/[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 87 by nigel, Sat Feb 24 21:41:21 2007 UTC revision 608 by ph10, Sun Jun 12 16:25:55 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) 1997-2006 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 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  pattern matching using an NFA algorithm, trying to mimic Perl as closely as  pattern matching using an NFA algorithm, trying to mimic Perl as closely as
43  possible. There are also some static supporting functions. */  possible. There are also some static supporting functions. */
44    
45    #ifdef HAVE_CONFIG_H
46    #include "config.h"
47    #endif
48    
49    #define NLBLOCK md             /* Block containing newline information */
50    #define PSSTART start_subject  /* Field containing processed string start */
51    #define PSEND   end_subject    /* Field containing processed string end */
52    
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    /* Undefine some potentially clashing cpp symbols */
56    
57  /* Structure for building a chain of data that actually lives on the  #undef min
58  stack, for holding the values of the subject pointer at the start of each  #undef max
 subpattern, so as to detect when an empty string has been matched by a  
 subpattern - to break infinite loops. When NO_RECURSE is set, these blocks  
 are on the heap, not on the stack. */  
   
 typedef struct eptrblock {  
   struct eptrblock *epb_prev;  
   USPTR epb_saved_eptr;  
 } eptrblock;  
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_isgroup      0x02    /* Set if start of bracketed 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 68  defined PCRE_ERROR_xxx codes, which are Line 70  defined PCRE_ERROR_xxx codes, which are
70  #define MATCH_MATCH        1  #define MATCH_MATCH        1
71  #define MATCH_NOMATCH      0  #define MATCH_NOMATCH      0
72    
73    /* Special internal returns from the match() function. Make them sufficiently
74    negative to avoid the external error codes. */
75    
76    #define MATCH_ACCEPT       (-999)
77    #define MATCH_COMMIT       (-998)
78    #define MATCH_KETRPOS      (-997)
79    #define MATCH_PRUNE        (-996)
80    #define MATCH_SKIP         (-995)
81    #define MATCH_SKIP_ARG     (-994)
82    #define MATCH_THEN         (-993)
83    
84    /* This is a convenience macro for code that occurs many times. */
85    
86    #define MRRETURN(ra) \
87      { \
88      md->mark = markptr; \
89      RRETURN(ra); \
90      }
91    
92  /* 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.
93  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,
94  because the offset vector is always a multiple of 3 long. */  because the offset vector is always a multiple of 3 long. */
# Line 81  static const char rep_max[] = { 0, 0, 0, Line 102  static const char rep_max[] = { 0, 0, 0,
102    
103    
104    
105  #ifdef DEBUG  #ifdef PCRE_DEBUG
106  /*************************************************  /*************************************************
107  *        Debugging function to print chars       *  *        Debugging function to print chars       *
108  *************************************************/  *************************************************/
# Line 101  Returns:     nothing Line 122  Returns:     nothing
122  static void  static void
123  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)
124  {  {
125  int c;  unsigned int c;
126  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;
127  while (length-- > 0)  while (length-- > 0)
128    if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c);    if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c);
# Line 114  while (length-- > 0) Line 135  while (length-- > 0)
135  *          Match a back-reference                *  *          Match a back-reference                *
136  *************************************************/  *************************************************/
137    
138  /* 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
139  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
140    the length passed is zero. Note that in caseless UTF-8 mode, the number of
141    subject bytes matched may be different to the number of reference bytes.
142    
143  Arguments:  Arguments:
144    offset      index into the offset vector    offset      index into the offset vector
145    eptr        points into the subject    eptr        pointer into the subject
146    length      length to be matched    length      length of reference to be matched (number of bytes)
147    md          points to match data block    md          points to match data block
148    ims         the ims flags    caseless    TRUE if caseless
149    
150  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
151  */  */
152    
153  static BOOL  static int
154  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
155    unsigned long int ims)    BOOL caseless)
156  {  {
157  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
158    register USPTR p = md->start_subject + md->offset_vector[offset];
159    
160  #ifdef DEBUG  #ifdef PCRE_DEBUG
161  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
162    printf("matching subject <null>");    printf("matching subject <null>");
163  else  else
# Line 146  pchars(p, length, FALSE, md); Line 170  pchars(p, length, FALSE, md);
170  printf("\n");  printf("\n");
171  #endif  #endif
172    
173  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
174    
175  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
176    
177  /* Separate the caselesss case for speed */  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
178    properly if Unicode properties are supported. Otherwise, we can check only
179    ASCII characters. */
180    
181  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
182    {    {
183    while (length-- > 0)  #ifdef SUPPORT_UTF8
184      if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE;  #ifdef SUPPORT_UCP
185      if (md->utf8)
186        {
187        /* Match characters up to the end of the reference. NOTE: the number of
188        bytes matched may differ, because there are some characters whose upper and
189        lower case versions code as different numbers of bytes. For example, U+023A
190        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
191        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
192        the latter. It is important, therefore, to check the length along the
193        reference, not along the subject (earlier code did this wrong). */
194    
195        USPTR endptr = p + length;
196        while (p < endptr)
197          {
198          int c, d;
199          if (eptr >= md->end_subject) return -1;
200          GETCHARINC(c, eptr);
201          GETCHARINC(d, p);
202          if (c != d && c != UCD_OTHERCASE(d)) return -1;
203          }
204        }
205      else
206    #endif
207    #endif
208    
209      /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
210      is no UCP support. */
211        {
212        if (eptr + length > md->end_subject) return -1;
213        while (length-- > 0)
214          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
215        }
216    }    }
217    
218    /* In the caseful case, we can just compare the bytes, whether or not we
219    are in UTF-8 mode. */
220    
221  else  else
222    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
223      if (eptr + length > md->end_subject) return -1;
224      while (length-- > 0) if (*p++ != *eptr++) return -1;
225      }
226    
227  return TRUE;  return eptr - eptr_start;
228  }  }
229    
230    
# Line 186  calls by keeping local variables that ne Line 250  calls by keeping local variables that ne
250  obtained from malloc() instead instead of on the stack. Macros are used to  obtained from malloc() instead instead of on the stack. Macros are used to
251  achieve this so that the actual code doesn't look very different to what it  achieve this so that the actual code doesn't look very different to what it
252  always used to.  always used to.
253    
254    The original heap-recursive code used longjmp(). However, it seems that this
255    can be very slow on some operating systems. Following a suggestion from Stan
256    Switzer, the use of longjmp() has been abolished, at the cost of having to
257    provide a unique number for each call to RMATCH. There is no way of generating
258    a sequence of numbers at compile time in C. I have given them names, to make
259    them stand out more clearly.
260    
261    Crude tests on x86 Linux show a small speedup of around 5-8%. However, on
262    FreeBSD, avoiding longjmp() more than halves the time taken to run the standard
263    tests. Furthermore, not using longjmp() means that local dynamic variables
264    don't have indeterminate values; this has meant that the frame size can be
265    reduced because the result can be "passed back" by straight setting of the
266    variable instead of being passed in the frame.
267  ****************************************************************************  ****************************************************************************
268  ***************************************************************************/  ***************************************************************************/
269    
270    /* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN
271    below must be updated in sync.  */
272    
273    enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,
274           RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,
275           RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
276           RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
277           RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
278           RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
279           RM61,  RM62, RM63, RM64 };
280    
281  /* 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
282  versions and production versions. */  versions and production versions. Note that the "rw" argument of RMATCH isn't
283    actually used in this definition. */
284    
285  #ifndef NO_RECURSE  #ifndef NO_RECURSE
286  #define REGISTER register  #define REGISTER register
287  #ifdef DEBUG  
288  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) \  #ifdef PCRE_DEBUG
289    #define RMATCH(ra,rb,rc,rd,re,rw) \
290    { \    { \
291    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
292    rx = match(ra,rb,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
293    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
294    }    }
295  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 208  versions and production versions. */ Line 298  versions and production versions. */
298    return ra; \    return ra; \
299    }    }
300  #else  #else
301  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
302    rx = match(ra,rb,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
303  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
304  #endif  #endif
305    
306  #else  #else
307    
308    
309  /* These versions of the macros manage a private stack on the heap. Note  /* These versions of the macros manage a private stack on the heap. Note that
310  that the rd argument of RMATCH isn't actually used. It's the md argument of  the "rd" argument of RMATCH isn't actually used in this definition. It's the md
311  match(), which never changes. */  argument of match(), which never changes. */
312    
313  #define REGISTER  #define REGISTER
314    
315  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
316    {\    {\
317    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
318    if (setjmp(frame->Xwhere) == 0)\    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
319      {\    frame->Xwhere = rw; \
320      newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
321      newframe->Xecode = rb;\    newframe->Xecode = rb;\
322      newframe->Xoffset_top = rc;\    newframe->Xmstart = mstart;\
323      newframe->Xims = re;\    newframe->Xmarkptr = markptr;\
324      newframe->Xeptrb = rf;\    newframe->Xoffset_top = rc;\
325      newframe->Xflags = rg;\    newframe->Xeptrb = re;\
326      newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
327      newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
328      frame = newframe;\    frame = newframe;\
329      DPRINTF(("restarting from line %d\n", __LINE__));\    DPRINTF(("restarting from line %d\n", __LINE__));\
330      goto HEAP_RECURSE;\    goto HEAP_RECURSE;\
331      }\    L_##rw:\
332    else\    DPRINTF(("jumped back to line %d\n", __LINE__));\
     {\  
     DPRINTF(("longjumped back to line %d\n", __LINE__));\  
     frame = md->thisframe;\  
     rx = frame->Xresult;\  
     }\  
333    }    }
334    
335  #define RRETURN(ra)\  #define RRETURN(ra)\
336    {\    {\
337    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
338    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
339    (pcre_stack_free)(newframe);\    (pcre_stack_free)(oldframe);\
340    if (frame != NULL)\    if (frame != NULL)\
341      {\      {\
342      frame->Xresult = ra;\      rrc = ra;\
343      md->thisframe = frame;\      goto HEAP_RETURN;\
     longjmp(frame->Xwhere, 1);\  
344      }\      }\
345    return ra;\    return ra;\
346    }    }
# Line 269  typedef struct heapframe { Line 353  typedef struct heapframe {
353    
354    /* Function arguments that may change */    /* Function arguments that may change */
355    
356    const uschar *Xeptr;    USPTR Xeptr;
357    const uschar *Xecode;    const uschar *Xecode;
358      USPTR Xmstart;
359      USPTR Xmarkptr;
360    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
361    eptrblock *Xeptrb;    eptrblock *Xeptrb;
362    int Xflags;    unsigned int Xrdepth;
   int Xrdepth;  
363    
364    /* Function local variables */    /* Function local variables */
365    
366    const uschar *Xcallpat;    USPTR Xcallpat;
367    const uschar *Xcharptr;  #ifdef SUPPORT_UTF8
368    const uschar *Xdata;    USPTR Xcharptr;
369    const uschar *Xnext;  #endif
370    const uschar *Xpp;    USPTR Xdata;
371    const uschar *Xprev;    USPTR Xnext;
372    const uschar *Xsaved_eptr;    USPTR Xpp;
373      USPTR Xprev;
374      USPTR Xsaved_eptr;
375    
376    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
377    
378    BOOL Xcur_is_word;    BOOL Xcur_is_word;
379    BOOL Xcondition;    BOOL Xcondition;
   BOOL Xminimize;  
380    BOOL Xprev_is_word;    BOOL Xprev_is_word;
381    
   unsigned long int Xoriginal_ims;  
   
