK_PARTIAL();
2449          RRETURN(MATCH_NOMATCH);
2450          }
2451      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2452      switch(c)      switch(c)
2453        {        {
# Line 1716  for (;;) Line 2470  for (;;)
2470    
2471      case OP_PROP:      case OP_PROP:
2472      case OP_NOTPROP:      case OP_NOTPROP:
2473      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2474          {
2475          SCHECK_PARTIAL();
2476          RRETURN(MATCH_NOMATCH);
2477          }
2478      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2479        {        {
2480        const ucd_record *prop = GET_UCD(c);        const ucd_record *prop = GET_UCD(c);
# Line 1732  for (;;) Line 2490  for (;;)
2490               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2491               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2492            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2493           break;          break;
2494    
2495          case PT_GC:          case PT_GC:
2496          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2497            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2498          break;          break;
2499    
# Line 1749  for (;;) Line 2507  for (;;)
2507            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2508          break;          break;
2509    
2510            /* These are specials */
2511    
2512            case PT_ALNUM:
2513            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2514                 PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2515              RRETURN(MATCH_NOMATCH);
2516            break;
2517    
2518            case PT_SPACE:    /* Perl space */
2519            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2520                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2521                   == (op == OP_NOTPROP))
2522              RRETURN(MATCH_NOMATCH);
2523            break;
2524    
2525            case PT_PXSPACE:  /* POSIX space */
2526            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2527                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2528                 c == CHAR_FF || c == CHAR_CR)
2529                   == (op == OP_NOTPROP))
2530              RRETURN(MATCH_NOMATCH);
2531            break;
2532    
2533            case PT_WORD:
2534            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2535                 PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2536                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2537              RRETURN(MATCH_NOMATCH);
2538            break;
2539    
2540            /* This should never occur */
2541    
2542          default:          default:
2543          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2544          }          }
# Line 1761  for (;;) Line 2551  for (;;)
2551      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2552    
2553      case OP_EXTUNI:      case OP_EXTUNI:
2554      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2555          {
2556          SCHECK_PARTIAL();
2557          RRETURN(MATCH_NOMATCH);
2558          }
2559      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2560        if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH);
2561        while (eptr < md->end_subject)
2562        {        {
2563        int category = UCD_CATEGORY(c);        int len = 1;
2564        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2565        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2566          {        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;  
         }  
2567        }        }
2568      ecode++;      ecode++;
2569      break;      break;
# Line 1792  for (;;) Line 2579  for (;;)
2579      loops). */      loops). */
2580    
2581      case OP_REF:      case OP_REF:
2582        {      case OP_REFI:
2583        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2584        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2585        ecode += 1 + IMM2_SIZE;
2586    
2587        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2588    
2589        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2590        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
2591        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.  
2592    
2593        (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
2594        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2595    
2596        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
2597        referenced subpattern. */      referenced subpattern. */
2598    
2599        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2600          length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;        length = (md->jscript_compat)? 0 : -1;
2601        else      else
2602          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2603    
2604        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2605    
2606        switch (*ecode)      switch (*ecode)
2607          {        {
2608          case OP_CRSTAR:        case OP_CRSTAR:
2609          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2610          case OP_CRPLUS:        case OP_CRPLUS:
2611          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2612          case OP_CRQUERY:        case OP_CRQUERY:
2613          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2614          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2615          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2616          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2617          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2618          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2619          break;        break;
2620    
2621          case OP_CRRANGE:        case OP_CRRANGE:
2622          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2623          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2624          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2625          max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2626          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2627          ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2628          break;        break;
2629    
2630          default:               /* No repeat follows */        default:               /* No repeat follows */
2631          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2632          eptr += length;          {
2633          continue;              /* With the main loop */          CHECK_PARTIAL();
2634            RRETURN(MATCH_NOMATCH);
2635          }          }
2636          eptr += length;
