/[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 85 by nigel, Sat Feb 24 21:41:13 2007 UTC revision 606 by ph10, Mon Jun 6 17:46:22 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) 1997-2005 University of Cambridge             Copyright (c) 1997-2011 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  pattern matching using an NFA algorithm, trying to mimic Perl as closely as  pattern matching using an NFA algorithm, trying to mimic Perl as closely as
43  possible. There are also some static supporting functions. */  possible. There are also some static supporting functions. */
44    
45    #ifdef HAVE_CONFIG_H
46    #include "config.h"
47    #endif
48    
49    #define NLBLOCK md             /* Block containing newline information */
50    #define PSSTART start_subject  /* Field containing processed string start */
51    #define PSEND   end_subject    /* Field containing processed string end */
52    
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    /* Undefine some potentially clashing cpp symbols */
56    
57  /* Structure for building a chain of data that actually lives on the  #undef min
58  stack, for holding the values of the subject pointer at the start of each  #undef max
 subpattern, so as to detect when an empty string has been matched by a  
 subpattern - to break infinite loops. When NO_RECURSE is set, these blocks  
 are on the heap, not on the stack. */  
   
 typedef struct eptrblock {  
   struct eptrblock *epb_prev;  
   const uschar *epb_saved_eptr;  
 } eptrblock;  
59    
60  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
61    of call to match(). We do it this way to save on using another stack variable,
62    as stack usage is to be discouraged. */
63    
64  #define match_condassert   0x01    /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
65  #define match_isgroup      0x02    /* Set if start of bracketed group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
66    
67  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
68  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 68  defined PCRE_ERROR_xxx codes, which are Line 70  defined PCRE_ERROR_xxx codes, which are
70  #define MATCH_MATCH        1  #define MATCH_MATCH        1
71  #define MATCH_NOMATCH      0  #define MATCH_NOMATCH      0
72    
73    /* Special internal returns from the match() function. Make them sufficiently
74    negative to avoid the external error codes. */
75    
76    #define MATCH_ACCEPT       (-999)
77    #define MATCH_COMMIT       (-998)
78    #define MATCH_KETRPOS      (-997)
79    #define MATCH_PRUNE        (-996)
80    #define MATCH_SKIP         (-995)
81    #define MATCH_SKIP_ARG     (-994)
82    #define MATCH_THEN         (-993)
83    
84    /* This is a convenience macro for code that occurs many times. */
85    
86    #define MRRETURN(ra) \
87      { \
88      md->mark = markptr; \
89      RRETURN(ra); \
90      }
91    
92  /* Maximum number of ints of offset to save on the stack for recursive calls.  /* Maximum number of ints of offset to save on the stack for recursive calls.
93  If the offset vector is bigger, malloc is used. This should be a multiple of 3,  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
94  because the offset vector is always a multiple of 3 long. */  because the offset vector is always a multiple of 3 long. */
# Line 81  static const char rep_max[] = { 0, 0, 0, Line 102  static const char rep_max[] = { 0, 0, 0,
102    
103    
104    
105  #ifdef DEBUG  #ifdef PCRE_DEBUG
106  /*************************************************  /*************************************************
107  *        Debugging function to print chars       *  *        Debugging function to print chars       *
108  *************************************************/  *************************************************/
# Line 101  Returns:     nothing Line 122  Returns:     nothing
122  static void  static void
123  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)
124  {  {
125  int c;  unsigned int c;
126  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;
127  while (length-- > 0)  while (length-- > 0)
128    if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c);    if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c);
# Line 114  while (length-- > 0) Line 135  while (length-- > 0)
135  *          Match a back-reference                *  *          Match a back-reference                *
136  *************************************************/  *************************************************/
137    
138  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
139  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
140    the length passed is zero. Note that in caseless UTF-8 mode, the number of
141    subject bytes matched may be different to the number of reference bytes.
142    
143  Arguments:  Arguments:
144    offset      index into the offset vector    offset      index into the offset vector
145    eptr        points into the subject    eptr        pointer into the subject
146    length      length to be matched    length      length of reference to be matched (number of bytes)
147    md          points to match data block    md          points to match data block
148    ims         the ims flags    caseless    TRUE if caseless
149    
150  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
151  */  */
152    
153  static BOOL  static int
154  match_ref(int offset, register const uschar *eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
155    unsigned long int ims)    BOOL caseless)
156  {  {
157  const uschar *p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
158    register USPTR p = md->start_subject + md->offset_vector[offset];
159    
160  #ifdef DEBUG  #ifdef PCRE_DEBUG
161  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
162    printf("matching subject <null>");    printf("matching subject <null>");
163  else  else
# Line 146  pchars(p, length, FALSE, md); Line 170  pchars(p, length, FALSE, md);
170  printf("\n");  printf("\n");
171  #endif  #endif
172    
173  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
174    
175  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
176    
177  /* Separate the caselesss case for speed */  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
178    properly if Unicode properties are supported. Otherwise, we can check only
179    ASCII characters. */
180    
181  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
182    {    {
183    while (length-- > 0)  #ifdef SUPPORT_UTF8
184      if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE;  #ifdef SUPPORT_UCP
185      if (md->utf8)
186        {
187        /* Match characters up to the end of the reference. NOTE: the number of
188        bytes matched may differ, because there are some characters whose upper and
189        lower case versions code as different numbers of bytes. For example, U+023A
190        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
191        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
192        the latter. It is important, therefore, to check the length along the
193        reference, not along the subject (earlier code did this wrong). */
194    
195        USPTR endptr = p + length;
196        while (p < endptr)
197          {
198          int c, d;
199          if (eptr >= md->end_subject) return -1;
200          GETCHARINC(c, eptr);
201          GETCHARINC(d, p);
202          if (c != d && c != UCD_OTHERCASE(d)) return -1;
203          }
204        }
205      else
206    #endif
207    #endif
208    
209      /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
210      is no UCP support. */
211        {
212        if (eptr + length > md->end_subject) return -1;
213        while (length-- > 0)
214          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
215        }
216    }    }
217    
218    /* In the caseful case, we can just compare the bytes, whether or not we
219    are in UTF-8 mode. */
220    
221  else  else
222    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
223      if (eptr + length > md->end_subject) return -1;
224      while (length-- > 0) if (*p++ != *eptr++) return -1;
225      }
226    
227  return TRUE;  return eptr - eptr_start;
228  }  }
229    
230    
# Line 169  return TRUE; Line 233  return TRUE;
233  ****************************************************************************  ****************************************************************************
234                     RECURSION IN THE match() FUNCTION                     RECURSION IN THE match() FUNCTION
235    
236  The match() function is highly recursive. Some regular expressions can cause  The match() function is highly recursive, though not every recursive call
237  it to recurse thousands of times. I was writing for Unix, so I just let it  increases the recursive depth. Nevertheless, some regular expressions can cause
238  call itself recursively. This uses the stack for saving everything that has  it to recurse to a great depth. I was writing for Unix, so I just let it call
239  to be saved for a recursive call. On Unix, the stack can be large, and this  itself recursively. This uses the stack for saving everything that has to be
240  works fine.  saved for a recursive call. On Unix, the stack can be large, and this works
241    fine.
242  It turns out that on non-Unix systems there are problems with programs that  
243  use a lot of stack. (This despite the fact that every last chip has oodles  It turns out that on some non-Unix-like systems there are problems with
244  of memory these days, and techniques for extending the stack have been known  programs that use a lot of stack. (This despite the fact that every last chip
245  for decades.) So....  has oodles of memory these days, and techniques for extending the stack have
246    been known for decades.) So....
247    
248  There is a fudge, triggered by defining NO_RECURSE, which avoids recursive  There is a fudge, triggered by defining NO_RECURSE, which avoids recursive
249  calls by keeping local variables that need to be preserved in blocks of memory  calls by keeping local variables that need to be preserved in blocks of memory
250  obtained from malloc instead instead of on the stack. Macros are used to  obtained from malloc() instead instead of on the stack. Macros are used to
251  achieve this so that the actual code doesn't look very different to what it  achieve this so that the actual code doesn't look very different to what it
252  always used to.  always used to.
253    
254    The original heap-recursive code used longjmp(). However, it seems that this
255    can be very slow on some operating systems. Following a suggestion from Stan
256    Switzer, the use of longjmp() has been abolished, at the cost of having to
257    provide a unique number for each call to RMATCH. There is no way of generating
258    a sequence of numbers at compile time in C. I have given them names, to make
259    them stand out more clearly.
260    
261    Crude tests on x86 Linux show a small speedup of around 5-8%. However, on
262    FreeBSD, avoiding longjmp() more than halves the time taken to run the standard
263    tests. Furthermore, not using longjmp() means that local dynamic variables
264    don't have indeterminate values; this has meant that the frame size can be
265    reduced because the result can be "passed back" by straight setting of the
266    variable instead of being passed in the frame.
267  ****************************************************************************  ****************************************************************************
268  ***************************************************************************/  ***************************************************************************/
269    
270    /* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN
271    below must be updated in sync.  */
272    
273  /* These versions of the macros use the stack, as normal */  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,
274           RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,
275           RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
276           RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
277           RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
278           RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
279           RM61,  RM62, RM63, RM64 };
280    
281    /* These versions of the macros use the stack, as normal. There are debugging
282    versions and production versions. Note that the "rw" argument of RMATCH isn't
283    actually used in this definition. */
284    
285  #ifndef NO_RECURSE  #ifndef NO_RECURSE
286  #define REGISTER register  #define REGISTER register
287  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) rx = match(ra,rb,rc,rd,re,rf,rg)  
288    #ifdef PCRE_DEBUG
289    #define RMATCH(ra,rb,rc,rd,re,rw) \
290      { \
291      printf("match() called in line %d\n", __LINE__); \
292      rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
293      printf("to line %d\n", __LINE__); \
294      }
295    #define RRETURN(ra) \
296      { \
297      printf("match() returned %d from line %d ", ra, __LINE__); \
298      return ra; \
299      }
300    #else
301    #define RMATCH(ra,rb,rc,rd,re,rw) \
302      rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
303  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
304    #endif
305    
306  #else  #else
307    
308    
309  /* These versions of the macros manage a private stack on the heap. Note  /* These versions of the macros manage a private stack on the heap. Note that
310  that the rd argument of RMATCH isn't actually used. It's the md argument of  the "rd" argument of RMATCH isn't actually used in this definition. It's the md
311  match(), which never changes. */  argument of match(), which never changes. */
312    
313  #define REGISTER  #define REGISTER
314    
315  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
316    {\    {\
317    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
318    if (setjmp(frame->Xwhere) == 0)\    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
319      {\    frame->Xwhere = rw; \
320      newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
321      newframe->Xecode = rb;\    newframe->Xecode = rb;\
322      newframe->Xoffset_top = rc;\    newframe->Xmstart = mstart;\
323      newframe->Xims = re;\    newframe->Xmarkptr = markptr;\
324      newframe->Xeptrb = rf;\    newframe->Xoffset_top = rc;\
325      newframe->Xflags = rg;\    newframe->Xeptrb = re;\
326      newframe->Xprevframe = frame;\    newframe->Xrdepth = frame->Xrdepth + 1;\
327      frame = newframe;\    newframe->Xprevframe = frame;\
328      DPRINTF(("restarting from line %d\n", __LINE__));\    frame = newframe;\
329      goto HEAP_RECURSE;\    DPRINTF(("restarting from line %d\n", __LINE__));\
330      }\    goto HEAP_RECURSE;\
331    else\    L_##rw:\
332      {\    DPRINTF(("jumped back to line %d\n", __LINE__));\
     DPRINTF(("longjumped back to line %d\n", __LINE__));\  
     frame = md->thisframe;\  
     rx = frame->Xresult;\  
     }\  
333    }    }
334    
335  #define RRETURN(ra)\  #define RRETURN(ra)\
336    {\    {\
337    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
338    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
339    (pcre_stack_free)(newframe);\    (pcre_stack_free)(oldframe);\
340    if (frame != NULL)\    if (frame != NULL)\
341      {\      {\
342      frame->Xresult = ra;\      rrc = ra;\
343      md->thisframe = frame;\      goto HEAP_RETURN;\
     longjmp(frame->Xwhere, 1);\  
344      }\      }\
345    return ra;\    return ra;\
346    }    }
# Line 250  typedef struct heapframe { Line 353  typedef struct heapframe {
353    
354    /* Function arguments that may change */    /* Function arguments that may change */
355    
356    const uschar *Xeptr;    USPTR Xeptr;
357    const uschar *Xecode;    const uschar *Xecode;
358      USPTR Xmstart;
359      USPTR Xmarkptr;
360    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
361    eptrblock *Xeptrb;    eptrblock *Xeptrb;
362    int Xflags;    unsigned int Xrdepth;
363    
364    /* Function local variables */    /* Function local variables */
365    
366    const uschar *Xcallpat;    USPTR Xcallpat;
367    const uschar *Xcharptr;  #ifdef SUPPORT_UTF8
368    const uschar *Xdata;    USPTR Xcharptr;
369    const uschar *Xnext;  #endif
370    const uschar *Xpp;    USPTR Xdata;
371    const uschar *Xprev;    USPTR Xnext;
372    const uschar *Xsaved_eptr;    USPTR Xpp;
373      USPTR Xprev;
374      USPTR Xsaved_eptr;
375    
376    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
377    
378    BOOL Xcur_is_word;    BOOL Xcur_is_word;
379    BOOL Xcondition;    BOOL Xcondition;
   BOOL Xminimize;  
380    BOOL Xprev_is_word;    BOOL Xprev_is_word;
381    
   unsigned long int Xoriginal_ims;  
   
