/[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 473 by ph10, Sat Jan 2 12:40:07 2010 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) 1997-2010 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 */  /* Flag bits for the match() function */
61    
62  #define match_condassert   0x01    /* Called to check a condition assertion */  #define match_condassert     0x01  /* Called to check a condition assertion */
63  #define match_isgroup      0x02    /* Set if start of bracketed group */  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */
64    
65  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
66  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 68  defined PCRE_ERROR_xxx codes, which are
68  #define MATCH_MATCH        1  #define MATCH_MATCH        1
69  #define MATCH_NOMATCH      0  #define MATCH_NOMATCH      0
70    
71    /* Special internal returns from the match() function. Make them sufficiently
72    negative to avoid the external error codes. */
73    
74    #define MATCH_COMMIT       (-999)
75    #define MATCH_PRUNE        (-998)
76    #define MATCH_SKIP         (-997)
77    #define MATCH_THEN         (-996)
78    
79  /* 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.
80  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,
81  because the offset vector is always a multiple of 3 long. */  because the offset vector is always a multiple of 3 long. */
# Line 101  Returns:     nothing Line 109  Returns:     nothing
109  static void  static void
110  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)
111  {  {
112  int c;  unsigned int c;
113  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;
114  while (length-- > 0)  while (length-- > 0)
115    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 128  Returns:      TRUE if matched Line 136  Returns:      TRUE if matched
136  */  */
137    
138  static BOOL  static BOOL
139  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,
140    unsigned long int ims)    unsigned long int ims)
141  {  {
142  const uschar *p = md->start_subject + md->offset_vector[offset];  USPTR p = md->start_subject + md->offset_vector[offset];
143    
144  #ifdef DEBUG  #ifdef DEBUG
145  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 150  printf("\n"); Line 158  printf("\n");
158    
159  if (length > md->end_subject - eptr) return FALSE;  if (length > md->end_subject - eptr) return FALSE;
160    
161  /* Separate the caselesss case for speed */  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
162    properly if Unicode properties are supported. Otherwise, we can check only
163    ASCII characters. */
164    
165  if ((ims & PCRE_CASELESS) != 0)  if ((ims & PCRE_CASELESS) != 0)
166    {    {
167    #ifdef SUPPORT_UTF8
168    #ifdef SUPPORT_UCP
169      if (md->utf8)
170        {
171        USPTR endptr = eptr + length;
172        while (eptr < endptr)
173          {
174          int c, d;
175          GETCHARINC(c, eptr);
176          GETCHARINC(d, p);
177          if (c != d && c != UCD_OTHERCASE(d)) return FALSE;
178          }
179        }
180      else
181    #endif
182    #endif
183    
184      /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
185      is no UCP support. */
186    
187    while (length-- > 0)    while (length-- > 0)
188      if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE;      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }
189    }    }
190    
191    /* In the caseful case, we can just compare the bytes, whether or not we
192    are in UTF-8 mode. */
193    
194  else  else
195    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }
196    
# Line 169  return TRUE; Line 203  return TRUE;
203  ****************************************************************************  ****************************************************************************
204                     RECURSION IN THE match() FUNCTION                     RECURSION IN THE match() FUNCTION
205    
206  The match() function is highly recursive. Some regular expressions can cause  The match() function is highly recursive, though not every recursive call
207  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
208  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
209  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
210  works fine.  saved for a recursive call. On Unix, the stack can be large, and this works
211    fine.
212  It turns out that on non-Unix systems there are problems with programs that  
213  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
214  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
215  for decades.) So....  has oodles of memory these days, and techniques for extending the stack have
216    been known for decades.) So....
217    
218  There is a fudge, triggered by defining NO_RECURSE, which avoids recursive  There is a fudge, triggered by defining NO_RECURSE, which avoids recursive
219  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
220  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
221  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
222  always used to.  always used to.
223    
224    The original heap-recursive code used longjmp(). However, it seems that this
225    can be very slow on some operating systems. Following a suggestion from Stan
226    Switzer, the use of longjmp() has been abolished, at the cost of having to
227    provide a unique number for each call to RMATCH. There is no way of generating
228    a sequence of numbers at compile time in C. I have given them names, to make
229    them stand out more clearly.
230    
231    Crude tests on x86 Linux show a small speedup of around 5-8%. However, on
232    FreeBSD, avoiding longjmp() more than halves the time taken to run the standard
233    tests. Furthermore, not using longjmp() means that local dynamic variables
234    don't have indeterminate values; this has meant that the frame size can be
235    reduced because the result can be "passed back" by straight setting of the
236    variable instead of being passed in the frame.
237  ****************************************************************************  ****************************************************************************
238  ***************************************************************************/  ***************************************************************************/
239    
240    /* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN
241    below must be updated in sync.  */
242    
243  /* These versions of the macros use the stack, as normal */  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,
244           RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,
245           RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
246           RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
247           RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
248           RM51,  RM52, RM53, RM54 };
249    
250    /* These versions of the macros use the stack, as normal. There are debugging
251    versions and production versions. Note that the "rw" argument of RMATCH isn't
252    actuall used in this definition. */
253    
254  #ifndef NO_RECURSE  #ifndef NO_RECURSE
255  #define REGISTER register  #define REGISTER register
256  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) rx = match(ra,rb,rc,rd,re,rf,rg)  
257    #ifdef DEBUG
258    #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
259      { \
260      printf("match() called in line %d\n", __LINE__); \
261      rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \
262      printf("to line %d\n", __LINE__); \
263      }
264    #define RRETURN(ra) \
265      { \
266      printf("match() returned %d from line %d ", ra, __LINE__); \
267      return ra; \
268      }
269    #else
270    #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
271      rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1)
272  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
273    #endif
274    
275  #else  #else
276    
277    
278  /* 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
279  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
280  match(), which never changes. */  argument of match(), which never changes. */
281    
282  #define REGISTER  #define REGISTER
283    
284  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg)\  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\
285    {\    {\
286    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\
287    if (setjmp(frame->Xwhere) == 0)\    frame->Xwhere = rw; \
288      {\    newframe->Xeptr = ra;\
289      newframe->Xeptr = ra;\    newframe->Xecode = rb;\
290      newframe->Xecode = rb;\    newframe->Xmstart = mstart;\
291      newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
292      newframe->Xims = re;\    newframe->Xims = re;\
293      newframe->Xeptrb = rf;\    newframe->Xeptrb = rf;\
294      newframe->Xflags = rg;\    newframe->Xflags = rg;\
295      newframe->Xprevframe = frame;\    newframe->Xrdepth = frame->Xrdepth + 1;\
296      frame = newframe;\    newframe->Xprevframe = frame;\
297      DPRINTF(("restarting from line %d\n", __LINE__));\    frame = newframe;\
298      goto HEAP_RECURSE;\    DPRINTF(("restarting from line %d\n", __LINE__));\
299      }\    goto HEAP_RECURSE;\
300    else\    L_##rw:\
301      {\    DPRINTF(("jumped back to line %d\n", __LINE__));\
     DPRINTF(("longjumped back to line %d\n", __LINE__));\  
     frame = md->thisframe;\  
     rx = frame->Xresult;\  
     }\  
302    }    }
303    
304  #define RRETURN(ra)\  #define RRETURN(ra)\
# Line 235  match(), which never changes. */ Line 308  match(), which never changes. */
308    (pcre_stack_free)(newframe);\    (pcre_stack_free)(newframe);\
309    if (frame != NULL)\    if (frame != NULL)\
310      {\      {\
311      frame->Xresult = ra;\      rrc = ra;\
312      md->thisframe = frame;\      goto HEAP_RETURN;\
     longjmp(frame->Xwhere, 1);\  
313      }\      }\
314    return ra;\    return ra;\
315    }    }
# Line 250  typedef struct heapframe { Line 322  typedef struct heapframe {
322    
323    /* Function arguments that may change */    /* Function arguments that may change */
324    
325    const uschar *Xeptr;    USPTR Xeptr;
326    const uschar *Xecode;    const uschar *Xecode;
327      USPTR Xmstart;
328    int Xoffset_top;    int Xoffset_top;
329    long int Xims;    long int Xims;
330    eptrblock *Xeptrb;    eptrblock *Xeptrb;
331    int Xflags;    int Xflags;
332      unsigned int Xrdepth;
333    
334    /* Function local variables */    /* Function local variables */
335    
336    const uschar *Xcallpat;    USPTR Xcallpat;
337    const uschar *Xcharptr;  #ifdef SUPPORT_UTF8
338    const uschar *Xdata;    USPTR Xcharptr;
339    const uschar *Xnext;  #endif
340    const uschar *Xpp;    USPTR Xdata;
341    const uschar *Xprev;    USPTR Xnext;
342    const uschar *Xsaved_eptr;    USPTR Xpp;
343      USPTR Xprev;
344      USPTR Xsaved_eptr;
345    
346    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
347    
348    BOOL Xcur_is_word;    BOOL Xcur_is_word;
349    BOOL Xcondition;    BOOL Xcondition;
   BOOL Xminimize;  
350    BOOL Xprev_is_word;    BOOL Xprev_is_word;
351    
352    unsigned long int Xoriginal_ims;    unsigned long int Xoriginal_ims;
353    
354  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
355    int Xprop_type;    int Xprop_type;
356      int Xprop_value;
357    int Xprop_fail_result;    int Xprop_fail_result;
358    int Xprop_category;    int Xprop_category;
359    int Xprop_chartype;    int Xprop_chartype;
360    int Xprop_othercase;    int Xprop_script;
361    int Xprop_test_against;    int Xoclength;
362    int *Xprop_test_variable;    uschar Xocchars[8];
363  #endif  #endif
364    
365      int Xcodelink;
366    int Xctype;    int Xctype;
367    int Xfc;    unsigned int Xfc;
368    int Xfi;    int Xfi;
369    int Xlength;    int Xlength;
370    int Xmax;    int Xmax;
# Line 301  typedef struct heapframe { Line 378  typedef struct heapframe {
378    
379    eptrblock Xnewptrb;    eptrblock Xnewptrb;
380    
381    /* Place to pass back result, and where to jump back to */    /* Where to jump back to */
382    
383    int  Xresult;    int Xwhere;
   jmp_buf Xwhere;  
384    
385  } heapframe;  } heapframe;
386    
# Line 320  typedef struct heapframe { Line 396  typedef struct heapframe {
396  *         Match from current position            *  *         Match from current position            *
397  *************************************************/  *************************************************/
398    
399  /* 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  
400  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
401  same response.  same response. */
402    
403    /* These macros pack up tests that are used for partial matching, and which
404    appears several times in the code. We set the "hit end" flag if the pointer is
405    at the end of the subject and also past the start of the subject (i.e.