2637          continue;              /* With the main loop */
2638          }
2639    
2640        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2641        main loop. */      zero, just continue with the main loop. If the length is negative, it
2642        means the reference is unset in non-Java-compatible mode. If the minimum is
2643        zero, we can continue at the same level without recursion. For any other
2644        minimum, carrying on will result in NOMATCH. */
2645    
2646        if (length == 0) continue;      if (length == 0) continue;
2647        if (length < 0 && min == 0) continue;
2648    
2649        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2650        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2651        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2652    
2653        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2654          {
2655          int slength;
2656          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2657          {          {
2658          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2659          eptr += length;          RRETURN(MATCH_NOMATCH);
2660          }          }
2661          eptr += slength;
2662          }
2663    
2664        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2665        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2666    
2667        if (min == max) continue;      if (min == max) continue;
2668    
2669        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2670    
2671        if (minimize)      if (minimize)
2672          {
2673          for (fi = min;; fi++)
2674          {          {
2675          for (fi = min;; fi++)          int slength;
2676            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2677            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2678            if (fi >= max) RRETURN(MATCH_NOMATCH);
2679            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2680            {            {
2681            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2682            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            RRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2683            }            }
2684          /* Control never gets here */          eptr += slength;
2685          }          }
2686          /* Control never gets here */
2687          }
2688    
2689        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2690    
2691        else      else
2692          {
2693          pp = eptr;
2694          for (i = min; i < max; i++)
2695          {          {
2696          pp = eptr;          int slength;
2697          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)  
2698            {            {
2699            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2700            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2701            }            }
2702          RRETURN(MATCH_NOMATCH);          eptr += slength;
2703          }          }
2704          while (eptr >= pp)
2705            {
2706            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2707            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2708            eptr -= length;
2709            }
2710          RRETURN(MATCH_NOMATCH);
2711        }        }
2712      /* Control never gets here */      /* Control never gets here */
2713    
   
   
2714      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2715      used when all the characters in the class have values in the range 0-255,      used when all the characters in the class have values in the range 0-255,
2716      and either the matching is caseful, or the characters are in the range      and either the matching is caseful, or the characters are in the range
# Line 1918  for (;;) Line 2725  for (;;)
2725      case OP_NCLASS:      case OP_NCLASS:
2726      case OP_CLASS:      case OP_CLASS:
2727        {        {
2728          /* The data variable is saved across frames, so the byte map needs to
2729          be stored there. */
2730    #define BYTE_MAP ((pcre_uint8 *)data)
2731        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2732        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2733    
2734        switch (*ecode)        switch (*ecode)
2735          {          {
# Line 1940  for (;;) Line 2750  for (;;)
2750          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2751          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2752          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2753          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2754          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2755          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2756          break;          break;
2757    
2758          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 1952  for (;;) Line 2762  for (;;)
2762    
2763        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2764    
2765  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2766        /* UTF-8 mode */        if (utf)
       if (utf8)  
2767          {          {
2768          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2769            {            {
2770            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2771                {
2772                SCHECK_PARTIAL();
2773                RRETURN(MATCH_NOMATCH);
2774                }
2775            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2776            if (c > 255)            if (c > 255)
2777              {              {
2778              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2779              }              }
2780            else            else
2781              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  
             }  
2782            }            }
2783          }          }
2784        else        else
2785  #endif  #endif
2786        /* Not UTF-8 mode */        /* Not UTF mode */
2787          {          {
2788          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2789            {            {
2790            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2791                {
2792                SCHECK_PARTIAL();
2793                RRETURN(MATCH_NOMATCH);
2794                }
2795            c = *eptr++;            c = *eptr++;
2796            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2797              if (c > 255)
2798                {
2799                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2800                }
2801              else
2802    #endif
2803                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2804            }            }
2805          }          }
2806    
