/[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 597 by ph10, Mon May 2 17:08:52 2011 UTC revision 893 by ph10, Thu Jan 19 17:15:11 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) 1997-2010 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 73  negative to avoid the external error cod Line 75  negative to avoid the external error cod
75    
76  #define MATCH_ACCEPT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_COMMIT       (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_PRUNE        (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_SKIP         (-996)  #define MATCH_ONCE         (-996)
80  #define MATCH_SKIP_ARG     (-995)  #define MATCH_PRUNE        (-995)
81  #define MATCH_THEN         (-994)  #define MATCH_SKIP         (-994)
82    #define MATCH_SKIP_ARG     (-993)
83  /* This is a convenience macro for code that occurs many times. */  #define MATCH_THEN         (-992)
   
 #define MRRETURN(ra) \  
   { \  
   md->mark = markptr; \  
   RRETURN(ra); \  
   }  
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 117  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 134  while (length-- > 0) Line 130  while (length-- > 0)
130    
131  /* Normally, if a back reference hasn't been set, the length that is passed is  /* Normally, if a back reference hasn't been set, the length that is passed is
132  negative, so the match always fails. However, in JavaScript compatibility mode,  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  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.  subject bytes matched may be different to the number of reference bytes.
135    
136  Arguments:  Arguments:
# Line 142  Arguments: Line 138  Arguments:
138    eptr        pointer into the subject    eptr        pointer into the subject
139    length      length of reference to be matched (number of bytes)    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:      < 0 if not matched, otherwise the number of subject bytes matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
144  */  */
145    
146  static int  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 eptr_start = eptr;  PCRE_PUCHAR eptr_start = eptr;
151  register USPTR p = md->start_subject + md->offset_vector[offset];  register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
152    
153  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
154  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 175  if (length < 0) return -1; Line 171  if (length < 0) return -1;
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      /* Match characters up to the end of the reference. NOTE: the number of      /* Match characters up to the end of the reference. NOTE: the number of
181      bytes matched may differ, because there are some characters whose upper and      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      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);      (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      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      the latter. It is important, therefore, to check the length along the
186      reference, not along the subject (earlier code did this wrong). */      reference, not along the subject (earlier code did this wrong). */
187    
188      USPTR endptr = p + length;      PCRE_PUCHAR endptr = p + length;
189      while (p < endptr)      while (p < endptr)
190        {        {
191        int c, d;        int c, d;
# Line 206  if ((ims & PCRE_CASELESS) != 0) Line 202  if ((ims & PCRE_CASELESS) != 0)
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      if (eptr + length > md->end_subject) return -1;      if (eptr + length > md->end_subject) return -1;
206      while (length-- > 0)      while (length-- > 0)
207        { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }        {
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    {    {
220    if (eptr + length > md->end_subject) return -1;    if (eptr + length > md->end_subject) return -1;
221    while (length-- > 0) if (*p++ != *eptr++) return -1;    while (length-- > 0) if (*p++ != *eptr++) return -1;
222    }    }
223    
224  return eptr - eptr_start;  return (int)(eptr - eptr_start);
225  }  }
226    
227    
# Line 273  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 273  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
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, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
276         RM61,  RM62 };         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
# Line 283  actually used in this definition. */ Line 283  actually used in this definition. */
283  #define REGISTER register  #define REGISTER register
284    
285  #ifdef PCRE_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,markptr,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 295  actually used in this definition. */ Line 295  actually 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,markptr,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 309  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 = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\
315    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    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;\
   newframe->Xmarkptr = markptr;\  
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 335  argument of match(), which never changes Line 332  argument of match(), which never changes
332    {\    {\
333    heapframe *oldframe = frame;\    heapframe *oldframe = frame;\
334    frame = oldframe->Xprevframe;\    frame = oldframe->Xprevframe;\
335    (pcre_stack_free)(oldframe);\    if (oldframe != &frame_zero) (PUBL(stack_free))(oldframe);\
336    if (frame != NULL)\    if (frame != NULL)\
337      {\      {\
338      rrc = ra;\      rrc = ra;\
# Line 352  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;
   USPTR Xmarkptr;  
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 380  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 432  returns a negative (error) response, the Line 421  returns a negative (error) response, the
421  same response. */  same response. */
422    
423  /* These macros pack up tests that are used for partial matching, and which  /* These macros pack up tests that are used for partial matching, and which
424  appears several times in the code. We set the "hit end" flag if the pointer is  appear several times in the code. We set the "hit end" flag if the pointer is
425  at the end of the subject and also past the start of the subject (i.e.  at the end of the subject and also past the start of the subject (i.e.
426  something has been matched). For hard partial matching, we then return  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  immediately. The second one is used when we already know we are past the end of
# Line 443  the subject. */ Line 432  the subject. */
432        eptr > md->start_used_ptr) \        eptr > md->start_used_ptr) \
433      { \      { \
434      md->hitend = TRUE; \      md->hitend = TRUE; \
435      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
436      }      }
437    
438  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
439    if (md->partial != 0 && eptr > md->start_used_ptr) \    if (md->partial != 0 && eptr > md->start_used_ptr) \
440      { \      { \
441      md->hitend = TRUE; \      md->hitend = TRUE; \
442      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
443      }      }
444    
445    
446  /* Performance note: It might be tempting to extract commonly used fields from  /* Performance note: It might be tempting to extract commonly used fields from
447  the md structure (e.g. utf8, end_subject) into individual variables to improve  the md structure (e.g. utf, end_subject) into individual variables to improve
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 464  Arguments: Line 453  Arguments:
453     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
454     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
455                   by encountering \K)                   by encountering \K)
    markptr     pointer to the most recent MARK name, or NULL  
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
# Line 484  Returns:       MATCH_MATCH if matched Line 467  Returns:       MATCH_MATCH if matched
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    const uschar *markptr, int offset_top, match_data *md, unsigned long int ims,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
472    eptrblock *eptrb, 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 495  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
488  preserved over calls to RMATCH() are part of a "frame" which is obtained from  preserved over calls to RMATCH() are part of a "frame". We set up the top-level
489  heap storage. Set up the top-level frame here; others are obtained from the  frame on the stack here; subsequent instantiations are obtained from the heap
490  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  whenever RMATCH() does a "recursion". See the macro definitions above. Putting
491    the top-level on the stack rather than malloc-ing them all gives a performance
492    boost in many cases where there is not much "recursion". */
493    
494  #ifdef NO_RECURSE  #ifdef NO_RECURSE
495  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));  heapframe frame_zero;
496  if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);  heapframe *frame = &frame_zero;
497  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
498    
499  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 515  frame->Xprevframe = NULL;            /* Line 501  frame->Xprevframe = NULL;            /*
501  frame->Xeptr = eptr;  frame->Xeptr = eptr;
502  frame->Xecode = ecode;  frame->Xecode = ecode;
503  frame->Xmstart = mstart;  frame->Xmstart = mstart;
 frame->Xmarkptr = markptr;  
504  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
505  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
506  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
507    
508  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 531  HEAP_RECURSE: Line 514  HEAP_RECURSE:
514  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
515  #define ecode              frame->Xecode  #define ecode              frame->Xecode
516  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
 #define markptr            frame->Xmarkptr  
517  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
518  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
519  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
520    
521  /* Ditto for the local variables */  /* Ditto for the local variables */
522    
523  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
524  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
525  #endif  #endif
526  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 557  HEAP_RECURSE: Line 537  HEAP_RECURSE:
537  #define condition          frame->Xcondition  #define condition          frame->Xcondition
538  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
539    
 #define original_ims       frame->Xoriginal_ims  
   
540  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
541  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
542  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
543  #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  
544  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
545  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
546  #endif  #endif
# Line 595  i, and fc and c, can be the same variabl Line 570  i, and fc and c, can be the same variabl
570  #define fi i  #define fi i
571  #define fc c  #define fc c
572    
573    /* Many of the following variables are used only in small blocks of the code.
