/[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 243 by ph10, Thu Sep 13 09:28:14 2007 UTC revision 649 by ph10, Mon Aug 1 11:23:52 2011 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2007 University of Cambridge             Copyright (c) 1997-2011 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 57  possible. There are also some static sup Line 57  possible. There are also some static sup
57  #undef min  #undef min
58  #undef max  #undef max
59    
60  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
61    of call to match(). We do it this way to save on using another stack variable,
62    as stack usage is to be discouraged. */
63    
64  #define match_condassert     0x01  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
65  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
66    
67  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
68  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 71  defined PCRE_ERROR_xxx codes, which are Line 73  defined PCRE_ERROR_xxx codes, which are
73  /* Special internal returns from the match() function. Make them sufficiently  /* Special internal returns from the match() function. Make them sufficiently
74  negative to avoid the external error codes. */  negative to avoid the external error codes. */
75    
76  #define MATCH_COMMIT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_PRUNE        (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_SKIP         (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_THEN         (-996)  #define MATCH_ONCE         (-996)
80    #define MATCH_PRUNE        (-995)
81    #define MATCH_SKIP         (-994)
82    #define MATCH_SKIP_ARG     (-993)
83    #define MATCH_THEN         (-992)
84    
85    /* This is a convenience macro for code that occurs many times. */
86    
87    #define MRRETURN(ra) \
88      { \
89      md->mark = markptr; \
90      RRETURN(ra); \
91      }
92    
93  /* 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.
94  If the offset vector is bigger, malloc is used. This should be a multiple of 3,  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
# Line 89  static const char rep_max[] = { 0, 0, 0, Line 103  static const char rep_max[] = { 0, 0, 0,
103    
104    
105    
106  #ifdef DEBUG  #ifdef PCRE_DEBUG
107  /*************************************************  /*************************************************
108  *        Debugging function to print chars       *  *        Debugging function to print chars       *
109  *************************************************/  *************************************************/
# Line 122  while (length-- > 0) Line 136  while (length-- > 0)
136  *          Match a back-reference                *  *          Match a back-reference                *
137  *************************************************/  *************************************************/
138    
139  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
140  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
141    the length passed is zero. Note that in caseless UTF-8 mode, the number of
142    subject bytes matched may be different to the number of reference bytes.
143    
144  Arguments:  Arguments:
145    offset      index into the offset vector    offset      index into the offset vector
146    eptr        points into the subject    eptr        pointer into the subject
147    length      length to be matched    length      length of reference to be matched (number of bytes)
148    md          points to match data block    md          points to match data block
149    ims         the ims flags    caseless    TRUE if caseless
150    
151  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
152  */  */
153    
154  static BOOL  static int
155  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
156    unsigned long int ims)    BOOL caseless)
157  {  {
158  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
159    register USPTR p = md->start_subject + md->offset_vector[offset];
160    
161  #ifdef DEBUG  #ifdef PCRE_DEBUG
162  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
163    printf("matching subject <null>");    printf("matching subject <null>");
164  else  else
# Line 154  pchars(p, length, FALSE, md); Line 171  pchars(p, length, FALSE, md);
171  printf("\n");  printf("\n");
172  #endif  #endif
173    
174  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
175    
176  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
177    
178  /* Separate the caselesss case for speed */  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
179    properly if Unicode properties are supported. Otherwise, we can check only
180    ASCII characters. */
181    
182  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
183    {    {
184    while (length-- > 0)  #ifdef SUPPORT_UTF8
185      if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE;  #ifdef SUPPORT_UCP
186      if (md->utf8)
187        {
188        /* Match characters up to the end of the reference. NOTE: the number of
189        bytes matched may differ, because there are some characters whose upper and
190        lower case versions code as different numbers of bytes. For example, U+023A
191        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
192        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
193        the latter. It is important, therefore, to check the length along the
194        reference, not along the subject (earlier code did this wrong). */
195    
196        USPTR endptr = p + length;
197        while (p < endptr)
198          {
199          int c, d;
200          if (eptr >= md->end_subject) return -1;
201          GETCHARINC(c, eptr);
202          GETCHARINC(d, p);
203          if (c != d && c != UCD_OTHERCASE(d)) return -1;
204          }
205        }
206      else
207    #endif
208    #endif
209    
210      /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
211      is no UCP support. */
212        {
213        if (eptr + length > md->end_subject) return -1;
214        while (length-- > 0)
215          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
216        }
217    }    }
218    
219    /* In the caseful case, we can just compare the bytes, whether or not we
220    are in UTF-8 mode. */
221    
222  else  else
223    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
224      if (eptr + length > md->end_subject) return -1;
225      while (length-- > 0) if (*p++ != *eptr++) return -1;
226      }
227    
228  return TRUE;  return eptr - eptr_start;
229  }  }
230    
231    
# Line 219  enum { RM1=1, RM2, RM3, RM4, RM5, RM Line 276  enum { RM1=1, RM2, RM3, RM4, RM5, RM
276         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
277         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
278         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
279         RM51,  RM52, RM53, RM54 };         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280           RM61,  RM62, RM63 };
281    
282  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
283  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
284  actuall used in this definition. */  actually used in this definition. */
285    
286  #ifndef NO_RECURSE  #ifndef NO_RECURSE
287  #define REGISTER register  #define REGISTER register
288    
289  #ifdef DEBUG  #ifdef PCRE_DEBUG
290  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
291    { \    { \
292    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
293    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
294    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
295    }    }
296  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 241  actuall used in this definition. */ Line 299  actuall used in this definition. */
299    return ra; \    return ra; \
300    }    }
301  #else  #else
302  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
303    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
304  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
305  #endif  #endif
306    
# Line 255  argument of match(), which never changes Line 313  argument of match(), which never changes
313    
314  #define REGISTER  #define REGISTER
315    
316  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
317    {\    {\
318    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
319      if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
320    frame->Xwhere = rw; \    frame->Xwhere = rw; \
321    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
322    newframe->Xecode = rb;\    newframe->Xecode = rb;\
323    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
324      newframe->Xmarkptr = markptr;\
325    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
326    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
327    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
328    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
329    frame = newframe;\    frame = newframe;\
# Line 277  argument of match(), which never changes Line 335  argument of match(), which never changes
335    
336  #define RRETURN(ra)\  #define RRETURN(ra)\
337    {\    {\
338    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
339    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
340    (pcre_stack_free)(newframe);\    (pcre_stack_free)(oldframe);\
341    if (frame != NULL)\    if (frame != NULL)\
342      {\      {\
343      rrc = ra;\      rrc = ra;\
# Line 296  typedef struct heapframe { Line 354  typedef struct heapframe {
354    
355    /* Function arguments that may change */    /* Function arguments that may change */
356    
357    const uschar *Xeptr;    USPTR Xeptr;
358    const uschar *Xecode;    const uschar *Xecode;
359    const uschar *Xmstart;    USPTR Xmstart;
360      USPTR Xmarkptr;
361    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
362    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
363    unsigned int Xrdepth;    unsigned int Xrdepth;
364    
365    /* Function local variables */    /* Function local variables */
366    
367    const uschar *Xcallpat;    USPTR Xcallpat;
368    const uschar *Xcharptr;  #ifdef SUPPORT_UTF8
369    const uschar *Xdata;    USPTR Xcharptr;
370    const uschar *Xnext;  #endif
371    const uschar *Xpp;    USPTR Xdata;
372    const uschar *Xprev;    USPTR Xnext;
373    const uschar *Xsaved_eptr;    USPTR Xpp;
374      USPTR Xprev;
375      USPTR Xsaved_eptr;
376    
377    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
378    
# Line 321  typedef struct heapframe { Line 380  typedef struct heapframe {
380    BOOL Xcondition;    BOOL Xcondition;
381    BOOL Xprev_is_word;    BOOL Xprev_is_word;
382    
   unsigned long int Xoriginal_ims;  
   
383  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
386    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
387    int Xoclength;    int Xoclength;
388    uschar Xocchars[8];    uschar Xocchars[8];
389  #endif  #endif
390    
391      int Xcodelink;
392    int Xctype;    int Xctype;
393    unsigned int Xfc;    unsigned int Xfc;
394    int Xfi;    int Xfi;
# Line 369  typedef struct heapframe { Line 424  typedef struct heapframe {
424    
425  /* This function is called recursively in many circumstances. Whenever it  /* This function is called recursively in many circumstances. Whenever it
426  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
427  same response.  same response. */
428    
429  Performance note: It might be tempting to extract commonly used fields from the  /* These macros pack up tests that are used for partial matching, and which
430  md structure (e.g. utf8, end_subject) into individual variables to improve  appears several times in the code. We set the "hit end" flag if the pointer is
431    at the end of the subject and also past the start of the subject (i.e.
432    something has been matched). For hard partial matching, we then return
433    immediately. The second one is used when we already know we are past the end of
434    the subject. */
435    
436    #define CHECK_PARTIAL()\
437      if (md->partial != 0 && eptr >= md->end_subject && \
438          eptr > md->start_used_ptr) \
439        { \
440        md->hitend = TRUE; \
441        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
442        }
443    
444    #define SCHECK_PARTIAL()\
445      if (md->partial != 0 && eptr > md->start_used_ptr) \
446        { \
447        md->hitend = TRUE; \
448        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
449        }
450    
451    
452    /* Performance note: It might be tempting to extract commonly used fields from
453    the md structure (e.g. utf8, end_subject) into individual variables to improve
454  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
455  made performance worse.  made performance worse.
456    
# Line 381  Arguments: Line 459  Arguments:
459     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
460     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
461                   by encountering \K)                   by encountering \K)
462       markptr     pointer to the most recent MARK name, or NULL
463     offset_top  current top pointer     offset_top  current top pointer
464     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
465     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
466                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
467     rdepth      the recursion depth     rdepth      the recursion depth
468    
469  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
470                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
471                   a negative MATCH_xxx value for PRUNE, SKIP, etc
472                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
473                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
474  */  */
475    
476  static int  static int
477  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
478    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
479    int flags, unsigned int rdepth)    unsigned int rdepth)
480  {  {
481  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
482  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 413  register unsigned int c; /* Character Line 488  register unsigned int c; /* Character
488  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
489    
490  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
491    BOOL caseless;
492    int condcode;
493    
494  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
495  preserved over calls to RMATCH() are part of a "frame" which is obtained from  preserved over calls to RMATCH() are part of a "frame" which is obtained from
# Line 420  heap storage. Set up the top-level frame Line 497  heap storage. Set up the top-level frame
497  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
498    
499  #ifdef NO_RECURSE  #ifdef NO_RECURSE
500  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
501    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
502  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
503    
504  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 428  frame->Xprevframe = NULL; /* Line 506  frame->Xprevframe = NULL; /*
506  frame->Xeptr = eptr;  frame->Xeptr = eptr;
507  frame->Xecode = ecode;  frame->Xecode = ecode;
508  frame->Xmstart = mstart;  frame->Xmstart = mstart;
509    frame->Xmarkptr = markptr;
510  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
511  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
512  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
513    
514  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 443  HEAP_RECURSE: Line 520  HEAP_RECURSE:
520  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
521  #define ecode              frame->Xecode  #define ecode              frame->Xecode
522  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
523    #define markptr            frame->Xmarkptr
524  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
525  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
526  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
527    
528  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 455  HEAP_RECURSE: Line 531  HEAP_RECURSE:
531  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
532  #endif  #endif
533  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
534    #define codelink           frame->Xcodelink
535  #define data               frame->Xdata  #define data               frame->Xdata
536  #define next               frame->Xnext  #define next               frame->Xnext
537  #define pp                 frame->Xpp  #define pp                 frame->Xpp
# Line 467  HEAP_RECURSE: Line 544  HEAP_RECURSE:
544  #define condition          frame->Xcondition  #define condition          frame->Xcondition
545  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
546    
 #define original_ims       frame->Xoriginal_ims  
   
547  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
548  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
549  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
550  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
 #define prop_category      frame->Xprop_category  
 #define prop_chartype      frame->Xprop_chartype  
 #define prop_script        frame->Xprop_script  
551  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
552  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
553  #endif  #endif
# Line 505  i, and fc and c, can be the same variabl Line 577  i, and fc and c, can be the same variabl
577  #define fi i  #define fi i
578  #define fc c  #define fc c
579    
580    /* Many of the following variables are used only in small blocks of the code.