# Line 1992  for (;;) Line 2814  for (;;)
2814    
2815        if (minimize)        if (minimize)
2816          {          {
2817  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2818          /* UTF-8 mode */          if (utf)
         if (utf8)  
2819            {            {
2820            for (fi = min;; fi++)            for (fi = min;; fi++)
2821              {              {
2822              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2823              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2824              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2825                if (eptr >= md->end_subject)
2826                  {
2827                  SCHECK_PARTIAL();
2828                  RRETURN(MATCH_NOMATCH);
2829                  }
2830              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2831              if (c > 255)              if (c > 255)
2832                {                {
2833                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2834                }                }
2835              else              else
2836                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  
               }  
2837              }              }
2838            }            }
2839          else          else
2840  #endif  #endif
2841          /* Not UTF-8 mode */          /* Not UTF mode */
2842            {            {
2843            for (fi = min;; fi++)            for (fi = min;; fi++)
2844              {              {
2845              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2846              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2847              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2848                if (eptr >= md->end_subject)
2849                  {
2850                  SCHECK_PARTIAL();
2851                  RRETURN(MATCH_NOMATCH);
2852                  }
2853              c = *eptr++;              c = *eptr++;
2854              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2855                if (c > 255)
2856                  {
2857                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2858                  }
2859                else
2860    #endif
2861                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2862              }              }
2863            }            }
2864          /* Control never gets here */          /* Control never gets here */
# Line 2034  for (;;) Line 2870  for (;;)
2870          {          {
2871          pp = eptr;          pp = eptr;
2872    
2873  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2874          /* UTF-8 mode */          if (utf)
         if (utf8)  
2875            {            {
2876            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2877              {              {
2878              int len = 1;              int len = 1;
2879              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2880                  {
2881                  SCHECK_PARTIAL();
2882                  break;
2883                  }
2884              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2885              if (c > 255)              if (c > 255)
2886                {                {
2887                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
2888                }                }
2889              else              else
2890                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
2891              eptr += len;              eptr += len;
2892              }              }
2893            for (;;)            for (;;)
2894              {              {
2895              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2896              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2897              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2898              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2063  for (;;) Line 2900  for (;;)
2900            }            }
2901          else          else
2902  #endif  #endif
2903            /* Not UTF-8 mode */            /* Not UTF mode */
2904            {            {
2905            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2906              {              {
2907              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2908                  {
2909                  SCHECK_PARTIAL();
2910                  break;
2911                  }
2912              c = *eptr;              c = *eptr;
2913              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
2914                if (c > 255)
2915                  {
2916                  if (op == OP_CLASS) break;
2917                  }
2918                else
2919    #endif
2920                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
2921              eptr++;              eptr++;
2922              }              }
2923            while (eptr >= pp)            while (eptr >= pp)
2924              {              {
2925              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2926              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2927              eptr--;              eptr--;
2928              }              }
# Line 2082  for (;;) Line 2930  for (;;)
2930    
2931          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2932          }          }
2933    #undef BYTE_MAP
2934        }        }
2935      /* Control never gets here */      /* Control never gets here */
2936    
# Line 2090  for (;;) Line 2939  for (;;)
2939      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
2940      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
2941    
2942  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2943      case OP_XCLASS:      case OP_XCLASS:
2944        {        {
2945        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2115  for (;;) Line 2964  for (;;)
2964          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2965          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2966          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2967          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2968          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2969          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2970          break;          break;
2971    
2972          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2129  for (;;) Line 2978  for (;;)
2978    
2979        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2980          {          {
2981          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2982              {
2983              SCHECK_PARTIAL();
2984              RRETURN(MATCH_NOMATCH);
2985              }