382  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
383    int Xprop_type;    int Xprop_type;
384      int Xprop_value;
385    int Xprop_fail_result;    int Xprop_fail_result;
386    int Xprop_category;    int Xprop_category;
387    int Xprop_chartype;    int Xprop_chartype;
388    int Xprop_othercase;    int Xprop_script;
389    int Xprop_test_against;    int Xoclength;
390    int *Xprop_test_variable;    uschar Xocchars[8];
391  #endif  #endif
392    
393      int Xcodelink;
394    int Xctype;    int Xctype;
395    int Xfc;    unsigned int Xfc;
396    int Xfi;    int Xfi;
397    int Xlength;    int Xlength;
398    int Xmax;    int Xmax;
# Line 301  typedef struct heapframe { Line 406  typedef struct heapframe {
406    
407    eptrblock Xnewptrb;    eptrblock Xnewptrb;
408    
409    /* Place to pass back result, and where to jump back to */    /* Where to jump back to */
410    
411    int  Xresult;    int Xwhere;
   jmp_buf Xwhere;  
412    
413  } heapframe;  } heapframe;
414    
# Line 320  typedef struct heapframe { Line 424  typedef struct heapframe {
424  *         Match from current position            *  *         Match from current position            *
425  *************************************************/  *************************************************/
426    
427  /* On entry ecode points to the first opcode, and eptr to the first character  /* This function is called recursively in many circumstances. Whenever it
 in the subject string, while eptrb holds the value of eptr at the start of the  
 last bracketed group - used for breaking infinite loops matching zero-length  
 strings. This function is called recursively in many circumstances. Whenever it  
428  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
429  same response.  same response. */
430    
431    /* These macros pack up tests that are used for partial matching, and which
432    appears several times in the code. We set the "hit end" flag if the pointer is
433    at the end of the subject and also past the start of the subject (i.e.
434    something has been matched). For hard partial matching, we then return
435    immediately. The second one is used when we already know we are past the end of
436    the subject. */
437    
438    #define CHECK_PARTIAL()\
439      if (md->partial != 0 && eptr >= md->end_subject && \
440          eptr > md->start_used_ptr) \
441        { \
442        md->hitend = TRUE; \
443        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
444        }
445    
446    #define SCHECK_PARTIAL()\
447      if (md->partial != 0 && eptr > md->start_used_ptr) \
448        { \
449        md->hitend = TRUE; \
450        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
451        }
452    
453    
454  Performance note: It might be tempting to extract commonly used fields from the  /* Performance note: It might be tempting to extract commonly used fields from
455  md structure (e.g. utf8, end_subject) into individual variables to improve  the md structure (e.g. utf8, end_subject) into individual variables to improve
456  performance. Tests using gcc on a SPARC disproved this; in the first case, it  performance. Tests using gcc on a SPARC disproved this; in the first case, it
457  made performance worse.  made performance worse.
458    
459  Arguments:  Arguments:
460     eptr        pointer in subject     eptr        pointer to current character in subject
461     ecode       position in code     ecode       pointer to current position in compiled code
462       mstart      pointer to the current match start position (can be modified
463                     by encountering \K)
464       markptr     pointer to the most recent MARK name, or NULL
465     offset_top  current top pointer     offset_top  current top pointer
466     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
467     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
468                   brackets - for testing for empty matches                   brackets - for testing for empty matches
469     flags       can contain     rdepth      the recursion depth
                  match_condassert - this is an assertion condition  
                  match_isgroup - this is the start of a bracketed group  
470    
471  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
472                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
473                   a negative MATCH_xxx value for PRUNE, SKIP, etc
474                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
475                   (e.g. stopped by recursion limit)                   (e.g. stopped by repeated call or recursion limit)
476  */  */
477    
478  static int  static int
479  match(REGISTER const uschar *eptr, REGISTER const uschar *ecode,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
480    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
481    int flags)    unsigned int rdepth)
482  {  {
483  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
484  so they can be ordinary variables in all cases. Mark them with "register"  so they can be ordinary variables in all cases. Mark some of them with
485  because they are used a lot in loops. */  "register" because they are used a lot in loops. */
486    
487  register int  rrc;    /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
488  register int  i;      /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
489  register int  c;      /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
490  register BOOL utf8;   /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
491    
492    BOOL minimize, possessive; /* Quantifier options */
493    BOOL caseless;
494    int condcode;
495    
496  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
497  preserved over calls to RMATCH() are part of a "frame" which is obtained from  preserved over calls to RMATCH() are part of a "frame" which is obtained from
# Line 370  heap storage. Set up the top-level frame Line 499  heap storage. Set up the top-level frame
499  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
500    
501  #ifdef NO_RECURSE  #ifdef NO_RECURSE
502  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
503    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
504  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
505    
506  /* Copy in the original argument variables */  /* Copy in the original argument variables */
507    
508  frame->Xeptr = eptr;  frame->Xeptr = eptr;
509  frame->Xecode = ecode;  frame->Xecode = ecode;
510    frame->Xmstart = mstart;
511    frame->Xmarkptr = markptr;
512  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
513  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
514  frame->Xflags = flags;  frame->Xrdepth = rdepth;
515    
516  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
517    
# Line 390  HEAP_RECURSE: Line 521  HEAP_RECURSE:
521    
522  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
523  #define ecode              frame->Xecode  #define ecode              frame->Xecode
524    #define mstart             frame->Xmstart
525    #define markptr            frame->Xmarkptr
526  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
527  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
528  #define flags              frame->Xflags  #define rdepth             frame->Xrdepth
529    
530  /* Ditto for the local variables */  /* Ditto for the local variables */
531    
# Line 401  HEAP_RECURSE: Line 533  HEAP_RECURSE:
533  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
534  #endif  #endif
535  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
536    #define codelink           frame->Xcodelink
537  #define data               frame->Xdata  #define data               frame->Xdata
538  #define next               frame->Xnext  #define next               frame->Xnext
539  #define pp                 frame->Xpp  #define pp                 frame->Xpp
# Line 411  HEAP_RECURSE: Line 544  HEAP_RECURSE:
544    
545  #define cur_is_word        frame->Xcur_is_word  #define cur_is_word        frame->Xcur_is_word
546  #define condition          frame->Xcondition  #define condition          frame->Xcondition
 #define minimize           frame->Xminimize  
547  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
548    
 #define original_ims       frame->Xoriginal_ims  
   
549  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
550  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
551    #define prop_value         frame->Xprop_value
552  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
553  #define prop_category      frame->Xprop_category  #define prop_category      frame->Xprop_category
554  #define prop_chartype      frame->Xprop_chartype  #define prop_chartype      frame->Xprop_chartype
555  #define prop_othercase     frame->Xprop_othercase  #define prop_script        frame->Xprop_script
556  #define prop_test_against  frame->Xprop_test_against  #define oclength           frame->Xoclength
557  #define prop_test_variable frame->Xprop_test_variable  #define occhars            frame->Xocchars
558  #endif  #endif
559    
560  #define ctype              frame->Xctype  #define ctype              frame->Xctype
# Line 447  HEAP_RECURSE: Line 578  HEAP_RECURSE:
578  get preserved during recursion in the normal way. In this environment, fi and  get preserved during recursion in the normal way. In this environment, fi and
579  i, and fc and c, can be the same variables. */  i, and fc and c, can be the same variables. */
580    
581  #else  #else         /* NO_RECURSE not defined */
582  #define fi i  #define fi i
583  #define fc c  #define fc c
584    
585    /* Many of the following variables are used only in small blocks of the code.
586  #ifdef SUPPORT_UTF8                /* Many of these variables are used ony */  My normal style of coding would have declared them within each of those blocks.
587  const uschar *charptr;             /* small blocks of the code. My normal  */  However, in order to accommodate the version of this code that uses an external
588  #endif                             /* style of coding would have declared  */  "stack" implemented on the heap, it is easier to declare them all here, so the
589  const uschar *callpat;             /* them within each of those blocks.    */  declarations can be cut out in a block. The only declarations within blocks
590  const uschar *data;                /* However, in order to accommodate the */  below are for variables that do not have to be preserved over a recursive call
591  const uschar *next;                /* version of this code that uses an    */  to RMATCH(). */
592  const uschar *pp;                  /* external "stack" implemented on the  */  
593  const uschar *prev;                /* heap, it is easier to declare them   */  #ifdef SUPPORT_UTF8
594  const uschar *saved_eptr;          /* all here, so the declarations can    */  const uschar *charptr;
595                                     /* be cut out in a block. The only      */  #endif
596  recursion_info new_recursive;      /* declarations within blocks below are */  const uschar *callpat;
597                                     /* for variables that do not have to    */  const uschar *data;
598  BOOL cur_is_word;                  /* be preserved over a recursive call   */  const uschar *next;
599  BOOL condition;                    /* to RMATCH().                         */  USPTR         pp;
600  BOOL minimize;  const uschar *prev;
601    USPTR         saved_eptr;
602    
603    recursion_info new_recursive;
604    
605    BOOL cur_is_word;
606    BOOL condition;
607  BOOL prev_is_word;  BOOL prev_is_word;
608    
 unsigned long int original_ims;  
   
609  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
610  int prop_type;  int prop_type;
611    int prop_value;
612  int prop_fail_result;  int prop_fail_result;
613  int prop_category;  int prop_category;
614  int prop_chartype;  int prop_chartype;
615  int prop_othercase;  int prop_script;
616  int prop_test_against;  int oclength;
617  int *prop_test_variable;  uschar occhars[8];
618  #endif  #endif
619    
620    int codelink;
621  int ctype;  int ctype;
622  int length;  int length;
623  int max;  int max;
# Line 493  int save_offset1, save_offset2, save_off Line 630  int save_offset1, save_offset2, save_off
630  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
631    
632  eptrblock newptrb;  eptrblock newptrb;
633  #endif  #endif     /* NO_RECURSE */
634    
635    /* To save space on the stack and in the heap frame, I have doubled up on some
636    of the local variables that are used only in localised parts of the code, but
637    still need to be preserved over recursive calls of match(). These macros define
638    the alternative names that are used. */
639    
640    #define allow_zero    cur_is_word
641    #define cbegroup      condition
642    #define code_offset   codelink
643    #define condassert    condition
644    #define matched_once  prev_is_word
645    
646  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
647  variables. */  variables. */
648    
649  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
650    prop_value = 0;
651  prop_fail_result = 0;  prop_fail_result = 0;
 prop_test_against = 0;  
 prop_test_variable = NULL;  
652  #endif  #endif
653    
 /* OK, now we can get on with the real code of the function. Recursion is  
 specified by the macros RMATCH and RRETURN. When NO_RECURSE is *not* defined,  
 these just turn into a recursive call to match() and a "return", respectively.  
 However, RMATCH isn't like a function call because it's quite a complicated  
 macro. It has to be used in one particular way. This shouldn't, however, impact  
 performance when true recursion is being used. */  
654    
655  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  /* This label is used for tail recursion, which is used in a few cases even
656    when NO_RECURSE is not defined, in order to reduce the amount of stack that is
657    used. Thanks to Ian Taylor for noticing this possibility and sending the
658    original patch. */
659    
660    TAIL_RECURSE:
661    
662    /* OK, now we can get on with the real code of the function. Recursive calls
663    are specified by the macro RMATCH and RRETURN is used to return. When
664    NO_RECURSE is *not* defined, these just turn into a recursive call to match()
665    and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
666    defined). However, RMATCH isn't like a function call because it's quite a
667    complicated macro. It has to be used in one particular way. This shouldn't,
668    however, impact performance when true recursion is being used. */
669    
670  original_ims = ims;    /* Save for resetting on ')' */  #ifdef SUPPORT_UTF8
671  utf8 = md->utf8;       /* Local copy of the flag */  utf8 = md->utf8;       /* Local copy of the flag */
672    #else
673    utf8 = FALSE;
674    #endif
675    
676    /* First check that we haven't called match() too many times, or that we
677    haven't exceeded the recursive call limit. */
678    
679    if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
680    if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
681    
682  /* At the start of a bracketed group, add the current subject pointer to the  /* At the start of a group with an unlimited repeat that may match an empty
683  stack of such pointers, to be re-instated at the end of the group when we hit  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
684  the closing ket. When match() is called in other circumstances, we don't add to  done this way to save having to use another function argument, which would take
685  this stack. */  up space on the stack. See also MATCH_CONDASSERT below.
686    
687    When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
688    such remembered pointers, to be checked when we hit the closing ket, in order
689    to break infinite loops that match no characters. When match() is called in
690    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
691    NOT be used with tail recursion, because the memory block that is used is on
692    the stack, so a new one may be required for each match(). */
693    
694  if ((flags & match_isgroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
695    {    {
   newptrb.epb_prev = eptrb;  
696    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
697      newptrb.epb_prev = eptrb;
698    eptrb = &newptrb;    eptrb = &newptrb;
699      md->match_function_type = 0;
700    }    }
701    
702  /* Now start processing the operations. */  /* Now start processing the opcodes. */
703    
704  for (;;)  for (;;)
705    {    {
706      minimize = possessive = FALSE;
707    op = *ecode;    op = *ecode;
708    minimize = FALSE;  
709      switch(op)
710    /* For partial matching, remember if we ever hit the end of the subject after      {
711    matching at least one subject character. */      case OP_MARK:
712        markptr = ecode + 2;
713        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
714          eptrb, RM55);
715    
716        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
717        argument, and we must check whether that argument matches this MARK's
718        argument. It is passed back in md->start_match_ptr (an overloading of that
719        variable). If it does match, we reset that variable to the current subject
720        position and return MATCH_SKIP. Otherwise, pass back the return code
721        unaltered. */
722    
723        if (rrc == MATCH_SKIP_ARG &&
724            strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
725          {
726          md->start_match_ptr = eptr;
727          RRETURN(MATCH_SKIP);
728          }
729    
730        if (md->mark == NULL) md->mark = markptr;
731        RRETURN(rrc);
732    
733        case OP_FAIL:
734        MRRETURN(MATCH_NOMATCH);
735    
736        /* COMMIT overrides PRUNE, SKIP, and THEN */
737    
738        case OP_COMMIT:
739        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
740          eptrb, RM52);
741        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
742            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
743            rrc != MATCH_THEN)
744          RRETURN(rrc);
745        MRRETURN(MATCH_COMMIT);
746    
747    if (md->partial &&      /* PRUNE overrides THEN */
       eptr >= md->end_subject &&  
       eptr > md->start_match)  
     md->hitend = TRUE;  
   
   /* Opening capturing bracket. If there is space in the offset vector, save  
   the current subject position in the working slot at the top of the vector. We  
   mustn't change the current values of the data slot, because they may be set  
   from a previous iteration of this group, and be referred to by a reference  
   inside the group.  
   
   If the bracket fails to match, we need to restore this value and also the  
   values of the final offsets, in case they were set by a previous iteration of  
   the same bracket.  
   