574    My normal style of coding would have declared them within each of those blocks.
575    However, in order to accommodate the version of this code that uses an external
576    "stack" implemented on the heap, it is easier to declare them all here, so the
577    declarations can be cut out in a block. The only declarations within blocks
578    below are for variables that do not have to be preserved over a recursive call
579    to RMATCH(). */
580    
581    #ifdef SUPPORT_UTF
582    const pcre_uchar *charptr;
583    #endif
584    const pcre_uchar *callpat;
585    const pcre_uchar *data;
586    const pcre_uchar *next;
587    PCRE_PUCHAR       pp;
588    const pcre_uchar *prev;
589    PCRE_PUCHAR       saved_eptr;
590    
591  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  recursion_info new_recursive;
592  const uschar *charptr;             /* in small blocks of the code. My normal */  
593  #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().          */  
594  BOOL condition;  BOOL condition;
595  BOOL prev_is_word;  BOOL prev_is_word;
596    
 unsigned long int original_ims;  
   
597  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
598  int prop_type;  int prop_type;
599  int prop_value;  int prop_value;
600  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
601  int oclength;  int oclength;
602  uschar occhars[8];  pcre_uchar occhars[6];
603  #endif  #endif
604    
605  int codelink;  int codelink;
# Line 638  int save_offset1, save_offset2, save_off Line 615  int save_offset1, save_offset2, save_off
615  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
616    
617  eptrblock newptrb;  eptrblock newptrb;
618    
619    /* There is a special fudge for calling match() in a way that causes it to
620    measure the size of its basic stack frame when the stack is being used for
621    recursion. The first argument (eptr) points to a pointer that is used
622    "statically" for doing the calculation. The second argument (ecode) being NULL
623    triggers this behaviour. It cannot normally every be NULL. The return is the
624    negated value of the frame size. */
625    
626    if (ecode == NULL)
627      {
628      char **aptr = (char **)eptr;
629      if (rdepth == 0)
630        {
631        *aptr = (char *)&rdepth;
632        return match(eptr, NULL, NULL, 0, NULL, NULL, 1);
633        }
634      else
635        {
636        int len = (char *)&rdepth - *aptr;
637        return (len > 0)? -len : len;
638        }
639      }
640  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
641    
642    /* To save space on the stack and in the heap frame, I have doubled up on some
643    of the local variables that are used only in localised parts of the code, but
644    still need to be preserved over recursive calls of match(). These macros define
645    the alternative names that are used. */
646    
647    #define allow_zero    cur_is_word
648    #define cbegroup      condition
649    #define code_offset   codelink
650    #define condassert    condition
651    #define matched_once  prev_is_word
652    #define foc           number
653    #define save_mark     data
654    
655  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
656  variables. */  variables. */
657    
# Line 664  defined). However, RMATCH isn't like a f Line 676  defined). However, RMATCH isn't like a f
676  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,
677  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
678    
679  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
680  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
681  #else  #else
682  utf8 = FALSE;  utf = FALSE;
683  #endif  #endif
684    
685  /* 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 676  haven't exceeded the recursive call limi Line 688  haven't exceeded the recursive call limi
688  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
689  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
690    
 original_ims = ims;    /* Save for resetting on ')' */  
   
691  /* 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
692  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
693  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
694  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
695  When match() is called in other circumstances, don't add to the chain. The  
696  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
697  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
698  match(). */  to break infinite loops that match no characters. When match() is called in
699    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
700    NOT be used with tail recursion, because the memory block that is used is on
701    the stack, so a new one may be required for each match(). */
702    
703  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
704    {    {
705    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
706    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
707    eptrb = &newptrb;    eptrb = &newptrb;
708      md->match_function_type = 0;
709    }    }
710    
711  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 704  for (;;) Line 718  for (;;)
718    switch(op)    switch(op)
719      {      {
720      case OP_MARK:      case OP_MARK:
721      markptr = ecode + 2;      md->nomatch_mark = ecode + 2;
722      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->mark = NULL;    /* In case previously set by assertion */
723        ims, eptrb, flags, RM55);      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
724          eptrb, RM55);
725        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
726             md->mark == NULL) md->mark = ecode + 2;
727    
728      /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an      /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
729      argument, and we must check whether that argument matches this MARK's      argument, and we must check whether that argument matches this MARK's
# Line 715  for (;;) Line 732  for (;;)
732      position and return MATCH_SKIP. Otherwise, pass back the return code      position and return MATCH_SKIP. Otherwise, pass back the return code
733      unaltered. */      unaltered. */
734    
735      if (rrc == MATCH_SKIP_ARG &&      else if (rrc == MATCH_SKIP_ARG &&
736          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)          STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0)
737        {        {
738        md->start_match_ptr = eptr;        md->start_match_ptr = eptr;
739        RRETURN(MATCH_SKIP);        RRETURN(MATCH_SKIP);
740        }        }
   
     if (md->mark == NULL) md->mark = markptr;  
741      RRETURN(rrc);      RRETURN(rrc);
742    
743      case OP_FAIL:      case OP_FAIL:
744      MRRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
745    
746      /* COMMIT overrides PRUNE, SKIP, and THEN */      /* COMMIT overrides PRUNE, SKIP, and THEN */
747    
748      case OP_COMMIT:      case OP_COMMIT:
749      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
750        ims, eptrb, flags, RM52);        eptrb, RM52);
751      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
752          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
753          rrc != MATCH_THEN)          rrc != MATCH_THEN)
754        RRETURN(rrc);        RRETURN(rrc);
755      MRRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
756    
757      /* PRUNE overrides THEN */      /* PRUNE overrides THEN */
758    
759      case OP_PRUNE:      case OP_PRUNE:
760      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
761        ims, eptrb, flags, RM51);        eptrb, RM51);
762      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
763      MRRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
764    
765      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
766      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->nomatch_mark = ecode + 2;
767        ims, eptrb, flags, RM56);      md->mark = NULL;    /* In case previously set by assertion */
768        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
769          eptrb, RM56);
770        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
771             md->mark == NULL) md->mark = ecode + 2;
772      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
     md->mark = ecode + 2;  
773      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
774    
775      /* SKIP overrides PRUNE and THEN */      /* SKIP overrides PRUNE and THEN */
776    
777      case OP_SKIP:      case OP_SKIP:
778      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
779        ims, eptrb, flags, RM53);        eptrb, RM53);
780      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
781        RRETURN(rrc);        RRETURN(rrc);
782      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
783      MRRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
784    
785        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
786        nomatch_mark. There is a flag that disables this opcode when re-matching a
787        pattern that ended with a SKIP for which there was not a matching MARK. */
788    
789      case OP_SKIP_ARG:      case OP_SKIP_ARG:
790      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      if (md->ignore_skip_arg)
791        ims, eptrb, flags, RM57);        {
792          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
793          break;
794          }
795        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
796          eptrb, RM57);
797      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
798        RRETURN(rrc);        RRETURN(rrc);
799    
800      /* Pass back the current skip name by overloading md->start_match_ptr and      /* Pass back the current skip name by overloading md->start_match_ptr and
801      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
802      caught by a matching MARK, or get to the top, where it is treated the same      caught by a matching MARK, or get to the top, where it causes a rematch
803      as PRUNE. */      with the md->ignore_skip_arg flag set. */
804    
805      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
806      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
807    
808      /* For THEN (and THEN_ARG) we pass back the address of the bracket or      /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
809      the alt that is at the start of the current branch. This makes it possible      the branch in which it occurs can be determined. Overload the start of
810      to skip back past alternatives that precede the THEN within the current      match pointer to do this. */
     branch. */  
811    
812      case OP_THEN:      case OP_THEN:
813      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
814        ims, eptrb, flags, RM54);        eptrb, RM54);
815      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
816      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
817      MRRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
818    
819      case OP_THEN_ARG:      case OP_THEN_ARG:
820      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      md->nomatch_mark = ecode + 2;
821        offset_top, md, ims, eptrb, flags, RM58);      md->mark = NULL;    /* In case previously set by assertion */
822        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
823          md, eptrb, RM58);
824        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
825             md->mark == NULL) md->mark = ecode + 2;
826      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
827      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
     md->mark = ecode + LINK_SIZE + 2;  
828      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
829    
830      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle an atomic group that does not contain any capturing parentheses.
831      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
832      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
833      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.
834      reference inside the group.      However, this uses a lot more stack, so in 8.20, atomic groups that do not
835        contain any captures generate OP_ONCE_NC, which can be handled in the old,
836      If the bracket fails to match, we need to restore this value and also the      less stack intensive way.
837      values of the final offsets, in case they were set by a previous iteration  
838      of the same bracket.      Check the alternative branches in turn - the matching won't pass the KET
839        for this kind of subpattern. If any one branch matches, we carry on as at
840        the end of a normal bracket, leaving the subject pointer, but resetting
841        the start-of-match value in case it was changed by \K. */
842    
843        case OP_ONCE_NC:
844        prev = ecode;
845        saved_eptr = eptr;
846        save_mark = md->mark;
847        do
848          {
849          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
850          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
851            {
852            mstart = md->start_match_ptr;
853            break;
854            }
855          if (rrc == MATCH_THEN)
856            {
857            next = ecode + GET(ecode,1);
858            if (md->start_match_ptr < next &&
859                (*ecode == OP_ALT || *next == OP_ALT))
860              rrc = MATCH_NOMATCH;
861            }
862    
863          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
864          ecode += GET(ecode,1);
865          md->mark = save_mark;
866          }
867        while (*ecode == OP_ALT);
868    
869        /* If hit the end of the group (which could be repeated), fail */
870    
871        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
872    
873        /* Continue as from after the group, updating the offsets high water
874        mark, since extracts may have been taken. */
875    
876        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
877    
878        offset_top = md->end_offset_top;
879        eptr = md->end_match_ptr;
880    
881        /* For a non-repeating ket, just continue at this level. This also
882        happens for a repeating ket if no characters were matched in the group.
883        This is the forcible breaking of infinite loops as implemented in Perl
884        5.005. */
885    
886        if (*ecode == OP_KET || eptr == saved_eptr)
887          {
888          ecode += 1+LINK_SIZE;
889          break;
890          }
891    
892        /* The repeating kets try the rest of the pattern or restart from the
893        preceding bracket, in the appropriate order. The second "call" of match()
894        uses tail recursion, to avoid using another stack frame. */
895    
896        if (*ecode == OP_KETRMIN)
897          {
898          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
899          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
900          ecode = prev;
901          goto TAIL_RECURSE;
902          }
903        else  /* OP_KETRMAX */
904          {
905          md->match_function_type = MATCH_CBEGROUP;
906          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
907          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
908          ecode += 1 + LINK_SIZE;
909          goto TAIL_RECURSE;
910          }
911        /* Control never gets here */
912    
913        /* Handle a capturing bracket, other than those that are possessive with an
914        unlimited repeat. If there is space in the offset vector, save the current
915        subject position in the working slot at the top of the vector. We mustn't
916        change the current values of the data slot, because they may be set from a
917        previous iteration of this group, and be referred to by a reference inside
918        the group. A failure to match might occur after the group has succeeded,
919        if something later on doesn't match. For this reason, we need to restore
920        the working value and also the values of the final offsets, in case they
921        were set by a previous iteration of the same bracket.
922    
923      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
924      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 830  for (;;) Line 942  for (;;)
942        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
943        save_offset3 = md->offset_vector[md->offset_end - number];        save_offset3 = md->offset_vector[md->offset_end - number];
944        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
945          save_mark = md->mark;
946    
947        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
948        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
949          (int)(eptr - md->start_subject);          (int)(eptr - md->start_subject);
950    
951        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
952          {          {
953          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
954            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
955          if (rrc != MATCH_NOMATCH &&            eptrb, RM1);
956              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
957            RRETURN(rrc);  
958            /* If we backed up to a THEN, check whether it is within the current
959            branch by comparing the address of the THEN that is passed back with
960            the end of the branch. If it is within the current branch, and the
961            branch is one of two or more alternatives (it either starts or ends
962            with OP_ALT), we have reached the limit of THEN's action, so convert
963            the return code to NOMATCH, which will cause normal backtracking to
964            happen from now on. Otherwise, THEN is passed back to an outer
965            alternative. This implements Perl's treatment of parenthesized groups,
966            where a group not containing | does not affect the current alternative,
967            that is, (X) is NOT the same as (X|(*F)). */
968    
969            if (rrc == MATCH_THEN)
970              {
971              next = ecode + GET(ecode,1);
972              if (md->start_match_ptr < next &&
973                  (*ecode == OP_ALT || *next == OP_ALT))
974                rrc = MATCH_NOMATCH;
975              }
976    
977            /* Anything other than NOMATCH is passed back. */
978    
979            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
980          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
981          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
982            md->mark = save_mark;
983            if (*ecode != OP_ALT) break;
984          }          }
       while (*ecode == OP_ALT);  
985    
986        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
987        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
988        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
989        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
990    
991        if (rrc != MATCH_THEN) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
992        RRETURN(MATCH_NOMATCH);  
993          RRETURN(rrc);
994        }        }
995    
996      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 869  for (;;) Line 1004  for (;;)
1004      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1005      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1006    
1007      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
1008      final alternative within the brackets, we would return the result of a      repeat and ONCE group with no captures. Loop for all the alternatives.
1009      recursive call to match() whatever happened. We can reduce stack usage by  
1010      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
1011      is set.*/      the result of a recursive call to match() whatever happened so it was
1012        possible to reduce stack usage by turning this into a tail recursion,
1013        except in the case of a possibly empty group. However, now that there is
1014        the possiblity of (*THEN) occurring in the final alternative, this
1015        optimization is no longer always possible.
1016    
1017        We can optimize if we know there are no (*THEN)s in the pattern; at present
1018        this is the best that can be done.