382  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
383    int Xprop_type;    int Xprop_type;
384    int Xprop_value;    int Xprop_value;
# Line 303  typedef struct heapframe { Line 386  typedef struct heapframe {
386    int Xprop_category;    int Xprop_category;
387    int Xprop_chartype;    int Xprop_chartype;
388    int Xprop_script;    int Xprop_script;
389    int *Xprop_test_variable;    int Xoclength;
390      uschar Xocchars[8];
391  #endif  #endif
392    
393      int Xcodelink;
394    int Xctype;    int Xctype;
395    int Xfc;    unsigned int Xfc;
396    int Xfi;    int Xfi;
397    int Xlength;    int Xlength;
398    int Xmax;    int Xmax;
# Line 321  typedef struct heapframe { Line 406  typedef struct heapframe {
406    
407    eptrblock Xnewptrb;    eptrblock Xnewptrb;
408    
409    /* Place to pass back result, and where to jump back to */    /* Where to jump back to */
410    
411    int  Xresult;    int Xwhere;
   jmp_buf Xwhere;  
412    
413  } heapframe;  } heapframe;
414    
# Line 340  typedef struct heapframe { Line 424  typedef struct heapframe {
424  *         Match from current position            *  *         Match from current position            *
425  *************************************************/  *************************************************/
426    
427  /* On entry ecode points to the first opcode, and eptr to the first character  /* This function is called recursively in many circumstances. Whenever it
 in the subject string, while eptrb holds the value of eptr at the start of the  
 last bracketed group - used for breaking infinite loops matching zero-length  
 strings. This function is called recursively in many circumstances. Whenever it  
428  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
429  same response.  same response. */
430    
431    /* These macros pack up tests that are used for partial matching, and which
432    appears several times in the code. We set the "hit end" flag if the pointer is
433    at the end of the subject and also past the start of the subject (i.e.
434    something has been matched). For hard partial matching, we then return
435    immediately. The second one is used when we already know we are past the end of
436    the subject. */
437    
438    #define CHECK_PARTIAL()\
439      if (md->partial != 0 && eptr >= md->end_subject && \
440          eptr > md->start_used_ptr) \
441        { \
442        md->hitend = TRUE; \
443        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
444        }
445    
446  Performance note: It might be tempting to extract commonly used fields from the  #define SCHECK_PARTIAL()\
447  md structure (e.g. utf8, end_subject) into individual variables to improve    if (md->partial != 0 && eptr > md->start_used_ptr) \
448        { \
449        md->hitend = TRUE; \
450        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
451        }
452    
453    
454    /* Performance note: It might be tempting to extract commonly used fields from
455    the md structure (e.g. utf8, end_subject) into individual variables to improve
456  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
457  made performance worse.  made performance worse.
458    
459  Arguments:  Arguments:
460     eptr        pointer in subject     eptr        pointer to current character in subject
461     ecode       position in code     ecode       pointer to current position in compiled code
462       mstart      pointer to the current match start position (can be modified
463                     by encountering \K)
464       markptr     pointer to the most recent MARK name, or NULL
465     offset_top  current top pointer     offset_top  current top pointer
466     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
467     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
468                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_isgroup - this is the start of a bracketed group  
469     rdepth      the recursion depth     rdepth      the recursion depth
470    
471  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
472                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
473                   a negative MATCH_xxx value for PRUNE, SKIP, etc
474                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
475                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
476  */  */
477    
478  static int  static int
479  match(REGISTER USPTR eptr, REGISTER const uschar *ecode,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
480    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
481    int flags, int rdepth)    unsigned int rdepth)
482  {  {
483  /* 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,
484  so they can be ordinary variables in all cases. Mark them with "register"  so they can be ordinary variables in all cases. Mark some of them with
485  because they are used a lot in loops. */  "register" because they are used a lot in loops. */
486    
487  register int  rrc;    /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
488  register int  i;      /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
489  register int  c;      /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
490  register BOOL utf8;   /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
491    
492    BOOL minimize, possessive; /* Quantifier options */
493    BOOL caseless;
494    int condcode;
495    
496  /* 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
497  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 391  heap storage. Set up the top-level frame Line 499  heap storage. Set up the top-level frame
499  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
500    
501  #ifdef NO_RECURSE  #ifdef NO_RECURSE
502  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
503    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
504  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
505    
506  /* Copy in the original argument variables */  /* Copy in the original argument variables */
507    
508  frame->Xeptr = eptr;  frame->Xeptr = eptr;
509  frame->Xecode = ecode;  frame->Xecode = ecode;
510    frame->Xmstart = mstart;
511    frame->Xmarkptr = markptr;
512  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
513  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
514  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
515    
516  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 412  HEAP_RECURSE: Line 521  HEAP_RECURSE:
521    
522  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
523  #define ecode              frame->Xecode  #define ecode              frame->Xecode
524    #define mstart             frame->Xmstart
525    #define markptr            frame->Xmarkptr
526  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
527  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
528  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
529    
530  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 424  HEAP_RECURSE: Line 533  HEAP_RECURSE:
533  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
534  #endif  #endif
535  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
536    #define codelink           frame->Xcodelink
537  #define data               frame->Xdata  #define data               frame->Xdata
538  #define next               frame->Xnext  #define next               frame->Xnext
539  #define pp                 frame->Xpp  #define pp                 frame->Xpp
# Line 434  HEAP_RECURSE: Line 544  HEAP_RECURSE:
544    
545  #define cur_is_word        frame->Xcur_is_word  #define cur_is_word        frame->Xcur_is_word
546  #define condition          frame->Xcondition  #define condition          frame->Xcondition
 #define minimize           frame->Xminimize  
547  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
548    
 #define original_ims       frame->Xoriginal_ims  
   
549  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
550  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
551  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
# Line 446  HEAP_RECURSE: Line 553  HEAP_RECURSE:
553  #define prop_category      frame->Xprop_category  #define prop_category      frame->Xprop_category
554  #define prop_chartype      frame->Xprop_chartype  #define prop_chartype      frame->Xprop_chartype
555  #define prop_script        frame->Xprop_script  #define prop_script        frame->Xprop_script
556  #define prop_test_variable frame->Xprop_test_variable  #define oclength           frame->Xoclength
557    #define occhars            frame->Xocchars
558  #endif  #endif
559    
560  #define ctype              frame->Xctype  #define ctype              frame->Xctype
# Line 470  HEAP_RECURSE: Line 578  HEAP_RECURSE:
578  get preserved during recursion in the normal way. In this environment, fi and  get preserved during recursion in the normal way. In this environment, fi and
579  i, and fc and c, can be the same variables. */  i, and fc and c, can be the same variables. */
580    
581  #else  #else         /* NO_RECURSE not defined */
582  #define fi i  #define fi i
583  #define fc c  #define fc c
584    
585    /* Many of the following variables are used only in small blocks of the code.
586  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  My normal style of coding would have declared them within each of those blocks.
587  const uschar *charptr;             /* in small blocks of the code. My normal */  However, in order to accommodate the version of this code that uses an external
588  #endif                             /* style of coding would have declared    */  "stack" implemented on the heap, it is easier to declare them all here, so the
589  const uschar *callpat;             /* them within each of those blocks.      */  declarations can be cut out in a block. The only declarations within blocks
590  const uschar *data;                /* However, in order to accommodate the   */  below are for variables that do not have to be preserved over a recursive call
591  const uschar *next;                /* version of this code that uses an      */  to RMATCH(). */
592  USPTR         pp;                  /* external "stack" implemented on the    */  
593  const uschar *prev;                /* heap, it is easier to declare them all */  #ifdef SUPPORT_UTF8
594  USPTR         saved_eptr;          /* here, so the declarations can be cut   */  const uschar *charptr;
595                                     /* out in a block. The only declarations  */  #endif
596  recursion_info new_recursive;      /* within blocks below are for variables  */  const uschar *callpat;
597                                     /* that do not have to be preserved over  */  const uschar *data;
598  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  const uschar *next;
599    USPTR         pp;
600    const uschar *prev;
601    USPTR         saved_eptr;
602    
603    recursion_info new_recursive;
604    
605    BOOL cur_is_word;
606  BOOL condition;  BOOL condition;
 BOOL minimize;  
607  BOOL prev_is_word;  BOOL prev_is_word;
608    
 unsigned long int original_ims;  
   
609  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
610  int prop_type;  int prop_type;
611  int prop_value;  int prop_value;
# Line 501  int prop_fail_result; Line 613  int prop_fail_result;
613  int prop_category;  int prop_category;
614  int prop_chartype;  int prop_chartype;
615  int prop_script;  int prop_script;
616  int *prop_test_variable;  int oclength;
617    uschar occhars[8];
618  #endif  #endif
619    
620    int codelink;
621  int ctype;  int ctype;
622  int length;  int length;
623  int max;  int max;
# Line 516  int save_offset1, save_offset2, save_off Line 630  int save_offset1, save_offset2, save_off
630  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
631    
632  eptrblock newptrb;  eptrblock newptrb;
633  #endif  #endif     /* NO_RECURSE */
634    
635    /* To save space on the stack and in the heap frame, I have doubled up on some
636    of the local variables that are used only in localised parts of the code, but
637    still need to be preserved over recursive calls of match(). These macros define
638    the alternative names that are used. */
639    
640    #define allow_zero    cur_is_word
641    #define cbegroup      condition
642    #define code_offset   codelink
643    #define condassert    condition
644    #define matched_once  prev_is_word
645    
646  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
647  variables. */  variables. */
# Line 524  variables. */ Line 649  variables. */
649  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
650  prop_value = 0;  prop_value = 0;
651  prop_fail_result = 0;  prop_fail_result = 0;
 prop_test_variable = NULL;  
652  #endif  #endif
653    
654    
655    /* This label is used for tail recursion, which is used in a few cases even
656    when NO_RECURSE is not defined, in order to reduce the amount of stack that is
657    used. Thanks to Ian Taylor for noticing this possibility and sending the
658    original patch. */
659    
660    TAIL_RECURSE:
661    
662  /* 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
663  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
664  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()
665  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
666  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
667  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,
668  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
669    
670    #ifdef SUPPORT_UTF8
671    utf8 = md->utf8;       /* Local copy of the flag */
672    #else
673    utf8 = FALSE;
674    #endif
675    
676  /* First check that we haven't called match() too many times, or that we  /* First check that we haven't called match() too many times, or that we
677  haven't exceeded the recursive call limit. */  haven't exceeded the recursive call limit. */
678    
679  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
680  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
681    
682  original_ims = ims;    /* Save for resetting on ')' */  /* At the start of a group with an unlimited repeat that may match an empty
683  utf8 = md->utf8;       /* Local copy of the flag */  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
684    done this way to save having to use another function argument, which would take
685  /* At the start of a bracketed group, add the current subject pointer to the  up space on the stack. See also MATCH_CONDASSERT below.
686  stack of such pointers, to be re-instated at the end of the group when we hit  
687  the closing ket. When match() is called in other circumstances, we don't add to  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
688  this stack. */  such remembered pointers, to be checked when we hit the closing ket, in order
689    to break infinite loops that match no characters. When match() is called in
690    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
691    NOT be used with tail recursion, because the memory block that is used is on
692    the stack, so a new one may be required for each match(). */
693    
694  if ((flags & match_isgroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
695    {    {
   newptrb.epb_prev = eptrb;  
696    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
697      newptrb.epb_prev = eptrb;
698    eptrb = &newptrb;    eptrb = &newptrb;
699      md->match_function_type = 0;
700    }    }
701    
702  /* Now start processing the operations. */  /* Now start processing the opcodes. */
703    
704  for (;;)  for (;;)
705    {    {
706      minimize = possessive = FALSE;
707    op = *ecode;    op = *ecode;
708    minimize = FALSE;  
709      switch(op)
710    /* For partial matching, remember if we ever hit the end of the subject after      {
711    matching at least one subject character. */      case OP_MARK:
712        markptr = ecode + 2;
713        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
714          eptrb, RM55);
715    
716        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
717        argument, and we must check whether that argument matches this MARK's
718        argument. It is passed back in md->start_match_ptr (an overloading of that
719        variable). If it does match, we reset that variable to the current subject
720        position and return MATCH_SKIP. Otherwise, pass back the return code
721        unaltered. */
722    
723        if (rrc == MATCH_SKIP_ARG &&
724            strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
725          {
726          md->start_match_ptr = eptr;
727          RRETURN(MATCH_SKIP);
728          }
729    
730        if (md->mark == NULL) md->mark = markptr;
731        RRETURN(rrc);
732    
733        case OP_FAIL:
734        MRRETURN(MATCH_NOMATCH);
735    
736        /* COMMIT overrides PRUNE, SKIP, and THEN */
737    
738        case OP_COMMIT:
739        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
740          eptrb, RM52);
741        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
742            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
743            rrc != MATCH_THEN)
744          RRETURN(rrc);
745        MRRETURN(MATCH_COMMIT);
746    
747    if (md->partial &&      /* PRUNE overrides THEN */
       eptr >= md->end_subject &&  
       eptr > md->start_match)  
     md->hitend = TRUE;  
   
   /* Opening capturing bracket. If there is space in the offset vector, save  
   the current subject position in the working slot at the top of the vector. We  
   mustn't change the current values of the data slot, because they may be set  
   from a previous iteration of this group, and be referred to by a reference  
   inside the group.  
   
   If the bracket fails to match, we need to restore this value and also the  
   values of the final offsets, in case they were set by a previous iteration of  
   the same bracket.  
   