   If there isn't enough space in the offset vector, treat this as if it were a  
   non-capturing bracket. Don't worry about setting the flag for the error case  
   here; that is handled in the code for KET. */  
748    
749    if (op > OP_BRA)      case OP_PRUNE:
750      {      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
751      number = op - OP_BRA;        eptrb, RM51);
752        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
753        MRRETURN(MATCH_PRUNE);
754    
755        case OP_PRUNE_ARG:
756        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
757          eptrb, RM56);
758        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
759        md->mark = ecode + 2;
760        RRETURN(MATCH_PRUNE);
761    
762        /* SKIP overrides PRUNE and THEN */
763    
764        case OP_SKIP:
765        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
766          eptrb, RM53);
767        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769        md->start_match_ptr = eptr;   /* Pass back current position */
770        MRRETURN(MATCH_SKIP);
771    
772      /* For extended extraction brackets (large number), we have to fish out the      case OP_SKIP_ARG:
773      number from a dummy opcode at the start. */      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
774          eptrb, RM57);
775        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
776          RRETURN(rrc);
777    
778      if (number > EXTRACT_BASIC_MAX)      /* Pass back the current skip name by overloading md->start_match_ptr and
779        number = GET2(ecode, 2+LINK_SIZE);      returning the special MATCH_SKIP_ARG return code. This will either be
780        caught by a matching MARK, or get to the top, where it is treated the same
781        as PRUNE. */
782    
783        md->start_match_ptr = ecode + 2;
784        RRETURN(MATCH_SKIP_ARG);
785    
786        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
787        the alt that is at the start of the current branch. This makes it possible
788        to skip back past alternatives that precede the THEN within the current
789        branch. */
790    
791        case OP_THEN:
792        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
793          eptrb, RM54);
794        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
795        md->start_match_ptr = ecode - GET(ecode, 1);
796        MRRETURN(MATCH_THEN);
797    
798        case OP_THEN_ARG:
799        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
800          offset_top, md, eptrb, RM58);
801        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
802        md->start_match_ptr = ecode - GET(ecode, 1);
803        md->mark = ecode + LINK_SIZE + 2;
804        RRETURN(MATCH_THEN);
805    
806        /* Handle a capturing bracket, other than those that are possessive with an
807        unlimited repeat. If there is space in the offset vector, save the current
808        subject position in the working slot at the top of the vector. We mustn't
809        change the current values of the data slot, because they may be set from a
810        previous iteration of this group, and be referred to by a reference inside
811        the group. If we fail to match, we need to restore this value and also the
812        values of the final offsets, in case they were set by a previous iteration
813        of the same bracket.
814    
815        If there isn't enough space in the offset vector, treat this as if it were
816        a non-capturing bracket. Don't worry about setting the flag for the error
817        case here; that is handled in the code for KET. */
818    
819        case OP_CBRA:
820        case OP_SCBRA:
821        number = GET2(ecode, 1+LINK_SIZE);
822      offset = number << 1;      offset = number << 1;
823    
824  #ifdef DEBUG  #ifdef PCRE_DEBUG
825      printf("start bracket %d subject=", number);      printf("start bracket %d\n", number);
826        printf("subject=");
827      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
828      printf("\n");      printf("\n");
829  #endif  #endif
# Line 582  for (;;) Line 836  for (;;)
836        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
837    
838        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
839        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
840            (int)(eptr - md->start_subject);
841    
842        do        for (;;)
843          {          {
844          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
845            match_isgroup);          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
846          if (rrc != MATCH_NOMATCH) RRETURN(rrc);            eptrb, RM1);
847            if (rrc != MATCH_NOMATCH &&
848                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
849              RRETURN(rrc);
850          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
851          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
852            if (*ecode != OP_ALT) break;
853          }          }
       while (*ecode == OP_ALT);  
854    
855        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
856    
# Line 600  for (;;) Line 858  for (;;)
858        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
859        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
860    
861          if (rrc != MATCH_THEN) md->mark = markptr;
862        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
863        }        }
864    
865      /* Insufficient room for saving captured contents */      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
866        as a non-capturing bracket. */
867    
868      else op = OP_BRA;      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
869      }      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870    
871    /* Other types of node can be handled by a switch */      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
872    
873    switch(op)      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
874      {      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
875      case OP_BRA:     /* Non-capturing bracket: optimized */  
876      DPRINTF(("start bracket 0\n"));      /* Non-capturing bracket, except for possessive with unlimited repeat. Loop
877      do      for all the alternatives. When we get to the final alternative within the
878        brackets, we would return the result of a recursive call to match()
879        whatever happened. We can reduce stack usage by turning this into a tail
880        recursion, except in the case of a possibly empty group.*/
881    
882        case OP_BRA:
883        case OP_SBRA:
884        DPRINTF(("start non-capturing bracket\n"));
885        for (;;)
886          {
887          if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */
888            {
889            if (op >= OP_SBRA)   /* Possibly empty group */
890              {
891              md->match_function_type = MATCH_CBEGROUP;
892              RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
893                RM48);
894              if (rrc == MATCH_NOMATCH) md->mark = markptr;
895              RRETURN(rrc);
896              }
897            /* Not a possibly empty group; use tail recursion */
898            ecode += _pcre_OP_lengths[*ecode];
899            DPRINTF(("bracket 0 tail recursion\n"));
900            goto TAIL_RECURSE;
901            }
902    
903          /* For non-final alternatives, continue the loop for a NOMATCH result;
904          otherwise return. */
905    
906          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
907          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
908            RM2);
909          if (rrc != MATCH_NOMATCH &&
910              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
911            RRETURN(rrc);
912          ecode += GET(ecode, 1);
913          }
914        /* Control never reaches here. */
915    
916        /* Handle possessive capturing brackets with an unlimited repeat. We come
917        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
918        handled similarly to the normal case above. However, the matching is
919        different. The end of these brackets will always be OP_KETRPOS, which
920        returns MATCH_KETRPOS without going further in the pattern. By this means
921        we can handle the group by iteration rather than recursion, thereby
922        reducing the amount of stack needed. */
923    
924        case OP_CBRAPOS:
925        case OP_SCBRAPOS:
926        allow_zero = FALSE;
927    
928        POSSESSIVE_CAPTURE:
929        number = GET2(ecode, 1+LINK_SIZE);
930        offset = number << 1;
931    
932    #ifdef PCRE_DEBUG
933        printf("start possessive bracket %d\n", number);
934        printf("subject=");
935        pchars(eptr, 16, TRUE, md);
936        printf("\n");
937    #endif
938    
939        if (offset < md->offset_max)
940        {        {
941        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,        matched_once = FALSE;
942          match_isgroup);        code_offset = ecode - md->start_code;
943        if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
944          save_offset1 = md->offset_vector[offset];
945          save_offset2 = md->offset_vector[offset+1];
946          save_offset3 = md->offset_vector[md->offset_end - number];
947          save_capture_last = md->capture_last;
948    
949          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
950    
951          /* Each time round the loop, save the current subject position for use
952          when the group matches. For MATCH_MATCH, the group has matched, so we
953          restart it with a new subject starting position, remembering that we had
954          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
955          usual. If we haven't matched any alternatives in any iteration, check to
956          see if a previous iteration matched. If so, the group has matched;
957          continue from afterwards. Otherwise it has failed; restore the previous
958          capture values before returning NOMATCH. */
959    
960          for (;;)
961            {
962            md->offset_vector[md->offset_end - number] =
963              (int)(eptr - md->start_subject);
964            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
965            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
966              eptrb, RM63);
967            if (rrc == MATCH_KETRPOS)
968              {
969              offset_top = md->end_offset_top;
970              eptr = md->end_match_ptr;
971              ecode = md->start_code + code_offset;
972              save_capture_last = md->capture_last;
973              matched_once = TRUE;
974              continue;
975              }
976            if (rrc != MATCH_NOMATCH &&
977                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
978              RRETURN(rrc);
979            md->capture_last = save_capture_last;
980            ecode += GET(ecode, 1);
981            if (*ecode != OP_ALT) break;
982            }
983    
984          if (!matched_once)
985            {
986            md->offset_vector[offset] = save_offset1;
987            md->offset_vector[offset+1] = save_offset2;
988            md->offset_vector[md->offset_end - number] = save_offset3;
989            }
990    
991          if (rrc != MATCH_THEN) md->mark = markptr;
992          if (allow_zero || matched_once)
993            {
994            ecode += 1 + LINK_SIZE;
995            break;
996            }
997    
998          RRETURN(MATCH_NOMATCH);
999          }
1000    
1001        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1002        as a non-capturing bracket. */
1003    
1004        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1005        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1006    
1007        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1008    
1009        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1010        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1011    
1012        /* Non-capturing possessive bracket with unlimited repeat. We come here
1013        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1014        without the capturing complication. It is written out separately for speed
1015        and cleanliness. */
1016    
1017        case OP_BRAPOS:
1018        case OP_SBRAPOS:
1019        allow_zero = FALSE;
1020    
1021        POSSESSIVE_NON_CAPTURE:
1022        matched_once = FALSE;
1023        code_offset = ecode - md->start_code;
1024    
1025        for (;;)
1026          {
1027          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1028          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1029            eptrb, RM64);
1030          if (rrc == MATCH_KETRPOS)
1031            {
1032            eptr = md->end_match_ptr;
1033            ecode = md->start_code + code_offset;
1034            matched_once = TRUE;
1035            continue;
1036            }
1037          if (rrc != MATCH_NOMATCH &&
1038              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1039            RRETURN(rrc);
1040        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1041          if (*ecode != OP_ALT) break;
1042        }        }
1043      while (*ecode == OP_ALT);  
1044      DPRINTF(("bracket 0 failed\n"));      if (matched_once || allow_zero)
1045          {
1046          ecode += 1 + LINK_SIZE;
1047          break;
1048          }
1049      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1050    
1051        /* Control never reaches here. */
1052    
1053      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1054      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1055      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1056      exactly what going to the ket would do. */      exactly what going to the ket would do. As there is only one branch to be
1057        obeyed, we can use tail recursion to avoid using another stack frame. */
1058    
1059      case OP_COND:      case OP_COND:
1060      if (ecode[LINK_SIZE+1] == OP_CREF) /* Condition extract or recurse test */      case OP_SCOND:
1061        codelink = GET(ecode, 1);
1062    
1063        /* Because of the way auto-callout works during compile, a callout item is
1064        inserted between OP_COND and an assertion condition. */
1065    
1066        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1067          {
1068          if (pcre_callout != NULL)
1069            {
1070            pcre_callout_block cb;
1071            cb.version          = 1;   /* Version 1 of the callout block */
1072            cb.callout_number   = ecode[LINK_SIZE+2];
1073            cb.offset_vector    = md->offset_vector;
1074            cb.subject          = (PCRE_SPTR)md->start_subject;
1075            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1076            cb.start_match      = (int)(mstart - md->start_subject);
1077            cb.current_position = (int)(eptr - md->start_subject);
1078            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1079            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1080            cb.capture_top      = offset_top/2;
1081            cb.capture_last     = md->capture_last;
1082            cb.callout_data     = md->callout_data;
1083            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1084            if (rrc < 0) RRETURN(rrc);
1085            }
1086          ecode += _pcre_OP_lengths[OP_CALLOUT];
1087          }
1088    
1089        condcode = ecode[LINK_SIZE+1];
1090    
1091        /* Now see what the actual condition is */
1092    
1093        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1094          {
1095          if (md->recursive == NULL)                /* Not recursing => FALSE */
1096            {
1097            condition = FALSE;
1098            ecode += GET(ecode, 1);
1099            }
1100          else
1101            {
1102            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1103            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1104    
1105            /* If the test is for recursion into a specific subpattern, and it is
1106            false, but the test was set up by name, scan the table to see if the
1107            name refers to any other numbers, and test them. The condition is true
1108            if any one is set. */
1109    
1110            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1111              {
1112              uschar *slotA = md->name_table;
1113              for (i = 0; i < md->name_count; i++)
1114                {
1115                if (GET2(slotA, 0) == recno) break;
1116                slotA += md->name_entry_size;
1117                }
1118    
1119              /* Found a name for the number - there can be only one; duplicate
1120              names for different numbers are allowed, but not vice versa. First
1121              scan down for duplicates. */
1122    
1123              if (i < md->name_count)
1124                {
1125                uschar *slotB = slotA;
1126                while (slotB > md->name_table)
1127                  {
1128                  slotB -= md->name_entry_size;
1129                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1130                    {
1131                    condition = GET2(slotB, 0) == md->recursive->group_num;
1132                    if (condition) break;
1133                    }
1134                  else break;
1135                  }
1136    
1137                /* Scan up for duplicates */
1138    
1139                if (!condition)
1140                  {
1141                  slotB = slotA;
1142                  for (i++; i < md->name_count; i++)
1143                    {
1144                    slotB += md->name_entry_size;
1145                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1146                      {
1147                      condition = GET2(slotB, 0) == md->recursive->group_num;
1148                      if (condition) break;
1149                      }
1150                    else break;
1151                    }
1152                  }
1153                }
1154              }
1155    
1156            /* Chose branch according to the condition */
1157    
1158            ecode += condition? 3 : GET(ecode, 1);
1159            }
1160          }
1161    
1162        else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1163        {        {
1164        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1165        condition = (offset == CREF_RECURSE * 2)?        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1166          (md->recursive != NULL) :  
1167          (offset < offset_top && md->offset_vector[offset] >= 0);        /* If the numbered capture is unset, but the reference was by name,
1168        RMATCH(rrc, eptr, ecode + (condition?        scan the table to see if the name refers to any other numbers, and test
1169          (LINK_SIZE + 4) : (LINK_SIZE + 1 + GET(ecode, 1))),        them. The condition is true if any one is set. This is tediously similar
1170          offset_top, md, ims, eptrb, match_isgroup);        to the code above, but not close enough to try to amalgamate. */
1171        RRETURN(rrc);  
1172          if (!condition && condcode == OP_NCREF)
1173            {
1174            int refno = offset >> 1;
1175            uschar *slotA = md->name_table;
1176    
1177            for (i = 0; i < md->name_count; i++)
1178              {
1179              if (GET2(slotA, 0) == refno) break;
1180              slotA += md->name_entry_size;
1181              }
1182    
1183            /* Found a name for the number - there can be only one; duplicate names
1184            for different numbers are allowed, but not vice versa. First scan down
1185            for duplicates. */
1186    
1187            if (i < md->name_count)
1188              {
1189              uschar *slotB = slotA;
1190              while (slotB > md->name_table)
1191                {
1192                slotB -= md->name_entry_size;
1193                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1194                  {
1195                  offset = GET2(slotB, 0) << 1;
1196                  condition = offset < offset_top &&
1197                    md->offset_vector[offset] >= 0;
1198                  if (condition) break;
1199                  }
1200                else break;
1201                }
1202    
1203              /* Scan up for duplicates */
1204    
1205              if (!condition)
1206                {
1207                slotB = slotA;
1208                for (i++; i < md->name_count; i++)
1209                  {
1210                  slotB += md->name_entry_size;
1211                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1212                    {
1213                    offset = GET2(slotB, 0) << 1;
1214                    condition = offset < offset_top &&
1215                      md->offset_vector[offset] >= 0;
1216                    if (condition) break;
1217                    }
1218                  else break;
1219                  }
1220                }
1221              }
1222            }
1223    
1224          /* Chose branch according to the condition */
1225    
1226          ecode += condition? 3 : GET(ecode, 1);
1227          }
1228    
1229        else if (condcode == OP_DEF)     /* DEFINE - always false */
1230          {
1231          condition = FALSE;
1232          ecode += GET(ecode, 1);
1233        }        }
1234    
1235      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1236      the final argument TRUE causes it to stop at the end of an assertion. */      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1237        an assertion. */
1238    
1239      else      else
1240        {        {
1241        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1242            match_condassert | match_isgroup);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1243        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1244          {          {
1245          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE+2);          condition = TRUE;
1246            ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1247          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1248          }          }
1249        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH &&
1250                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1251          {          {
1252          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1253          }          }
1254        else ecode += GET(ecode, 1);        else
1255        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,          {
1256          match_isgroup);          condition = FALSE;
1257        RRETURN(rrc);          ecode += codelink;
1258            }
1259        }        }
     /* Control never reaches here */  
1260    
1261      /* Skip over conditional reference or large extraction number data if      /* We are now at the branch that is to be obeyed. As there is only one,
1262      encountered. */      we can use tail recursion to avoid using another stack frame, except when
1263        we have an unlimited repeat of a possibly empty group. If the second
1264        alternative doesn't exist, we can just plough on. */
1265    
1266        if (condition || *ecode == OP_ALT)
1267          {
1268          ecode += 1 + LINK_SIZE;
1269          if (op == OP_SCOND)        /* Possibly empty group */
1270            {
1271            md->match_function_type = MATCH_CBEGROUP;
1272            RMATCH(eptr, ecode, offset_top, md, eptrb, RM49);
1273            RRETURN(rrc);
1274            }
1275          else goto TAIL_RECURSE;
1276          }
1277        else                         /* Condition false & no alternative */
1278          {
1279          ecode += 1 + LINK_SIZE;
1280          }
1281        break;
1282    
1283    
1284        /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1285        to close any currently open capturing brackets. */
1286    
1287        case OP_CLOSE:
1288        number = GET2(ecode, 1);
1289        offset = number << 1;
1290    
1291    #ifdef PCRE_DEBUG
1292          printf("end bracket %d at *ACCEPT", number);
1293          printf("\n");
1294    #endif
1295    
1296      case OP_CREF:      md->capture_last = number;
1297      case OP_BRANUMBER:      if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1298          {
1299          md->offset_vector[offset] =
1300            md->offset_vector[md->offset_end - number];
1301          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1302          if (offset_top <= offset) offset_top = offset + 2;
1303          }
1304      ecode += 3;      ecode += 3;
1305      break;      break;
1306    
     /* End of the pattern. If we are in a recursion, we should restore the  
     offsets appropriately and continue from after the call. */  
1307    
1308        /* End of the pattern, either real or forced. If we are in a top-level
1309        recursion, we should restore the offsets appropriately and continue from
1310        after the call. */
1311    
1312        case OP_ACCEPT:
1313      case OP_END:      case OP_END:
1314      if (md->recursive != NULL && md->recursive->group_num == 0)      if (md->recursive != NULL && md->recursive->group_num == 0)
1315        {        {
1316        recursion_info *rec = md->recursive;        recursion_info *rec = md->recursive;
1317        DPRINTF(("Hit the end in a (?0) recursion\n"));        DPRINTF(("End of pattern in a (?0) recursion\n"));
1318        md->recursive = rec->prevrec;        md->recursive = rec->prevrec;
1319        memmove(md->offset_vector, rec->offset_save,        memmove(md->offset_vector, rec->offset_save,
1320          rec->saved_max * sizeof(int));          rec->saved_max * sizeof(int));
1321        md->start_match = rec->save_start;        offset_top = rec->save_offset_top;
       ims = original_ims;  
1322        ecode = rec->after_call;        ecode = rec->after_call;
1323        break;        break;
1324        }        }
1325    
1326      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is
1327      string - backtracking will then try other alternatives, if any. */      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of
1328        the subject. In both cases, backtracking will then try other alternatives,
1329        if any. */
1330    
1331        if (eptr == mstart &&
1332            (md->notempty ||
1333              (md->notempty_atstart &&
1334                mstart == md->start_subject + md->start_offset)))
1335          MRRETURN(MATCH_NOMATCH);
1336    
1337        /* Otherwise, we have a match. */
1338    
1339        md->end_match_ptr = eptr;           /* Record where we ended */
1340        md->end_offset_top = offset_top;    /* and how many extracts were taken */
1341        md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1342    
1343      if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH);      /* For some reason, the macros don't work properly if an expression is
1344      md->end_match_ptr = eptr;          /* Record where we ended */      given as the argument to MRRETURN when the heap is in use. */
1345      md->end_offset_top = offset_top;   /* and how many extracts were taken */  
1346      RRETURN(MATCH_MATCH);      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1347        MRRETURN(rrc);
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1348    
1349      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1350      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,
1351      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
1352      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
1353      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1354        of a condition, we want to return immediately afterwards. The caller of
1355        this incarnation of the match() function will have set MATCH_CONDASSERT in
1356        md->match_function type, and one of these opcodes will be the first opcode
1357        that is processed. We use a local variable that is preserved over calls to
1358        match() to remember this case. */
1359    
1360      case OP_ASSERT:      case OP_ASSERT:
1361      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1362        if (md->match_function_type == MATCH_CONDASSERT)
1363          {
1364          condassert = TRUE;
1365          md->match_function_type = 0;
1366          }
1367        else condassert = FALSE;
1368    
1369      do      do
1370        {        {
1371        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1372          match_isgroup);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1373        if (rrc == MATCH_MATCH) break;          {
1374        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          mstart = md->start_match_ptr;   /* In case \K reset it */
1375            break;
1376            }
1377          if (rrc != MATCH_NOMATCH &&
1378              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1379            RRETURN(rrc);
1380        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1381        }        }
1382      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1383      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1384        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1385    
1386      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1387    
1388      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1389    
1390      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1391      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 738  for (;;) Line 1395  for (;;)
1395      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1396      continue;      continue;
1397    
1398      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1399        PRUNE, or COMMIT means we must assume failure without checking subsequent
1400        branches. */
1401    
1402      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1403      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1404        if (md->match_function_type == MATCH_CONDASSERT)
1405          {
1406          condassert = TRUE;
1407          md->match_function_type = 0;
1408          }
1409        else condassert = FALSE;
1410    
1411      do      do
1412        {        {
1413        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1414          match_isgroup);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1415        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1416        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          {
1417            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1418            break;
1419            }
1420          if (rrc != MATCH_NOMATCH &&
1421              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1422            RRETURN(rrc);
1423        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1424        }        }
1425      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1426    
1427      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1428    
1429      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1430      continue;      continue;
1431    
# Line 766  for (;;) Line 1438  for (;;)
1438  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1439      if (utf8)      if (utf8)
1440        {        {
1441        c = GET(ecode,1);        i = GET(ecode, 1);
1442        for (i = 0; i < c; i++)        while (i-- > 0)
1443          {          {
1444          eptr--;          eptr--;
1445          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1446          BACKCHAR(eptr)          BACKCHAR(eptr);
1447          }          }
1448        }        }
1449      else      else
# Line 780  for (;;) Line 1452  for (;;)
1452      /* No UTF-8 support, or not in UTF-8 mode: count is byte count */      /* No UTF-8 support, or not in UTF-8 mode: count is byte count */
1453    
1454        {        {
1455        eptr -= GET(ecode,1);        eptr -= GET(ecode, 1);
1456        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1457        }        }
1458    
1459      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1460    
1461        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1462      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1463      break;      break;
1464    
# Line 800  for (;;) Line 1473  for (;;)
1473        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 1;   /* Version 1 of the callout block */
1474        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1475        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1476        cb.subject          = (const char *)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1477        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1478        cb.start_match      = md->start_match - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1479        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1480        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1481        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1482        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1483        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1484        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1485        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1486        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1487        }        }
1488      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 837  for (;;) Line 1510  for (;;)
1510      case OP_RECURSE:      case OP_RECURSE:
1511        {        {
1512        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1513        new_recursive.group_num = *callpat - OP_BRA;        new_recursive.group_num = (callpat == md->start_code)? 0 :
1514            GET2(callpat, 1 + LINK_SIZE);
       /* For extended extraction brackets (large number), we have to fish out  
       the number from a dummy opcode at the start. */  
   