1019    
1020        MATCH_ONCE is returned when the end of an atomic group is successfully
1021        reached, but subsequent matching fails. It passes back up the tree (causing
1022        captured values to be reset) until the original atomic group level is
1023        reached. This is tested by comparing md->once_target with the start of the
1024        group. At this point, the return is converted into MATCH_NOMATCH so that
1025        previous backup points can be taken. */
1026    
1027        case OP_ONCE:
1028      case OP_BRA:      case OP_BRA:
1029      case OP_SBRA:      case OP_SBRA:
1030      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
1031      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
1032      for (;;)      for (;;)
1033        {        {
1034        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
1035    
1036          /* If this is not a possibly empty group, and there are no (*THEN)s in
1037          the pattern, and this is the final alternative, optimize as described
1038          above. */
1039    
1040          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1041          {          {
1042          if (flags == 0)    /* Not a possibly empty group */          ecode += PRIV(OP_lengths)[*ecode];
1043            goto TAIL_RECURSE;
1044            }
1045    
1046          /* In all other cases, we have to make another call to match(). */
1047    
1048          save_mark = md->mark;
1049          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1050            RM2);
1051    
1052          /* See comment in the code for capturing groups above about handling
1053          THEN. */
1054    
1055          if (rrc == MATCH_THEN)
1056            {
1057            next = ecode + GET(ecode,1);
1058            if (md->start_match_ptr < next &&
1059                (*ecode == OP_ALT || *next == OP_ALT))
1060              rrc = MATCH_NOMATCH;
1061            }
1062    
1063          if (rrc != MATCH_NOMATCH)
1064            {
1065            if (rrc == MATCH_ONCE)
1066            {            {
1067            ecode += _pcre_OP_lengths[*ecode];            const pcre_uchar *scode = ecode;
1068            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1069            goto TAIL_RECURSE;              {
1070                while (*scode == OP_ALT) scode += GET(scode, 1);
1071                scode -= GET(scode, 1);
1072                }
1073              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1074            }            }
1075            RRETURN(rrc);
1076            }
1077          ecode += GET(ecode, 1);
1078          md->mark = save_mark;
1079          if (*ecode != OP_ALT) break;
1080          }
1081    
1082          /* Possibly empty group; can't use tail recursion. */      RRETURN(MATCH_NOMATCH);
1083    
1084          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* Handle possessive capturing brackets with an unlimited repeat. We come
1085            eptrb, flags, RM48);      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1086          if (rrc == MATCH_NOMATCH) md->mark = markptr;      handled similarly to the normal case above. However, the matching is
1087          RRETURN(rrc);      different. The end of these brackets will always be OP_KETRPOS, which
1088        returns MATCH_KETRPOS without going further in the pattern. By this means
1089        we can handle the group by iteration rather than recursion, thereby
1090        reducing the amount of stack needed. */
1091    
1092        case OP_CBRAPOS:
1093        case OP_SCBRAPOS:
1094        allow_zero = FALSE;
1095    
1096        POSSESSIVE_CAPTURE:
1097        number = GET2(ecode, 1+LINK_SIZE);
1098        offset = number << 1;
1099    
1100    #ifdef PCRE_DEBUG
1101        printf("start possessive bracket %d\n", number);
1102        printf("subject=");
1103        pchars(eptr, 16, TRUE, md);
1104        printf("\n");
1105    #endif
1106    
1107        if (offset < md->offset_max)
1108          {
1109          matched_once = FALSE;
1110          code_offset = (int)(ecode - md->start_code);
1111    
1112          save_offset1 = md->offset_vector[offset];
1113          save_offset2 = md->offset_vector[offset+1];
1114          save_offset3 = md->offset_vector[md->offset_end - number];
1115          save_capture_last = md->capture_last;
1116    
1117          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1118    
1119          /* Each time round the loop, save the current subject position for use
1120          when the group matches. For MATCH_MATCH, the group has matched, so we
1121          restart it with a new subject starting position, remembering that we had
1122          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1123          usual. If we haven't matched any alternatives in any iteration, check to
1124          see if a previous iteration matched. If so, the group has matched;
1125          continue from afterwards. Otherwise it has failed; restore the previous
1126          capture values before returning NOMATCH. */
1127    
1128          for (;;)
1129            {
1130            md->offset_vector[md->offset_end - number] =
1131              (int)(eptr - md->start_subject);
1132            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1133            RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1134              eptrb, RM63);
1135            if (rrc == MATCH_KETRPOS)
1136              {
1137              offset_top = md->end_offset_top;
1138              eptr = md->end_match_ptr;
1139              ecode = md->start_code + code_offset;
1140              save_capture_last = md->capture_last;
1141              matched_once = TRUE;
1142              continue;
1143              }
1144    
1145            /* See comment in the code for capturing groups above about handling
1146            THEN. */
1147    
1148            if (rrc == MATCH_THEN)
1149              {
1150              next = ecode + GET(ecode,1);
1151              if (md->start_match_ptr < next &&
1152                  (*ecode == OP_ALT || *next == OP_ALT))
1153                rrc = MATCH_NOMATCH;
1154              }
1155    
1156            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1157            md->capture_last = save_capture_last;
1158            ecode += GET(ecode, 1);
1159            if (*ecode != OP_ALT) break;
1160          }          }
1161    
1162        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1163        otherwise return. */          {
1164            md->offset_vector[offset] = save_offset1;
1165            md->offset_vector[offset+1] = save_offset2;
1166            md->offset_vector[md->offset_end - number] = save_offset3;
1167            }
1168    
1169        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (allow_zero || matched_once)
1170          eptrb, flags, RM2);          {
1171        if (rrc != MATCH_NOMATCH &&          ecode += 1 + LINK_SIZE;
1172            (rrc != MATCH_THEN || md->start_match_ptr != ecode))          break;
1173          RRETURN(rrc);          }
1174    
1175          RRETURN(MATCH_NOMATCH);
1176          }
1177    
1178        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1179        as a non-capturing bracket. */
1180    
1181        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1182        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1183    
1184        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1185    
1186        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1187        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1188    
1189        /* Non-capturing possessive bracket with unlimited repeat. We come here
1190        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1191        without the capturing complication. It is written out separately for speed
1192        and cleanliness. */
1193    
1194        case OP_BRAPOS:
1195        case OP_SBRAPOS:
1196        allow_zero = FALSE;
1197    
1198        POSSESSIVE_NON_CAPTURE:
1199        matched_once = FALSE;
1200        code_offset = (int)(ecode - md->start_code);
1201    
1202        for (;;)
1203          {
1204          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1205          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1206            eptrb, RM48);
1207          if (rrc == MATCH_KETRPOS)
1208            {
1209            offset_top = md->end_offset_top;
1210            eptr = md->end_match_ptr;
1211            ecode = md->start_code + code_offset;
1212            matched_once = TRUE;
1213            continue;
1214            }
1215    
1216          /* See comment in the code for capturing groups above about handling
1217          THEN. */
1218    
1219          if (rrc == MATCH_THEN)
1220            {
1221            next = ecode + GET(ecode,1);
1222            if (md->start_match_ptr < next &&
1223                (*ecode == OP_ALT || *next == OP_ALT))
1224              rrc = MATCH_NOMATCH;
1225            }
1226    
1227          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1228        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1229          if (*ecode != OP_ALT) break;
1230        }        }
1231    
1232        if (matched_once || allow_zero)
1233          {
1234          ecode += 1 + LINK_SIZE;
1235          break;
1236          }
1237        RRETURN(MATCH_NOMATCH);
1238    
1239      /* Control never reaches here. */      /* Control never reaches here. */
1240    
1241      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1242      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
1243      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
1244      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. */  
1245    
1246      case OP_COND:      case OP_COND:
1247      case OP_SCOND:      case OP_SCOND:
1248      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1249    
1250      /* 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
1251      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
1252    
1253      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1254        {        {
1255        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1256          {          {
1257          pcre_callout_block cb;          PUBL(callout_block) cb;
1258          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1259          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1260          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1261    #ifdef COMPILE_PCRE8
1262          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1263    #else
1264            cb.subject          = (PCRE_SPTR16)md->start_subject;
1265    #endif
1266          cb.subject_length   = (int)(md->end_subject - md->start_subject);          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1267          cb.start_match      = (int)(mstart - md->start_subject);          cb.start_match      = (int)(mstart - md->start_subject);
1268          cb.current_position = (int)(eptr - md->start_subject);          cb.current_position = (int)(eptr - md->start_subject);
# Line 940  for (;;) Line 1271  for (;;)
1271          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1272          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1273          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1274          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1275            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1276          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1277          }          }
1278        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1279        }        }
1280    
1281      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
# Line 960  for (;;) Line 1292  for (;;)
1292        else        else
1293          {          {
1294          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1295          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1296    
1297          /* If the test is for recursion into a specific subpattern, and it is          /* If the test is for recursion into a specific subpattern, and it is
1298          false, but the test was set up by name, scan the table to see if the          false, but the test was set up by name, scan the table to see if the
1299          name refers to any other numbers, and test them. The condition is true          name refers to any other numbers, and test them. The condition is true
1300          if any one is set. */          if any one is set. */
1301    
1302          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1303            {            {
1304            uschar *slotA = md->name_table;            pcre_uchar *slotA = md->name_table;
1305            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
1306              {              {
1307              if (GET2(slotA, 0) == recno) break;              if (GET2(slotA, 0) == recno) break;
# Line 982  for (;;) Line 1314  for (;;)
1314    
1315            if (i < md->name_count)            if (i < md->name_count)
1316              {              {
1317              uschar *slotB = slotA;              pcre_uchar *slotB = slotA;
1318              while (slotB > md->name_table)              while (slotB > md->name_table)
1319                {                {
1320                slotB -= md->name_entry_size;                slotB -= md->name_entry_size;
1321                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1322                  {                  {
1323                  condition = GET2(slotB, 0) == md->recursive->group_num;                  condition = GET2(slotB, 0) == md->recursive->group_num;
1324                  if (condition) break;                  if (condition) break;
# Line 1002  for (;;) Line 1334  for (;;)
1334                for (i++; i < md->name_count; i++)                for (i++; i < md->name_count; i++)
1335                  {                  {
1336                  slotB += md->name_entry_size;                  slotB += md->name_entry_size;
1337                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1338                    {                    {
1339                    condition = GET2(slotB, 0) == md->recursive->group_num;                    condition = GET2(slotB, 0) == md->recursive->group_num;
1340                    if (condition) break;                    if (condition) break;
# Line 1015  for (;;) Line 1347  for (;;)
1347    
1348          /* Chose branch according to the condition */          /* Chose branch according to the condition */
1349    
1350          ecode += condition? 3 : GET(ecode, 1);          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1351          }          }
1352        }        }
1353    
# Line 1032  for (;;) Line 1364  for (;;)
1364        if (!condition && condcode == OP_NCREF)        if (!condition && condcode == OP_NCREF)
1365          {          {
1366          int refno = offset >> 1;          int refno = offset >> 1;
1367          uschar *slotA = md->name_table;          pcre_uchar *slotA = md->name_table;
1368    
1369          for (i = 0; i < md->name_count; i++)          for (i = 0; i < md->name_count; i++)
1370            {            {
# Line 1046  for (;;) Line 1378  for (;;)
1378    
1379          if (i < md->name_count)          if (i < md->name_count)
1380            {            {
1381            uschar *slotB = slotA;            pcre_uchar *slotB = slotA;
1382            while (slotB > md->name_table)            while (slotB > md->name_table)
1383              {              {
1384              slotB -= md->name_entry_size;              slotB -= md->name_entry_size;
1385              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)              if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1386                {                {
1387                offset = GET2(slotB, 0) << 1;                offset = GET2(slotB, 0) << 1;
1388                condition = offset < offset_top &&                condition = offset < offset_top &&
# Line 1068  for (;;) Line 1400  for (;;)
1400              for (i++; i < md->name_count; i++)              for (i++; i < md->name_count; i++)
1401                {                {
1402                slotB += md->name_entry_size;                slotB += md->name_entry_size;
1403                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1404                  {                  {
1405                  offset = GET2(slotB, 0) << 1;                  offset = GET2(slotB, 0) << 1;
1406                  condition = offset < offset_top &&                  condition = offset < offset_top &&
# Line 1083  for (;;) Line 1415  for (;;)
1415    
1416        /* Chose branch according to the condition */        /* Chose branch according to the condition */
1417    
1418        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1419        }        }
1420    
1421      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 1093  for (;;) Line 1425  for (;;)
1425        }        }
1426    
1427      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1428      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
1429      assertion. */      an assertion. */
1430    
1431      else      else
1432        {        {
1433        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1434            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1435        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1436          {          {
1437            if (md->end_offset_top > offset_top)
1438              offset_top = md->end_offset_top;  /* Captures may have happened */
1439          condition = TRUE;          condition = TRUE;
1440          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1441          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1442          }          }
1443        else if (rrc != MATCH_NOMATCH &&  
1444                (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1445          assertion; it is therefore treated as NOMATCH. */
1446    
1447          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1448          {          {
1449          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1450          }          }
# Line 1118  for (;;) Line 1455  for (;;)
1455          }          }
1456        }        }
1457    
1458      /* 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
1459      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
1460      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
1461      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
1462        exist, in which case we can just plough on. Note that, for compatibility
1463        with Perl, the | in a conditional group is NOT treated as creating two
1464        alternatives. If a THEN is encountered in the branch, it propagates out to
1465        the enclosing alternative (unless nested in a deeper set of alternatives,
1466        of course). */
1467    
1468      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1469        {        {
1470        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
1471          {          {
1472          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          ecode += 1 + LINK_SIZE;
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
1473          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1474          }          }
1475    
1476          md->match_function_type = MATCH_CBEGROUP;
1477          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1478          RRETURN(rrc);
1479        }        }
1480      else                         /* Condition false & no alternative */  
1481         /* Condition false & no alternative; continue after the group. */
1482    
1483        else
1484        {        {
1485        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1486        }        }
# Line 1164  for (;;) Line 1507  for (;;)
1507        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1508        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1509        }        }