581    My normal style of coding would have declared them within each of those blocks.
582    However, in order to accommodate the version of this code that uses an external
583    "stack" implemented on the heap, it is easier to declare them all here, so the
584    declarations can be cut out in a block. The only declarations within blocks
585    below are for variables that do not have to be preserved over a recursive call
586    to RMATCH(). */
587    
588  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  #ifdef SUPPORT_UTF8
589  const uschar *charptr;             /* in small blocks of the code. My normal */  const uschar *charptr;
590  #endif                             /* style of coding would have declared    */  #endif
591  const uschar *callpat;             /* them within each of those blocks.      */  const uschar *callpat;
592  const uschar *data;                /* However, in order to accommodate the   */  const uschar *data;
593  const uschar *next;                /* version of this code that uses an      */  const uschar *next;
594  USPTR         pp;                  /* external "stack" implemented on the    */  USPTR         pp;
595  const uschar *prev;                /* heap, it is easier to declare them all */  const uschar *prev;
596  USPTR         saved_eptr;          /* here, so the declarations can be cut   */  USPTR         saved_eptr;
597                                     /* out in a block. The only declarations  */  
598  recursion_info new_recursive;      /* within blocks below are for variables  */  recursion_info new_recursive;
599                                     /* that do not have to be preserved over  */  
600  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  BOOL cur_is_word;
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
 unsigned long int original_ims;  
   
604  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
608  int oclength;  int oclength;
609  uschar occhars[8];  uschar occhars[8];
610  #endif  #endif
611    
612    int codelink;
613  int ctype;  int ctype;
614  int length;  int length;
615  int max;  int max;
# Line 549  int stacksave[REC_STACK_SAVE_MAX]; Line 624  int stacksave[REC_STACK_SAVE_MAX];
624  eptrblock newptrb;  eptrblock newptrb;
625  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
626    
627    /* To save space on the stack and in the heap frame, I have doubled up on some
628    of the local variables that are used only in localised parts of the code, but
629    still need to be preserved over recursive calls of match(). These macros define
630    the alternative names that are used. */
631    
632    #define allow_zero    cur_is_word
633    #define cbegroup      condition
634    #define code_offset   codelink
635    #define condassert    condition
636    #define matched_once  prev_is_word
637    
638  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
639  variables. */  variables. */
640    
# Line 568  TAIL_RECURSE: Line 654  TAIL_RECURSE:
654  /* OK, now we can get on with the real code of the function. Recursive calls  /* OK, now we can get on with the real code of the function. Recursive calls
655  are specified by the macro RMATCH and RRETURN is used to return. When  are specified by the macro RMATCH and RRETURN is used to return. When
656  NO_RECURSE is *not* defined, these just turn into a recursive call to match()  NO_RECURSE is *not* defined, these just turn into a recursive call to match()
657  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
658  defined). However, RMATCH isn't like a function call because it's quite a  defined). However, RMATCH isn't like a function call because it's quite a
659  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,
660  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
# Line 585  haven't exceeded the recursive call limi Line 671  haven't exceeded the recursive call limi
671  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
672  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
673    
 original_ims = ims;    /* Save for resetting on ')' */  
   
674  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
675  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
676  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
677  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
678  When match() is called in other circumstances, don't add to the chain. The  
679  match_cbegroup flag must NOT be used with tail recursion, because the memory  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
680  block that is used is on the stack, so a new one may be required for each  such remembered pointers, to be checked when we hit the closing ket, in order
681  match(). */  to break infinite loops that match no characters. When match() is called in
682    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
683    NOT be used with tail recursion, because the memory block that is used is on
684    the stack, so a new one may be required for each match(). */
685    
686  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
687    {    {
688    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
689    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
690    eptrb = &newptrb;    eptrb = &newptrb;
691      md->match_function_type = 0;
692    }    }
693    
694  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 610  for (;;) Line 698  for (;;)
698    minimize = possessive = FALSE;    minimize = possessive = FALSE;
699    op = *ecode;    op = *ecode;
700    
   /* For partial matching, remember if we ever hit the end of the subject after  
   matching at least one subject character. */  
   
   if (md->partial &&  
       eptr >= md->end_subject &&  
       eptr > mstart)  
     md->hitend = TRUE;  
   
701    switch(op)    switch(op)
702      {      {
703        case OP_MARK:
704        markptr = ecode + 2;
705        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
706          eptrb, RM55);
707    
708        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
709        argument, and we must check whether that argument matches this MARK's
710        argument. It is passed back in md->start_match_ptr (an overloading of that
711        variable). If it does match, we reset that variable to the current subject
712        position and return MATCH_SKIP. Otherwise, pass back the return code
713        unaltered. */
714    
715        if (rrc == MATCH_SKIP_ARG &&
716            strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
717          {
718          md->start_match_ptr = eptr;
719          RRETURN(MATCH_SKIP);
720          }
721    
722        if (md->mark == NULL) md->mark = markptr;
723        RRETURN(rrc);
724    
725      case OP_FAIL:      case OP_FAIL:
726      RRETURN(MATCH_NOMATCH);      MRRETURN(MATCH_NOMATCH);
727    
728        /* COMMIT overrides PRUNE, SKIP, and THEN */
729    
730        case OP_COMMIT:
731        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
732          eptrb, RM52);
733        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
734            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
735            rrc != MATCH_THEN)
736          RRETURN(rrc);
737        MRRETURN(MATCH_COMMIT);
738    
739        /* PRUNE overrides THEN */
740    
741      case OP_PRUNE:      case OP_PRUNE:
742      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
743        ims, eptrb, flags, RM51);        eptrb, RM51);
744      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
745        MRRETURN(MATCH_PRUNE);
746    
747        case OP_PRUNE_ARG:
748        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
749          eptrb, RM56);
750        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
751        md->mark = ecode + 2;
752      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
753    
754      case OP_COMMIT:      /* SKIP overrides PRUNE and THEN */
     RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,  
       ims, eptrb, flags, RM52);  
     if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
     RRETURN(MATCH_COMMIT);  
755    
756      case OP_SKIP:      case OP_SKIP:
757      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
758        ims, eptrb, flags, RM53);        eptrb, RM53);
759      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
760          RRETURN(rrc);
761      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
762      RRETURN(MATCH_SKIP);      MRRETURN(MATCH_SKIP);
763    
764        case OP_SKIP_ARG:
765        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
766          eptrb, RM57);
767        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769    
770        /* Pass back the current skip name by overloading md->start_match_ptr and
771        returning the special MATCH_SKIP_ARG return code. This will either be
772        caught by a matching MARK, or get to the top, where it is treated the same
773        as PRUNE. */
774    
775        md->start_match_ptr = ecode + 2;
776        RRETURN(MATCH_SKIP_ARG);
777    
778        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
779        the alt that is at the start of the current branch. This makes it possible
780        to skip back past alternatives that precede the THEN within the current
781        branch. */
782    
783      case OP_THEN:      case OP_THEN:
784      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
785        ims, eptrb, flags, RM54);        eptrb, RM54);
786        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
787        md->start_match_ptr = ecode - GET(ecode, 1);
788        MRRETURN(MATCH_THEN);
789    
790        case OP_THEN_ARG:
791        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
792          offset_top, md, eptrb, RM58);
793      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
794        md->start_match_ptr = ecode - GET(ecode, 1);
795        md->mark = ecode + LINK_SIZE + 2;
796      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
797    
798      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle a capturing bracket, other than those that are possessive with an
799      the current subject position in the working slot at the top of the vector.      unlimited repeat. If there is space in the offset vector, save the current
800      We mustn't change the current values of the data slot, because they may be      subject position in the working slot at the top of the vector. We mustn't
801      set from a previous iteration of this group, and be referred to by a      change the current values of the data slot, because they may be set from a
802      reference inside the group.      previous iteration of this group, and be referred to by a reference inside
803        the group. A failure to match might occur after the group has succeeded,
804      If the bracket fails to match, we need to restore this value and also the      if something later on doesn't match. For this reason, we need to restore
805      values of the final offsets, in case they were set by a previous iteration      the working value and also the values of the final offsets, in case they
806      of the same bracket.      were set by a previous iteration of the same bracket.
807    
808      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
809      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 667  for (;;) Line 814  for (;;)
814      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
815      offset = number << 1;      offset = number << 1;
816    
817  #ifdef DEBUG  #ifdef PCRE_DEBUG
818      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
819      printf("subject=");      printf("subject=");
820      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 682  for (;;) Line 829  for (;;)
829        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
830    
831        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
832        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
833            (int)(eptr - md->start_subject);
834    
835        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
836          {          {
837            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839            ims, eptrb, flags, RM1);            eptrb, RM1);
840          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841            if (rrc != MATCH_NOMATCH &&
842                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
843              RRETURN(rrc);
844          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
845          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
846            if (*ecode != OP_ALT) break;
847          }          }
       while (*ecode == OP_ALT);  
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855          MATCH_THEN. */
856    
857          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
858          RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
859        }        }
860    
861      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 715  for (;;) Line 869  for (;;)
869      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
871    
872      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
873      final alternative within the brackets, we would return the result of a      repeat. Loop for all the alternatives. When we get to the final alternative
874      recursive call to match() whatever happened. We can reduce stack usage by      within the brackets, we used to return the result of a recursive call to
875      turning this into a tail recursion, except in the case when match_cbegroup      match() whatever happened so it was possible to reduce stack usage by
876      is set.*/      turning this into a tail recursion, except in the case of a possibly empty
877        group. However, now that there is the possiblity of (*THEN) occurring in
878        the final alternative, this optimization is no longer possible.