406    something has been matched). For hard partial matching, we then return
407    immediately. The second one is used when we already know we are past the end of
408    the subject. */
409    
410    #define CHECK_PARTIAL()\
411      if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\
412        {\
413        md->hitend = TRUE;\
414        if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\
415        }
416    
417    #define SCHECK_PARTIAL()\
418      if (md->partial != 0 && eptr > mstart)\
419        {\
420        md->hitend = TRUE;\
421        if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\
422        }
423    
424  Performance note: It might be tempting to extract commonly used fields from the  
425  md structure (e.g. utf8, end_subject) into individual variables to improve  /* Performance note: It might be tempting to extract commonly used fields from
426    the md structure (e.g. utf8, end_subject) into individual variables to improve
427  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
428  made performance worse.  made performance worse.
429    
430  Arguments:  Arguments:
431     eptr        pointer in subject     eptr        pointer to current character in subject
432     ecode       position in code     ecode       pointer to current position in compiled code
433       mstart      pointer to the current match start position (can be modified
434                     by encountering \K)
435     offset_top  current top pointer     offset_top  current top pointer
436     md          pointer to "static" info for the match     md          pointer to "static" info for the match
437     ims         current /i, /m, and /s options     ims         current /i, /m, and /s options
# Line 342  Arguments: Line 439  Arguments:
439                   brackets - for testing for empty matches                   brackets - for testing for empty matches
440     flags       can contain     flags       can contain
441                   match_condassert - this is an assertion condition                   match_condassert - this is an assertion condition
442                   match_isgroup - this is the start of a bracketed group                   match_cbegroup - this is the start of an unlimited repeat
443                       group that can match an empty string
444       rdepth      the recursion depth
445    
446  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
447                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
448                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
449                   (e.g. stopped by recursion limit)                   (e.g. stopped by repeated call or recursion limit)
450  */  */
451    
452  static int  static int
453  match(REGISTER const uschar *eptr, REGISTER const uschar *ecode,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
454    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,
455    int flags)    int flags, unsigned int rdepth)
456  {  {
457  /* 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,
458  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
459  because they are used a lot in loops. */  "register" because they are used a lot in loops. */
460    
461    register int  rrc;         /* Returns from recursive calls */
462    register int  i;           /* Used for loops not involving calls to RMATCH() */
463    register unsigned int c;   /* Character values not kept over RMATCH() calls */
464    register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
465    
466  register int  rrc;    /* Returns from recursive calls */  BOOL minimize, possessive; /* Quantifier options */
467  register int  i;      /* Used for loops not involving calls to RMATCH() */  int condcode;
 register int  c;      /* Character values not kept over RMATCH() calls */  
 register BOOL utf8;   /* Local copy of UTF-8 flag for speed */  
468    
469  /* 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
470  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 377  frame->Xprevframe = NULL;            /* Line 479  frame->Xprevframe = NULL;            /*
479    
480  frame->Xeptr = eptr;  frame->Xeptr = eptr;
481  frame->Xecode = ecode;  frame->Xecode = ecode;
482    frame->Xmstart = mstart;
483  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
484  frame->Xims = ims;  frame->Xims = ims;
485  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
486  frame->Xflags = flags;  frame->Xflags = flags;
487    frame->Xrdepth = rdepth;
488    
489  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
490    
# Line 390  HEAP_RECURSE: Line 494  HEAP_RECURSE:
494    
495  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
496  #define ecode              frame->Xecode  #define ecode              frame->Xecode
497    #define mstart             frame->Xmstart
498  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
499  #define ims                frame->Xims  #define ims                frame->Xims
500  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
501  #define flags              frame->Xflags  #define flags              frame->Xflags
502    #define rdepth             frame->Xrdepth
503    
504  /* Ditto for the local variables */  /* Ditto for the local variables */
505    
# Line 401  HEAP_RECURSE: Line 507  HEAP_RECURSE:
507  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
508  #endif  #endif
509  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
510    #define codelink           frame->Xcodelink
511  #define data               frame->Xdata  #define data               frame->Xdata
512  #define next               frame->Xnext  #define next               frame->Xnext
513  #define pp                 frame->Xpp  #define pp                 frame->Xpp
# Line 411  HEAP_RECURSE: Line 518  HEAP_RECURSE:
518    
519  #define cur_is_word        frame->Xcur_is_word  #define cur_is_word        frame->Xcur_is_word
520  #define condition          frame->Xcondition  #define condition          frame->Xcondition
 #define minimize           frame->Xminimize  
521  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
522    
523  #define original_ims       frame->Xoriginal_ims  #define original_ims       frame->Xoriginal_ims
524    
525  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
526  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
527    #define prop_value         frame->Xprop_value
528  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
529  #define prop_category      frame->Xprop_category  #define prop_category      frame->Xprop_category
530  #define prop_chartype      frame->Xprop_chartype  #define prop_chartype      frame->Xprop_chartype
531  #define prop_othercase     frame->Xprop_othercase  #define prop_script        frame->Xprop_script
532  #define prop_test_against  frame->Xprop_test_against  #define oclength           frame->Xoclength
533  #define prop_test_variable frame->Xprop_test_variable  #define occhars            frame->Xocchars
534  #endif  #endif
535    
536  #define ctype              frame->Xctype  #define ctype              frame->Xctype
# Line 447  HEAP_RECURSE: Line 554  HEAP_RECURSE:
554  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
555  i, and fc and c, can be the same variables. */  i, and fc and c, can be the same variables. */
556    
557  #else  #else         /* NO_RECURSE not defined */
558  #define fi i  #define fi i
559  #define fc c  #define fc c
560    
561    
562  #ifdef SUPPORT_UTF8                /* Many of these variables are used ony */  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */
563  const uschar *charptr;             /* small blocks of the code. My normal  */  const uschar *charptr;             /* in small blocks of the code. My normal */
564  #endif                             /* style of coding would have declared  */  #endif                             /* style of coding would have declared    */
565  const uschar *callpat;             /* them within each of those blocks.    */  const uschar *callpat;             /* them within each of those blocks.      */
566  const uschar *data;                /* However, in order to accommodate the */  const uschar *data;                /* However, in order to accommodate the   */
567  const uschar *next;                /* version of this code that uses an    */  const uschar *next;                /* version of this code that uses an      */
568  const uschar *pp;                  /* external "stack" implemented on the  */  USPTR         pp;                  /* external "stack" implemented on the    */
569  const uschar *prev;                /* heap, it is easier to declare them   */  const uschar *prev;                /* heap, it is easier to declare them all */
570  const uschar *saved_eptr;          /* all here, so the declarations can    */  USPTR         saved_eptr;          /* here, so the declarations can be cut   */
571                                     /* be cut out in a block. The only      */                                     /* out in a block. The only declarations  */
572  recursion_info new_recursive;      /* declarations within blocks below are */  recursion_info new_recursive;      /* within blocks below are for variables  */
573                                     /* for variables that do not have to    */                                     /* that do not have to be preserved over  */
574  BOOL cur_is_word;                  /* be preserved over a recursive call   */  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */
575  BOOL condition;                    /* to RMATCH().                         */  BOOL condition;
 BOOL minimize;  
576  BOOL prev_is_word;  BOOL prev_is_word;
577    
578  unsigned long int original_ims;  unsigned long int original_ims;
579    
580  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
581  int prop_type;  int prop_type;
582    int prop_value;
583  int prop_fail_result;  int prop_fail_result;
584  int prop_category;  int prop_category;
585  int prop_chartype;  int prop_chartype;
586  int prop_othercase;  int prop_script;
587  int prop_test_against;  int oclength;
588  int *prop_test_variable;  uschar occhars[8];
589  #endif  #endif
590    
591    int codelink;
592  int ctype;  int ctype;
593  int length;  int length;
594  int max;  int max;
# Line 493  int save_offset1, save_offset2, save_off Line 601  int save_offset1, save_offset2, save_off
601  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
602    
603  eptrblock newptrb;  eptrblock newptrb;
604  #endif  #endif     /* NO_RECURSE */
605    
606  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
607  variables. */  variables. */
608    
609  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
610    prop_value = 0;
611  prop_fail_result = 0;  prop_fail_result = 0;
 prop_test_against = 0;  
 prop_test_variable = NULL;  
612  #endif  #endif
613    
614  /* OK, now we can get on with the real code of the function. Recursion is  
615  specified by the macros RMATCH and RRETURN. When NO_RECURSE is *not* defined,  /* This label is used for tail recursion, which is used in a few cases even
616  these just turn into a recursive call to match() and a "return", respectively.  when NO_RECURSE is not defined, in order to reduce the amount of stack that is
617  However, RMATCH isn't like a function call because it's quite a complicated  used. Thanks to Ian Taylor for noticing this possibility and sending the
618  macro. It has to be used in one particular way. This shouldn't, however, impact  original patch. */
619  performance when true recursion is being used. */  
620    TAIL_RECURSE:
621    
622    /* OK, now we can get on with the real code of the function. Recursive calls
623    are specified by the macro RMATCH and RRETURN is used to return. When
624    NO_RECURSE is *not* defined, these just turn into a recursive call to match()
625    and a "return", respectively (possibly with some debugging if DEBUG is
626    defined). However, RMATCH isn't like a function call because it's quite a
627    complicated macro. It has to be used in one particular way. This shouldn't,
628    however, impact performance when true recursion is being used. */
629    
630    #ifdef SUPPORT_UTF8
631    utf8 = md->utf8;       /* Local copy of the flag */
632    #else
633    utf8 = FALSE;
634    #endif
635    
636    /* First check that we haven't called match() too many times, or that we
637    haven't exceeded the recursive call limit. */
638    
639  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
640    if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
641    
642  original_ims = ims;    /* Save for resetting on ')' */  original_ims = ims;    /* Save for resetting on ')' */
 utf8 = md->utf8;       /* Local copy of the flag */  
643    
644  /* 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
645  stack of such pointers, to be re-instated at the end of the group when we hit  string, the match_cbegroup flag is set. When this is the case, add the current
646  the closing ket. When match() is called in other circumstances, we don't add to  subject pointer to the chain of such remembered pointers, to be checked when we
647  this stack. */  hit the closing ket, in order to break infinite loops that match no characters.