   If there isn't enough space in the offset vector, treat this as if it were a  
   non-capturing bracket. Don't worry about setting the flag for the error case  
   here; that is handled in the code for KET. */  
748    
749    if (op > OP_BRA)      case OP_PRUNE:
750      {      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
751      number = op - OP_BRA;        eptrb, RM51);
752        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
753        MRRETURN(MATCH_PRUNE);
754    
755        case OP_PRUNE_ARG:
756        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
757          eptrb, RM56);
758        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
759        md->mark = ecode + 2;
760        RRETURN(MATCH_PRUNE);
761    
762        /* SKIP overrides PRUNE and THEN */
763    
764        case OP_SKIP:
765        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
766          eptrb, RM53);
767        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769        md->start_match_ptr = eptr;   /* Pass back current position */
770        MRRETURN(MATCH_SKIP);
771    
772      /* For extended extraction brackets (large number), we have to fish out the      case OP_SKIP_ARG:
773      number from a dummy opcode at the start. */      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
774          eptrb, RM57);
775        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
776          RRETURN(rrc);
777    
778      if (number > EXTRACT_BASIC_MAX)      /* Pass back the current skip name by overloading md->start_match_ptr and
779        number = GET2(ecode, 2+LINK_SIZE);      returning the special MATCH_SKIP_ARG return code. This will either be
780        caught by a matching MARK, or get to the top, where it is treated the same
781        as PRUNE. */
782    
783        md->start_match_ptr = ecode + 2;
784        RRETURN(MATCH_SKIP_ARG);
785    
786        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
787        the alt that is at the start of the current branch. This makes it possible
788        to skip back past alternatives that precede the THEN within the current
789        branch. */
790    
791        case OP_THEN:
792        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
793          eptrb, RM54);
794        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
795        md->start_match_ptr = ecode - GET(ecode, 1);
796        MRRETURN(MATCH_THEN);
797    
798        case OP_THEN_ARG:
799        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
800          offset_top, md, eptrb, RM58);
801        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
802        md->start_match_ptr = ecode - GET(ecode, 1);
803        md->mark = ecode + LINK_SIZE + 2;
804        RRETURN(MATCH_THEN);
805    
806        /* Handle a capturing bracket, other than those that are possessive with an
807        unlimited repeat. If there is space in the offset vector, save the current
808        subject position in the working slot at the top of the vector. We mustn't
809        change the current values of the data slot, because they may be set from a
810        previous iteration of this group, and be referred to by a reference inside
811        the group. If we fail to match, we need to restore this value and also the
812        values of the final offsets, in case they were set by a previous iteration
813        of the same bracket.
814    
815        If there isn't enough space in the offset vector, treat this as if it were
816        a non-capturing bracket. Don't worry about setting the flag for the error
817        case here; that is handled in the code for KET. */
818    
819        case OP_CBRA:
820        case OP_SCBRA:
821        number = GET2(ecode, 1+LINK_SIZE);
822      offset = number << 1;      offset = number << 1;
823    
824  #ifdef DEBUG  #ifdef PCRE_DEBUG
825      printf("start bracket %d subject=", number);      printf("start bracket %d\n", number);
826        printf("subject=");
827      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
828      printf("\n");      printf("\n");
829  #endif  #endif
# Line 610  for (;;) Line 836  for (;;)
836        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
837    
838        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
839        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
840            (int)(eptr - md->start_subject);
841    
842        do        for (;;)
843          {          {
844          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
845            match_isgroup);          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
846          if (rrc != MATCH_NOMATCH) RRETURN(rrc);            eptrb, RM1);
847            if (rrc != MATCH_NOMATCH &&
848                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
849              RRETURN(rrc);
850          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
851          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
852            if (*ecode != OP_ALT) break;
853          }          }
       while (*ecode == OP_ALT);  
854    
855        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
856    
# Line 628  for (;;) Line 858  for (;;)
858        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
859        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
860    
861          if (rrc != MATCH_THEN) md->mark = markptr;
862        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
863        }        }
864    
865      /* Insufficient room for saving captured contents */      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
866        as a non-capturing bracket. */
867    
868      else op = OP_BRA;      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
869      }      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870    
871    /* Other types of node can be handled by a switch */      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
872    
873    switch(op)      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
874      {      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
875      case OP_BRA:     /* Non-capturing bracket: optimized */  
876      DPRINTF(("start bracket 0\n"));      /* Non-capturing bracket, except for possessive with unlimited repeat. Loop
877      do      for all the alternatives. When we get to the final alternative within the
878        brackets, we would return the result of a recursive call to match()
879        whatever happened. We can reduce stack usage by turning this into a tail
880        recursion, except in the case of a possibly empty group.*/
881    
882        case OP_BRA:
883        case OP_SBRA:
884        DPRINTF(("start non-capturing bracket\n"));
885        for (;;)
886        {        {
887        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */
888          match_isgroup);          {
889        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (op >= OP_SBRA)   /* Possibly empty group */
890              {
891              md->match_function_type = MATCH_CBEGROUP;
892              RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
893                RM48);
894              if (rrc == MATCH_NOMATCH) md->mark = markptr;
895              RRETURN(rrc);
896              }
897            /* Not a possibly empty group; use tail recursion */
898            ecode += _pcre_OP_lengths[*ecode];
899            DPRINTF(("bracket 0 tail recursion\n"));
900            goto TAIL_RECURSE;
901            }
902    
903          /* For non-final alternatives, continue the loop for a NOMATCH result;
904          otherwise return. */
905    
906          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
907          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
908            RM2);
909          if (rrc != MATCH_NOMATCH &&
910              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
911            RRETURN(rrc);
912        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
913        }        }
914      while (*ecode == OP_ALT);      /* Control never reaches here. */
915      DPRINTF(("bracket 0 failed\n"));  
916        /* Handle possessive capturing brackets with an unlimited repeat. We come
917        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
918        handled similarly to the normal case above. However, the matching is
919        different. The end of these brackets will always be OP_KETRPOS, which
920        returns MATCH_KETRPOS without going further in the pattern. By this means
921        we can handle the group by iteration rather than recursion, thereby
922        reducing the amount of stack needed. */
923    
924        case OP_CBRAPOS:
925        case OP_SCBRAPOS:
926        allow_zero = FALSE;
927    
928        POSSESSIVE_CAPTURE:
929        number = GET2(ecode, 1+LINK_SIZE);
930        offset = number << 1;
931    
932    #ifdef PCRE_DEBUG
933        printf("start possessive bracket %d\n", number);
934        printf("subject=");
935        pchars(eptr, 16, TRUE, md);
936        printf("\n");
937    #endif
938    
939        if (offset < md->offset_max)
940          {
941          matched_once = FALSE;
942          code_offset = ecode - md->start_code;
943    
944          save_offset1 = md->offset_vector[offset];
945          save_offset2 = md->offset_vector[offset+1];
946          save_offset3 = md->offset_vector[md->offset_end - number];
947          save_capture_last = md->capture_last;
948    
949          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
950    
951          /* Each time round the loop, save the current subject position for use
952          when the group matches. For MATCH_MATCH, the group has matched, so we
953          restart it with a new subject starting position, remembering that we had
954          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
955          usual. If we haven't matched any alternatives in any iteration, check to
956          see if a previous iteration matched. If so, the group has matched;
957          continue from afterwards. Otherwise it has failed; restore the previous
958          capture values before returning NOMATCH. */
959    
960          for (;;)
961            {
962            md->offset_vector[md->offset_end - number] =
963              (int)(eptr - md->start_subject);
964            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
965            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
966              eptrb, RM63);
967            if (rrc == MATCH_KETRPOS)
968              {
969              offset_top = md->end_offset_top;
970              eptr = md->end_match_ptr;
971              ecode = md->start_code + code_offset;
972              save_capture_last = md->capture_last;
973              matched_once = TRUE;
974              continue;
975              }
976            if (rrc != MATCH_NOMATCH &&
977                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
978              RRETURN(rrc);
979            md->capture_last = save_capture_last;
980            ecode += GET(ecode, 1);
981            if (*ecode != OP_ALT) break;
982            }
983    
984          if (!matched_once)
985            {
986            md->offset_vector[offset] = save_offset1;
987            md->offset_vector[offset+1] = save_offset2;
988            md->offset_vector[md->offset_end - number] = save_offset3;
989            }
990    
991          if (rrc != MATCH_THEN) md->mark = markptr;
992          if (allow_zero || matched_once)
993            {
994            ecode += 1 + LINK_SIZE;
995            break;
996            }
997    
998          RRETURN(MATCH_NOMATCH);
999          }
1000    
1001        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1002        as a non-capturing bracket. */
1003    
1004        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1005        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1006    
1007        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1008    
1009        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1010        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1011    
1012        /* Non-capturing possessive bracket with unlimited repeat. We come here
1013        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1014        without the capturing complication. It is written out separately for speed
1015        and cleanliness. */
1016    
1017        case OP_BRAPOS:
1018        case OP_SBRAPOS:
1019        allow_zero = FALSE;
1020    
1021        POSSESSIVE_NON_CAPTURE:
1022        matched_once = FALSE;
1023        code_offset = ecode - md->start_code;
1024    
1025        for (;;)
1026          {
1027          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1028          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1029            eptrb, RM64);
1030          if (rrc == MATCH_KETRPOS)
1031            {
1032            eptr = md->end_match_ptr;
1033            ecode = md->start_code + code_offset;
1034            matched_once = TRUE;
1035            continue;
1036            }
1037          if (rrc != MATCH_NOMATCH &&
1038              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1039            RRETURN(rrc);
1040          ecode += GET(ecode, 1);
1041          if (*ecode != OP_ALT) break;
1042          }
1043    
1044        if (matched_once || allow_zero)
1045          {
1046          ecode += 1 + LINK_SIZE;
1047          break;
1048          }
1049      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1050    
1051        /* Control never reaches here. */
1052    
1053      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1054      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
1055      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
1056      exactly what going to the ket would do. */      exactly what going to the ket would do. As there is only one branch to be
1057        obeyed, we can use tail recursion to avoid using another stack frame. */
1058    
1059      case OP_COND:      case OP_COND:
1060      if (ecode[LINK_SIZE+1] == OP_CREF) /* Condition extract or recurse test */      case OP_SCOND:
1061        codelink = GET(ecode, 1);
1062    
1063        /* Because of the way auto-callout works during compile, a callout item is
1064        inserted between OP_COND and an assertion condition. */
1065    
1066        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1067          {
1068          if (pcre_callout != NULL)
1069            {
1070            pcre_callout_block cb;
1071            cb.version          = 1;   /* Version 1 of the callout block */
1072            cb.callout_number   = ecode[LINK_SIZE+2];
1073            cb.offset_vector    = md->offset_vector;
1074            cb.subject          = (PCRE_SPTR)md->start_subject;
1075            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1076            cb.start_match      = (int)(mstart - md->start_subject);
1077            cb.current_position = (int)(eptr - md->start_subject);
1078            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1079            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1080            cb.capture_top      = offset_top/2;
1081            cb.capture_last     = md->capture_last;
1082            cb.callout_data     = md->callout_data;
1083            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1084            if (rrc < 0) RRETURN(rrc);
1085            }
1086          ecode += _pcre_OP_lengths[OP_CALLOUT];
1087          }
1088    
1089        condcode = ecode[LINK_SIZE+1];
1090    
1091        /* Now see what the actual condition is */
1092    
1093        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1094          {
1095          if (md->recursive == NULL)                /* Not recursing => FALSE */
1096            {
1097            condition = FALSE;
1098            ecode += GET(ecode, 1);
1099            }
1100          else
1101            {
1102            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1103            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1104    
1105            /* If the test is for recursion into a specific subpattern, and it is
1106            false, but the test was set up by name, scan the table to see if the
1107            name refers to any other numbers, and test them. The condition is true
1108            if any one is set. */
1109    
1110            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1111              {
1112              uschar *slotA = md->name_table;
1113              for (i = 0; i < md->name_count; i++)
1114                {
1115                if (GET2(slotA, 0) == recno) break;
1116                slotA += md->name_entry_size;
1117                }
1118    
1119              /* Found a name for the number - there can be only one; duplicate
1120              names for different numbers are allowed, but not vice versa. First
1121              scan down for duplicates. */