2986          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
2987          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
2988          }          }
2989    
2990        /* 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 2146  for (;;) Line 2999  for (;;)
2999          {          {
3000          for (fi = min;; fi++)          for (fi = min;; fi++)
3001            {            {
3002            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
3003            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3004            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3005              if (eptr >= md->end_subject)
3006                {
3007                SCHECK_PARTIAL();
3008                RRETURN(MATCH_NOMATCH);
3009                }
3010            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3011            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3012            }            }
3013          /* Control never gets here */          /* Control never gets here */
3014          }          }
# Line 2163  for (;;) Line 3021  for (;;)
3021          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3022            {            {
3023            int len = 1;            int len = 1;
3024            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
3025                {
3026                SCHECK_PARTIAL();
3027                break;
3028                }
3029    #ifdef SUPPORT_UTF
3030            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3031            if (!_pcre_xclass(c, data)) break;  #else
3032              c = *eptr;
3033    #endif
3034              if (!PRIV(xclass)(c, data, utf)) break;
3035            eptr += len;            eptr += len;
3036            }            }
3037          for(;;)          for(;;)
3038            {            {
3039            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3040            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3041            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3042            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3043              if (utf) BACKCHAR(eptr);
3044    #endif
3045            }            }
3046          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3047          }          }
# Line 2185  for (;;) Line 3053  for (;;)
3053      /* Match a single character, casefully */      /* Match a single character, casefully */
3054    
3055      case OP_CHAR:      case OP_CHAR:
3056  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3057      if (utf8)      if (utf)
3058        {        {
3059        length = 1;        length = 1;
3060        ecode++;        ecode++;
3061        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3062        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
3063            {
3064            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
3065            RRETURN(MATCH_NOMATCH);
3066            }
3067        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);
3068        }        }
3069      else      else
3070  #endif  #endif
3071        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3072        {        {
3073        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
3074            {
3075            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
3076            RRETURN(MATCH_NOMATCH);
3077            }
3078        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
3079        ecode += 2;        ecode += 2;
3080        }        }
3081      break;      break;
3082    
3083      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3084        subject, give up immediately. */
3085    
3086        case OP_CHARI:
3087        if (eptr >= md->end_subject)
3088          {
3089          SCHECK_PARTIAL();
3090          RRETURN(MATCH_NOMATCH);
3091          }
3092    
3093      case OP_CHARNC:  #ifdef SUPPORT_UTF
3094  #ifdef SUPPORT_UTF8      if (utf)
     if (utf8)  
3095        {        {
3096        length = 1;        length = 1;
3097        ecode++;        ecode++;
3098        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3099    
       if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
   
3100        /* 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
3101        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3102          fast lookup table. We know that there is at least one byte left in the
3103          subject. */
3104    
3105        if (fc < 128)        if (fc < 128)
3106          {          {
3107          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[fc]
3108                != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3109            ecode++;
3110            eptr++;
3111          }          }
3112    
3113        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3114          use the value of "length" to check for sufficient bytes left, because the
3115          other case of the character may have more or fewer bytes.  */
3116    
3117        else        else
3118          {          {
# Line 2246  for (;;) Line 3133  for (;;)
3133          }          }
3134        }        }
3135      else      else
3136  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3137    
3138      /* Non-UTF-8 mode */      /* Not UTF mode */
3139        {        {
3140        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3141        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3142          eptr++;
3143        ecode += 2;        ecode += 2;
3144        }        }
3145      break;      break;
# Line 2259  for (;;) Line 3147  for (;;)
3147      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3148    
3149      case OP_EXACT:      case OP_EXACT:
3150        case OP_EXACTI:
3151      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3152      ecode += 3;      ecode += 1 + IMM2_SIZE;
3153      goto REPEATCHAR;      goto REPEATCHAR;
3154    
3155      case OP_POSUPTO:      case OP_POSUPTO:
3156        case OP_POSUPTOI:
3157      possessive = TRUE;      possessive = TRUE;
3158      /* Fall through */      /* Fall through */
3159    
3160      case OP_UPTO:      case OP_UPTO:
3161        case OP_UPTOI:
3162      case OP_MINUPTO:      case OP_MINUPTO:
3163        case OP_MINUPTOI:
3164      min = 0;      min = 0;
3165      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3166      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3167      ecode += 3;      ecode += 1 + IMM2_SIZE;
3168      goto REPEATCHAR;      goto REPEATCHAR;
3169    
3170      case OP_POSSTAR:      case OP_POSSTAR:
3171        case OP_POSSTARI:
3172      possessive = TRUE;      possessive = TRUE;
3173      min = 0;      min = 0;
3174      max = INT_MAX;      max = INT_MAX;