879    
880        MATCH_ONCE is returned when the end of an atomic group is successfully
881        reached, but subsequent matching fails. It passes back up the tree (causing
882        captured values to be reset) until the original atomic group level is
883        reached. This is tested by comparing md->once_target with the start of the
884        group. At this point, the return is converted into MATCH_NOMATCH so that
885        previous backup points can be taken. */
886    
887        case OP_ONCE:
888      case OP_BRA:      case OP_BRA:
889      case OP_SBRA:      case OP_SBRA:
890      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
891      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
892      for (;;)      for (;;)
893        {        {
894        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
895          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
896            RM2);
897          if (rrc != MATCH_NOMATCH &&
898              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
899          {          {
900          if (flags == 0)    /* Not a possibly empty group */          if (rrc == MATCH_ONCE)
901            {            {
902            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
903            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
904            goto TAIL_RECURSE;              {
905                while (*scode == OP_ALT) scode += GET(scode, 1);
906                scode -= GET(scode, 1);
907                }
908              if (md->once_target == scode) rrc = MATCH_NOMATCH;
909            }            }
910            RRETURN(rrc);
911            }
912          ecode += GET(ecode, 1);
913          if (*ecode != OP_ALT) break;
914          }
915        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
916        RRETURN(MATCH_NOMATCH);
917    
918          /* Possibly empty group; can't use tail recursion. */      /* Handle possessive capturing brackets with an unlimited repeat. We come
919        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
920        handled similarly to the normal case above. However, the matching is
921        different. The end of these brackets will always be OP_KETRPOS, which
922        returns MATCH_KETRPOS without going further in the pattern. By this means
923        we can handle the group by iteration rather than recursion, thereby
924        reducing the amount of stack needed. */
925    
926        case OP_CBRAPOS:
927        case OP_SCBRAPOS:
928        allow_zero = FALSE;
929    
930          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      POSSESSIVE_CAPTURE:
931            eptrb, flags, RM48);      number = GET2(ecode, 1+LINK_SIZE);
932          RRETURN(rrc);      offset = number << 1;
933    
934    #ifdef PCRE_DEBUG
935        printf("start possessive bracket %d\n", number);
936        printf("subject=");
937        pchars(eptr, 16, TRUE, md);
938        printf("\n");
939    #endif
940    
941        if (offset < md->offset_max)
942          {
943          matched_once = FALSE;
944          code_offset = ecode - md->start_code;
945    
946          save_offset1 = md->offset_vector[offset];
947          save_offset2 = md->offset_vector[offset+1];
948          save_offset3 = md->offset_vector[md->offset_end - number];
949          save_capture_last = md->capture_last;
950    
951          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
952    
953          /* Each time round the loop, save the current subject position for use
954          when the group matches. For MATCH_MATCH, the group has matched, so we
955          restart it with a new subject starting position, remembering that we had
956          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
957          usual. If we haven't matched any alternatives in any iteration, check to
958          see if a previous iteration matched. If so, the group has matched;
959          continue from afterwards. Otherwise it has failed; restore the previous
960          capture values before returning NOMATCH. */
961    
962          for (;;)
963            {
964            md->offset_vector[md->offset_end - number] =
965              (int)(eptr - md->start_subject);
966            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
967            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
968              eptrb, RM63);
969            if (rrc == MATCH_KETRPOS)
970              {
971              offset_top = md->end_offset_top;
972              eptr = md->end_match_ptr;
973              ecode = md->start_code + code_offset;
974              save_capture_last = md->capture_last;
975              matched_once = TRUE;
976              continue;
977              }
978            if (rrc != MATCH_NOMATCH &&
979                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
980              RRETURN(rrc);
981            md->capture_last = save_capture_last;
982            ecode += GET(ecode, 1);
983            if (*ecode != OP_ALT) break;
984            }
985    
986          if (!matched_once)
987            {
988            md->offset_vector[offset] = save_offset1;
989            md->offset_vector[offset+1] = save_offset2;
990            md->offset_vector[md->offset_end - number] = save_offset3;
991          }          }
992    
993        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
994        otherwise return. */        if (allow_zero || matched_once)
995            {
996            ecode += 1 + LINK_SIZE;
997            break;
998            }
999    
1000          RRETURN(MATCH_NOMATCH);
1001          }
1002    
1003        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1004        as a non-capturing bracket. */
1005    
1006        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1007        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008    
1009        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1010    
1011        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1012        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1013    
1014        /* Non-capturing possessive bracket with unlimited repeat. We come here
1015        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1016        without the capturing complication. It is written out separately for speed
1017        and cleanliness. */
1018    
1019        case OP_BRAPOS:
1020        case OP_SBRAPOS:
1021        allow_zero = FALSE;
1022    
1023        POSSESSIVE_NON_CAPTURE:
1024        matched_once = FALSE;
1025        code_offset = ecode - md->start_code;
1026    
1027        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      for (;;)
1028          eptrb, flags, RM2);        {
1029        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1030          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1031            eptrb, RM48);
1032          if (rrc == MATCH_KETRPOS)
1033            {
1034            offset_top = md->end_offset_top;
1035            eptr = md->end_match_ptr;
1036            ecode = md->start_code + code_offset;
1037            matched_once = TRUE;
1038            continue;
1039            }
1040          if (rrc != MATCH_NOMATCH &&
1041              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1042            RRETURN(rrc);
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045          }
1046    
1047        if (matched_once || allow_zero)
1048          {
1049          ecode += 1 + LINK_SIZE;
1050          break;
1051        }        }
1052        RRETURN(MATCH_NOMATCH);
1053    
1054      /* Control never reaches here. */      /* Control never reaches here. */
1055    
1056      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1057      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1058      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1059      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1060    
1061      case OP_COND:      case OP_COND:
1062      case OP_SCOND:      case OP_SCOND:
1063      if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */      codelink = GET(ecode, 1);
1064    
1065        /* Because of the way auto-callout works during compile, a callout item is
1066        inserted between OP_COND and an assertion condition. */
1067    
1068        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1069        {        {
1070        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (pcre_callout != NULL)
1071        condition = md->recursive != NULL &&          {
1072          (offset == RREF_ANY || offset == md->recursive->group_num);          pcre_callout_block cb;
1073        ecode += condition? 3 : GET(ecode, 1);          cb.version          = 2;   /* Version 1 of the callout block */
1074            cb.callout_number   = ecode[LINK_SIZE+2];
1075            cb.offset_vector    = md->offset_vector;
1076            cb.subject          = (PCRE_SPTR)md->start_subject;
1077            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1078            cb.start_match      = (int)(mstart - md->start_subject);
1079            cb.current_position = (int)(eptr - md->start_subject);
1080            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1081            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1082            cb.capture_top      = offset_top/2;
1083            cb.capture_last     = md->capture_last;
1084            cb.callout_data     = md->callout_data;
1085            cb.mark             = markptr;
1086            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1087            if (rrc < 0) RRETURN(rrc);
1088            }
1089          ecode += _pcre_OP_lengths[OP_CALLOUT];
1090        }        }
1091    
1092      else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */      condcode = ecode[LINK_SIZE+1];
1093    
1094        /* Now see what the actual condition is */
1095    
1096        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1097          {
1098          if (md->recursive == NULL)                /* Not recursing => FALSE */
1099            {
1100            condition = FALSE;
1101            ecode += GET(ecode, 1);
1102            }
1103          else
1104            {
1105            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1106            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1107    
1108            /* If the test is for recursion into a specific subpattern, and it is
1109            false, but the test was set up by name, scan the table to see if the
1110            name refers to any other numbers, and test them. The condition is true
1111            if any one is set. */
1112    
1113            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1114              {
1115              uschar *slotA = md->name_table;
1116              for (i = 0; i < md->name_count; i++)
1117                {
1118                if (GET2(slotA, 0) == recno) break;
1119                slotA += md->name_entry_size;
1120                }
1121    
1122              /* Found a name for the number - there can be only one; duplicate
1123              names for different numbers are allowed, but not vice versa. First
1124              scan down for duplicates. */
1125    
1126              if (i < md->name_count)
1127                {
1128                uschar *slotB = slotA;
1129                while (slotB > md->name_table)
1130                  {
1131                  slotB -= md->name_entry_size;
1132                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1133                    {
1134                    condition = GET2(slotB, 0) == md->recursive->group_num;
1135                    if (condition) break;
1136                    }
1137                  else break;
1138                  }
1139    
1140                /* Scan up for duplicates */
1141    
1142                if (!condition)
1143                  {
1144                  slotB = slotA;
1145                  for (i++; i < md->name_count; i++)
1146                    {
1147                    slotB += md->name_entry_size;
1148                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1149                      {
1150                      condition = GET2(slotB, 0) == md->recursive->group_num;
1151                      if (condition) break;
1152                      }
1153                    else break;
1154                    }
1155                  }
1156                }
1157              }
1158    
1159            /* Chose branch according to the condition */
1160    
1161            ecode += condition? 3 : GET(ecode, 1);
1162            }
1163          }
1164    
1165        else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1166        {        {
1167        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1168        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1169    
1170          /* If the numbered capture is unset, but the reference was by name,
1171          scan the table to see if the name refers to any other numbers, and test
1172          them. The condition is true if any one is set. This is tediously similar
1173          to the code above, but not close enough to try to amalgamate. */
1174    
1175          if (!condition && condcode == OP_NCREF)
1176            {
1177            int refno = offset >> 1;
1178            uschar *slotA = md->name_table;
1179    
1180            for (i = 0; i < md->name_count; i++)
1181              {
1182              if (GET2(slotA, 0) == refno) break;
1183              slotA += md->name_entry_size;
1184              }
1185    
1186            /* Found a name for the number - there can be only one; duplicate names
1187            for different numbers are allowed, but not vice versa. First scan down
1188            for duplicates. */
1189    
1190            if (i < md->name_count)
1191              {
1192              uschar *slotB = slotA;
1193              while (slotB > md->name_table)
1194                {
1195                slotB -= md->name_entry_size;
1196                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1197                  {
1198                  offset = GET2(slotB, 0) << 1;
1199                  condition = offset < offset_top &&
1200                    md->offset_vector[offset] >= 0;
1201                  if (condition) break;
1202                  }
1203                else break;
1204                }
1205    
1206              /* Scan up for duplicates */
1207    
1208              if (!condition)
1209                {
1210                slotB = slotA;
1211                for (i++; i < md->name_count; i++)
1212                  {
1213                  slotB += md->name_entry_size;
1214                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1215                    {
1216                    offset = GET2(slotB, 0) << 1;
1217                    condition = offset < offset_top &&
1218                      md->offset_vector[offset] >= 0;
1219                    if (condition) break;
1220                    }
1221                  else break;
1222                  }
1223                }
1224              }
1225            }
1226    
1227          /* Chose branch according to the condition */
1228    
1229        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 3 : GET(ecode, 1);
1230        }        }
1231    
1232      else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
1233        {        {
1234        condition = FALSE;        condition = FALSE;
1235        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1236        }        }
1237    
1238      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1239      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1240      assertion. */      an assertion. */
1241    
1242      else      else
1243        {        {
1244        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1245            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1246        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1247          {          {
1248            if (md->end_offset_top > offset_top)
1249              offset_top = md->end_offset_top;  /* Captures may have happened */
1250          condition = TRUE;          condition = TRUE;
1251          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1252          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1253          }          }
1254        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH &&
1255                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1256          {          {
1257          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1258          }          }
1259        else        else
1260          {          {
1261          condition = FALSE;          condition = FALSE;
1262          ecode += GET(ecode, 1);          ecode += codelink;
1263          }          }
1264        }        }
1265    
1266      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one,
1267      we can use tail recursion to avoid using another stack frame, except when      we used to use tail recursion to avoid using another stack frame, except
1268      match_cbegroup is required for an unlimited repeat of a possibly empty      when there was unlimited repeat of a possibly empty group. However, that
1269      group. If the second alternative doesn't exist, we can just plough on. */      strategy no longer works because of the possibilty of (*THEN) being
1270        encountered in the branch. A recursive call to match() is always required,
1271        unless the second alternative doesn't exist, in which case we can just
1272        plough on. */
1273    
1274      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1275        {        {
1276        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1277        if (op == OP_SCOND)        /* Possibly empty group */        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1278          {        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1279          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          rrc = MATCH_NOMATCH;
1280          RRETURN(rrc);        RRETURN(rrc);
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
         goto TAIL_RECURSE;  
         }  
1281        }        }
1282      else                         /* Condition false & no 2nd alternative */      else                         /* Condition false & no alternative */
1283        {        {
1284        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1285        }        }
1286      break;      break;
1287    
1288    
1289      /* End of the pattern, either real or forced. If we are in a top-level      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1290      recursion, we should restore the offsets appropriately and continue from      to close any currently open capturing brackets. */
     after the call. */  
1291    
1292      case OP_ACCEPT:      case OP_CLOSE:
1293      case OP_END:      number = GET2(ecode, 1);
1294      if (md->recursive != NULL && md->recursive->group_num == 0)      offset = number << 1;
1295    
1296    #ifdef PCRE_DEBUG
1297          printf("end bracket %d at *ACCEPT", number);
1298          printf("\n");
1299    #endif
1300    
1301        md->capture_last = number;
1302        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1303        {        {
1304        recursion_info *rec = md->recursive;        md->offset_vector[offset] =
1305        DPRINTF(("End of pattern in a (?0) recursion\n"));          md->offset_vector[md->offset_end - number];
1306        md->recursive = rec->prevrec;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1307        memmove(md->offset_vector, rec->offset_save,        if (offset_top <= offset) offset_top = offset + 2;
         rec->saved_max * sizeof(int));  
       mstart = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
1308        }        }
1309        ecode += 3;
1310        break;
1311    
1312    
1313        /* End of the pattern, either real or forced. */
1314    
1315        case OP_END:
1316        case OP_ACCEPT:
1317        case OP_ASSERT_ACCEPT:
1318    
1319        /* If we have matched an empty string, fail if not in an assertion and not