       if (new_recursive.group_num > EXTRACT_BASIC_MAX)  
         new_recursive.group_num = GET2(callpat, 2+LINK_SIZE);  
1515    
1516        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1517    
# Line 869  for (;;) Line 1537  for (;;)
1537    
1538        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1539              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1540        new_recursive.save_start = md->start_match;        new_recursive.save_offset_top = offset_top;
       md->start_match = eptr;  
1541    
1542        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. For each top-level alternative we
1543        restore the offset and recursion data. */        restore the offset and recursion data. */
1544    
1545        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1546          cbegroup = (*callpat >= OP_SBRA);
1547        do        do
1548          {          {
1549          RMATCH(rrc, eptr, callpat + 1 + LINK_SIZE, offset_top, md, ims,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1550              eptrb, match_isgroup);          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1551          if (rrc == MATCH_MATCH)            md, eptrb, RM6);
1552            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1553            {            {
1554              DPRINTF(("Recursion matched\n"));
1555            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1556            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1557              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1558            RRETURN(MATCH_MATCH);            MRRETURN(MATCH_MATCH);
1559              }
1560            else if (rrc != MATCH_NOMATCH &&
1561                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1562              {
1563              DPRINTF(("Recursion gave error %d\n", rrc));
1564              if (new_recursive.offset_save != stacksave)
1565                (pcre_free)(new_recursive.offset_save);
1566              RRETURN(rrc);
1567            }            }
         else if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
1568    
1569          md->recursive = &new_recursive;          md->recursive = &new_recursive;
1570          memcpy(md->offset_vector, new_recursive.offset_save,          memcpy(md->offset_vector, new_recursive.offset_save,
# Line 900  for (;;) Line 1577  for (;;)
1577        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1578        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1579          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1580        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1581        }        }
1582      /* Control never reaches here */      /* Control never reaches here */
1583    
# Line 909  for (;;) Line 1586  for (;;)
1586      a move back into the brackets. Friedl calls these "atomic" subpatterns.      a move back into the brackets. Friedl calls these "atomic" subpatterns.
1587      Check the alternative branches in turn - the matching won't pass the KET      Check the alternative branches in turn - the matching won't pass the KET
1588      for this kind of subpattern. If any one branch matches, we carry on as at      for this kind of subpattern. If any one branch matches, we carry on as at
1589      the end of a normal bracket, leaving the subject pointer. */      the end of a normal bracket, leaving the subject pointer, but resetting
1590        the start-of-match value in case it was changed by \K. */
1591    
1592      case OP_ONCE:      case OP_ONCE:
1593        {      prev = ecode;
1594        prev = ecode;      saved_eptr = eptr;
       saved_eptr = eptr;  
1595    
1596        do      do
1597          {
1598          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1599          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
1600          {          {
1601          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,          mstart = md->start_match_ptr;
1602            eptrb, match_isgroup);          break;
         if (rrc == MATCH_MATCH) break;  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         ecode += GET(ecode,1);  
1603          }          }
1604        while (*ecode == OP_ALT);        if (rrc != MATCH_NOMATCH &&
1605              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1606        /* If hit the end of the group (which could be repeated), fail */          RRETURN(rrc);
1607          ecode += GET(ecode,1);
1608          }
1609        while (*ecode == OP_ALT);
1610    
1611        if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);      /* If hit the end of the group (which could be repeated), fail */
1612    
1613        /* Continue as from after the assertion, updating the offsets high water      if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
       mark, since extracts may have been taken. */  
1614    
1615        do ecode += GET(ecode,1); while (*ecode == OP_ALT);      /* Continue as from after the assertion, updating the offsets high water
1616        mark, since extracts may have been taken. */
1617    
1618        offset_top = md->end_offset_top;      do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
       eptr = md->end_match_ptr;  
1619    
1620        /* For a non-repeating ket, just continue at this level. This also      offset_top = md->end_offset_top;
1621        happens for a repeating ket if no characters were matched in the group.      eptr = md->end_match_ptr;
       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. */  
1622    
1623        if (*ecode == OP_KET || eptr == saved_eptr)      /* For a non-repeating ket, just continue at this level. This also
1624          {      happens for a repeating ket if no characters were matched in the group.
1625          ecode += 1+LINK_SIZE;      This is the forcible breaking of infinite loops as implemented in Perl
1626          break;      5.005. If there is an options reset, it will get obeyed in the normal
1627          }      course of events. */
1628    
1629        /* The repeating kets try the rest of the pattern or restart from the      if (*ecode == OP_KET || eptr == saved_eptr)
1630        preceding bracket, in the appropriate order. We need to reset any options        {
1631        that changed within the bracket before re-running it, so check the next        ecode += 1+LINK_SIZE;
1632        opcode. */        break;
1633          }
1634    
1635        if (ecode[1+LINK_SIZE] == OP_OPT)      /* The repeating kets try the rest of the pattern or restart from the
1636          {      preceding bracket, in the appropriate order. The second "call" of match()
1637          ims = (ims & ~PCRE_IMS) | ecode[4];      uses tail recursion, to avoid using another stack frame. */
         DPRINTF(("ims set to %02lx at group repeat\n", ims));  
         }  
1638    
1639        if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1640          {        {
1641          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM8);
1642          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1643          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);        ecode = prev;
1644          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        goto TAIL_RECURSE;
         }  
       else  /* OP_KETRMAX */  
         {  
         RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         }  
1645        }        }
1646      RRETURN(MATCH_NOMATCH);      else  /* OP_KETRMAX */
1647          {
1648          md->match_function_type = MATCH_CBEGROUP;
1649          RMATCH(eptr, prev, offset_top, md, eptrb, RM9);
1650          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1651          ecode += 1 + LINK_SIZE;
1652          goto TAIL_RECURSE;
1653          }
1654        /* Control never gets here */
1655    
1656      /* An alternation is the end of a branch; scan along to find the end of the      /* An alternation is the end of a branch; scan along to find the end of the
1657      bracketed group and go to there. */      bracketed group and go to there. */
# Line 985  for (;;) Line 1660  for (;;)
1660      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1661      break;      break;
1662    
1663      /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1664      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
1665      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
1666      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
1667      preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1668    
1669      case OP_BRAZERO:      case OP_BRAZERO:
1670        {      next = ecode + 1;
1671        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1672        RMATCH(rrc, eptr, next, offset_top, md, ims, eptrb, match_isgroup);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1673        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1674        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1+LINK_SIZE;  
       }  
1675      break;      break;
1676    
1677      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1678        {      next = ecode + 1;
1679        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1680        do next += GET(next,1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1681        RMATCH(rrc, eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb,      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1682          match_isgroup);      ecode++;
1683        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      break;
1684        ecode++;  
1685        }      case OP_SKIPZERO:
1686        next = ecode+1;
1687        do next += GET(next,1); while (*next == OP_ALT);
1688        ecode = next + 1 + LINK_SIZE;
1689      break;      break;
1690    
1691        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1692        here; just jump to the group, with allow_zero set TRUE. */
1693    
1694        case OP_BRAPOSZERO:
1695        op = *(++ecode);
1696        allow_zero = TRUE;
1697        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1698          goto POSSESSIVE_NON_CAPTURE;
1699    
1700      /* End of a group, repeated or non-repeating. If we are at the end of      /* End of a group, repeated or non-repeating. */
     an assertion "group", stop matching and return MATCH_MATCH, but record the  
     current high water mark for use by positive assertions. Do this also  
     for the "once" (not-backup up) groups. */  
1701    
1702      case OP_KET:      case OP_KET:
1703      case OP_KETRMIN:      case OP_KETRMIN:
1704      case OP_KETRMAX:      case OP_KETRMAX:
1705        {      case OP_KETRPOS:
1706        prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
       saved_eptr = eptrb->epb_saved_eptr;  
   
