/[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 96 by nigel, Fri Mar 2 13:10:43 2007 UTC revision 609 by ph10, Wed Jun 15 18:09:23 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 */  #define NLBLOCK md             /* Block containing newline information */
50  #define PSSTART start_subject  /* Field containing processed string start */  #define PSSTART start_subject  /* Field containing processed string start */
51  #define PSEND   end_subject    /* Field containing processed string end */  #define PSEND   end_subject    /* Field containing processed string end */
52    
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55  /* The chain of eptrblocks for tail recursions uses memory in stack workspace,  /* Undefine some potentially clashing cpp symbols */
 obtained at top level, the size of which is defined by EPTR_WORK_SIZE. */  
56    
57  #define EPTR_WORK_SIZE (1000)  #undef min
58    #undef max
59    
60  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
61    of call to match(). We do it this way to save on using another stack variable,
62    as stack usage is to be discouraged. */
63    
64  #define match_condassert     0x01  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
65  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
 #define match_tail_recursed  0x04  /* Tail recursive call */  
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 65  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 78  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 111  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 143  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 183  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};
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 205  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 266  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;
   int Xflags;  
362    unsigned int Xrdepth;    unsigned 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    
# Line 290  typedef struct heapframe { Line 379  typedef struct heapframe {
379    BOOL Xcondition;    BOOL Xcondition;
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 299  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 Xoclength;
390      uschar Xocchars[8];
391  #endif  #endif
392    
393      int Xcodelink;
394    int Xctype;    int Xctype;
395    unsigned int Xfc;    unsigned int Xfc;
396    int Xfi;    int Xfi;
# Line 316  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 337  typedef struct heapframe { Line 426  typedef struct heapframe {
426    
427  /* This function is called recursively in many circumstances. Whenever it  /* 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 to current character in subject     eptr        pointer to current character in subject
461     ecode       pointer to current position in compiled 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_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
                  match_tail_recursed - this is a tail_recursed 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, unsigned 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 some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 380  register unsigned int c;   /* Character Line 490  register unsigned int c;   /* Character
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 */  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 387  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 408  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 420  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 432  HEAP_RECURSE: Line 546  HEAP_RECURSE:
546  #define condition          frame->Xcondition  #define condition          frame->Xcondition
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 441  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 oclength           frame->Xoclength
557    #define occhars            frame->Xocchars
558  #endif  #endif
559    
560  #define ctype              frame->Xctype  #define ctype              frame->Xctype
# Line 468  i, and fc and c, can be the same variabl Line 582  i, and fc and c, can be the same variabl
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;
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 494  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 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 510  int stacksave[REC_STACK_SAVE_MAX]; Line 632  int stacksave[REC_STACK_SAVE_MAX];
632  eptrblock newptrb;  eptrblock newptrb;
633  #endif     /* NO_RECURSE */  #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. */
648    
# Line 529  TAIL_RECURSE: Line 662  TAIL_RECURSE:
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    
 /* First check that we haven't called match() too many times, or that we  
 haven't exceeded the recursive call limit. */  
   
 if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  
 if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  
   
 original_ims = ims;    /* Save for resetting on ')' */  
   
670  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
671  utf8 = md->utf8;       /* Local copy of the flag */  utf8 = md->utf8;       /* Local copy of the flag */
672  #else  #else
673  utf8 = FALSE;  utf8 = FALSE;
674  #endif  #endif
675    
676    /* First check that we haven't called match() too many times, or that we
677    haven't exceeded the recursive call limit. */
678    
679    if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
680    if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
681    
682  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
683  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
684  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
685  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
686  When match() is called in other circumstances, don't add to the chain. If this  
687  is a tail recursion, use a block from the workspace, as the one on the stack is  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
688  already used. */  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_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
695    {    {
696    eptrblock *p;    newptrb.epb_saved_eptr = eptr;
697    if ((flags & match_tail_recursed) != 0)    newptrb.epb_prev = eptrb;
698      {    eptrb = &newptrb;
699      if (md->eptrn >= EPTR_WORK_SIZE) RRETURN(PCRE_ERROR_NULLWSLIMIT);    md->match_function_type = 0;
     p = md->eptrchain + md->eptrn++;  
     }  
   else p = &newptrb;  
   p->epb_saved_eptr = eptr;  
   p->epb_prev = eptrb;  
   eptrb = p;  
700    }    }
701    
702  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 576  for (;;) Line 705  for (;;)
705    {    {
706    minimize = possessive = FALSE;    minimize = possessive = FALSE;
707    op = *ecode;    op = *ecode;
708    
   /* For partial matching, remember if we ever hit the end of the subject after  
   matching at least one subject character. */  
   
   if (md->partial &&  
       eptr >= md->end_subject &&  
       eptr > md->start_match)  
     md->hitend = TRUE;  
   
709    switch(op)    switch(op)
710      {      {
711      /* Handle a capturing bracket. If there is space in the offset vector, save      case OP_MARK:
712      the current subject position in the working slot at the top of the vector.      markptr = ecode + 2;
713      We mustn't change the current values of the data slot, because they may be      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
714      set from a previous iteration of this group, and be referred to by a        eptrb, RM55);
715      reference inside the group.  
716        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
717      If the bracket fails to match, we need to restore this value and also the      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        /* PRUNE overrides THEN */
748    
749        case OP_PRUNE:
750        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
751          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        case OP_SKIP_ARG:
773        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        /* Pass back the current skip name by overloading md->start_match_ptr and
779        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      values of the final offsets, in case they were set by a previous iteration
813      of the same bracket.      of the same bracket.
814    
# Line 605  for (;;) Line 820  for (;;)
820      case OP_SCBRA:      case OP_SCBRA:
821      number = GET2(ecode, 1+LINK_SIZE);      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\n", number);      printf("start bracket %d\n", number);
826      printf("subject=");      printf("subject=");
827      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 621  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        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
843          {          {
844          RMATCH(rrc, eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
845            ims, eptrb, flags);          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 640  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 == NULL) md->mark = markptr;
862        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
863        }        }
864    
865      /* Insufficient room for saving captured contents. Treat as a non-capturing      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
866      bracket. */      as a non-capturing bracket. */
867    
868        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
869        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870    
871      DPRINTF(("insufficient capture room: treat as non-capturing\n"));      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
872    
873      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
874      final alternative within the brackets, we would return the result of a      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
875      recursive call to match() whatever happened. We can reduce stack usage by  
876      turning this into a tail recursion. */      /* Non-capturing bracket, except for possessive with unlimited repeat. Loop
877        for all the alternatives. When we get to the final alternative within the
878        brackets, we used to return the result of a recursive call to match()
879        whatever happened so it was possible to reduce stack usage by turning this
880        into a tail recursion, except in the case of a possibly empty group.
881        However, now that there is the possiblity of (*THEN) occurring in the final
882        alternative, this optimization is no longer possible. */
883    
884      case OP_BRA:      case OP_BRA:
885      case OP_SBRA:      case OP_SBRA:
886      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
     flags = (op >= OP_SBRA)? match_cbegroup : 0;  
887      for (;;)      for (;;)
888        {        {
889        if (ecode[GET(ecode, 1)] != OP_ALT)        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
890          {        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
891          ecode += _pcre_OP_lengths[*ecode];          RM2);
892          flags |= match_tail_recursed;        if (rrc != MATCH_NOMATCH &&
893          DPRINTF(("bracket 0 tail recursion\n"));            (rrc != MATCH_THEN || md->start_match_ptr != ecode))
894          goto TAIL_RECURSE;          RRETURN(rrc);
895          ecode += GET(ecode, 1);
896          if (*ecode != OP_ALT) break;
897          }
898    
899        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
900        RRETURN(MATCH_NOMATCH);
901    
902        /* Handle possessive capturing brackets with an unlimited repeat. We come
903        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
904        handled similarly to the normal case above. However, the matching is
905        different. The end of these brackets will always be OP_KETRPOS, which
906        returns MATCH_KETRPOS without going further in the pattern. By this means
907        we can handle the group by iteration rather than recursion, thereby
908        reducing the amount of stack needed. */
909    
910        case OP_CBRAPOS:
911        case OP_SCBRAPOS:
912        allow_zero = FALSE;
913    
914        POSSESSIVE_CAPTURE:
915        number = GET2(ecode, 1+LINK_SIZE);
916        offset = number << 1;
917    
918    #ifdef PCRE_DEBUG
919        printf("start possessive bracket %d\n", number);
920        printf("subject=");
921        pchars(eptr, 16, TRUE, md);
922        printf("\n");
923    #endif
924    
925        if (offset < md->offset_max)
926          {
927          matched_once = FALSE;
928          code_offset = ecode - md->start_code;
929    
930          save_offset1 = md->offset_vector[offset];
931          save_offset2 = md->offset_vector[offset+1];
932          save_offset3 = md->offset_vector[md->offset_end - number];
933          save_capture_last = md->capture_last;
934    
935          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
936    
937          /* Each time round the loop, save the current subject position for use
938          when the group matches. For MATCH_MATCH, the group has matched, so we
939          restart it with a new subject starting position, remembering that we had
940          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
941          usual. If we haven't matched any alternatives in any iteration, check to
942          see if a previous iteration matched. If so, the group has matched;
943          continue from afterwards. Otherwise it has failed; restore the previous
944          capture values before returning NOMATCH. */
945    
946          for (;;)
947            {
948            md->offset_vector[md->offset_end - number] =
949              (int)(eptr - md->start_subject);
950            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
951            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
952              eptrb, RM63);
953            if (rrc == MATCH_KETRPOS)
954              {
955              offset_top = md->end_offset_top;
956              eptr = md->end_match_ptr;
957              ecode = md->start_code + code_offset;
958              save_capture_last = md->capture_last;
959              matched_once = TRUE;
960              continue;
961              }
962            if (rrc != MATCH_NOMATCH &&
963                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
964              RRETURN(rrc);
965            md->capture_last = save_capture_last;
966            ecode += GET(ecode, 1);
967            if (*ecode != OP_ALT) break;
968          }          }
969    
970          if (!matched_once)
971            {
972            md->offset_vector[offset] = save_offset1;
973            md->offset_vector[offset+1] = save_offset2;
974            md->offset_vector[md->offset_end - number] = save_offset3;
975            }
976    
977          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
978          if (allow_zero || matched_once)
979            {
980            ecode += 1 + LINK_SIZE;
981            break;
982            }
983    
984          RRETURN(MATCH_NOMATCH);
985          }
986    
987        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
988        as a non-capturing bracket. */
989    
990        /* For non-final alternatives, continue the loop for a NOMATCH result;      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
991        otherwise return. */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
992    
993        RMATCH(rrc, eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
994          eptrb, flags);  
995        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
996        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
997    
998        /* Non-capturing possessive bracket with unlimited repeat. We come here
999        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1000        without the capturing complication. It is written out separately for speed