1320        in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1321        is set and we have matched at the start of the subject. In both cases,
1322        backtracking will then try other alternatives, if any. */
1323    
1324        if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1325             md->recursive == NULL &&
1326             (md->notempty ||
1327               (md->notempty_atstart &&
1328                 mstart == md->start_subject + md->start_offset)))
1329          MRRETURN(MATCH_NOMATCH);
1330    
1331      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty      /* Otherwise, we have a match. */
     string - backtracking will then try other alternatives, if any. */  
1332    
     if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH);  
1333      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1334      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1335      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
     RRETURN(MATCH_MATCH);  
1336    
1337      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1338        given as the argument to MRRETURN when the heap is in use. */
1339    
1340      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1341      ims = ecode[1];      MRRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1342    
1343      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1344      matching won't pass the KET for an assertion. If any one branch matches,      matching won't pass the KET for an assertion. If any one branch matches,
1345      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
1346      start of each branch to move the current point backwards, so the code at      start of each branch to move the current point backwards, so the code at
1347      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1348        of a condition, we want to return immediately afterwards. The caller of
1349        this incarnation of the match() function will have set MATCH_CONDASSERT in
1350        md->match_function type, and one of these opcodes will be the first opcode
1351        that is processed. We use a local variable that is preserved over calls to
1352        match() to remember this case. */
1353    
1354      case OP_ASSERT:      case OP_ASSERT:
1355      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1356        if (md->match_function_type == MATCH_CONDASSERT)
1357          {
1358          condassert = TRUE;
1359          md->match_function_type = 0;
1360          }
1361        else condassert = FALSE;
1362    
1363      do      do
1364        {        {
1365        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1366          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1367        if (rrc == MATCH_MATCH) break;          {
1368        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          mstart = md->start_match_ptr;   /* In case \K reset it */
1369            markptr = md->mark;
1370            break;
1371            }
1372          if (rrc != MATCH_NOMATCH &&
1373              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1374            RRETURN(rrc);
1375        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1376        }        }
1377      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1378      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1379        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1380    
1381      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1382    
1383      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1384    
1385      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1386      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 900  for (;;) Line 1390  for (;;)
1390      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1391      continue;      continue;
1392    
1393      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1394        PRUNE, or COMMIT means we must assume failure without checking subsequent
1395        branches. */
1396    
1397      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1398      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1399        if (md->match_function_type == MATCH_CONDASSERT)
1400          {
1401          condassert = TRUE;
1402          md->match_function_type = 0;
1403          }
1404        else condassert = FALSE;
1405    
1406      do      do
1407        {        {
1408        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1409          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1410        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1411        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          {
1412            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1413            break;
1414            }
1415          if (rrc != MATCH_NOMATCH &&
1416              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1417            RRETURN(rrc);
1418        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1419        }        }
1420      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1421    
1422      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1423    
1424      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1425      continue;      continue;
# Line 932  for (;;) Line 1437  for (;;)
1437        while (i-- > 0)        while (i-- > 0)
1438          {          {
1439          eptr--;          eptr--;
1440          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1441          BACKCHAR(eptr);          BACKCHAR(eptr);
1442          }          }
1443        }        }
# Line 943  for (;;) Line 1448  for (;;)
1448    
1449        {        {
1450        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1451        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1452        }        }
1453    
1454      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1455    
1456        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1457      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1458      break;      break;
1459    
# Line 959  for (;;) Line 1465  for (;;)
1465      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1466        {        {
1467        pcre_callout_block cb;        pcre_callout_block cb;
1468        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1469        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1470        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1471        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1472        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1473        cb.start_match      = mstart - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1474        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1475        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1476        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1477        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1478        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1479        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1480        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        cb.mark             = markptr;
1481          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1482        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1483        }        }
1484      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 981  for (;;) Line 1488  for (;;)
1488      offset data is the offset to the starting bracket from the start of the      offset data is the offset to the starting bracket from the start of the
1489      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1490    
1491      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1492      save their starting points and reinstate them after the recursion. However,      re-instated afterwards. We don't know how many are started and not yet
1493      we don't know how many such there are (offset_top records the completed      finished (offset_top records the completed total) so we just have to save
1494      total) so we just have to save all the potential data. There may be up to      all the potential data. There may be up to 65535 such values, which is too
1495      65535 such values, which is too large to put on the stack, but using malloc      large to put on the stack, but using malloc for small numbers seems
1496      for small numbers seems expensive. As a compromise, the stack is used when      expensive. As a compromise, the stack is used when there are no more than
1497      there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc      REC_STACK_SAVE_MAX values to store; otherwise malloc is used.
     is used. A problem is what to do if the malloc fails ... there is no way of  
     returning to the top level with an error. Save the top REC_STACK_SAVE_MAX  
     values on the stack, and accept that the rest may be wrong.  
1498    
1499      There are also other values that have to be saved. We use a chained      There are also other values that have to be saved. We use a chained
1500      sequence of blocks that actually live on the stack. Thanks to Robin Houston      sequence of blocks that actually live on the stack. Thanks to Robin Houston
1501      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1502        a lot, so he is not to blame for the current way it works. */
1503    
1504      case OP_RECURSE:      case OP_RECURSE:
1505        {        {
1506          recursion_info *ri;
1507          int recno;
1508    
1509        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1510        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1511          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1512    
1513          /* Check for repeating a recursion without advancing the subject pointer.
1514          This should catch convoluted mutual recursions. (Some simple cases are
1515          caught at compile time.) */
1516    
1517          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1518            if (recno == ri->group_num && eptr == ri->subject_position)
1519              RRETURN(PCRE_ERROR_RECURSELOOP);
1520    
1521        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1522    
1523          new_recursive.group_num = recno;
1524          new_recursive.subject_position = eptr;
1525        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1526        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1527    
1528        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1529    
1530        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1531    
1532        /* Now save the offset data. */        /* Now save the offset data */
1533    
1534        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1535        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1023  for (;;) Line 1540  for (;;)
1540            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1541          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1542          }          }
   
1543        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1544              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = mstart;  
       mstart = eptr;  
1545    
1546        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1547        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1548          might be changed, so reset it before looping. */
1549    
1550        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1551        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1552        do        do
1553          {          {
1554            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1555          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1556            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1557          if (rrc == MATCH_MATCH)          memcpy(md->offset_vector, new_recursive.offset_save,
1558                new_recursive.saved_max * sizeof(int));
1559            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1560            {            {
1561            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1562            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1563            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1564              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1565            RRETURN(MATCH_MATCH);  
1566              /* Set where we got to in the subject, and reset the start in case
1567              it was changed by \K. This *is* propagated back out of a recursion,
1568              for Perl compatibility. */
1569    
1570              eptr = md->end_match_ptr;
1571              mstart = md->start_match_ptr;
1572              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1573            }            }
1574          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH &&
1575                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1576            {            {
1577            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1578              if (new_recursive.offset_save != stacksave)
1579                (pcre_free)(new_recursive.offset_save);
1580            RRETURN(rrc);            RRETURN(rrc);
1581            }            }
1582    
1583          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1584          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1585          }          }
1586        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1063  for (;;) Line 1589  for (;;)
1589        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1590        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1591          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1592        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1593        }        }
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH) break;  
       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
1594    
1595      do ecode += GET(ecode, 1); while (*ecode == OP_ALT);      RECURSION_MATCHED:
1596        break;
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
1597    
1598      /* An alternation is the end of a branch; scan along to find the end of the      /* An alternation is the end of a branch; scan along to find the end of the
1599      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1148  for (;;) Line 1602  for (;;)
1602      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1603      break;      break;
1604    
1605      /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1606      that it may occur zero times. It may repeat infinitely, or not at all -      indicating that it may occur zero times. It may repeat infinitely, or not
1607      i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1608      repeat limits are compiled as a number of copies, with the optional ones      with fixed upper repeat limits are compiled as a number of copies, with the
1609      preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1610    
1611      case OP_BRAZERO:      case OP_BRAZERO:
1612        {      next = ecode + 1;
1613        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1614        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1615        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1616        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1617      break;      break;
1618    
1619      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1620        {      next = ecode + 1;
1621        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1622        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1623        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1624        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1625      break;      break;
1626    
1627        case OP_SKIPZERO:
1628        next = ecode+1;
1629        do next += GET(next,1); while (*next == OP_ALT);
1630        ecode = next + 1 + LINK_SIZE;
1631        break;
1632    
1633        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1634        here; just jump to the group, with allow_zero set TRUE. */
1635    
1636        case OP_BRAPOSZERO:
1637        op = *(++ecode);
1638        allow_zero = TRUE;
1639        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1640          goto POSSESSIVE_NON_CAPTURE;
1641    
1642      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1643    
1644      case OP_KET:      case OP_KET:
1645      case OP_KETRMIN:      case OP_KETRMIN:
1646      case OP_KETRMAX:      case OP_KETRMAX:
1647        case OP_KETRPOS:
1648      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1649    
1650      /* If this was a group that remembered the subject start, in order to break      /* If this was a group that remembered the subject start, in order to break
1651      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1652      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1653    
1654      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1655        {        {
1656        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1657        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
# Line 1194  for (;;) Line 1660  for (;;)
1660    
1661      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group, stop matching and return
1662      MATCH_MATCH, but record the current high water mark for use by positive      MATCH_MATCH, but record the current high water mark for use by positive
1663      assertions. Do this also for the "once" (atomic) groups. */      assertions. We also need to record the match start in case it was changed
1664        by \K. */
1665    
1666      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1667          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1668        {        {
1669        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1670        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1671        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1672          MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1673        }        }
1674    
1675      /* For capturing groups we have to check the group number back at the start      /* For capturing groups we have to check the group number back at the start
1676      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1677      bumping the high water mark. Note that whole-pattern recursion is coded as      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1678      a recurse into group 0, so it won't be picked up here. Instead, we catch it      into group 0, so it won't be picked up here. Instead, we catch it when the
1679      when the OP_END is reached. Other recursion is handled here. */      OP_END is reached. Other recursion is handled here. We just have to record
1680        the current subject position and start match pointer and give a MATCH
1681        return. */
1682    
1683      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1684            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1685        {        {
1686        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1687        offset = number << 1;        offset = number << 1;
1688    
1689  #ifdef DEBUG  #ifdef PCRE_DEBUG
1690        printf("end bracket %d", number);        printf("end bracket %d", number);
1691        printf("\n");        printf("\n");
1692  #endif  #endif
1693    
1694          /* Handle a recursively called group. */
1695    
1696          if (md->recursive != NULL && md->recursive->group_num == number)
1697            {
1698            md->end_match_ptr = eptr;
1699            md->start_match_ptr = mstart;
1700            RRETURN(MATCH_MATCH);
1701            }
1702    
1703          /* Deal with capturing */
1704    
1705        md->capture_last = number;        md->capture_last = number;
1706        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1707          {          {
1708            /* If offset is greater than offset_top, it means that we are
1709            "skipping" a capturing group, and that group's offsets must be marked
1710            unset. In earlier versions of PCRE, all the offsets were unset at the
1711            start of matching, but this doesn't work because atomic groups and
1712            assertions can cause a value to be set that should later be unset.