1510      ecode += 3;      ecode += 1 + IMM2_SIZE;
1511      break;      break;
1512    
1513    
1514      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. */
     recursion, we should restore the offsets appropriately and continue from  
     after the call. */  
1515    
     case OP_ACCEPT:  
1516      case OP_END:      case OP_END:
1517      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1518        {      case OP_ASSERT_ACCEPT:
       recursion_info *rec = md->recursive;  
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
       md->recursive = rec->prevrec;  
       memmove(md->offset_vector, rec->offset_save,  
         rec->saved_max * sizeof(int));  
       offset_top = rec->save_offset_top;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1519    
1520      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is      /* If we have matched an empty string, fail if not in an assertion and not
1521      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1522      the subject. In both cases, backtracking will then try other alternatives,      is set and we have matched at the start of the subject. In both cases,
1523      if any. */      backtracking will then try other alternatives, if any. */
1524    
1525      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1526          (md->notempty ||           md->recursive == NULL &&
1527            (md->notempty_atstart &&           (md->notempty ||
1528              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1529        MRRETURN(MATCH_NOMATCH);               mstart == md->start_subject + md->start_offset)))
1530          RRETURN(MATCH_NOMATCH);
1531    
1532      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1533    
# Line 1205  for (;;) Line 1536  for (;;)
1536      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1537    
1538      /* For some reason, the macros don't work properly if an expression is      /* For some reason, the macros don't work properly if an expression is
1539      given as the argument to MRRETURN when the heap is in use. */      given as the argument to RRETURN when the heap is in use. */
1540    
1541      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1542      MRRETURN(rrc);      RRETURN(rrc);
   
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1543    
1544      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1545      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,
1546      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
1547      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
1548      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1549        of a condition, we want to return immediately afterwards. The caller of
1550        this incarnation of the match() function will have set MATCH_CONDASSERT in
1551        md->match_function type, and one of these opcodes will be the first opcode
1552        that is processed. We use a local variable that is preserved over calls to
1553        match() to remember this case. */
1554    
1555      case OP_ASSERT:      case OP_ASSERT:
1556      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1557        save_mark = md->mark;
1558        if (md->match_function_type == MATCH_CONDASSERT)
1559          {
1560          condassert = TRUE;
1561          md->match_function_type = 0;
1562          }
1563        else condassert = FALSE;
1564    
1565      do      do
1566        {        {
1567        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1568        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1569          {          {
1570          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1571          break;          break;
1572          }          }
1573        if (rrc != MATCH_NOMATCH &&  
1574            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1575          RRETURN(rrc);        as NOMATCH. */
1576    
1577          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1578        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1579          md->mark = save_mark;
1580        }        }
1581      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1582      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);  
1583        if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1584    
1585      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1586    
1587      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1588    
1589      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1590      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1261  for (;;) Line 1600  for (;;)
1600    
1601      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1602      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1603        save_mark = md->mark;
1604        if (md->match_function_type == MATCH_CONDASSERT)
1605          {
1606          condassert = TRUE;
1607          md->match_function_type = 0;
1608          }
1609        else condassert = FALSE;
1610    
1611      do      do
1612        {        {
1613        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1614          RM5);        md->mark = save_mark;
1615        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1616        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1617          {          {
1618          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1619          break;          break;
1620          }          }
1621        if (rrc != MATCH_NOMATCH &&  
1622            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1623          RRETURN(rrc);        as NOMATCH. */
1624    
1625          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1626        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1627        }        }
1628      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1629    
1630      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1631    
1632      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1633      continue;      continue;
# Line 1289  for (;;) Line 1638  for (;;)
1638      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1639    
1640      case OP_REVERSE:      case OP_REVERSE:
1641  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1642      if (utf8)      if (utf)
1643        {        {
1644        i = GET(ecode, 1);        i = GET(ecode, 1);
1645        while (i-- > 0)        while (i-- > 0)
1646          {          {
1647          eptr--;          eptr--;
1648          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1649          BACKCHAR(eptr);          BACKCHAR(eptr);
1650          }          }
1651        }        }
# Line 1307  for (;;) Line 1656  for (;;)
1656    
1657        {        {
1658        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1659        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1660        }        }
1661    
1662      /* Save the earliest consulted character, then skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
# Line 1321  for (;;) Line 1670  for (;;)
1670      function is able to force a failure. */      function is able to force a failure. */
1671    
1672      case OP_CALLOUT:      case OP_CALLOUT:
1673      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1674        {        {
1675        pcre_callout_block cb;        PUBL(callout_block) cb;
1676        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1677        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1678        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1679    #ifdef COMPILE_PCRE8
1680        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1681    #else
1682          cb.subject          = (PCRE_SPTR16)md->start_subject;
1683    #endif
1684        cb.subject_length   = (int)(md->end_subject - md->start_subject);        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1685        cb.start_match      = (int)(mstart - md->start_subject);        cb.start_match      = (int)(mstart - md->start_subject);
1686        cb.current_position = (int)(eptr - md->start_subject);        cb.current_position = (int)(eptr - md->start_subject);
# Line 1336  for (;;) Line 1689  for (;;)
1689        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1690        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1691        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1692        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1693          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1694        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1695        }        }
1696      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1346  for (;;) Line 1700  for (;;)
1700      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
1701      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1702    
1703      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1704      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
1705      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
1706      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
1707      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
1708      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
1709      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.  
1710    
1711      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
1712      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
1713      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1714        a lot, so he is not to blame for the current way it works. */
1715    
1716      case OP_RECURSE:      case OP_RECURSE:
1717        {        {
1718          recursion_info *ri;
1719          int recno;
1720    
1721        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1722        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1723          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1724    
1725          /* Check for repeating a recursion without advancing the subject pointer.
1726          This should catch convoluted mutual recursions. (Some simple cases are
1727          caught at compile time.) */
1728    
1729          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1730            if (recno == ri->group_num && eptr == ri->subject_position)
1731              RRETURN(PCRE_ERROR_RECURSELOOP);
1732    
1733        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1734    
1735          new_recursive.group_num = recno;
1736          new_recursive.subject_position = eptr;
1737        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1738        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1739    
1740        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1741    
1742        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1743    
1744        /* Now save the offset data. */        /* Now save the offset data */
1745    
1746        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1747        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1385  for (;;) Line 1749  for (;;)
1749        else        else
1750          {          {
1751          new_recursive.offset_save =          new_recursive.offset_save =
1752            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1753          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1754          }          }
   
1755        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1756              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1757    
1758        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1759        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1760          might be changed, so reset it before looping. */
1761    
1762        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1763        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1764        do        do
1765          {          {
1766          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1767            md, ims, eptrb, flags, RM6);          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1768              md, eptrb, RM6);
1769            memcpy(md->offset_vector, new_recursive.offset_save,
1770                new_recursive.saved_max * sizeof(int));
1771            md->recursive = new_recursive.prevrec;
1772          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1773            {            {
1774            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1775            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1776              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1777            MRRETURN(MATCH_MATCH);  
1778              /* Set where we got to in the subject, and reset the start in case
1779              it was changed by \K. This *is* propagated back out of a recursion,
1780              for Perl compatibility. */
1781    
1782              eptr = md->end_match_ptr;
1783              mstart = md->start_match_ptr;
1784              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1785            }            }
1786          else if (rrc != MATCH_NOMATCH &&  
1787                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1788            as NOMATCH. */
1789    
1790            else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1791            {            {
1792            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1793            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1794              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1795            RRETURN(rrc);            RRETURN(rrc);
1796            }            }
1797    
1798          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1799          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1800          }          }
1801        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1429  for (;;) Line 1803  for (;;)
1803        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1804        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1805        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1806          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1807        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
       }  
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer, but resetting  
     the start-of-match value in case it was changed by \K. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */  
         {  
         mstart = md->start_match_ptr;  
         break;  
         }  
       if (rrc != MATCH_NOMATCH &&  
           (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
         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));  
1808        }        }
1809    
1810      if (*ecode == OP_KETRMIN)      RECURSION_MATCHED:
1811        {      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 */  
1812    
1813      /* 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
1814      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1529  for (;;) Line 1824  for (;;)
1824      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1825    
1826      case OP_BRAZERO:      case OP_BRAZERO:
1827        {      next = ecode + 1;
1828        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1829        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1830        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1831        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1832      break;      break;
1833    
1834      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1835        {      next = ecode + 1;
1836        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1837        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1838        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1839        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1840      break;      break;
1841    
1842      case OP_SKIPZERO:      case OP_SKIPZERO:
1843        {      next = ecode+1;
1844        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1845        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1846      break;      break;
1847    
1848        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1849        here; just jump to the group, with allow_zero set TRUE. */
1850    
1851        case OP_BRAPOSZERO:
1852        op = *(++ecode);
1853        allow_zero = TRUE;
1854        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1855          goto POSSESSIVE_NON_CAPTURE;
1856    
1857      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1858    
1859      case OP_KET:      case OP_KET:
1860      case OP_KETRMIN:      case OP_KETRMIN:
1861      case OP_KETRMAX:      case OP_KETRMAX:
1862        case OP_KETRPOS:
1863      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1864    
1865      /* 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
1866      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1867      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1868    
1869      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1870        {        {
1871        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1872        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1873        }        }
1874      else saved_eptr = NULL;      else saved_eptr = NULL;
1875    
1876      /* If we are at the end of an assertion group or an atomic group, stop      /* If we are at the end of an assertion group or a non-capturing atomic
1877      matching and return MATCH_MATCH, but record the current high water mark for      group, stop matching and return MATCH_MATCH, but record the current high
1878      use by positive assertions. We also need to record the match start in case      water mark for use by positive assertions. We also need to record the match
1879      it was changed by \K. */      start in case it was changed by \K. */
1880    
1881      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1882          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||           *prev == OP_ONCE_NC)
         *prev == OP_ONCE)  
1883        {        {
1884        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1885        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1886        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1887        MRRETURN(MATCH_MATCH);        RRETURN(MATCH_MATCH);         /* Sets md->mark */
1888        }        }
1889    
1890      /* 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
1891      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1892      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
1893      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
1894      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
1895        the current subject position and start match pointer and give a MATCH
1896        return. */
1897    
1898      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1899            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1900        {        {
1901        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1902        offset = number << 1;        offset = number << 1;
# Line 1605  for (;;) Line 1906  for (;;)
1906        printf("\n");        printf("\n");
1907  #endif  #endif
1908    
1909          /* Handle a recursively called group. */
1910    
1911          if (md->recursive != NULL && md->recursive->group_num == number)
1912            {
1913            md->end_match_ptr = eptr;
1914            md->start_match_ptr = mstart;
1915            RRETURN(MATCH_MATCH);
1916            }
1917    
1918          /* Deal with capturing */
1919    
1920        md->capture_last = number;        md->capture_last = number;
1921        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1922          {          {
1923            /* If offset is greater than offset_top, it means that we are
1924            "skipping" a capturing group, and that group's offsets must be marked
1925            unset. In earlier versions of PCRE, all the offsets were unset at the
1926            start of matching, but this doesn't work because atomic groups and
1927            assertions can cause a value to be set that should later be unset.