# Line 2283  for (;;) Line 3176  for (;;)
3176      goto REPEATCHAR;      goto REPEATCHAR;
3177    
3178      case OP_POSPLUS:      case OP_POSPLUS:
3179        case OP_POSPLUSI:
3180      possessive = TRUE;      possessive = TRUE;
3181      min = 1;      min = 1;
3182      max = INT_MAX;      max = INT_MAX;
# Line 2290  for (;;) Line 3184  for (;;)
3184      goto REPEATCHAR;      goto REPEATCHAR;
3185    
3186      case OP_POSQUERY:      case OP_POSQUERY:
3187        case OP_POSQUERYI:
3188      possessive = TRUE;      possessive = TRUE;
3189      min = 0;      min = 0;
3190      max = 1;      max = 1;
# Line 2297  for (;;) Line 3192  for (;;)
3192      goto REPEATCHAR;      goto REPEATCHAR;
3193    
3194      case OP_STAR:      case OP_STAR:
3195        case OP_STARI:
3196      case OP_MINSTAR:      case OP_MINSTAR:
3197        case OP_MINSTARI:
3198      case OP_PLUS:      case OP_PLUS:
3199        case OP_PLUSI:
3200      case OP_MINPLUS:      case OP_MINPLUS:
3201        case OP_MINPLUSI:
3202      case OP_QUERY:      case OP_QUERY:
3203        case OP_QUERYI:
3204      case OP_MINQUERY:      case OP_MINQUERY:
3205      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3206        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3207      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3208      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3209      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3210      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3211    
3212      /* Common code for all repeated single-character matches. We can give      /* Common code for all repeated single-character matches. */
     up quickly if there are fewer than the minimum number of characters left in  
     the subject. */  
3213    
3214      REPEATCHAR:      REPEATCHAR:
3215  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3216      if (utf8)      if (utf)
3217        {        {
3218        length = 1;        length = 1;
3219        charptr = ecode;        charptr = ecode;
3220        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3221        ecode += length;        ecode += length;
3222    
3223        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 2329  for (;;) Line 3227  for (;;)
3227          {          {
3228  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3229          unsigned int othercase;          unsigned int othercase;
3230          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3231              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3232            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3233          else oclength = 0;          else oclength = 0;
3234  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3235    
3236          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3237            {            {
3238            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3239                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3240  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3241            /* Need braces because of following else */            else if (oclength > 0 &&
3242            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                     eptr <= md->end_subject - oclength &&
3243                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3244    #endif  /* SUPPORT_UCP */
3245            else            else
3246              {              {
3247              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3248              eptr += oclength;              RRETURN(MATCH_NOMATCH);
3249              }              }
 #else   /* without SUPPORT_UCP */  
           else { RRETURN(MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3250            }            }
3251    
3252          if (min == max) continue;          if (min == max) continue;
# Line 2357  for (;;) Line 3255  for (;;)
3255            {            {
3256            for (fi = min;; fi++)            for (fi = min;; fi++)
3257              {              {
3258              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3259              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3260              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3261              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3262                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3263  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3264              /* Need braces because of following else */              else if (oclength > 0 &&
3265              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                       eptr <= md->end_subject - oclength &&
3266                         memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3267    #endif  /* SUPPORT_UCP */
3268              else              else
3269                {                {
3270                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3271                eptr += oclength;                RRETURN(MATCH_NOMATCH);
3272                }                }
 #else   /* without SUPPORT_UCP */  
             else { RRETURN (MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3273              }              }
3274            /* Control never gets here */            /* Control never gets here */
3275            }            }
# Line 2381  for (;;) Line 3279  for (;;)
3279            pp = eptr;            pp = eptr;
3280            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3281              {              {
3282              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3283              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3284  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3285              else if (oclength == 0) break;              else if (oclength > 0 &&
3286                         eptr <= md->end_subject - oclength &&
3287                         memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3288    #endif  /* SUPPORT_UCP */
3289              else              else
3290                {                {
3291                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3292                eptr += oclength;                break;
3293                }                }
 #else   /* without SUPPORT_UCP */  
             else break;  
 #endif  /* SUPPORT_UCP */  
3294              }              }
3295    
3296            if (possessive) continue;            if (possessive) continue;
3297    
3298            for(;;)            for(;;)
3299             {              {
3300             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3301             if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3302             if (eptr == pp) RRETURN(MATCH_NOMATCH);              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3303  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3304             eptr--;              eptr--;