1713            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1714            part of the atomic group, but this is not on the final matching path,
1715            so must be unset when 2 is set. (If there is no group 2, there is no
1716            problem, because offset_top will then be 2, indicating no capture.) */
1717    
1718            if (offset > offset_top)
1719              {
1720              register int *iptr = md->offset_vector + offset_top;
1721              register int *iend = md->offset_vector + offset;
1722              while (iptr < iend) *iptr++ = -1;
1723              }
1724    
1725            /* Now make the extraction */
1726    
1727          md->offset_vector[offset] =          md->offset_vector[offset] =
1728            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1729          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1730          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1731          }          }
1732          }
1733    
1734        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1735        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1736        group. This is the forcible breaking of infinite loops as implemented in
1737        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1738        processing the rest of the pattern at a lower level. If this results in a
1739        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1740        bypassing intermediate backup points, but resetting any captures that
1741        happened along the way. */
1742    
1743        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1744          {
1745          if (*prev == OP_ONCE)
1746          {          {
1747          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1748          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1749          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1750          mstart = rec->save_start;          RRETURN(MATCH_ONCE);
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1751          }          }
1752          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1753          break;
1754        }        }
1755    
1756      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1757      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1758        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
1759    
1760      /* For a non-repeating ket, just continue at this level. This also      if (*ecode == OP_KETRPOS)
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
1761        {        {
1762        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1763        break;        md->end_offset_top = offset_top;
1764          RRETURN(MATCH_KETRPOS);
1765        }        }
1766    
1767      /* The repeating kets try the rest of the pattern or restart from the      /* The normal repeating kets try the rest of the pattern or restart from
1768      preceding bracket, in the appropriate order. In the second case, we can use      the preceding bracket, in the appropriate order. In the second case, we can
1769      tail recursion to avoid using another stack frame, unless we have an      use tail recursion to avoid using another stack frame, unless we have an
1770      unlimited repeat of a group that can match an empty string. */      an atomic group or an unlimited repeat of a group that can match an empty
1771        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1772    
1773      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1774        {        {
1775        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1776        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1777        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1778          {          {
1779          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1780            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1781            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1782            RRETURN(MATCH_ONCE);
1783            }
1784          if (*prev >= OP_SBRA)    /* Could match an empty string */
1785            {
1786            md->match_function_type = MATCH_CBEGROUP;
1787            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1788          RRETURN(rrc);          RRETURN(rrc);
1789          }          }
1790        ecode = prev;        ecode = prev;
# Line 1286  for (;;) Line 1792  for (;;)
1792        }        }
1793      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1794        {        {
1795        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1796          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1797          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1798        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1799          if (*prev == OP_ONCE)
1800            {
1801            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1802            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1803            md->once_target = prev;
1804            RRETURN(MATCH_ONCE);
1805            }
1806        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1807        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1808        }        }
1809      /* Control never gets here */      /* Control never gets here */
1810    
1811      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1812    
1813      case OP_CIRC:      case OP_CIRC:
1814      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
     if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         RRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
1815    
1816      /* Start of subject assertion */      /* Start of subject assertion */
1817    
1818      case OP_SOD:      case OP_SOD:
1819      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1820        ecode++;
1821        break;
1822    
1823        /* Multiline mode: start of subject unless notbol, or after any newline. */
1824    
1825        case OP_CIRCM:
1826        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1827        if (eptr != md->start_subject &&
1828            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1829          MRRETURN(MATCH_NOMATCH);
1830      ecode++;      ecode++;
1831      break;      break;
1832    
1833      /* Start of match assertion */      /* Start of match assertion */
1834    
1835      case OP_SOM:      case OP_SOM:
1836      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1837      ecode++;      ecode++;
1838      break;      break;
1839    
# Line 1329  for (;;) Line 1844  for (;;)
1844      ecode++;      ecode++;
1845      break;      break;
1846    
1847      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1848      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1849    
1850      case OP_DOLL:      case OP_DOLLM:
1851      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1852        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
       if (eptr < md->end_subject)  
         { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }  
       else  
         { if (md->noteol) RRETURN(MATCH_NOMATCH); }  
       ecode++;  
       break;  
       }  
1853      else      else
1854        {        {
1855        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1856        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1857        }        }
1858        ecode++;
1859        break;
1860    
1861        /* Not multiline mode: assert before a terminating newline or before end of
1862        subject unless noteol is set. */
1863    
1864        case OP_DOLL:
1865        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1866        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1867    
1868      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1869    
1870      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1871    
1872      case OP_EOD:      case OP_EOD:
1873      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1874        SCHECK_PARTIAL();
1875      ecode++;      ecode++;
1876      break;      break;
1877    
1878      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1879    
1880      case OP_EODN:      case OP_EODN:
1881      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1882        if (eptr < md->end_subject &&
1883          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1884        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1885    
1886        /* Either at end of string or \n before end. */
1887    
1888        SCHECK_PARTIAL();
1889      ecode++;      ecode++;
1890      break;      break;
1891    
# Line 1378  for (;;) Line 1895  for (;;)
1895      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1896        {        {
1897    
1898        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1899        It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to        It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to
1900        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1901          partial matching. */
1902    
1903    #ifdef SUPPORT_UTF8
1904          if (utf8)
1905            {
1906            /* Get status of previous character */
1907    
1908            if (eptr == md->start_subject) prev_is_word = FALSE; else
1909              {
1910              USPTR lastptr = eptr - 1;
1911              while((*lastptr & 0xc0) == 0x80) lastptr--;
1912              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1913              GETCHAR(c, lastptr);
1914    #ifdef SUPPORT_UCP
1915              if (md->use_ucp)
1916                {
1917                if (c == '_') prev_is_word = TRUE; else
1918                  {
1919                  int cat = UCD_CATEGORY(c);
1920                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1921                  }
1922                }
1923              else
1924    #endif
1925              prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1926              }
1927    
1928            /* Get status of next character */
1929    
1930  #ifdef SUPPORT_UTF8          if (eptr >= md->end_subject)
       if (utf8)  
         {  
         if (eptr == md->start_subject) prev_is_word = FALSE; else  
1931            {            {
1932            const uschar *lastptr = eptr - 1;            SCHECK_PARTIAL();
1933            while((*lastptr & 0xc0) == 0x80) lastptr--;            cur_is_word = FALSE;
           GETCHAR(c, lastptr);  
           prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;  
1934            }            }
1935          if (eptr >= md->end_subject) cur_is_word = FALSE; else          else
1936            {            {
1937            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1938    #ifdef SUPPORT_UCP
1939              if (md->use_ucp)
1940                {
1941                if (c == '_') cur_is_word = TRUE; else
1942                  {
1943                  int cat = UCD_CATEGORY(c);
1944                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1945                  }
1946                }
1947              else
1948    #endif
1949            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1950            }            }
1951          }          }
1952        else        else
1953  #endif  #endif
1954    
1955        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1956          consistency with the behaviour of \w we do use it in this case. */
1957    
1958          {          {
1959          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
1960            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
1961          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
1962            ((md->ctypes[*eptr] & ctype_word) != 0);            {
1963              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1964    #ifdef SUPPORT_UCP
1965              if (md->use_ucp)
1966                {
1967                c = eptr[-1];
1968                if (c == '_') prev_is_word = TRUE; else
1969                  {
1970                  int cat = UCD_CATEGORY(c);
1971                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1972                  }
1973                }
1974              else
1975    #endif
1976              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1977              }
1978    
1979            /* Get status of next character */
1980    
1981            if (eptr >= md->end_subject)
1982              {
1983              SCHECK_PARTIAL();
1984              cur_is_word = FALSE;
1985              }
1986            else
1987    #ifdef SUPPORT_UCP
1988            if (md->use_ucp)
1989              {
1990              c = *eptr;
1991              if (c == '_') cur_is_word = TRUE; else
1992                {
1993                int cat = UCD_CATEGORY(c);
1994                cur_is_word = (cat == ucp_L || cat == ucp_N);
1995                }
1996              }
1997            else
1998    #endif
1999            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2000          }          }
2001    
2002        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2003    
2004        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2005             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2006          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2007        }        }
2008      break;      break;
2009    
2010      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2011    
2012      case OP_ANY:      case OP_ANY:
2013      if ((ims & PCRE_DOTALL) == 0)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2014        {      /* Fall through */
2015        if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);  
2016        case OP_ALLANY:
2017        if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2018          {                            /* not be updated before SCHECK_PARTIAL. */
2019          SCHECK_PARTIAL();
2020          MRRETURN(MATCH_NOMATCH);
2021        }        }
2022      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      eptr++;
2023      if (utf8)      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
2024      ecode++;      ecode++;
2025      break;      break;
2026    
# Line 1435  for (;;) Line 2028  for (;;)
2028      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2029    
2030      case OP_ANYBYTE:      case OP_ANYBYTE:
2031      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2032          {                            /* not be updated before SCHECK_PARTIAL. */
2033          SCHECK_PARTIAL();
2034          MRRETURN(MATCH_NOMATCH);
2035          }
2036        eptr++;
2037      ecode++;      ecode++;
2038      break;      break;
2039    
2040      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2041      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2042          {
2043          SCHECK_PARTIAL();
2044          MRRETURN(MATCH_NOMATCH);
2045          }
2046      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2047      if (      if (
2048  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1448  for (;;) Line 2050  for (;;)
2050  #endif  #endif
2051         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2052         )         )
2053        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2054      ecode++;      ecode++;
2055      break;      break;
2056    
2057      case OP_DIGIT:      case OP_DIGIT:
2058      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2059          {
2060          SCHECK_PARTIAL();
2061          MRRETURN(MATCH_NOMATCH);
2062          }
2063      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2064      if (      if (
2065  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1461  for (;;) Line 2067  for (;;)
2067  #endif  #endif
2068         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2069         )         )
2070        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2071      ecode++;      ecode++;
2072      break;      break;
2073    
2074      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2075      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2076          {
2077          SCHECK_PARTIAL();
2078          MRRETURN(MATCH_NOMATCH);
2079          }
2080      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2081      if (      if (
2082  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1474  for (;;) Line 2084  for (;;)
2084  #endif  #endif
2085         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2086         )         )
2087        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2088      ecode++;      ecode++;
2089      break;      break;
2090    
2091      case OP_WHITESPACE:      case OP_WHITESPACE:
2092      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2093          {
2094          SCHECK_PARTIAL();
2095          MRRETURN(MATCH_NOMATCH);
2096          }
2097      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2098      if (      if (
2099  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1487  for (;;) Line 2101  for (;;)
2101  #endif  #endif
2102         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2103         )         )
2104        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2105      ecode++;      ecode++;
2106      break;      break;
2107    
2108      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2109      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2110          {
2111          SCHECK_PARTIAL();
2112          MRRETURN(MATCH_NOMATCH);
2113          }
2114      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2115      if (      if (
2116  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1500  for (;;) Line 2118  for (;;)
2118  #endif  #endif
2119         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2120         )         )
2121        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2122      ecode++;      ecode++;
2123      break;      break;
2124    
2125      case OP_WORDCHAR:      case OP_WORDCHAR:
2126      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2127          {
2128          SCHECK_PARTIAL();
2129          MRRETURN(MATCH_NOMATCH);
2130          }
2131      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2132      if (      if (
2133  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1513  for (;;) Line 2135  for (;;)
2135  #endif  #endif
2136         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2137         )         )
2138        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2139      ecode++;      ecode++;
2140      break;      break;
2141    
2142      case OP_ANYNL:      case OP_ANYNL:
2143      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2144          {