1001        and cleanliness. */
1002    
1003        case OP_BRAPOS:
1004        case OP_SBRAPOS:
1005        allow_zero = FALSE;
1006    
1007        POSSESSIVE_NON_CAPTURE:
1008        matched_once = FALSE;
1009        code_offset = ecode - md->start_code;
1010    
1011        for (;;)
1012          {
1013          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1014          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1015            eptrb, RM48);
1016          if (rrc == MATCH_KETRPOS)
1017            {
1018            eptr = md->end_match_ptr;
1019            ecode = md->start_code + code_offset;
1020            matched_once = TRUE;
1021            continue;
1022            }
1023          if (rrc != MATCH_NOMATCH &&
1024              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1025            RRETURN(rrc);
1026        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1027          if (*ecode != OP_ALT) break;
1028        }        }
1029    
1030        if (matched_once || allow_zero)
1031          {
1032          ecode += 1 + LINK_SIZE;
1033          break;
1034          }
1035        RRETURN(MATCH_NOMATCH);
1036    
1037      /* Control never reaches here. */      /* Control never reaches here. */
1038    
1039      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1040      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
1041      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
1042      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1043    
1044      case OP_COND:      case OP_COND:
1045      case OP_SCOND:      case OP_SCOND:
1046      if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */      codelink = GET(ecode, 1);
1047    
1048        /* Because of the way auto-callout works during compile, a callout item is
1049        inserted between OP_COND and an assertion condition. */
1050    
1051        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1052        {        {
1053        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (pcre_callout != NULL)
1054        condition = md->recursive != NULL &&          {
1055          (offset == RREF_ANY || offset == md->recursive->group_num);          pcre_callout_block cb;
1056        ecode += condition? 3 : GET(ecode, 1);          cb.version          = 1;   /* Version 1 of the callout block */
1057            cb.callout_number   = ecode[LINK_SIZE+2];
1058            cb.offset_vector    = md->offset_vector;
1059            cb.subject          = (PCRE_SPTR)md->start_subject;
1060            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1061            cb.start_match      = (int)(mstart - md->start_subject);
1062            cb.current_position = (int)(eptr - md->start_subject);
1063            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1064            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1065            cb.capture_top      = offset_top/2;
1066            cb.capture_last     = md->capture_last;
1067            cb.callout_data     = md->callout_data;
1068            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1069            if (rrc < 0) RRETURN(rrc);
1070            }
1071          ecode += _pcre_OP_lengths[OP_CALLOUT];
1072          }
1073    
1074        condcode = ecode[LINK_SIZE+1];
1075    
1076        /* Now see what the actual condition is */
1077    
1078        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1079          {
1080          if (md->recursive == NULL)                /* Not recursing => FALSE */
1081            {
1082            condition = FALSE;
1083            ecode += GET(ecode, 1);
1084            }
1085          else
1086            {
1087            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1088            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1089    
1090            /* If the test is for recursion into a specific subpattern, and it is
1091            false, but the test was set up by name, scan the table to see if the
1092            name refers to any other numbers, and test them. The condition is true
1093            if any one is set. */
1094    
1095            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1096              {
1097              uschar *slotA = md->name_table;
1098              for (i = 0; i < md->name_count; i++)
1099                {
1100                if (GET2(slotA, 0) == recno) break;
1101                slotA += md->name_entry_size;
1102                }
1103    
1104              /* Found a name for the number - there can be only one; duplicate
1105              names for different numbers are allowed, but not vice versa. First
1106              scan down for duplicates. */
1107    
1108              if (i < md->name_count)
1109                {
1110                uschar *slotB = slotA;
1111                while (slotB > md->name_table)
1112                  {
1113                  slotB -= md->name_entry_size;
1114                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1115                    {
1116                    condition = GET2(slotB, 0) == md->recursive->group_num;
1117                    if (condition) break;
1118                    }
1119                  else break;
1120                  }
1121    
1122                /* Scan up for duplicates */
1123    
1124                if (!condition)
1125                  {
1126                  slotB = slotA;
1127                  for (i++; i < md->name_count; i++)
1128                    {
1129                    slotB += md->name_entry_size;
1130                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1131                      {
1132                      condition = GET2(slotB, 0) == md->recursive->group_num;
1133                      if (condition) break;
1134                      }
1135                    else break;
1136                    }
1137                  }
1138                }
1139              }
1140    
1141            /* Chose branch according to the condition */
1142    
1143            ecode += condition? 3 : GET(ecode, 1);
1144            }
1145        }        }
1146    
1147      else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1148        {        {
1149        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1150        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1151    
1152          /* If the numbered capture is unset, but the reference was by name,
1153          scan the table to see if the name refers to any other numbers, and test
1154          them. The condition is true if any one is set. This is tediously similar
1155          to the code above, but not close enough to try to amalgamate. */
1156    
1157          if (!condition && condcode == OP_NCREF)
1158            {
1159            int refno = offset >> 1;
1160            uschar *slotA = md->name_table;
1161    
1162            for (i = 0; i < md->name_count; i++)
1163              {
1164              if (GET2(slotA, 0) == refno) break;
1165              slotA += md->name_entry_size;
1166              }
1167    
1168            /* Found a name for the number - there can be only one; duplicate names
1169            for different numbers are allowed, but not vice versa. First scan down
1170            for duplicates. */
1171    
1172            if (i < md->name_count)
1173              {
1174              uschar *slotB = slotA;
1175              while (slotB > md->name_table)
1176                {
1177                slotB -= md->name_entry_size;
1178                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1179                  {
1180                  offset = GET2(slotB, 0) << 1;
1181                  condition = offset < offset_top &&
1182                    md->offset_vector[offset] >= 0;
1183                  if (condition) break;
1184                  }
1185                else break;
1186                }
1187    
1188              /* Scan up for duplicates */
1189    
1190              if (!condition)
1191                {
1192                slotB = slotA;
1193                for (i++; i < md->name_count; i++)
1194                  {
1195                  slotB += md->name_entry_size;
1196                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1197                    {
1198                    offset = GET2(slotB, 0) << 1;
1199                    condition = offset < offset_top &&
1200                      md->offset_vector[offset] >= 0;
1201                    if (condition) break;
1202                    }
1203                  else break;
1204                  }
1205                }
1206              }
1207            }
1208    
1209          /* Chose branch according to the condition */
1210    
1211        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 3 : GET(ecode, 1);
1212        }        }
1213    
1214      else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
1215        {        {
1216        condition = FALSE;        condition = FALSE;
1217        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1218        }        }
1219    
1220      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1221      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1222      assertion. */      an assertion. */
1223    
1224      else      else
1225        {        {
1226        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1227            match_condassert);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1228        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1229          {          {
1230          condition = TRUE;          condition = TRUE;
1231          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1232          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1233          }          }
1234        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH &&
1235                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1236          {          {
1237          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1238          }          }
1239        else        else
1240          {          {
1241          condition = FALSE;          condition = FALSE;
1242          ecode += GET(ecode, 1);          ecode += codelink;
1243          }          }
1244        }        }
1245    
1246      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one,
1247      we can use tail recursion to avoid using another stack frame. If the second      we used to use tail recursion to avoid using another stack frame, except
1248      alternative doesn't exist, we can just plough on. */      when there was unlimited repeat of a possibly empty group. However, that
1249        strategy no longer works because of the possibilty of (*THEN) being
1250        encountered in the branch. A recursive call to match() is always required,
1251        unless the second alternative doesn't exist, in which case we can just
1252        plough on. */
1253    
1254      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1255        {        {
1256        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1257        flags = match_tail_recursed | ((op == OP_SCOND)? match_cbegroup : 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1258        goto TAIL_RECURSE;        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1259            rrc = MATCH_NOMATCH;
1260          RRETURN(rrc);
1261        }        }
1262      else      else                         /* Condition false & no alternative */
1263        {        {
1264        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1265        }        }
1266      break;      break;
1267    
1268    
1269      /* End of the pattern. If we are in a top-level recursion, we should      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1270      restore the offsets appropriately and continue from after the call. */      to close any currently open capturing brackets. */
1271    
1272        case OP_CLOSE:
1273        number = GET2(ecode, 1);
1274        offset = number << 1;
1275    
1276    #ifdef PCRE_DEBUG
1277          printf("end bracket %d at *ACCEPT", number);
1278          printf("\n");
1279    #endif
1280    
1281        md->capture_last = number;
1282        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1283          {
1284          md->offset_vector[offset] =
1285            md->offset_vector[md->offset_end - number];
1286          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1287          if (offset_top <= offset) offset_top = offset + 2;
1288          }
1289        ecode += 3;
1290        break;
1291    
1292    
1293        /* End of the pattern, either real or forced. If we are in a recursion, we
1294        should restore the offsets appropriately, and if it's a top-level
1295        recursion, continue from after the call. */
1296    
1297        case OP_ACCEPT:
1298      case OP_END:      case OP_END:
1299      if (md->recursive != NULL && md->recursive->group_num == 0)      if (md->recursive != NULL)
1300        {        {
1301        recursion_info *rec = md->recursive;        recursion_info *rec = md->recursive;
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
1302        md->recursive = rec->prevrec;        md->recursive = rec->prevrec;
1303        memmove(md->offset_vector, rec->offset_save,        memmove(md->offset_vector, rec->offset_save,
1304          rec->saved_max * sizeof(int));          rec->saved_max * sizeof(int));
1305        md->start_match = rec->save_start;        offset_top = rec->save_offset_top;
1306        ims = original_ims;        if (rec->group_num == 0)
1307        ecode = rec->after_call;          {
1308        break;          ecode = rec->after_call;
1309            break;
1310            }
1311        }        }
1312    
1313      /* 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
1314      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
1315        the subject. In both cases, backtracking will then try other alternatives,
1316        if any. */
1317    
1318        else if (eptr == mstart &&
1319            (md->notempty ||
1320              (md->notempty_atstart &&
1321                mstart == md->start_subject + md->start_offset)))
1322          MRRETURN(MATCH_NOMATCH);
1323    
1324        /* Otherwise, we have a match. */
1325    
1326        md->end_match_ptr = eptr;           /* Record where we ended */
1327        md->end_offset_top = offset_top;    /* and how many extracts were taken */
1328        md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1329    
1330      if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH);      /* For some reason, the macros don't work properly if an expression is
1331      md->end_match_ptr = eptr;          /* Record where we ended */      given as the argument to MRRETURN when the heap is in use. */
1332      md->end_offset_top = offset_top;   /* and how many extracts were taken */  
1333      RRETURN(MATCH_MATCH);      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1334        MRRETURN(rrc);
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1335    
1336      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1337      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,
1338      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
1339      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
1340      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1341        of a condition, we want to return immediately afterwards. The caller of
1342        this incarnation of the match() function will have set MATCH_CONDASSERT in
1343        md->match_function type, and one of these opcodes will be the first opcode
1344        that is processed. We use a local variable that is preserved over calls to
1345        match() to remember this case. */
1346    
1347      case OP_ASSERT:      case OP_ASSERT:
1348      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1349        if (md->match_function_type == MATCH_CONDASSERT)
1350          {
1351          condassert = TRUE;
1352          md->match_function_type = 0;
1353          }
1354        else condassert = FALSE;
1355    
1356      do      do
1357        {        {
1358        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1359        if (rrc == MATCH_MATCH) break;        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1360        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          {
1361            mstart = md->start_match_ptr;   /* In case \K reset it */
1362            break;
1363            }
1364          if (rrc != MATCH_NOMATCH &&
1365              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1366            RRETURN(rrc);
1367        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1368        }        }
1369      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1370      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1371        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1372    
1373      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1374    
1375      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1376    
1377      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1378      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 811  for (;;) Line 1382  for (;;)
1382      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1383      continue;      continue;
1384    
1385      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1386        PRUNE, or COMMIT means we must assume failure without checking subsequent
1387        branches. */
1388    
1389      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1390      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1391        if (md->match_function_type == MATCH_CONDASSERT)
1392          {
1393          condassert = TRUE;
1394          md->match_function_type = 0;
1395          }
1396        else condassert = FALSE;
1397    
1398      do      do
1399        {        {
1400        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1401        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1402        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1403            {
1404            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1405            break;
1406            }
1407          if (rrc != MATCH_NOMATCH &&
1408              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1409            RRETURN(rrc);
1410        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1411        }        }
1412      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1413    
1414      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1415    
1416      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1417      continue;      continue;
1418    
# Line 842  for (;;) Line 1429  for (;;)
1429        while (i-- > 0)        while (i-- > 0)
1430          {          {
1431          eptr--;          eptr--;
1432          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1433          BACKCHAR(eptr)          BACKCHAR(eptr);
1434          }          }
1435        }        }
1436      else      else
# Line 853  for (;;) Line 1440  for (;;)
1440    
1441        {        {
1442        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1443        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1444        }        }
1445    
1446      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1447    
1448        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1449      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1450      break;      break;
1451    
# Line 873  for (;;) Line 1461  for (;;)
1461        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1462        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1463        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1464        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1465        cb.start_match      = md->start_match - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1466        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1467        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1468        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1469        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1470        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1471        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1472        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1473        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1474        }        }
1475      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 936  for (;;) Line 1524  for (;;)
1524    
1525        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1526              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1527        new_recursive.save_start = md->start_match;        new_recursive.save_offset_top = offset_top;
1528        md->start_match = eptr;  
   
1529        /* 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
1530        restore the offset and recursion data. */        restore the offset and recursion data. */
1531    
1532        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1533        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1534        do        do
1535          {          {
1536          RMATCH(rrc, eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1537            md, ims, eptrb, flags);          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1538          if (rrc == MATCH_MATCH)            md, eptrb, RM6);
1539            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1540            {            {
1541            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1542            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1543            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1544              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1545            RRETURN(MATCH_MATCH);            MRRETURN(MATCH_MATCH);
1546            }            }
1547          else if (rrc != MATCH_NOMATCH)          else if (rrc != MATCH_NOMATCH &&
1548                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1549            {            {
1550            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1551              if (new_recursive.offset_save != stacksave)
1552                (pcre_free)(new_recursive.offset_save);
1553            RRETURN(rrc);            RRETURN(rrc);
1554            }            }
1555    
# Line 973  for (;;) Line 1564  for (;;)
1564        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1565        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1566          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1567        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1568        }        }
1569      /* Control never reaches here */      /* Control never reaches here */
1570    
# Line 982  for (;;) Line 1573  for (;;)
1573      a move back into the brackets. Friedl calls these "atomic" subpatterns.      a move back into the brackets. Friedl calls these "atomic" subpatterns.
1574      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
1575      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
1576      the end of a normal bracket, leaving the subject pointer. */      the end of a normal bracket, leaving the subject pointer, but resetting
1577        the start-of-match value in case it was changed by \K. */
1578    
1579      case OP_ONCE:      case OP_ONCE:
1580      prev = ecode;      prev = ecode;
# Line 990  for (;;) Line 1582  for (;;)
1582    
1583      do      do
1584        {        {
1585        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1586          eptrb, 0);        if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
1587        if (rrc == MATCH_MATCH) break;          {
1588        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          mstart = md->start_match_ptr;
1589            break;
1590            }
1591          if (rrc != MATCH_NOMATCH &&
1592              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1593            RRETURN(rrc);
1594        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1595        }        }
1596      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
# Line 1024  for (;;) Line 1621  for (;;)
1621    
1622      /* 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
1623      preceding bracket, in the appropriate order. The second "call" of match()      preceding bracket, in the appropriate order. The second "call" of match()
1624      uses tail recursion, to avoid using another stack frame. We need to reset      uses tail recursion, to avoid using another stack frame. */
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
1625    
1626      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1627        {        {
1628        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM8);
1629        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1630        ecode = prev;        ecode = prev;
       flags = match_tail_recursed;  
1631        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1632        }        }
1633      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1634        {        {
1635        RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_cbegroup);        md->match_function_type = MATCH_CBEGROUP;
1636          RMATCH(eptr, prev, offset_top, md, eptrb, RM9);
1637        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1638        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = match_tail_recursed;  
1639        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1640        }        }
1641      /* Control never gets here */      /* Control never gets here */
# Line 1059  for (;;) Line 1647  for (;;)
1647      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1648      break;      break;
1649    
1650      /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1651      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
1652      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
1653      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
1654      preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1655    
1656      case OP_BRAZERO:      case OP_BRAZERO:
1657        {      next = ecode + 1;
1658        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1659        RMATCH(rrc, eptr, next, offset_top, md, ims, eptrb, 0);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1660        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1661        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1662      break;      break;
1663    
1664      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1665        {      next = ecode + 1;
1666        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1667        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1668        RMATCH(rrc, eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1669        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
1670        ecode++;      break;
1671        }  
1672        case OP_SKIPZERO:
1673        next = ecode+1;
1674        do next += GET(next,1); while (*next == OP_ALT);
1675        ecode = next + 1 + LINK_SIZE;
1676      break;      break;
1677    
1678        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1679        here; just jump to the group, with allow_zero set TRUE. */
1680    
1681        case OP_BRAPOSZERO:
1682        op = *(++ecode);
1683        allow_zero = TRUE;
1684        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1685          goto POSSESSIVE_NON_CAPTURE;
1686    
1687      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1688    
1689      case OP_KET:      case OP_KET:
1690      case OP_KETRMIN:      case OP_KETRMIN:
1691      case OP_KETRMAX:      case OP_KETRMAX:
1692        case OP_KETRPOS:
1693      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1694    
1695      /* If this was a group that remembered the subject start, in order to break      /* If this was a group that remembered the subject start, in order to break
# Line 1103  for (;;) Line 1703  for (;;)
1703        }        }
1704      else saved_eptr = NULL;      else saved_eptr = NULL;
1705    
1706      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group or an atomic group, stop
1707      MATCH_MATCH, but record the current high water mark for use by positive      matching and return MATCH_MATCH, but record the current high water mark for
1708      assertions. Do this also for the "once" (atomic) groups. */      use by positive assertions. We also need to record the match start in case
1709        it was changed by \K. */
1710    
1711      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1712          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||
# Line 1113  for (;;) Line 1714  for (;;)
1714        {        {
1715        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1716        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1717        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1718          MRRETURN(MATCH_MATCH);
1719        }        }
1720    
1721      /* For capturing groups we have to check the group number back at the start      /* For capturing groups we have to check the group number back at the start
# Line 1122  for (;;) Line 1724  for (;;)
1724      a recurse into group 0, so it won't be picked up here. Instead, we catch it      a recurse into group 0, so it won't be picked up here. Instead, we catch it
1725      when the OP_END is reached. Other recursion is handled here. */      when the OP_END is reached. Other recursion is handled here. */
1726    
1727      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1728            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1729        {        {
1730        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1731        offset = number << 1;        offset = number << 1;
1732    
1733  #ifdef DEBUG  #ifdef PCRE_DEBUG
1734        printf("end bracket %d", number);        printf("end bracket %d", number);
1735        printf("\n");        printf("\n");
1736  #endif  #endif
# Line 1137  for (;;) Line 1740  for (;;)
1740          {          {
1741          md->offset_vector[offset] =          md->offset_vector[offset] =
1742            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1743          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1744          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1745          }          }
1746    
# Line 1149  for (;;) Line 1752  for (;;)
1752          recursion_info *rec = md->recursive;          recursion_info *rec = md->recursive;
1753          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
1754          md->recursive = rec->prevrec;          md->recursive = rec->prevrec;
         md->start_match = rec->save_start;  
1755          memcpy(md->offset_vector, rec->offset_save,          memcpy(md->offset_vector, rec->offset_save,
1756            rec->saved_max * sizeof(int));            rec->saved_max * sizeof(int));
1757            offset_top = rec->save_offset_top;
1758          ecode = rec->after_call;          ecode = rec->after_call;
         ims = original_ims;  
1759          break;          break;
1760          }          }
1761        }        }
1762    
     /* For both capturing and non-capturing groups, reset the value of the ims  
     flags, in case they got changed during the group. */  
   
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
   
1763      /* For a non-repeating ket, just continue at this level. This also      /* For a non-repeating ket, just continue at this level. This also
1764      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.