1122    
1123              if (i < md->name_count)
1124                {
1125                uschar *slotB = slotA;
1126                while (slotB > md->name_table)
1127                  {
1128                  slotB -= md->name_entry_size;
1129                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1130                    {
1131                    condition = GET2(slotB, 0) == md->recursive->group_num;
1132                    if (condition) break;
1133                    }
1134                  else break;
1135                  }
1136    
1137                /* Scan up for duplicates */
1138    
1139                if (!condition)
1140                  {
1141                  slotB = slotA;
1142                  for (i++; i < md->name_count; i++)
1143                    {
1144                    slotB += md->name_entry_size;
1145                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1146                      {
1147                      condition = GET2(slotB, 0) == md->recursive->group_num;
1148                      if (condition) break;
1149                      }
1150                    else break;
1151                    }
1152                  }
1153                }
1154              }
1155    
1156            /* Chose branch according to the condition */
1157    
1158            ecode += condition? 3 : GET(ecode, 1);
1159            }
1160          }
1161    
1162        else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1163        {        {
1164        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1165        condition = (offset == CREF_RECURSE * 2)?        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1166          (md->recursive != NULL) :  
1167          (offset < offset_top && md->offset_vector[offset] >= 0);        /* If the numbered capture is unset, but the reference was by name,
1168        RMATCH(rrc, eptr, ecode + (condition?        scan the table to see if the name refers to any other numbers, and test
1169          (LINK_SIZE + 4) : (LINK_SIZE + 1 + GET(ecode, 1))),        them. The condition is true if any one is set. This is tediously similar
1170          offset_top, md, ims, eptrb, match_isgroup);        to the code above, but not close enough to try to amalgamate. */
1171        RRETURN(rrc);  
1172          if (!condition && condcode == OP_NCREF)
1173            {
1174            int refno = offset >> 1;
1175            uschar *slotA = md->name_table;
1176    
1177            for (i = 0; i < md->name_count; i++)
1178              {
1179              if (GET2(slotA, 0) == refno) break;
1180              slotA += md->name_entry_size;
1181              }
1182    
1183            /* Found a name for the number - there can be only one; duplicate names
1184            for different numbers are allowed, but not vice versa. First scan down
1185            for duplicates. */
1186    
1187            if (i < md->name_count)
1188              {
1189              uschar *slotB = slotA;
1190              while (slotB > md->name_table)
1191                {
1192                slotB -= md->name_entry_size;
1193                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1194                  {
1195                  offset = GET2(slotB, 0) << 1;
1196                  condition = offset < offset_top &&
1197                    md->offset_vector[offset] >= 0;
1198                  if (condition) break;
1199                  }
1200                else break;
1201                }
1202    
1203              /* Scan up for duplicates */
1204    
1205              if (!condition)
1206                {
1207                slotB = slotA;
1208                for (i++; i < md->name_count; i++)
1209                  {
1210                  slotB += md->name_entry_size;
1211                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1212                    {
1213                    offset = GET2(slotB, 0) << 1;
1214                    condition = offset < offset_top &&
1215                      md->offset_vector[offset] >= 0;
1216                    if (condition) break;
1217                    }
1218                  else break;
1219                  }
1220                }
1221              }
1222            }
1223    
1224          /* Chose branch according to the condition */
1225    
1226          ecode += condition? 3 : GET(ecode, 1);
1227          }
1228    
1229        else if (condcode == OP_DEF)     /* DEFINE - always false */
1230          {
1231          condition = FALSE;
1232          ecode += GET(ecode, 1);
1233        }        }
1234    
1235      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1236      the final argument TRUE causes it to stop at the end of an assertion. */      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1237        an assertion. */
1238    
1239      else      else
1240        {        {
1241        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1242            match_condassert | match_isgroup);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1243        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1244          {          {
1245          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE+2);          condition = TRUE;
1246            ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1247          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1248          }          }
1249        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH &&
1250                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1251          {          {
1252          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1253          }          }
1254        else ecode += GET(ecode, 1);        else
1255        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,          {
1256          match_isgroup);          condition = FALSE;
1257        RRETURN(rrc);          ecode += codelink;
1258            }
1259        }        }
     /* Control never reaches here */  
1260    
1261      /* Skip over conditional reference or large extraction number data if      /* We are now at the branch that is to be obeyed. As there is only one,
1262      encountered. */      we can use tail recursion to avoid using another stack frame, except when
1263        we have an unlimited repeat of a possibly empty group. If the second
1264        alternative doesn't exist, we can just plough on. */
1265    
1266        if (condition || *ecode == OP_ALT)
1267          {
1268          ecode += 1 + LINK_SIZE;
1269          if (op == OP_SCOND)        /* Possibly empty group */
1270            {
1271            md->match_function_type = MATCH_CBEGROUP;
1272            RMATCH(eptr, ecode, offset_top, md, eptrb, RM49);
1273            RRETURN(rrc);
1274            }
1275          else goto TAIL_RECURSE;
1276          }
1277        else                         /* Condition false & no alternative */
1278          {
1279          ecode += 1 + LINK_SIZE;
1280          }
1281        break;
1282    
1283    
1284        /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1285        to close any currently open capturing brackets. */
1286    
1287        case OP_CLOSE:
1288        number = GET2(ecode, 1);
1289        offset = number << 1;
1290    
1291    #ifdef PCRE_DEBUG
1292          printf("end bracket %d at *ACCEPT", number);
1293          printf("\n");
1294    #endif
1295    
1296      case OP_CREF:      md->capture_last = number;
1297      case OP_BRANUMBER:      if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1298          {
1299          md->offset_vector[offset] =
1300            md->offset_vector[md->offset_end - number];
1301          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1302          if (offset_top <= offset) offset_top = offset + 2;
1303          }
1304      ecode += 3;      ecode += 3;
1305      break;      break;
1306    
     /* End of the pattern. If we are in a recursion, we should restore the  
     offsets appropriately and continue from after the call. */  
1307    
1308        /* End of the pattern, either real or forced. If we are in a recursion, we
1309        should restore the offsets appropriately, and if it's a top-level
1310        recursion, continue from after the call. */
1311    
1312        case OP_ACCEPT:
1313      case OP_END:      case OP_END:
1314      if (md->recursive != NULL && md->recursive->group_num == 0)      if (md->recursive != NULL)
1315        {        {
1316        recursion_info *rec = md->recursive;        recursion_info *rec = md->recursive;
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
1317        md->recursive = rec->prevrec;        md->recursive = rec->prevrec;
1318        memmove(md->offset_vector, rec->offset_save,        memmove(md->offset_vector, rec->offset_save,
1319          rec->saved_max * sizeof(int));          rec->saved_max * sizeof(int));
1320        md->start_match = rec->save_start;        offset_top = rec->save_offset_top;
1321        ims = original_ims;        if (rec->group_num == 0)
1322        ecode = rec->after_call;          {
1323        break;          ecode = rec->after_call;
1324            break;
1325            }
1326        }        }
1327    
1328      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is
1329      string - backtracking will then try other alternatives, if any. */      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of
1330        the subject. In both cases, backtracking will then try other alternatives,
1331        if any. */
1332    
1333        else if (eptr == mstart &&
1334            (md->notempty ||
1335              (md->notempty_atstart &&
1336                mstart == md->start_subject + md->start_offset)))
1337          MRRETURN(MATCH_NOMATCH);
1338    
1339        /* Otherwise, we have a match. */
1340    
1341        md->end_match_ptr = eptr;           /* Record where we ended */
1342        md->end_offset_top = offset_top;    /* and how many extracts were taken */
1343        md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1344    
1345      if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH);      /* For some reason, the macros don't work properly if an expression is
1346      md->end_match_ptr = eptr;          /* Record where we ended */      given as the argument to MRRETURN when the heap is in use. */
1347      md->end_offset_top = offset_top;   /* and how many extracts were taken */  
1348      RRETURN(MATCH_MATCH);      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1349        MRRETURN(rrc);
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1350    
1351      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1352      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,
1353      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
1354      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
1355      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1356        of a condition, we want to return immediately afterwards. The caller of
1357        this incarnation of the match() function will have set MATCH_CONDASSERT in
1358        md->match_function type, and one of these opcodes will be the first opcode
1359        that is processed. We use a local variable that is preserved over calls to
1360        match() to remember this case. */
1361    
1362      case OP_ASSERT:      case OP_ASSERT:
1363      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1364        if (md->match_function_type == MATCH_CONDASSERT)
1365          {
1366          condassert = TRUE;
1367          md->match_function_type = 0;
1368          }
1369        else condassert = FALSE;
1370    
1371      do      do
1372        {        {
1373        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1374          match_isgroup);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1375        if (rrc == MATCH_MATCH) break;          {
1376        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          mstart = md->start_match_ptr;   /* In case \K reset it */
1377            break;
1378            }
1379          if (rrc != MATCH_NOMATCH &&
1380              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1381            RRETURN(rrc);
1382        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1383        }        }
1384      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1385      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1386        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1387    
1388      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1389    
1390      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1391    
1392      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1393      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 766  for (;;) Line 1397  for (;;)
1397      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1398      continue;      continue;
1399    
1400      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1401        PRUNE, or COMMIT means we must assume failure without checking subsequent
1402        branches. */
1403    
1404      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1405      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1406        if (md->match_function_type == MATCH_CONDASSERT)
1407          {
1408          condassert = TRUE;
1409          md->match_function_type = 0;
1410          }
1411        else condassert = FALSE;
1412    
1413      do      do
1414        {        {
1415        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1416          match_isgroup);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1417        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1418        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          {
1419            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1420            break;
1421            }
1422          if (rrc != MATCH_NOMATCH &&
1423              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1424            RRETURN(rrc);
1425        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1426        }        }
1427      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1428    
1429      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1430    
1431      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1432      continue;      continue;
1433    
# Line 794  for (;;) Line 1440  for (;;)
1440  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1441      if (utf8)      if (utf8)
1442        {        {
1443        c = GET(ecode,1);        i = GET(ecode, 1);
1444        for (i = 0; i < c; i++)        while (i-- > 0)
1445          {          {
1446          eptr--;          eptr--;
1447          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1448          BACKCHAR(eptr)          BACKCHAR(eptr);
1449          }          }
1450        }        }
1451      else      else
# Line 808  for (;;) Line 1454  for (;;)
1454      /* No UTF-8 support, or not in UTF-8 mode: count is byte count */      /* No UTF-8 support, or not in UTF-8 mode: count is byte count */
1455    
1456        {        {
1457        eptr -= GET(ecode,1);        eptr -= GET(ecode, 1);
1458        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1459        }        }
1460    
1461      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1462    
1463        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1464      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1465      break;      break;
1466    
# Line 829  for (;;) Line 1476  for (;;)
1476        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1477        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1478        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1479        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1480        cb.start_match      = md->start_match - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1481        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1482        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1483        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1484        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1485        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1486        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1487        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1488        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1489        }        }
1490      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 865  for (;;) Line 1512  for (;;)
1512      case OP_RECURSE:      case OP_RECURSE:
1513        {        {
1514        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1515        new_recursive.group_num = *callpat - OP_BRA;        new_recursive.group_num = (callpat == md->start_code)? 0 :
1516            GET2(callpat, 1 + LINK_SIZE);
       /* For extended extraction brackets (large number), we have to fish out  
       the number from a dummy opcode at the start. */  
   