2145          SCHECK_PARTIAL();
2146          MRRETURN(MATCH_NOMATCH);
2147          }
2148      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2149      switch(c)      switch(c)
2150        {        {
2151        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2152    
2153        case 0x000d:        case 0x000d:
2154        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2155        break;        break;
# Line 1535  for (;;) Line 2162  for (;;)
2162        case 0x0085:        case 0x0085:
2163        case 0x2028:        case 0x2028:
2164        case 0x2029:        case 0x2029:
2165        if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);        if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2166        break;        break;
2167        }        }
2168      ecode++;      ecode++;
2169      break;      break;
2170    
2171      case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
2172      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2173          {
2174          SCHECK_PARTIAL();
2175          MRRETURN(MATCH_NOMATCH);
2176          }
2177      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2178      switch(c)      switch(c)
2179        {        {
# Line 1566  for (;;) Line 2197  for (;;)
2197        case 0x202f:    /* NARROW NO-BREAK SPACE */        case 0x202f:    /* NARROW NO-BREAK SPACE */
2198        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2199        case 0x3000:    /* IDEOGRAPHIC SPACE */        case 0x3000:    /* IDEOGRAPHIC SPACE */
2200        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2201        }        }
2202      ecode++;      ecode++;
2203      break;      break;
2204    
2205      case OP_HSPACE:      case OP_HSPACE:
2206      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2207          {
2208          SCHECK_PARTIAL();
2209          MRRETURN(MATCH_NOMATCH);
2210          }
2211      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2212      switch(c)      switch(c)
2213        {        {
2214        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2215        case 0x09:      /* HT */        case 0x09:      /* HT */
2216        case 0x20:      /* SPACE */        case 0x20:      /* SPACE */
2217        case 0xa0:      /* NBSP */        case 0xa0:      /* NBSP */
# Line 1602  for (;;) Line 2237  for (;;)
2237      break;      break;
2238    
2239      case OP_NOT_VSPACE:      case OP_NOT_VSPACE:
2240      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2241          {
2242          SCHECK_PARTIAL();
2243          MRRETURN(MATCH_NOMATCH);
2244          }
2245      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2246      switch(c)      switch(c)
2247        {        {
# Line 1614  for (;;) Line 2253  for (;;)
2253        case 0x85:      /* NEL */        case 0x85:      /* NEL */
2254        case 0x2028:    /* LINE SEPARATOR */        case 0x2028:    /* LINE SEPARATOR */
2255        case 0x2029:    /* PARAGRAPH SEPARATOR */        case 0x2029:    /* PARAGRAPH SEPARATOR */
2256        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2257        }        }
2258      ecode++;      ecode++;
2259      break;      break;
2260    
2261      case OP_VSPACE:      case OP_VSPACE:
2262      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2263          {
2264          SCHECK_PARTIAL();
2265          MRRETURN(MATCH_NOMATCH);
2266          }
2267      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2268      switch(c)      switch(c)
2269        {        {
2270        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2271        case 0x0a:      /* LF */        case 0x0a:      /* LF */
2272        case 0x0b:      /* VT */        case 0x0b:      /* VT */
2273        case 0x0c:      /* FF */        case 0x0c:      /* FF */
# Line 1643  for (;;) Line 2286  for (;;)
2286    
2287      case OP_PROP:      case OP_PROP:
2288      case OP_NOTPROP:      case OP_NOTPROP:
2289      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2290          {
2291          SCHECK_PARTIAL();
2292          MRRETURN(MATCH_NOMATCH);
2293          }
2294      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2295        {        {
2296        int chartype, script;        const ucd_record *prop = GET_UCD(c);
       int category = _pcre_ucp_findprop(c, &chartype, &script);  
2297    
2298        switch(ecode[1])        switch(ecode[1])
2299          {          {
2300          case PT_ANY:          case PT_ANY:
2301          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2302          break;          break;
2303    
2304          case PT_LAMP:          case PT_LAMP:
2305          if ((chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2306               chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2307               chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2308            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2309           break;          break;
2310    
2311          case PT_GC:          case PT_GC:
2312          if ((ecode[2] != category) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2313            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2314          break;          break;
2315    
2316          case PT_PC:          case PT_PC:
2317          if ((ecode[2] != chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2318            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2319          break;          break;
2320    
2321          case PT_SC:          case PT_SC:
2322          if ((ecode[2] != script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2323            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2324          break;          break;
2325    
2326            /* These are specials */
2327    
2328            case PT_ALNUM:
2329            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2330                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2331              MRRETURN(MATCH_NOMATCH);
2332            break;
2333    
2334            case PT_SPACE:    /* Perl space */
2335            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2336                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2337                   == (op == OP_NOTPROP))
2338              MRRETURN(MATCH_NOMATCH);
2339            break;
2340    
2341            case PT_PXSPACE:  /* POSIX space */
2342            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2343                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2344                 c == CHAR_FF || c == CHAR_CR)
2345                   == (op == OP_NOTPROP))
2346              MRRETURN(MATCH_NOMATCH);
2347            break;
2348    
2349            case PT_WORD:
2350            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2351                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2352                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2353              MRRETURN(MATCH_NOMATCH);
2354            break;
2355    
2356            /* This should never occur */
2357    
2358          default:          default:
2359          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2360          }          }
# Line 1689  for (;;) Line 2367  for (;;)
2367      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2368    
2369      case OP_EXTUNI:      case OP_EXTUNI:
2370      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2371          {
2372          SCHECK_PARTIAL();
2373          MRRETURN(MATCH_NOMATCH);
2374          }
2375      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2376        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2377        while (eptr < md->end_subject)
2378        {        {
2379        int chartype, script;        int len = 1;
2380        int category = _pcre_ucp_findprop(c, &chartype, &script);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2381        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (UCD_CATEGORY(c) != ucp_M) break;
2382        while (eptr < md->end_subject)        eptr += len;
         {  
         int len = 1;  
         if (!utf8) c = *eptr; else  
           {  
           GETCHARLEN(c, eptr, len);  
           }  
         category = _pcre_ucp_findprop(c, &chartype, &script);  
         if (category != ucp_M) break;  
         eptr += len;  
         }  
2383        }        }
2384      ecode++;      ecode++;
2385      break;      break;
# Line 1721  for (;;) Line 2395  for (;;)
2395      loops). */      loops). */
2396    
2397      case OP_REF:      case OP_REF:
2398        {      case OP_REFI:
2399        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2400        ecode += 3;                                 /* Advance past item */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2401        ecode += 3;
2402    
2403        /* If the reference is unset, set the length to be longer than the amount      /* If the reference is unset, there are two possibilities:
       of subject left; this ensures that every attempt at a match fails. We  
       can't just fail here, because of the possibility of quantifiers with zero  
       minima. */  
   
       length = (offset >= offset_top || md->offset_vector[offset] < 0)?  
         md->end_subject - eptr + 1 :  
         md->offset_vector[offset+1] - md->offset_vector[offset];  
2404    
2405        /* Set up for repetition, or handle the non-repeated case */      (a) In the default, Perl-compatible state, set the length negative;
2406        this ensures that every attempt at a match fails. We can't just fail
2407        here, because of the possibility of quantifiers with zero minima.
2408    
2409        switch (*ecode)      (b) If the JavaScript compatibility flag is set, set the length to zero
2410          {      so that the back reference matches an empty string.
         case OP_CRSTAR:  
         case OP_CRMINSTAR:  
         case OP_CRPLUS:  
         case OP_CRMINPLUS:  
         case OP_CRQUERY:  
         case OP_CRMINQUERY:  
         c = *ecode++ - OP_CRSTAR;  
         minimize = (c & 1) != 0;  
         min = rep_min[c];                 /* Pick up values from tables; */  
         max = rep_max[c];                 /* zero for max => infinity */  
         if (max == 0) max = INT_MAX;  
         break;  
2411    
2412          case OP_CRRANGE:      Otherwise, set the length to the length of what was matched by the
2413          case OP_CRMINRANGE:      referenced subpattern. */
         minimize = (*ecode == OP_CRMINRANGE);  
         min = GET2(ecode, 1);  
         max = GET2(ecode, 3);  
         if (max == 0) max = INT_MAX;  
         ecode += 5;  
         break;  
2414    
2415          default:               /* No repeat follows */      if (offset >= offset_top || md->offset_vector[offset] < 0)
2416          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);        length = (md->jscript_compat)? 0 : -1;
2417          eptr += length;      else
2418          continue;              /* With the main loop */        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2419    
2420        /* Set up for repetition, or handle the non-repeated case */
2421    
2422        switch (*ecode)
2423          {
2424          case OP_CRSTAR:
2425          case OP_CRMINSTAR:
2426          case OP_CRPLUS:
2427          case OP_CRMINPLUS:
2428          case OP_CRQUERY:
2429          case OP_CRMINQUERY:
2430          c = *ecode++ - OP_CRSTAR;
2431          minimize = (c & 1) != 0;
2432          min = rep_min[c];                 /* Pick up values from tables; */
2433          max = rep_max[c];                 /* zero for max => infinity */
2434          if (max == 0) max = INT_MAX;
2435          break;
2436    
2437          case OP_CRRANGE:
2438          case OP_CRMINRANGE:
2439          minimize = (*ecode == OP_CRMINRANGE);
2440          min = GET2(ecode, 1);
2441          max = GET2(ecode, 3);
2442          if (max == 0) max = INT_MAX;
2443          ecode += 5;
2444          break;
2445    
2446          default:               /* No repeat follows */
2447          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2448            {
2449            CHECK_PARTIAL();
2450            MRRETURN(MATCH_NOMATCH);
2451          }          }
2452          eptr += length;
2453          continue;              /* With the main loop */
2454          }
2455    
2456        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2457        main loop. */      zero, just continue with the main loop. */
2458    
2459        if (length == 0) continue;      if (length == 0) continue;
2460    
2461        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2462        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2463        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2464    
2465        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2466          {
2467          int slength;
2468          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2469          {          {
2470          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2471          eptr += length;          MRRETURN(MATCH_NOMATCH);
2472          }          }
2473          eptr += slength;
2474          }
2475    
2476        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2477        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2478    
2479        if (min == max) continue;      if (min == max) continue;
2480    
2481        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2482    
2483        if (minimize)      if (minimize)
2484          {
2485          for (fi = min;; fi++)
2486          {          {
2487          for (fi = min;; fi++)          int slength;
2488            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2489            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2490            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2491            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2492            {            {
2493            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2494            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2495            }            }
2496          /* Control never gets here */          eptr += slength;
2497          }          }
2498          /* Control never gets here */
2499          }
2500    
2501        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2502    
2503        else      else
2504          {
2505          pp = eptr;
2506          for (i = min; i < max; i++)
2507          {          {
2508          pp = eptr;          int slength;
2509          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
           {  
           if (!match_ref(offset, eptr, length, md, ims)) break;  
           eptr += length;  
           }  
         while (eptr >= pp)  
2510            {            {
2511            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2512            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2513            }            }
2514          RRETURN(MATCH_NOMATCH);          eptr += slength;
2515            }
2516          while (eptr >= pp)
2517            {
2518            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2519            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2520            eptr -= length;
2521          }          }
2522          MRRETURN(MATCH_NOMATCH);
2523        }        }
2524      /* Control never gets here */      /* Control never gets here */
2525    
   
   
2526      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2527      used when all the characters in the class have values in the range 0-255,      used when all the characters in the class have values in the range 0-255,
2528      and either the matching is caseful, or the characters are in the range      and either the matching is caseful, or the characters are in the range
# Line 1878  for (;;) Line 2577  for (;;)
2577          {          {
2578          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2579            {            {
2580            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2581                {
2582                SCHECK_PARTIAL();
2583                MRRETURN(MATCH_NOMATCH);
2584                }
2585            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2586            if (c > 255)            if (c > 255)
2587              {              {
2588              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2589              }              }
2590            else            else
2591              {              {
2592              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2593              }              }
2594            }            }
2595          }          }
# Line 1896  for (;;) Line 2599  for (;;)
2599          {          {
2600          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2601            {            {
2602            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2603                {
2604                SCHECK_PARTIAL();
2605                MRRETURN(MATCH_NOMATCH);
2606                }
2607            c = *eptr++;            c = *eptr++;
2608            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2609            }            }
2610          }          }
2611    
# Line 1918  for (;;) Line 2625  for (;;)
2625            {            {
2626            for (fi = min;; fi++)            for (fi = min;; fi++)
2627              {              {
2628              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2629              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2630              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2631                if (eptr >= md->end_subject)
2632                  {
2633                  SCHECK_PARTIAL();
2634                  MRRETURN(MATCH_NOMATCH);
2635                  }
2636              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2637              if (c > 255)              if (c > 255)
2638                {                {
2639                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2640                }                }
2641              else              else
2642                {                {
2643                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2644                }                }
2645              }              }
2646            }            }
# Line 1938  for (;;) Line 2650  for (;;)
2650            {            {
2651            for (fi = min;; fi++)            for (fi = min;; fi++)
2652              {              {