1765      This is the forcible breaking of infinite loops as implemented in Perl      This is the forcible breaking of infinite loops as implemented in Perl
# Line 1175  for (;;) Line 1771  for (;;)
1771        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1772        break;        break;
1773        }        }
1774    
1775      /* The repeating kets try the rest of the pattern or restart from the      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1776      preceding bracket, in the appropriate order. In the second case, we can use      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1777      tail recursion to avoid using another stack frame. */      at a time from the outer level, thus saving stack. */
1778    
1779      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;      if (*ecode == OP_KETRPOS)
1780          {
1781          md->end_match_ptr = eptr;
1782          md->end_offset_top = offset_top;
1783          RRETURN(MATCH_KETRPOS);
1784          }
1785    
1786        /* The normal repeating kets try the rest of the pattern or restart from
1787        the preceding bracket, in the appropriate order. In the second case, we can
1788        use tail recursion to avoid using another stack frame, unless we have an
1789        unlimited repeat of a group that can match an empty string. */
1790    
1791      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1792        {        {
1793        RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1794        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1795          if (*prev >= OP_SBRA)    /* Could match an empty string */
1796            {
1797            md->match_function_type = MATCH_CBEGROUP;
1798            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1799            RRETURN(rrc);
1800            }
1801        ecode = prev;        ecode = prev;
       flags |= match_tail_recursed;  
1802        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1803        }        }
1804      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1805        {        {
1806        RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, flags);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1807          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1808        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1809        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = match_tail_recursed;  
1810        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1811        }        }
1812      /* Control never gets here */      /* Control never gets here */
1813    
1814      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1815    
1816      case OP_CIRC:      case OP_CIRC:
1817      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1818      if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         RRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
   
1819      /* Start of subject assertion */      /* Start of subject assertion */
1820    
1821      case OP_SOD:      case OP_SOD:
1822      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1823        ecode++;
1824        break;
1825    
1826        /* Multiline mode: start of subject unless notbol, or after any newline. */
1827    
1828        case OP_CIRCM:
1829        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1830        if (eptr != md->start_subject &&
1831            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1832          MRRETURN(MATCH_NOMATCH);
1833      ecode++;      ecode++;
1834      break;      break;
1835    
1836      /* Start of match assertion */      /* Start of match assertion */
1837    
1838      case OP_SOM:      case OP_SOM:
1839      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1840      ecode++;      ecode++;
1841      break;      break;
1842    
1843      /* 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. */  
1844    
1845      case OP_DOLL:      case OP_SET_SOM:
1846      if ((ims & PCRE_MULTILINE) != 0)      mstart = eptr;
1847        {      ecode++;
1848        if (eptr < md->end_subject)      break;
1849          { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }  
1850        else      /* Multiline mode: assert before any newline, or before end of subject
1851          { if (md->noteol) RRETURN(MATCH_NOMATCH); }      unless noteol is set. */
1852        ecode++;  
1853        break;      case OP_DOLLM:
1854        }      if (eptr < md->end_subject)
1855          { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1856      else      else
1857        {        {
1858        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1859        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1860        }        }
1861        ecode++;
1862        break;
1863    
1864        /* Not multiline mode: assert before a terminating newline or before end of
1865        subject unless noteol is set. */
1866    
1867        case OP_DOLL:
1868        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1869        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1870    
1871      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1872    
1873      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1874    
1875      case OP_EOD:      case OP_EOD:
1876      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1877        SCHECK_PARTIAL();
1878      ecode++;      ecode++;
1879      break;      break;
1880    
1881      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1882    
1883      case OP_EODN:      case OP_EODN:
1884      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1885        if (eptr < md->end_subject &&
1886          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1887        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1888    
1889        /* Either at end of string or \n before end. */
1890    
1891        SCHECK_PARTIAL();
1892      ecode++;      ecode++;
1893      break;      break;
1894    
# Line 1279  for (;;) Line 1900  for (;;)
1900    
1901        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1902        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
1903        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1904          partial matching. */
1905    
1906  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1907        if (utf8)        if (utf8)
1908          {          {
1909            /* Get status of previous character */
1910    
1911          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1912            {            {
1913            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1914            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1915              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1916            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1917    #ifdef SUPPORT_UCP
1918              if (md->use_ucp)
1919                {
1920                if (c == '_') prev_is_word = TRUE; else
1921                  {
1922                  int cat = UCD_CATEGORY(c);
1923                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1924                  }
1925                }
1926              else
1927    #endif
1928            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1929            }            }
1930          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
1931            /* Get status of next character */
1932    
1933            if (eptr >= md->end_subject)
1934              {
1935              SCHECK_PARTIAL();
1936              cur_is_word = FALSE;
1937              }
1938            else
1939            {            {
1940            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1941    #ifdef SUPPORT_UCP
1942              if (md->use_ucp)
1943                {
1944                if (c == '_') cur_is_word = TRUE; else
1945                  {
1946                  int cat = UCD_CATEGORY(c);
1947                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1948                  }
1949                }
1950              else
1951    #endif
1952            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1953            }            }
1954          }          }
1955        else        else
1956  #endif  #endif
1957    
1958        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1959          consistency with the behaviour of \w we do use it in this case. */
1960    
1961          {          {
1962          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
1963            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
1964          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
1965            ((md->ctypes[*eptr] & ctype_word) != 0);            {
1966              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1967    #ifdef SUPPORT_UCP
1968              if (md->use_ucp)
1969                {
1970                c = eptr[-1];
1971                if (c == '_') prev_is_word = TRUE; else
1972                  {
1973                  int cat = UCD_CATEGORY(c);
1974                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1975                  }
1976                }
1977              else
1978    #endif
1979              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1980              }
1981    
1982            /* Get status of next character */
1983    
1984            if (eptr >= md->end_subject)
1985              {
1986              SCHECK_PARTIAL();
1987              cur_is_word = FALSE;
1988              }
1989            else
1990    #ifdef SUPPORT_UCP
1991            if (md->use_ucp)
1992              {
1993              c = *eptr;
1994              if (c == '_') cur_is_word = TRUE; else
1995                {
1996                int cat = UCD_CATEGORY(c);
1997                cur_is_word = (cat == ucp_L || cat == ucp_N);
1998                }
1999              }
2000            else
2001    #endif
2002            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2003          }          }
2004    
2005        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2006    
2007        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2008             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2009          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2010        }        }
2011      break;      break;
2012    
2013      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2014    
2015      case OP_ANY:      case OP_ANY:
2016      if ((ims & PCRE_DOTALL) == 0)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2017        /* Fall through */
2018    
2019        case OP_ALLANY:
2020        if (eptr++ >= md->end_subject)
2021        {        {
2022        if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);        SCHECK_PARTIAL();
2023          MRRETURN(MATCH_NOMATCH);
2024        }        }
2025      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
     if (utf8)  
       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
2026      ecode++;      ecode++;
2027      break;      break;
2028    
# Line 1334  for (;;) Line 2030  for (;;)
2030      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2031    
2032      case OP_ANYBYTE:      case OP_ANYBYTE:
2033      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr++ >= md->end_subject)
2034          {
2035          SCHECK_PARTIAL();
2036          MRRETURN(MATCH_NOMATCH);
2037          }
2038      ecode++;      ecode++;
2039      break;      break;
2040    
2041      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2042      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2043          {
2044          SCHECK_PARTIAL();
2045          MRRETURN(MATCH_NOMATCH);
2046          }
2047      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2048      if (      if (
2049  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1347  for (;;) Line 2051  for (;;)
2051  #endif  #endif
2052         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2053         )         )
2054        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2055      ecode++;      ecode++;
2056      break;      break;
2057    
2058      case OP_DIGIT:      case OP_DIGIT:
2059      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2060          {
2061          SCHECK_PARTIAL();
2062          MRRETURN(MATCH_NOMATCH);
2063          }
2064      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2065      if (      if (
2066  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1360  for (;;) Line 2068  for (;;)
2068  #endif  #endif
2069         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2070         )         )
2071        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2072      ecode++;      ecode++;
2073      break;      break;
2074    
2075      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2076      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2077          {
2078          SCHECK_PARTIAL();
2079          MRRETURN(MATCH_NOMATCH);
2080          }
2081      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2082      if (      if (
2083  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1373  for (;;) Line 2085  for (;;)
2085  #endif  #endif
2086         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2087         )         )
2088        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2089      ecode++;      ecode++;
2090      break;      break;
2091    
2092      case OP_WHITESPACE:      case OP_WHITESPACE:
2093      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2094          {
2095          SCHECK_PARTIAL();
2096          MRRETURN(MATCH_NOMATCH);
2097          }
2098      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2099      if (      if (
2100  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1386  for (;;) Line 2102  for (;;)
2102  #endif  #endif
2103         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2104         )         )
2105        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2106      ecode++;      ecode++;
2107      break;      break;
2108    
2109      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2110      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2111          {
2112          SCHECK_PARTIAL();
2113          MRRETURN(MATCH_NOMATCH);
2114          }
2115      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2116      if (      if (
2117  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1399  for (;;) Line 2119  for (;;)
2119  #endif  #endif
2120         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2121         )         )
2122        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2123      ecode++;      ecode++;
2124      break;      break;
2125    
2126      case OP_WORDCHAR:      case OP_WORDCHAR:
2127      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2128          {
2129          SCHECK_PARTIAL();
2130          MRRETURN(MATCH_NOMATCH);
2131          }
2132      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2133      if (      if (
2134  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1412  for (;;) Line 2136  for (;;)
2136  #endif  #endif
2137         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2138         )         )
2139        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2140      ecode++;      ecode++;
2141      break;      break;
2142    
2143      case OP_ANYNL:      case OP_ANYNL:
2144      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2145          {
2146          SCHECK_PARTIAL();
2147          MRRETURN(MATCH_NOMATCH);
2148          }
2149      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2150      switch(c)      switch(c)
2151        {        {
2152        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2153    
2154        case 0x000d:        case 0x000d:
2155        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2156        break;        break;
2157    
2158        case 0x000a:        case 0x000a:
2159          break;
2160    
2161        case 0x000b:        case 0x000b:
2162        case 0x000c:        case 0x000c:
2163        case 0x0085:        case 0x0085:
2164        case 0x2028:        case 0x2028:
2165        case 0x2029:        case 0x2029:
2166          if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2167          break;
2168          }
2169        ecode++;
2170        break;
2171    
2172        case OP_NOT_HSPACE:
2173        if (eptr >= md->end_subject)
2174          {
2175          SCHECK_PARTIAL();
2176          MRRETURN(MATCH_NOMATCH);
2177          }
2178        GETCHARINCTEST(c, eptr);
2179        switch(c)
2180          {
2181          default: break;
2182          case 0x09:      /* HT */