       /* Back up the stack of bracket start pointers. */  
1707    
1708        eptrb = eptrb->epb_prev;      /* If this was a group that remembered the subject start, in order to break
1709        infinite repeats of empty string matches, retrieve the subject start from
1710        the chain. Otherwise, set it NULL. */
1711    
1712        if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev >= OP_SBRA)
1713            *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||        {
1714            *prev == OP_ONCE)        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1715          {        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1716          md->end_match_ptr = eptr;      /* For ONCE */        }
1717          md->end_offset_top = offset_top;      else saved_eptr = NULL;
         RRETURN(MATCH_MATCH);  
         }  
1718    
1719        /* In all other cases except a conditional group we have to check the      /* If we are at the end of an assertion group or an atomic group, stop
1720        group number back at the start and if necessary complete handling an      matching and return MATCH_MATCH, but record the current high water mark for
1721        extraction by setting the offsets and bumping the high water mark. */      use by positive assertions. We also need to record the match start in case
1722        it was changed by \K. */
1723    
1724        if (*prev != OP_COND)      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1725          {          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||
1726          number = *prev - OP_BRA;          *prev == OP_ONCE)
1727          {
1728          md->end_match_ptr = eptr;      /* For ONCE */
1729          md->end_offset_top = offset_top;
1730          md->start_match_ptr = mstart;
1731          MRRETURN(MATCH_MATCH);
1732          }
1733    
1734          /* For extended extraction brackets (large number), we have to fish out      /* For capturing groups we have to check the group number back at the start
1735          the number from a dummy opcode at the start. */      and if necessary complete handling an extraction by setting the offsets and
1736        bumping the high water mark. Note that whole-pattern recursion is coded as
1737        a recurse into group 0, so it won't be picked up here. Instead, we catch it
1738        when the OP_END is reached. Other recursion is handled here. */
1739    
1740          if (number > EXTRACT_BASIC_MAX) number = GET2(prev, 2+LINK_SIZE);      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1741          offset = number << 1;          *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1742          {
1743          number = GET2(prev, 1+LINK_SIZE);
1744          offset = number << 1;
1745    
1746  #ifdef DEBUG  #ifdef PCRE_DEBUG
1747          printf("end bracket %d", number);        printf("end bracket %d", number);
1748          printf("\n");        printf("\n");
1749  #endif  #endif
1750    
1751          /* Test for a numbered group. This includes groups called as a result        md->capture_last = number;
1752          of recursion. Note that whole-pattern recursion is coded as a recurse        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
         into group 0, so it won't be picked up here. Instead, we catch it when  
         the OP_END is reached. */  
   
         if (number > 0)  
           {  
           md->capture_last = number;  
           if (offset >= md->offset_max) md->offset_overflow = TRUE; else  
             {  
             md->offset_vector[offset] =  
               md->offset_vector[md->offset_end - number];  
             md->offset_vector[offset+1] = eptr - md->start_subject;  
             if (offset_top <= offset) offset_top = offset + 2;  
             }  
   
           /* Handle a recursively called group. Restore the offsets  
           appropriately and continue from after the call. */  
   
           if (md->recursive != NULL && md->recursive->group_num == number)  
             {  
             recursion_info *rec = md->recursive;  
             DPRINTF(("Recursion (%d) succeeded - continuing\n", number));  
             md->recursive = rec->prevrec;  
             md->start_match = rec->save_start;  
             memcpy(md->offset_vector, rec->offset_save,  
               rec->saved_max * sizeof(int));  
             ecode = rec->after_call;  
             ims = original_ims;  
             break;  
             }  
           }  
         }  
   
       /* Reset the value of the ims flags, in case they got changed during  
       the group. */  
   
       ims = original_ims;  
       DPRINTF(("ims reset to %02lx\n", ims));  
   
       /* 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)  
1753          {          {
1754          ecode += 1 + LINK_SIZE;          md->offset_vector[offset] =
1755              md->offset_vector[md->offset_end - number];
1756            md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1757            if (offset_top <= offset) offset_top = offset + 2;
1758            }
1759    
1760          /* Handle a recursively called group. Restore the offsets
1761          appropriately and continue from after the call. */
1762    
1763          if (md->recursive != NULL && md->recursive->group_num == number)
1764            {
1765            recursion_info *rec = md->recursive;
1766            DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
1767            md->recursive = rec->prevrec;
1768            memcpy(md->offset_vector, rec->offset_save,
1769              rec->saved_max * sizeof(int));
1770            offset_top = rec->save_offset_top;
1771            ecode = rec->after_call;
1772          break;          break;
1773          }          }
1774          }
1775    
1776        /* The repeating kets try the rest of the pattern or restart from the      /* For a non-repeating ket, just continue at this level. This also
1777        preceding bracket, in the appropriate order. */      happens for a repeating ket if no characters were matched in the group.
1778        This is the forcible breaking of infinite loops as implemented in Perl
1779        5.005. If there is an options reset, it will get obeyed in the normal
1780        course of events. */
1781    
1782        if (*ecode == OP_KETRMIN)      if (*ecode == OP_KET || eptr == saved_eptr)
1783          {        {
1784          RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);        ecode += 1 + LINK_SIZE;
1785          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        break;
1786          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);        }
1787          if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
1788          }      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1789        else  /* OP_KETRMAX */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1790        at a time from the outer level, thus saving stack. */
1791    
1792        if (*ecode == OP_KETRPOS)
1793          {
1794          md->end_match_ptr = eptr;
1795          md->end_offset_top = offset_top;
1796          RRETURN(MATCH_KETRPOS);
1797          }
1798    
1799        /* The normal repeating kets try the rest of the pattern or restart from
1800        the preceding bracket, in the appropriate order. In the second case, we can
1801        use tail recursion to avoid using another stack frame, unless we have an
1802        unlimited repeat of a group that can match an empty string. */
1803    
1804        if (*ecode == OP_KETRMIN)
1805          {
1806          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1807          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1808          if (*prev >= OP_SBRA)    /* Could match an empty string */
1809          {          {
1810          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);          md->match_function_type = MATCH_CBEGROUP;
1811          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1812          RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);          RRETURN(rrc);
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
1813          }          }
1814          ecode = prev;
1815          goto TAIL_RECURSE;
1816        }        }
1817        else  /* OP_KETRMAX */
     RRETURN(MATCH_NOMATCH);  
   
