/[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 93 by nigel, Sat Feb 24 21:41:42 2007 UTC revision 702 by ph10, Tue Sep 20 15:45:06 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_ONCE         (-996)
80    #define MATCH_PRUNE        (-995)
81    #define MATCH_SKIP         (-994)
82    #define MATCH_SKIP_ARG     (-993)
83    #define MATCH_THEN         (-992)
84    
85    /* This is a convenience macro for code that occurs many times. */
86    
87    #define MRRETURN(ra) \
88      { \
89      md->mark = markptr; \
90      RRETURN(ra); \
91      }
92    
93  /* Maximum number of ints of offset to save on the stack for recursive calls.  /* Maximum number of ints of offset to save on the stack for recursive calls.
94  If the offset vector is bigger, malloc is used. This should be a multiple of 3,  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
95  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 103  static const char rep_max[] = { 0, 0, 0,
103    
104    
105    
106  #ifdef DEBUG  #ifdef PCRE_DEBUG
107  /*************************************************  /*************************************************
108  *        Debugging function to print chars       *  *        Debugging function to print chars       *
109  *************************************************/  *************************************************/
# Line 111  while (length-- > 0) Line 136  while (length-- > 0)
136  *          Match a back-reference                *  *          Match a back-reference                *
137  *************************************************/  *************************************************/
138    
139  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
140  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
141    the length passed is zero. Note that in caseless UTF-8 mode, the number of
142    subject bytes matched may be different to the number of reference bytes.
143    
144  Arguments:  Arguments:
145    offset      index into the offset vector    offset      index into the offset vector
146    eptr        points into the subject    eptr        pointer into the subject
147    length      length to be matched    length      length of reference to be matched (number of bytes)
148    md          points to match data block    md          points to match data block
149    ims         the ims flags    caseless    TRUE if caseless
150    
151  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
152  */  */
153    
154  static BOOL  static int
155  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
156    unsigned long int ims)    BOOL caseless)
157  {  {
158  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
159    register USPTR p = md->start_subject + md->offset_vector[offset];
160    
161  #ifdef DEBUG  #ifdef PCRE_DEBUG
162  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
163    printf("matching subject <null>");    printf("matching subject <null>");
164  else  else
# Line 143  pchars(p, length, FALSE, md); Line 171  pchars(p, length, FALSE, md);
171  printf("\n");  printf("\n");
172  #endif  #endif
173    
174  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
175    
176  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
177    
178  /* Separate the caselesss case for speed */  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
179    properly if Unicode properties are supported. Otherwise, we can check only
180    ASCII characters. */
181    
182  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
183    {    {
184    while (length-- > 0)  #ifdef SUPPORT_UTF8
185      if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE;  #ifdef SUPPORT_UCP
186      if (md->utf8)
187        {
188        /* Match characters up to the end of the reference. NOTE: the number of
189        bytes matched may differ, because there are some characters whose upper and
190        lower case versions code as different numbers of bytes. For example, U+023A
191        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
192        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
193        the latter. It is important, therefore, to check the length along the
194        reference, not along the subject (earlier code did this wrong). */
195    
196        USPTR endptr = p + length;
197        while (p < endptr)
198          {
199          int c, d;
200          if (eptr >= md->end_subject) return -1;
201          GETCHARINC(c, eptr);
202          GETCHARINC(d, p);
203          if (c != d && c != UCD_OTHERCASE(d)) return -1;
204          }
205        }
206      else
207    #endif
208    #endif
209    
210      /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
211      is no UCP support. */
212        {
213        if (eptr + length > md->end_subject) return -1;
214        while (length-- > 0)
215          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
216        }
217    }    }
218    
219    /* In the caseful case, we can just compare the bytes, whether or not we
220    are in UTF-8 mode. */
221    
222  else  else
223    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
224      if (eptr + length > md->end_subject) return -1;
225      while (length-- > 0) if (*p++ != *eptr++) return -1;
226      }
227    
228  return TRUE;  return eptr - eptr_start;
229  }  }
230    
231    
# Line 183  calls by keeping local variables that ne Line 251  calls by keeping local variables that ne
251  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
252  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
253  always used to.  always used to.
254    
255    The original heap-recursive code used longjmp(). However, it seems that this
256    can be very slow on some operating systems. Following a suggestion from Stan
257    Switzer, the use of longjmp() has been abolished, at the cost of having to
258    provide a unique number for each call to RMATCH. There is no way of generating
259    a sequence of numbers at compile time in C. I have given them names, to make
260    them stand out more clearly.
261    
262    Crude tests on x86 Linux show a small speedup of around 5-8%. However, on
263    FreeBSD, avoiding longjmp() more than halves the time taken to run the standard
264    tests. Furthermore, not using longjmp() means that local dynamic variables
265    don't have indeterminate values; this has meant that the frame size can be
266    reduced because the result can be "passed back" by straight setting of the
267    variable instead of being passed in the frame.
268  ****************************************************************************  ****************************************************************************
269  ***************************************************************************/  ***************************************************************************/
270    
271    /* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN
272    below must be updated in sync.  */
273    
274    enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,
275           RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,
276           RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
277           RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
278           RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
279           RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280           RM61,  RM62, RM63 };
281    
282  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
283  versions and production versions. */  versions and production versions. Note that the "rw" argument of RMATCH isn't
284    actually used in this definition. */
285    
286  #ifndef NO_RECURSE  #ifndef NO_RECURSE
287  #define REGISTER register  #define REGISTER register
288  #ifdef DEBUG  
289  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) \  #ifdef PCRE_DEBUG
290    #define RMATCH(ra,rb,rc,rd,re,rw) \
291    { \    { \
292    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
293    rx = match(ra,rb,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
294    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
295    }    }
296  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 205  versions and production versions. */ Line 299  versions and production versions. */
299    return ra; \    return ra; \
300    }    }
301  #else  #else
302  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
303    rx = match(ra,rb,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
304  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
305  #endif  #endif
306    
307  #else  #else
308    
309    
310  /* 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
311  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
312  match(), which never changes. */  argument of match(), which never changes. */
313    
314  #define REGISTER  #define REGISTER
315    
316  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
317    {\    {\
318    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
319    if (setjmp(frame->Xwhere) == 0)\    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
320      {\    frame->Xwhere = rw; \
321      newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
322      newframe->Xecode = rb;\    newframe->Xecode = rb;\
323      newframe->Xoffset_top = rc;\    newframe->Xmstart = mstart;\
324      newframe->Xims = re;\    newframe->Xmarkptr = markptr;\
325      newframe->Xeptrb = rf;\    newframe->Xoffset_top = rc;\
326      newframe->Xflags = rg;\    newframe->Xeptrb = re;\
327      newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
328      newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
329      frame = newframe;\    frame = newframe;\
330      DPRINTF(("restarting from line %d\n", __LINE__));\    DPRINTF(("restarting from line %d\n", __LINE__));\
331      goto HEAP_RECURSE;\    goto HEAP_RECURSE;\
332      }\    L_##rw:\
333    else\    DPRINTF(("jumped back to line %d\n", __LINE__));\
     {\  
     DPRINTF(("longjumped back to line %d\n", __LINE__));\  
     frame = md->thisframe;\  
     rx = frame->Xresult;\  
     }\  
334    }    }
335    
336  #define RRETURN(ra)\  #define RRETURN(ra)\
337    {\    {\
338    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
339    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
340    (pcre_stack_free)(newframe);\    (pcre_stack_free)(oldframe);\
341    if (frame != NULL)\    if (frame != NULL)\
342      {\      {\
343      frame->Xresult = ra;\      rrc = ra;\
344      md->thisframe = frame;\      goto HEAP_RETURN;\
     longjmp(frame->Xwhere, 1);\  
345      }\      }\
346    return ra;\    return ra;\
347    }    }
# Line 266  typedef struct heapframe { Line 354  typedef struct heapframe {
354    
355    /* Function arguments that may change */    /* Function arguments that may change */
356    
357    const uschar *Xeptr;    USPTR Xeptr;
358    const uschar *Xecode;    const uschar *Xecode;
359      USPTR Xmstart;
360      USPTR Xmarkptr;
361    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
362    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
363    unsigned int Xrdepth;    unsigned int Xrdepth;
364    
365    /* Function local variables */    /* Function local variables */
366    
367    const uschar *Xcallpat;    USPTR Xcallpat;
368    const uschar *Xcharptr;  #ifdef SUPPORT_UTF8
369    const uschar *Xdata;    USPTR Xcharptr;
370    const uschar *Xnext;  #endif
371    const uschar *Xpp;    USPTR Xdata;
372    const uschar *Xprev;    USPTR Xnext;
373    const uschar *Xsaved_eptr;    USPTR Xpp;
374      USPTR Xprev;
375      USPTR Xsaved_eptr;
376    
377    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
378    
# Line 290  typedef struct heapframe { Line 380  typedef struct heapframe {
380    BOOL Xcondition;    BOOL Xcondition;
381    BOOL Xprev_is_word;    BOOL Xprev_is_word;
382    
   unsigned long int Xoriginal_ims;  
   
383  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
386    int Xprop_fail_result;    int Xprop_fail_result;
387    int Xprop_category;    int Xoclength;
388    int Xprop_chartype;    uschar Xocchars[8];
   int Xprop_script;  
389  #endif  #endif
390    
391      int Xcodelink;
392    int Xctype;    int Xctype;
393    unsigned int Xfc;    unsigned int Xfc;
394    int Xfi;    int Xfi;
# Line 316  typedef struct heapframe { Line 404  typedef struct heapframe {
404    
405    eptrblock Xnewptrb;    eptrblock Xnewptrb;
406    
407    /* Place to pass back result, and where to jump back to */    /* Where to jump back to */
408    
409    int  Xresult;    int Xwhere;
   jmp_buf Xwhere;  
410    
411  } heapframe;  } heapframe;
412    
# Line 337  typedef struct heapframe { Line 424  typedef struct heapframe {
424    
425  /* This function is called recursively in many circumstances. Whenever it  /* This function is called recursively in many circumstances. Whenever it
426  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
427  same response.  same response. */
428    
429    /* These macros pack up tests that are used for partial matching, and which
430    appears several times in the code. We set the "hit end" flag if the pointer is
431    at the end of the subject and also past the start of the subject (i.e.
432    something has been matched). For hard partial matching, we then return
433    immediately. The second one is used when we already know we are past the end of
434    the subject. */
435    
436    #define CHECK_PARTIAL()\
437      if (md->partial != 0 && eptr >= md->end_subject && \
438          eptr > md->start_used_ptr) \
439        { \
440        md->hitend = TRUE; \
441        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
442        }
443    
444    #define SCHECK_PARTIAL()\
445      if (md->partial != 0 && eptr > md->start_used_ptr) \
446        { \
447        md->hitend = TRUE; \
448        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
449        }
450    
451  Performance note: It might be tempting to extract commonly used fields from the  
452  md structure (e.g. utf8, end_subject) into individual variables to improve  /* Performance note: It might be tempting to extract commonly used fields from
453    the md structure (e.g. utf8, end_subject) into individual variables to improve
454  performance. Tests using gcc on a SPARC disproved this; in the first case, it  performance. Tests using gcc on a SPARC disproved this; in the first case, it
455  made performance worse.  made performance worse.
456    
457  Arguments:  Arguments:
458     eptr        pointer to current character in subject     eptr        pointer to current character in subject
459     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
460       mstart      pointer to the current match start position (can be modified
461                     by encountering \K)
462       markptr     pointer to the most recent MARK name, or NULL
463     offset_top  current top pointer     offset_top  current top pointer
464     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
465     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
466                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
                  match_tail_recursed - this is a tail_recursed group  
467     rdepth      the recursion depth     rdepth      the recursion depth
468    
469  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
470                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
471                   a negative MATCH_xxx value for PRUNE, SKIP, etc
472                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
473                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
474  */  */
475    
476  static int  static int
477  match(REGISTER USPTR eptr, REGISTER const uschar *ecode,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
478    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
479    int flags, unsigned int rdepth)    unsigned int rdepth)
480  {  {
481  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
482  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 380  register unsigned int c; /* Character Line 488  register unsigned int c; /* Character
488  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
489    
490  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
491    BOOL caseless;
492    int condcode;
493    
494  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
495  preserved over calls to RMATCH() are part of a "frame" which is obtained from  preserved over calls to RMATCH() are part of a "frame" which is obtained from
# Line 387  heap storage. Set up the top-level frame Line 497  heap storage. Set up the top-level frame
497  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
498    
499  #ifdef NO_RECURSE  #ifdef NO_RECURSE
500  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
501    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
502  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
503    
504  /* Copy in the original argument variables */  /* Copy in the original argument variables */
505    
506  frame->Xeptr = eptr;  frame->Xeptr = eptr;
507  frame->Xecode = ecode;  frame->Xecode = ecode;
508    frame->Xmstart = mstart;
509    frame->Xmarkptr = markptr;
510  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
511  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
512  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
513    
514  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 408  HEAP_RECURSE: Line 519  HEAP_RECURSE:
519    
520  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
521  #define ecode              frame->Xecode  #define ecode              frame->Xecode
522    #define mstart             frame->Xmstart
523    #define markptr            frame->Xmarkptr
524  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
525  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
526  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
527    
528  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 420  HEAP_RECURSE: Line 531  HEAP_RECURSE:
531  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
532  #endif  #endif
533  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
534    #define codelink           frame->Xcodelink
535  #define data               frame->Xdata  #define data               frame->Xdata
536  #define next               frame->Xnext  #define next               frame->Xnext
537  #define pp                 frame->Xpp  #define pp                 frame->Xpp
# Line 432  HEAP_RECURSE: Line 544  HEAP_RECURSE:
544  #define condition          frame->Xcondition  #define condition          frame->Xcondition
545  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
546    
 #define original_ims       frame->Xoriginal_ims  
   
547  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
548  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
549  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
550  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
551  #define prop_category      frame->Xprop_category  #define oclength           frame->Xoclength
552  #define prop_chartype      frame->Xprop_chartype  #define occhars            frame->Xocchars
 #define prop_script        frame->Xprop_script  
553  #endif  #endif
554    
555  #define ctype              frame->Xctype  #define ctype              frame->Xctype
# Line 468  i, and fc and c, can be the same variabl Line 577  i, and fc and c, can be the same variabl
577  #define fi i  #define fi i
578  #define fc c  #define fc c
579    
580    /* Many of the following variables are used only in small blocks of the code.