2183          case 0x20:      /* SPACE */
2184          case 0xa0:      /* NBSP */
2185          case 0x1680:    /* OGHAM SPACE MARK */
2186          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2187          case 0x2000:    /* EN QUAD */
2188          case 0x2001:    /* EM QUAD */
2189          case 0x2002:    /* EN SPACE */
2190          case 0x2003:    /* EM SPACE */
2191          case 0x2004:    /* THREE-PER-EM SPACE */
2192          case 0x2005:    /* FOUR-PER-EM SPACE */
2193          case 0x2006:    /* SIX-PER-EM SPACE */
2194          case 0x2007:    /* FIGURE SPACE */
2195          case 0x2008:    /* PUNCTUATION SPACE */
2196          case 0x2009:    /* THIN SPACE */
2197          case 0x200A:    /* HAIR SPACE */
2198          case 0x202f:    /* NARROW NO-BREAK SPACE */
2199          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2200          case 0x3000:    /* IDEOGRAPHIC SPACE */
2201          MRRETURN(MATCH_NOMATCH);
2202          }
2203        ecode++;
2204        break;
2205    
2206        case OP_HSPACE:
2207        if (eptr >= md->end_subject)
2208          {
2209          SCHECK_PARTIAL();
2210          MRRETURN(MATCH_NOMATCH);
2211          }
2212        GETCHARINCTEST(c, eptr);
2213        switch(c)
2214          {
2215          default: MRRETURN(MATCH_NOMATCH);
2216          case 0x09:      /* HT */
2217          case 0x20:      /* SPACE */
2218          case 0xa0:      /* NBSP */
2219          case 0x1680:    /* OGHAM SPACE MARK */
2220          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2221          case 0x2000:    /* EN QUAD */
2222          case 0x2001:    /* EM QUAD */
2223          case 0x2002:    /* EN SPACE */
2224          case 0x2003:    /* EM SPACE */
2225          case 0x2004:    /* THREE-PER-EM SPACE */
2226          case 0x2005:    /* FOUR-PER-EM SPACE */
2227          case 0x2006:    /* SIX-PER-EM SPACE */
2228          case 0x2007:    /* FIGURE SPACE */
2229          case 0x2008:    /* PUNCTUATION SPACE */
2230          case 0x2009:    /* THIN SPACE */
2231          case 0x200A:    /* HAIR SPACE */
2232          case 0x202f:    /* NARROW NO-BREAK SPACE */
2233          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2234          case 0x3000:    /* IDEOGRAPHIC SPACE */
2235          break;
2236          }
2237        ecode++;
2238        break;
2239    
2240        case OP_NOT_VSPACE:
2241        if (eptr >= md->end_subject)
2242          {
2243          SCHECK_PARTIAL();
2244          MRRETURN(MATCH_NOMATCH);
2245          }
2246        GETCHARINCTEST(c, eptr);
2247        switch(c)
2248          {
2249          default: break;
2250          case 0x0a:      /* LF */
2251          case 0x0b:      /* VT */
2252          case 0x0c:      /* FF */
2253          case 0x0d:      /* CR */
2254          case 0x85:      /* NEL */
2255          case 0x2028:    /* LINE SEPARATOR */
2256          case 0x2029:    /* PARAGRAPH SEPARATOR */
2257          MRRETURN(MATCH_NOMATCH);
2258          }
2259        ecode++;
2260        break;
2261    
2262        case OP_VSPACE:
2263        if (eptr >= md->end_subject)
2264          {
2265          SCHECK_PARTIAL();
2266          MRRETURN(MATCH_NOMATCH);
2267          }
2268        GETCHARINCTEST(c, eptr);
2269        switch(c)
2270          {
2271          default: MRRETURN(MATCH_NOMATCH);
2272          case 0x0a:      /* LF */
2273          case 0x0b:      /* VT */
2274          case 0x0c:      /* FF */
2275          case 0x0d:      /* CR */
2276          case 0x85:      /* NEL */
2277          case 0x2028:    /* LINE SEPARATOR */
2278          case 0x2029:    /* PARAGRAPH SEPARATOR */
2279        break;        break;
2280        }        }
2281      ecode++;      ecode++;
# Line 1442  for (;;) Line 2287  for (;;)
2287    
2288      case OP_PROP:      case OP_PROP:
2289      case OP_NOTPROP:      case OP_NOTPROP:
2290      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2291          {
2292          SCHECK_PARTIAL();
2293          MRRETURN(MATCH_NOMATCH);
2294          }
2295      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2296        {        {
2297        int chartype, script;        const ucd_record *prop = GET_UCD(c);
       int category = _pcre_ucp_findprop(c, &chartype, &script);  
2298    
2299        switch(ecode[1])        switch(ecode[1])
2300          {          {
2301          case PT_ANY:          case PT_ANY:
2302          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2303          break;          break;
2304    
2305          case PT_LAMP:          case PT_LAMP:
2306          if ((chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2307               chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2308               chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2309            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2310           break;          break;
2311    
2312          case PT_GC:          case PT_GC:
2313          if ((ecode[2] != category) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2314            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2315          break;          break;
2316    
2317          case PT_PC:          case PT_PC:
2318          if ((ecode[2] != chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2319            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2320          break;          break;
2321    
2322          case PT_SC:          case PT_SC:
2323          if ((ecode[2] != script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2324            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2325            break;
2326    
2327            /* These are specials */
2328    
2329            case PT_ALNUM:
2330            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2331                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2332              MRRETURN(MATCH_NOMATCH);
2333            break;
2334    
2335            case PT_SPACE:    /* Perl space */
2336            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2337                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2338                   == (op == OP_NOTPROP))
2339              MRRETURN(MATCH_NOMATCH);
2340            break;
2341    
2342            case PT_PXSPACE:  /* POSIX space */
2343            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2344                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2345                 c == CHAR_FF || c == CHAR_CR)
2346                   == (op == OP_NOTPROP))
2347              MRRETURN(MATCH_NOMATCH);
2348            break;
2349    
2350            case PT_WORD:
2351            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2352                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2353                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2354              MRRETURN(MATCH_NOMATCH);
2355          break;          break;
2356    
2357            /* This should never occur */
2358    
2359          default:          default:
2360          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2361          }          }
# Line 1488  for (;;) Line 2368  for (;;)
2368      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2369    
2370      case OP_EXTUNI:      case OP_EXTUNI:
2371      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2372          {
2373          SCHECK_PARTIAL();
2374          MRRETURN(MATCH_NOMATCH);
2375          }
2376      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2377        {        {
2378        int chartype, script;        int category = UCD_CATEGORY(c);
2379        int category = _pcre_ucp_findprop(c, &chartype, &script);        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);
       if (category == ucp_M) RRETURN(MATCH_NOMATCH);  
2380        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2381          {          {
2382          int len = 1;          int len = 1;
# Line 1501  for (;;) Line 2384  for (;;)
2384            {            {
2385            GETCHARLEN(c, eptr, len);            GETCHARLEN(c, eptr, len);
2386            }            }
2387          category = _pcre_ucp_findprop(c, &chartype, &script);          category = UCD_CATEGORY(c);
2388          if (category != ucp_M) break;          if (category != ucp_M) break;
2389          eptr += len;          eptr += len;
2390          }          }
# Line 1520  for (;;) Line 2403  for (;;)
2403      loops). */      loops). */
2404    
2405      case OP_REF:      case OP_REF:
2406        {      case OP_REFI:
2407        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2408        ecode += 3;                                 /* Advance past item */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2409        ecode += 3;
2410    
2411        /* If the reference is unset, set the length to be longer than the amount      /* If the reference is unset, there are two possibilities:
       of subject left; this ensures that every attempt at a match fails. We  
       can't just fail here, because of the possibility of quantifiers with zero  
       minima. */  
   
       length = (offset >= offset_top || md->offset_vector[offset] < 0)?  
         md->end_subject - eptr + 1 :  
         md->offset_vector[offset+1] - md->offset_vector[offset];  
2412    
2413        /* Set up for repetition, or handle the non-repeated case */      (a) In the default, Perl-compatible state, set the length negative;
2414        this ensures that every attempt at a match fails. We can't just fail
2415        here, because of the possibility of quantifiers with zero minima.
2416    
2417        switch (*ecode)      (b) If the JavaScript compatibility flag is set, set the length to zero
2418          {      so that the back reference matches an empty string.
         case OP_CRSTAR:  
         case OP_CRMINSTAR:  
         case OP_CRPLUS:  
         case OP_CRMINPLUS:  
         case OP_CRQUERY:  
         case OP_CRMINQUERY:  
         c = *ecode++ - OP_CRSTAR;  
         minimize = (c & 1) != 0;  
         min = rep_min[c];                 /* Pick up values from tables; */  
         max = rep_max[c];                 /* zero for max => infinity */  
         if (max == 0) max = INT_MAX;  
         break;  
2419    
2420          case OP_CRRANGE:      Otherwise, set the length to the length of what was matched by the
2421          case OP_CRMINRANGE:      referenced subpattern. */
         minimize = (*ecode == OP_CRMINRANGE);  
         min = GET2(ecode, 1);  
         max = GET2(ecode, 3);  
         if (max == 0) max = INT_MAX;  
         ecode += 5;  
         break;  
2422    
2423          default:               /* No repeat follows */      if (offset >= offset_top || md->offset_vector[offset] < 0)
2424          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);        length = (md->jscript_compat)? 0 : -1;
2425          eptr += length;      else
2426          continue;              /* With the main loop */        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2427    
2428        /* Set up for repetition, or handle the non-repeated case */
2429    
2430        switch (*ecode)
2431          {
2432          case OP_CRSTAR:
2433          case OP_CRMINSTAR:
2434          case OP_CRPLUS:
2435          case OP_CRMINPLUS:
2436          case OP_CRQUERY:
2437          case OP_CRMINQUERY:
2438          c = *ecode++ - OP_CRSTAR;
2439          minimize = (c & 1) != 0;
2440          min = rep_min[c];                 /* Pick up values from tables; */
2441          max = rep_max[c];                 /* zero for max => infinity */
2442          if (max == 0) max = INT_MAX;
2443          break;
2444    
2445          case OP_CRRANGE:
2446          case OP_CRMINRANGE:
2447          minimize = (*ecode == OP_CRMINRANGE);
2448          min = GET2(ecode, 1);
2449          max = GET2(ecode, 3);
2450          if (max == 0) max = INT_MAX;
2451          ecode += 5;
2452          break;
2453    
2454          default:               /* No repeat follows */
2455          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2456            {
2457            CHECK_PARTIAL();
2458            MRRETURN(MATCH_NOMATCH);
2459          }          }
2460          eptr += length;
2461          continue;              /* With the main loop */
2462          }
2463    
2464        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2465        main loop. */      zero, just continue with the main loop. */
2466    
2467        if (length == 0) continue;      if (length == 0) continue;
2468    
2469        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2470        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2471        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2472    
2473        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2474          {
2475          int slength;
2476          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2477          {          {
2478          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2479          eptr += length;          MRRETURN(MATCH_NOMATCH);
2480          }          }
2481          eptr += slength;
2482          }
2483    
2484        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2485        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2486    
2487        if (min == max) continue;      if (min == max) continue;
2488    
2489        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2490    
2491        if (minimize)      if (minimize)
2492          {
2493          for (fi = min;; fi++)
2494          {          {
2495          for (fi = min;; fi++)          int slength;
2496            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2497            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2498            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2499            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2500            {            {
2501            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            CHECK_PARTIAL();
2502            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2503            }            }
2504          /* Control never gets here */          eptr += slength;
2505          }          }
2506          /* Control never gets here */
2507          }
2508    
2509        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2510    
2511        else      else
2512          {
2513          pp = eptr;
2514          for (i = min; i < max; i++)
2515          {          {
2516          pp = eptr;          int slength;
2517          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
           {  
           if (!match_ref(offset, eptr, length, md, ims)) break;  
           eptr += length;  
           }  
         while (eptr >= pp)  
2518            {            {
2519            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            CHECK_PARTIAL();
2520            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2521            }            }
2522          RRETURN(MATCH_NOMATCH);          eptr += slength;
2523            }
2524          while (eptr >= pp)
2525            {
2526            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2527            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2528            eptr -= length;
2529          }          }
2530          MRRETURN(MATCH_NOMATCH);
2531        }        }
2532      /* Control never gets here */      /* Control never gets here */
2533    
   
   
2534      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2535      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,
2536      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 1677  for (;;) Line 2585  for (;;)
2585          {          {
2586          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2587            {            {
2588            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2589                {
2590                SCHECK_PARTIAL();
2591                MRRETURN(MATCH_NOMATCH);
2592                }
2593            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2594            if (c > 255)            if (c > 255)
2595              {              {
2596              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2597              }              }
2598            else            else
2599              {              {
2600              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2601              }              }
2602            }            }
2603          }          }
# Line 1695  for (;;) Line 2607  for (;;)
2607          {          {
2608          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2609            {            {