3305             BACKCHAR(eptr);              BACKCHAR(eptr);
3306  #else   /* without SUPPORT_UCP */  #else   /* without SUPPORT_UCP */
3307             eptr -= length;              eptr -= length;
3308  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3309             }              }
3310            }            }
3311          /* Control never gets here */          /* Control never gets here */
3312          }          }
# Line 2417  for (;;) Line 3316  for (;;)
3316        value of fc will always be < 128. */        value of fc will always be < 128. */
3317        }        }
3318      else      else
3319  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3320          /* When not in UTF-8 mode, load a single-byte character. */
     /* When not in UTF-8 mode, load a single-byte character. */  
       {  
       if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3321        fc = *ecode++;        fc = *ecode++;
       }  
3322    
3323      /* The value of fc at this point is always less than 256, though we may or      /* The value of fc at this point is always one character, though we may
3324      may not be in UTF-8 mode. The code is duplicated for the caseless and      or may not be in UTF mode. The code is duplicated for the caseless and
3325      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3326      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3327      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 2437  for (;;) Line 3332  for (;;)
3332      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3333        max, eptr));        max, eptr));
3334    
3335      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3336        {        {
3337        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3338          /* fc must be < 128 if UTF is enabled. */
3339          foc = md->fcc[fc];
3340    #else
3341    #ifdef SUPPORT_UTF
3342    #ifdef SUPPORT_UCP
3343          if (utf && fc > 127)
3344            foc = UCD_OTHERCASE(fc);
3345    #else
3346          if (utf && fc > 127)
3347            foc = fc;
3348    #endif /* SUPPORT_UCP */
3349          else
3350    #endif /* SUPPORT_UTF */
3351            foc = TABLE_GET(fc, md->fcc, fc);
3352    #endif /* COMPILE_PCRE8 */
3353    
3354        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3355          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3356            if (eptr >= md->end_subject)
3357              {
3358              SCHECK_PARTIAL();
3359              RRETURN(MATCH_NOMATCH);
3360              }
3361            if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3362            eptr++;
3363            }
3364        if (min == max) continue;        if (min == max) continue;
3365        if (minimize)        if (minimize)
3366          {          {
3367          for (fi = min;; fi++)          for (fi = min;; fi++)
3368            {            {
3369            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3370            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3371            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) RRETURN(MATCH_NOMATCH);
3372                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3373                {
3374                SCHECK_PARTIAL();
3375              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3376                }
3377              if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3378              eptr++;
3379            }            }
3380          /* Control never gets here */          /* Control never gets here */
3381          }          }
# Line 2460  for (;;) Line 3384  for (;;)
3384          pp = eptr;          pp = eptr;
3385          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3386            {            {
3387            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3388                {
3389                SCHECK_PARTIAL();
3390                break;
3391                }
3392              if (fc != *eptr && foc != *eptr) break;
3393            eptr++;            eptr++;
3394            }            }
3395    
3396          if (possessive) continue;          if (possessive) continue;
3397    
3398          while (eptr >= pp)          while (eptr >= pp)
3399            {            {
3400            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3401            eptr--;            eptr--;
3402            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3403            }            }
# Line 2479  for (;;) Line 3410  for (;;)
3410    
3411      else      else
3412        {        {
3413        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3414            {
3415            if (eptr >= md->end_subject)
3416              {
3417              SCHECK_PARTIAL();
3418              RRETURN(MATCH_NOMATCH);
3419              }
3420            if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
3421            }
3422    
3423        if (min == max) continue;        if (min == max) continue;
3424    
3425        if (minimize)        if (minimize)
3426          {          {
3427          for (fi = min;; fi++)          for (fi = min;; fi++)
3428            {            {
3429            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3430            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3431            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) RRETURN(MATCH_NOMATCH);
3432              if (eptr >= md->end_subject)
3433                {
3434                SCHECK_PARTIAL();
3435              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3436                }
3437              if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
3438            }            }
3439          /* Control never gets here */          /* Control never gets here */
3440          }          }
# Line 2497  for (;;) Line 3443  for (;;)
3443          pp = eptr;          pp = eptr;
3444          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3445            {            {
3446            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3447                {
3448                SCHECK_PARTIAL();
3449                break;
3450                }
3451              if (fc != *eptr) break;
3452            eptr++;            eptr++;
3453            }            }
3454          if (possessive) continue;          if (possessive) continue;
3455    
3456          while (eptr >= pp)          while (eptr >= pp)
3457            {            {
3458            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3459            eptr--;            eptr--;