581    My normal style of coding would have declared them within each of those blocks.
582    However, in order to accommodate the version of this code that uses an external
583    "stack" implemented on the heap, it is easier to declare them all here, so the
584    declarations can be cut out in a block. The only declarations within blocks
585    below are for variables that do not have to be preserved over a recursive call
586    to RMATCH(). */
587    
588    #ifdef SUPPORT_UTF8
589    const uschar *charptr;
590    #endif
591    const uschar *callpat;
592    const uschar *data;
593    const uschar *next;
594    USPTR         pp;
595    const uschar *prev;
596    USPTR         saved_eptr;
597    
598    recursion_info new_recursive;
599    
600  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  BOOL cur_is_word;
 const uschar *charptr;             /* in small blocks of the code. My normal */  
 #endif                             /* style of coding would have declared    */  
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
 unsigned long int original_ims;  
   
604  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
608  int prop_category;  int oclength;
609  int prop_chartype;  uschar occhars[8];
 int prop_script;  
610  #endif  #endif
611    
612    int codelink;
613  int ctype;  int ctype;
614  int length;  int length;
615  int max;  int max;
# Line 510  int stacksave[REC_STACK_SAVE_MAX]; Line 624  int stacksave[REC_STACK_SAVE_MAX];
624  eptrblock newptrb;  eptrblock newptrb;
625  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
626    
627    /* To save space on the stack and in the heap frame, I have doubled up on some
628    of the local variables that are used only in localised parts of the code, but
629    still need to be preserved over recursive calls of match(). These macros define
630    the alternative names that are used. */
631    
632    #define allow_zero    cur_is_word
633    #define cbegroup      condition
634    #define code_offset   codelink
635    #define condassert    condition
636    #define matched_once  prev_is_word
637    
638  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
639  variables. */  variables. */
640    
# Line 529  TAIL_RECURSE: Line 654  TAIL_RECURSE:
654  /* OK, now we can get on with the real code of the function. Recursive calls  /* OK, now we can get on with the real code of the function. Recursive calls
655  are specified by the macro RMATCH and RRETURN is used to return. When  are specified by the macro RMATCH and RRETURN is used to return. When
656  NO_RECURSE is *not* defined, these just turn into a recursive call to match()  NO_RECURSE is *not* defined, these just turn into a recursive call to match()
657  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
658  defined). However, RMATCH isn't like a function call because it's quite a  defined). However, RMATCH isn't like a function call because it's quite a
659  complicated macro. It has to be used in one particular way. This shouldn't,  complicated macro. It has to be used in one particular way. This shouldn't,
660  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
661    
 /* 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 ')' */  
   
662  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
663  utf8 = md->utf8;       /* Local copy of the flag */  utf8 = md->utf8;       /* Local copy of the flag */
664  #else  #else
665  utf8 = FALSE;  utf8 = FALSE;
666  #endif  #endif
667    
668    /* First check that we haven't called match() too many times, or that we
669    haven't exceeded the recursive call limit. */
670    
671    if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
672    if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
673    
674  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
675  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
676  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
677  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
678  When match() is called in other circumstances, don't add to the chain. If this  
679  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
680  already used. */  such remembered pointers, to be checked when we hit the closing ket, in order
681    to break infinite loops that match no characters. When match() is called in
682    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
683    NOT be used with tail recursion, because the memory block that is used is on
684    the stack, so a new one may be required for each match(). */
685    
686  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
687    {    {
688    eptrblock *p;    newptrb.epb_saved_eptr = eptr;
689    if ((flags & match_tail_recursed) != 0)    newptrb.epb_prev = eptrb;
690      {    eptrb = &newptrb;
691      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;  
692    }    }
693    
694  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 577  for (;;) Line 698  for (;;)
698    minimize = possessive = FALSE;    minimize = possessive = FALSE;
699    op = *ecode;    op = *ecode;
700    
   /* For partial matching, remember if we ever hit the end of the subject after  
   matching at least one subject character. */  
   
   if (md->partial &&  
       eptr >= md->end_subject &&  
       eptr > md->start_match)  
     md->hitend = TRUE;  
   
701    switch(op)    switch(op)
702      {      {
703      /* Handle a capturing bracket. If there is space in the offset vector, save      case OP_MARK:
704      the current subject position in the working slot at the top of the vector.      markptr = ecode + 2;
705      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,
706      set from a previous iteration of this group, and be referred to by a        eptrb, RM55);
707      reference inside the group.  
708        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
709      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
710      values of the final offsets, in case they were set by a previous iteration      argument. It is passed back in md->start_match_ptr (an overloading of that
711      of the same bracket.      variable). If it does match, we reset that variable to the current subject
712        position and return MATCH_SKIP. Otherwise, pass back the return code
713        unaltered. */
714    
715        if (rrc == MATCH_SKIP_ARG &&
716            strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
717          {
718          md->start_match_ptr = eptr;
719          RRETURN(MATCH_SKIP);
720          }
721    
722        if (md->mark == NULL) md->mark = markptr;
723        RRETURN(rrc);
724    
725        case OP_FAIL:
726        MRRETURN(MATCH_NOMATCH);
727    
728        /* COMMIT overrides PRUNE, SKIP, and THEN */
729    
730        case OP_COMMIT:
731        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
732          eptrb, RM52);
733        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
734            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
735            rrc != MATCH_THEN)
736          RRETURN(rrc);
737        MRRETURN(MATCH_COMMIT);
738    
739        /* PRUNE overrides THEN */
740    
741        case OP_PRUNE:
742        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
743          eptrb, RM51);
744        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
745        MRRETURN(MATCH_PRUNE);
746    
747        case OP_PRUNE_ARG:
748        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
749          eptrb, RM56);
750        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
751        md->mark = ecode + 2;
752        RRETURN(MATCH_PRUNE);
753    
754        /* SKIP overrides PRUNE and THEN */
755    
756        case OP_SKIP:
757        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
758          eptrb, RM53);
759        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
760          RRETURN(rrc);
761        md->start_match_ptr = eptr;   /* Pass back current position */
762        MRRETURN(MATCH_SKIP);
763    
764        case OP_SKIP_ARG:
765        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
766          eptrb, RM57);
767        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769    
770        /* Pass back the current skip name by overloading md->start_match_ptr and
771        returning the special MATCH_SKIP_ARG return code. This will either be
772        caught by a matching MARK, or get to the top, where it is treated the same
773        as PRUNE. */
774    
775        md->start_match_ptr = ecode + 2;
776        RRETURN(MATCH_SKIP_ARG);
777    
778        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
779        the alt that is at the start of the current branch. This makes it possible
780        to skip back past alternatives that precede the THEN within the current
781        branch. */
782    
783        case OP_THEN:
784        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
785          eptrb, RM54);
786        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
787        md->start_match_ptr = ecode - GET(ecode, 1);
788        MRRETURN(MATCH_THEN);
789    
790        case OP_THEN_ARG:
791        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
792          offset_top, md, eptrb, RM58);
793        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
794        md->start_match_ptr = ecode - GET(ecode, 1);
795        md->mark = ecode + LINK_SIZE + 2;
796        RRETURN(MATCH_THEN);
797    
798        /* Handle a capturing bracket, other than those that are possessive with an
799        unlimited repeat. If there is space in the offset vector, save the current
800        subject position in the working slot at the top of the vector. We mustn't
801        change the current values of the data slot, because they may be set from a
802        previous iteration of this group, and be referred to by a reference inside
803        the group. A failure to match might occur after the group has succeeded,
804        if something later on doesn't match. For this reason, we need to restore
805        the working value and also the values of the final offsets, in case they
806        were set by a previous iteration of the same bracket.
807    
808      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
809      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 606  for (;;) Line 814  for (;;)
814      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
815      offset = number << 1;      offset = number << 1;
816    
817  #ifdef DEBUG  #ifdef PCRE_DEBUG
818      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
819      printf("subject=");      printf("subject=");
820      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 621  for (;;) Line 829  for (;;)
829        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
830    
831        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
832        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
833            (int)(eptr - md->start_subject);
834    
835        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
836          {          {
837          RMATCH(rrc, eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838            ims, eptrb, flags);          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839          if (rrc != MATCH_NOMATCH) RRETURN(rrc);            eptrb, RM1);
840            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841            if (rrc != MATCH_NOMATCH &&
842                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
843              RRETURN(rrc);
844          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
845          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
846            if (*ecode != OP_ALT) break;
847          }          }
       while (*ecode == OP_ALT);  
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855          MATCH_THEN. */
856    
857          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
858          RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
859        }        }
860    
861      /* Insufficient room for saving captured contents. Treat as a non-capturing      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
862      bracket. */      as a non-capturing bracket. */
863    
864        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
865        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
866    
867      DPRINTF(("insufficient capture room: treat as non-capturing\n"));      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
868    
869      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870      final alternative within the brackets, we would return the result of a      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
     recursive call to match() whatever happened. We can reduce stack usage by  
     turning this into a tail recursion. */  
871    
872        /* Non-capturing or atomic group, except for possessive with unlimited
873        repeat. Loop for all the alternatives.
874    
875        When we get to the final alternative within the brackets, we used to return
876        the result of a recursive call to match() whatever happened so it was
877        possible to reduce stack usage by turning this into a tail recursion,
878        except in the case of a possibly empty group. However, now that there is
879        the possiblity of (*THEN) occurring in the final alternative, this
880        optimization is no longer always possible.
881    
882        We can optimize if we know there are no (*THEN)s in the pattern; at present
883        this is the best that can be done.