       if (new_recursive.group_num > EXTRACT_BASIC_MAX)  
         new_recursive.group_num = GET2(callpat, 2+LINK_SIZE);  
1517    
1518        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1519    
# Line 897  for (;;) Line 1539  for (;;)
1539    
1540        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1541              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1542        new_recursive.save_start = md->start_match;        new_recursive.save_offset_top = offset_top;
1543        md->start_match = eptr;  
   
1544        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. For each top-level alternative we
1545        restore the offset and recursion data. */        restore the offset and recursion data. */
1546    
1547        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1548          cbegroup = (*callpat >= OP_SBRA);
1549        do        do
1550          {          {
1551          RMATCH(rrc, eptr, callpat + 1 + LINK_SIZE, offset_top, md, ims,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1552              eptrb, match_isgroup);          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1553          if (rrc == MATCH_MATCH)            md, eptrb, RM6);
1554            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1555            {            {
1556            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1557            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1558            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1559              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1560            RRETURN(MATCH_MATCH);            MRRETURN(MATCH_MATCH);
1561            }            }
1562          else if (rrc != MATCH_NOMATCH)          else if (rrc != MATCH_NOMATCH &&
1563                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1564            {            {
1565            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1566              if (new_recursive.offset_save != stacksave)
1567                (pcre_free)(new_recursive.offset_save);
1568            RRETURN(rrc);            RRETURN(rrc);
1569            }            }
1570    
# Line 933  for (;;) Line 1579  for (;;)
1579        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1580        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1581          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1582        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1583        }        }
1584      /* Control never reaches here */      /* Control never reaches here */
1585    
# Line 942  for (;;) Line 1588  for (;;)
1588      a move back into the brackets. Friedl calls these "atomic" subpatterns.      a move back into the brackets. Friedl calls these "atomic" subpatterns.
1589      Check the alternative branches in turn - the matching won't pass the KET      Check the alternative branches in turn - the matching won't pass the KET
1590      for this kind of subpattern. If any one branch matches, we carry on as at      for this kind of subpattern. If any one branch matches, we carry on as at
1591      the end of a normal bracket, leaving the subject pointer. */      the end of a normal bracket, leaving the subject pointer, but resetting
1592        the start-of-match value in case it was changed by \K. */
1593    
1594      case OP_ONCE:      case OP_ONCE:
1595        {      prev = ecode;
1596        prev = ecode;      saved_eptr = eptr;
       saved_eptr = eptr;  
1597    
1598        do      do
1599          {
1600          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1601          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
1602          {          {
1603          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,          mstart = md->start_match_ptr;
1604            eptrb, match_isgroup);          break;
         if (rrc == MATCH_MATCH) break;  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         ecode += GET(ecode,1);  
1605          }          }
1606        while (*ecode == OP_ALT);        if (rrc != MATCH_NOMATCH &&
1607              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1608            RRETURN(rrc);
1609          ecode += GET(ecode,1);
1610          }
1611        while (*ecode == OP_ALT);
1612    
1613        /* If hit the end of the group (which could be repeated), fail */      /* If hit the end of the group (which could be repeated), fail */
1614    
1615        if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);      if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
1616    
1617        /* Continue as from after the assertion, updating the offsets high water      /* Continue as from after the assertion, updating the offsets high water
1618        mark, since extracts may have been taken. */      mark, since extracts may have been taken. */
1619    
1620        do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
1621    
1622        offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1623        eptr = md->end_match_ptr;      eptr = md->end_match_ptr;
1624    
1625        /* For a non-repeating ket, just continue at this level. This also      /* For a non-repeating ket, just continue at this level. This also
1626        happens for a repeating ket if no characters were matched in the group.      happens for a repeating ket if no characters were matched in the group.
1627        This is the forcible breaking of infinite loops as implemented in Perl      This is the forcible breaking of infinite loops as implemented in Perl
1628        5.005. If there is an options reset, it will get obeyed in the normal      5.005. If there is an options reset, it will get obeyed in the normal
1629        course of events. */      course of events. */
1630    
1631        if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1632          {        {
1633          ecode += 1+LINK_SIZE;        ecode += 1+LINK_SIZE;
1634          break;        break;
1635          }        }
1636    
1637        /* The repeating kets try the rest of the pattern or restart from the      /* The repeating kets try the rest of the pattern or restart from the
1638        preceding bracket, in the appropriate order. We need to reset any options      preceding bracket, in the appropriate order. The second "call" of match()
1639        that changed within the bracket before re-running it, so check the next      uses tail recursion, to avoid using another stack frame. */
       opcode. */  
1640    
1641        if (ecode[1+LINK_SIZE] == OP_OPT)      if (*ecode == OP_KETRMIN)
1642          {        {
1643          ims = (ims & ~PCRE_IMS) | ecode[4];        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM8);
1644          DPRINTF(("ims set to %02lx at group repeat\n", ims));        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1645          }        ecode = prev;
1646          goto TAIL_RECURSE;
1647        if (*ecode == OP_KETRMIN)        }
1648          {      else  /* OP_KETRMAX */
1649          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0);        {
1650          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        md->match_function_type = MATCH_CBEGROUP;
1651          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);        RMATCH(eptr, prev, offset_top, md, eptrb, RM9);
1652          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1653          }        ecode += 1 + LINK_SIZE;
1654        else  /* OP_KETRMAX */        goto TAIL_RECURSE;
1655          {        }
1656          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);      /* Control never gets here */
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         }  
       }  
     RRETURN(MATCH_NOMATCH);  
1657    
1658      /* 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
1659      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1018  for (;;) Line 1662  for (;;)
1662      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1663      break;      break;
1664    
1665      /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1666      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
1667      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
1668      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
1669      preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1670    
1671      case OP_BRAZERO:      case OP_BRAZERO:
1672        {      next = ecode + 1;
1673        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1674        RMATCH(rrc, eptr, next, offset_top, md, ims, eptrb, match_isgroup);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1675        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1676        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1+LINK_SIZE;  
       }  
1677      break;      break;
1678    
1679      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1680        {      next = ecode + 1;
1681        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1682        do next += GET(next,1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1683        RMATCH(rrc, eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb,      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1684          match_isgroup);      ecode++;
1685        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      break;
1686        ecode++;  
1687        }      case OP_SKIPZERO:
1688        next = ecode+1;
1689        do next += GET(next,1); while (*next == OP_ALT);
1690        ecode = next + 1 + LINK_SIZE;
1691      break;      break;
1692    
1693        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1694        here; just jump to the group, with allow_zero set TRUE. */
1695    
1696        case OP_BRAPOSZERO:
1697        op = *(++ecode);
1698        allow_zero = TRUE;
1699        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1700          goto POSSESSIVE_NON_CAPTURE;
1701    
1702      /* End of a group, repeated or non-repeating. If we are at the end of      /* End of a group, repeated or non-repeating. */
     an assertion "group", stop matching and return MATCH_MATCH, but record the  
     current high water mark for use by positive assertions. Do this also  
     for the "once" (not-backup up) groups. */  
1703    
1704      case OP_KET:      case OP_KET:
1705      case OP_KETRMIN:      case OP_KETRMIN:
1706      case OP_KETRMAX:      case OP_KETRMAX:
1707        {      case OP_KETRPOS:
1708        prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
       saved_eptr = eptrb->epb_saved_eptr;  
1709    
1710        /* Back up the stack of bracket start pointers. */      /* If this was a group that remembered the subject start, in order to break
1711        infinite repeats of empty string matches, retrieve the subject start from
1712        the chain. Otherwise, set it NULL. */
1713    
1714        eptrb = eptrb->epb_prev;      if (*prev >= OP_SBRA)
1715          {
1716        if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1717            *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1718            *prev == OP_ONCE)        }
1719          {      else saved_eptr = NULL;
         md->end_match_ptr = eptr;      /* For ONCE */  
         md->end_offset_top = offset_top;  
         RRETURN(MATCH_MATCH);  
         }  
1720    
1721        /* In all other cases except a conditional group we have to check the      /* If we are at the end of an assertion group or an atomic group, stop
1722        group number back at the start and if necessary complete handling an      matching and return MATCH_MATCH, but record the current high water mark for
1723        extraction by setting the offsets and bumping the high water mark. */      use by positive assertions. We also need to record the match start in case
1724        it was changed by \K. */
1725    
1726        if (*prev != OP_COND)      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1727          {          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||
1728          number = *prev - OP_BRA;          *prev == OP_ONCE)
1729          {
1730          md->end_match_ptr = eptr;      /* For ONCE */
1731          md->end_offset_top = offset_top;
1732          md->start_match_ptr = mstart;
1733          MRRETURN(MATCH_MATCH);
1734          }
1735    
1736          /* For extended extraction brackets (large number), we have to fish out      /* For capturing groups we have to check the group number back at the start
1737          the number from a dummy opcode at the start. */      and if necessary complete handling an extraction by setting the offsets and
1738        bumping the high water mark. Note that whole-pattern recursion is coded as
1739        a recurse into group 0, so it won't be picked up here. Instead, we catch it
1740        when the OP_END is reached. Other recursion is handled here. */
1741    
1742          if (number > EXTRACT_BASIC_MAX) number = GET2(prev, 2+LINK_SIZE);      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1743          offset = number << 1;          *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1744          {
1745          number = GET2(prev, 1+LINK_SIZE);
1746          offset = number << 1;
1747    
1748  #ifdef DEBUG  #ifdef PCRE_DEBUG
1749          printf("end bracket %d", number);        printf("end bracket %d", number);
1750          printf("\n");        printf("\n");
1751  #endif  #endif
1752    
1753          /* Test for a numbered group. This includes groups called as a result        md->capture_last = number;
1754          of recursion. Note that whole-pattern recursion is coded as a recurse        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1755          into group 0, so it won't be picked up here. Instead, we catch it when          {
1756          the OP_END is reached. */          md->offset_vector[offset] =
1757              md->offset_vector[md->offset_end - number];
1758          if (number > 0)          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1759            {          if (offset_top <= offset) offset_top = offset + 2;
           md->capture_last = number;  
           if (offset >= md->offset_max) md->offset_overflow = TRUE; else  
             {  
             md->offset_vector[offset] =  
               md->offset_vector[md->offset_end - number];  
             md->offset_vector[offset+1] = eptr - md->start_subject;  
             if (offset_top <= offset) offset_top = offset + 2;  
             }  
   
           /* Handle a recursively called group. Restore the offsets  
           appropriately and continue from after the call. */  
   
           if (md->recursive != NULL && md->recursive->group_num == number)  
             {  
             recursion_info *rec = md->recursive;  
             DPRINTF(("Recursion (%d) succeeded - continuing\n", number));  
             md->recursive = rec->prevrec;  
             md->start_match = rec->save_start;  
             memcpy(md->offset_vector, rec->offset_save,  
               rec->saved_max * sizeof(int));  
             ecode = rec->after_call;  
             ims = original_ims;  
             break;  
             }  
           }  
1760          }          }
1761    
1762        /* Reset the value of the ims flags, in case they got changed during        /* Handle a recursively called group. Restore the offsets
1763        the group. */        appropriately and continue from after the call. */
   
       ims = original_ims;  
       DPRINTF(("ims reset to %02lx\n", ims));  
   
       /* For a non-repeating ket, just continue at this level. This also  
       happens for a repeating ket if no characters were matched in the group.  
       This is the forcible breaking of infinite loops as implemented in Perl  
       5.005. If there is an options reset, it will get obeyed in the normal  
       course of events. */  
1764    
1765        if (*ecode == OP_KET || eptr == saved_eptr)        if (md->recursive != NULL && md->recursive->group_num == number)
1766          {          {
1767          ecode += 1 + LINK_SIZE;          recursion_info *rec = md->recursive;
1768            DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
1769            md->recursive = rec->prevrec;
1770            memcpy(md->offset_vector, rec->offset_save,
1771              rec->saved_max * sizeof(int));
1772            offset_top = rec->save_offset_top;
1773            ecode = rec->after_call;
1774          break;          break;
1775          }          }
1776          }
1777    
1778        /* The repeating kets try the rest of the pattern or restart from the      /* For a non-repeating ket, just continue at this level. This also
1779        preceding bracket, in the appropriate order. */      happens for a repeating ket if no characters were matched in the group.
1780        This is the forcible breaking of infinite loops as implemented in Perl
1781        5.005. If there is an options reset, it will get obeyed in the normal
1782        course of events. */
1783    
1784        if (*ecode == OP_KETRMIN)      if (*ecode == OP_KET || eptr == saved_eptr)
1785          {        {
1786          RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);        ecode += 1 + LINK_SIZE;
1787          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        break;
1788          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);        }
1789          if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
1790          }      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1791        else  /* OP_KETRMAX */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1792        at a time from the outer level, thus saving stack. */
1793    
1794        if (*ecode == OP_KETRPOS)
1795          {
1796          md->end_match_ptr = eptr;
1797          md->end_offset_top = offset_top;
1798          RRETURN(MATCH_KETRPOS);
1799          }
1800    
1801        /* The normal repeating kets try the rest of the pattern or restart from
1802        the preceding bracket, in the appropriate order. In the second case, we can
1803        use tail recursion to avoid using another stack frame, unless we have an
1804        unlimited repeat of a group that can match an empty string. */
1805    
1806        if (*ecode == OP_KETRMIN)
1807          {
1808          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1809          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1810          if (*prev >= OP_SBRA)    /* Could match an empty string */
1811          {          {
1812          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);          md->match_function_type = MATCH_CBEGROUP;
1813          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1814          RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);          RRETURN(rrc);
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
1815          }          }
1816          ecode = prev;
1817          goto TAIL_RECURSE;
1818        }        }
1819        else  /* OP_KETRMAX */
     RRETURN(MATCH_NOMATCH);  
   