1928            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1929            part of the atomic group, but this is not on the final matching path,
1930            so must be unset when 2 is set. (If there is no group 2, there is no
1931            problem, because offset_top will then be 2, indicating no capture.) */
1932    
1933            if (offset > offset_top)
1934              {
1935              register int *iptr = md->offset_vector + offset_top;
1936              register int *iend = md->offset_vector + offset;
1937              while (iptr < iend) *iptr++ = -1;
1938              }
1939    
1940            /* Now make the extraction */
1941    
1942          md->offset_vector[offset] =          md->offset_vector[offset] =
1943            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1944          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1945          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1946          }          }
1947          }
1948    
1949        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1950        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1951        group. This is the forcible breaking of infinite loops as implemented in
1952        Perl 5.005. For a non-repeating atomic group that includes captures,
1953        establish a backup point by processing the rest of the pattern at a lower
1954        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1955        original OP_ONCE level, thereby bypassing intermediate backup points, but
1956        resetting any captures that happened along the way. */
1957    
1958        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1959          {
1960          if (*prev == OP_ONCE)
1961          {          {
1962          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1963          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1964          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1965          memcpy(md->offset_vector, rec->offset_save,          RRETURN(MATCH_ONCE);
           rec->saved_max * sizeof(int));  
         offset_top = rec->save_offset_top;  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1966          }          }
1967          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1968          break;
1969        }        }
1970    
1971      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1972      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1973        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. */  
1974    
1975      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRPOS)
1976        {        {
1977        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1978        break;        md->end_offset_top = offset_top;
1979          RRETURN(MATCH_KETRPOS);
1980        }        }
1981    
1982      /* 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
1983      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
1984      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
1985      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
1986        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1987    
1988      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1989        {        {
1990        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1991        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1992        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1993          {          {
1994          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1995            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1996            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1997            RRETURN(MATCH_ONCE);
1998            }
1999          if (*prev >= OP_SBRA)    /* Could match an empty string */
2000            {
2001            md->match_function_type = MATCH_CBEGROUP;
2002            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
2003          RRETURN(rrc);          RRETURN(rrc);
2004          }          }
2005        ecode = prev;        ecode = prev;
# Line 1670  for (;;) Line 2007  for (;;)
2007        }        }
2008      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
2009        {        {
2010        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
2011          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
2012          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
2013        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2014          if (*prev == OP_ONCE)
2015            {
2016            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
2017            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2018            md->once_target = prev;
2019            RRETURN(MATCH_ONCE);
2020            }
2021        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
2022        goto TAIL_RECURSE;        goto TAIL_RECURSE;
2023        }        }
2024      /* Control never gets here */      /* Control never gets here */
2025    
2026      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
2027    
2028      case OP_CIRC:      case OP_CIRC:
2029      if (md->notbol && eptr == md->start_subject) MRRETURN(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)))  
         MRRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
2030    
2031      /* Start of subject assertion */      /* Start of subject assertion */
2032    
2033      case OP_SOD:      case OP_SOD:
2034      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);
2035        ecode++;
2036        break;
2037    
2038        /* Multiline mode: start of subject unless notbol, or after any newline. */
2039    
2040        case OP_CIRCM:
2041        if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2042        if (eptr != md->start_subject &&
2043            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
2044          RRETURN(MATCH_NOMATCH);
2045      ecode++;      ecode++;
2046      break;      break;
2047    
2048      /* Start of match assertion */      /* Start of match assertion */
2049    
2050      case OP_SOM:      case OP_SOM:
2051      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);
2052      ecode++;      ecode++;
2053      break;      break;
2054    
# Line 1713  for (;;) Line 2059  for (;;)
2059      ecode++;      ecode++;
2060      break;      break;
2061    
2062      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2063      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2064    
2065      case OP_DOLL:      case OP_DOLLM:
2066      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2067        {        { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }
2068        if (eptr < md->end_subject)      else
         { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }  
       else  
         {  
         if (md->noteol) MRRETURN(MATCH_NOMATCH);  
         SCHECK_PARTIAL();  
         }  
       ecode++;  
       break;  
       }  
     else  /* Not multiline */  
2069        {        {
2070        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2071        if (!md->endonly) goto ASSERT_NL_OR_EOS;        SCHECK_PARTIAL();
2072        }        }
2073        ecode++;
2074        break;
2075    
2076        /* Not multiline mode: assert before a terminating newline or before end of
2077        subject unless noteol is set. */
2078    
2079        case OP_DOLL:
2080        if (md->noteol) RRETURN(MATCH_NOMATCH);
2081        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2082    
2083      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2084    
2085      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2086    
2087      case OP_EOD:      case OP_EOD:
2088      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2089      SCHECK_PARTIAL();      SCHECK_PARTIAL();
2090      ecode++;      ecode++;
2091      break;      break;
# Line 1751  for (;;) Line 2096  for (;;)
2096      ASSERT_NL_OR_EOS:      ASSERT_NL_OR_EOS:
2097      if (eptr < md->end_subject &&      if (eptr < md->end_subject &&
2098          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2099        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2100    
2101      /* Either at end of string or \n before end. */      /* Either at end of string or \n before end. */
2102    
# Line 1770  for (;;) Line 2115  for (;;)
2115        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2116        partial matching. */        partial matching. */
2117    
2118  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2119        if (utf8)        if (utf)
2120          {          {
2121          /* Get status of previous character */          /* Get status of previous character */
2122    
2123          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2124            {            {
2125            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2126            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2127            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2128            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2129  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 1843  for (;;) Line 2188  for (;;)
2188              }              }
2189            else            else
2190  #endif  #endif
2191            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);            prev_is_word = MAX_255(eptr[-1])
2192                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2193            }            }
2194    
2195          /* Get status of next character */          /* Get status of next character */
# Line 1866  for (;;) Line 2212  for (;;)
2212            }            }
2213          else          else
2214  #endif  #endif
2215          cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          cur_is_word = MAX_255(*eptr)
2216              && ((md->ctypes[*eptr] & ctype_word) != 0);
2217          }          }
2218    
2219        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2220    
2221        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2222             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2223          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2224        }        }
2225      break;      break;
2226    
2227      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2228    
2229      case OP_ANY:      case OP_ANY:
2230      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
2231      /* Fall through */      /* Fall through */
2232    
2233      case OP_ALLANY:      case OP_ALLANY:
2234      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2235        {        {                            /* not be updated before SCHECK_PARTIAL. */
2236        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2237        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2238        }        }
2239      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2240    #ifdef SUPPORT_UTF
2241        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2242    #endif
2243      ecode++;      ecode++;
2244      break;      break;
2245    
# Line 1897  for (;;) Line 2247  for (;;)
2247      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2248    
2249      case OP_ANYBYTE:      case OP_ANYBYTE:
2250      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2251        {        {                            /* not be updated before SCHECK_PARTIAL. */
2252        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2253        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2254        }        }
2255        eptr++;
2256      ecode++;      ecode++;
2257      break;      break;
2258    
# Line 1909  for (;;) Line 2260  for (;;)
2260      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2261        {        {
2262        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2263        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2264        }        }
2265      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2266      if (      if (
2267  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2268         c < 256 &&         c < 256 &&
2269  #endif  #endif
2270         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2271         )         )
2272        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2273      ecode++;      ecode++;
2274      break;      break;
2275    
# Line 1926  for (;;) Line 2277  for (;;)
2277      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2278        {        {
2279        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2280        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2281        }        }
2282      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2283      if (      if (
2284  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2285         c >= 256 ||         c > 255 ||
2286  #endif  #endif
2287         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2288         )         )
2289        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2290      ecode++;      ecode++;
2291      break;      break;
2292    
# Line 1943  for (;;) Line 2294  for (;;)
2294      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2295        {        {
2296        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2297        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2298        }        }
2299      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2300      if (      if (
2301  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2302         c < 256 &&         c < 256 &&
2303  #endif  #endif
2304         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2305         )         )
2306        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2307      ecode++;      ecode++;
2308      break;      break;
2309    
# Line 1960  for (;;) Line 2311  for (;;)
2311      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2312        {        {
2313        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2314        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2315        }        }
2316      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2317      if (      if (
2318  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2319         c >= 256 ||         c > 255 ||
2320  #endif  #endif
2321         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2322         )         )
2323        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2324      ecode++;      ecode++;
2325      break;      break;
2326    
# Line 1977  for (;;) Line 2328  for (;;)
2328      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2329        {        {
2330        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2331        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2332        }        }
2333      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2334      if (      if (
2335  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2336         c < 256 &&         c < 256 &&
2337  #endif  #endif
2338         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2339         )         )
2340        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2341      ecode++;      ecode++;
2342      break;      break;
2343    
# Line 1994  for (;;) Line 2345  for (;;)
2345      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2346        {        {
2347        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2348        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2349        }        }
2350      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2351      if (      if (
2352  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2353         c >= 256 ||         c > 255 ||
2354  #endif  #endif
2355         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2356         )         )
2357        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2358      ecode++;      ecode++;
2359      break;      break;
2360    
# Line 2011  for (;;) Line 2362  for (;;)
2362      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2363        {        {
2364        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2365        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2366        }        }
2367      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2368      switch(c)      switch(c)
2369        {        {
2370        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2371    
2372        case 0x000d:        case 0x000d:
2373        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2374        break;        break;
# Line 2029  for (;;) Line 2381  for (;;)
2381        case 0x0085:        case 0x0085:
2382        case 0x2028:        case 0x2028:
2383        case 0x2029:        case 0x2029:
2384        if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);        if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
2385        break;        break;
2386        }        }
2387      ecode++;      ecode++;
# Line 2039  for (;;) Line 2391  for (;;)
2391      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2392        {        {
2393        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2394        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2395        }        }
2396      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2397      switch(c)      switch(c)
# Line 2064  for (;;) Line 2416  for (;;)
2416        case 0x202f:    /* NARROW NO-BREAK SPACE */        case 0x202f:    /* NARROW NO-BREAK SPACE */
2417        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2418        case 0x3000:    /* IDEOGRAPHIC SPACE */        case 0x3000:    /* IDEOGRAPHIC SPACE */
2419        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2420        }        }
2421      ecode++;      ecode++;
2422      break;      break;
# Line 2073  for (;;) Line 2425  for (;;)
2425      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2426        {        {
2427        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2428        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2429        }        }