884    
885        MATCH_ONCE is returned when the end of an atomic group is successfully
886        reached, but subsequent matching fails. It passes back up the tree (causing
887        captured values to be reset) until the original atomic group level is
888        reached. This is tested by comparing md->once_target with the start of the
889        group. At this point, the return is converted into MATCH_NOMATCH so that
890        previous backup points can be taken. */
891    
892        case OP_ONCE:
893      case OP_BRA:      case OP_BRA:
894      case OP_SBRA:      case OP_SBRA:
895      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
896      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
897      for (;;)      for (;;)
898        {        {
899        if (ecode[GET(ecode, 1)] != OP_ALT)        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
900    
901          /* If this is not a possibly empty group, and there are no (*THEN)s in
902          the pattern, and this is the final alternative, optimize as described
903          above. */
904    
905          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
906          {          {
907          ecode += _pcre_OP_lengths[*ecode];          ecode += _pcre_OP_lengths[*ecode];
         flags |= match_tail_recursed;  
         DPRINTF(("bracket 0 tail recursion\n"));  
908          goto TAIL_RECURSE;          goto TAIL_RECURSE;
909            }
910    
911          /* In all other cases, we have to make another call to match(). */
912    
913          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
914            RM2);
915          if (rrc != MATCH_NOMATCH &&
916              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
917            {
918            if (rrc == MATCH_ONCE)
919              {
920              const uschar *scode = ecode;
921              if (*scode != OP_ONCE)           /* If not at start, find it */
922                {
923                while (*scode == OP_ALT) scode += GET(scode, 1);
924                scode -= GET(scode, 1);
925                }
926              if (md->once_target == scode) rrc = MATCH_NOMATCH;
927              }
928            RRETURN(rrc);
929          }          }
930          ecode += GET(ecode, 1);
931          if (*ecode != OP_ALT) break;
932          }
933        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
934        RRETURN(MATCH_NOMATCH);
935    
936        /* For non-final alternatives, continue the loop for a NOMATCH result;      /* Handle possessive capturing brackets with an unlimited repeat. We come
937        otherwise return. */      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
938        handled similarly to the normal case above. However, the matching is
939        different. The end of these brackets will always be OP_KETRPOS, which
940        returns MATCH_KETRPOS without going further in the pattern. By this means
941        we can handle the group by iteration rather than recursion, thereby
942        reducing the amount of stack needed. */
943    
944        case OP_CBRAPOS:
945        case OP_SCBRAPOS:
946        allow_zero = FALSE;
947    
948        RMATCH(rrc, eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      POSSESSIVE_CAPTURE:
949          eptrb, flags);      number = GET2(ecode, 1+LINK_SIZE);
950        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      offset = number << 1;
951    
952    #ifdef PCRE_DEBUG
953        printf("start possessive bracket %d\n", number);
954        printf("subject=");
955        pchars(eptr, 16, TRUE, md);
956        printf("\n");
957    #endif
958    
959        if (offset < md->offset_max)
960          {
961          matched_once = FALSE;
962          code_offset = ecode - md->start_code;
963    
964          save_offset1 = md->offset_vector[offset];
965          save_offset2 = md->offset_vector[offset+1];
966          save_offset3 = md->offset_vector[md->offset_end - number];
967          save_capture_last = md->capture_last;
968    
969          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
970    
971          /* Each time round the loop, save the current subject position for use
972          when the group matches. For MATCH_MATCH, the group has matched, so we
973          restart it with a new subject starting position, remembering that we had
974          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
975          usual. If we haven't matched any alternatives in any iteration, check to
976          see if a previous iteration matched. If so, the group has matched;
977          continue from afterwards. Otherwise it has failed; restore the previous
978          capture values before returning NOMATCH. */
979    
980          for (;;)
981            {
982            md->offset_vector[md->offset_end - number] =
983              (int)(eptr - md->start_subject);
984            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
985            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
986              eptrb, RM63);
987            if (rrc == MATCH_KETRPOS)
988              {
989              offset_top = md->end_offset_top;
990              eptr = md->end_match_ptr;
991              ecode = md->start_code + code_offset;
992              save_capture_last = md->capture_last;
993              matched_once = TRUE;
994              continue;
995              }
996            if (rrc != MATCH_NOMATCH &&
997                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
998              RRETURN(rrc);
999            md->capture_last = save_capture_last;
1000            ecode += GET(ecode, 1);
1001            if (*ecode != OP_ALT) break;
1002            }
1003    
1004          if (!matched_once)
1005            {
1006            md->offset_vector[offset] = save_offset1;
1007            md->offset_vector[offset+1] = save_offset2;
1008            md->offset_vector[md->offset_end - number] = save_offset3;
1009            }
1010    
1011          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
1012          if (allow_zero || matched_once)
1013            {
1014            ecode += 1 + LINK_SIZE;
1015            break;
1016            }
1017    
1018          RRETURN(MATCH_NOMATCH);
1019          }
1020    
1021        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1022        as a non-capturing bracket. */
1023    
1024        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1025        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1026    
1027        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1028    
1029        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1030        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1031    
1032        /* Non-capturing possessive bracket with unlimited repeat. We come here
1033        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1034        without the capturing complication. It is written out separately for speed
1035        and cleanliness. */
1036    
1037        case OP_BRAPOS:
1038        case OP_SBRAPOS:
1039        allow_zero = FALSE;
1040    
1041        POSSESSIVE_NON_CAPTURE:
1042        matched_once = FALSE;
1043        code_offset = ecode - md->start_code;
1044    
1045        for (;;)
1046          {
1047          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1048          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1049            eptrb, RM48);
1050          if (rrc == MATCH_KETRPOS)
1051            {
1052            offset_top = md->end_offset_top;
1053            eptr = md->end_match_ptr;
1054            ecode = md->start_code + code_offset;
1055            matched_once = TRUE;
1056            continue;
1057            }
1058          if (rrc != MATCH_NOMATCH &&
1059              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1060            RRETURN(rrc);
1061        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1062          if (*ecode != OP_ALT) break;
1063          }
1064    
1065        if (matched_once || allow_zero)
1066          {
1067          ecode += 1 + LINK_SIZE;
1068          break;
1069        }        }
1070        RRETURN(MATCH_NOMATCH);
1071    
1072      /* Control never reaches here. */      /* Control never reaches here. */
1073    
1074      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1075      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
1076      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
1077      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. */  
1078    
1079      case OP_COND:      case OP_COND:
1080      case OP_SCOND:      case OP_SCOND:
1081      if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */      codelink = GET(ecode, 1);
1082    
1083        /* Because of the way auto-callout works during compile, a callout item is
1084        inserted between OP_COND and an assertion condition. */
1085    
1086        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1087        {        {
1088        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (pcre_callout != NULL)
1089        condition = md->recursive != NULL &&          {
1090          (offset == RREF_ANY || offset == md->recursive->group_num);          pcre_callout_block cb;
1091        ecode += condition? 3 : GET(ecode, 1);          cb.version          = 2;   /* Version 1 of the callout block */
1092            cb.callout_number   = ecode[LINK_SIZE+2];
1093            cb.offset_vector    = md->offset_vector;
1094            cb.subject          = (PCRE_SPTR)md->start_subject;
1095            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1096            cb.start_match      = (int)(mstart - md->start_subject);
1097            cb.current_position = (int)(eptr - md->start_subject);
1098            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1099            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1100            cb.capture_top      = offset_top/2;
1101            cb.capture_last     = md->capture_last;
1102            cb.callout_data     = md->callout_data;
1103            cb.mark             = markptr;
1104            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1105            if (rrc < 0) RRETURN(rrc);
1106            }
1107          ecode += _pcre_OP_lengths[OP_CALLOUT];
1108          }
1109    
1110        condcode = ecode[LINK_SIZE+1];
1111    
1112        /* Now see what the actual condition is */
1113    
1114        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1115          {
1116          if (md->recursive == NULL)                /* Not recursing => FALSE */
1117            {
1118            condition = FALSE;
1119            ecode += GET(ecode, 1);
1120            }
1121          else
1122            {
1123            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1124            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1125    
1126            /* If the test is for recursion into a specific subpattern, and it is
1127            false, but the test was set up by name, scan the table to see if the
1128            name refers to any other numbers, and test them. The condition is true
1129            if any one is set. */
1130    
1131            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1132              {
1133              uschar *slotA = md->name_table;
1134              for (i = 0; i < md->name_count; i++)
1135                {
1136                if (GET2(slotA, 0) == recno) break;
1137                slotA += md->name_entry_size;
1138                }
1139    
1140              /* Found a name for the number - there can be only one; duplicate
1141              names for different numbers are allowed, but not vice versa. First
1142              scan down for duplicates. */
1143    
1144              if (i < md->name_count)
1145                {
1146                uschar *slotB = slotA;
1147                while (slotB > md->name_table)
1148                  {
1149                  slotB -= md->name_entry_size;
1150                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1151                    {
1152                    condition = GET2(slotB, 0) == md->recursive->group_num;
1153                    if (condition) break;
1154                    }
1155                  else break;
1156                  }
1157    
1158                /* Scan up for duplicates */
1159    
1160                if (!condition)
1161                  {
1162                  slotB = slotA;
1163                  for (i++; i < md->name_count; i++)
1164                    {
1165                    slotB += md->name_entry_size;
1166                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1167                      {
1168                      condition = GET2(slotB, 0) == md->recursive->group_num;
1169                      if (condition) break;
1170                      }
1171                    else break;
1172                    }
1173                  }
1174                }
1175              }
1176    
1177            /* Chose branch according to the condition */
1178    
1179            ecode += condition? 3 : GET(ecode, 1);
1180            }
1181        }        }
1182    
1183      else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1184        {        {
1185        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1186        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1187    
1188          /* If the numbered capture is unset, but the reference was by name,
1189          scan the table to see if the name refers to any other numbers, and test
1190          them. The condition is true if any one is set. This is tediously similar
1191          to the code above, but not close enough to try to amalgamate. */
1192    
1193          if (!condition && condcode == OP_NCREF)
1194            {
1195            int refno = offset >> 1;
1196            uschar *slotA = md->name_table;
1197    
1198            for (i = 0; i < md->name_count; i++)
1199              {
1200              if (GET2(slotA, 0) == refno) break;
1201              slotA += md->name_entry_size;
1202              }
1203    
1204            /* Found a name for the number - there can be only one; duplicate names
1205            for different numbers are allowed, but not vice versa. First scan down
1206            for duplicates. */
1207    
1208            if (i < md->name_count)
1209              {
1210              uschar *slotB = slotA;
1211              while (slotB > md->name_table)
1212                {
1213                slotB -= md->name_entry_size;
1214                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1215                  {
1216                  offset = GET2(slotB, 0) << 1;
1217                  condition = offset < offset_top &&
1218                    md->offset_vector[offset] >= 0;
1219                  if (condition) break;
1220                  }
1221                else break;
1222                }
1223    
1224              /* Scan up for duplicates */
1225    
1226              if (!condition)
1227                {
1228                slotB = slotA;
1229                for (i++; i < md->name_count; i++)
1230                  {
1231                  slotB += md->name_entry_size;
1232                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1233                    {
1234                    offset = GET2(slotB, 0) << 1;
1235                    condition = offset < offset_top &&
1236                      md->offset_vector[offset] >= 0;
1237                    if (condition) break;
1238                    }
1239                  else break;
1240                  }
1241                }
1242              }
1243            }
1244    
1245          /* Chose branch according to the condition */
1246    
1247        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 3 : GET(ecode, 1);
1248        }        }
1249    
1250      else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
1251        {        {
1252        condition = FALSE;        condition = FALSE;
1253        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1254        }        }
1255    
1256      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1257      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
1258      assertion. */      an assertion. */
1259    
1260      else      else
1261        {        {
1262        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1263            match_condassert);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1264        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1265          {          {
1266            if (md->end_offset_top > offset_top)
1267              offset_top = md->end_offset_top;  /* Captures may have happened */
1268          condition = TRUE;          condition = TRUE;
1269          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1270          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1271          }          }
1272        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH &&
1273                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1274          {          {
1275          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1276          }          }
1277        else        else
1278          {          {
1279          condition = FALSE;          condition = FALSE;
1280          ecode += GET(ecode, 1);          ecode += codelink;
1281          }          }
1282        }        }
1283    
1284      /* 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,
1285      we can use tail recursion to avoid using another stack frame. If the second      we used always to use tail recursion to avoid using another stack frame,
1286      alternative doesn't exist, we can just plough on. */      except when there was unlimited repeat of a possibly empty group. However,
1287        that strategy no longer works because of the possibilty of (*THEN) being
1288        encountered in the branch. However, we can still use tail recursion if
1289        there are no (*THEN)s in the pattern. Otherwise, a recursive call to
1290        match() is always required, unless the second alternative doesn't exist, in
1291        which case we can just plough on. */
1292    
1293      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1294        {        {
1295        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1296        flags = match_tail_recursed | ((op == OP_SCOND)? match_cbegroup : 0);        else if (!md->hasthen)
1297        goto TAIL_RECURSE;          {
1298        }          ecode += 1 + LINK_SIZE;
1299            goto TAIL_RECURSE;
1300            }
1301    
1302          /* A call to match() is required. */
1303    
1304          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1305    
1306          /* If the result is THEN from within the "true" branch of the condition,
1307          md->start_match_ptr will point to the original OP_COND, not to the start
1308          of the branch, so we have do work to see if it matches. If THEN comes
1309          from the "false" branch, md->start_match_ptr does point to OP_ALT. */
1310    
1311          if (rrc == MATCH_THEN)
1312            {
1313            if (*ecode != OP_ALT)
1314              {
1315              do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
1316              ecode -= GET(ecode, 1);
1317              }
1318            if (md->start_match_ptr == ecode) rrc = MATCH_NOMATCH;
1319            }
1320          RRETURN(rrc);
1321          }
1322    
1323         /* Condition false & no alternative; continue after the group. */
1324    
1325      else      else
1326        {        {
1327        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
# Line 748  for (;;) Line 1329  for (;;)
1329      break;      break;
1330    
1331    
1332      /* 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,
1333      restore the offsets appropriately and continue from after the call. */      to close any currently open capturing brackets. */
1334    
1335      case OP_END:      case OP_CLOSE:
1336      if (md->recursive != NULL && md->recursive->group_num == 0)      number = GET2(ecode, 1);
1337        offset = number << 1;
1338    
1339    #ifdef PCRE_DEBUG
1340          printf("end bracket %d at *ACCEPT", number);
1341          printf("\n");
1342    #endif
1343    
1344        md->capture_last = number;
1345        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1346        {        {
1347        recursion_info *rec = md->recursive;        md->offset_vector[offset] =
1348        DPRINTF(("End of pattern in a (?0) recursion\n"));          md->offset_vector[md->offset_end - number];
1349        md->recursive = rec->prevrec;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1350        memmove(md->offset_vector, rec->offset_save,        if (offset_top <= offset) offset_top = offset + 2;
         rec->saved_max * sizeof(int));  