648    When match() is called in other circumstances, don't add to the chain. The
649    match_cbegroup flag must NOT be used with tail recursion, because the memory
650    block that is used is on the stack, so a new one may be required for each
651    match(). */
652    
653  if ((flags & match_isgroup) != 0)  if ((flags & match_cbegroup) != 0)
654    {    {
   newptrb.epb_prev = eptrb;  
655    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
656      newptrb.epb_prev = eptrb;
657    eptrb = &newptrb;    eptrb = &newptrb;
658    }    }
659    
660  /* Now start processing the operations. */  /* Now start processing the opcodes. */
661    
662  for (;;)  for (;;)
663    {    {
664      minimize = possessive = FALSE;
665    op = *ecode;    op = *ecode;
   minimize = FALSE;  
   
   /* For partial matching, remember if we ever hit the end of the subject after  
   matching at least one subject character. */  
   
   if (md->partial &&  
       eptr >= md->end_subject &&  
       eptr > md->start_match)  
     md->hitend = TRUE;  
   
   /* 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. */  
666    
667    if (op > OP_BRA)    switch(op)
668      {      {
669      number = op - OP_BRA;      case OP_FAIL:
670        RRETURN(MATCH_NOMATCH);
     /* For extended extraction brackets (large number), we have to fish out the  
     number from a dummy opcode at the start. */  
671    
672      if (number > EXTRACT_BASIC_MAX)      case OP_PRUNE:
673        number = GET2(ecode, 2+LINK_SIZE);      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
674          ims, eptrb, flags, RM51);
675        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
676        RRETURN(MATCH_PRUNE);
677    
678        case OP_COMMIT:
679        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
680          ims, eptrb, flags, RM52);
681        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
682        RRETURN(MATCH_COMMIT);
683    
684        case OP_SKIP:
685        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
686          ims, eptrb, flags, RM53);
687        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
688        md->start_match_ptr = eptr;   /* Pass back current position */
689        RRETURN(MATCH_SKIP);
690    
691        case OP_THEN:
692        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
693          ims, eptrb, flags, RM54);
694        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
695        RRETURN(MATCH_THEN);
696    
697        /* Handle a capturing bracket. If there is space in the offset vector, save
698        the current subject position in the working slot at the top of the vector.
699        We mustn't change the current values of the data slot, because they may be
700        set from a previous iteration of this group, and be referred to by a
701        reference inside the group.
702    
703        If the bracket fails to match, we need to restore this value and also the
704        values of the final offsets, in case they were set by a previous iteration
705        of the same bracket.
706    
707        If there isn't enough space in the offset vector, treat this as if it were
708        a non-capturing bracket. Don't worry about setting the flag for the error
709        case here; that is handled in the code for KET. */
710    
711        case OP_CBRA:
712        case OP_SCBRA:
713        number = GET2(ecode, 1+LINK_SIZE);
714      offset = number << 1;      offset = number << 1;
715    
716  #ifdef DEBUG  #ifdef DEBUG
717      printf("start bracket %d subject=", number);      printf("start bracket %d\n", number);
718        printf("subject=");
719      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
720      printf("\n");      printf("\n");
721  #endif  #endif
# Line 584  for (;;) Line 730  for (;;)
730        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
731        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;
732    
733          flags = (op == OP_SCBRA)? match_cbegroup : 0;
734        do        do
735          {          {
736          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
737            match_isgroup);            ims, eptrb, flags, RM1);
738          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
739          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
740          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
741          }          }
# Line 603  for (;;) Line 750  for (;;)
750        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
751        }        }
752    
753      /* Insufficient room for saving captured contents */      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
754        as a non-capturing bracket. */
755    
756      else op = OP_BRA;      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
757      }      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
758    
759    /* Other types of node can be handled by a switch */      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
760    
761    switch(op)      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
762      {      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
763      case OP_BRA:     /* Non-capturing bracket: optimized */  
764      DPRINTF(("start bracket 0\n"));      /* Non-capturing bracket. Loop for all the alternatives. When we get to the
765      do      final alternative within the brackets, we would return the result of a
766        recursive call to match() whatever happened. We can reduce stack usage by
767        turning this into a tail recursion, except in the case when match_cbegroup
768        is set.*/
769    
770        case OP_BRA:
771        case OP_SBRA:
772        DPRINTF(("start non-capturing bracket\n"));
773        flags = (op >= OP_SBRA)? match_cbegroup : 0;
774        for (;;)
775        {        {
776        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */
777          match_isgroup);          {
778        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (flags == 0)    /* Not a possibly empty group */
779              {
780              ecode += _pcre_OP_lengths[*ecode];
781              DPRINTF(("bracket 0 tail recursion\n"));
782              goto TAIL_RECURSE;
783              }
784    
785            /* Possibly empty group; can't use tail recursion. */
786    
787            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,
788              eptrb, flags, RM48);
789            RRETURN(rrc);
790            }
791    
792          /* For non-final alternatives, continue the loop for a NOMATCH result;
793          otherwise return. */
794    
795          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,
796            eptrb, flags, RM2);
797          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
798        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
799        }        }
800      while (*ecode == OP_ALT);      /* Control never reaches here. */
     DPRINTF(("bracket 0 failed\n"));  
     RRETURN(MATCH_NOMATCH);  
801    
802      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
803      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
804      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
805      exactly what going to the ket would do. */      exactly what going to the ket would do. As there is only one branch to be
806        obeyed, we can use tail recursion to avoid using another stack frame. */
807    
808      case OP_COND:      case OP_COND:
809      if (ecode[LINK_SIZE+1] == OP_CREF) /* Condition extract or recurse test */      case OP_SCOND:
810        codelink= GET(ecode, 1);
811    
812        /* Because of the way auto-callout works during compile, a callout item is
813        inserted between OP_COND and an assertion condition. */
814    
815        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
816          {
817          if (pcre_callout != NULL)
818            {
819            pcre_callout_block cb;
820            cb.version          = 1;   /* Version 1 of the callout block */
821            cb.callout_number   = ecode[LINK_SIZE+2];
822            cb.offset_vector    = md->offset_vector;
823            cb.subject          = (PCRE_SPTR)md->start_subject;
824            cb.subject_length   = md->end_subject - md->start_subject;
825            cb.start_match      = mstart - md->start_subject;
826            cb.current_position = eptr - md->start_subject;
827            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
828            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
829            cb.capture_top      = offset_top/2;
830            cb.capture_last     = md->capture_last;
831            cb.callout_data     = md->callout_data;
832            if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);
833            if (rrc < 0) RRETURN(rrc);
834            }
835          ecode += _pcre_OP_lengths[OP_CALLOUT];
836          }
837    
838        condcode = ecode[LINK_SIZE+1];
839    
840        /* Now see what the actual condition is */
841    
842        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
843          {
844          if (md->recursive == NULL)                /* Not recursing => FALSE */
845            {
846            condition = FALSE;
847            ecode += GET(ecode, 1);
848            }
849          else
850            {
851            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
852            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
853    
854            /* If the test is for recursion into a specific subpattern, and it is
855            false, but the test was set up by name, scan the table to see if the
856            name refers to any other numbers, and test them. The condition is true
857            if any one is set. */
858    
859            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
860              {
861              uschar *slotA = md->name_table;
862              for (i = 0; i < md->name_count; i++)
863                {
864                if (GET2(slotA, 0) == recno) break;
865                slotA += md->name_entry_size;
866                }
867    
868              /* Found a name for the number - there can be only one; duplicate
869              names for different numbers are allowed, but not vice versa. First
870              scan down for duplicates. */
871    
872              if (i < md->name_count)
873                {
874                uschar *slotB = slotA;
875                while (slotB > md->name_table)
876                  {
877                  slotB -= md->name_entry_size;
878                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
879                    {
880                    condition = GET2(slotB, 0) == md->recursive->group_num;
881                    if (condition) break;
882                    }
883                  else break;
884                  }
885    
886                /* Scan up for duplicates */
887    
888                if (!condition)
889                  {
890                  slotB = slotA;
891                  for (i++; i < md->name_count; i++)
892                    {
893                    slotB += md->name_entry_size;
894                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
895                      {
896                      condition = GET2(slotB, 0) == md->recursive->group_num;
897                      if (condition) break;
898                      }
899                    else break;
900                    }
901                  }
902                }
903              }
904    
905            /* Chose branch according to the condition */
906    
907            ecode += condition? 3 : GET(ecode, 1);
908            }
909          }
910    
911        else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
912        {        {
913        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
914        condition = (offset == CREF_RECURSE * 2)?        condition = offset < offset_top && md->offset_vector[offset] >= 0;
915          (md->recursive != NULL) :  
916          (offset < offset_top && md->offset_vector[offset] >= 0);        /* If the numbered capture is unset, but the reference was by name,
917        RMATCH(rrc, eptr, ecode + (condition?        scan the table to see if the name refers to any other numbers, and test
918          (LINK_SIZE + 4) : (LINK_SIZE + 1 + GET(ecode, 1))),        them. The condition is true if any one is set. This is tediously similar
919          offset_top, md, ims, eptrb, match_isgroup);        to the code above, but not close enough to try to amalgamate. */
920        RRETURN(rrc);  
921          if (!condition && condcode == OP_NCREF)
922            {
923            int refno = offset >> 1;
924            uschar *slotA = md->name_table;
925    
926            for (i = 0; i < md->name_count; i++)
927              {
928              if (GET2(slotA, 0) == refno) break;
929              slotA += md->name_entry_size;
930              }
931    
932            /* Found a name for the number - there can be only one; duplicate names
933            for different numbers are allowed, but not vice versa. First scan down
934            for duplicates. */
935    
936            if (i < md->name_count)
937              {
938              uschar *slotB = slotA;
939              while (slotB > md->name_table)
940                {
941                slotB -= md->name_entry_size;
942                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
943                  {
944                  offset = GET2(slotB, 0) << 1;
945                  condition = offset < offset_top &&
946                    md->offset_vector[offset] >= 0;
947                  if (condition) break;
948                  }
949                else break;
950                }
951    
952              /* Scan up for duplicates */
953    
954              if (!condition)
955                {
956                slotB = slotA;
957                for (i++; i < md->name_count; i++)
958                  {
959                  slotB += md->name_entry_size;
960                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
961                    {
962                    offset = GET2(slotB, 0) << 1;
963                    condition = offset < offset_top &&