2610            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2611                {
2612                SCHECK_PARTIAL();
2613                MRRETURN(MATCH_NOMATCH);
2614                }
2615            c = *eptr++;            c = *eptr++;
2616            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2617            }            }
2618          }          }
2619    
# Line 1717  for (;;) Line 2633  for (;;)
2633            {            {
2634            for (fi = min;; fi++)            for (fi = min;; fi++)
2635              {              {
2636              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2637              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2638              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2639                if (eptr >= md->end_subject)
2640                  {
2641                  SCHECK_PARTIAL();
2642                  MRRETURN(MATCH_NOMATCH);
2643                  }
2644              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2645              if (c > 255)              if (c > 255)
2646                {                {
2647                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2648                }                }
2649              else              else
2650                {                {
2651                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2652                }                }
2653              }              }
2654            }            }
# Line 1737  for (;;) Line 2658  for (;;)
2658            {            {
2659            for (fi = min;; fi++)            for (fi = min;; fi++)
2660              {              {
2661              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2662              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2663              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2664                if (eptr >= md->end_subject)
2665                  {
2666                  SCHECK_PARTIAL();
2667                  MRRETURN(MATCH_NOMATCH);
2668                  }
2669              c = *eptr++;              c = *eptr++;
2670              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2671              }              }
2672            }            }
2673          /* Control never gets here */          /* Control never gets here */
# Line 1760  for (;;) Line 2686  for (;;)
2686            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2687              {              {
2688              int len = 1;              int len = 1;
2689              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2690                  {
2691                  SCHECK_PARTIAL();
2692                  break;
2693                  }
2694              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2695              if (c > 255)              if (c > 255)
2696                {                {
# Line 1774  for (;;) Line 2704  for (;;)
2704              }              }
2705            for (;;)            for (;;)
2706              {              {
2707              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2708              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2709              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2710              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1786  for (;;) Line 2716  for (;;)
2716            {            {
2717            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2718              {              {
2719              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2720                  {
2721                  SCHECK_PARTIAL();
2722                  break;
2723                  }
2724              c = *eptr;              c = *eptr;
2725              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2726              eptr++;              eptr++;
2727              }              }
2728            while (eptr >= pp)            while (eptr >= pp)
2729              {              {
2730              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2731              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2732              eptr--;              eptr--;
2733              }              }
2734            }            }
2735    
2736          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2737          }          }
2738        }        }
2739      /* Control never gets here */      /* Control never gets here */
2740    
2741    
2742      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2743      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
2744        mode, because Unicode properties are supported in non-UTF-8 mode. */
2745    
2746  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2747      case OP_XCLASS:      case OP_XCLASS:
# Line 1847  for (;;) Line 2782  for (;;)
2782    
2783        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2784          {          {
2785          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2786          GETCHARINC(c, eptr);            {
2787          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2788              MRRETURN(MATCH_NOMATCH);
2789              }
2790            GETCHARINCTEST(c, eptr);
2791            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2792          }          }
2793    
2794        /* 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 1864  for (;;) Line 2803  for (;;)
2803          {          {
2804          for (fi = min;; fi++)          for (fi = min;; fi++)
2805            {            {
2806            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2807            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2808            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2809            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2810            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2811                SCHECK_PARTIAL();
2812                MRRETURN(MATCH_NOMATCH);
2813                }
2814              GETCHARINCTEST(c, eptr);
2815              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2816            }            }
2817          /* Control never gets here */          /* Control never gets here */
2818          }          }
# Line 1881  for (;;) Line 2825  for (;;)
2825          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2826            {            {
2827            int len = 1;            int len = 1;
2828            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2829            GETCHARLEN(c, eptr, len);              {
2830                SCHECK_PARTIAL();
2831                break;
2832                }
2833              GETCHARLENTEST(c, eptr, len);
2834            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2835            eptr += len;            eptr += len;
2836            }            }
2837          for(;;)          for(;;)
2838            {            {
2839            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2840            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2841            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2842            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2843            }            }
2844          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2845          }          }
2846    
2847        /* Control never gets here */        /* Control never gets here */
# Line 1909  for (;;) Line 2857  for (;;)
2857        length = 1;        length = 1;
2858        ecode++;        ecode++;
2859        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2860        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2861        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2862            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2863            MRRETURN(MATCH_NOMATCH);
2864            }
2865          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2866        }        }
2867      else      else
2868  #endif  #endif
2869    
2870      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2871        {        {
2872        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2873        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2874            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2875            MRRETURN(MATCH_NOMATCH);
2876            }
2877          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2878        ecode += 2;        ecode += 2;
2879        }        }
2880      break;      break;
2881    
2882      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2883    
2884      case OP_CHARNC:      case OP_CHARI:
2885  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2886      if (utf8)      if (utf8)
2887        {        {
# Line 1933  for (;;) Line 2889  for (;;)
2889        ecode++;        ecode++;
2890        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2891    
2892        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2893            {
2894            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2895            MRRETURN(MATCH_NOMATCH);
2896            }
2897    
2898        /* 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
2899        can use the fast lookup table. */        can use the fast lookup table. */
2900    
2901        if (fc < 128)        if (fc < 128)
2902          {          {
2903          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2904          }          }
2905    
2906        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
# Line 1957  for (;;) Line 2917  for (;;)
2917          if (fc != dc)          if (fc != dc)
2918            {            {
2919  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2920            if (dc != _pcre_ucp_othercase(fc))            if (dc != UCD_OTHERCASE(fc))
2921  #endif  #endif
2922              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
2923            }            }
2924          }          }
2925        }        }
# Line 1968  for (;;) Line 2928  for (;;)
2928    
2929      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2930        {        {
2931        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2932        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
2933            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2934            MRRETURN(MATCH_NOMATCH);
2935            }
2936          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2937        ecode += 2;        ecode += 2;
2938        }        }
2939      break;      break;
# Line 1977  for (;;) Line 2941  for (;;)
2941      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2942    
2943      case OP_EXACT:      case OP_EXACT:
2944        case OP_EXACTI:
2945      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2946      ecode += 3;      ecode += 3;
2947      goto REPEATCHAR;      goto REPEATCHAR;
2948    
2949      case OP_POSUPTO:      case OP_POSUPTO:
2950        case OP_POSUPTOI:
2951      possessive = TRUE;      possessive = TRUE;
2952      /* Fall through */      /* Fall through */
2953    
2954      case OP_UPTO:      case OP_UPTO:
2955        case OP_UPTOI:
2956      case OP_MINUPTO:      case OP_MINUPTO:
2957        case OP_MINUPTOI:
2958      min = 0;      min = 0;
2959      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2960      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2961      ecode += 3;      ecode += 3;
2962      goto REPEATCHAR;      goto REPEATCHAR;
2963    
2964      case OP_POSSTAR:      case OP_POSSTAR:
2965        case OP_POSSTARI:
2966      possessive = TRUE;      possessive = TRUE;
2967      min = 0;      min = 0;
2968      max = INT_MAX;      max = INT_MAX;
# Line 2001  for (;;) Line 2970  for (;;)
2970      goto REPEATCHAR;      goto REPEATCHAR;
2971    
2972      case OP_POSPLUS:      case OP_POSPLUS:
2973        case OP_POSPLUSI:
2974      possessive = TRUE;      possessive = TRUE;
2975      min = 1;      min = 1;
2976      max = INT_MAX;      max = INT_MAX;
# Line 2008  for (;;) Line 2978  for (;;)
2978      goto REPEATCHAR;      goto REPEATCHAR;
2979    
2980      case OP_POSQUERY:      case OP_POSQUERY:
2981        case OP_POSQUERYI:
2982      possessive = TRUE;      possessive = TRUE;
2983      min = 0;      min = 0;
2984      max = 1;      max = 1;
# Line 2015  for (;;) Line 2986  for (;;)
2986      goto REPEATCHAR;      goto REPEATCHAR;
2987    
2988      case OP_STAR:      case OP_STAR:
2989        case OP_STARI:
2990      case OP_MINSTAR:      case OP_MINSTAR:
2991        case OP_MINSTARI:
2992      case OP_PLUS:      case OP_PLUS:
2993        case OP_PLUSI:
2994      case OP_MINPLUS:      case OP_MINPLUS:
2995        case OP_MINPLUSI:
2996      case OP_QUERY:      case OP_QUERY:
2997        case OP_QUERYI:
2998      case OP_MINQUERY:      case OP_MINQUERY:
2999      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3000        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3001      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3002      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3003      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3004      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3005    
3006      /* 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. */  
3007    
3008      REPEATCHAR:      REPEATCHAR:
3009  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2037  for (;;) Line 3012  for (;;)
3012        length = 1;        length = 1;
3013        charptr = ecode;        charptr = ecode;
3014        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3015        ecode += length;        ecode += length;
3016    
3017        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 2045  for (;;) Line 3019  for (;;)
3019    
3020        if (length > 1)        if (length > 1)
3021          {          {
         int oclength = 0;  
         uschar occhars[8];  
   
3022  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3023          unsigned int othercase;          unsigned int othercase;
3024          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3025              (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)              (othercase = UCD_OTHERCASE(fc)) != fc)
3026            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3027            else oclength = 0;
3028  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3029    
3030          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3031            {            {
3032            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3033            /* Need braces because of following else */              memcmp(eptr, charptr, length) == 0) eptr += length;
3034            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
3035              else if (oclength > 0 &&
3036                       eptr <= md->end_subject - oclength &&
3037                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3038    #endif  /* SUPPORT_UCP */
3039            else            else
3040              {              {
3041              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3042              eptr += oclength;              MRRETURN(MATCH_NOMATCH);
3043              }              }
3044            }            }
3045    
# Line 2073  for (;;) Line 3049  for (;;)
3049            {            {
3050            for (fi = min;; fi++)            for (fi = min;; fi++)
3051              {              {
3052              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3053              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3054              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3055              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3056              /* Need braces because of following else */                memcmp(eptr, charptr, length) == 0) eptr += length;
3057              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
3058                else if (oclength > 0 &&
3059                         eptr <= md->end_subject - oclength &&
3060                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3061    #endif  /* SUPPORT_UCP */
3062              else              else
3063                {                {
3064                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3065                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3066                }                }
3067              }              }
3068            /* Control never gets here */            /* Control never gets here */
# Line 2093  for (;;) Line 3073  for (;;)
3073            pp = eptr;            pp = eptr;
3074            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3075              {              {