       md->start_match = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
1351        }        }
1352        ecode += 3;
1353        break;
1354    
     /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty  
     string - backtracking will then try other alternatives, if any. */  
1355    
1356      if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH);      /* End of the pattern, either real or forced. */
1357      md->end_match_ptr = eptr;          /* Record where we ended */  
1358      md->end_offset_top = offset_top;   /* and how many extracts were taken */      case OP_END:
1359      RRETURN(MATCH_MATCH);      case OP_ACCEPT:
1360        case OP_ASSERT_ACCEPT:
1361      /* Change option settings */  
1362        /* If we have matched an empty string, fail if not in an assertion and not
1363      case OP_OPT:      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1364      ims = ecode[1];      is set and we have matched at the start of the subject. In both cases,
1365      ecode += 2;      backtracking will then try other alternatives, if any. */
1366      DPRINTF(("ims set to %02lx\n", ims));  
1367      break;      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1368             md->recursive == NULL &&
1369             (md->notempty ||
1370               (md->notempty_atstart &&
1371                 mstart == md->start_subject + md->start_offset)))
1372          MRRETURN(MATCH_NOMATCH);
1373    
1374        /* Otherwise, we have a match. */
1375    
1376        md->end_match_ptr = eptr;           /* Record where we ended */
1377        md->end_offset_top = offset_top;    /* and how many extracts were taken */
1378        md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1379    
1380        /* For some reason, the macros don't work properly if an expression is
1381        given as the argument to MRRETURN when the heap is in use. */
1382    
1383        rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1384        MRRETURN(rrc);
1385    
1386      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1387      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,
1388      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
1389      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
1390      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1391        of a condition, we want to return immediately afterwards. The caller of
1392        this incarnation of the match() function will have set MATCH_CONDASSERT in
1393        md->match_function type, and one of these opcodes will be the first opcode
1394        that is processed. We use a local variable that is preserved over calls to
1395        match() to remember this case. */
1396    
1397      case OP_ASSERT:      case OP_ASSERT:
1398      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1399        if (md->match_function_type == MATCH_CONDASSERT)
1400          {
1401          condassert = TRUE;
1402          md->match_function_type = 0;
1403          }
1404        else condassert = FALSE;
1405    
1406      do      do
1407        {        {
1408        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1409        if (rrc == MATCH_MATCH) break;        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1410        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          {
1411            mstart = md->start_match_ptr;   /* In case \K reset it */
1412            markptr = md->mark;
1413            break;
1414            }
1415          if (rrc != MATCH_NOMATCH &&
1416              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1417            RRETURN(rrc);
1418        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1419        }        }
1420      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1421      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1422        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1423    
1424      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1425    
1426      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1427    
1428      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1429      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 811  for (;;) Line 1433  for (;;)
1433      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1434      continue;      continue;
1435    
1436      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1437        PRUNE, or COMMIT means we must assume failure without checking subsequent
1438        branches. */
1439    
1440      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1441      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1442        if (md->match_function_type == MATCH_CONDASSERT)
1443          {
1444          condassert = TRUE;
1445          md->match_function_type = 0;
1446          }
1447        else condassert = FALSE;
1448    
1449      do      do
1450        {        {
1451        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1452        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1453        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1454            {
1455            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1456            break;
1457            }
1458          if (rrc != MATCH_NOMATCH &&
1459              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1460            RRETURN(rrc);
1461        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1462        }        }
1463      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1464    
1465      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1466    
1467      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1468      continue;      continue;
# Line 842  for (;;) Line 1480  for (;;)
1480        while (i-- > 0)        while (i-- > 0)
1481          {          {
1482          eptr--;          eptr--;
1483          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1484          BACKCHAR(eptr)          BACKCHAR(eptr);
1485          }          }
1486        }        }
1487      else      else
# Line 853  for (;;) Line 1491  for (;;)
1491    
1492        {        {
1493        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1494        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1495        }        }
1496    
1497      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1498    
1499        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1500      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1501      break;      break;
1502    
# Line 869  for (;;) Line 1508  for (;;)
1508      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1509        {        {
1510        pcre_callout_block cb;        pcre_callout_block cb;
1511        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1512        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1513        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1514        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1515        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1516        cb.start_match      = md->start_match - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1517        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1518        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1519        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1520        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1521        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1522        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1523        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        cb.mark             = markptr;
1524          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1525        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1526        }        }
1527      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 891  for (;;) Line 1531  for (;;)
1531      offset data is the offset to the starting bracket from the start of the      offset data is the offset to the starting bracket from the start of the
1532      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1533    
1534      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1535      save their starting points and reinstate them after the recursion. However,      re-instated afterwards. We don't know how many are started and not yet
1536      we don't know how many such there are (offset_top records the completed      finished (offset_top records the completed total) so we just have to save
1537      total) so we just have to save all the potential data. There may be up to      all the potential data. There may be up to 65535 such values, which is too
1538      65535 such values, which is too large to put on the stack, but using malloc      large to put on the stack, but using malloc for small numbers seems
1539      for small numbers seems expensive. As a compromise, the stack is used when      expensive. As a compromise, the stack is used when there are no more than
1540      there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc      REC_STACK_SAVE_MAX values to store; otherwise malloc is used.
     is used. A problem is what to do if the malloc fails ... there is no way of  
     returning to the top level with an error. Save the top REC_STACK_SAVE_MAX  
     values on the stack, and accept that the rest may be wrong.  
1541    
1542      There are also other values that have to be saved. We use a chained      There are also other values that have to be saved. We use a chained
1543      sequence of blocks that actually live on the stack. Thanks to Robin Houston      sequence of blocks that actually live on the stack. Thanks to Robin Houston
1544      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1545        a lot, so he is not to blame for the current way it works. */
1546    
1547      case OP_RECURSE:      case OP_RECURSE:
1548        {        {
1549          recursion_info *ri;
1550          int recno;
1551    
1552        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1553        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1554          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1555    
1556          /* Check for repeating a recursion without advancing the subject pointer.
1557          This should catch convoluted mutual recursions. (Some simple cases are
1558          caught at compile time.) */
1559    
1560          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1561            if (recno == ri->group_num && eptr == ri->subject_position)
1562              RRETURN(PCRE_ERROR_RECURSELOOP);
1563    
1564        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1565    
1566          new_recursive.group_num = recno;
1567          new_recursive.subject_position = eptr;
1568        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1569        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1570    
1571        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1572    
1573        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1574    
1575        /* Now save the offset data. */        /* Now save the offset data */
1576    
1577        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1578        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 933  for (;;) Line 1583  for (;;)
1583            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1584          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1585          }          }
   
1586        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1587              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = md->start_match;  
       md->start_match = eptr;  
1588    
1589        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1590        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1591          might be changed, so reset it before looping. */
1592    
1593        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1594        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1595        do        do
1596          {          {
1597          RMATCH(rrc, eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1598            md, ims, eptrb, flags);          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1599          if (rrc == MATCH_MATCH)            md, eptrb, RM6);
1600            memcpy(md->offset_vector, new_recursive.offset_save,
1601                new_recursive.saved_max * sizeof(int));
1602            md->recursive = new_recursive.prevrec;
1603            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1604            {            {
1605            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1606            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1607              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1608            RRETURN(MATCH_MATCH);  
1609              /* Set where we got to in the subject, and reset the start in case
1610              it was changed by \K. This *is* propagated back out of a recursion,
1611              for Perl compatibility. */
1612    
1613              eptr = md->end_match_ptr;
1614              mstart = md->start_match_ptr;
1615              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1616            }            }
1617          else if (rrc != MATCH_NOMATCH)          else if (rrc != MATCH_NOMATCH &&
1618                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1619            {            {
1620            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1621              if (new_recursive.offset_save != stacksave)
1622                (pcre_free)(new_recursive.offset_save);
1623            RRETURN(rrc);            RRETURN(rrc);
1624            }            }
1625    
1626          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1627          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1628          }          }
1629        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 973  for (;;) Line 1632  for (;;)
1632        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1633        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1634          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1635        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1636        }        }
     /* Control never reaches here */  
1637    
1638      /* "Once" brackets are like assertion brackets except that after a match,      RECURSION_MATCHED:
1639      the point in the subject string is not moved back. Thus there can never be      break;
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer. */  
1640    
1641      case OP_ONCE:      /* An alternation is the end of a branch; scan along to find the end of the
1642      prev = ecode;      bracketed group and go to there. */
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,  
         eptrb, 0);  
       if (rrc == MATCH_MATCH) break;  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = match_tail_recursed;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_cbegroup);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = match_tail_recursed;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
   
     /* An alternation is the end of a branch; scan along to find the end of the  
     bracketed group and go to there. */  
1643    
1644      case OP_ALT:      case OP_ALT:
1645      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1646      break;      break;
1647    
1648      /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1649      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
1650      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
1651      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
1652      preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1653    
1654      case OP_BRAZERO:      case OP_BRAZERO:
1655        {      next = ecode + 1;
1656        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1657        RMATCH(rrc, eptr, next, offset_top, md, ims, eptrb, 0);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1658        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1659        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1660      break;      break;
1661    
1662      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1663        {      next = ecode + 1;
1664        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1665        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1666        RMATCH(rrc, eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1667        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
1668        ecode++;      break;
1669        }  
1670        case OP_SKIPZERO:
1671        next = ecode+1;
1672        do next += GET(next,1); while (*next == OP_ALT);
1673        ecode = next + 1 + LINK_SIZE;
1674      break;      break;
1675    
1676        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1677        here; just jump to the group, with allow_zero set TRUE. */
1678    
1679        case OP_BRAPOSZERO:
1680        op = *(++ecode);
1681        allow_zero = TRUE;
1682        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1683          goto POSSESSIVE_NON_CAPTURE;
1684    
1685      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1686    
1687      case OP_KET:      case OP_KET:
1688      case OP_KETRMIN:      case OP_KETRMIN:
1689      case OP_KETRMAX:      case OP_KETRMAX:
1690        case OP_KETRPOS:
1691      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1692    
1693      /* 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
1694      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1695      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1696    
1697      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1698        {        {
1699        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1700        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
# Line 1105  for (;;) Line 1703  for (;;)
1703    
1704      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group, stop matching and return
1705      MATCH_MATCH, but record the current high water mark for use by positive      MATCH_MATCH, but record the current high water mark for use by positive
1706      assertions. Do this also for the "once" (atomic) groups. */      assertions. We also need to record the match start in case it was changed
1707        by \K. */
1708    
1709      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1710          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1711        {        {
1712        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1713        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1714        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1715          MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1716        }        }
1717    
1718      /* 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
1719      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1720      bumping the high water mark. Note that whole-pattern recursion is coded as      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1721      a recurse into group 0, so it won't be picked up here. Instead, we catch it      into group 0, so it won't be picked up here. Instead, we catch it when the
1722      when the OP_END is reached. Other recursion is handled here. */      OP_END is reached. Other recursion is handled here. We just have to record
1723        the current subject position and start match pointer and give a MATCH
1724        return. */
1725    
1726      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1727            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1728        {        {
1729        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1730        offset = number << 1;        offset = number << 1;
1731    
1732  #ifdef DEBUG  #ifdef PCRE_DEBUG
1733        printf("end bracket %d", number);        printf("end bracket %d", number);
1734        printf("\n");        printf("\n");
1735  #endif  #endif
1736    
1737          /* Handle a recursively called group. */
1738    
1739          if (md->recursive != NULL && md->recursive->group_num == number)
1740            {
1741            md->end_match_ptr = eptr;
1742            md->start_match_ptr = mstart;
1743            RRETURN(MATCH_MATCH);
1744            }
1745    
1746          /* Deal with capturing */
1747    
1748        md->capture_last = number;        md->capture_last = number;
1749        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1750          {          {
1751            /* If offset is greater than offset_top, it means that we are
1752            "skipping" a capturing group, and that group's offsets must be marked
1753            unset. In earlier versions of PCRE, all the offsets were unset at the
1754            start of matching, but this doesn't work because atomic groups and
1755            assertions can cause a value to be set that should later be unset.