2653              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2654              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2655              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2656                if (eptr >= md->end_subject)
2657                  {
2658                  SCHECK_PARTIAL();
2659                  MRRETURN(MATCH_NOMATCH);
2660                  }
2661              c = *eptr++;              c = *eptr++;
2662              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2663              }              }
2664            }            }
2665          /* Control never gets here */          /* Control never gets here */
# Line 1961  for (;;) Line 2678  for (;;)
2678            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2679              {              {
2680              int len = 1;              int len = 1;
2681              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2682                  {
2683                  SCHECK_PARTIAL();
2684                  break;
2685                  }
2686              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2687              if (c > 255)              if (c > 255)
2688                {                {
# Line 1975  for (;;) Line 2696  for (;;)
2696              }              }
2697            for (;;)            for (;;)
2698              {              {
2699              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2700              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2701              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2702              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1987  for (;;) Line 2708  for (;;)
2708            {            {
2709            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2710              {              {
2711              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2712                  {
2713                  SCHECK_PARTIAL();
2714                  break;
2715                  }
2716              c = *eptr;              c = *eptr;
2717              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2718              eptr++;              eptr++;
2719              }              }
2720            while (eptr >= pp)            while (eptr >= pp)
2721              {              {
2722              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2723              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2724              eptr--;              eptr--;
2725              }              }
2726            }            }
2727    
2728          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2729          }          }
2730        }        }
2731      /* Control never gets here */      /* Control never gets here */
2732    
2733    
2734      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2735      in UTF-8 mode, because that's the only time it is compiled. */      when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8
2736        mode, because Unicode properties are supported in non-UTF-8 mode. */
2737    
2738  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2739      case OP_XCLASS:      case OP_XCLASS:
# Line 2048  for (;;) Line 2774  for (;;)
2774    
2775        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2776          {          {
2777          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2778          GETCHARINC(c, eptr);            {
2779          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2780              MRRETURN(MATCH_NOMATCH);
2781              }
2782            GETCHARINCTEST(c, eptr);
2783            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2784          }          }
2785    
2786        /* 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 2065  for (;;) Line 2795  for (;;)
2795          {          {
2796          for (fi = min;; fi++)          for (fi = min;; fi++)
2797            {            {
2798            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2799            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2800            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2801            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2802            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2803                SCHECK_PARTIAL();
2804                MRRETURN(MATCH_NOMATCH);
2805                }
2806              GETCHARINCTEST(c, eptr);
2807              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2808            }            }
2809          /* Control never gets here */          /* Control never gets here */
2810          }          }
# Line 2082  for (;;) Line 2817  for (;;)
2817          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2818            {            {
2819            int len = 1;            int len = 1;
2820            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2821            GETCHARLEN(c, eptr, len);              {
2822                SCHECK_PARTIAL();
2823                break;
2824                }
2825              GETCHARLENTEST(c, eptr, len);
2826            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2827            eptr += len;            eptr += len;
2828            }            }
2829          for(;;)          for(;;)
2830            {            {
2831            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2832            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2833            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2834            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
2835            }            }
2836          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2837          }          }
2838    
2839        /* Control never gets here */        /* Control never gets here */
# Line 2110  for (;;) Line 2849  for (;;)
2849        length = 1;        length = 1;
2850        ecode++;        ecode++;
2851        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2852        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2853        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2854            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2855            MRRETURN(MATCH_NOMATCH);
2856            }
2857          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2858        }        }
2859      else      else
2860  #endif  #endif
2861    
2862      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2863        {        {
2864        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2865        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2866            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2867            MRRETURN(MATCH_NOMATCH);
2868            }
2869          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2870        ecode += 2;        ecode += 2;
2871        }        }
2872      break;      break;
2873    
2874      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2875    
2876      case OP_CHARNC:      case OP_CHARI:
2877  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2878      if (utf8)      if (utf8)
2879        {        {
# Line 2134  for (;;) Line 2881  for (;;)
2881        ecode++;        ecode++;
2882        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2883    
2884        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2885            {
2886            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2887            MRRETURN(MATCH_NOMATCH);
2888            }
2889    
2890        /* 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
2891        can use the fast lookup table. */        can use the fast lookup table. */
2892    
2893        if (fc < 128)        if (fc < 128)
2894          {          {
2895          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2896          }          }
2897    
2898        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
# Line 2158  for (;;) Line 2909  for (;;)
2909          if (fc != dc)          if (fc != dc)
2910            {            {
2911  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2912            if (dc != _pcre_ucp_othercase(fc))            if (dc != UCD_OTHERCASE(fc))
2913  #endif  #endif
2914              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
2915            }            }
2916          }          }
2917        }        }
# Line 2169  for (;;) Line 2920  for (;;)
2920    
2921      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2922        {        {
2923        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2924        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
2925            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2926            MRRETURN(MATCH_NOMATCH);
2927            }
2928          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2929        ecode += 2;        ecode += 2;
2930        }        }
2931      break;      break;
# Line 2178  for (;;) Line 2933  for (;;)
2933      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2934    
2935      case OP_EXACT:      case OP_EXACT:
2936        case OP_EXACTI:
2937      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2938      ecode += 3;      ecode += 3;
2939      goto REPEATCHAR;      goto REPEATCHAR;
2940    
2941      case OP_POSUPTO:      case OP_POSUPTO:
2942        case OP_POSUPTOI:
2943      possessive = TRUE;      possessive = TRUE;
2944      /* Fall through */      /* Fall through */
2945    
2946      case OP_UPTO:      case OP_UPTO:
2947        case OP_UPTOI:
2948      case OP_MINUPTO:      case OP_MINUPTO:
2949        case OP_MINUPTOI:
2950      min = 0;      min = 0;
2951      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2952      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2953      ecode += 3;      ecode += 3;
2954      goto REPEATCHAR;      goto REPEATCHAR;
2955    
2956      case OP_POSSTAR:      case OP_POSSTAR:
2957        case OP_POSSTARI:
2958      possessive = TRUE;      possessive = TRUE;
2959      min = 0;      min = 0;
2960      max = INT_MAX;      max = INT_MAX;
# Line 2202  for (;;) Line 2962  for (;;)
2962      goto REPEATCHAR;      goto REPEATCHAR;
2963    
2964      case OP_POSPLUS:      case OP_POSPLUS:
2965        case OP_POSPLUSI:
2966      possessive = TRUE;      possessive = TRUE;
2967      min = 1;      min = 1;
2968      max = INT_MAX;      max = INT_MAX;
# Line 2209  for (;;) Line 2970  for (;;)
2970      goto REPEATCHAR;      goto REPEATCHAR;
2971    
2972      case OP_POSQUERY:      case OP_POSQUERY:
2973        case OP_POSQUERYI:
2974      possessive = TRUE;      possessive = TRUE;
2975      min = 0;      min = 0;
2976      max = 1;      max = 1;
# Line 2216  for (;;) Line 2978  for (;;)
2978      goto REPEATCHAR;      goto REPEATCHAR;
2979    
2980      case OP_STAR:      case OP_STAR:
2981        case OP_STARI:
2982      case OP_MINSTAR:      case OP_MINSTAR:
2983        case OP_MINSTARI:
2984      case OP_PLUS:      case OP_PLUS:
2985        case OP_PLUSI:
2986      case OP_MINPLUS:      case OP_MINPLUS:
2987        case OP_MINPLUSI:
2988      case OP_QUERY:      case OP_QUERY:
2989        case OP_QUERYI:
2990      case OP_MINQUERY:      case OP_MINQUERY:
2991      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
2992        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
2993      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
2994      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2995      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2996      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
2997    
2998      /* Common code for all repeated single-character matches. We can give      /* Common code for all repeated single-character matches. */
     up quickly if there are fewer than the minimum number of characters left in  
     the subject. */  
2999    
3000      REPEATCHAR:      REPEATCHAR:
3001  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2238  for (;;) Line 3004  for (;;)
3004        length = 1;        length = 1;
3005        charptr = ecode;        charptr = ecode;
3006        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3007        ecode += length;        ecode += length;
3008    
3009        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 2248  for (;;) Line 3013  for (;;)
3013          {          {
3014  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3015          unsigned int othercase;          unsigned int othercase;
3016          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3017              (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)              (othercase = UCD_OTHERCASE(fc)) != fc)
3018            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3019          else oclength = 0;          else oclength = 0;
3020  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3021    
3022          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3023            {            {
3024            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3025                memcmp(eptr, charptr, length) == 0) eptr += length;
3026  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3027            /* Need braces because of following else */            else if (oclength > 0 &&
3028            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                     eptr <= md->end_subject - oclength &&
3029                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3030    #endif  /* SUPPORT_UCP */
3031            else            else
3032              {              {
3033              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3034              eptr += oclength;              MRRETURN(MATCH_NOMATCH);
3035              }              }
 #else   /* without SUPPORT_UCP */  
           else { RRETURN(MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3036            }            }
3037    
3038          if (min == max) continue;          if (min == max) continue;
# Line 2276  for (;;) Line 3041  for (;;)
3041            {            {
3042            for (fi = min;; fi++)            for (fi = min;; fi++)
3043              {              {
3044              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3045              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3046              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3047              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3048                  memcmp(eptr, charptr, length) == 0) eptr += length;
3049  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3050              /* Need braces because of following else */              else if (oclength > 0 &&
3051              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                       eptr <= md->end_subject - oclength &&
3052                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3053    #endif  /* SUPPORT_UCP */
3054              else              else
3055                {                {
3056                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3057                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3058                }                }
 #else   /* without SUPPORT_UCP */  
             else { RRETURN (MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3059              }              }
3060            /* Control never gets here */            /* Control never gets here */
3061            }            }
# Line 2300  for (;;) Line 3065  for (;;)
3065            pp = eptr;            pp = eptr;
3066            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3067              {              {
3068              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3069              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3070  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3071              else if (oclength == 0) break;              else if (oclength > 0 &&
3072                         eptr <= md->end_subject - oclength &&
3073                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3074    #endif  /* SUPPORT_UCP */
3075              else              else
3076                {                {
3077                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3078                eptr += oclength;                break;
3079                }                }
 #else   /* without SUPPORT_UCP */  
             else break;  
 #endif  /* SUPPORT_UCP */  
3080              }              }
3081    
3082            if (possessive) continue;            if (possessive) continue;
3083    
3084            for(;;)            for(;;)
3085             {              {
3086             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3087             if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3088             if (eptr == pp) RRETURN(MATCH_NOMATCH);              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3089  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3090             eptr--;              eptr--;
3091             BACKCHAR(eptr);              BACKCHAR(eptr);
3092  #else   /* without SUPPORT_UCP */  #else   /* without SUPPORT_UCP */
3093             eptr -= length;              eptr -= length;
3094  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3095             }              }
3096            }            }
3097          /* Control never gets here */          /* Control never gets here */
3098          }          }
# Line 2339  for (;;) Line 3105  for (;;)
3105  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3106    
3107      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3108        {  
3109        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3110    
3111      /* 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 less than 256, though we may or
3112      may not be in UTF-8 mode. The code is duplicated for the caseless and      may not be in UTF-8 mode. The code is duplicated for the caseless and
# Line 2356  for (;;) Line 3120  for (;;)
3120      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3121        max, eptr));        max, eptr));
3122    
3123      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3124        {        {
3125        fc = md->lcc[fc];        fc = md->lcc[fc];
3126        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3127          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3128            if (eptr >= md->end_subject)
3129              {
3130              SCHECK_PARTIAL();
3131              MRRETURN(MATCH_NOMATCH);
3132              }
3133            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3134            }
3135        if (min == max) continue;        if (min == max) continue;
3136        if (minimize)        if (minimize)