3076              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3077              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3078              else if (oclength == 0) break;  #ifdef SUPPORT_UCP
3079                else if (oclength > 0 &&
3080                         eptr <= md->end_subject - oclength &&
3081                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3082    #endif  /* SUPPORT_UCP */
3083              else              else
3084                {                {
3085                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3086                eptr += oclength;                break;
3087                }                }
3088              }              }
3089    
3090            if (possessive) continue;            if (possessive) continue;
3091            while (eptr >= pp)  
3092             {            for(;;)
3093             RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              {
3094             if (rrc != MATCH_NOMATCH) RRETURN(rrc);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3095             eptr -= length;              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3096             }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3097            RRETURN(MATCH_NOMATCH);  #ifdef SUPPORT_UCP
3098                eptr--;
3099                BACKCHAR(eptr);
3100    #else   /* without SUPPORT_UCP */
3101                eptr -= length;
3102    #endif  /* SUPPORT_UCP */
3103                }
3104            }            }
3105          /* Control never gets here */          /* Control never gets here */
3106          }          }
# Line 2123  for (;;) Line 3113  for (;;)
3113  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3114    
3115      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3116        {  
3117        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3118    
3119      /* 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
3120      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 2140  for (;;) Line 3128  for (;;)
3128      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3129        max, eptr));        max, eptr));
3130    
3131      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3132        {        {
3133        fc = md->lcc[fc];        fc = md->lcc[fc];
3134        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3135          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3136        if (min == max) continue;          if (eptr >= md->end_subject)
3137        if (minimize)            {
3138              SCHECK_PARTIAL();
3139              MRRETURN(MATCH_NOMATCH);
3140              }
3141            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3142            }
3143          if (min == max) continue;
3144          if (minimize)
3145          {          {
3146          for (fi = min;; fi++)          for (fi = min;; fi++)
3147            {            {
3148            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3149            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3150            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3151                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3152              RRETURN(MATCH_NOMATCH);              {
3153                SCHECK_PARTIAL();
3154                MRRETURN(MATCH_NOMATCH);
3155                }
3156              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3157            }            }
3158          /* Control never gets here */          /* Control never gets here */
3159          }          }
# Line 2163  for (;;) Line 3162  for (;;)
3162          pp = eptr;          pp = eptr;
3163          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3164            {            {
3165            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3166                {
3167                SCHECK_PARTIAL();
3168                break;
3169                }
3170              if (fc != md->lcc[*eptr]) break;
3171            eptr++;            eptr++;
3172            }            }
3173    
3174          if (possessive) continue;          if (possessive) continue;
3175    
3176          while (eptr >= pp)          while (eptr >= pp)
3177            {            {
3178            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3179            eptr--;            eptr--;
3180            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3181            }            }
3182          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3183          }          }
3184        /* Control never gets here */        /* Control never gets here */
3185        }        }
# Line 2182  for (;;) Line 3188  for (;;)
3188    
3189      else      else
3190        {        {
3191        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3192            {
3193            if (eptr >= md->end_subject)
3194              {
3195              SCHECK_PARTIAL();
3196              MRRETURN(MATCH_NOMATCH);
3197              }
3198            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3199            }
3200    
3201        if (min == max) continue;        if (min == max) continue;
3202    
3203        if (minimize)        if (minimize)
3204          {          {
3205          for (fi = min;; fi++)          for (fi = min;; fi++)
3206            {            {
3207            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3208            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3209            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3210              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3211                {
3212                SCHECK_PARTIAL();
3213                MRRETURN(MATCH_NOMATCH);
3214                }
3215              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3216            }            }
3217          /* Control never gets here */          /* Control never gets here */
3218          }          }
# Line 2200  for (;;) Line 3221  for (;;)
3221          pp = eptr;          pp = eptr;
3222          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3223            {            {
3224            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3225                {
3226                SCHECK_PARTIAL();
3227                break;
3228                }
3229              if (fc != *eptr) break;
3230            eptr++;            eptr++;
3231            }            }
3232          if (possessive) continue;          if (possessive) continue;
3233    
3234          while (eptr >= pp)          while (eptr >= pp)
3235            {            {
3236            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3237            eptr--;            eptr--;
3238            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3239            }            }
3240          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3241          }          }
3242        }        }
3243      /* Control never gets here */      /* Control never gets here */
# Line 2219  for (;;) Line 3246  for (;;)
3246      checking can be multibyte. */      checking can be multibyte. */
3247    
3248      case OP_NOT:      case OP_NOT:
3249      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3250        if (eptr >= md->end_subject)
3251          {
3252          SCHECK_PARTIAL();
3253          MRRETURN(MATCH_NOMATCH);
3254          }
3255      ecode++;      ecode++;
3256      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3257      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3258        {        {
3259  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3260        if (c < 256)        if (c < 256)
3261  #endif  #endif
3262        c = md->lcc[c];        c = md->lcc[c];
3263        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3264        }        }
3265      else      else    /* Caseful */
3266        {        {
3267        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3268        }        }
3269      break;      break;
3270    
# Line 2244  for (;;) Line 3276  for (;;)
3276      about... */      about... */
3277    
3278      case OP_NOTEXACT:      case OP_NOTEXACT:
3279        case OP_NOTEXACTI:
3280      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3281      ecode += 3;      ecode += 3;
3282      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3283    
3284      case OP_NOTUPTO:      case OP_NOTUPTO:
3285        case OP_NOTUPTOI:
3286      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3287        case OP_NOTMINUPTOI:
3288      min = 0;      min = 0;
3289      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3290      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3291      ecode += 3;      ecode += 3;
3292      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3293    
3294      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3295        case OP_NOTPOSSTARI:
3296      possessive = TRUE;      possessive = TRUE;
3297      min = 0;      min = 0;
3298      max = INT_MAX;      max = INT_MAX;
# Line 2264  for (;;) Line 3300  for (;;)
3300      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3301    
3302      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3303        case OP_NOTPOSPLUSI:
3304      possessive = TRUE;      possessive = TRUE;
3305      min = 1;      min = 1;
3306      max = INT_MAX;      max = INT_MAX;
# Line 2271  for (;;) Line 3308  for (;;)
3308      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3309    
3310      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3311        case OP_NOTPOSQUERYI:
3312      possessive = TRUE;      possessive = TRUE;
3313      min = 0;      min = 0;
3314      max = 1;      max = 1;
# Line 2278  for (;;) Line 3316  for (;;)
3316      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3317    
3318      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3319        case OP_NOTPOSUPTOI:
3320      possessive = TRUE;      possessive = TRUE;
3321      min = 0;      min = 0;
3322      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 2285  for (;;) Line 3324  for (;;)
3324      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3325    
3326      case OP_NOTSTAR:      case OP_NOTSTAR:
3327        case OP_NOTSTARI:
3328      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3329        case OP_NOTMINSTARI:
3330      case OP_NOTPLUS:      case OP_NOTPLUS:
3331        case OP_NOTPLUSI:
3332      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3333        case OP_NOTMINPLUSI:
3334      case OP_NOTQUERY:      case OP_NOTQUERY:
3335        case OP_NOTQUERYI:
3336      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3337      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3338        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3339      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3340      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3341      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3342      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3343    
3344      /* 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. */  
3345    
3346      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3347      fc = *ecode++;      fc = *ecode++;
3348    
3349      /* 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 2315  for (;;) Line 3357  for (;;)
3357      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,
3358        max, eptr));        max, eptr));
3359    
3360      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3361        {        {
3362        fc = md->lcc[fc];        fc = md->lcc[fc];
3363    
# Line 2326  for (;;) Line 3368  for (;;)
3368          register unsigned int d;          register unsigned int d;
3369          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3370            {            {
3371              if (eptr >= md->end_subject)
3372                {
3373                SCHECK_PARTIAL();
3374                MRRETURN(MATCH_NOMATCH);
3375                }
3376            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3377            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3378            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3379            }            }
3380          }          }
3381        else        else
# Line 2337  for (;;) Line 3384  for (;;)
3384        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3385          {          {
3386          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3387            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3388              if (eptr >= md->end_subject)
3389                {
3390                SCHECK_PARTIAL();
3391                MRRETURN(MATCH_NOMATCH);
3392                }
3393              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3394              }
3395          }          }
3396    
3397        if (min == max) continue;        if (min == max) continue;
# Line 2351  for (;;) Line 3405  for (;;)
3405            register unsigned int d;            register unsigned int d;
3406            for (fi = min;; fi++)            for (fi = min;; fi++)
3407              {              {
3408              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3409              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3410                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3411                if (eptr >= md->end_subject)
3412                  {
3413                  SCHECK_PARTIAL();
3414                  MRRETURN(MATCH_NOMATCH);
3415                  }
3416              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3417              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3418              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3419              }              }
3420            }            }
3421          else          else
# Line 2365  for (;;) Line 3424  for (;;)
3424            {            {
3425            for (fi = min;; fi++)            for (fi = min;; fi++)
3426              {              {
3427              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3428              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3429              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3430                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3431                  {
3432                  SCHECK_PARTIAL();
3433                  MRRETURN(MATCH_NOMATCH);
3434                  }
3435                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3436              }              }
3437            }            }
3438          /* Control never gets here */          /* Control never gets here */
# Line 2388  for (;;) Line 3452  for (;;)
3452            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3453              {              {
3454              int len = 1;              int len = 1;
3455              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3456                  {
3457                  SCHECK_PARTIAL();
3458                  break;
3459                  }
3460              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3461              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3462              if (fc == d) break;              if (fc == d) break;
# Line 2397  for (;;) Line 3465  for (;;)
3465          if (possessive) continue;          if (possessive) continue;
3466          for(;;)          for(;;)
3467              {              {
3468              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3469              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3470              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3471              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2409  for (;;) Line 3477  for (;;)
3477            {            {
3478            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3479              {              {
3480              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3481                  {
3482                  SCHECK_PARTIAL();
3483                  break;
3484                  }
3485                if (fc == md->lcc[*eptr]) break;
3486              eptr++;              eptr++;
3487              }              }
3488            if (possessive) continue;            if (possessive) continue;
3489            while (eptr >= pp)            while (eptr >= pp)
3490              {              {
3491              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3492              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3493              eptr--;              eptr--;
3494              }              }
3495            }            }