1756            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1757            part of the atomic group, but this is not on the final matching path,
1758            so must be unset when 2 is set. (If there is no group 2, there is no
1759            problem, because offset_top will then be 2, indicating no capture.) */
1760    
1761            if (offset > offset_top)
1762              {
1763              register int *iptr = md->offset_vector + offset_top;
1764              register int *iend = md->offset_vector + offset;
1765              while (iptr < iend) *iptr++ = -1;
1766              }
1767    
1768            /* Now make the extraction */
1769    
1770          md->offset_vector[offset] =          md->offset_vector[offset] =
1771            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1772          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1773          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1774          }          }
1775          }
1776    
1777        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1778        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1779        group. This is the forcible breaking of infinite loops as implemented in
1780        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1781        processing the rest of the pattern at a lower level. If this results in a
1782        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1783        bypassing intermediate backup points, but resetting any captures that
1784        happened along the way. */
1785    
1786        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1787          {
1788          if (*prev == OP_ONCE)
1789          {          {
1790          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1791          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1792          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1793          md->start_match = rec->save_start;          RRETURN(MATCH_ONCE);
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1794          }          }
1795          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1796          break;
1797        }        }
1798    
1799      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1800      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1801        at a time from the outer level, thus saving stack. */
1802    
1803      ims = original_ims;      if (*ecode == OP_KETRPOS)
     DPRINTF(("ims reset to %02lx\n", ims));  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
1804        {        {
1805        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1806        break;        md->end_offset_top = offset_top;
1807          RRETURN(MATCH_KETRPOS);
1808        }        }
1809    
1810      /* The repeating kets try the rest of the pattern or restart from the      /* The normal repeating kets try the rest of the pattern or restart from
1811      preceding bracket, in the appropriate order. In the second case, we can use      the preceding bracket, in the appropriate order. In the second case, we can
1812      tail recursion to avoid using another stack frame. */      use tail recursion to avoid using another stack frame, unless we have an
1813        an atomic group or an unlimited repeat of a group that can match an empty
1814      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;      string. */
1815    
1816      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1817        {        {
1818        RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1819        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1820          if (*prev == OP_ONCE)
1821            {
1822            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1823            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1824            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1825            RRETURN(MATCH_ONCE);
1826            }
1827          if (*prev >= OP_SBRA)    /* Could match an empty string */
1828            {
1829            md->match_function_type = MATCH_CBEGROUP;
1830            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1831            RRETURN(rrc);
1832            }
1833        ecode = prev;        ecode = prev;
       flags |= match_tail_recursed;  
1834        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1835        }        }
1836      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1837        {        {
1838        RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, flags);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1839          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1840          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1841        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1842          if (*prev == OP_ONCE)
1843            {
1844            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1845            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1846            md->once_target = prev;
1847            RRETURN(MATCH_ONCE);
1848            }
1849        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = match_tail_recursed;  
1850        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1851        }        }
1852      /* Control never gets here */      /* Control never gets here */
1853    
1854      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1855    
1856      case OP_CIRC:      case OP_CIRC:
1857      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
     if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         RRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
1858    
1859      /* Start of subject assertion */      /* Start of subject assertion */
1860    
1861      case OP_SOD:      case OP_SOD:
1862      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1863        ecode++;
1864        break;
1865    
1866        /* Multiline mode: start of subject unless notbol, or after any newline. */
1867    
1868        case OP_CIRCM:
1869        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1870        if (eptr != md->start_subject &&
1871            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1872          MRRETURN(MATCH_NOMATCH);
1873      ecode++;      ecode++;
1874      break;      break;
1875    
1876      /* Start of match assertion */      /* Start of match assertion */
1877    
1878      case OP_SOM:      case OP_SOM:
1879      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1880      ecode++;      ecode++;
1881      break;      break;
1882    
1883      /* 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. */  
1884    
1885      case OP_DOLL:      case OP_SET_SOM:
1886      if ((ims & PCRE_MULTILINE) != 0)      mstart = eptr;
1887        {      ecode++;
1888        if (eptr < md->end_subject)      break;
1889          { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }  
1890        else      /* Multiline mode: assert before any newline, or before end of subject
1891          { if (md->noteol) RRETURN(MATCH_NOMATCH); }      unless noteol is set. */
1892        ecode++;  
1893        break;      case OP_DOLLM:
1894        }      if (eptr < md->end_subject)
1895          { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1896      else      else
1897        {        {
1898        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1899        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1900        }        }
1901        ecode++;
1902        break;
1903    
1904        /* Not multiline mode: assert before a terminating newline or before end of
1905        subject unless noteol is set. */
1906    
1907        case OP_DOLL:
1908        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1909        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1910    
1911      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1912    
1913      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1914    
1915      case OP_EOD:      case OP_EOD:
1916      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1917        SCHECK_PARTIAL();
1918      ecode++;      ecode++;
1919      break;      break;
1920    
1921      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1922    
1923      case OP_EODN:      case OP_EODN:
1924      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1925        if (eptr < md->end_subject &&
1926          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1927        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1928    
1929        /* Either at end of string or \n before end. */
1930    
1931        SCHECK_PARTIAL();
1932      ecode++;      ecode++;
1933      break;      break;
1934    
# Line 1279  for (;;) Line 1940  for (;;)
1940    
1941        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1942        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
1943        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1944          partial matching. */
1945    
1946  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1947        if (utf8)        if (utf8)
1948          {          {
1949            /* Get status of previous character */
1950    
1951          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1952            {            {
1953            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1954            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1955              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1956            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1957    #ifdef SUPPORT_UCP
1958              if (md->use_ucp)
1959                {
1960                if (c == '_') prev_is_word = TRUE; else
1961                  {
1962                  int cat = UCD_CATEGORY(c);
1963                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1964                  }
1965                }
1966              else
1967    #endif
1968            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1969            }            }
1970          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
1971            /* Get status of next character */
1972    
1973            if (eptr >= md->end_subject)
1974              {
1975              SCHECK_PARTIAL();
1976              cur_is_word = FALSE;
1977              }
1978            else
1979            {            {
1980            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1981    #ifdef SUPPORT_UCP
1982              if (md->use_ucp)
1983                {
1984                if (c == '_') cur_is_word = TRUE; else
1985                  {
1986                  int cat = UCD_CATEGORY(c);
1987                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1988                  }
1989                }
1990              else
1991    #endif
1992            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1993            }            }
1994          }          }
1995        else        else
1996  #endif  #endif
1997    
1998        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1999          consistency with the behaviour of \w we do use it in this case. */
2000    
2001          {          {
2002          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
2003            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
2004          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
2005            ((md->ctypes[*eptr] & ctype_word) != 0);            {
2006              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
2007    #ifdef SUPPORT_UCP
2008              if (md->use_ucp)
2009                {
2010                c = eptr[-1];
2011                if (c == '_') prev_is_word = TRUE; else
2012                  {
2013                  int cat = UCD_CATEGORY(c);
2014                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2015                  }
2016                }
2017              else
2018    #endif
2019              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2020              }
2021    
2022            /* Get status of next character */
2023    
2024            if (eptr >= md->end_subject)
2025              {
2026              SCHECK_PARTIAL();
2027              cur_is_word = FALSE;
2028              }
2029            else
2030    #ifdef SUPPORT_UCP
2031            if (md->use_ucp)
2032              {
2033              c = *eptr;
2034              if (c == '_') cur_is_word = TRUE; else
2035                {
2036                int cat = UCD_CATEGORY(c);
2037                cur_is_word = (cat == ucp_L || cat == ucp_N);
2038                }
2039              }
2040            else
2041    #endif
2042            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2043          }          }
2044    
2045        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2046    
2047        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2048             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2049          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2050        }        }
2051      break;      break;
2052    
2053      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2054    
2055      case OP_ANY:      case OP_ANY:
2056      if ((ims & PCRE_DOTALL) == 0)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2057        {      /* Fall through */
2058        if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);  
2059        case OP_ALLANY:
2060        if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2061          {                            /* not be updated before SCHECK_PARTIAL. */
2062          SCHECK_PARTIAL();
2063          MRRETURN(MATCH_NOMATCH);
2064        }        }
2065      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      eptr++;
2066      if (utf8)      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
2067      ecode++;      ecode++;
2068      break;      break;
2069    
# Line 1334  for (;;) Line 2071  for (;;)
2071      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2072    
2073      case OP_ANYBYTE:      case OP_ANYBYTE:
2074      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2075          {                            /* not be updated before SCHECK_PARTIAL. */
2076          SCHECK_PARTIAL();
2077          MRRETURN(MATCH_NOMATCH);
2078          }
2079        eptr++;
2080      ecode++;      ecode++;
2081      break;      break;
2082    
2083      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2084      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2085          {
2086          SCHECK_PARTIAL();
2087          MRRETURN(MATCH_NOMATCH);
2088          }
2089      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2090      if (      if (
2091  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1347  for (;;) Line 2093  for (;;)
2093  #endif  #endif
2094         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2095         )         )
2096        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2097      ecode++;      ecode++;
2098      break;      break;
2099    
2100      case OP_DIGIT:      case OP_DIGIT:
2101      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2102          {
2103          SCHECK_PARTIAL();
2104          MRRETURN(MATCH_NOMATCH);
2105          }
2106      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2107      if (      if (
2108  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1360  for (;;) Line 2110  for (;;)
2110  #endif  #endif
2111         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2112         )         )
2113        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2114      ecode++;      ecode++;
2115      break;      break;
2116    
2117      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2118      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2119          {
2120          SCHECK_PARTIAL();
2121          MRRETURN(MATCH_NOMATCH);
2122          }
2123      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2124      if (      if (
2125  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1373  for (;;) Line 2127  for (;;)
2127  #endif  #endif
2128         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2129         )         )
2130        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2131      ecode++;      ecode++;
2132      break;      break;
2133    
2134      case OP_WHITESPACE:      case OP_WHITESPACE:
2135      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2136          {
2137          SCHECK_PARTIAL();
2138          MRRETURN(MATCH_NOMATCH);
2139          }
2140      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2141      if (      if (
2142  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1386  for (;;) Line 2144  for (;;)
2144  #endif  #endif
2145         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2146         )         )
2147        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2148      ecode++;      ecode++;
2149      break;      break;
2150    
2151      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2152      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2153          {
2154          SCHECK_PARTIAL();
2155          MRRETURN(MATCH_NOMATCH);
2156          }
2157      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2158      if (      if (
2159  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1399  for (;;) Line 2161  for (;;)
2161  #endif  #endif
2162         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2163         )         )
2164        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2165      ecode++;      ecode++;
2166      break;      break;
2167    
2168      case OP_WORDCHAR:      case OP_WORDCHAR:
2169      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2170          {
2171          SCHECK_PARTIAL();
2172          MRRETURN(MATCH_NOMATCH);
2173          }
2174      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2175      if (      if (
2176  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1412  for (;;) Line 2178  for (;;)
2178  #endif  #endif
2179         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2180         )         )
2181        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2182      ecode++;      ecode++;
2183      break;      break;
2184    
2185      case OP_ANYNL:      case OP_ANYNL:
2186      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2187          {
2188          SCHECK_PARTIAL();
2189          MRRETURN(MATCH_NOMATCH);
2190          }
2191      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2192      switch(c)      switch(c)
2193        {        {
2194        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2195    
2196        case 0x000d:        case 0x000d:
2197        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2198        break;        break;
2199    
2200        case 0x000a:        case 0x000a:
2201          break;
2202    
2203        case 0x000b:        case 0x000b:
2204        case 0x000c:        case 0x000c:
2205        case 0x0085:        case 0x0085:
2206        case 0x2028:        case 0x2028:
2207        case 0x2029:        case 0x2029:
2208          if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2209          break;
2210          }
2211        ecode++;
2212        break;
2213    
2214        case OP_NOT_HSPACE:
2215        if (eptr >= md->end_subject)
2216          {
2217          SCHECK_PARTIAL();
2218          MRRETURN(MATCH_NOMATCH);
2219          }
2220        GETCHARINCTEST(c, eptr);
2221        switch(c)
2222          {
2223          default: break;
2224          case 0x09:      /* HT */
2225          case 0x20:      /* SPACE */
2226          case 0xa0:      /* NBSP */
2227          case 0x1680:    /* OGHAM SPACE MARK */
2228          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2229          case 0x2000:    /* EN QUAD */
2230          case 0x2001:    /* EM QUAD */
2231          case 0x2002:    /* EN SPACE */
2232          case 0x2003:    /* EM SPACE */
2233          case 0x2004:    /* THREE-PER-EM SPACE */
2234          case 0x2005:    /* FOUR-PER-EM SPACE */
2235          case 0x2006:    /* SIX-PER-EM SPACE */
2236          case 0x2007:    /* FIGURE SPACE */
2237          case 0x2008:    /* PUNCTUATION SPACE */