964                      md->offset_vector[offset] >= 0;
965                    if (condition) break;
966                    }
967                  else break;
968                  }
969                }
970              }
971            }
972    
973          /* Chose branch according to the condition */
974    
975          ecode += condition? 3 : GET(ecode, 1);
976          }
977    
978        else if (condcode == OP_DEF)     /* DEFINE - always false */
979          {
980          condition = FALSE;
981          ecode += GET(ecode, 1);
982        }        }
983    
984      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
985      the final argument TRUE causes it to stop at the end of an assertion. */      the final argument match_condassert causes it to stop at the end of an
986        assertion. */
987    
988      else      else
989        {        {
990        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,
991            match_condassert | match_isgroup);            match_condassert, RM3);
992        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
993          {          {
994          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE+2);          condition = TRUE;
995            ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
996          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
997          }          }
998        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
999          {          {
1000          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1001          }          }
1002        else ecode += GET(ecode, 1);        else
1003        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,          {
1004          match_isgroup);          condition = FALSE;
1005        RRETURN(rrc);          ecode += codelink;
1006            }
1007        }        }
     /* Control never reaches here */  
1008    
1009      /* 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,
1010      encountered. */      we can use tail recursion to avoid using another stack frame, except when
1011        match_cbegroup is required for an unlimited repeat of a possibly empty
1012        group. If the second alternative doesn't exist, we can just plough on. */
1013    
1014        if (condition || *ecode == OP_ALT)
1015          {
1016          ecode += 1 + LINK_SIZE;
1017          if (op == OP_SCOND)        /* Possibly empty group */
1018            {
1019            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);
1020            RRETURN(rrc);
1021            }
1022          else                       /* Group must match something */
1023            {
1024            flags = 0;
1025            goto TAIL_RECURSE;
1026            }
1027          }
1028        else                         /* Condition false & no alternative */
1029          {
1030          ecode += 1 + LINK_SIZE;
1031          }
1032        break;
1033    
1034    
1035        /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1036        to close any currently open capturing brackets. */
1037    
1038        case OP_CLOSE:
1039        number = GET2(ecode, 1);
1040        offset = number << 1;
1041    
1042    #ifdef DEBUG
1043          printf("end bracket %d at *ACCEPT", number);
1044          printf("\n");
1045    #endif
1046    
1047      case OP_CREF:      md->capture_last = number;
1048      case OP_BRANUMBER:      if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1049          {
1050          md->offset_vector[offset] =
1051            md->offset_vector[md->offset_end - number];
1052          md->offset_vector[offset+1] = eptr - md->start_subject;
1053          if (offset_top <= offset) offset_top = offset + 2;
1054          }
1055      ecode += 3;      ecode += 3;
1056      break;      break;
1057    
     /* End of the pattern. If we are in a recursion, we should restore the  
     offsets appropriately and continue from after the call. */  
1058    
1059        /* End of the pattern, either real or forced. If we are in a top-level
1060        recursion, we should restore the offsets appropriately and continue from
1061        after the call. */
1062    
1063        case OP_ACCEPT:
1064      case OP_END:      case OP_END:
1065      if (md->recursive != NULL && md->recursive->group_num == 0)      if (md->recursive != NULL && md->recursive->group_num == 0)
1066        {        {
1067        recursion_info *rec = md->recursive;        recursion_info *rec = md->recursive;
1068        DPRINTF(("Hit the end in a (?0) recursion\n"));        DPRINTF(("End of pattern in a (?0) recursion\n"));
1069        md->recursive = rec->prevrec;        md->recursive = rec->prevrec;
1070        memmove(md->offset_vector, rec->offset_save,        memmove(md->offset_vector, rec->offset_save,
1071          rec->saved_max * sizeof(int));          rec->saved_max * sizeof(int));
1072        md->start_match = rec->save_start;        offset_top = rec->save_offset_top;
1073          mstart = rec->save_start;
1074        ims = original_ims;        ims = original_ims;
1075        ecode = rec->after_call;        ecode = rec->after_call;
1076        break;        break;
1077        }        }
1078    
1079      /* 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
1080      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
1081        the subject. In both cases, backtracking will then try other alternatives,
1082        if any. */
1083    
1084        if (eptr == mstart &&
1085            (md->notempty ||
1086              (md->notempty_atstart &&
1087                mstart == md->start_subject + md->start_offset)))
1088          RRETURN(MATCH_NOMATCH);
1089    
1090        /* Otherwise, we have a match. */
1091    
1092      if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH);      md->end_match_ptr = eptr;           /* Record where we ended */
1093      md->end_match_ptr = eptr;          /* Record where we ended */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1094      md->end_offset_top = offset_top;   /* and how many extracts were taken */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1095      RRETURN(MATCH_MATCH);      RRETURN(MATCH_MATCH);
1096    
1097      /* Change option settings */      /* Change option settings */
# Line 717  for (;;) Line 1112  for (;;)
1112      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1113      do      do
1114        {        {
1115        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
1116          match_isgroup);          RM4);
1117        if (rrc == MATCH_MATCH) break;        if (rrc == MATCH_MATCH) break;
1118        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1119        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1120        }        }
1121      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
# Line 738  for (;;) Line 1133  for (;;)
1133      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1134      continue;      continue;
1135    
1136      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1137        PRUNE, or COMMIT means we must assume failure without checking subsequent
1138        branches. */
1139    
1140      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1141      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1142      do      do
1143        {        {
1144        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
1145          match_isgroup);          RM5);
1146        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);
1147        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1148            {
1149            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1150            break;
1151            }
1152          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1153        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1154        }        }
1155      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
# Line 766  for (;;) Line 1168  for (;;)
1168  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1169      if (utf8)      if (utf8)
1170        {        {
1171        c = GET(ecode,1);        i = GET(ecode, 1);
1172        for (i = 0; i < c; i++)        while (i-- > 0)
1173          {          {
1174          eptr--;          eptr--;
1175          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1176          BACKCHAR(eptr)          BACKCHAR(eptr);
1177          }          }
1178        }        }
1179      else      else
# Line 780  for (;;) Line 1182  for (;;)
1182      /* 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 */
1183    
1184        {        {
1185        eptr -= GET(ecode,1);        eptr -= GET(ecode, 1);
1186        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1187        }        }
1188    
1189      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1190    
1191        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1192      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1193      break;      break;
1194    
# Line 800  for (;;) Line 1203  for (;;)
1203        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 1;   /* Version 1 of the callout block */
1204        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1205        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1206        cb.subject          = (const char *)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1207        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = md->end_subject - md->start_subject;
1208        cb.start_match      = md->start_match - md->start_subject;        cb.start_match      = mstart - md->start_subject;
1209        cb.current_position = eptr - md->start_subject;        cb.current_position = eptr - md->start_subject;
1210        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1211        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
# Line 837  for (;;) Line 1240  for (;;)
1240      case OP_RECURSE:      case OP_RECURSE:
1241        {        {
1242        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1243        new_recursive.group_num = *callpat - OP_BRA;        new_recursive.group_num = (callpat == md->start_code)? 0 :
1244            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);  
1245    
1246        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1247    
# Line 869  for (;;) Line 1267  for (;;)
1267    
1268        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1269              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1270        new_recursive.save_start = md->start_match;        new_recursive.save_start = mstart;
1271        md->start_match = eptr;        new_recursive.save_offset_top = offset_top;
1272          mstart = eptr;
1273    
1274        /* 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
1275        restore the offset and recursion data. */        restore the offset and recursion data. */
1276    
1277        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1278          flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;
1279        do        do
1280          {          {
1281          RMATCH(rrc, eptr, callpat + 1 + LINK_SIZE, offset_top, md, ims,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1282              eptrb, match_isgroup);            md, ims, eptrb, flags, RM6);
1283          if (rrc == MATCH_MATCH)          if (rrc == MATCH_MATCH)
1284            {            {
1285              DPRINTF(("Recursion matched\n"));
1286            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1287            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1288              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1289            RRETURN(MATCH_MATCH);            RRETURN(MATCH_MATCH);
1290            }            }
1291          else if (rrc != MATCH_NOMATCH) RRETURN(rrc);          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1292              {
1293              DPRINTF(("Recursion gave error %d\n", rrc));
1294              if (new_recursive.offset_save != stacksave)
1295                (pcre_free)(new_recursive.offset_save);
1296              RRETURN(rrc);
1297              }
1298    
1299          md->recursive = &new_recursive;          md->recursive = &new_recursive;
1300          memcpy(md->offset_vector, new_recursive.offset_save,          memcpy(md->offset_vector, new_recursive.offset_save,
# Line 912  for (;;) Line 1319  for (;;)
1319      the end of a normal bracket, leaving the subject pointer. */      the end of a normal bracket, leaving the subject pointer. */
1320    
1321      case OP_ONCE:      case OP_ONCE:
1322        {      prev = ecode;
1323        prev = ecode;      saved_eptr = eptr;
       saved_eptr = eptr;  
1324    
1325        do      do
1326          {        {
1327          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);
1328            eptrb, match_isgroup);        if (rrc == MATCH_MATCH) break;
1329          if (rrc == MATCH_MATCH) break;        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1330          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        ecode += GET(ecode,1);
1331          ecode += GET(ecode,1);        }
1332          }      while (*ecode == OP_ALT);
       while (*ecode == OP_ALT);  
1333    
1334        /* If hit the end of the group (which could be repeated), fail */      /* If hit the end of the group (which could be repeated), fail */
1335    
1336        if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);      if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
1337    
1338        /* Continue as from after the assertion, updating the offsets high water      /* Continue as from after the assertion, updating the offsets high water
1339        mark, since extracts may have been taken. */      mark, since extracts may have been taken. */
1340    
1341        do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
1342    
1343        offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1344        eptr = md->end_match_ptr;      eptr = md->end_match_ptr;
1345    
1346        /* For a non-repeating ket, just continue at this level. This also      /* For a non-repeating ket, just continue at this level. This also
1347        happens for a repeating ket if no characters were matched in the group.      happens for a repeating ket if no characters were matched in the group.