2430      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2431      switch(c)      switch(c)
2432        {        {
2433        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2434        case 0x09:      /* HT */        case 0x09:      /* HT */
2435        case 0x20:      /* SPACE */        case 0x20:      /* SPACE */
2436        case 0xa0:      /* NBSP */        case 0xa0:      /* NBSP */
# Line 2107  for (;;) Line 2459  for (;;)
2459      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2460        {        {
2461        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2462        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2463        }        }
2464      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2465      switch(c)      switch(c)
# Line 2120  for (;;) Line 2472  for (;;)
2472        case 0x85:      /* NEL */        case 0x85:      /* NEL */
2473        case 0x2028:    /* LINE SEPARATOR */        case 0x2028:    /* LINE SEPARATOR */
2474        case 0x2029:    /* PARAGRAPH SEPARATOR */        case 0x2029:    /* PARAGRAPH SEPARATOR */
2475        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2476        }        }
2477      ecode++;      ecode++;
2478      break;      break;
# Line 2129  for (;;) Line 2481  for (;;)
2481      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2482        {        {
2483        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2484        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2485        }        }
2486      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2487      switch(c)      switch(c)
2488        {        {
2489        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2490        case 0x0a:      /* LF */        case 0x0a:      /* LF */
2491        case 0x0b:      /* VT */        case 0x0b:      /* VT */
2492        case 0x0c:      /* FF */        case 0x0c:      /* FF */
# Line 2156  for (;;) Line 2508  for (;;)
2508      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2509        {        {
2510        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2511        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2512        }        }
2513      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2514        {        {
# Line 2165  for (;;) Line 2517  for (;;)
2517        switch(ecode[1])        switch(ecode[1])
2518          {          {
2519          case PT_ANY:          case PT_ANY:
2520          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2521          break;          break;
2522    
2523          case PT_LAMP:          case PT_LAMP:
2524          if ((prop->chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2525               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2526               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2527            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2528          break;          break;
2529    
2530          case PT_GC:          case PT_GC:
2531          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2532            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2533          break;          break;
2534    
2535          case PT_PC:          case PT_PC:
2536          if ((ecode[2] != prop->chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2537            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2538          break;          break;
2539    
2540          case PT_SC:          case PT_SC:
2541          if ((ecode[2] != prop->script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2542            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2543          break;          break;
2544    
2545          /* These are specials */          /* These are specials */
2546    
2547          case PT_ALNUM:          case PT_ALNUM:
2548          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2549               _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))               PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2550            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2551          break;          break;
2552    
2553          case PT_SPACE:    /* Perl space */          case PT_SPACE:    /* Perl space */
2554          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2555               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2556                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2557            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2558          break;          break;
2559    
2560          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2561          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2562               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2563               c == CHAR_FF || c == CHAR_CR)               c == CHAR_FF || c == CHAR_CR)
2564                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2565            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2566          break;          break;
2567    
2568          case PT_WORD:          case PT_WORD:
2569          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2570               _pcre_ucp_gentype[prop->chartype] == ucp_N ||               PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2571               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2572            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2573          break;          break;
2574    
2575          /* This should never occur */          /* This should never occur */
# Line 2237  for (;;) Line 2589  for (;;)
2589      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2590        {        {
2591        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2592        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2593        }        }
2594      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2595        if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH);
2596        while (eptr < md->end_subject)
2597        {        {
2598        int category = UCD_CATEGORY(c);        int len = 1;
2599        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2600        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2601          {        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;  
         }  
2602        }        }
2603      ecode++;      ecode++;
2604      break;      break;
# Line 2269  for (;;) Line 2614  for (;;)
2614      loops). */      loops). */
2615    
2616      case OP_REF:      case OP_REF:
2617        case OP_REFI:
2618        caseless = op == OP_REFI;
2619      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2620      ecode += 3;      ecode += 1 + IMM2_SIZE;
2621    
2622      /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2623    
# Line 2310  for (;;) Line 2657  for (;;)
2657        case OP_CRMINRANGE:        case OP_CRMINRANGE:
2658        minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2659        min = GET2(ecode, 1);        min = GET2(ecode, 1);
2660        max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2661        if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2662        ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2663        break;        break;
2664    
2665        default:               /* No repeat follows */        default:               /* No repeat follows */
2666        if ((length = match_ref(offset, eptr, length, md, ims)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2667          {          {
2668          CHECK_PARTIAL();          CHECK_PARTIAL();
2669          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2670          }          }
2671        eptr += length;        eptr += length;
2672        continue;              /* With the main loop */        continue;              /* With the main loop */
2673        }        }
2674    
2675      /* Handle repeated back references. If the length of the reference is      /* Handle repeated back references. If the length of the reference is
2676      zero, just continue with the main loop. */      zero, just continue with the main loop. If the length is negative, it
2677        means the reference is unset in non-Java-compatible mode. If the minimum is
2678        zero, we can continue at the same level without recursion. For any other
2679        minimum, carrying on will result in NOMATCH. */
2680    
2681      if (length == 0) continue;      if (length == 0) continue;
2682        if (length < 0 && min == 0) continue;
2683    
2684      /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2685      the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
# Line 2336  for (;;) Line 2687  for (;;)
2687    
2688      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2689        {        {
2690        int slength;        int slength;
2691        if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2692          {          {
2693          CHECK_PARTIAL();          CHECK_PARTIAL();
2694          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2695          }          }
2696        eptr += slength;        eptr += slength;
2697        }        }
# Line 2356  for (;;) Line 2707  for (;;)
2707        {        {
2708        for (fi = min;; fi++)        for (fi = min;; fi++)
2709          {          {
2710          int slength;          int slength;
2711          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2712          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2713          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) RRETURN(MATCH_NOMATCH);
2714          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2715            {            {
2716            CHECK_PARTIAL();            CHECK_PARTIAL();
2717            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2718            }            }
2719          eptr += slength;          eptr += slength;
2720          }          }
# Line 2377  for (;;) Line 2728  for (;;)
2728        pp = eptr;        pp = eptr;
2729        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2730          {          {
2731          int slength;          int slength;
2732          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2733            {            {
2734            CHECK_PARTIAL();            CHECK_PARTIAL();
2735            break;            break;
# Line 2387  for (;;) Line 2738  for (;;)
2738          }          }
2739        while (eptr >= pp)        while (eptr >= pp)
2740          {          {
2741          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2742          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2743          eptr -= length;          eptr -= length;
2744          }          }
2745        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2746        }        }
2747      /* Control never gets here */      /* Control never gets here */
2748    
# Line 2409  for (;;) Line 2760  for (;;)
2760      case OP_NCLASS:      case OP_NCLASS:
2761      case OP_CLASS:      case OP_CLASS:
2762        {        {
2763          /* The data variable is saved across frames, so the byte map needs to
2764          be stored there. */
2765    #define BYTE_MAP ((pcre_uint8 *)data)
2766        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2767        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2768    
2769        switch (*ecode)        switch (*ecode)
2770          {          {
# Line 2431  for (;;) Line 2785  for (;;)
2785          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2786          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2787          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2788          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2789          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2790          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2791          break;          break;
2792    
2793          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2443  for (;;) Line 2797  for (;;)
2797    
2798        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2799    
2800  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2801        /* UTF-8 mode */        if (utf)
       if (utf8)  
2802          {          {
2803          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2804            {            {
2805            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2806              {              {
2807              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2808              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2809              }              }
2810            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2811            if (c > 255)            if (c > 255)
2812              {              {
2813              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2814              }              }
2815            else            else
2816              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
             }  
2817            }            }
2818          }          }
2819        else        else
2820  #endif  #endif
2821        /* Not UTF-8 mode */        /* Not UTF mode */
2822          {          {
2823          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2824            {            {
2825            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2826              {              {
2827              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2828              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2829              }              }
2830            c = *eptr++;            c = *eptr++;
2831            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2832              if (c > 255)
2833                {
2834                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2835                }
2836              else
2837    #endif
2838                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2839            }            }
2840          }          }
2841    
# Line 2491  for (;;) Line 2849  for (;;)
2849    
2850        if (minimize)        if (minimize)
2851          {          {
2852  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2853          /* UTF-8 mode */          if (utf)
         if (utf8)  
2854            {            {
2855            for (fi = min;; fi++)            for (fi = min;; fi++)
2856              {              {
2857              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2858              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2859              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2860              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2861                {                {
2862                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2863                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2864                }                }
2865              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2866              if (c > 255)              if (c > 255)
2867                {                {
2868                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2869                }                }
2870              else              else
2871                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
               }  
2872              }              }
2873            }            }
2874          else          else
2875  #endif  #endif
2876          /* Not UTF-8 mode */          /* Not UTF mode */
2877            {            {
2878            for (fi = min;; fi++)            for (fi = min;; fi++)
2879              {              {
2880              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2881              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2882              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2883              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2884                {                {
2885                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2886                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2887                }                }
2888              c = *eptr++;              c = *eptr++;
2889              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2890                if (c > 255)
2891                  {
2892                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2893                  }
2894                else
2895    #endif
2896                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2897              }              }
2898            }            }
2899          /* Control never gets here */          /* Control never gets here */
# Line 2543  for (;;) Line 2905  for (;;)
2905          {          {
2906          pp = eptr;          pp = eptr;
2907    
2908  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2909          /* UTF-8 mode */          if (utf)
         if (utf8)  
2910            {            {
2911            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2912              {              {
# Line 2561  for (;;) Line 2922  for (;;)
2922                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
2923                }                }
2924              else              else
2925                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
2926              eptr += len;              eptr += len;
2927              }              }
2928            for (;;)            for (;;)
2929              {              {
2930              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2931              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2932              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2933              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2576  for (;;) Line 2935  for (;;)
2935            }            }
2936          else          else
2937  #endif  #endif