2238          case 0x2009:    /* THIN SPACE */
2239          case 0x200A:    /* HAIR SPACE */
2240          case 0x202f:    /* NARROW NO-BREAK SPACE */
2241          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2242          case 0x3000:    /* IDEOGRAPHIC SPACE */
2243          MRRETURN(MATCH_NOMATCH);
2244          }
2245        ecode++;
2246        break;
2247    
2248        case OP_HSPACE:
2249        if (eptr >= md->end_subject)
2250          {
2251          SCHECK_PARTIAL();
2252          MRRETURN(MATCH_NOMATCH);
2253          }
2254        GETCHARINCTEST(c, eptr);
2255        switch(c)
2256          {
2257          default: MRRETURN(MATCH_NOMATCH);
2258          case 0x09:      /* HT */
2259          case 0x20:      /* SPACE */
2260          case 0xa0:      /* NBSP */
2261          case 0x1680:    /* OGHAM SPACE MARK */
2262          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2263          case 0x2000:    /* EN QUAD */
2264          case 0x2001:    /* EM QUAD */
2265          case 0x2002:    /* EN SPACE */
2266          case 0x2003:    /* EM SPACE */
2267          case 0x2004:    /* THREE-PER-EM SPACE */
2268          case 0x2005:    /* FOUR-PER-EM SPACE */
2269          case 0x2006:    /* SIX-PER-EM SPACE */
2270          case 0x2007:    /* FIGURE SPACE */
2271          case 0x2008:    /* PUNCTUATION SPACE */
2272          case 0x2009:    /* THIN SPACE */
2273          case 0x200A:    /* HAIR SPACE */
2274          case 0x202f:    /* NARROW NO-BREAK SPACE */
2275          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2276          case 0x3000:    /* IDEOGRAPHIC SPACE */
2277          break;
2278          }
2279        ecode++;
2280        break;
2281    
2282        case OP_NOT_VSPACE:
2283        if (eptr >= md->end_subject)
2284          {
2285          SCHECK_PARTIAL();
2286          MRRETURN(MATCH_NOMATCH);
2287          }
2288        GETCHARINCTEST(c, eptr);
2289        switch(c)
2290          {
2291          default: break;
2292          case 0x0a:      /* LF */
2293          case 0x0b:      /* VT */
2294          case 0x0c:      /* FF */
2295          case 0x0d:      /* CR */
2296          case 0x85:      /* NEL */
2297          case 0x2028:    /* LINE SEPARATOR */
2298          case 0x2029:    /* PARAGRAPH SEPARATOR */
2299          MRRETURN(MATCH_NOMATCH);
2300          }
2301        ecode++;
2302        break;
2303    
2304        case OP_VSPACE:
2305        if (eptr >= md->end_subject)
2306          {
2307          SCHECK_PARTIAL();
2308          MRRETURN(MATCH_NOMATCH);
2309          }
2310        GETCHARINCTEST(c, eptr);
2311        switch(c)
2312          {
2313          default: MRRETURN(MATCH_NOMATCH);
2314          case 0x0a:      /* LF */
2315          case 0x0b:      /* VT */
2316          case 0x0c:      /* FF */
2317          case 0x0d:      /* CR */
2318          case 0x85:      /* NEL */
2319          case 0x2028:    /* LINE SEPARATOR */
2320          case 0x2029:    /* PARAGRAPH SEPARATOR */
2321        break;        break;
2322        }        }
2323      ecode++;      ecode++;
# Line 1442  for (;;) Line 2329  for (;;)
2329    
2330      case OP_PROP:      case OP_PROP:
2331      case OP_NOTPROP:      case OP_NOTPROP:
2332      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2333          {
2334          SCHECK_PARTIAL();
2335          MRRETURN(MATCH_NOMATCH);
2336          }
2337      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2338        {        {
2339        int chartype, script;        const ucd_record *prop = GET_UCD(c);
       int category = _pcre_ucp_findprop(c, &chartype, &script);  
2340    
2341        switch(ecode[1])        switch(ecode[1])
2342          {          {
2343          case PT_ANY:          case PT_ANY:
2344          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2345          break;          break;
2346    
2347          case PT_LAMP:          case PT_LAMP:
2348          if ((chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2349               chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2350               chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2351            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2352           break;          break;
2353    
2354          case PT_GC:          case PT_GC:
2355          if ((ecode[2] != category) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2356            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2357          break;          break;
2358    
2359          case PT_PC:          case PT_PC:
2360          if ((ecode[2] != chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2361            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2362          break;          break;
2363    
2364          case PT_SC:          case PT_SC:
2365          if ((ecode[2] != script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2366            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2367            break;
2368    
2369            /* These are specials */
2370    
2371            case PT_ALNUM:
2372            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2373                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2374              MRRETURN(MATCH_NOMATCH);
2375          break;          break;
2376    
2377            case PT_SPACE:    /* Perl space */
2378            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2379                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2380                   == (op == OP_NOTPROP))
2381              MRRETURN(MATCH_NOMATCH);
2382            break;
2383    
2384            case PT_PXSPACE:  /* POSIX space */
2385            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2386                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2387                 c == CHAR_FF || c == CHAR_CR)
2388                   == (op == OP_NOTPROP))
2389              MRRETURN(MATCH_NOMATCH);
2390            break;
2391    
2392            case PT_WORD:
2393            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2394                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2395                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2396              MRRETURN(MATCH_NOMATCH);
2397            break;
2398    
2399            /* This should never occur */
2400    
2401          default:          default:
2402          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2403          }          }
# Line 1488  for (;;) Line 2410  for (;;)
2410      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2411    
2412      case OP_EXTUNI:      case OP_EXTUNI:
2413      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2414          {
2415          SCHECK_PARTIAL();
2416          MRRETURN(MATCH_NOMATCH);
2417          }
2418      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2419        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2420        while (eptr < md->end_subject)
2421        {        {
2422        int chartype, script;        int len = 1;
2423        int category = _pcre_ucp_findprop(c, &chartype, &script);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2424        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (UCD_CATEGORY(c) != ucp_M) break;
2425        while (eptr < md->end_subject)        eptr += len;
         {  
         int len = 1;  
         if (!utf8) c = *eptr; else  
           {  
           GETCHARLEN(c, eptr, len);  
           }  
         category = _pcre_ucp_findprop(c, &chartype, &script);  
         if (category != ucp_M) break;  
         eptr += len;  
         }  
2426        }        }
2427      ecode++;      ecode++;
2428      break;      break;
# Line 1520  for (;;) Line 2438  for (;;)
2438      loops). */      loops). */
2439    
2440      case OP_REF:      case OP_REF:
2441        {      case OP_REFI:
2442        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2443        ecode += 3;                                 /* Advance past item */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2444        ecode += 3;
2445    
2446        /* 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];  
2447    
2448        /* Set up for repetition, or handle the non-repeated case */      (a) In the default, Perl-compatible state, set the length negative;
2449        this ensures that every attempt at a match fails. We can't just fail
2450        here, because of the possibility of quantifiers with zero minima.
2451    
2452        switch (*ecode)      (b) If the JavaScript compatibility flag is set, set the length to zero
2453          {      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;  
2454    
2455          case OP_CRRANGE:      Otherwise, set the length to the length of what was matched by the
2456          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;  
2457    
2458          default:               /* No repeat follows */      if (offset >= offset_top || md->offset_vector[offset] < 0)
2459          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);        length = (md->jscript_compat)? 0 : -1;
2460          eptr += length;      else
2461          continue;              /* With the main loop */        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2462    
2463        /* Set up for repetition, or handle the non-repeated case */
2464    
2465        switch (*ecode)
2466          {
2467          case OP_CRSTAR:
2468          case OP_CRMINSTAR:
2469          case OP_CRPLUS:
2470          case OP_CRMINPLUS:
2471          case OP_CRQUERY:
2472          case OP_CRMINQUERY:
2473          c = *ecode++ - OP_CRSTAR;
2474          minimize = (c & 1) != 0;
2475          min = rep_min[c];                 /* Pick up values from tables; */
2476          max = rep_max[c];                 /* zero for max => infinity */
2477          if (max == 0) max = INT_MAX;
2478          break;
2479    
2480          case OP_CRRANGE:
2481          case OP_CRMINRANGE:
2482          minimize = (*ecode == OP_CRMINRANGE);
2483          min = GET2(ecode, 1);
2484          max = GET2(ecode, 3);
2485          if (max == 0) max = INT_MAX;
2486          ecode += 5;
2487          break;
2488    
2489          default:               /* No repeat follows */
2490          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2491            {
2492            CHECK_PARTIAL();
2493            MRRETURN(MATCH_NOMATCH);
2494          }          }
2495          eptr += length;
2496          continue;              /* With the main loop */
2497          }
2498    
2499        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2500        main loop. */      zero, just continue with the main loop. */
2501    
2502        if (length == 0) continue;      if (length == 0) continue;
2503    
2504        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2505        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2506        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2507    
2508        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2509          {
2510          int slength;
2511          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2512          {          {
2513          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2514          eptr += length;          MRRETURN(MATCH_NOMATCH);
2515          }          }
2516          eptr += slength;
2517          }
2518    
2519        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2520        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2521    
2522        if (min == max) continue;      if (min == max) continue;
2523    
2524        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2525    
2526        if (minimize)      if (minimize)
2527          {
2528          for (fi = min;; fi++)
2529          {          {
2530          for (fi = min;; fi++)          int slength;
2531            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2532            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2533            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2534            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2535            {            {
2536            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            CHECK_PARTIAL();
2537            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2538            }            }
2539          /* Control never gets here */          eptr += slength;
2540          }          }
2541          /* Control never gets here */
2542          }
2543    
2544        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2545    
2546        else      else
2547          {
2548          pp = eptr;
2549          for (i = min; i < max; i++)
2550          {          {
2551          pp = eptr;          int slength;
2552          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)  
2553            {            {
2554            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            CHECK_PARTIAL();
2555            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2556            }            }
2557          RRETURN(MATCH_NOMATCH);          eptr += slength;
2558          }          }
2559          while (eptr >= pp)
2560            {
2561            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2562            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2563            eptr -= length;
2564            }
2565          MRRETURN(MATCH_NOMATCH);
2566        }        }
2567      /* Control never gets here */      /* Control never gets here */
2568    
   
   
2569      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2570      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,
2571      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 2620  for (;;)
2620          {          {
2621          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2622            {            {
2623            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2624                {
2625                SCHECK_PARTIAL();
2626                MRRETURN(MATCH_NOMATCH);
2627                }
2628            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2629            if (c > 255)            if (c > 255)
2630              {              {
2631              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2632              }              }
2633            else            else
2634              {              {
2635              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2636              }              }
2637            }            }
2638          }          }
# Line 1695  for (;;) Line 2642  for (;;)
2642          {          {
2643          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2644            {            {
2645            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2646                {
2647                SCHECK_PARTIAL();
2648                MRRETURN(MATCH_NOMATCH);
2649                }
2650            c = *eptr++;            c = *eptr++;
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 1717  for (;;) Line 2668  for (;;)
2668            {            {
2669            for (fi = min;; fi++)            for (fi = min;; fi++)
2670              {              {
2671              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2672              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2673              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2674                if (eptr >= md->end_subject)
2675                  {
2676                  SCHECK_PARTIAL();
2677                  MRRETURN(MATCH_NOMATCH);
2678                  }
2679              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2680              if (c > 255)              if (c > 255)
2681                {                {
2682                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2683                }                }
2684              else              else
2685                {                {
2686                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2687                }                }
2688              }              }
2689            }            }
# Line 1737  for (;;) Line 2693  for (;;)
2693            {            {
2694            for (fi = min;; fi++)            for (fi = min;; fi++)
2695              {              {
2696              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2697              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2698              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2699                if (eptr >= md->end_subject)
2700                  {
2701                  SCHECK_PARTIAL();
2702                  MRRETURN(MATCH_NOMATCH);
2703                  }
2704              c = *eptr++;              c = *eptr++;
2705              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2706              }              }
2707            }            }
2708          /* Control never gets here */          /* Control never gets here */
# Line 1760  for (;;) Line 2721  for (;;)
2721            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2722              {              {
2723              int len = 1;              int len = 1;
2724              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2725                  {
2726                  SCHECK_PARTIAL();
2727                  break;
2728                  }
2729              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2730              if (c > 255)              if (c > 255)
2731                {                {
# Line 1774  for (;;) Line 2739  for (;;)
2739              }              }
2740            for (;;)            for (;;)
2741              {              {
2742              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2743              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2744              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2745              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1786  for (;;) Line 2751  for (;;)
2751            {            {
2752            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2753              {              {
2754              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2755                  {
2756                  SCHECK_PARTIAL();
2757                  break;
2758                  }
2759              c = *eptr;              c = *eptr;
2760              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2761              eptr++;              eptr++;
2762              }              }
2763            while (eptr >= pp)            while (eptr >= pp)
2764              {              {
2765              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2766              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2767              eptr--;              eptr--;
2768              }              }
2769            }            }
2770    
2771          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2772          }          }
2773        }        }
2774      /* Control never gets here */      /* Control never gets here */
2775    
2776    
2777      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2778      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
2779        mode, because Unicode properties are supported in non-UTF-8 mode. */
2780    
2781  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2782      case OP_XCLASS:      case OP_XCLASS:
# Line 1847  for (;;) Line 2817  for (;;)
2817    
2818        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2819          {          {
2820          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2821          GETCHARINC(c, eptr);            {
2822          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2823              MRRETURN(MATCH_NOMATCH);
2824              }
2825            GETCHARINCTEST(c, eptr);
2826            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2827          }          }
2828    
2829        /* 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 2838  for (;;)
2838          {          {
2839          for (fi = min;; fi++)          for (fi = min;; fi++)
2840            {            {