1348        This is the forcible breaking of infinite loops as implemented in Perl      This is the forcible breaking of infinite loops as implemented in Perl
1349        5.005. If there is an options reset, it will get obeyed in the normal      5.005. If there is an options reset, it will get obeyed in the normal
1350        course of events. */      course of events. */
1351    
1352        if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1353          {        {
1354          ecode += 1+LINK_SIZE;        ecode += 1+LINK_SIZE;
1355          break;        break;
1356          }        }
1357    
1358        /* The repeating kets try the rest of the pattern or restart from the      /* The repeating kets try the rest of the pattern or restart from the
1359        preceding bracket, in the appropriate order. We need to reset any options      preceding bracket, in the appropriate order. The second "call" of match()
1360        that changed within the bracket before re-running it, so check the next      uses tail recursion, to avoid using another stack frame. We need to reset
1361        opcode. */      any options that changed within the bracket before re-running it, so
1362        check the next opcode. */
1363    
1364        if (ecode[1+LINK_SIZE] == OP_OPT)      if (ecode[1+LINK_SIZE] == OP_OPT)
1365          {        {
1366          ims = (ims & ~PCRE_IMS) | ecode[4];        ims = (ims & ~PCRE_IMS) | ecode[4];
1367          DPRINTF(("ims set to %02lx at group repeat\n", ims));        DPRINTF(("ims set to %02lx at group repeat\n", ims));
1368          }        }
1369    
1370        if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1371          {        {
1372          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);
1373          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1374          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);        ecode = prev;
1375          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        flags = 0;
1376          }        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);  
         }  
1377        }        }
1378      RRETURN(MATCH_NOMATCH);      else  /* OP_KETRMAX */
1379          {
1380          RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);
1381          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1382          ecode += 1 + LINK_SIZE;
1383          flags = 0;
1384          goto TAIL_RECURSE;
1385          }
1386        /* Control never gets here */
1387    
1388      /* 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
1389      bracketed group and go to there. */      bracketed group and go to there. */
# Line 985  for (;;) Line 1392  for (;;)
1392      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1393      break;      break;
1394    
1395      /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1396      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
1397      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
1398      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
1399      preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1400    
1401      case OP_BRAZERO:      case OP_BRAZERO:
1402        {        {
1403        next = ecode+1;        next = ecode+1;
1404        RMATCH(rrc, eptr, next, offset_top, md, ims, eptrb, match_isgroup);        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);
1405        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1406        do next += GET(next,1); while (*next == OP_ALT);        do next += GET(next,1); while (*next == OP_ALT);
1407        ecode = next + 1+LINK_SIZE;        ecode = next + 1 + LINK_SIZE;
1408        }        }
1409      break;      break;
1410    
1411      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1412        {        {
1413        next = ecode+1;        next = ecode+1;
1414        do next += GET(next,1); while (*next == OP_ALT);        do next += GET(next, 1); while (*next == OP_ALT);
1415        RMATCH(rrc, eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb,        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);
         match_isgroup);  
1416        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1417        ecode++;        ecode++;
1418        }        }
1419      break;      break;
1420    
1421      /* End of a group, repeated or non-repeating. If we are at the end of      case OP_SKIPZERO:
1422      an assertion "group", stop matching and return MATCH_MATCH, but record the        {
1423      current high water mark for use by positive assertions. Do this also        next = ecode+1;
1424      for the "once" (not-backup up) groups. */        do next += GET(next,1); while (*next == OP_ALT);
1425          ecode = next + 1 + LINK_SIZE;
1426          }
1427        break;
1428    
1429        /* End of a group, repeated or non-repeating. */
1430    
1431      case OP_KET:      case OP_KET:
1432      case OP_KETRMIN:      case OP_KETRMIN:
1433      case OP_KETRMAX:      case OP_KETRMAX:
1434        {      prev = ecode - GET(ecode, 1);
       prev = ecode - GET(ecode, 1);  
       saved_eptr = eptrb->epb_saved_eptr;  
1435    
1436        /* Back up the stack of bracket start pointers. */      /* If this was a group that remembered the subject start, in order to break
1437        infinite repeats of empty string matches, retrieve the subject start from
1438        the chain. Otherwise, set it NULL. */
1439    
1440        eptrb = eptrb->epb_prev;      if (*prev >= OP_SBRA)
1441          {
1442        if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1443            *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1444            *prev == OP_ONCE)        }
1445          {      else saved_eptr = NULL;
         md->end_match_ptr = eptr;      /* For ONCE */  
         md->end_offset_top = offset_top;  
         RRETURN(MATCH_MATCH);  
         }  
1446    
1447        /* In all other cases except a conditional group we have to check the      /* If we are at the end of an assertion group, stop matching and return
1448        group number back at the start and if necessary complete handling an      MATCH_MATCH, but record the current high water mark for use by positive
1449        extraction by setting the offsets and bumping the high water mark. */      assertions. Do this also for the "once" (atomic) groups. */
1450    
1451        if (*prev != OP_COND)      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1452          {          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||
1453          number = *prev - OP_BRA;          *prev == OP_ONCE)
1454          {
1455          md->end_match_ptr = eptr;      /* For ONCE */
1456          md->end_offset_top = offset_top;
1457          RRETURN(MATCH_MATCH);
1458          }
1459    
1460          /* 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
1461          the number from a dummy opcode at the start. */      and if necessary complete handling an extraction by setting the offsets and
1462        bumping the high water mark. Note that whole-pattern recursion is coded as
1463        a recurse into group 0, so it won't be picked up here. Instead, we catch it
1464        when the OP_END is reached. Other recursion is handled here. */
1465    
1466          if (number > EXTRACT_BASIC_MAX) number = GET2(prev, 2+LINK_SIZE);      if (*prev == OP_CBRA || *prev == OP_SCBRA)
1467          offset = number << 1;        {
1468          number = GET2(prev, 1+LINK_SIZE);
1469          offset = number << 1;
1470    
1471  #ifdef DEBUG  #ifdef DEBUG
1472          printf("end bracket %d", number);        printf("end bracket %d", number);
1473          printf("\n");        printf("\n");
1474  #endif  #endif
1475    
1476          /* Test for a numbered group. This includes groups called as a result        md->capture_last = number;
1477          of recursion. Note that whole-pattern recursion is coded as a recurse        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1478          into group 0, so it won't be picked up here. Instead, we catch it when          {
1479          the OP_END is reached. */          md->offset_vector[offset] =
1480              md->offset_vector[md->offset_end - number];
1481            md->offset_vector[offset+1] = eptr - md->start_subject;
1482            if (offset_top <= offset) offset_top = offset + 2;
1483            }
1484    
1485          /* Handle a recursively called group. Restore the offsets
1486          appropriately and continue from after the call. */
1487    
1488          if (md->recursive != NULL && md->recursive->group_num == number)
1489            {
1490            recursion_info *rec = md->recursive;
1491            DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
1492            md->recursive = rec->prevrec;
1493            mstart = rec->save_start;
1494            memcpy(md->offset_vector, rec->offset_save,
1495              rec->saved_max * sizeof(int));
1496            offset_top = rec->save_offset_top;
1497            ecode = rec->after_call;
1498            ims = original_ims;
1499            break;
1500            }
1501          }
1502    
1503          if (number > 0)      /* For both capturing and non-capturing groups, reset the value of the ims
1504            {      flags, in case they got changed during the group. */
           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;  
             }  
1505    
1506            /* Handle a recursively called group. Restore the offsets      ims = original_ims;
1507            appropriately and continue from after the call. */      DPRINTF(("ims reset to %02lx\n", ims));
1508    
1509            if (md->recursive != NULL && md->recursive->group_num == number)      /* For a non-repeating ket, just continue at this level. This also
1510              {      happens for a repeating ket if no characters were matched in the group.