3460            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3461            }            }
# Line 2516  for (;;) Line 3468  for (;;)
3468      checking can be multibyte. */      checking can be multibyte. */
3469    
3470      case OP_NOT:      case OP_NOT:
3471      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3472        if (eptr >= md->end_subject)
3473          {
3474          SCHECK_PARTIAL();
3475          RRETURN(MATCH_NOMATCH);
3476          }
3477      ecode++;      ecode++;
3478      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3479      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3480        {        {
3481  #ifdef SUPPORT_UTF8        register int ch, och;
3482        if (c < 256)        ch = *ecode++;
3483  #endif  #ifdef COMPILE_PCRE8
3484        c = md->lcc[c];        /* ch must be < 128 if UTF is enabled. */
3485        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        och = md->fcc[ch];
3486    #else
3487    #ifdef SUPPORT_UTF
3488    #ifdef SUPPORT_UCP
3489          if (utf && ch > 127)
3490            och = UCD_OTHERCASE(ch);
3491    #else
3492          if (utf && ch > 127)
3493            och = ch;
3494    #endif /* SUPPORT_UCP */
3495          else
3496    #endif /* SUPPORT_UTF */
3497            och = TABLE_GET(ch, md->fcc, ch);
3498    #endif /* COMPILE_PCRE8 */
3499          if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3500        }        }
3501      else      else    /* Caseful */
3502        {        {
3503        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);
3504        }        }
# Line 2541  for (;;) Line 3512  for (;;)
3512      about... */      about... */
3513    
3514      case OP_NOTEXACT:      case OP_NOTEXACT:
3515        case OP_NOTEXACTI:
3516      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3517      ecode += 3;      ecode += 1 + IMM2_SIZE;
3518      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3519    
3520      case OP_NOTUPTO:      case OP_NOTUPTO:
3521        case OP_NOTUPTOI:
3522      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3523        case OP_NOTMINUPTOI:
3524      min = 0;      min = 0;
3525      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3526      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3527      ecode += 3;      ecode += 1 + IMM2_SIZE;
3528      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3529    
3530      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3531        case OP_NOTPOSSTARI:
3532      possessive = TRUE;      possessive = TRUE;
3533      min = 0;      min = 0;
3534      max = INT_MAX;      max = INT_MAX;
# Line 2561  for (;;) Line 3536  for (;;)
3536      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3537    
3538      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3539        case OP_NOTPOSPLUSI:
3540      possessive = TRUE;      possessive = TRUE;
3541      min = 1;      min = 1;
3542      max = INT_MAX;      max = INT_MAX;
# Line 2568  for (;;) Line 3544  for (;;)
3544      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3545    
3546      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3547        case OP_NOTPOSQUERYI:
3548      possessive = TRUE;      possessive = TRUE;
3549      min = 0;      min = 0;
3550      max = 1;      max = 1;
# Line 2575  for (;;) Line 3552  for (;;)
3552      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3553    
3554      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3555        case OP_NOTPOSUPTOI:
3556      possessive = TRUE;      possessive = TRUE;
3557      min = 0;      min = 0;
3558      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3559      ecode += 3;      ecode += 1 + IMM2_SIZE;
3560      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3561    
3562      case OP_NOTSTAR:      case OP_NOTSTAR:
3563        case OP_NOTSTARI:
3564      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3565        case OP_NOTMINSTARI:
3566      case OP_NOTPLUS:      case OP_NOTPLUS:
3567        case OP_NOTPLUSI:
3568      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3569        case OP_NOTMINPLUSI:
3570      case OP_NOTQUERY:      case OP_NOTQUERY:
3571        case OP_NOTQUERYI:
3572      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3573      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3574        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3575      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3576      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3577      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3578      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3579    
3580      /* Common code for all repeated single-byte matches. We can give up quickly      /* Common code for all repeated single-byte matches. */
     if there are fewer than the minimum number of bytes left in the  
     subject. */  
3581    
3582      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3583      fc = *ecode++;      fc = *ecode++;
3584    
3585      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
# Line 2612  for (;;) Line 3593  for (;;)
3593      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,
3594        max, eptr));        max, eptr));
3595    
3596      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3597        {        {
3598        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3599          /* fc must be < 128 if UTF is enabled. */
3600          foc = md->fcc[fc];
3601    #else
3602    #ifdef SUPPORT_UTF
3603    #ifdef SUPPORT_UCP
3604          if (utf && fc > 127)
3605            foc = UCD_OTHERCASE(fc);
3606    #else
3607          if (utf && fc > 127)
3608            foc = fc;
3609    #endif /* SUPPORT_UCP */
3610          else
3611    #endif /* SUPPORT_UTF */
3612            foc = TABLE_GET(fc, md->fcc, fc);
3613    #endif /* COMPILE_PCRE8 */
3614    
3615  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3616        /* UTF-8 mode */        if (utf)
       if (utf8)  
3617          {          {
3618          register unsigned int d;          register unsigned int d;
3619          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3620            {            {
3621              if (eptr >= md->end_subject)
3622                {
3623                SCHECK_PARTIAL();
3624                RRETURN(MATCH_NOMATCH);
3625                }
3626            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3627            if (d < 256) d = md->lcc[d];            if (fc == d || foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) RRETURN(MATCH_NOMATCH);  
3628            }            }
3629          }          }
3630        else        else
3631  #endif  #endif
3632          /* Not UTF mode */
       /* Not UTF-8 mode */