2841            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2842            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2843            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2844            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2845            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2846                SCHECK_PARTIAL();
2847                MRRETURN(MATCH_NOMATCH);
2848                }
2849              GETCHARINCTEST(c, eptr);
2850              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2851            }            }
2852          /* Control never gets here */          /* Control never gets here */
2853          }          }
# Line 1881  for (;;) Line 2860  for (;;)
2860          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2861            {            {
2862            int len = 1;            int len = 1;
2863            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2864            GETCHARLEN(c, eptr, len);              {
2865                SCHECK_PARTIAL();
2866                break;
2867                }
2868              GETCHARLENTEST(c, eptr, len);
2869            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2870            eptr += len;            eptr += len;
2871            }            }
2872          for(;;)          for(;;)
2873            {            {
2874            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2875            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2876            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2877            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2878            }            }
2879          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2880          }          }
2881    
2882        /* Control never gets here */        /* Control never gets here */
# Line 1909  for (;;) Line 2892  for (;;)
2892        length = 1;        length = 1;
2893        ecode++;        ecode++;
2894        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2895        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2896        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2897            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2898            MRRETURN(MATCH_NOMATCH);
2899            }
2900          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2901        }        }
2902      else      else
2903  #endif  #endif
2904    
2905      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2906        {        {
2907        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2908        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2909            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2910            MRRETURN(MATCH_NOMATCH);
2911            }
2912          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2913        ecode += 2;        ecode += 2;
2914        }        }
2915      break;      break;
2916    
2917      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2918    
2919      case OP_CHARNC:      case OP_CHARI:
2920  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2921      if (utf8)      if (utf8)
2922        {        {
# Line 1933  for (;;) Line 2924  for (;;)
2924        ecode++;        ecode++;
2925        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2926    
2927        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2928            {
2929            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2930            MRRETURN(MATCH_NOMATCH);
2931            }
2932    
2933        /* 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
2934        can use the fast lookup table. */        can use the fast lookup table. */
2935    
2936        if (fc < 128)        if (fc < 128)
2937          {          {
2938          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2939          }          }
2940    
2941        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
# Line 1957  for (;;) Line 2952  for (;;)
2952          if (fc != dc)          if (fc != dc)
2953            {            {
2954  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2955            if (dc != _pcre_ucp_othercase(fc))            if (dc != UCD_OTHERCASE(fc))
2956  #endif  #endif
2957              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
2958            }            }
2959          }          }
2960        }        }
# Line 1968  for (;;) Line 2963  for (;;)
2963    
2964      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2965        {        {
2966        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2967        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
2968            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2969            MRRETURN(MATCH_NOMATCH);
2970            }
2971          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2972        ecode += 2;        ecode += 2;
2973        }        }
2974      break;      break;
# Line 1977  for (;;) Line 2976  for (;;)
2976      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2977    
2978      case OP_EXACT:      case OP_EXACT:
2979        case OP_EXACTI:
2980      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2981      ecode += 3;      ecode += 3;
2982      goto REPEATCHAR;      goto REPEATCHAR;
2983    
2984      case OP_POSUPTO:      case OP_POSUPTO:
2985        case OP_POSUPTOI:
2986      possessive = TRUE;      possessive = TRUE;
2987      /* Fall through */      /* Fall through */
2988    
2989      case OP_UPTO:      case OP_UPTO:
2990        case OP_UPTOI:
2991      case OP_MINUPTO:      case OP_MINUPTO:
2992        case OP_MINUPTOI:
2993      min = 0;      min = 0;
2994      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2995      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2996      ecode += 3;      ecode += 3;
2997      goto REPEATCHAR;      goto REPEATCHAR;
2998    
2999      case OP_POSSTAR:      case OP_POSSTAR:
3000        case OP_POSSTARI:
3001      possessive = TRUE;      possessive = TRUE;
3002      min = 0;      min = 0;
3003      max = INT_MAX;      max = INT_MAX;
# Line 2001  for (;;) Line 3005  for (;;)
3005      goto REPEATCHAR;      goto REPEATCHAR;
3006    
3007      case OP_POSPLUS:      case OP_POSPLUS:
3008        case OP_POSPLUSI:
3009      possessive = TRUE;      possessive = TRUE;
3010      min = 1;      min = 1;
3011      max = INT_MAX;      max = INT_MAX;
# Line 2008  for (;;) Line 3013  for (;;)
3013      goto REPEATCHAR;      goto REPEATCHAR;
3014    
3015      case OP_POSQUERY:      case OP_POSQUERY:
3016        case OP_POSQUERYI:
3017      possessive = TRUE;      possessive = TRUE;
3018      min = 0;      min = 0;
3019      max = 1;      max = 1;
# Line 2015  for (;;) Line 3021  for (;;)
3021      goto REPEATCHAR;      goto REPEATCHAR;
3022    
3023      case OP_STAR:      case OP_STAR:
3024        case OP_STARI:
3025      case OP_MINSTAR:      case OP_MINSTAR:
3026        case OP_MINSTARI:
3027      case OP_PLUS:      case OP_PLUS:
3028        case OP_PLUSI:
3029      case OP_MINPLUS:      case OP_MINPLUS:
3030        case OP_MINPLUSI:
3031      case OP_QUERY:      case OP_QUERY:
3032        case OP_QUERYI:
3033      case OP_MINQUERY:      case OP_MINQUERY:
3034      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3035        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3036      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3037      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3038      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3039      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3040    
3041      /* 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. */  
3042    
3043      REPEATCHAR:      REPEATCHAR:
3044  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2037  for (;;) Line 3047  for (;;)
3047        length = 1;        length = 1;
3048        charptr = ecode;        charptr = ecode;
3049        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3050        ecode += length;        ecode += length;
3051    
3052        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 2045  for (;;) Line 3054  for (;;)
3054    
3055        if (length > 1)        if (length > 1)
3056          {          {
         int oclength = 0;  
         uschar occhars[8];  
   
3057  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3058          unsigned int othercase;          unsigned int othercase;
3059          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3060              (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)              (othercase = UCD_OTHERCASE(fc)) != fc)
3061            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3062            else oclength = 0;
3063  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3064    
3065          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3066            {            {
3067            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3068            /* Need braces because of following else */              memcmp(eptr, charptr, length) == 0) eptr += length;
3069            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
3070              else if (oclength > 0 &&
3071                       eptr <= md->end_subject - oclength &&
3072                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3073    #endif  /* SUPPORT_UCP */
3074            else            else
3075              {              {
3076              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3077              eptr += oclength;              MRRETURN(MATCH_NOMATCH);
3078              }              }
3079            }            }
3080    
# Line 2073  for (;;) Line 3084  for (;;)
3084            {            {
3085            for (fi = min;; fi++)            for (fi = min;; fi++)
3086              {              {
3087              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3088              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3089              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3090              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3091              /* Need braces because of following else */                memcmp(eptr, charptr, length) == 0) eptr += length;
3092              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
3093                else if (oclength > 0 &&
3094                         eptr <= md->end_subject - oclength &&
3095                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3096    #endif  /* SUPPORT_UCP */
3097              else              else
3098                {                {
3099                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3100                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3101                }                }
3102              }              }
3103            /* Control never gets here */            /* Control never gets here */
# Line 2093  for (;;) Line 3108  for (;;)
3108            pp = eptr;            pp = eptr;
3109            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3110              {              {
3111              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3112              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3113              else if (oclength == 0) break;  #ifdef SUPPORT_UCP
3114                else if (oclength > 0 &&
3115                         eptr <= md->end_subject - oclength &&
3116                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3117    #endif  /* SUPPORT_UCP */
3118              else              else
3119                {                {
3120                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3121                eptr += oclength;                break;
3122                }                }
3123              }              }
3124    
3125            if (possessive) continue;            if (possessive) continue;
3126            while (eptr >= pp)  
3127             {            for(;;)
3128             RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              {
3129             if (rrc != MATCH_NOMATCH) RRETURN(rrc);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3130             eptr -= length;              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3131             }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3132            RRETURN(MATCH_NOMATCH);  #ifdef SUPPORT_UCP
3133                eptr--;
3134                BACKCHAR(eptr);
3135    #else   /* without SUPPORT_UCP */
3136                eptr -= length;
3137    #endif  /* SUPPORT_UCP */
3138                }
3139            }            }
3140          /* Control never gets here */          /* Control never gets here */
3141          }          }
# Line 2123  for (;;) Line 3148  for (;;)
3148  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3149    
3150      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3151        {  
3152        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3153    
3154      /* 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
3155      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 3163  for (;;)
3163      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3164        max, eptr));        max, eptr));
3165    
3166      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3167        {        {
3168        fc = md->lcc[fc];        fc = md->lcc[fc];
3169        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3170          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3171            if (eptr >= md->end_subject)
3172              {
3173              SCHECK_PARTIAL();
3174              MRRETURN(MATCH_NOMATCH);
3175              }
3176            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3177            }
3178        if (min == max) continue;        if (min == max) continue;
3179        if (minimize)        if (minimize)
3180          {          {
3181          for (fi = min;; fi++)          for (fi = min;; fi++)
3182            {            {
3183            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3184            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3185            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3186                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3187              RRETURN(MATCH_NOMATCH);              {
3188                SCHECK_PARTIAL();
3189                MRRETURN(MATCH_NOMATCH);
3190                }
3191              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3192            }            }
3193          /* Control never gets here */          /* Control never gets here */
3194          }          }
# Line 2163  for (;;) Line 3197  for (;;)
3197          pp = eptr;          pp = eptr;
3198          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3199            {            {
3200            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3201                {
3202                SCHECK_PARTIAL();
3203                break;
3204                }
3205              if (fc != md->lcc[*eptr]) break;
3206            eptr++;            eptr++;
3207            }            }
3208    
3209          if (possessive) continue;          if (possessive) continue;
3210    
3211          while (eptr >= pp)          while (eptr >= pp)
3212            {            {
3213            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3214            eptr--;            eptr--;
3215            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3216            }            }
3217          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3218          }          }
3219        /* Control never gets here */        /* Control never gets here */
3220        }        }
# Line 2182  for (;;) Line 3223  for (;;)
3223    
3224      else      else
3225        {        {
3226        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3227            {
3228            if (eptr >= md->end_subject)
3229              {
3230              SCHECK_PARTIAL();
3231              MRRETURN(MATCH_NOMATCH);
3232              }
3233            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3234            }
3235    
3236        if (min == max) continue;        if (min == max) continue;
3237    
3238        if (minimize)        if (minimize)
3239          {          {
3240          for (fi = min;; fi++)          for (fi = min;; fi++)
3241            {            {
3242            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3243            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3244            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3245              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3246                {
3247                SCHECK_PARTIAL();
3248                MRRETURN(MATCH_NOMATCH);
3249                }
3250              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3251            }            }
3252          /* Control never gets here */          /* Control never gets here */
3253          }          }
# Line 2200  for (;;) Line 3256  for (;;)
3256          pp = eptr;          pp = eptr;
3257          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3258            {            {
3259            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3260                {
3261                SCHECK_PARTIAL();
3262                break;
3263                }
3264              if (fc != *eptr) break;
3265            eptr++;            eptr++;
3266            }            }
3267          if (possessive) continue;          if (possessive) continue;
3268    
3269          while (eptr >= pp)          while (eptr >= pp)
3270            {            {
3271            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3272            eptr--;            eptr--;
3273            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3274            }            }
3275          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3276          }          }
3277        }        }
3278      /* Control never gets here */      /* Control never gets here */
# Line 2219  for (;;) Line 3281  for (;;)
3281      checking can be multibyte. */      checking can be multibyte. */
3282    
3283      case OP_NOT:      case OP_NOT:
3284      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3285        if (eptr >= md->end_subject)
3286          {
3287          SCHECK_PARTIAL();
3288          MRRETURN(MATCH_NOMATCH);
3289          }
3290      ecode++;      ecode++;
3291      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3292      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3293        {        {
3294  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3295        if (c < 256)        if (c < 256)
3296  #endif  #endif
3297        c = md->lcc[c];        c = md->lcc[c];
3298        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3299        }        }
3300      else      else    /* Caseful */
3301        {        {
3302        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3303        }        }
3304      break;      break;
3305    
# Line 2244  for (;;) Line 3311  for (;;)
3311      about... */      about... */
3312    
3313      case OP_NOTEXACT:      case OP_NOTEXACT:
3314        case OP_NOTEXACTI:
3315      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3316      ecode += 3;      ecode += 3;
3317      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3318    
3319      case OP_NOTUPTO:      case OP_NOTUPTO:
3320        case OP_NOTUPTOI:
3321      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3322        case OP_NOTMINUPTOI:
3323      min = 0;      min = 0;
3324      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3325      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3326      ecode += 3;      ecode += 3;
3327      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3328    
3329      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3330        case OP_NOTPOSSTARI:
3331      possessive = TRUE;      possessive = TRUE;
3332      min = 0;      min = 0;
3333      max = INT_MAX;      max = INT_MAX;
# Line 2264  for (;;) Line 3335  for (;;)
3335      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3336    