2938            /* Not UTF-8 mode */            /* Not UTF mode */
2939            {            {
2940            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2941              {              {
# Line 2586  for (;;) Line 2945  for (;;)
2945                break;                break;
2946                }                }
2947              c = *eptr;              c = *eptr;
2948              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
2949                if (c > 255)
2950                  {
2951                  if (op == OP_CLASS) break;
2952                  }
2953                else
2954    #endif
2955                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
2956              eptr++;              eptr++;
2957              }              }
2958            while (eptr >= pp)            while (eptr >= pp)
2959              {              {
2960              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2961              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2962              eptr--;              eptr--;
2963              }              }
2964            }            }
2965    
2966          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2967          }          }
2968    #undef BYTE_MAP
2969        }        }
2970      /* Control never gets here */      /* Control never gets here */
2971    
# Line 2607  for (;;) Line 2974  for (;;)
2974      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
2975      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
2976    
2977  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2978      case OP_XCLASS:      case OP_XCLASS:
2979        {        {
2980        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2632  for (;;) Line 2999  for (;;)
2999          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3000          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3001          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3002          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
3003          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
3004          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
3005          break;          break;
3006    
3007          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2649  for (;;) Line 3016  for (;;)
3016          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3017            {            {
3018            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3019            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3020            }            }
3021          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
3022          if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3023          }          }
3024    
3025        /* 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 2667  for (;;) Line 3034  for (;;)
3034          {          {
3035          for (fi = min;; fi++)          for (fi = min;; fi++)
3036            {            {
3037            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
3038            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3039            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3040            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3041              {              {
3042              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3043              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3044              }              }
3045            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3046            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3047            }            }
3048          /* Control never gets here */          /* Control never gets here */
3049          }          }
# Line 2694  for (;;) Line 3061  for (;;)
3061              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3062              break;              break;
3063              }              }
3064    #ifdef SUPPORT_UTF
3065            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3066            if (!_pcre_xclass(c, data)) break;  #else
3067              c = *eptr;
3068    #endif
3069              if (!PRIV(xclass)(c, data, utf)) break;
3070            eptr += len;            eptr += len;
3071            }            }
3072          for(;;)          for(;;)
3073            {            {
3074            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3075            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3076            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3077            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3078              if (utf) BACKCHAR(eptr);
3079    #endif
3080            }            }
3081          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3082          }          }
3083    
3084        /* Control never gets here */        /* Control never gets here */
# Line 2715  for (;;) Line 3088  for (;;)
3088      /* Match a single character, casefully */      /* Match a single character, casefully */
3089    
3090      case OP_CHAR:      case OP_CHAR:
3091  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3092      if (utf8)      if (utf)
3093        {        {
3094        length = 1;        length = 1;
3095        ecode++;        ecode++;
# Line 2724  for (;;) Line 3097  for (;;)
3097        if (length > md->end_subject - eptr)        if (length > md->end_subject - eptr)
3098          {          {
3099          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
3100          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3101          }          }
3102        while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);
3103        }        }
3104      else      else
3105  #endif  #endif
3106        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3107        {        {
3108        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3109          {          {
3110          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
3111          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3112          }          }
3113        if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
3114        ecode += 2;        ecode += 2;
3115        }        }
3116      break;      break;
3117    
3118      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3119        subject, give up immediately. */
3120    
3121        case OP_CHARI:
3122        if (eptr >= md->end_subject)
3123          {
3124          SCHECK_PARTIAL();
3125          RRETURN(MATCH_NOMATCH);
3126          }
3127    
3128      case OP_CHARNC:  #ifdef SUPPORT_UTF
3129  #ifdef SUPPORT_UTF8      if (utf)
     if (utf8)  
3130        {        {
3131        length = 1;        length = 1;
3132        ecode++;        ecode++;
3133        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3134    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         MRRETURN(MATCH_NOMATCH);  
         }  
   
3135        /* 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
3136        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3137          fast lookup table. We know that there is at least one byte left in the
3138          subject. */
3139    
3140        if (fc < 128)        if (fc < 128)
3141          {          {
3142          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          if (md->lcc[fc]
3143                != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3144            ecode++;
3145            eptr++;
3146          }          }
3147    
3148        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3149          use the value of "length" to check for sufficient bytes left, because the
3150          other case of the character may have more or fewer bytes.  */
3151    
3152        else        else
3153          {          {
# Line 2783  for (;;) Line 3163  for (;;)
3163  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3164            if (dc != UCD_OTHERCASE(fc))            if (dc != UCD_OTHERCASE(fc))
3165  #endif  #endif
3166              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3167            }            }
3168          }          }
3169        }        }
3170      else      else
3171  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3172    
3173      /* Non-UTF-8 mode */      /* Not UTF mode */
3174        {        {
3175        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3176          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3177          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         MRRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);  
3178        ecode += 2;        ecode += 2;
3179        }        }
3180      break;      break;
# Line 2805  for (;;) Line 3182  for (;;)
3182      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3183    
3184      case OP_EXACT:      case OP_EXACT:
3185        case OP_EXACTI:
3186      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3187      ecode += 3;      ecode += 1 + IMM2_SIZE;
3188      goto REPEATCHAR;      goto REPEATCHAR;
3189    
3190      case OP_POSUPTO:      case OP_POSUPTO:
3191        case OP_POSUPTOI:
3192      possessive = TRUE;      possessive = TRUE;
3193      /* Fall through */      /* Fall through */
3194    
3195      case OP_UPTO:      case OP_UPTO:
3196        case OP_UPTOI:
3197      case OP_MINUPTO:      case OP_MINUPTO:
3198        case OP_MINUPTOI:
3199      min = 0;      min = 0;
3200      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3201      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3202      ecode += 3;      ecode += 1 + IMM2_SIZE;
3203      goto REPEATCHAR;      goto REPEATCHAR;
3204    
3205      case OP_POSSTAR:      case OP_POSSTAR:
3206        case OP_POSSTARI:
3207      possessive = TRUE;      possessive = TRUE;
3208      min = 0;      min = 0;
3209      max = INT_MAX;      max = INT_MAX;
# Line 2829  for (;;) Line 3211  for (;;)
3211      goto REPEATCHAR;      goto REPEATCHAR;
3212    
3213      case OP_POSPLUS:      case OP_POSPLUS:
3214        case OP_POSPLUSI:
3215      possessive = TRUE;      possessive = TRUE;
3216      min = 1;      min = 1;
3217      max = INT_MAX;      max = INT_MAX;
# Line 2836  for (;;) Line 3219  for (;;)
3219      goto REPEATCHAR;      goto REPEATCHAR;
3220    
3221      case OP_POSQUERY:      case OP_POSQUERY:
3222        case OP_POSQUERYI:
3223      possessive = TRUE;      possessive = TRUE;
3224      min = 0;      min = 0;
3225      max = 1;      max = 1;
# Line 2843  for (;;) Line 3227  for (;;)
3227      goto REPEATCHAR;      goto REPEATCHAR;
3228    
3229      case OP_STAR:      case OP_STAR:
3230        case OP_STARI:
3231      case OP_MINSTAR:      case OP_MINSTAR:
3232        case OP_MINSTARI:
3233      case OP_PLUS:      case OP_PLUS:
3234        case OP_PLUSI:
3235      case OP_MINPLUS:      case OP_MINPLUS:
3236        case OP_MINPLUSI:
3237      case OP_QUERY:      case OP_QUERY:
3238        case OP_QUERYI:
3239      case OP_MINQUERY:      case OP_MINQUERY:
3240      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3241        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3242      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3243      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3244      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3245      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2858  for (;;) Line 3247  for (;;)
3247      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3248    
3249      REPEATCHAR:      REPEATCHAR:
3250  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3251      if (utf8)      if (utf)
3252        {        {
3253        length = 1;        length = 1;
3254        charptr = ecode;        charptr = ecode;
# Line 2873  for (;;) Line 3262  for (;;)
3262          {          {
3263  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3264          unsigned int othercase;          unsigned int othercase;
3265          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3266              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3267            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3268          else oclength = 0;          else oclength = 0;
3269  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3270    
3271          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3272            {            {
3273            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3274              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3275  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3276            else if (oclength > 0 &&            else if (oclength > 0 &&
3277                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3278                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3279  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3280            else            else
3281              {              {
3282              CHECK_PARTIAL();              CHECK_PARTIAL();
3283              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3284              }              }
3285            }            }
3286    
# Line 2901  for (;;) Line 3290  for (;;)
3290            {            {
3291            for (fi = min;; fi++)            for (fi = min;; fi++)
3292              {              {
3293              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3294              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3295              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3296              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3297                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3298  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3299              else if (oclength > 0 &&              else if (oclength > 0 &&
3300                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3301                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3302  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3303              else              else
3304                {                {
3305                CHECK_PARTIAL();                CHECK_PARTIAL();
3306                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3307                }                }
3308              }              }
3309            /* Control never gets here */            /* Control never gets here */
# Line 2926  for (;;) Line 3315  for (;;)
3315            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3316              {              {
3317              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3318                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3319  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3320              else if (oclength > 0 &&              else if (oclength > 0 &&
3321                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3322                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3323  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3324              else              else
3325                {                {
# Line 2943  for (;;) Line 3332  for (;;)
3332    
3333            for(;;)            for(;;)
3334              {              {
3335              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3336              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3337              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3338  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3339              eptr--;              eptr--;
3340              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2962  for (;;) Line 3351  for (;;)
3351        value of fc will always be < 128. */        value of fc will always be < 128. */
3352        }        }
3353      else      else
3354  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3355          /* When not in UTF-8 mode, load a single-byte character. */
3356      /* When not in UTF-8 mode, load a single-byte character. */        fc = *ecode++;
   
     fc = *ecode++;  
3357    
3358      /* 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
3359      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
3360      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3361      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3362      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 2980  for (;;) Line 3367  for (;;)
3367      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3368        max, eptr));        max, eptr));
3369    
3370      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3371        {        {
3372        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3373          /* fc must be < 128 if UTF is enabled. */
3374          foc = md->fcc[fc];
3375    #else
3376    #ifdef SUPPORT_UTF
3377    #ifdef SUPPORT_UCP
3378          if (utf && fc > 127)
3379            foc = UCD_OTHERCASE(fc);
3380    #else
3381          if (utf && fc > 127)
3382            foc = fc;
3383    #endif /* SUPPORT_UCP */
3384          else
3385    #endif /* SUPPORT_UTF */
3386            foc = TABLE_GET(fc, md->fcc, fc);
3387    #endif /* COMPILE_PCRE8 */
3388    
3389        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3390          {          {
3391          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3392            {            {
3393            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3394            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3395            }            }
3396          if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3397            eptr++;
3398          }          }
3399        if (min == max) continue;        if (min == max) continue;
3400        if (minimize)        if (minimize)
3401          {          {
3402          for (fi = min;; fi++)          for (fi = min;; fi++)
3403            {            {