3137          {          {
3138          for (fi = min;; fi++)          for (fi = min;; fi++)
3139            {            {
3140            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3141            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3142            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3143                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3144              RRETURN(MATCH_NOMATCH);              {
3145                SCHECK_PARTIAL();
3146                MRRETURN(MATCH_NOMATCH);
3147                }
3148              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3149            }            }
3150          /* Control never gets here */          /* Control never gets here */
3151          }          }
# Line 2379  for (;;) Line 3154  for (;;)
3154          pp = eptr;          pp = eptr;
3155          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3156            {            {
3157            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3158                {
3159                SCHECK_PARTIAL();
3160                break;
3161                }
3162              if (fc != md->lcc[*eptr]) break;
3163            eptr++;            eptr++;
3164            }            }
3165    
3166          if (possessive) continue;          if (possessive) continue;
3167    
3168          while (eptr >= pp)          while (eptr >= pp)
3169            {            {
3170            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3171            eptr--;            eptr--;
3172            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3173            }            }
3174          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3175          }          }
3176        /* Control never gets here */        /* Control never gets here */
3177        }        }
# Line 2398  for (;;) Line 3180  for (;;)
3180    
3181      else      else
3182        {        {
3183        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3184            {
3185            if (eptr >= md->end_subject)
3186              {
3187              SCHECK_PARTIAL();
3188              MRRETURN(MATCH_NOMATCH);
3189              }
3190            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3191            }
3192    
3193        if (min == max) continue;        if (min == max) continue;
3194    
3195        if (minimize)        if (minimize)
3196          {          {
3197          for (fi = min;; fi++)          for (fi = min;; fi++)
3198            {            {
3199            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3200            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3201            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3202              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3203                {
3204                SCHECK_PARTIAL();
3205                MRRETURN(MATCH_NOMATCH);
3206                }
3207              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3208            }            }
3209          /* Control never gets here */          /* Control never gets here */
3210          }          }
# Line 2416  for (;;) Line 3213  for (;;)
3213          pp = eptr;          pp = eptr;
3214          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3215            {            {
3216            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3217                {
3218                SCHECK_PARTIAL();
3219                break;
3220                }
3221              if (fc != *eptr) break;
3222            eptr++;            eptr++;
3223            }            }
3224          if (possessive) continue;          if (possessive) continue;
3225    
3226          while (eptr >= pp)          while (eptr >= pp)
3227            {            {
3228            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3229            eptr--;            eptr--;
3230            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3231            }            }
3232          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3233          }          }
3234        }        }
3235      /* Control never gets here */      /* Control never gets here */
# Line 2435  for (;;) Line 3238  for (;;)
3238      checking can be multibyte. */      checking can be multibyte. */
3239    
3240      case OP_NOT:      case OP_NOT:
3241      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3242        if (eptr >= md->end_subject)
3243          {
3244          SCHECK_PARTIAL();
3245          MRRETURN(MATCH_NOMATCH);
3246          }
3247      ecode++;      ecode++;
3248      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3249      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3250        {        {
3251  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3252        if (c < 256)        if (c < 256)
3253  #endif  #endif
3254        c = md->lcc[c];        c = md->lcc[c];
3255        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3256        }        }
3257      else      else    /* Caseful */
3258        {        {
3259        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3260        }        }
3261      break;      break;
3262    
# Line 2460  for (;;) Line 3268  for (;;)
3268      about... */      about... */
3269    
3270      case OP_NOTEXACT:      case OP_NOTEXACT:
3271        case OP_NOTEXACTI:
3272      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3273      ecode += 3;      ecode += 3;
3274      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3275    
3276      case OP_NOTUPTO:      case OP_NOTUPTO:
3277        case OP_NOTUPTOI:
3278      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3279        case OP_NOTMINUPTOI:
3280      min = 0;      min = 0;
3281      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3282      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3283      ecode += 3;      ecode += 3;
3284      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3285    
3286      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3287        case OP_NOTPOSSTARI:
3288      possessive = TRUE;      possessive = TRUE;
3289      min = 0;      min = 0;
3290      max = INT_MAX;      max = INT_MAX;
# Line 2480  for (;;) Line 3292  for (;;)
3292      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3293    
3294      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3295        case OP_NOTPOSPLUSI:
3296      possessive = TRUE;      possessive = TRUE;
3297      min = 1;      min = 1;
3298      max = INT_MAX;      max = INT_MAX;
# Line 2487  for (;;) Line 3300  for (;;)
3300      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3301    
3302      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3303        case OP_NOTPOSQUERYI:
3304      possessive = TRUE;      possessive = TRUE;
3305      min = 0;      min = 0;
3306      max = 1;      max = 1;
# Line 2494  for (;;) Line 3308  for (;;)
3308      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3309    
3310      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3311        case OP_NOTPOSUPTOI:
3312      possessive = TRUE;      possessive = TRUE;
3313      min = 0;      min = 0;
3314      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 2501  for (;;) Line 3316  for (;;)
3316      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3317    
3318      case OP_NOTSTAR:      case OP_NOTSTAR:
3319        case OP_NOTSTARI:
3320      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3321        case OP_NOTMINSTARI:
3322      case OP_NOTPLUS:      case OP_NOTPLUS:
3323        case OP_NOTPLUSI:
3324      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3325        case OP_NOTMINPLUSI:
3326      case OP_NOTQUERY:      case OP_NOTQUERY:
3327        case OP_NOTQUERYI:
3328      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3329      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3330        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3331      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3332      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3333      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3334      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3335    
3336      /* Common code for all repeated single-byte matches. We can give up quickly      /* Common code for all repeated single-byte matches. */
     if there are fewer than the minimum number of bytes left in the  
     subject. */  
3337    
3338      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3339      fc = *ecode++;      fc = *ecode++;
3340    
3341      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
# Line 2531  for (;;) Line 3349  for (;;)
3349      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,
3350        max, eptr));        max, eptr));
3351    
3352      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3353        {        {
3354        fc = md->lcc[fc];        fc = md->lcc[fc];
3355    
# Line 2542  for (;;) Line 3360  for (;;)
3360          register unsigned int d;          register unsigned int d;
3361          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3362            {            {
3363              if (eptr >= md->end_subject)
3364                {
3365                SCHECK_PARTIAL();
3366                MRRETURN(MATCH_NOMATCH);
3367                }
3368            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3369            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3370            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3371            }            }
3372          }          }
3373        else        else
# Line 2553  for (;;) Line 3376  for (;;)
3376        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3377          {          {
3378          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3379            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3380              if (eptr >= md->end_subject)
3381                {
3382                SCHECK_PARTIAL();
3383                MRRETURN(MATCH_NOMATCH);
3384                }
3385              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3386              }
3387          }          }
3388    
3389        if (min == max) continue;        if (min == max) continue;
# Line 2567  for (;;) Line 3397  for (;;)
3397            register unsigned int d;            register unsigned int d;
3398            for (fi = min;; fi++)            for (fi = min;; fi++)
3399              {              {
3400              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3401              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3402                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3403                if (eptr >= md->end_subject)
3404                  {
3405                  SCHECK_PARTIAL();
3406                  MRRETURN(MATCH_NOMATCH);
3407                  }
3408              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3409              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3410              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3411              }              }
3412            }            }
3413          else          else
# Line 2581  for (;;) Line 3416  for (;;)
3416            {            {
3417            for (fi = min;; fi++)            for (fi = min;; fi++)
3418              {              {
3419              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3420              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3421              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3422                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3423                  {
3424                  SCHECK_PARTIAL();
3425                  MRRETURN(MATCH_NOMATCH);
3426                  }
3427                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3428              }              }
3429            }            }
3430          /* Control never gets here */          /* Control never gets here */
# Line 2604  for (;;) Line 3444  for (;;)
3444            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3445              {              {
3446              int len = 1;              int len = 1;
3447              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3448                  {
3449                  SCHECK_PARTIAL();
3450                  break;
3451                  }
3452              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3453              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3454              if (fc == d) break;              if (fc == d) break;
# Line 2613  for (;;) Line 3457  for (;;)
3457          if (possessive) continue;          if (possessive) continue;
3458          for(;;)          for(;;)
3459              {              {
3460              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3461              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3462              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3463              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2625  for (;;) Line 3469  for (;;)
3469            {            {
3470            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3471              {              {
3472              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3473                  {
3474                  SCHECK_PARTIAL();
3475                  break;
3476                  }
3477                if (fc == md->lcc[*eptr]) break;
3478              eptr++;              eptr++;
3479              }              }
3480            if (possessive) continue;            if (possessive) continue;
3481            while (eptr >= pp)            while (eptr >= pp)
3482              {              {
3483              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3484              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3485              eptr--;              eptr--;
3486              }              }
3487            }            }
3488    
3489          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3490          }          }
3491        /* Control never gets here */        /* Control never gets here */
3492        }        }
# Line 2653  for (;;) Line 3502  for (;;)
3502          register unsigned int d;          register unsigned int d;
3503          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3504            {            {
3505              if (eptr >= md->end_subject)
3506                {
3507                SCHECK_PARTIAL();
3508                MRRETURN(MATCH_NOMATCH);
3509                }
3510            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3511            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3512            }            }
3513          }          }
3514        else        else
# Line 2662  for (;;) Line 3516  for (;;)
3516        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3517          {          {
3518          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3519            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3520              if (eptr >= md->end_subject)
3521                {
3522                SCHECK_PARTIAL();
3523                MRRETURN(MATCH_NOMATCH);
3524                }
3525              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3526              }
3527          }          }
3528    
3529        if (min == max) continue;        if (min == max) continue;
# Line 2676  for (;;) Line 3537  for (;;)
3537            register unsigned int d;            register unsigned int d;
3538            for (fi = min;; fi++)            for (fi = min;; fi++)
3539              {              {
3540              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3541              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3542                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3543                if (eptr >= md->end_subject)
3544                  {
3545                  SCHECK_PARTIAL();
3546                  MRRETURN(MATCH_NOMATCH);
3547                  }
3548              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3549              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3550              }              }
3551            }            }
3552          else          else
# Line 2689  for (;;) Line 3555  for (;;)
3555            {            {
3556            for (fi = min;; fi++)            for (fi = min;; fi++)
3557              {              {
3558              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3559              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3560              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3561                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3562                  {
3563                  SCHECK_PARTIAL();
3564                  MRRETURN(MATCH_NOMATCH);
3565                  }
3566                if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3567              }              }
3568            }            }
3569          /* Control never gets here */          /* Control never gets here */
# Line 2712  for (;;) Line 3583  for (;;)
3583            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3584              {              {
3585              int len = 1;              int len = 1;
3586              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3587                  {
3588                  SCHECK_PARTIAL();
3589                  break;
3590                  }
3591              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3592              if (fc == d) break;              if (fc == d) break;
3593              eptr += len;              eptr += len;
# Line 2720  for (;;) Line 3595  for (;;)
3595            if (possessive) continue;            if (possessive) continue;
3596            for(;;)            for(;;)
3597              {              {
3598              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3599              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3600              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3601              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2732  for (;;) Line 3607  for (;;)
3607            {            {
3608            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3609              {              {
3610              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3611                  {
3612                  SCHECK_PARTIAL();
3613                  break;
3614                  }
3615               &n