1511              recursion_info *rec = md->recursive;      This is the forcible breaking of infinite loops as implemented in Perl
1512              DPRINTF(("Recursion (%d) succeeded - continuing\n", number));      5.005. If there is an options reset, it will get obeyed in the normal
1513              md->recursive = rec->prevrec;      course of events. */
             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;  
             }  
           }  
         }  
1514    
1515        /* Reset the value of the ims flags, in case they got changed during      if (*ecode == OP_KET || eptr == saved_eptr)
1516        the group. */        {
1517          ecode += 1 + LINK_SIZE;
1518          break;
1519          }
1520    
1521        ims = original_ims;      /* The repeating kets try the rest of the pattern or restart from the
1522        DPRINTF(("ims reset to %02lx\n", ims));      preceding bracket, in the appropriate order. In the second case, we can use
1523        tail recursion to avoid using another stack frame, unless we have an
1524        unlimited repeat of a group that can match an empty string. */
1525    
1526        /* For a non-repeating ket, just continue at this level. This also      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;
       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. */  
1527    
1528        if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRMIN)
1529          {
1530          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);
1531          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1532          if (flags != 0)    /* Could match an empty string */
1533          {          {
1534          ecode += 1 + LINK_SIZE;          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);
1535          break;          RRETURN(rrc);
         }  
   
       /* The repeating kets try the rest of the pattern or restart from the  
       preceding bracket, in the appropriate order. */  
   
       if (*ecode == OP_KETRMIN)  
         {  
         RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         }  
       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);  
1536          }          }
1537          ecode = prev;
1538          goto TAIL_RECURSE;
1539        }        }
1540        else  /* OP_KETRMAX */
1541      RRETURN(MATCH_NOMATCH);        {
1542          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);
1543          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1544          ecode += 1 + LINK_SIZE;
1545          flags = 0;
1546          goto TAIL_RECURSE;
1547          }
1548        /* Control never gets here */
1549    
1550      /* Start of subject unless notbol, or after internal newline if multiline */      /* Start of subject unless notbol, or after internal newline if multiline */
1551    
# Line 1135  for (;;) Line 1553  for (;;)
1553      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
1554      if ((ims & PCRE_MULTILINE) != 0)      if ((ims & PCRE_MULTILINE) != 0)
1555        {        {
1556        if (eptr != md->start_subject && eptr[-1] != NEWLINE)        if (eptr != md->start_subject &&
1557              (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1558          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
1559        ecode++;        ecode++;
1560        break;        break;
# Line 1156  for (;;) Line 1575  for (;;)
1575      ecode++;      ecode++;
1576      break;      break;
1577    
1578        /* Reset the start of match point */
1579    
1580        case OP_SET_SOM:
1581        mstart = eptr;
1582        ecode++;
1583        break;
1584    
1585      /* Assert before internal newline if multiline, or before a terminating      /* Assert before internal newline if multiline, or before a terminating
1586      newline unless endonly is set, else end of subject unless noteol is set. */      newline unless endonly is set, else end of subject unless noteol is set. */
1587    
# Line 1163  for (;;) Line 1589  for (;;)
1589      if ((ims & PCRE_MULTILINE) != 0)      if ((ims & PCRE_MULTILINE) != 0)
1590        {        {
1591        if (eptr < md->end_subject)        if (eptr < md->end_subject)
1592          { if (*eptr != NEWLINE) RRETURN(MATCH_NOMATCH); }          { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }
1593        else        else
1594          { if (md->noteol) RRETURN(MATCH_NOMATCH); }          { if (md->noteol) RRETURN(MATCH_NOMATCH); }
1595        ecode++;        ecode++;
# Line 1174  for (;;) Line 1600  for (;;)
1600        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
1601        if (!md->endonly)        if (!md->endonly)
1602          {          {
1603          if (eptr < md->end_subject - 1 ||          if (eptr != md->end_subject &&
1604             (eptr == md->end_subject - 1 && *eptr != NEWLINE))              (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1605            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1606          ecode++;          ecode++;
1607          break;          break;
1608          }          }
1609        }        }
1610      /* ... else fall through */      /* ... else fall through for endonly */
1611    
1612      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1613    
# Line 1193  for (;;) Line 1619  for (;;)
1619      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1620    
1621      case OP_EODN:      case OP_EODN:
1622      if (eptr < md->end_subject - 1 ||      if (eptr != md->end_subject &&
1623         (eptr == md->end_subject - 1 && *eptr != NEWLINE)) RRETURN(MATCH_NOMATCH);          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1624          RRETURN(MATCH_NOMATCH);
1625      ecode++;      ecode++;
1626      break;      break;
1627    
# Line 1206  for (;;) Line 1633  for (;;)
1633    
1634        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1635        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
1636        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1637          partial matching. */
1638    
1639  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1640        if (utf8)        if (utf8)
1641          {          {
1642          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1643            {            {
1644            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1645            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1646              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1647            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1648            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1649            }            }
1650          if (eptr >= md->end_subject) cur_is_word = FALSE; else          if (eptr >= md->end_subject)
1651              {
1652              SCHECK_PARTIAL();
1653              cur_is_word = FALSE;
1654              }
1655            else
1656            {            {
1657            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1658            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
# Line 1227  for (;;) Line 1661  for (;;)
1661        else        else
1662  #endif  #endif
1663    
1664        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode */
1665    
1666          {          {
1667          prev_is_word = (eptr != md->start_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
1668            ((md->ctypes[eptr[-1]] & ctype_word) != 0);            {
1669          cur_is_word = (eptr < md->end_subject) &&            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1670            ((md->ctypes[*eptr] & ctype_word) != 0);            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1671              }
1672            if (eptr >= md->end_subject)
1673              {
1674              SCHECK_PARTIAL();
1675              cur_is_word = FALSE;
1676              }
1677            else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
1678          }          }
1679    
1680        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
# Line 1247  for (;;) Line 1688  for (;;)
1688      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
1689    
1690      case OP_ANY:      case OP_ANY:
1691      if ((ims & PCRE_DOTALL) == 0 && eptr < md->end_subject && *eptr == NEWLINE)      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
1692        /* Fall through */
1693    
1694        case OP_ALLANY:
1695        if (eptr++ >= md->end_subject)
1696          {
1697          SCHECK_PARTIAL();
1698        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1699      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);        }
1700  #ifdef SUPPORT_UTF8      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
     if (utf8)  
       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
 #endif  
1701      ecode++;      ecode++;
1702      break;      break;
1703    
# Line 1261  for (;;) Line 1705  for (;;)
1705      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
1706    
1707      case OP_ANYBYTE:      case OP_ANYBYTE:
1708      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr++ >= md->end_subject)
1709          {
1710          SCHECK_PARTIAL();
1711          RRETURN(MATCH_NOMATCH);
1712          }
1713      ecode++;      ecode++;
1714      break;      break;
1715    
1716      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1717      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
1718          {
1719          SCHECK_PARTIAL();
1720          RRETURN(MATCH_NOMATCH);
1721          }
1722      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1723      if (      if (
1724  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1279  for (;;) Line 1731  for (;;)
1731      break;      break;
1732    
1733      case OP_DIGIT:      case OP_DIGIT:
1734      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
1735          {
1736          SCHECK_PARTIAL();
1737          RRETURN(MATCH_NOMATCH);
1738          }
1739      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1740      if (      if (
1741  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1292  for (;;) Line 1748  for (;;)
1748      break;      break;
1749    
1750      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
1751      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
1752          {
1753          SCHECK_PARTIAL();
1754          RRETURN(MATCH_NOMATCH);
1755          }
1756      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1757      if (      if (
1758  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1305  for (;;) Line 1765  for (;;)
1765      break;      break;
1766    
1767      case OP_WHITESPACE:      case OP_WHITESPACE:
1768      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
1769          {
1770          SCHECK_PARTIAL();
1771          RRETURN(MATCH_NOMATCH);
1772          }
1773      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1774      if (      if (
1775  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1318  for (;;) Line 1782  for (;;)
1782      break;      break;
1783    
1784      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
1785      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
1786          {
1787          SCHECK_PARTIAL();
1788          RRETURN(MATCH_NOMATCH);
1789          }
1790      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1791      if (      if (
1792  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1331  for (;;) Line 1799  for (;;)
1799      break;      break;
1800    
1801      case OP_WORDCHAR:      case OP_WORDCHAR:
1802      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
1803          {
1804          SCHECK_PARTIAL();
1805          RRETURN(MATCH_NOMATCH);
1806          }
1807      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1808      if (      if (
1809  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1343  for (;;) Line 1815  for (;;)
1815      ecode++;      ecode++;
1816      break;      break;
1817    
1818        case OP_ANYNL:
1819        if (eptr >= md->end_subject)
1820          {
1821          SCHECK_PARTIAL();
1822          RRETURN(MATCH_NOMATCH);
1823          }
1824        GETCHARINCTEST(c, eptr);
1825        switch(c)
1826          {
1827          default: RRETURN(MATCH_NOMATCH);
1828          case 0x000d:
1829          if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
1830          break;
1831    
1832          case 0x000a:
1833          break;
1834    
1835          case 0x000b:
1836          case 0x000c:
1837          case 0x0085:
1838          case 0x2028:
1839          case 0x2029:
1840          if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
1841          break;
1842          }
1843        ecode++;
1844        break;
1845    
1846        case OP_NOT_HSPACE:
1847        if (eptr >= md->end_subject)
1848          {
1849          SCHECK_PARTIAL();
1850          RRETURN(MATCH_NOMATCH);
1851          }
1852        GETCHARINCTEST(c, eptr);
1853        switch(c)
1854          {
1855          default: break;
1856          case 0x09:      /* HT */
1857          case 0x20:      /* SPACE */
1858          case 0xa0:      /* NBSP */
1859          case 0x1680:    /* OGHAM SPACE MARK */
1860          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
1861          case 0x2000:    /* EN QUAD */
1862          case 0x2001:    /* EM QUAD */
1863          case 0x2002:    /* EN SPACE */
1864          case 0x2003:    /* EM SPACE */
1865          case 0x2004:    /* THREE-PER-EM SPACE */
1866          case 0x2005:    /* FOUR-PER-EM SPACE */
1867          case 0x2006:    /* SIX-PER-EM SPACE */
1868          case 0x2007:    /* FIGURE SPACE */
1869          case 0x2008:    /* PUNCTUATION SPACE */
1870          case 0x2009:    /* THIN SPACE */
1871          case 0x200A:    /* HAIR SPACE */
1872          case 0x202f:    /* NARROW NO-BREAK SPACE */
1873          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
1874          case 0x3000:    /* IDEOGRAPHIC SPACE */
1875          RRETURN(MATCH_NOMATCH);
1876          }
1877        ecode++;
1878        break;
1879    
1880        case OP_HSPACE:
1881        if (eptr >= md->end_subject)
1882          {
1883          SCHECK_PARTIAL();
1884          RRETURN(MATCH_NOMATCH);
1885          }
1886        GETCHARINCTEST(c, eptr);
1887        switch(c)
1888          {
1889          default: RRETURN(MATCH_NOMATCH);
1890          case 0x09:      /* HT */
1891          case 0x20:      /* SPACE */
1892          case 0xa0:      /* NBSP */
1893          case 0x1680:    /* OGHAM SPACE MARK */
1894          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
1895          case 0x2000:    /* EN QUAD */
1896          case 0x2001:    /* EM QUAD */
1897          case 0x2002:    /* EN SPACE */
1898          case 0x2003:    /* EM SPACE */
1899          case 0x2004:    /* THREE-PER-EM SPACE */
1900          case 0x2005:    /* FOUR-PER-EM SPACE */
1901          case 0x2006:    /* SIX-PER-EM SPACE */
1902          case 0x2007:    /* FIGURE SPACE */
1903          case 0x2008:    /* PUNCTUATION SPACE */
1904          case 0x2009:    /* THIN SPACE */
1905          case 0x200A:    /* HAIR SPACE */
1906          case 0x202f:    /* NARROW NO-BREAK SPACE */
1907          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
1908          case 0x3000:    /* IDEOGRAPHIC SPACE */
1909          break;
1910          }
1911        ecode++;
1912        break;
1913    
1914        case OP_NOT_VSPACE:
1915        if (eptr >= md->end_subject)
1916          {
1917          SCHECK_PARTIAL();
1918          RRETURN(MATCH_NOMATCH);
1919          }