3337      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3338        case OP_NOTPOSPLUSI:
3339      possessive = TRUE;      possessive = TRUE;
3340      min = 1;      min = 1;
3341      max = INT_MAX;      max = INT_MAX;
# Line 2271  for (;;) Line 3343  for (;;)
3343      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3344    
3345      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3346        case OP_NOTPOSQUERYI:
3347      possessive = TRUE;      possessive = TRUE;
3348      min = 0;      min = 0;
3349      max = 1;      max = 1;
# Line 2278  for (;;) Line 3351  for (;;)
3351      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3352    
3353      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3354        case OP_NOTPOSUPTOI:
3355      possessive = TRUE;      possessive = TRUE;
3356      min = 0;      min = 0;
3357      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 2285  for (;;) Line 3359  for (;;)
3359      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3360    
3361      case OP_NOTSTAR:      case OP_NOTSTAR:
3362        case OP_NOTSTARI:
3363      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3364        case OP_NOTMINSTARI:
3365      case OP_NOTPLUS:      case OP_NOTPLUS:
3366        case OP_NOTPLUSI:
3367      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3368        case OP_NOTMINPLUSI:
3369      case OP_NOTQUERY:      case OP_NOTQUERY:
3370        case OP_NOTQUERYI:
3371      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3372      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3373        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3374      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3375      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3376      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3377      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3378    
3379      /* 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. */  
3380    
3381      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3382      fc = *ecode++;      fc = *ecode++;
3383    
3384      /* 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 3392  for (;;)
3392      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,
3393        max, eptr));        max, eptr));
3394    
3395      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3396        {        {
3397        fc = md->lcc[fc];        fc = md->lcc[fc];
3398    
# Line 2326  for (;;) Line 3403  for (;;)
3403          register unsigned int d;          register unsigned int d;
3404          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3405            {            {
3406              if (eptr >= md->end_subject)
3407                {
3408                SCHECK_PARTIAL();
3409                MRRETURN(MATCH_NOMATCH);
3410                }
3411            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3412            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3413            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3414            }            }
3415          }          }
3416        else        else
# Line 2337  for (;;) Line 3419  for (;;)
3419        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3420          {          {
3421          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3422            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3423              if (eptr >= md->end_subject)
3424                {
3425                SCHECK_PARTIAL();
3426                MRRETURN(MATCH_NOMATCH);
3427                }
3428              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3429              }
3430          }          }
3431    
3432        if (min == max) continue;        if (min == max) continue;
# Line 2351  for (;;) Line 3440  for (;;)
3440            register unsigned int d;            register unsigned int d;
3441            for (fi = min;; fi++)            for (fi = min;; fi++)
3442              {              {
3443              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3444              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3445                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3446                if (eptr >= md->end_subject)
3447                  {
3448                  SCHECK_PARTIAL();
3449                  MRRETURN(MATCH_NOMATCH);
3450                  }
3451              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3452              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3453              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3454              }              }
3455            }            }
3456          else          else
# Line 2365  for (;;) Line 3459  for (;;)
3459            {            {
3460            for (fi = min;; fi++)            for (fi = min;; fi++)
3461              {              {
3462              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3463              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3464              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3465                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3466                  {
3467                  SCHECK_PARTIAL();
3468                  MRRETURN(MATCH_NOMATCH);
3469                  }
3470                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3471              }              }
3472            }            }
3473          /* Control never gets here */          /* Control never gets here */
# Line 2388  for (;;) Line 3487  for (;;)
3487            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3488              {              {
3489              int len = 1;              int len = 1;
3490              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3491                  {
3492                  SCHECK_PARTIAL();
3493                  break;
3494                  }
3495              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3496              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3497              if (fc == d) break;              if (fc == d) break;
# Line 2397  for (;;) Line 3500  for (;;)
3500          if (possessive) continue;          if (possessive) continue;
3501          for(;;)          for(;;)
3502              {              {
3503              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3504              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3505              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3506              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2409  for (;;) Line 3512  for (;;)
3512            {            {
3513            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3514              {              {
3515              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3516                  {
3517                  SCHECK_PARTIAL();
3518                  break;
3519                  }
3520                if (fc == md->lcc[*eptr]) break;
3521              eptr++;              eptr++;
3522              }              }
3523            if (possessive) continue;            if (possessive) continue;
3524            while (eptr >= pp)            while (eptr >= pp)
3525              {              {
3526              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3527              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3528              eptr--;              eptr--;
3529              }              }
3530            }            }
3531    
3532          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3533          }          }
3534        /* Control never gets here */        /* Control never gets here */
3535        }        }
# Line 2437  for (;;) Line 3545  for (;;)
3545          register unsigned int d;          register unsigned int d;
3546          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3547            {            {
3548              if (eptr >= md->end_subject)
3549                {
3550                SCHECK_PARTIAL();
3551                MRRETURN(MATCH_NOMATCH);
3552                }
3553            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3554            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3555            }            }
3556          }          }
3557        else        else
# Line 2446  for (;;) Line 3559  for (;;)
3559        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3560          {          {
3561          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3562            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3563              if (eptr >= md->end_subject)
3564                {
3565                SCHECK_PARTIAL();
3566                MRRETURN(MATCH_NOMATCH);
3567                }
3568              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3569              }
3570          }          }
3571    
3572        if (min == max) continue;        if (min == max) continue;
# Line 2460  for (;;) Line 3580  for (;;)
3580            register unsigned int d;            register unsigned int d;
3581            for (fi = min;; fi++)            for (fi = min;; fi++)
3582              {              {
3583              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3584              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3585                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3586                if (eptr >= md->end_subject)
3587                  {
3588                  SCHECK_PARTIAL();
3589                  MRRETURN(MATCH_NOMATCH);
3590                  }
3591              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3592              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3593              }              }
3594            }            }
3595          else          else
# Line 2473  for (;;) Line 3598  for (;;)
3598            {            {
3599            for (fi = min;; fi++)            for (fi = min;; fi++)
3600              {              {
3601              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3602              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3603              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3604                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3605                  {
3606                  SCHECK_PARTIAL();
3607                  MRRETURN(MATCH_NOMATCH);
3608                  }
3609                if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3610              }              }
3611            }            }
3612          /* Control never gets here */          /* Control never gets here */
# Line 2496  for (;;) Line 3626  for (;;)
3626            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3627              {              {
3628              int len = 1;              int len = 1;
3629              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3630                  {
3631                  SCHECK_PARTIAL();
3632                  break;
3633                  }
3634              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3635              if (fc == d) break;              if (fc == d) break;
3636              eptr += len;              eptr += len;
# Line 2504  for (;;) Line 3638  for (;;)
3638            if (possessive) continue;            if (possessive) continue;
3639            for(;;)            for(;;)
3640              {              {
3641              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3642              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3643              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3644              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2516  for (;;) Line 3650  for (;;)
3650            {            {
3651            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3652              {              {
3653              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3654                  {
3655                  SCHECK_PARTIAL();
3656                  break;
3657                  }
3658                if (fc == *eptr) break;
3659              eptr++;              eptr++;
3660              }              }
3661            if (possessive) continue;            if (possessive) continue;
3662            while (eptr >= pp)            while (eptr >= pp)
3663              {              {
3664              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3665              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3666              eptr--;              eptr--;
3667              }              }
3668            }            }
3669    
3670          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3671          }          }
3672        }        }
3673      /* Control never gets here */      /* Control never gets here */
# Line 2610  for (;;) Line 3749  for (;;)
3749    
3750      /* First, ensure the minimum number of matches are present. Use inline      /* First, ensure the minimum number of matches are present. Use inline
3751      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
3752      (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  
3753      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
3754      and single-bytes. */      and single-bytes. */
3755    
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3756      if (min > 0)      if (min > 0)
3757        {        {
3758  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2625  for (;;) Line 3761  for (;;)
3761          switch(prop_type)          switch(prop_type)
3762            {            {
3763            case PT_ANY:            case PT_ANY:
3764            if (prop_fail_result) RRETURN(MATCH_NOMATCH);            if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
3765            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3766              {              {
3767              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3768              GETCHARINC(c, eptr);                {
3769                  SCHECK_PARTIAL();
3770                  MRRETURN(MATCH_NOMATCH);
3771                  }
3772                GETCHARINCTEST(c, eptr);
3773              }              }
3774            break;            break;
3775    
3776            case PT_LAMP:            case PT_LAMP:
3777            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3778              {              {
3779              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              int chartype;
3780              GETCHARINC(c, eptr);              if (eptr >= md->end_subject)
3781              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                {
3782              if ((prop_chartype == ucp_Lu ||                SCHECK_PARTIAL();
3783                   prop_chartype == ucp_Ll ||                MRRETURN(MATCH_NOMATCH);
3784                   prop_chartype == ucp_Lt) == prop_fail_result)                }
3785                RRETURN(MATCH_NOMATCH);              GETCHARINCTEST(c, eptr);
3786                chartype = UCD_CHARTYPE(c);
3787                if ((chartype == ucp_Lu ||
3788                     chartype == ucp_Ll ||
3789                     chartype == ucp_Lt) == prop_fail_result)
3790                  MRRETURN(MATCH_NOMATCH);
3791              }              }
3792            break;            break;
3793    
3794            case PT_GC:            case PT_GC:
3795            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3796              {              {
3797              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3798              GETCHARINC(c, eptr);                {
3799              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3800              if ((prop_category == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
3801                RRETURN(MATCH_NOMATCH);                }
3802                GETCHARINCTEST(c, eptr);
3803                if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
3804                  MRRETURN(MATCH_NOMATCH);
3805              }              }
3806            break;            break;
3807    
3808            case PT_PC:            case PT_PC:
3809            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3810              {              {
3811              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3812              GETCHARINC(c, eptr);                {
3813              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3814              if ((prop_chartype == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
3815                RRETURN(MATCH_NOMATCH);                }
3816                GETCHARINCTEST(c, eptr);
3817                if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
3818                  MRRETURN(MATCH_NOMATCH);
3819              }              }
3820            break;            break;
3821    
3822            case PT_SC:            case PT_SC:
3823            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3824              {              {
3825              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3826              GETCHARINC(c, eptr);                {
3827              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3828              if ((prop_script == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
3829                RRETURN(MATCH_NOMATCH);                }
3830                GETCHARINCTEST(c, eptr);
3831                if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
3832                  MRRETURN(MATCH_NOMATCH);
3833                }
3834              break;
3835    
3836              case PT_ALNUM:
3837              for (i = 1; i <= min; i++)
3838                {
3839                int category;
3840                if (eptr >= md->end_subject)
3841                  {
3842                  SCHECK_PARTIAL();
3843                  MRRETURN(MATCH_NOMATCH);
3844                  }
3845                GETCHARINCTEST(c, eptr);
3846                category = UCD_CATEGORY(c);
3847                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
3848                  MRRETURN(MATCH_NOMATCH);
3849                }
3850              break;
3851    
3852              case PT_SPACE:    /* Perl space */
3853              for (i = 1; i <= min; i++)
3854                {
3855                if (eptr >= md->end_subject)
3856                  {
3857                  SCHECK_PARTIAL();
3858                  MRRETURN(MATCH_NOMATCH);
3859                  }
3860                GETCHARINCTEST(c, eptr);
3861                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3862                     c == CHAR_FF || c == CHAR_CR)
3863                       == prop_fail_result)
3864                  MRRETURN(MATCH_NOMATCH);
3865                }
3866              break;
3867    
3868              case PT_PXSPACE:  /* POSIX space */
3869              for (i = 1; i <= min; i++)
3870                {
3871                if (eptr >= md->end_subject)
3872                  {
3873                  SCHECK_PARTIAL();
3874                  MRRETURN(MATCH_NOMATCH);
3875                  }
3876                GETCHARINCTEST(c, eptr);
3877                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3878                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3879                       == prop_fail_result)
3880                  MRRETURN(MATCH_NOMATCH);
3881              }              }
3882            break;            break;
3883    
3884              case PT_WORD:
3885              for (i = 1; i <= min; i++)
3886                {
3887                int category;
3888                if (eptr >= md->end_subject)
3889                  {
3890                  SCHECK_PARTIAL();
3891                  MRRETURN(MATCH_NOMATCH);
3892                  }
3893                GETCHARINCTEST(c, eptr);
3894                category = UCD_CATEGORY(c);
3895                if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
3896                       == prop_fail_result)
3897                  MRRETURN(MATCH_NOMATCH);
3898                }
3899              break;
3900    
3901              /* This should not occur */
3902    
3903            default:            default:
3904            RRETURN(PCRE_ERROR_INTERNAL);