     /* Start of subject unless notbol, or after internal newline if multiline */  
   
     case OP_CIRC:  
     if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);  
     if ((ims & PCRE_MULTILINE) != 0)  
1818        {        {
1819        if (eptr != md->start_subject && eptr[-1] != NEWLINE)        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1820          RRETURN(MATCH_NOMATCH);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1821        ecode++;        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1822        break;        ecode += 1 + LINK_SIZE;
1823          goto TAIL_RECURSE;
1824        }        }
1825      /* ... else fall through */      /* Control never gets here */
1826    
1827        /* Not multiline mode: start of subject assertion, unless notbol. */
1828    
1829        case OP_CIRC:
1830        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1831    
1832      /* Start of subject assertion */      /* Start of subject assertion */
1833    
1834      case OP_SOD:      case OP_SOD:
1835      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1836        ecode++;
1837        break;
1838    
1839        /* Multiline mode: start of subject unless notbol, or after any newline. */
1840    
1841        case OP_CIRCM:
1842        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1843        if (eptr != md->start_subject &&
1844            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1845          MRRETURN(MATCH_NOMATCH);
1846      ecode++;      ecode++;
1847      break;      break;
1848    
1849      /* Start of match assertion */      /* Start of match assertion */
1850    
1851      case OP_SOM:      case OP_SOM:
1852      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1853      ecode++;      ecode++;
1854      break;      break;
1855    
1856      /* 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. */  
1857    
1858      case OP_DOLL:      case OP_SET_SOM:
1859      if ((ims & PCRE_MULTILINE) != 0)      mstart = eptr;
1860        {      ecode++;
1861        if (eptr < md->end_subject)      break;
1862          { if (*eptr != NEWLINE) RRETURN(MATCH_NOMATCH); }  
1863        else      /* Multiline mode: assert before any newline, or before end of subject
1864          { if (md->noteol) RRETURN(MATCH_NOMATCH); }      unless noteol is set. */
1865        ecode++;  
1866        break;      case OP_DOLLM:
1867        }      if (eptr < md->end_subject)
1868          { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1869      else      else
1870        {        {
1871        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1872        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr < md->end_subject - 1 ||  
            (eptr == md->end_subject - 1 && *eptr != NEWLINE))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1873        }        }
1874      /* ... else fall through */      ecode++;
1875        break;
1876    
1877        /* Not multiline mode: assert before a terminating newline or before end of
1878        subject unless noteol is set. */
1879    
1880        case OP_DOLL:
1881        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1882        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1883    
1884        /* ... else fall through for endonly */
1885    
1886      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1887    
1888      case OP_EOD:      case OP_EOD:
1889      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1890        SCHECK_PARTIAL();
1891      ecode++;      ecode++;
1892      break;      break;
1893    
1894      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1895    
1896      case OP_EODN:      case OP_EODN:
1897      if (eptr < md->end_subject - 1 ||      ASSERT_NL_OR_EOS:
1898         (eptr == md->end_subject - 1 && *eptr != NEWLINE)) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject &&
1899            (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1900          MRRETURN(MATCH_NOMATCH);
1901    
1902        /* Either at end of string or \n before end. */
1903    
1904        SCHECK_PARTIAL();
1905      ecode++;      ecode++;
1906      break;      break;
1907    
# Line 1206  for (;;) Line 1913  for (;;)
1913    
1914        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1915        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
1916        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1917          partial matching. */
1918    
1919  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1920        if (utf8)        if (utf8)
1921          {          {
1922            /* Get status of previous character */
1923    
1924          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1925            {            {
1926            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1927            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1928              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1929            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1930    #ifdef SUPPORT_UCP
1931              if (md->use_ucp)
1932                {
1933                if (c == '_') prev_is_word = TRUE; else
1934                  {
1935                  int cat = UCD_CATEGORY(c);
1936                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1937                  }
1938                }
1939              else
1940    #endif
1941            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1942            }            }
1943          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
1944            /* Get status of next character */
1945    
1946            if (eptr >= md->end_subject)
1947              {
1948              SCHECK_PARTIAL();
1949              cur_is_word = FALSE;
1950              }
1951            else
1952            {            {
1953            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1954    #ifdef SUPPORT_UCP
1955              if (md->use_ucp)
1956                {
1957                if (c == '_') cur_is_word = TRUE; else
1958                  {
1959                  int cat = UCD_CATEGORY(c);
1960                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1961                  }
1962                }
1963              else
1964    #endif
1965            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1966            }            }
1967          }          }
1968        else        else
1969  #endif  #endif
1970    
1971        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1972          consistency with the behaviour of \w we do use it in this case. */
1973    
1974          {          {
1975          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
1976            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
1977          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
1978            ((md->ctypes[*eptr] & ctype_word) != 0);            {
1979              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1980    #ifdef SUPPORT_UCP
1981              if (md->use_ucp)
1982                {
1983                c = eptr[-1];
1984                if (c == '_') prev_is_word = TRUE; else
1985                  {
1986                  int cat = UCD_CATEGORY(c);
1987                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1988                  }
1989                }
1990              else
1991    #endif
1992              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1993              }
1994    
1995            /* Get status of next character */
1996    
1997            if (eptr >= md->end_subject)
1998              {
1999              SCHECK_PARTIAL();
2000              cur_is_word = FALSE;
2001              }
2002            else
2003    #ifdef SUPPORT_UCP
2004            if (md->use_ucp)
2005              {
2006              c = *eptr;
2007              if (c == '_') cur_is_word = TRUE; else
2008                {
2009                int cat = UCD_CATEGORY(c);
2010                cur_is_word = (cat == ucp_L || cat == ucp_N);
2011                }
2012              }
2013            else
2014    #endif
2015            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2016          }          }
2017    
2018        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2019    
2020        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2021             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2022          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2023        }        }
2024      break;      break;
2025    
2026      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2027    
2028      case OP_ANY:      case OP_ANY:
2029      if ((ims & PCRE_DOTALL) == 0 && eptr < md->end_subject && *eptr == NEWLINE)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2030        RRETURN(MATCH_NOMATCH);      /* Fall through */
2031      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);  
2032  #ifdef SUPPORT_UTF8      case OP_ALLANY:
2033      if (utf8)      if (eptr++ >= md->end_subject)
2034        while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;        {
2035  #endif        SCHECK_PARTIAL();
2036          MRRETURN(MATCH_NOMATCH);
2037          }
2038        if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2039      ecode++;      ecode++;
2040      break;      break;
2041    
# Line 1261  for (;;) Line 2043  for (;;)
2043      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2044    
2045      case OP_ANYBYTE:      case OP_ANYBYTE:
2046      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr++ >= md->end_subject)
2047          {
2048          SCHECK_PARTIAL();
2049          MRRETURN(MATCH_NOMATCH);
2050          }
2051      ecode++;      ecode++;
2052      break;      break;
2053    
2054      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2055      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2056          {
2057          SCHECK_PARTIAL();
2058          MRRETURN(MATCH_NOMATCH);
2059          }
2060      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2061      if (      if (
2062  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1274  for (;;) Line 2064  for (;;)
2064  #endif  #endif
2065         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2066         )         )
2067        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2068      ecode++;      ecode++;
2069      break;      break;
2070    
2071      case OP_DIGIT:      case OP_DIGIT:
2072      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2073          {
2074          SCHECK_PARTIAL();
2075          MRRETURN(MATCH_NOMATCH);
2076          }
2077      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2078      if (      if (
2079  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1287  for (;;) Line 2081  for (;;)
2081  #endif  #endif
2082         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2083         )         )
2084        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2085      ecode++;      ecode++;
2086      break;      break;
2087    
2088      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2089      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2090          {
2091          SCHECK_PARTIAL();
2092          MRRETURN(MATCH_NOMATCH);
2093          }
2094      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2095      if (      if (
2096  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1300  for (;;) Line 2098  for (;;)
2098  #endif  #endif
2099         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2100         )         )
2101        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2102      ecode++;      ecode++;
2103      break;      break;
2104    
2105      case OP_WHITESPACE:      case OP_WHITESPACE:
2106      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2107          {
2108          SCHECK_PARTIAL();
2109          MRRETURN(MATCH_NOMATCH);
2110          }
2111      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2112      if (      if (
2113  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1313  for (;;) Line 2115  for (;;)
2115  #endif  #endif
2116         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2117         )         )
2118        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2119      ecode++;      ecode++;
2120      break;      break;
2121    
2122      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2123      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2124          {
2125          SCHECK_PARTIAL();
2126          MRRETURN(MATCH_NOMATCH);
2127          }
2128      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2129      if (      if (
2130  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1326  for (;;) Line 2132  for (;;)
2132  #endif  #endif
2133         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2134         )         )
2135        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2136      ecode++;      ecode++;
2137      break;      break;
2138    
2139      case OP_WORDCHAR:      case OP_WORDCHAR:
2140      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2141          {
2142          SCHECK_PARTIAL();
2143          MRRETURN(MATCH_NOMATCH);
2144          }
2145      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2146      if (      if (
2147  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1339  for (;;) Line 2149  for (;;)
2149  #endif  #endif
2150         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2151         )         )
2152        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2153        ecode++;
2154        break;
2155    
2156        case OP_ANYNL:
2157        if (eptr >= md->end_subject)
2158          {
2159          SCHECK_PARTIAL();
2160          MRRETURN(MATCH_NOMATCH);
2161          }
2162        GETCHARINCTEST(c, eptr);
2163        switch(c)
2164          {
2165          default: MRRETURN(MATCH_NOMATCH);
2166    
2167          case 0x000d:
2168          if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2169          break;
2170    
2171          case 0x000a:
2172          break;
2173    
2174          case 0x000b:
2175          case 0x000c:
2176          case 0x0085:
2177          case 0x2028:
2178          case 0x2029:
2179          if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2180          break;
2181          }
2182        ecode++;
2183        break;
2184    
2185        case OP_NOT_HSPACE:
2186        if (eptr >= md->end_subject)
2187          {
2188          SCHECK_PARTIAL();
2189          MRRETURN(MATCH_NOMATCH);
2190          }
2191        GETCHARINCTEST(c, eptr);
2192        switch(c)
2193          {
2194          default: break;
2195          case 0x09:      /* HT */
2196          case 0x20:      /* SPACE */
2197          case 0xa0:      /* NBSP */
2198          case 0x1680:    /* OGHAM SPACE MARK */
2199          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2200          case 0x2000:    /* EN QUAD */
2201          case 0x2001:    /* EM QUAD */
2202          case 0x2002:    /* EN SPACE */
2203          case 0x2003:    /* EM SPACE */
2204          case 0x2004:    /* THREE-PER-EM SPACE */
2205          case 0x2005:    /* FOUR-PER-EM SPACE */
2206          case 0x2006:    /* SIX-PER-EM SPACE */
2207          case 0x2007:    /* FIGURE SPACE */
2208          case 0x2008:    /* PUNCTUATION SPACE */
2209          case 0x2009:    /* THIN SPACE */
2210          case 0x200A:    /* HAIR SPACE */
2211          case 0x202f:    /* NARROW NO-BREAK SPACE */
2212          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2213          case 0x3000:    /* IDEOGRAPHIC SPACE */
2214          MRRETURN(MATCH_NOMATCH);
2215          }
2216        ecode++;
2217        break;
2218    
2219        case OP_HSPACE:
2220        if (eptr >= md->end_subject)
2221          {
2222          SCHECK_PARTIAL();
2223          MRRETURN(MATCH_NOMATCH);
2224          }
2225        GETCHARINCTEST(c, eptr);
2226        switch(c)
2227          {
2228          default: MRRETURN(MATCH_NOMATCH);
2229          case 0x09:      /* HT */
2230          case 0x20:      /* SPACE */
2231          case 0xa0:      /* NBSP */
2232          case 0x1680:    /* OGHAM SPACE MARK */
2233          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2234          case 0x2000:    /* EN QUAD */
2235          case 0x2001:    /* EM QUAD */
2236          case 0x2002:    /* EN SPACE */
2237          case 0x2003:    /* EM SPACE */
2238          case 0x2004:    /* THREE-PER-EM SPACE */
2239          case 0x2005:    /* FOUR-PER-EM SPACE */
2240          case 0x2006:    /* SIX-PER-EM SPACE */
2241          case 0x2007:    /* FIGURE SPACE */
2242          case 0x2008:    /* PUNCTUATION SPACE */
2243          case 0x2009:    /* THIN SPACE */
2244          case 0x200A:    /* HAIR SPACE */
2245          case 0x202f:    /* NARROW NO-BREAK SPACE */
2246          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2247          case 0x3000:    /* IDEOGRAPHIC SPACE */
2248          break;
2249          }
2250        ecode++;
2251        break;
2252    
2253        case OP_NOT_VSPACE:
2254        if (eptr >= md->end_subject)
2255          {
2256          SCHECK_PARTIAL();
2257          MRRETURN(MATCH_NOMATCH);
2258          }
2259        GETCHARINCTEST(c, eptr);
2260        switch(c)
2261          {
2262          default: break;
2263          case 0x0a:      /* LF */
2264          case 0x0b:      /* VT */
2265          case 0x0c:      /* FF */
2266          case 0x0d:      /* CR */
2267          case 0x85:      /* NEL */
2268          case 0x2028:    /* LINE SEPARATOR */
2269          case 0x2029:    /* PARAGRAPH SEPARATOR */
2270          MRRETURN(MATCH_NOMATCH);
2271          }
2272        ecode++;
2273        break;
2274    
2275        case OP_VSPACE:
2276        if (eptr >= md->end_subject)
2277          {
2278          SCHECK_PARTIAL();
2279          MRRETURN(MATCH_NOMATCH);
2280          }
2281        GETCHARINCTEST(c, eptr);
2282        switch(c)
2283          {
2284          default: MRRETURN(MATCH_NOMATCH);
2285          case 0x0a:      /* LF */
2286          case 0x0b:      /* VT */
2287          case 0x0c:      /* FF */
2288          case 0x0d:      /* CR */
2289          case 0x85:      /* NEL */
2290          case 0x2028:    /* LINE SEPARATOR */
2291          case 0x2029:    /* PARAGRAPH SEPARATOR */
2292          break;
2293          }
2294      ecode++;      ecode++;
2295      break;      break;
2296    
# Line 1349  for (;;) Line 2300  for (;;)
2300    
2301      case OP_PROP:      case OP_PROP:
2302      case OP_NOTPROP:      case OP_NOTPROP:
2303      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2304          {
2305          SCHECK_PARTIAL();
2306          MRRETURN(MATCH_NOMATCH);
2307          }
2308      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2309        {        {
2310        int chartype, rqdtype;        const ucd_record *prop = GET_UCD(c);
       int othercase;  
       int category = _pcre_ucp_findchar(c, &chartype, &othercase);  
   
       rqdtype = *(++ecode);  
       ecode++;  
2311    
2312        if (rqdtype >= 128)        switch(ecode[1])
         {  
         if ((rqdtype - 128 != category) == (op == OP_PROP))  
           RRETURN(MATCH_NOMATCH);  
         }  
       else  
2313          {          {
2314          if ((rqdtype != chartype) == (op == OP_PROP))          case PT_ANY:
2315            RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2316            break;
2317    
2318            case PT_LAMP:
2319            if ((prop->chartype == ucp_Lu ||
2320                 prop->chartype == ucp_Ll ||
2321                 prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2322              MRRETURN(MATCH_NOMATCH);
2323            break;
2324    
2325            case PT_GC:
2326            if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2327              MRRETURN(MATCH_NOMATCH);
2328            break;
2329    
2330            case PT_PC:
2331            if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2332              MRRETURN(MATCH_NOMATCH);
2333            break;
2334    
2335            case PT_SC:
2336            if ((ecode[2] != prop->script) == (op == OP_PROP))
2337              MRRETURN(MATCH_NOMATCH);
2338            break;
2339    
2340            /* These are specials */
2341    
2342            case PT_ALNUM:
2343            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2344                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2345              MRRETURN(MATCH_NOMATCH);
2346            break;
2347    
2348            case PT_SPACE:    /* Perl space */
2349            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2350                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2351                   == (op == OP_NOTPROP))
2352              MRRETURN(MATCH_NOMATCH);
2353            break;
2354    
2355            case PT_PXSPACE:  /* POSIX space */
2356            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2357                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2358                 c == CHAR_FF || c == CHAR_CR)
2359                   == (op == OP_NOTPROP))
2360              MRRETURN(MATCH_NOMATCH);
2361            break;
2362    
2363            case PT_WORD:
2364            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2365                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2366                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2367              MRRETURN(MATCH_NOMATCH);
2368            break;
2369    
2370            /* This should never occur */
2371    
2372            default:
2373            RRETURN(PCRE_ERROR_INTERNAL);
2374          }          }
2375    
2376          ecode += 3;
2377        }        }
2378      break;      break;
2379    
# Line 1376  for (;;) Line 2381  for (;;)
2381      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2382    
2383      case OP_EXTUNI:      case OP_EXTUNI:
2384      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2385          {
2386          SCHECK_PARTIAL();
2387          MRRETURN(MATCH_NOMATCH);
2388          }
2389      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2390        {        {
2391        int chartype;        int category = UCD_CATEGORY(c);
2392        int othercase;        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);
       int category = _pcre_ucp_findchar(c, &chartype, &othercase);  
       if (category == ucp_M) RRETURN(MATCH_NOMATCH);  
2393        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2394          {          {
2395          int len = 1;          int len = 1;
# Line 1390  for (;;) Line 2397  for (;;)
2397            {            {
2398            GETCHARLEN(c, eptr, len);            GETCHARLEN(c, eptr, len);
2399            }            }
2400          category = _pcre_ucp_findchar(c, &chartype, &othercase);          category = UCD_CATEGORY(c);
2401          if (category != ucp_M) break;          if (category != ucp_M) break;
2402          eptr += len;          eptr += len;
2403          }          }
# Line 1409  for (;;) Line 2416  for (;;)
2416      loops). */      loops). */
2417    
2418      case OP_REF:      case OP_REF:
2419        case OP_REFI:
2420        caseless = op == OP_REFI;
2421        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2422        ecode += 3;
2423    
2424        /* If the reference is unset, there are two possibilities:
2425    
2426        (a) In the default, Perl-compatible state, set the length negative;
2427        this ensures that every attempt at a match fails. We can't just fail
2428        here, because of the possibility of quantifiers with zero minima.
2429    
2430        (b) If the JavaScript compatibility flag is set, set the length to zero
2431        so that the back reference matches an empty string.
2432    
2433        Otherwise, set the length to the length of what was matched by the
2434        referenced subpattern. */
2435    
2436        if (offset >= offset_top || md->offset_vector[offset] < 0)
2437          length = (md->jscript_compat)? 0 : -1;
2438        else
2439          length = md->offset_vector[offset+1] - md->offset_vector[offset];
2440    
2441        /* Set up for repetition, or handle the non-repeated case */
2442    
2443        switch (*ecode)
2444        {        {
2445        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */        case OP_CRSTAR:
2446        ecode += 3;                                 /* Advance past item */        case OP_CRMINSTAR:
2447          case OP_CRPLUS:
2448          case OP_CRMINPLUS:
2449          case OP_CRQUERY:
2450          case OP_CRMINQUERY:
2451          c = *ecode++ - OP_CRSTAR;
2452          minimize = (c & 1) != 0;
2453          min = rep_min[c];                 /* Pick up values from tables; */
2454          max = rep_max[c];                 /* zero for max => infinity */
2455          if (max == 0) max = INT_MAX;
2456          break;
2457    
2458        /* If the reference is unset, set the length to be longer than the amount        case OP_CRRANGE:
2459        of subject left; this ensures that every attempt at a match fails. We        case OP_CRMINRANGE:
2460        can't just fail here, because of the possibility of quantifiers with zero        minimize = (*ecode == OP_CRMINRANGE);
2461        minima. */        min = GET2(ecode, 1);
2462          max = GET2(ecode, 3);
2463        length = (offset >= offset_top || md->offset_vector[offset] < 0)?        if (max == 0) max = INT_MAX;
2464          md->end_subject - eptr + 1 :        ecode += 5;
2465          md->offset_vector[offset+1] - md->offset_vector[offset];        break;
2466    
2467        /* Set up for repetition, or handle the non-repeated case */        default:               /* No repeat follows */
2468          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2469            {
2470            CHECK_PARTIAL();
2471            MRRETURN(MATCH_NOMATCH);
2472            }
2473          eptr += length;
2474          continue;              /* With the main loop */
2475          }
2476    
2477        switch (*ecode)      /* Handle repeated back references. If the length of the reference is
2478        zero, just continue with the main loop. */
2479    
2480        if (length == 0) continue;
2481    
2482        /* First, ensure the minimum number of matches are present. We get back
2483        the length of the reference string explicitly rather than passing the
2484        address of eptr, so that eptr can be a register variable. */
2485    
2486        for (i = 1; i <= min; i++)
2487          {
2488          int slength;
2489          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2490          {          {
2491          case OP_CRSTAR:          CHECK_PARTIAL();
2492          case OP_CRMINSTAR:          MRRETURN(MATCH_NOMATCH);
         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;  
   
         case OP_CRRANGE:  
         case OP_CRMINRANGE:  
         minimize = (*ecode == OP_CRMINRANGE);  
         min = GET2(ecode, 1);  
         max = GET2(ecode, 3);  
         if (max == 0) max = INT_MAX;  
         ecode += 5;  
         break;  
   
         default:               /* No repeat follows */  
         if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);  
         eptr += length;  
         continue;              /* With the main loop */  
         }  
   