     /* Start of subject unless notbol, or after internal newline if multiline */  
   
     case OP_CIRC:  
     if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);  
     if ((ims & PCRE_MULTILINE) != 0)  
1820        {        {
1821        if (eptr != md->start_subject && eptr[-1] != NEWLINE)        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1822          RRETURN(MATCH_NOMATCH);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1823        ecode++;        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1824        break;        ecode += 1 + LINK_SIZE;
1825          goto TAIL_RECURSE;
1826        }        }
1827      /* ... else fall through */      /* Control never gets here */
1828    
1829        /* Not multiline mode: start of subject assertion, unless notbol. */
1830    
1831        case OP_CIRC:
1832        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1833    
1834      /* Start of subject assertion */      /* Start of subject assertion */
1835    
1836      case OP_SOD:      case OP_SOD:
1837      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1838        ecode++;
1839        break;
1840    
1841        /* Multiline mode: start of subject unless notbol, or after any newline. */
1842    
1843        case OP_CIRCM:
1844        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1845        if (eptr != md->start_subject &&
1846            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1847          MRRETURN(MATCH_NOMATCH);
1848      ecode++;      ecode++;
1849      break;      break;
1850    
1851      /* Start of match assertion */      /* Start of match assertion */
1852    
1853      case OP_SOM:      case OP_SOM:
1854      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1855      ecode++;      ecode++;
1856      break;      break;
1857    
1858      /* Assert before internal newline if multiline, or before a terminating      /* Reset the start of match point */
     newline unless endonly is set, else end of subject unless noteol is set. */  
1859    
1860      case OP_DOLL:      case OP_SET_SOM:
1861      if ((ims & PCRE_MULTILINE) != 0)      mstart = eptr;
1862        {      ecode++;
1863        if (eptr < md->end_subject)      break;
1864          { if (*eptr != NEWLINE) RRETURN(MATCH_NOMATCH); }  
1865        else      /* Multiline mode: assert before any newline, or before end of subject
1866          { if (md->noteol) RRETURN(MATCH_NOMATCH); }      unless noteol is set. */
1867        ecode++;  
1868        break;      case OP_DOLLM:
1869        }      if (eptr < md->end_subject)
1870          { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1871      else      else
1872        {        {
1873        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1874        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr < md->end_subject - 1 ||  
            (eptr == md->end_subject - 1 && *eptr != NEWLINE))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1875        }        }
1876      /* ... else fall through */      ecode++;
1877        break;
1878    
1879        /* Not multiline mode: assert before a terminating newline or before end of
1880        subject unless noteol is set. */
1881    
1882        case OP_DOLL:
1883        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1884        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1885    
1886        /* ... else fall through for endonly */
1887    
1888      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1889    
1890      case OP_EOD:      case OP_EOD:
1891      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1892        SCHECK_PARTIAL();
1893      ecode++;      ecode++;
1894      break;      break;
1895    
1896      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1897    
1898      case OP_EODN:      case OP_EODN:
1899      if (eptr < md->end_subject - 1 ||      ASSERT_NL_OR_EOS:
1900         (eptr == md->end_subject - 1 && *eptr != NEWLINE)) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject &&
1901            (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1902          MRRETURN(MATCH_NOMATCH);
1903    
1904        /* Either at end of string or \n before end. */
1905    
1906        SCHECK_PARTIAL();
1907      ecode++;      ecode++;
1908      break;      break;
1909    
# Line 1239  for (;;) Line 1915  for (;;)
1915    
1916        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1917        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
1918        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1919          partial matching. */
1920    
1921  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1922        if (utf8)        if (utf8)
1923          {          {
1924            /* Get status of previous character */
1925    
1926          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1927            {            {
1928            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1929            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1930              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1931            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1932    #ifdef SUPPORT_UCP
1933              if (md->use_ucp)
1934                {
1935                if (c == '_') prev_is_word = TRUE; else
1936                  {
1937                  int cat = UCD_CATEGORY(c);
1938                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1939                  }
1940                }
1941              else
1942    #endif
1943            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1944            }            }
1945          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
1946            /* Get status of next character */
1947    
1948            if (eptr >= md->end_subject)
1949              {
1950              SCHECK_PARTIAL();
1951              cur_is_word = FALSE;
1952              }
1953            else
1954            {            {
1955            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1956    #ifdef SUPPORT_UCP
1957              if (md->use_ucp)
1958                {
1959                if (c == '_') cur_is_word = TRUE; else
1960                  {
1961                  int cat = UCD_CATEGORY(c);
1962                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1963                  }
1964                }
1965              else
1966    #endif
1967            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1968            }            }
1969          }          }
1970        else        else
1971  #endif  #endif
1972    
1973        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1974          consistency with the behaviour of \w we do use it in this case. */
1975    
1976          {          {
1977          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
1978            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
1979          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
1980            ((md->ctypes[*eptr] & ctype_word) != 0);            {
1981              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1982    #ifdef SUPPORT_UCP
1983              if (md->use_ucp)
1984                {
1985                c = eptr[-1];
1986                if (c == '_') prev_is_word = TRUE; else
1987                  {
1988                  int cat = UCD_CATEGORY(c);
1989                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1990                  }
1991                }
1992              else
1993    #endif
1994              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1995              }
1996    
1997            /* Get status of next character */
1998    
1999            if (eptr >= md->end_subject)
2000              {
2001              SCHECK_PARTIAL();
2002              cur_is_word = FALSE;
2003              }
2004            else
2005    #ifdef SUPPORT_UCP
2006            if (md->use_ucp)
2007              {
2008              c = *eptr;
2009              if (c == '_') cur_is_word = TRUE; else
2010                {
2011                int cat = UCD_CATEGORY(c);
2012                cur_is_word = (cat == ucp_L || cat == ucp_N);
2013                }
2014              }
2015            else
2016    #endif
2017            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2018          }          }
2019    
2020        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2021    
2022        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2023             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2024          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2025        }        }
2026      break;      break;
2027    
2028      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2029    
2030      case OP_ANY:      case OP_ANY:
2031      if ((ims & PCRE_DOTALL) == 0 && eptr < md->end_subject && *eptr == NEWLINE)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2032        RRETURN(MATCH_NOMATCH);      /* Fall through */
2033      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);  
2034  #ifdef SUPPORT_UTF8      case OP_ALLANY:
2035      if (utf8)      if (eptr++ >= md->end_subject)
2036        while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;        {
2037  #endif        SCHECK_PARTIAL();
2038          MRRETURN(MATCH_NOMATCH);
2039          }
2040        if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2041      ecode++;      ecode++;
2042      break;      break;
2043    
# Line 1294  for (;;) Line 2045  for (;;)
2045      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2046    
2047      case OP_ANYBYTE:      case OP_ANYBYTE:
2048      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr++ >= md->end_subject)
2049          {
2050          SCHECK_PARTIAL();
2051          MRRETURN(MATCH_NOMATCH);
2052          }
2053      ecode++;      ecode++;
2054      break;      break;
2055    
2056      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2057      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2058          {
2059          SCHECK_PARTIAL();
2060          MRRETURN(MATCH_NOMATCH);
2061          }
2062      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2063      if (      if (
2064  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1307  for (;;) Line 2066  for (;;)
2066  #endif  #endif
2067         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2068         )         )
2069        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2070      ecode++;      ecode++;
2071      break;      break;
2072    
2073      case OP_DIGIT:      case OP_DIGIT:
2074      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2075          {
2076          SCHECK_PARTIAL();
2077          MRRETURN(MATCH_NOMATCH);
2078          }
2079      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2080      if (      if (
2081  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1320  for (;;) Line 2083  for (;;)
2083  #endif  #endif
2084         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2085         )         )
2086        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2087      ecode++;      ecode++;
2088      break;      break;
2089    
2090      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2091      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2092          {
2093          SCHECK_PARTIAL();
2094          MRRETURN(MATCH_NOMATCH);
2095          }
2096      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2097      if (      if (
2098  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1333  for (;;) Line 2100  for (;;)
2100  #endif  #endif
2101         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2102         )         )
2103        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2104      ecode++;      ecode++;
2105      break;      break;
2106    
2107      case OP_WHITESPACE:      case OP_WHITESPACE:
2108      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2109          {
2110          SCHECK_PARTIAL();
2111          MRRETURN(MATCH_NOMATCH);
2112          }
2113      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2114      if (      if (
2115  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1346  for (;;) Line 2117  for (;;)
2117  #endif  #endif
2118         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2119         )         )
2120        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2121      ecode++;      ecode++;
2122      break;      break;
2123    
2124      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2125      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2126          {
2127          SCHECK_PARTIAL();
2128          MRRETURN(MATCH_NOMATCH);
2129          }
2130      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2131      if (      if (
2132  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1359  for (;;) Line 2134  for (;;)
2134  #endif  #endif
2135         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2136         )         )
2137        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2138      ecode++;      ecode++;
2139      break;      break;
2140    
2141      case OP_WORDCHAR:      case OP_WORDCHAR:
2142      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2143          {
2144          SCHECK_PARTIAL();
2145          MRRETURN(MATCH_NOMATCH);
2146          }
2147      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2148      if (      if (
2149  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1372  for (;;) Line 2151  for (;;)
2151  #endif  #endif
2152         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2153         )         )
2154        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2155        ecode++;
2156        break;
2157    
2158        case OP_ANYNL:
2159        if (eptr >= md->end_subject)
2160          {
2161          SCHECK_PARTIAL();
2162          MRRETURN(MATCH_NOMATCH);
2163          }
2164        GETCHARINCTEST(c, eptr);
2165        switch(c)
2166          {
2167          default: MRRETURN(MATCH_NOMATCH);
2168    
2169          case 0x000d:
2170          if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2171          break;
2172    
2173          case 0x000a:
2174          break;
2175    
2176          case 0x000b:
2177          case 0x000c:
2178          case 0x0085:
2179          case 0x2028:
2180          case 0x2029:
2181          if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2182          break;
2183          }
2184        ecode++;
2185        break;
2186    
2187        case OP_NOT_HSPACE:
2188        if (eptr >= md->end_subject)
2189          {
2190          SCHECK_PARTIAL();
2191          MRRETURN(MATCH_NOMATCH);
2192          }
2193        GETCHARINCTEST(c, eptr);
2194        switch(c)
2195          {
2196          default: break;
2197          case 0x09:      /* HT */
2198          case 0x20:      /* SPACE */
2199          case 0xa0:      /* NBSP */
2200          case 0x1680:    /* OGHAM SPACE MARK */
2201          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2202          case 0x2000:    /* EN QUAD */
2203          case 0x2001:    /* EM QUAD */
2204          case 0x2002:    /* EN SPACE */
2205          case 0x2003:    /* EM SPACE */
2206          case 0x2004:    /* THREE-PER-EM SPACE */
2207          case 0x2005:    /* FOUR-PER-EM SPACE */
2208          case 0x2006:    /* SIX-PER-EM SPACE */
2209          case 0x2007:    /* FIGURE SPACE */
2210          case 0x2008:    /* PUNCTUATION SPACE */
2211          case 0x2009:    /* THIN SPACE */
2212          case 0x200A:    /* HAIR SPACE */
2213          case 0x202f:    /* NARROW NO-BREAK SPACE */
2214          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2215          case 0x3000:    /* IDEOGRAPHIC SPACE */
2216          MRRETURN(MATCH_NOMATCH);
2217          }
2218        ecode++;
2219        break;
2220    
2221        case OP_HSPACE:
2222        if (eptr >= md->end_subject)
2223          {
2224          SCHECK_PARTIAL();
2225          MRRETURN(MATCH_NOMATCH);
2226          }
2227        GETCHARINCTEST(c, eptr);
2228        switch(c)
2229          {
2230          default: MRRETURN(MATCH_NOMATCH);
2231          case 0x09:      /* HT */
2232          case 0x20:      /* SPACE */
2233          case 0xa0:      /* NBSP */
2234          case 0x1680:    /* OGHAM SPACE MARK */
2235          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2236          case 0x2000:    /* EN QUAD */
2237          case 0x2001:    /* EM QUAD */
2238          case 0x2002:    /* EN SPACE */
2239          case 0x2003:    /* EM SPACE */
2240          case 0x2004:    /* THREE-PER-EM SPACE */
2241          case 0x2005:    /* FOUR-PER-EM SPACE */
2242          case 0x2006:    /* SIX-PER-EM SPACE */
2243          case 0x2007:    /* FIGURE SPACE */
2244          case 0x2008:    /* PUNCTUATION SPACE */
2245          case 0x2009:    /* THIN SPACE */
2246          case 0x200A:    /* HAIR SPACE */
2247          case 0x202f:    /* NARROW NO-BREAK SPACE */
2248          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2249          case 0x3000:    /* IDEOGRAPHIC SPACE */
2250          break;
2251          }
2252        ecode++;
2253        break;
2254    
2255        case OP_NOT_VSPACE:
2256        if (eptr >= md->end_subject)
2257          {
2258          SCHECK_PARTIAL();
2259          MRRETURN(MATCH_NOMATCH);
2260          }
2261        GETCHARINCTEST(c, eptr);
2262        switch(c)
2263          {
2264          default: break;
2265          case 0x0a:      /* LF */
2266          case 0x0b:      /* VT */
2267          case 0x0c:      /* FF */
2268          case 0x0d:      /* CR */
2269          case 0x85:      /* NEL */
2270          case 0x2028:    /* LINE SEPARATOR */
2271          case 0x2029:    /* PARAGRAPH SEPARATOR */
2272          MRRETURN(MATCH_NOMATCH);
2273          }
2274        ecode++;
2275        break;
2276    
2277        case OP_VSPACE:
2278        if (eptr >= md->end_subject)
2279          {
2280          SCHECK_PARTIAL();
2281          MRRETURN(MATCH_NOMATCH);
2282          }
2283        GETCHARINCTEST(c, eptr);
2284        switch(c)
2285          {
2286          default: MRRETURN(MATCH_NOMATCH);
2287          case 0x0a:      /* LF */
2288          case 0x0b:      /* VT */
2289          case 0x0c:      /* FF */
2290          case 0x0d:      /* CR */
2291          case 0x85:      /* NEL */
2292          case 0x2028:    /* LINE SEPARATOR */
2293          case 0x2029:    /* PARAGRAPH SEPARATOR */
2294          break;
2295          }
2296      ecode++;      ecode++;
2297      break;      break;
2298    
# Line 1382  for (;;) Line 2302  for (;;)
2302    
2303      case OP_PROP:      case OP_PROP:
2304      case OP_NOTPROP:      case OP_NOTPROP:
2305      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2306          {
2307          SCHECK_PARTIAL();
2308          MRRETURN(MATCH_NOMATCH);
2309          }
2310      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2311        {        {
2312        int chartype, script;        const ucd_record *prop = GET_UCD(c);
       int category = _pcre_ucp_findprop(c, &chartype, &script);  
2313    
2314        switch(ecode[1])        switch(ecode[1])
2315          {          {
2316          case PT_ANY:          case PT_ANY:
2317          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2318          break;          break;
2319    
2320          case PT_LAMP:          case PT_LAMP:
2321          if ((chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2322               chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2323               chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2324            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2325           break;          break;
2326    
2327          case PT_GC:          case PT_GC:
2328          if ((ecode[2] != category) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2329            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2330          break;          break;
2331    
2332          case PT_PC:          case PT_PC:
2333          if ((ecode[2] != chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2334            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2335          break;          break;
2336    
2337          case PT_SC:          case PT_SC:
2338          if ((ecode[2] != script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2339            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2340            break;
2341    
2342            /* These are specials */
2343    
2344            case PT_ALNUM:
2345            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2346                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2347              MRRETURN(MATCH_NOMATCH);
2348            break;
2349    
2350            case PT_SPACE:    /* Perl space */
2351            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2352                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2353                   == (op == OP_NOTPROP))
2354              MRRETURN(MATCH_NOMATCH);
2355            break;
2356    
2357            case PT_PXSPACE:  /* POSIX space */
2358            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2359                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2360                 c == CHAR_FF || c == CHAR_CR)
2361                   == (op == OP_NOTPROP))
2362              MRRETURN(MATCH_NOMATCH);
2363            break;
2364    
2365            case PT_WORD:
2366            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2367                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2368                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2369              MRRETURN(MATCH_NOMATCH);
2370          break;          break;
2371    
2372            /* This should never occur */
2373    
2374          default:          default:
2375          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
         break;  
2376          }          }
2377    
2378        ecode += 3;        ecode += 3;
# Line 1429  for (;;) Line 2383  for (;;)
2383      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2384    
2385      case OP_EXTUNI:      case OP_EXTUNI:
2386      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2387          {
2388          SCHECK_PARTIAL();
2389          MRRETURN(MATCH_NOMATCH);
2390          }
2391      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2392        {        {
2393        int chartype, script;        int category = UCD_CATEGORY(c);
2394        int category = _pcre_ucp_findprop(c, &chartype, &script);        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);
       if (category == ucp_M) RRETURN(MATCH_NOMATCH);  
2395        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2396          {          {
2397          int len = 1;          int len = 1;
# Line 1442  for (;;) Line 2399  for (;;)
2399            {            {
2400            GETCHARLEN(c, eptr, len);            GETCHARLEN(c, eptr, len);
2401            }            }
2402          category = _pcre_ucp_findprop(c, &chartype, &script);          category = UCD_CATEGORY(c);
2403          if (category != ucp_M) break;          if (category != ucp_M) break;
2404          eptr += len;          eptr += len;
2405          }          }
# Line 1461  for (;;) Line 2418  for (;;)
2418      loops). */      loops). */
2419    
2420      case OP_REF:      case OP_REF:
2421        {      case OP_REFI:
2422        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2423        ecode += 3;                                 /* Advance past item */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2424        ecode += 3;
       /* If the reference is unset, set the length to be longer than the amount  
       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];  
2425    
2426        /* Set up for repetition, or handle the non-repeated case */      /* If the reference is unset, there are two possibilities:
2427    
2428        switch (*ecode)      (a) In the default, Perl-compatible state, set the length negative;
2429          {      this ensures that every attempt at a match fails. We can't just fail
2430          case OP_CRSTAR:      here, because of the possibility of quantifiers with zero minima.
         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;  
2431    
2432          case OP_CRRANGE:      (b) If the JavaScript compatibility flag is set, set the length to zero
2433          case OP_CRMINRANGE:      so that the back reference matches an empty string.
         minimize = (*ecode == OP_CRMINRANGE);  
         min = GET2(ecode, 1);  
         max = GET2(ecode, 3);  
         if (max == 0) max = INT_MAX;  
         ecode += 5;  
         break;  
2434    
2435          default:               /* No repeat follows */      Otherwise, set the length to the length of what was matched by the
2436          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);      referenced subpattern. */
         eptr += length;  
         continue;              /* With the main loop */  
         }  
2437    
2438        /* If the length of the reference is zero, just continue with the      if (offset >= offset_top || md->offset_vector[offset] < 0)
2439        main loop. */        length = (md->jscript_compat)? 0 : -1;
2440        else
2441          length = md->offset_vector[offset+1] - md->offset_vector[offset];
2442    
2443        if (length == 0) continue;      /* Set up for repetition, or handle the non-repeated case */
2444    
2445        /* First, ensure the minimum number of matches are present. We get back      switch (*ecode)
2446        the length of the reference string explicitly rather than passing the        {
2447        address of eptr, so that eptr can be a register variable. */        case OP_CRSTAR:
2448          case OP_CRMINSTAR:
2449          case OP_CRPLUS:
2450          case OP_CRMINPLUS:
2451          case OP_CRQUERY:
2452          case OP_CRMINQUERY:
2453          c = *ecode++ - OP_CRSTAR;
2454          minimize = (c & 1) != 0;
2455          min = rep_min[c];                 /* Pick up values from tables; */
2456          max = rep_max[c];                 /* zero for max => infinity */
2457          if (max == 0) max = INT_MAX;
2458          break;
2459    
2460        for (i = 1; i <= min; i++)        case OP_CRRANGE:
2461          case OP_CRMINRANGE:
2462          minimize = (*ecode == OP_CRMINRANGE);
2463          min = GET2(ecode, 1);
2464          max = GET2(ecode, 3);
2465          if (max == 0) max = INT_MAX;
2466          ecode += 5;
2467          break;
2468    
2469          default:               /* No repeat follows */
2470          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2471          {          {
2472          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2473          eptr += length;          MRRETURN(MATCH_NOMATCH);
2474          }          }
2475          eptr += length;
2476          continue;              /* With the main loop */
2477          }
2478    
2479        /* If min = max, continue at the same level without recursion.      /* Handle repeated back references. If the length of the reference is
2480        They are not both allowed to be zero. */      zero, just continue with the main loop. */
2481    
2482        if (min == max) continue;      if (length == 0) continue;
2483    
2484        /* If minimizing, keep trying and advancing the pointer */      /* First, ensure the minimum number of matches are present. We get back
2485        the length of the reference string explicitly rather than passing the
2486        address of eptr, so that eptr can be a register variable. */
2487    
2488        if (minimize)      for (i = 1; i <= min; i++)
2489          {
2490          int slength;
2491          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2492          {          {
2493          for (fi = min;; fi++)          CHECK_PARTIAL();
2494            {          MRRETURN(MATCH_NOMATCH);
           RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);  
           if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
           }  
         /* Control never gets here */  
2495          }          }
2496          eptr += slength;
2497          }
2498    
2499        /* If maximizing, find the longest string and work backwards */      /* If min = max, continue at the same level without recursion.
2500        They are not both allowed to be zero. */
2501    
2502        else      if (min == max) continue;
2503    
2504        /* If minimizing, keep trying and advancing the pointer */
2505    
2506        if (minimize)
2507          {
2508          for (fi = min;; fi++)
2509          {          {
2510          pp = eptr;          int slength;
2511          for (i = min; i < max; i++)          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2512            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2513            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2514            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2515            {            {
2516            if (!match_ref(offset, eptr, length, md, ims)) break;            CHECK_PARTIAL();
2517            eptr += length;            MRRETURN(MATCH_NOMATCH);
2518            }            }
2519          while (eptr >= pp)          eptr += slength;
2520            }
2521          /* Control never gets here */
2522          }
2523    
2524        /* If maximizing, find the longest string and work backwards */
2525    
2526        else
2527          {
2528          pp = eptr;
2529          for (i = min; i < max; i++)
2530            {
2531            int slength;
2532            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2533            {            {
2534            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            CHECK_PARTIAL();
2535            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2536            }            }
2537          RRETURN(MATCH_NOMATCH);          eptr += slength;
2538          }          }
2539          while (eptr >= pp)
2540            {
2541            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2542            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2543            eptr -= length;
2544            }
2545          MRRETURN(MATCH_NOMATCH);
2546        }        }
2547      /* Control never gets here */      /* Control never gets here */
2548    
   