1920        GETCHARINCTEST(c, eptr);
1921        switch(c)
1922          {
1923          default: break;
1924          case 0x0a:      /* LF */
1925          case 0x0b:      /* VT */
1926          case 0x0c:      /* FF */
1927          case 0x0d:      /* CR */
1928          case 0x85:      /* NEL */
1929          case 0x2028:    /* LINE SEPARATOR */
1930          case 0x2029:    /* PARAGRAPH SEPARATOR */
1931          RRETURN(MATCH_NOMATCH);
1932          }
1933        ecode++;
1934        break;
1935    
1936        case OP_VSPACE:
1937        if (eptr >= md->end_subject)
1938          {
1939          SCHECK_PARTIAL();
1940          RRETURN(MATCH_NOMATCH);
1941          }
1942        GETCHARINCTEST(c, eptr);
1943        switch(c)
1944          {
1945          default: RRETURN(MATCH_NOMATCH);
1946          case 0x0a:      /* LF */
1947          case 0x0b:      /* VT */
1948          case 0x0c:      /* FF */
1949          case 0x0d:      /* CR */
1950          case 0x85:      /* NEL */
1951          case 0x2028:    /* LINE SEPARATOR */
1952          case 0x2029:    /* PARAGRAPH SEPARATOR */
1953          break;
1954          }
1955        ecode++;
1956        break;
1957    
1958  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1959      /* Check the next character by Unicode property. We will get here only      /* Check the next character by Unicode property. We will get here only
1960      if the support is in the binary; otherwise a compile-time error occurs. */      if the support is in the binary; otherwise a compile-time error occurs. */
1961    
1962      case OP_PROP:      case OP_PROP:
1963      case OP_NOTPROP:      case OP_NOTPROP:
1964      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
1965          {
1966          SCHECK_PARTIAL();
1967          RRETURN(MATCH_NOMATCH);
1968          }
1969      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1970        {        {
1971        int chartype, rqdtype;        const ucd_record *prop = GET_UCD(c);
       int othercase;  
       int category = _pcre_ucp_findchar(c, &chartype, &othercase);  
1972    
1973        rqdtype = *(++ecode);        switch(ecode[1])
       ecode++;  
   
       if (rqdtype >= 128)  
1974          {          {
1975          if ((rqdtype - 128 != category) == (op == OP_PROP))          case PT_ANY:
1976            if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
1977            break;
1978    
1979            case PT_LAMP:
1980            if ((prop->chartype == ucp_Lu ||
1981                 prop->chartype == ucp_Ll ||
1982                 prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
1983            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1984          }           break;
1985        else  
1986          {          case PT_GC:
1987          if ((rqdtype != chartype) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
1988              RRETURN(MATCH_NOMATCH);
1989            break;
1990    
1991            case PT_PC:
1992            if ((ecode[2] != prop->chartype) == (op == OP_PROP))
1993              RRETURN(MATCH_NOMATCH);
1994            break;
1995    
1996            case PT_SC:
1997            if ((ecode[2] != prop->script) == (op == OP_PROP))
1998            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1999            break;
2000    
2001            default:
2002            RRETURN(PCRE_ERROR_INTERNAL);
2003          }          }
2004    
2005          ecode += 3;
2006        }        }
2007      break;      break;
2008    
# Line 1376  for (;;) Line 2010  for (;;)
2010      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2011    
2012      case OP_EXTUNI:      case OP_EXTUNI:
2013      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2014          {
2015          SCHECK_PARTIAL();
2016          RRETURN(MATCH_NOMATCH);
2017          }
2018      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2019        {        {
2020        int chartype;        int category = UCD_CATEGORY(c);
       int othercase;  
       int category = _pcre_ucp_findchar(c, &chartype, &othercase);  
2021        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (category == ucp_M) RRETURN(MATCH_NOMATCH);
2022        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2023          {          {
# Line 1390  for (;;) Line 2026  for (;;)
2026            {            {
2027            GETCHARLEN(c, eptr, len);            GETCHARLEN(c, eptr, len);
2028            }            }
2029          category = _pcre_ucp_findchar(c, &chartype, &othercase);          category = UCD_CATEGORY(c);
2030          if (category != ucp_M) break;          if (category != ucp_M) break;
2031          eptr += len;          eptr += len;
2032          }          }
# Line 1411  for (;;) Line 2047  for (;;)
2047      case OP_REF:      case OP_REF:
2048        {        {
2049        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2050        ecode += 3;                                 /* Advance past item */        ecode += 3;
2051    
2052          /* If the reference is unset, there are two possibilities:
2053    
2054          (a) In the default, Perl-compatible state, set the length to be longer
2055          than the amount of subject left; this ensures that every attempt at a
2056          match fails. We can't just fail here, because of the possibility of
2057          quantifiers with zero minima.
2058    
2059        /* If the reference is unset, set the length to be longer than the amount        (b) If the JavaScript compatibility flag is set, set the length to zero
2060        of subject left; this ensures that every attempt at a match fails. We        so that the back reference matches an empty string.
2061        can't just fail here, because of the possibility of quantifiers with zero  
2062        minima. */        Otherwise, set the length to the length of what was matched by the
2063          referenced subpattern. */
2064        length = (offset >= offset_top || md->offset_vector[offset] < 0)?  
2065          md->end_subject - eptr + 1 :        if (offset >= offset_top || md->offset_vector[offset] < 0)
2066          md->offset_vector[offset+1] - md->offset_vector[offset];          length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;
2067          else
2068            length = md->offset_vector[offset+1] - md->offset_vector[offset];
2069    
2070        /* Set up for repetition, or handle the non-repeated case */        /* Set up for repetition, or handle the non-repeated case */
2071    
# Line 1449  for (;;) Line 2094  for (;;)
2094          break;          break;
2095    
2096          default:               /* No repeat follows */          default:               /* No repeat follows */
2097          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          if (!match_ref(offset, eptr, length, md, ims))
2098              {
2099              CHECK_PARTIAL();
2100              RRETURN(MATCH_NOMATCH);
2101              }
2102          eptr += length;          eptr += length;
2103          continue;              /* With the main loop */          continue;              /* With the main loop */
2104          }          }
# Line 1465  for (;;) Line 2114  for (;;)
2114    
2115        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2116          {          {
2117          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          if (!match_ref(offset, eptr, length, md, ims))
2118              {
2119              CHECK_PARTIAL();
2120              RRETURN(MATCH_NOMATCH);
2121              }
2122          eptr += length;          eptr += length;
2123          }          }
2124    
# Line 1480  for (;;) Line 2133  for (;;)
2133          {          {
2134          for (fi = min;; fi++)          for (fi = min;; fi++)
2135            {            {
2136            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);
2137            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2138            if (fi >= max || !match_ref(offset, eptr, length, md, ims))            if (fi >= max) RRETURN(MATCH_NOMATCH);
2139              if (!match_ref(offset, eptr, length, md, ims))
2140                {
2141                CHECK_PARTIAL();
2142              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2143                }
2144            eptr += length;            eptr += length;
2145            }            }
2146          /* Control never gets here */          /* Control never gets here */
# Line 1496  for (;;) Line 2153  for (;;)
2153          pp = eptr;          pp = eptr;
2154          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2155            {            {
2156            if (!match_ref(offset, eptr, length, md, ims)) break;            if (!match_ref(offset, eptr, length, md, ims))
2157                {
2158                CHECK_PARTIAL();
2159                break;
2160                }
2161            eptr += length;            eptr += length;
2162            }            }
2163          while (eptr >= pp)          while (eptr >= pp)
2164            {            {
2165            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);
2166            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2167            eptr -= length;            eptr -= length;
2168            }            }
# Line 1510  for (;;) Line 2171  for (;;)
2171        }        }
2172      /* Control never gets here */      /* Control never gets here */
2173    
   
   
2174      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2175      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,
2176      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 2225  for (;;)
2225          {          {
2226          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2227            {            {
2228            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2229                {
2230                SCHECK_PARTIAL();
2231                RRETURN(MATCH_NOMATCH);
2232                }
2233            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2234            if (c > 255)            if (c > 255)
2235              {              {
# Line 1584  for (;;) Line 2247  for (;;)
2247          {          {
2248          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2249            {            {
2250            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2251                {
2252                SCHECK_PARTIAL();
2253                RRETURN(MATCH_NOMATCH);
2254                }
2255            c = *eptr++;            c = *eptr++;
2256            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2257            }            }
# Line 1606  for (;;) Line 2273  for (;;)
2273            {            {
2274            for (fi = min;; fi++)            for (fi = min;; fi++)
2275              {              {
2276              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);
2277              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2278              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2279                if (eptr >= md->end_subject)
2280                  {
2281                  SCHECK_PARTIAL();
2282                  RRETURN(MATCH_NOMATCH);
2283                  }
2284              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2285              if (c > 255)              if (c > 255)
2286                {                {
# Line 1626  for (;;) Line 2298  for (;;)
2298            {            {
2299            for (fi = min;; fi++)            for (fi = min;; fi++)
2300              {              {
2301              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);
2302              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2303              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2304                if (eptr >= md->end_subject)
2305                  {
2306                  SCHECK_PARTIAL();
2307                  RRETURN(MATCH_NOMATCH);
2308                  }
2309              c = *eptr++;              c = *eptr++;
2310              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2311              }              }
# Line 1649  for (;;) Line 2326  for (;;)
2326            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2327              {              {
2328              int len = 1;              int len = 1;
2329              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2330                  {
2331                  SCHECK_PARTIAL();
2332                  break;
2333                  }
2334              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2335              if (c > 255)              if (c > 255)
2336                {                {
# Line 1663  for (;;) Line 2344  for (;;)
2344              }              }
2345            for (;;)            for (;;)
2346              {              {
2347              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);
2348              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2349              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2350              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1675  for (;;) Line 2356  for (;;)
2356            {            {
2357            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2358              {              {
2359              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2360                  {
2361                  SCHECK_PARTIAL();
2362                  break;
2363                  }
2364              c = *eptr;              c = *eptr;
2365              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2366              eptr++;              eptr++;
2367              }              }
2368            while (eptr >= pp)            while (eptr >= pp)
2369              {              {
2370              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);
             eptr--;  
2371              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2372                eptr--;
2373              }              }
2374            }            }
2375    
# Line 1695  for (;;) Line 2380  for (;;)
2380    
2381    
2382      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2383      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
2384        mode, because Unicode properties are supported in non-UTF-8 mode. */
2385    
2386  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2387      case OP_XCLASS:      case OP_XCLASS:
# Line 1736  for (;;) Line 2422  for (;;)
2422    
2423        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2424          {          {
2425          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2426          GETCHARINC(c, eptr);            {
2427              SCHECK_PARTIAL();
2428              RRETURN(MATCH_NOMATCH);
2429              }
2430            GETCHARINCTEST(c, eptr);
2431          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);
2432          }          }
2433    
# Line 1753  for (;;) Line 2443  for (;;)
2443          {          {
2444          for (fi = min;; fi++)          for (fi = min;; fi++)
2445            {            {
2446            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);
2447            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2448            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
2449            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2450                {
2451                SCHECK_PARTIAL();
2452                RRETURN(MATCH_NOMATCH);
2453                }
2454              GETCHARINCTEST(c, eptr);
2455            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);