       /* If the length of the reference is zero, just continue with the  
       main loop. */  
   
       if (length == 0) continue;  
   
       /* First, ensure the minimum number of matches are present. We get back  
       the length of the reference string explicitly rather than passing the  
       address of eptr, so that eptr can be a register variable. */  
   
       for (i = 1; i <= min; i++)  
         {  
         if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);  
         eptr += length;  
2493          }          }
2494          eptr += slength;
2495          }
2496    
2497        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2498        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2499    
2500        if (min == max) continue;      if (min == max) continue;
2501    
2502        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2503    
2504        if (minimize)      if (minimize)
2505          {
2506          for (fi = min;; fi++)
2507          {          {
2508          for (fi = min;; fi++)          int slength;
2509            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2510            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2511            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2512            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2513            {            {
2514            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            CHECK_PARTIAL();
2515            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2516            }            }
2517          /* Control never gets here */          eptr += slength;
2518          }          }
2519          /* Control never gets here */
2520          }
2521    
2522        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2523    
2524        else      else
2525          {
2526          pp = eptr;
2527          for (i = min; i < max; i++)
2528          {          {
2529          pp = eptr;          int slength;
2530          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)  
2531            {            {
2532            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            CHECK_PARTIAL();
2533            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2534            }            }
2535          RRETURN(MATCH_NOMATCH);          eptr += slength;
2536          }          }
2537          while (eptr >= pp)
2538            {
2539            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2540            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2541            eptr -= length;
2542            }
2543          MRRETURN(MATCH_NOMATCH);
2544        }        }
2545      /* Control never gets here */      /* Control never gets here */
2546    
   
   
2547      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2548      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,
2549      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 1566  for (;;) Line 2598  for (;;)
2598          {          {
2599          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2600            {            {
2601            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2602                {
2603                SCHECK_PARTIAL();
2604                MRRETURN(MATCH_NOMATCH);
2605                }
2606            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2607            if (c > 255)            if (c > 255)
2608              {              {
2609              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2610              }              }
2611            else            else
2612              {              {
2613              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2614              }              }
2615            }            }
2616          }          }
# Line 1584  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            c = *eptr++;            c = *eptr++;
2629            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2630            }            }
2631          }          }
2632    
# Line 1606  for (;;) Line 2646  for (;;)
2646            {            {
2647            for (fi = min;; fi++)            for (fi = min;; fi++)
2648              {              {
2649              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2650              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2651              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2652                if (eptr >= md->end_subject)
2653                  {
2654                  SCHECK_PARTIAL();
2655                  MRRETURN(MATCH_NOMATCH);
2656                  }
2657              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2658              if (c > 255)              if (c > 255)
2659                {                {
2660                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2661                }                }
2662              else              else
2663                {                {
2664                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2665                }                }
2666              }              }
2667            }            }
# Line 1626  for (;;) Line 2671  for (;;)
2671            {            {
2672            for (fi = min;; fi++)            for (fi = min;; fi++)
2673              {              {
2674              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2675              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2676              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2677                if (eptr >= md->end_subject)
2678                  {
2679                  SCHECK_PARTIAL();
2680                  MRRETURN(MATCH_NOMATCH);
2681                  }
2682              c = *eptr++;              c = *eptr++;
2683              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2684              }              }
2685            }            }
2686          /* Control never gets here */          /* Control never gets here */
# Line 1649  for (;;) Line 2699  for (;;)
2699            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2700              {              {
2701              int len = 1;              int len = 1;
2702              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2703                  {
2704                  SCHECK_PARTIAL();
2705                  break;
2706                  }
2707              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2708              if (c > 255)              if (c > 255)
2709                {                {
# Line 1663  for (;;) Line 2717  for (;;)
2717              }              }
2718            for (;;)            for (;;)
2719              {              {
2720              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2721              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2722              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2723              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1675  for (;;) Line 2729  for (;;)
2729            {            {
2730            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2731              {              {
2732              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2733                  {
2734                  SCHECK_PARTIAL();
2735                  break;
2736                  }
2737              c = *eptr;              c = *eptr;
2738              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2739              eptr++;              eptr++;
2740              }              }
2741            while (eptr >= pp)            while (eptr >= pp)
2742              {              {
2743              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
             eptr--;  
2744              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2745                eptr--;
2746              }              }
2747            }            }
2748    
2749          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2750          }          }
2751        }        }
2752      /* Control never gets here */      /* Control never gets here */
2753    
2754    
2755      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2756      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
2757        mode, because Unicode properties are supported in non-UTF-8 mode. */
2758    
2759  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2760      case OP_XCLASS:      case OP_XCLASS:
# Line 1736  for (;;) Line 2795  for (;;)
2795    
2796        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2797          {          {
2798          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2799          GETCHARINC(c, eptr);            {
2800          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2801              MRRETURN(MATCH_NOMATCH);
2802              }
2803            GETCHARINCTEST(c, eptr);
2804            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2805          }          }
2806    
2807        /* 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 1753  for (;;) Line 2816  for (;;)
2816          {          {
2817          for (fi = min;; fi++)          for (fi = min;; fi++)
2818            {            {
2819            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2820            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2821            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2822            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2823            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2824                SCHECK_PARTIAL();
2825                MRRETURN(MATCH_NOMATCH);
2826                }
2827              GETCHARINCTEST(c, eptr);
2828              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2829            }            }
2830          /* Control never gets here */          /* Control never gets here */
2831          }          }
# Line 1770  for (;;) Line 2838  for (;;)
2838          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2839            {            {
2840            int len = 1;            int len = 1;
2841            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2842            GETCHARLEN(c, eptr, len);              {
2843                SCHECK_PARTIAL();
2844                break;
2845                }
2846              GETCHARLENTEST(c, eptr, len);
2847            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2848            eptr += len;            eptr += len;
2849            }            }
2850          for(;;)          for(;;)
2851            {            {
2852            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2853            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2854            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2855            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2856            }            }
2857          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2858          }          }
2859    
2860        /* Control never gets here */        /* Control never gets here */
# Line 1798  for (;;) Line 2870  for (;;)
2870        length = 1;        length = 1;
2871        ecode++;        ecode++;
2872        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2873        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2874        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2875            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2876            MRRETURN(MATCH_NOMATCH);
2877            }
2878          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2879        }        }
2880      else      else
2881  #endif  #endif
2882    
2883      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2884        {        {
2885        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2886        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2887            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2888            MRRETURN(MATCH_NOMATCH);
2889            }
2890          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2891        ecode += 2;        ecode += 2;
2892        }        }
2893      break;      break;
2894    
2895      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2896    
2897      case OP_CHARNC:      case OP_CHARI:
2898  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2899      if (utf8)      if (utf8)
2900        {        {
# Line 1822  for (;;) Line 2902  for (;;)
2902        ecode++;        ecode++;
2903        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2904    
2905        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2906            {
2907            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2908            MRRETURN(MATCH_NOMATCH);
2909            }
2910    
2911        /* 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
2912        can use the fast lookup table. */        can use the fast lookup table. */
2913    
2914        if (fc < 128)        if (fc < 128)
2915          {          {
2916          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2917          }          }
2918    
2919        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
2920    
2921        else        else
2922          {          {
2923          int dc;          unsigned int dc;
2924          GETCHARINC(dc, eptr);          GETCHARINC(dc, eptr);
2925          ecode += length;          ecode += length;
2926    
2927          /* If we have Unicode property support, we can use it to test the other          /* If we have Unicode property support, we can use it to test the other
2928          case of the character, if there is one. The result of _pcre_ucp_findchar() is          case of the character, if there is one. */
         < 0 if the char isn't found, and othercase is returned as zero if there  
         isn't one. */  
2929    
2930          if (fc != dc)          if (fc != dc)
2931            {            {
2932  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2933            int chartype;            if (dc != UCD_OTHERCASE(fc))
           int othercase;  
           if (_pcre_ucp_findchar(fc, &chartype, &othercase) < 0 || dc != othercase)  
2934  #endif  #endif
2935              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
2936            }            }
2937          }          }
2938        }        }
# Line 1861  for (;;) Line 2941  for (;;)
2941    
2942      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2943        {        {
2944        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2945        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
2946            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2947            MRRETURN(MATCH_NOMATCH);
2948            }
2949          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2950        ecode += 2;        ecode += 2;
2951        }        }
2952      break;      break;
2953    
2954      /* Match a single character repeatedly; different opcodes share code. */      /* Match a single character repeatedly. */
2955    
2956      case OP_EXACT:      case OP_EXACT:
2957        case OP_EXACTI:
2958      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2959      ecode += 3;      ecode += 3;
2960      goto REPEATCHAR;      goto REPEATCHAR;
2961    
2962        case OP_POSUPTO:
2963        case OP_POSUPTOI:
2964        possessive = TRUE;
2965        /* Fall through */
2966    
2967      case OP_UPTO:      case OP_UPTO:
2968        case OP_UPTOI:
2969      case OP_MINUPTO:      case OP_MINUPTO:
2970        case OP_MINUPTOI:
2971      min = 0;      min = 0;
2972      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2973      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2974      ecode += 3;      ecode += 3;
2975      goto REPEATCHAR;      goto REPEATCHAR;
2976    
2977        case OP_POSSTAR:
2978        case OP_POSSTARI:
2979        possessive = TRUE;
2980        min = 0;
2981        max = INT_MAX;
2982        ecode++;
2983        goto REPEATCHAR;
2984    
2985        case OP_POSPLUS:
2986        case OP_POSPLUSI:
2987        possessive = TRUE;
2988        min = 1;
2989        max = INT_MAX;
2990        ecode++;
2991        goto REPEATCHAR;
2992    
2993        case OP_POSQUERY:
2994        case OP_POSQUERYI:
2995        possessive = TRUE;
2996        min = 0;
2997        max = 1;
2998        ecode++;
2999        goto REPEATCHAR;
3000    
3001      case OP_STAR:      case OP_STAR:
3002        case OP_STARI:
3003      case OP_MINSTAR:      case OP_MINSTAR:
3004        case OP_MINSTARI:
3005      case OP_PLUS:      case OP_PLUS:
3006        case OP_PLUSI:
3007      case OP_MINPLUS:      case OP_MINPLUS:
3008        case OP_MINPLUSI:
3009      case OP_QUERY:      case OP_QUERY:
3010        case OP_QUERYI:
3011      case OP_MINQUERY:      case OP_MINQUERY:
3012      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3013        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3014      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3015      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3016      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3017      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3018    
3019      /* 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. */  
3020    
3021      REPEATCHAR:      REPEATCHAR:
3022  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1905  for (;;) Line 3025  for (;;)
3025        length = 1;        length = 1;
3026        charptr = ecode;        charptr = ecode;
3027        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3028        ecode += length;        ecode += length;
3029    
3030        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 1913  for (;;) Line 3032  for (;;)
3032    
3033        if (length > 1)        if (length > 1)
3034          {          {
         int oclength = 0;  
         uschar occhars[8];  
   