   
2549      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2550      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,
2551      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 1618  for (;;) Line 2600  for (;;)
2600          {          {
2601          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2602            {            {
2603            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2604                {
2605                SCHECK_PARTIAL();
2606                MRRETURN(MATCH_NOMATCH);
2607                }
2608            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2609            if (c > 255)            if (c > 255)
2610              {              {
2611              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2612              }              }
2613            else            else
2614              {              {
2615              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2616              }              }
2617            }            }
2618          }          }
# Line 1636  for (;;) Line 2622  for (;;)
2622          {          {
2623          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2624            {            {
2625            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2626                {
2627                SCHECK_PARTIAL();
2628                MRRETURN(MATCH_NOMATCH);
2629                }
2630            c = *eptr++;            c = *eptr++;
2631            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2632            }            }
2633          }          }
2634    
# Line 1658  for (;;) Line 2648  for (;;)
2648            {            {
2649            for (fi = min;; fi++)            for (fi = min;; fi++)
2650              {              {
2651              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2652              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2653              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2654                if (eptr >= md->end_subject)
2655                  {
2656                  SCHECK_PARTIAL();
2657                  MRRETURN(MATCH_NOMATCH);
2658                  }
2659              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2660              if (c > 255)              if (c > 255)
2661                {                {
2662                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2663                }                }
2664              else              else
2665                {                {
2666                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2667                }                }
2668              }              }
2669            }            }
# Line 1678  for (;;) Line 2673  for (;;)
2673            {            {
2674            for (fi = min;; fi++)            for (fi = min;; fi++)
2675              {              {
2676              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2677              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2678              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2679                if (eptr >= md->end_subject)
2680                  {
2681                  SCHECK_PARTIAL();
2682                  MRRETURN(MATCH_NOMATCH);
2683                  }
2684              c = *eptr++;              c = *eptr++;
2685              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2686              }              }
2687            }            }
2688          /* Control never gets here */          /* Control never gets here */
# Line 1701  for (;;) Line 2701  for (;;)
2701            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2702              {              {
2703              int len = 1;              int len = 1;
2704              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2705                  {
2706                  SCHECK_PARTIAL();
2707                  break;
2708                  }
2709              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2710              if (c > 255)              if (c > 255)
2711                {                {
# Line 1715  for (;;) Line 2719  for (;;)
2719              }              }
2720            for (;;)            for (;;)
2721              {              {
2722              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2723              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2724              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2725              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1727  for (;;) Line 2731  for (;;)
2731            {            {
2732            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2733              {              {
2734              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2735                  {
2736                  SCHECK_PARTIAL();
2737                  break;
2738                  }
2739              c = *eptr;              c = *eptr;
2740              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2741              eptr++;              eptr++;
2742              }              }
2743            while (eptr >= pp)            while (eptr >= pp)
2744              {              {
2745              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2746              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2747              eptr--;              eptr--;
2748              }              }
2749            }            }
2750    
2751          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2752          }          }
2753        }        }
2754      /* Control never gets here */      /* Control never gets here */
2755    
2756    
2757      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2758      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
2759        mode, because Unicode properties are supported in non-UTF-8 mode. */
2760    
2761  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2762      case OP_XCLASS:      case OP_XCLASS:
# Line 1788  for (;;) Line 2797  for (;;)
2797    
2798        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2799          {          {
2800          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2801          GETCHARINC(c, eptr);            {
2802          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2803              MRRETURN(MATCH_NOMATCH);
2804              }
2805            GETCHARINCTEST(c, eptr);
2806            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2807          }          }
2808    
2809        /* 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 1805  for (;;) Line 2818  for (;;)
2818          {          {
2819          for (fi = min;; fi++)          for (fi = min;; fi++)
2820            {            {
2821            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2822            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2823            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2824            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2825            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2826                SCHECK_PARTIAL();
2827                MRRETURN(MATCH_NOMATCH);
2828                }
2829              GETCHARINCTEST(c, eptr);
2830              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2831            }            }
2832          /* Control never gets here */          /* Control never gets here */
2833          }          }
# Line 1822  for (;;) Line 2840  for (;;)
2840          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2841            {            {
2842            int len = 1;            int len = 1;
2843            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2844            GETCHARLEN(c, eptr, len);              {
2845                SCHECK_PARTIAL();
2846                break;
2847                }
2848              GETCHARLENTEST(c, eptr, len);
2849            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2850            eptr += len;            eptr += len;
2851            }            }
2852          for(;;)          for(;;)
2853            {            {
2854            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2855            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2856            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2857            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2858            }            }
2859          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2860          }          }
2861    
2862        /* Control never gets here */        /* Control never gets here */
# Line 1850  for (;;) Line 2872  for (;;)
2872        length = 1;        length = 1;
2873        ecode++;        ecode++;
2874        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2875        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2876        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2877            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2878            MRRETURN(MATCH_NOMATCH);
2879            }
2880          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2881        }        }
2882      else      else
2883  #endif  #endif
2884    
2885      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2886        {        {
2887        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2888        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2889            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2890            MRRETURN(MATCH_NOMATCH);
2891            }
2892          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2893        ecode += 2;        ecode += 2;
2894        }        }
2895      break;      break;
2896    
2897      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2898    
2899      case OP_CHARNC:      case OP_CHARI:
2900  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2901      if (utf8)      if (utf8)
2902        {        {
# Line 1874  for (;;) Line 2904  for (;;)
2904        ecode++;        ecode++;
2905        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2906    
2907        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2908            {
2909            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2910            MRRETURN(MATCH_NOMATCH);
2911            }
2912    
2913        /* 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
2914        can use the fast lookup table. */        can use the fast lookup table. */
2915    
2916        if (fc < 128)        if (fc < 128)
2917          {          {
2918          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2919          }          }
2920    
2921        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
2922    
2923        else        else
2924          {          {
2925          int dc;          unsigned int dc;
2926          GETCHARINC(dc, eptr);          GETCHARINC(dc, eptr);
2927          ecode += length;          ecode += length;
2928    
# Line 1898  for (;;) Line 2932  for (;;)
2932          if (fc != dc)          if (fc != dc)
2933            {            {
2934  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2935            if (dc != _pcre_ucp_othercase(fc))            if (dc != UCD_OTHERCASE(fc))
2936  #endif  #endif
2937              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
2938            }            }
2939          }          }
2940        }        }
# Line 1909  for (;;) Line 2943  for (;;)
2943    
2944      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2945        {        {
2946        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2947        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
2948            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2949            MRRETURN(MATCH_NOMATCH);
2950            }
2951          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2952        ecode += 2;        ecode += 2;
2953        }        }
2954      break;      break;
2955    
2956      /* Match a single character repeatedly; different opcodes share code. */      /* Match a single character repeatedly. */
2957    
2958      case OP_EXACT:      case OP_EXACT:
2959        case OP_EXACTI:
2960      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2961      ecode += 3;      ecode += 3;
2962      goto REPEATCHAR;      goto REPEATCHAR;
2963    
2964        case OP_POSUPTO:
2965        case OP_POSUPTOI:
2966        possessive = TRUE;
2967        /* Fall through */
2968    
2969      case OP_UPTO:      case OP_UPTO:
2970        case OP_UPTOI:
2971      case OP_MINUPTO:      case OP_MINUPTO:
2972        case OP_MINUPTOI:
2973      min = 0;      min = 0;
2974      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2975      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2976      ecode += 3;      ecode += 3;
2977      goto REPEATCHAR;      goto REPEATCHAR;
2978    
2979        case OP_POSSTAR:
2980        case OP_POSSTARI:
2981        possessive = TRUE;
2982        min = 0;
2983        max = INT_MAX;
2984        ecode++;
2985        goto REPEATCHAR;
2986    
2987        case OP_POSPLUS:
2988        case OP_POSPLUSI:
2989        possessive = TRUE;
2990        min = 1;
2991        max = INT_MAX;
2992        ecode++;
2993        goto REPEATCHAR;
2994    
2995        case OP_POSQUERY:
2996        case OP_POSQUERYI:
2997        possessive = TRUE;
2998        min = 0;
2999        max = 1;
3000        ecode++;
3001        goto REPEATCHAR;
3002    
3003      case OP_STAR:      case OP_STAR:
3004        case OP_STARI:
3005      case OP_MINSTAR:      case OP_MINSTAR:
3006        case OP_MINSTARI:
3007      case OP_PLUS:      case OP_PLUS:
3008        case OP_PLUSI:
3009      case OP_MINPLUS:      case OP_MINPLUS:
3010        case OP_MINPLUSI:
3011      case OP_QUERY:      case OP_QUERY:
3012        case OP_QUERYI:
3013      case OP_MINQUERY:      case OP_MINQUERY:
3014      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3015        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3016      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3017      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3018      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3019      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3020    
3021      /* 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. */  
3022    
3023      REPEATCHAR:      REPEATCHAR:
3024  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1953  for (;;) Line 3027  for (;;)
3027        length = 1;        length = 1;
3028        charptr = ecode;        charptr = ecode;
3029        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3030        ecode += length;        ecode += length;
3031    
3032        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 1961  for (;;) Line 3034  for (;;)
3034    
3035        if (length > 1)        if (length > 1)
3036          {          {
         int oclength = 0;  
         uschar occhars[8];  
   