3496    
3497          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3498          }          }
3499        /* Control never gets here */        /* Control never gets here */
3500        }        }
# Line 2437  for (;;) Line 3510  for (;;)
3510          register unsigned int d;          register unsigned int d;
3511          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3512            {            {
3513              if (eptr >= md->end_subject)
3514                {
3515                SCHECK_PARTIAL();
3516                MRRETURN(MATCH_NOMATCH);
3517                }
3518            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3519            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3520            }            }
3521          }          }
3522        else        else
# Line 2446  for (;;) Line 3524  for (;;)
3524        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3525          {          {
3526          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3527            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3528              if (eptr >= md->end_subject)
3529                {
3530                SCHECK_PARTIAL();
3531                MRRETURN(MATCH_NOMATCH);
3532                }
3533              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3534              }
3535          }          }
3536    
3537        if (min == max) continue;        if (min == max) continue;
# Line 2460  for (;;) Line 3545  for (;;)
3545            register unsigned int d;            register unsigned int d;
3546            for (fi = min;; fi++)            for (fi = min;; fi++)
3547              {              {
3548              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3549              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3550                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3551                if (eptr >= md->end_subject)
3552                  {
3553                  SCHECK_PARTIAL();
3554                  MRRETURN(MATCH_NOMATCH);
3555                  }
3556              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3557              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3558              }              }
3559            }            }
3560          else          else
# Line 2473  for (;;) Line 3563  for (;;)
3563            {            {
3564            for (fi = min;; fi++)            for (fi = min;; fi++)
3565              {              {
3566              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3567              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3568              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3569                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3570                  {
3571                  SCHECK_PARTIAL();
3572                  MRRETURN(MATCH_NOMATCH);
3573                  }
3574                if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3575              }              }
3576            }            }
3577          /* Control never gets here */          /* Control never gets here */
# Line 2496  for (;;) Line 3591  for (;;)
3591            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3592              {              {
3593              int len = 1;              int len = 1;
3594              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3595                  {
3596                  SCHECK_PARTIAL();
3597                  break;
3598                  }
3599              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3600              if (fc == d) break;              if (fc == d) break;
3601              eptr += len;              eptr += len;
# Line 2504  for (;;) Line 3603  for (;;)
3603            if (possessive) continue;            if (possessive) continue;
3604            for(;;)            for(;;)
3605              {              {
3606              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3607              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3608              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3609              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2516  for (;;) Line 3615  for (;;)
3615            {            {
3616            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3617              {              {
3618              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3619                  {
3620                  SCHECK_PARTIAL();
3621                  break;
3622                  }
3623                if (fc == *eptr) break;
3624              eptr++;              eptr++;
3625              }              }
3626            if (possessive) continue;            if (possessive) continue;
3627            while (eptr >= pp)            while (eptr >= pp)
3628              {              {
3629              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3630              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3631              eptr--;              eptr--;
3632              }              }
3633            }            }
3634    
3635          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3636          }          }
3637        }        }
3638      /* Control never gets here */      /* Control never gets here */
# Line 2610  for (;;) Line 3714  for (;;)
3714    
3715      /* First, ensure the minimum number of matches are present. Use inline      /* First, ensure the minimum number of matches are present. Use inline
3716      code for maximizing the speed, and do the type test once at the start      code for maximizing the speed, and do the type test once at the start
3717      (i.e. keep it out of the loop). Also we can test that there are at least      (i.e. keep it out of the loop). Separate the UTF-8 code completely as that
     the minimum number of bytes before we start. This isn't as effective in  
     UTF-8 mode, but it does no harm. Separate the UTF-8 code completely as that  
3718      is tidier. Also separate the UCP code, which can be the same for both UTF-8      is tidier. Also separate the UCP code, which can be the same for both UTF-8
3719      and single-bytes. */      and single-bytes. */
3720    
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3721      if (min > 0)      if (min > 0)
3722        {        {
3723  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2625  for (;;) Line 3726  for (;;)
3726          switch(prop_type)          switch(prop_type)
3727            {            {
3728            case PT_ANY:            case PT_ANY:
3729            if (prop_fail_result) RRETURN(MATCH_NOMATCH);            if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
3730            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3731              {              {
3732              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3733              GETCHARINC(c, eptr);                {
3734                  SCHECK_PARTIAL();
3735                  MRRETURN(MATCH_NOMATCH);
3736                  }
3737                GETCHARINCTEST(c, eptr);
3738              }              }
3739            break;            break;
3740    
3741            case PT_LAMP:            case PT_LAMP:
3742            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3743              {              {
3744              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3745              GETCHARINC(c, eptr);                {
3746              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3747                  MRRETURN(MATCH_NOMATCH);
3748                  }
3749                GETCHARINCTEST(c, eptr);
3750                prop_chartype = UCD_CHARTYPE(c);
3751              if ((prop_chartype == ucp_Lu ||              if ((prop_chartype == ucp_Lu ||
3752                   prop_chartype == ucp_Ll ||                   prop_chartype == ucp_Ll ||
3753                   prop_chartype == ucp_Lt) == prop_fail_result)                   prop_chartype == ucp_Lt) == prop_fail_result)
3754                RRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3755              }              }
3756            break;            break;
3757    
3758            case PT_GC:            case PT_GC:
3759            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3760              {              {
3761              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3762              GETCHARINC(c, eptr);                {
3763              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3764                  MRRETURN(MATCH_NOMATCH);
3765                  }
3766                GETCHARINCTEST(c, eptr);
3767                prop_category = UCD_CATEGORY(c);
3768              if ((prop_category == prop_value) == prop_fail_result)              if ((prop_category == prop_value) == prop_fail_result)
3769                RRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3770              }              }
3771            break;            break;
3772    
3773            case PT_PC:            case PT_PC:
3774            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3775              {              {
3776              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3777              GETCHARINC(c, eptr);                {
3778              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3779                  MRRETURN(MATCH_NOMATCH);
3780                  }
3781                GETCHARINCTEST(c, eptr);
3782                prop_chartype = UCD_CHARTYPE(c);
3783              if ((prop_chartype == prop_value) == prop_fail_result)              if ((prop_chartype == prop_value) == prop_fail_result)
3784                RRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3785              }              }
3786            break;            break;
3787    
3788            case PT_SC:            case PT_SC:
3789            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3790              {              {
3791              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3792              GETCHARINC(c, eptr);                {
3793              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3794                  MRRETURN(MATCH_NOMATCH);
3795                  }
3796                GETCHARINCTEST(c, eptr);
3797                prop_script = UCD_SCRIPT(c);
3798              if ((prop_script == prop_value) == prop_fail_result)              if ((prop_script == prop_value) == prop_fail_result)
3799                RRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3800                }
3801              break;
3802    
3803              case PT_ALNUM:
3804              for (i = 1; i <= min; i++)
3805                {
3806                if (eptr >= md->end_subject)
3807                  {
3808                  SCHECK_PARTIAL();
3809                  MRRETURN(MATCH_NOMATCH);
3810                  }
3811                GETCHARINCTEST(c, eptr);
3812                prop_category = UCD_CATEGORY(c);
3813                if ((prop_category == ucp_L || prop_category == ucp_N)
3814                       == prop_fail_result)
3815                  MRRETURN(MATCH_NOMATCH);
3816                }
3817              break;
3818    
3819              case PT_SPACE:    /* Perl space */
3820              for (i = 1; i <= min; i++)
3821                {
3822                if (eptr >= md->end_subject)
3823                  {
3824                  SCHECK_PARTIAL();
3825                  MRRETURN(MATCH_NOMATCH);
3826                  }
3827                GETCHARINCTEST(c, eptr);
3828                prop_category = UCD_CATEGORY(c);
3829                if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3830                     c == CHAR_FF || c == CHAR_CR)
3831                       == prop_fail_result)
3832                  MRRETURN(MATCH_NOMATCH);
3833                }
3834              break;
3835    
3836              case PT_PXSPACE:  /* POSIX space */
3837              for (i = 1; i <= min; i++)
3838                {
3839                if (eptr >= md->end_subject)
3840                  {
3841                  SCHECK_PARTIAL();
3842                  MRRETURN(MATCH_NOMATCH);
3843                  }
3844                GETCHARINCTEST(c, eptr);
3845                prop_category = UCD_CATEGORY(c);
3846                if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3847                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3848                       == prop_fail_result)
3849                  MRRETURN(MATCH_NOMATCH);
3850                }
3851              break;
3852    
3853              case PT_WORD:
3854              for (i = 1; i <= min; i++)
3855                {
3856                if (eptr >= md->end_subject)
3857                  {
3858                  SCHECK_PARTIAL();
3859                  MRRETURN(MATCH_NOMATCH);
3860                  }
3861                GETCHARINCTEST(c, eptr);
3862                prop_category = UCD_CATEGORY(c);
3863                if ((prop_category == ucp_L || prop_category == ucp_N ||
3864                     c == CHAR_UNDERSCORE)
3865                       == prop_fail_result)
3866                  MRRETURN(MATCH_NOMATCH);
3867              }              }
3868            break;            break;
3869    
3870              /* This should not occur */
3871    
3872            default:            default:
3873            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
3874            }            }
# Line 2691  for (;;) Line 3881  for (;;)
3881          {          {
3882          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3883            {            {
3884              if (eptr >= md->end_subject)
3885                {
3886                SCHECK_PARTIAL();
3887                MRRETURN(MATCH_NOMATCH);
3888                }
3889            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3890            prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);            prop_category = UCD_CATEGORY(c);
3891            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);            if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);
3892            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3893              {              {
3894              int len = 1;              int len = 1;
3895              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr;
3896                {                else { GETCHARLEN(c, eptr, len); }
3897                GETCHARLEN(c, eptr, len);              prop_category = UCD_CATEGORY(c);
               }  
             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);  
3898              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
3899              eptr += len;              eptr += len;
3900              }              }
# Line 2719  for (;;) Line 3912  for (;;)
3912          case OP_ANY:          case OP_ANY:
3913          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3914            {            {
3915            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject)
3916                 ((ims & PCRE_DOTALL) == 0 && IS_NEWLINE(eptr)))              {
3917              RRETURN(MATCH_NOMATCH);              SCHECK_PARTIAL();
3918                MRRETURN(MATCH_NOMATCH);
3919                }
3920              if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
3921              eptr++;
3922              while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
3923              }
3924            break;
3925    
3926            case OP_ALLANY:
3927            for (i = 1; i <= min; i++)
3928              {
3929              if (eptr >= md->end_subject)
3930                {
3931                SCHECK_PARTIAL();
3932                MRRETURN(MATCH_NOMATCH);
3933                }
3934            eptr++;            eptr++;
3935            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
3936            }            }
3937          break;          break;
3938    
3939          case OP_ANYBYTE:          case OP_ANYBYTE:
3940            if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH);
3941          eptr += min;          eptr += min;
3942          break;          break;
3943    
3944          case OP_ANYNL:          case OP_ANYNL:
3945          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3946            {            {
3947            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3948                {
3949                SCHECK_PARTIAL();
3950                MRRETURN(MATCH_NOMATCH);
3951                }
3952            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
3953            switch(c)            switch(c)
3954              {              {
<