3404            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3405            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3406            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3407            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3408              {              {
3409              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3410              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3411              }              }
3412            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3413              eptr++;
3414            }            }
3415          /* Control never gets here */          /* Control never gets here */
3416          }          }
# Line 3019  for (;;) Line 3424  for (;;)
3424              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3425              break;              break;
3426              }              }
3427            if (fc != md->lcc[*eptr]) break;            if (fc != *eptr && foc != *eptr) break;
3428            eptr++;            eptr++;
3429            }            }
3430    
# Line 3027  for (;;) Line 3432  for (;;)
3432    
3433          while (eptr >= pp)          while (eptr >= pp)
3434            {            {
3435            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3436            eptr--;            eptr--;
3437            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3438            }            }
3439          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3440          }          }
3441        /* Control never gets here */        /* Control never gets here */
3442        }        }
# Line 3045  for (;;) Line 3450  for (;;)
3450          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3451            {            {
3452            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3453            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3454            }            }
3455          if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);          if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
3456          }          }
3457    
3458        if (min == max) continue;        if (min == max) continue;
# Line 3056  for (;;) Line 3461  for (;;)
3461          {          {
3462          for (fi = min;; fi++)          for (fi = min;; fi++)
3463            {            {
3464            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3465            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3466            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3467            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3468              {              {
3469              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3470              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3471              }              }
3472            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);            if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
3473            }            }
3474          /* Control never gets here */          /* Control never gets here */
3475          }          }
# Line 3085  for (;;) Line 3490  for (;;)
3490    
3491          while (eptr >= pp)          while (eptr >= pp)
3492            {            {
3493            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3494            eptr--;            eptr--;
3495            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3496            }            }
3497          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3498          }          }
3499        }        }
3500      /* Control never gets here */      /* Control never gets here */
# Line 3098  for (;;) Line 3503  for (;;)
3503      checking can be multibyte. */      checking can be multibyte. */
3504    
3505      case OP_NOT:      case OP_NOT:
3506        case OP_NOTI:
3507      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3508        {        {
3509        SCHECK_PARTIAL();        SCHECK_PARTIAL();
3510        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
3511        }        }
3512      ecode++;      ecode++;
3513      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3514      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3515        {        {
3516  #ifdef SUPPORT_UTF8        register int ch, och;
3517        if (c < 256)        ch = *ecode++;
3518  #endif  #ifdef COMPILE_PCRE8
3519        c = md->lcc[c];        /* ch must be < 128 if UTF is enabled. */
3520        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        och = md->fcc[ch];
3521    #else
3522    #ifdef SUPPORT_UTF
3523    #ifdef SUPPORT_UCP
3524          if (utf && ch > 127)
3525            och = UCD_OTHERCASE(ch);
3526    #else
3527          if (utf && ch > 127)
3528            och = ch;
3529    #endif /* SUPPORT_UCP */
3530          else
3531    #endif /* SUPPORT_UTF */
3532            och = TABLE_GET(ch, md->fcc, ch);
3533    #endif /* COMPILE_PCRE8 */
3534          if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3535        }        }
3536      else      else    /* Caseful */
3537        {        {
3538        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);
3539        }        }
3540      break;      break;
3541    
# Line 3127  for (;;) Line 3547  for (;;)
3547      about... */      about... */
3548    
3549      case OP_NOTEXACT:      case OP_NOTEXACT:
3550        case OP_NOTEXACTI:
3551      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3552      ecode += 3;      ecode += 1 + IMM2_SIZE;
3553      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3554    
3555      case OP_NOTUPTO:      case OP_NOTUPTO:
3556        case OP_NOTUPTOI:
3557      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3558        case OP_NOTMINUPTOI:
3559      min = 0;      min = 0;
3560      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3561      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3562      ecode += 3;      ecode += 1 + IMM2_SIZE;
3563      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3564    
3565      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3566        case OP_NOTPOSSTARI:
3567      possessive = TRUE;      possessive = TRUE;
3568      min = 0;      min = 0;
3569      max = INT_MAX;      max = INT_MAX;
# Line 3147  for (;;) Line 3571  for (;;)
3571      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3572    
3573      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3574        case OP_NOTPOSPLUSI:
3575      possessive = TRUE;      possessive = TRUE;
3576      min = 1;      min = 1;
3577      max = INT_MAX;      max = INT_MAX;
# Line 3154  for (;;) Line 3579  for (;;)
3579      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3580    
3581      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3582        case OP_NOTPOSQUERYI:
3583      possessive = TRUE;      possessive = TRUE;
3584      min = 0;      min = 0;
3585      max = 1;      max = 1;
# Line 3161  for (;;) Line 3587  for (;;)
3587      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3588    
3589      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3590        case OP_NOTPOSUPTOI:
3591      possessive = TRUE;      possessive = TRUE;
3592      min = 0;      min = 0;
3593      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3594      ecode += 3;      ecode += 1 + IMM2_SIZE;
3595      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3596    
3597      case OP_NOTSTAR:      case OP_NOTSTAR:
3598        case OP_NOTSTARI:
3599      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3600        case OP_NOTMINSTARI:
3601      case OP_NOTPLUS:      case OP_NOTPLUS:
3602        case OP_NOTPLUSI:
3603      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3604        case OP_NOTMINPLUSI:
3605      case OP_NOTQUERY:      case OP_NOTQUERY:
3606        case OP_NOTQUERYI:
3607      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3608      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3609        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3610      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3611      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3612      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3195  for (;;) Line 3628  for (;;)
3628      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,
3629        max, eptr));        max, eptr));
3630    
3631      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3632        {        {
3633        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3634          /* fc must be < 128 if UTF is enabled. */
3635          foc = md->fcc[fc];
3636    #else
3637    #ifdef SUPPORT_UTF
3638    #ifdef SUPPORT_UCP
3639          if (utf && fc > 127)
3640            foc = UCD_OTHERCASE(fc);
3641    #else
3642          if (utf && fc > 127)
3643            foc = fc;
3644    #endif /* SUPPORT_UCP */
3645          else
3646    #endif /* SUPPORT_UTF */
3647            foc = TABLE_GET(fc, md->fcc, fc);
3648    #endif /* COMPILE_PCRE8 */
3649    
3650  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3651        /* UTF-8 mode */        if (utf)
       if (utf8)  
3652          {          {
3653          register unsigned int d;          register unsigned int d;
3654          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3209  for (;;) Line 3656  for (;;)
3656            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3657              {              {
3658              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3659              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3660              }              }
3661            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3662            if (d < 256) d = md->lcc[d];            if (fc == d || foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) MRRETURN(MATCH_NOMATCH);  
3663            }            }
3664          }          }
3665        else        else
3666  #endif  #endif
3667          /* Not UTF mode */
       /* Not UTF-8 mode */  
3668          {          {
3669          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3670            {            {
3671            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3672              {              {
3673              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3674              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3675              }              }
3676            if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3677              eptr++;
3678            }            }
3679          }          }
3680    
# Line 3236  for (;;) Line 3682  for (;;)
3682    
3683        if (minimize)        if (minimize)
3684          {          {
3685  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3686          /* UTF-8 mode */          if (utf)
         if (utf8)  
3687            {            {
3688            register unsigned int d;            register unsigned int d;
3689            for (fi = min;; fi++)            for (fi = min;; fi++)
3690              {              {
3691              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3692              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3693              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3694              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3695                {                {
3696                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3697                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3698                }                }
3699              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3700              if (d < 256) d = md->lcc[d];              if (fc == d || foc == d) RRETURN(MATCH_NOMATCH);
             if (fc == d) MRRETURN(MATCH_NOMATCH);  
3701              }              }
3702            }            }
3703          else          else
3704  #endif  #endif
3705          /* Not UTF-8 mode */          /* Not UTF mode */
3706            {            {
3707            for (fi = min;; fi++)            for (fi = min;; fi++)
3708              {              {
3709              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3710              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3711              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3712              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3713                {                {
3714                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3715                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3716                }                }
3717              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);              if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3718                eptr++;
3719              }              }
3720            }            }
3721          /* Control never gets here */          /* Control never gets here */
# Line 3282  for (;;) Line 3727  for (;;)
3727          {          {
3728          pp = eptr;          pp = eptr;
3729    
3730  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3731          /* UTF-8 mode */          if (utf)
         if (utf8)  
3732            {            {
3733            register unsigned int d;            register unsigned int d;
3734            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3296  for (;;) Line 3740  for (;;)
3740                break;                break;
3741                }                }
3742              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3743              if (d < 256) d = md->lcc[d];              if (fc == d || foc == d) break;
             if (fc == d) break;  
3744              eptr += len;              eptr += len;
3745              }              }
3746          if (possessive) continue;            if (possessive) continue;
3747          for(;;)            for(;;)
3748              {              {
3749              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3750              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3751              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3752              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3311  for (;;) Line 3754  for (;;)
3754            }            }
3755          else          else
3756  #endif  #endif
3757          /* Not UTF-8 mode */          /* Not UTF mode */
3758            {            {
3759            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3760              {              {
# Line 3320  for (;;) Line 3763  for (;;)
3763                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3764                break;                break;
3765                }                }
3766              if (fc == md->lcc[*eptr]) break;              if (fc == *eptr || foc == *eptr) break;
3767              eptr++;              eptr++;
3768              }              }
3769            if (possessive) continue;            if (possessive) continue;
3770            while (eptr >= pp)            while (eptr >= pp)
3771              {              {
3772              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3773              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3774              eptr--;              eptr--;
3775              }              }
3776            }            }
3777    
3778          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3779          }          }
3780        /* Control never gets here */        /* Control never gets here */
3781        }        }
# Line 3341  for (;;) Line 3784  for (;;)
3784    
3785      else      else
3786        {        {
3787  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3788        /* UTF-8 mode */        if (utf)
       if (utf8)  
3789          {          {
3790          register unsigned int d;          register unsigned int d;
3791          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3351  for (;;) Line 3793  for (;;)
3793            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3794              {              {
3795              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3796              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3797              }              }
3798            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3799            if (fc == d) MRRETURN(MATCH_NOMATCH);            if (fc == d) RRETURN(MATCH_NOMATCH);
3800            }            }
3801          }          }
3802        else        else
3803  #endif  #endif
3804        /* Not UTF-8 mode */        /* Not UTF mode */
3805          {          {
3806          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3807            {            {
3808            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3809              {              {
3810              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3811              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3812              }              }
3813            if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);
3814            }            }
3815          }          }
3816    
# Line 3376  for (;;) Line 3818  for (;;)
3818    
3819        if (minimize)        if (minimize)
3820          {          {
3821  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3822          /* UTF-8 mode */          if (utf)
         if (utf8)  
3823            {            {
3824            register unsigned int d;            register unsigned int d;
3825            for (fi = min;; fi++)            for (fi = min;; fi++)
3826              {              {
3827              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3828              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3829              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3830              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3831                {                {
3832                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3833                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3834                }   &nb