3035  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3036          int othercase;          unsigned int othercase;
3037          int chartype;          if (op >= OP_STARI &&     /* Caseless */
3038          if ((ims & PCRE_CASELESS) != 0 &&              (othercase = UCD_OTHERCASE(fc)) != fc)
              _pcre_ucp_findchar(fc, &chartype, &othercase) >= 0 &&  
              othercase > 0)  
3039            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3040            else oclength = 0;
3041  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3042    
3043          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3044            {            {
3045            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3046            /* Need braces because of following else */              memcmp(eptr, charptr, length) == 0) eptr += length;
3047            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
3048              else if (oclength > 0 &&
3049                       eptr <= md->end_subject - oclength &&
3050                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3051    #endif  /* SUPPORT_UCP */
3052            else            else
3053              {              {
3054              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3055              eptr += oclength;              MRRETURN(MATCH_NOMATCH);
3056              }              }
3057            }            }
3058    
# Line 1943  for (;;) Line 3062  for (;;)
3062            {            {
3063            for (fi = min;; fi++)            for (fi = min;; fi++)
3064              {              {
3065              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3066              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3067              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3068              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3069              /* Need braces because of following else */                memcmp(eptr, charptr, length) == 0) eptr += length;
3070              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
3071                else if (oclength > 0 &&
3072                         eptr <= md->end_subject - oclength &&
3073                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3074    #endif  /* SUPPORT_UCP */
3075              else              else
3076                {                {
3077                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3078                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3079                }                }
3080              }              }
3081            /* Control never gets here */            /* Control never gets here */
3082            }            }
3083          else  
3084            else  /* Maximize */
3085            {            {
3086            pp = eptr;            pp = eptr;
3087            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3088              {              {
3089              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3090              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3091              else if (oclength == 0) break;  #ifdef SUPPORT_UCP
3092                else if (oclength > 0 &&
3093                         eptr <= md->end_subject - oclength &&
3094                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3095    #endif  /* SUPPORT_UCP */
3096              else              else
3097                {                {
3098                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3099                eptr += oclength;                break;
3100                }                }
3101              }              }
3102            while (eptr >= pp)  
3103             {            if (possessive) continue;
3104             RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);  
3105             if (rrc != MATCH_NOMATCH) RRETURN(rrc);            for(;;)
3106             eptr -= length;              {
3107             }              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3108            RRETURN(MATCH_NOMATCH);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3109                if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3110    #ifdef SUPPORT_UCP
3111                eptr--;
3112                BACKCHAR(eptr);
3113    #else   /* without SUPPORT_UCP */
3114                eptr -= length;
3115    #endif  /* SUPPORT_UCP */
3116                }
3117            }            }
3118          /* Control never gets here */          /* Control never gets here */
3119          }          }
# Line 1990  for (;;) Line 3126  for (;;)
3126  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3127    
3128      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3129        {  
3130        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3131    
3132      /* 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
3133      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 2007  for (;;) Line 3141  for (;;)
3141      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3142        max, eptr));        max, eptr));
3143    
3144      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3145        {        {
3146        fc = md->lcc[fc];        fc = md->lcc[fc];
3147        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3148          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3149            if (eptr >= md->end_subject)
3150              {
3151              SCHECK_PARTIAL();
3152              MRRETURN(MATCH_NOMATCH);
3153              }
3154            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3155            }
3156        if (min == max) continue;        if (min == max) continue;
3157        if (minimize)        if (minimize)
3158          {          {
3159          for (fi = min;; fi++)          for (fi = min;; fi++)
3160            {            {
3161            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3162            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3163            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3164                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3165              RRETURN(MATCH_NOMATCH);              {
3166                SCHECK_PARTIAL();
3167                MRRETURN(MATCH_NOMATCH);
3168                }
3169              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3170            }            }
3171          /* Control never gets here */          /* Control never gets here */
3172          }          }
3173        else        else  /* Maximize */
3174          {          {
3175          pp = eptr;          pp = eptr;
3176          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3177            {            {
3178            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3179                {
3180                SCHECK_PARTIAL();
3181                break;
3182                }
3183              if (fc != md->lcc[*eptr]) break;
3184            eptr++;            eptr++;
3185            }            }
3186    
3187            if (possessive) continue;
3188    
3189          while (eptr >= pp)          while (eptr >= pp)
3190            {            {
3191            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3192            eptr--;            eptr--;
3193            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3194            }            }
3195          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3196          }          }
3197        /* Control never gets here */        /* Control never gets here */
3198        }        }
# Line 2048  for (;;) Line 3201  for (;;)
3201    
3202      else      else
3203        {        {
3204        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3205            {
3206            if (eptr >= md->end_subject)
3207              {
3208              SCHECK_PARTIAL();
3209              MRRETURN(MATCH_NOMATCH);
3210              }
3211            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3212            }
3213    
3214        if (min == max) continue;        if (min == max) continue;
3215    
3216        if (minimize)        if (minimize)
3217          {          {
3218          for (fi = min;; fi++)          for (fi = min;; fi++)
3219            {            {
3220            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3221            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3222            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3223              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3224                {
3225                SCHECK_PARTIAL();
3226                MRRETURN(MATCH_NOMATCH);
3227                }
3228              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3229            }            }
3230          /* Control never gets here */          /* Control never gets here */
3231          }          }
3232        else        else  /* Maximize */
3233          {          {
3234          pp = eptr;          pp = eptr;
3235          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3236            {            {
3237            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3238                {
3239                SCHECK_PARTIAL();
3240                break;
3241                }
3242              if (fc != *eptr) break;
3243            eptr++;            eptr++;
3244            }            }
3245            if (possessive) continue;
3246    
3247          while (eptr >= pp)          while (eptr >= pp)
3248            {            {
3249            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3250            eptr--;            eptr--;
3251            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3252            }            }
3253          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3254          }          }
3255        }        }
3256      /* Control never gets here */      /* Control never gets here */
# Line 2084  for (;;) Line 3259  for (;;)
3259      checking can be multibyte. */      checking can be multibyte. */
3260    
3261      case OP_NOT:      case OP_NOT:
3262      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3263        if (eptr >= md->end_subject)
3264          {
3265          SCHECK_PARTIAL();
3266          MRRETURN(MATCH_NOMATCH);
3267          }
3268      ecode++;      ecode++;
3269      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3270      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3271        {        {
3272  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3273        if (c < 256)        if (c < 256)
3274  #endif  #endif
3275        c = md->lcc[c];        c = md->lcc[c];
3276        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3277        }        }
3278      else      else    /* Caseful */
3279        {        {
3280        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3281        }        }
3282      break;      break;
3283    
# Line 2109  for (;;) Line 3289  for (;;)
3289      about... */      about... */
3290    
3291      case OP_NOTEXACT:      case OP_NOTEXACT:
3292        case OP_NOTEXACTI:
3293      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3294      ecode += 3;      ecode += 3;
3295      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3296    
3297      case OP_NOTUPTO:      case OP_NOTUPTO:
3298        case OP_NOTUPTOI:
3299      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3300        case OP_NOTMINUPTOI:
3301        min = 0;
3302        max = GET2(ecode, 1);
3303        minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3304        ecode += 3;
3305        goto REPEATNOTCHAR;
3306    
3307        case OP_NOTPOSSTAR:
3308        case OP_NOTPOSSTARI:
3309        possessive = TRUE;
3310        min = 0;
3311        max = INT_MAX;
3312        ecode++;
3313        goto REPEATNOTCHAR;
3314    
3315        case OP_NOTPOSPLUS:
3316        case OP_NOTPOSPLUSI:
3317        possessive = TRUE;
3318        min = 1;
3319        max = INT_MAX;
3320        ecode++;
3321        goto REPEATNOTCHAR;
3322    
3323        case OP_NOTPOSQUERY:
3324        case OP_NOTPOSQUERYI:
3325        possessive = TRUE;
3326        min = 0;
3327        max = 1;
3328        ecode++;
3329        goto REPEATNOTCHAR;
3330    
3331        case OP_NOTPOSUPTO:
3332        case OP_NOTPOSUPTOI:
3333        possessive = TRUE;
3334      min = 0;      min = 0;
3335      max = GET2(ecode, 1);      max = GET2(ecode, 1);
     minimize = *ecode == OP_NOTMINUPTO;  
3336      ecode += 3;      ecode += 3;
3337      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3338    
3339      case OP_NOTSTAR:      case OP_NOTSTAR:
3340        case OP_NOTSTARI:
3341      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3342        case OP_NOTMINSTARI:
3343      case OP_NOTPLUS:      case OP_NOTPLUS:
3344        case OP_NOTPLUSI:
3345      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3346        case OP_NOTMINPLUSI:
3347      case OP_NOTQUERY:      case OP_NOTQUERY:
3348        case OP_NOTQUERYI:
3349      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3350      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3351        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3352      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3353      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3354      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3355      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3356    
3357      /* 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. */  
3358    
3359      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3360      fc = *ecode++;      fc = *ecode++;
3361    
3362      /* 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 2152  for (;;) Line 3370  for (;;)
3370      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,
3371        max, eptr));        max, eptr));
3372    
3373      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3374        {        {
3375        fc = md->lcc[fc];        fc = md->lcc[fc];
3376    
# Line 2160  for (;;) Line 3378  for (;;)
3378        /* UTF-8 mode */        /* UTF-8 mode */
3379        if (utf8)        if (utf8)
3380          {          {
3381          register int d;          register unsigned int d;
3382          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3383            {            {
3384              if (eptr >= md->end_subject)
3385                {
3386                SCHECK_PARTIAL();
3387                MRRETURN(MATCH_NOMATCH);
3388                }
3389            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3390            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3391            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3392            }            }
3393          }          }
3394        else        else
# Line 2174  for (;;) Line 3397  for (;;)
3397        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3398          {          {
3399          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3400            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3401              if (eptr >= md->end_subject)
3402                {
3403                SCHECK_PARTIAL();
3404                MRRETURN(MATCH_NOMATCH);
3405                }
3406              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3407              }
3408          }          }
3409    
3410        if (min == max) continue;        if (min == max) continue;
# Line 2185  for (;;) Line 3415  for (;;)
3415          /* UTF-8 mode */          /* UTF-8 mode */
3416          if (utf8)          if (utf8)
3417            {            {
3418            register int d;            register unsigned int d;
3419            for (fi = min;; fi++)            for (fi = min;; fi++)
3420              {              {
3421              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3422              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3423                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3424                if (eptr >= md->end_subject)
3425                  {
3426                  SCHECK_PARTIAL();
3427                  MRRETURN(MATCH_NOMATCH);
3428                  }
3429              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3430              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3431              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3432              }              }
3433            }            }
3434          else          else
# Line 2202  for (;;) Line 3437  for (;;)
3437            {            {
3438            for (fi = min;; fi++)            for (fi = min;; fi++)
3439              {              {
3440              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3441              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3442              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3443                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3444                  {
3445                  SCHECK_PARTIAL();
3446                  MRRETURN(MATCH_NOMATCH);
3447                  }
3448                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3449              }              }
3450            }            }
3451          /* Control never gets here */          /* Control never gets here */
# Line 2221  for (;;) Line 3461  for (;;)
3461          /* UTF-8 mode */          /* UTF-8 mode */
3462          if (utf8)          if (utf8)
3463            {            {
3464            register int d;            register unsigned int d;
3465            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3466              {              {
3467              int len = 1;              int len = 1;
3468              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3469                  {
3470                  SCHECK_PARTIAL();
3471                  break;
3472                  }
3473              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3474              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3475              if (fc == d) break;              if (fc == d) break;
3476              eptr += len;              eptr += len;
3477              }              }
3478            for(;;)          if (possessive) continue;
3479            for(;;)
3480              {              {
3481              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3482              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3483              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3484              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2245  for (;;) Line 3490  for (;;)
3490            {            {
3491            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3492              {              {
3493              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3494                  {
3495                  SCHECK_PARTIAL();
3496                  break;
3497                  }
3498                if (fc == md->lcc[*eptr]) break;
3499              eptr++;              eptr++;
3500              }              }
3501              if (possessive) continue;
3502            while (eptr >= pp)            while (eptr >= pp)
3503              {              {
3504              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3505              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3506              eptr--;              eptr--;
3507              }              }
3508            }            }
3509    
3510          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3511          }          }
3512        /* Control never gets here */        /* Control never gets here */
3513        }        }
# Line 2269  for (;;) Line 3520  for (;;)
3520        /* UTF-8 mode */        /* UTF-8 mode */
3521        if (utf8)        if (utf8)
3522          {          {
3523          register int d;          register unsigned int d;
3524          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3525            {            {
3526              if (eptr >= md->end_subject)
3527                {
3528                SCHECK_PARTIAL();
3529                MRRETURN(MATCH_NOMATCH);
3530                }
3531            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3532            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3533            }            }
3534          }          }
3535        else        else
# Line 2281  for (;;) Line 3537  for (;;)
3537        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3538          {          {
3539          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3540            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3541              if (eptr >= md->end_subject)
3542                {
3543                SCHECK_PARTIAL();
3544                MRRETURN(MATCH_NOMATCH);
3545                }
3546              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3547              }
3548          }          }
3549    
3550        if (min == max) continue;        if (min == max) continue;
# Line 2292  for (;;) Line 3555  for (;;)
3555          /* UTF-8 mode */          /* UTF-8 mode */
3556          if (utf8)          if (utf8)
3557            {            {
3558            register int d;            register unsigned int d;
3559            for (fi = min;; fi++)            for (fi = min;; fi++)
3560              {              {
3561              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3562              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3563                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3564                if (eptr >= md->end_subject)
3565                  {
3566                  SCHECK_PARTIAL();
3567                  MRRETURN(MATCH_NOMATCH);
3568                  }
3569              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3570              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3571              }              }
3572            }            }
3573          else          else
# Line 2308  for (;;) Line 3576  for (;;)
3576            {            {
3577            for (fi = min;; fi++)            for (fi = min;; fi++)
3578              {              {
3579              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3580              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3581              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3582                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3583                  {
3584                  SCHECK_PARTIAL();
3585                  MRRETURN(MATCH_NOMATCH);
3586                  }
3587                if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3588              }              }
3589            }            }
3590          /* Control never gets here */          /* Control never gets here */
# Line 2327  for (;;) Line 3600  for (;;)
3600          /* UTF-8 mode */          /* UTF-8 mode */
3601          if (utf8)          if (utf8)
3602            {            {
3603            register int d;            register unsigned int d;
3604            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3605              {              {
3606              int len = 1;              int len = 1;
3607              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3608                  {
3609                  SCHECK_PARTIAL();
3610                  break;
3611                  }
3612              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3613              if (fc == d) break;              if (fc == d) break;
3614              eptr += len;              eptr += len;
3615              }              }
3616              if (possessive) continue;
3617            for(;;)            for(;;)
3618              {              {
3619              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3620              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3621              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3622              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2350  for (;;) Line 3628  for (;;)
3628            {            {
3629            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3630              {              {
3631              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3632                  {
3633                  SCHECK_PARTIAL();
3634                  break;
3635                  }
3636                if (fc == *eptr) break;
3637              eptr++;              eptr++;
3638              }              }
3639              if (possessive) continue;
3640            while (eptr >= pp)            while (eptr >= pp)
3641              {              {
3642              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3643              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3644              eptr--;              eptr--;
3645              }              }
3646            }            }
3647    
3648          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3649          }          }
3650        }        }
3651      /* Control never gets here */      /* Control never gets here */
# Line 2384  for (;;) Line 3668  for (;;)
3668      ecode += 3;      ecode += 3;
3669      goto REPEATTYPE;      goto REPEATTYPE;
3670    
3671        case OP_TYPEPOSSTAR:
3672        possessive = TRUE;
3673        min = 0;
3674        max = INT_MAX;
3675        ecode++;
3676        goto REPEATTYPE;
3677    
3678        case OP_TYPEPOSPLUS:
3679        possessive = TRUE;
3680        min = 1;
3681        max = INT_MAX;
3682        ecode++;
3683        goto REPEATTYPE;
3684    
3685        case OP_TYPEPOSQUERY:
3686        possessive = TRUE;
3687        min = 0;
3688        max = 1;
3689        ecode++;
3690        goto REPEATTYPE;
3691    
3692        case OP_TYPEPOSUPTO:
3693        possessive = TRUE;
3694        min = 0;
3695        max = GET2(ecode, 1);
3696        ecode += 3;
3697        goto REPEATTYPE;
3698    
3699