3037  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3038          int othercase;          unsigned int othercase;
3039          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3040              (othercase = _pcre_ucp_othercase(fc)) >= 0 &&              (othercase = UCD_OTHERCASE(fc)) != fc)
              othercase >= 0)  
3041            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3042            else oclength = 0;
3043  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3044    
3045          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3046            {            {
3047            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3048            /* Need braces because of following else */              memcmp(eptr, charptr, length) == 0) eptr += length;
3049            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
3050              else if (oclength > 0 &&
3051                       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              }              }
3059            }            }
3060    
# Line 1990  for (;;) Line 3064  for (;;)
3064            {            {
3065            for (fi = min;; fi++)            for (fi = min;; fi++)
3066              {              {
3067              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3068              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3069              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3070              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3071              /* Need braces because of following else */                memcmp(eptr, charptr, length) == 0) eptr += length;
3072              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
3073                else if (oclength > 0 &&
3074                         eptr <= md->end_subject - oclength &&
3075                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3076    #endif  /* SUPPORT_UCP */
3077              else              else
3078                {                {
3079                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3080                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3081                }                }
3082              }              }
3083            /* Control never gets here */            /* Control never gets here */
3084            }            }
3085          else  
3086            else  /* Maximize */
3087            {            {
3088            pp = eptr;            pp = eptr;
3089            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3090              {              {
3091              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3092              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3093              else if (oclength == 0) break;  #ifdef SUPPORT_UCP
3094                else if (oclength > 0 &&
3095                         eptr <= md->end_subject - oclength &&
3096                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3097    #endif  /* SUPPORT_UCP */
3098              else              else
3099                {                {
3100                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3101                eptr += oclength;                break;
3102                }                }
3103              }              }
3104            while (eptr >= pp)  
3105             {            if (possessive) continue;
3106             RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);  
3107             if (rrc != MATCH_NOMATCH) RRETURN(rrc);            for(;;)
3108             eptr -= length;              {
3109             }              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3110            RRETURN(MATCH_NOMATCH);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3111                if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3112    #ifdef SUPPORT_UCP
3113                eptr--;
3114                BACKCHAR(eptr);
3115    #else   /* without SUPPORT_UCP */
3116                eptr -= length;
3117    #endif  /* SUPPORT_UCP */
3118                }
3119            }            }
3120          /* Control never gets here */          /* Control never gets here */
3121          }          }
# Line 2037  for (;;) Line 3128  for (;;)
3128  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3129    
3130      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3131        {  
3132        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3133    
3134      /* 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
3135      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 2054  for (;;) Line 3143  for (;;)
3143      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3144        max, eptr));        max, eptr));
3145    
3146      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3147        {        {
3148        fc = md->lcc[fc];        fc = md->lcc[fc];
3149        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3150          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3151            if (eptr >= md->end_subject)
3152              {
3153              SCHECK_PARTIAL();
3154              MRRETURN(MATCH_NOMATCH);
3155              }
3156            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3157            }
3158        if (min == max) continue;        if (min == max) continue;
3159        if (minimize)        if (minimize)
3160          {          {
3161          for (fi = min;; fi++)          for (fi = min;; fi++)
3162            {            {
3163            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3164            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3165            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3166                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3167              RRETURN(MATCH_NOMATCH);              {
3168                SCHECK_PARTIAL();
3169                MRRETURN(MATCH_NOMATCH);
3170                }
3171              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3172            }            }
3173          /* Control never gets here */          /* Control never gets here */
3174          }          }
3175        else        else  /* Maximize */
3176          {          {
3177          pp = eptr;          pp = eptr;
3178          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3179            {            {
3180            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3181                {
3182                SCHECK_PARTIAL();
3183                break;
3184                }
3185              if (fc != md->lcc[*eptr]) break;
3186            eptr++;            eptr++;
3187            }            }
3188    
3189            if (possessive) continue;
3190    
3191          while (eptr >= pp)          while (eptr >= pp)
3192            {            {
3193            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3194            eptr--;            eptr--;
3195            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3196            }            }
3197          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3198          }          }
3199        /* Control never gets here */        /* Control never gets here */
3200        }        }
# Line 2095  for (;;) Line 3203  for (;;)
3203    
3204      else      else
3205        {        {
3206        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3207            {
3208            if (eptr >= md->end_subject)
3209              {
3210              SCHECK_PARTIAL();
3211              MRRETURN(MATCH_NOMATCH);
3212              }
3213            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3214            }
3215    
3216        if (min == max) continue;        if (min == max) continue;
3217    
3218        if (minimize)        if (minimize)
3219          {          {
3220          for (fi = min;; fi++)          for (fi = min;; fi++)
3221            {            {
3222            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3223            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3224            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3225              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3226                {
3227                SCHECK_PARTIAL();
3228                MRRETURN(MATCH_NOMATCH);
3229                }
3230              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3231            }            }
3232          /* Control never gets here */          /* Control never gets here */
3233          }          }
3234        else        else  /* Maximize */
3235          {          {
3236          pp = eptr;          pp = eptr;
3237          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3238            {            {
3239            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3240                {
3241                SCHECK_PARTIAL();
3242                break;
3243                }
3244              if (fc != *eptr) break;
3245            eptr++;            eptr++;
3246            }            }
3247            if (possessive) continue;
3248    
3249          while (eptr >= pp)          while (eptr >= pp)
3250            {            {
3251            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3252            eptr--;            eptr--;
3253            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3254            }            }
3255          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3256          }          }
3257        }        }
3258      /* Control never gets here */      /* Control never gets here */
# Line 2131  for (;;) Line 3261  for (;;)
3261      checking can be multibyte. */      checking can be multibyte. */
3262    
3263      case OP_NOT:      case OP_NOT:
3264      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3265        if (eptr >= md->end_subject)
3266          {
3267          SCHECK_PARTIAL();
3268          MRRETURN(MATCH_NOMATCH);
3269          }
3270      ecode++;      ecode++;
3271      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3272      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3273        {        {
3274  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3275        if (c < 256)        if (c < 256)
3276  #endif  #endif
3277        c = md->lcc[c];        c = md->lcc[c];
3278        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3279        }        }
3280      else      else    /* Caseful */
3281        {        {
3282        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3283        }        }
3284      break;      break;
3285    
# Line 2156  for (;;) Line 3291  for (;;)
3291      about... */      about... */
3292    
3293      case OP_NOTEXACT:      case OP_NOTEXACT:
3294        case OP_NOTEXACTI:
3295      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3296      ecode += 3;      ecode += 3;
3297      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3298    
3299      case OP_NOTUPTO:      case OP_NOTUPTO:
3300        case OP_NOTUPTOI:
3301      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3302        case OP_NOTMINUPTOI:
3303        min = 0;
3304        max = GET2(ecode, 1);
3305        minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3306        ecode += 3;
3307        goto REPEATNOTCHAR;
3308    
3309        case OP_NOTPOSSTAR:
3310        case OP_NOTPOSSTARI:
3311        possessive = TRUE;
3312        min = 0;
3313        max = INT_MAX;
3314        ecode++;
3315        goto REPEATNOTCHAR;
3316    
3317        case OP_NOTPOSPLUS:
3318        case OP_NOTPOSPLUSI:
3319        possessive = TRUE;
3320        min = 1;
3321        max = INT_MAX;
3322        ecode++;
3323        goto REPEATNOTCHAR;
3324    
3325        case OP_NOTPOSQUERY:
3326        case OP_NOTPOSQUERYI:
3327        possessive = TRUE;
3328        min = 0;
3329        max = 1;
3330        ecode++;
3331        goto REPEATNOTCHAR;
3332    
3333        case OP_NOTPOSUPTO:
3334        case OP_NOTPOSUPTOI:
3335        possessive = TRUE;
3336      min = 0;      min = 0;
3337      max = GET2(ecode, 1);      max = GET2(ecode, 1);
     minimize = *ecode == OP_NOTMINUPTO;  
3338      ecode += 3;      ecode += 3;
3339      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3340    
3341      case OP_NOTSTAR:      case OP_NOTSTAR:
3342        case OP_NOTSTARI:
3343      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3344        case OP_NOTMINSTARI:
3345      case OP_NOTPLUS:      case OP_NOTPLUS:
3346        case OP_NOTPLUSI:
3347      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3348        case OP_NOTMINPLUSI:
3349      case OP_NOTQUERY:      case OP_NOTQUERY:
3350        case OP_NOTQUERYI:
3351      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3352      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3353        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3354      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3355      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3356      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3357      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3358    
3359      /* 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. */  
3360    
3361      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3362      fc = *ecode++;      fc = *ecode++;
3363    
3364      /* 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 2199  for (;;) Line 3372  for (;;)
3372      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,
3373        max, eptr));        max, eptr));
3374    
3375      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3376        {        {
3377        fc = md->lcc[fc];        fc = md->lcc[fc];
3378    
# Line 2207  for (;;) Line 3380  for (;;)
3380        /* UTF-8 mode */        /* UTF-8 mode */
3381        if (utf8)        if (utf8)
3382          {          {
3383          register int d;          register unsigned int d;
3384          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3385            {            {
3386              if (eptr >= md->end_subject)
3387                {
3388                SCHECK_PARTIAL();
3389                MRRETURN(MATCH_NOMATCH);
3390                }
3391            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3392            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3393            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3394            }            }
3395          }          }
3396        else        else
# Line 2221  for (;;) Line 3399  for (;;)
3399        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3400          {          {
3401          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3402            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3403              if (eptr >= md->end_subject)
3404                {
3405                SCHECK_PARTIAL();
3406                MRRETURN(MATCH_NOMATCH);
3407                }
3408              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3409              }
3410          }          }
3411    
3412        if (min == max) continue;        if (min == max) continue;
# Line 2232  for (;;) Line 3417  for (;;)
3417          /* UTF-8 mode */          /* UTF-8 mode */
3418          if (utf8)          if (utf8)
3419            {            {
3420            register int d;            register unsigned int d;
3421            for (fi = min;; fi++)            for (fi = min;; fi++)
3422              {              {
3423              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3424              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3425                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3426                if (eptr >= md->end_subject)
3427                  {
3428                  SCHECK_PARTIAL();
3429                  MRRETURN(MATCH_NOMATCH);
3430                  }
3431              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3432              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3433              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3434              }              }
3435            }            }
3436          else          else
# Line 2249  for (;;) Line 3439  for (;;)
3439            {            {
3440            for (fi = min;; fi++)            for (fi = min;; fi++)
3441              {              {
3442              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3443              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3444              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3445                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3446                  {
3447                  SCHECK_PARTIAL();
3448                  MRRETURN(MATCH_NOMATCH);
3449                  }
3450                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3451              }              }
3452            }            }
3453          /* Control never gets here */          /* Control never gets here */
# Line 2268  for (;;) Line 3463  for (;;)
3463          /* UTF-8 mode */          /* UTF-8 mode */
3464          if (utf8)          if (utf8)
3465            {            {
3466            register int d;            register unsigned int d;
3467            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3468              {              {
3469              int len = 1;              int len = 1;
3470              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3471                  {
3472                  SCHECK_PARTIAL();
3473                  break;
3474                  }
3475              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3476              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3477              if (fc == d) break;              if (fc == d) break;
3478              eptr += len;              eptr += len;
3479              }              }
3480            for(;;)          if (possessive) continue;
3481            for(;;)
3482              {              {
3483              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3484              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3485              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3486              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2292  for (;;) Line 3492  for (;;)
3492            {            {
3493            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3494              {              {
3495              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3496                  {
3497                  SCHECK_PARTIAL();
3498                  break;
3499                  }
3500                if (fc == md->lcc[*eptr]) break;
3501              eptr++;              eptr++;
3502              }              }
3503              if (possessive) continue;
3504            while (eptr >= pp)            while (eptr >= pp)
3505              {              {
3506              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3507              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3508              eptr--;              eptr--;
3509              }              }
3510            }            }
3511    
3512          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3513          }          }
3514        /* Control never gets here */        /* Control never gets here */
3515        }        }
# Line 2316  for (;;) Line 3522  for (;;)
3522        /* UTF-8 mode */        /* UTF-8 mode */
3523        if (utf8)        if (utf8)
3524          {          {
3525          register int d;          register unsigned int d;
3526          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3527            {            {
3528              if (eptr >= md->end_subject)
3529                {
3530                SCHECK_PARTIAL();
3531                MRRETURN(MATCH_NOMATCH);
3532                }
3533            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3534            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3535            }            }
3536          }          }
3537        else        else
# Line 2328  for (;;) Line 3539  for (;;)
3539        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3540          {          {
3541          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3542            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3543              if (eptr >= md->end_subject)