2456            }            }
2457          /* Control never gets here */          /* Control never gets here */
# Line 1770  for (;;) Line 2465  for (;;)
2465          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2466            {            {
2467            int len = 1;            int len = 1;
2468            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2469            GETCHARLEN(c, eptr, len);              {
2470                SCHECK_PARTIAL();
2471                break;
2472                }
2473              GETCHARLENTEST(c, eptr, len);
2474            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2475            eptr += len;            eptr += len;
2476            }            }
2477          for(;;)          for(;;)
2478            {            {
2479            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);
2480            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2481            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2482            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2483            }            }
2484          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2485          }          }
# Line 1798  for (;;) Line 2497  for (;;)
2497        length = 1;        length = 1;
2498        ecode++;        ecode++;
2499        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2500        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2501            {
2502            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2503            RRETURN(MATCH_NOMATCH);
2504            }
2505        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);
2506        }        }
2507      else      else
# Line 1806  for (;;) Line 2509  for (;;)
2509    
2510      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2511        {        {
2512        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2513            {
2514            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2515            RRETURN(MATCH_NOMATCH);
2516            }
2517        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
2518        ecode += 2;        ecode += 2;
2519        }        }
# Line 1822  for (;;) Line 2529  for (;;)
2529        ecode++;        ecode++;
2530        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2531    
2532        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2533            {
2534            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2535            RRETURN(MATCH_NOMATCH);
2536            }
2537    
2538        /* 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
2539        can use the fast lookup table. */        can use the fast lookup table. */
# Line 1836  for (;;) Line 2547  for (;;)
2547    
2548        else        else
2549          {          {
2550          int dc;          unsigned int dc;
2551          GETCHARINC(dc, eptr);          GETCHARINC(dc, eptr);
2552          ecode += length;          ecode += length;
2553    
2554          /* 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
2555          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. */  
2556    
2557          if (fc != dc)          if (fc != dc)
2558            {            {
2559  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2560            int chartype;            if (dc != UCD_OTHERCASE(fc))
           int othercase;  
           if (_pcre_ucp_findchar(fc, &chartype, &othercase) < 0 || dc != othercase)  
2561  #endif  #endif
2562              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2563            }            }
# Line 1861  for (;;) Line 2568  for (;;)
2568    
2569      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2570        {        {
2571        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2572            {
2573            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2574            RRETURN(MATCH_NOMATCH);
2575            }
2576        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
2577        ecode += 2;        ecode += 2;
2578        }        }
2579      break;      break;
2580    
2581      /* Match a single character repeatedly; different opcodes share code. */      /* Match a single character repeatedly. */
2582    
2583      case OP_EXACT:      case OP_EXACT:
2584      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2585      ecode += 3;      ecode += 3;
2586      goto REPEATCHAR;      goto REPEATCHAR;
2587    
2588        case OP_POSUPTO:
2589        possessive = TRUE;
2590        /* Fall through */
2591    
2592      case OP_UPTO:      case OP_UPTO:
2593      case OP_MINUPTO:      case OP_MINUPTO:
2594      min = 0;      min = 0;
# Line 1882  for (;;) Line 2597  for (;;)
2597      ecode += 3;      ecode += 3;
2598      goto REPEATCHAR;      goto REPEATCHAR;
2599    
2600        case OP_POSSTAR:
2601        possessive = TRUE;
2602        min = 0;
2603        max = INT_MAX;
2604        ecode++;
2605        goto REPEATCHAR;
2606    
2607        case OP_POSPLUS:
2608        possessive = TRUE;
2609        min = 1;
2610        max = INT_MAX;
2611        ecode++;
2612        goto REPEATCHAR;
2613    
2614        case OP_POSQUERY:
2615        possessive = TRUE;
2616        min = 0;
2617        max = 1;
2618        ecode++;
2619        goto REPEATCHAR;
2620    
2621      case OP_STAR:      case OP_STAR:
2622      case OP_MINSTAR:      case OP_MINSTAR:
2623      case OP_PLUS:      case OP_PLUS:
# Line 1890  for (;;) Line 2626  for (;;)
2626      case OP_MINQUERY:      case OP_MINQUERY:
2627      c = *ecode++ - OP_STAR;      c = *ecode++ - OP_STAR;
2628      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
2629    
2630      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2631      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2632      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
2633    
2634      /* 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. */  
2635    
2636      REPEATCHAR:      REPEATCHAR:
2637  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1905  for (;;) Line 2640  for (;;)
2640        length = 1;        length = 1;
2641        charptr = ecode;        charptr = ecode;
2642        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
2643        ecode += length;        ecode += length;
2644    
2645        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 1913  for (;;) Line 2647  for (;;)
2647    
2648        if (length > 1)        if (length > 1)
2649          {          {
         int oclength = 0;  
         uschar occhars[8];  
   
2650  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2651          int othercase;          unsigned int othercase;
         int chartype;  
2652          if ((ims & PCRE_CASELESS) != 0 &&          if ((ims & PCRE_CASELESS) != 0 &&
2653               _pcre_ucp_findchar(fc, &chartype, &othercase) >= 0 &&              (othercase = UCD_OTHERCASE(fc)) != fc)
              othercase > 0)  
2654            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
2655            else oclength = 0;
2656  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2657    
2658          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2659            {            {
2660            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
2661            /* Need braces because of following else */              memcmp(eptr, charptr, length) == 0) eptr += length;
2662            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
2663              else if (oclength > 0 &&
2664                       eptr <= md->end_subject - oclength &&
2665                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
2666    #endif  /* SUPPORT_UCP */
2667            else            else
2668              {              {
2669              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
2670              eptr += oclength;              RRETURN(MATCH_NOMATCH);
2671              }              }
2672            }            }
2673    
# Line 1943  for (;;) Line 2677  for (;;)
2677            {            {
2678            for (fi = min;; fi++)            for (fi = min;; fi++)
2679              {              {
2680              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);
2681              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2682              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2683              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
2684              /* Need braces because of following else */                memcmp(eptr, charptr, length) == 0) eptr += length;
2685              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }  #ifdef SUPPORT_UCP
2686                else if (oclength > 0 &&
2687                         eptr <= md->end_subject - oclength &&
2688                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
2689    #endif  /* SUPPORT_UCP */
2690              else              else
2691                {                {
2692                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
2693                eptr += oclength;                RRETURN(MATCH_NOMATCH);
2694                }                }
2695              }              }
2696            /* Control never gets here */            /* Control never gets here */
2697            }            }
2698          else  
2699            else  /* Maximize */
2700            {            {
2701            pp = eptr;            pp = eptr;
2702            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2703              {              {
2704              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
2705              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
2706              else if (oclength == 0) break;  #ifdef SUPPORT_UCP
2707                else if (oclength > 0 &&
2708                         eptr <= md->end_subject - oclength &&
2709                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
2710    #endif  /* SUPPORT_UCP */
2711              else              else
2712                {                {
2713                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
2714                eptr += oclength;                break;
2715                }                }
2716              }              }
2717            while (eptr >= pp)  
2718             {            if (possessive) continue;
2719             RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);  
2720             if (rrc != MATCH_NOMATCH) RRETURN(rrc);            for(;;)
2721             eptr -= length;              {
2722             }              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);
2723            RRETURN(MATCH_NOMATCH);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2724                if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
2725    #ifdef SUPPORT_UCP
2726                eptr--;
2727                BACKCHAR(eptr);
2728    #else   /* without SUPPORT_UCP */
2729                eptr -= length;
2730    #endif  /* SUPPORT_UCP */
2731                }
2732            }            }
2733          /* Control never gets here */          /* Control never gets here */
2734          }          }
# Line 1990  for (;;) Line 2741  for (;;)
2741  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2742    
2743      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
2744        {  
2745        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
2746    
2747      /* 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
2748      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 2011  for (;;) Line 2760  for (;;)
2760        {        {
2761        fc = md->lcc[fc];        fc = md->lcc[fc];
2762        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2763            {
2764            if (eptr >= md->end_subject)
2765              {
2766              SCHECK_PARTIAL();
2767              RRETURN(MATCH_NOMATCH);
2768              }
2769          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
2770            }
2771        if (min == max) continue;        if (min == max) continue;
2772        if (minimize)        if (minimize)
2773          {          {
2774          for (fi = min;; fi++)          for (fi = min;; fi++)
2775            {            {
2776            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);
2777            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2778            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) RRETURN(MATCH_NOMATCH);
2779                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
2780                {
2781                SCHECK_PARTIAL();
2782              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2783                }
2784              if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
2785            }            }
2786          /* Control never gets here */          /* Control never gets here */
2787          }          }
2788        else        else  /* Maximize */
2789          {          {
2790          pp = eptr;          pp = eptr;
2791          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2792            {            {
2793            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
2794                {
2795                SCHECK_PARTIAL();
2796                break;
2797                }
2798              if (fc != md->lcc[*eptr]) break;
2799            eptr++;            eptr++;
2800            }            }
2801    
2802            if (possessive) continue;
2803    
2804          while (eptr >= pp)          while (eptr >= pp)
2805            {            {
2806            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);
2807            eptr--;            eptr--;
2808            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2809            }            }
# Line 2048  for (;;) Line 2816  for (;;)
2816    
2817      else      else
2818        {        {
2819        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
2820            {
2821            if (eptr >= md->end_subject)
2822              {
2823              SCHECK_PARTIAL();
2824              RRETURN(MATCH_NOMATCH);
2825              }
2826            if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
2827            }
2828    
2829        if (min == max) continue;        if (min == max) continue;
2830    
2831        if (minimize)        if (minimize)
2832          {          {
2833          for (fi = min;; fi++)          for (fi = min;; fi++)
2834            {            {
2835            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);
2836            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2837            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) RRETURN(MATCH_NOMATCH);
2838              if (eptr >= md->end_subject)
2839                {
2840                SCHECK_PARTIAL();
2841              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2842                }
2843              if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
2844            }            }
2845          /* Control never gets here */          /* Control never gets here */
2846          }          }
2847        else        else  /* Maximize */
2848          {          {
2849          pp = eptr;          pp = eptr;
2850          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2851            {            {
2852            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
2853                {
2854                SCHECK_PARTIAL();
2855                break;
2856                }
2857              if (fc != *eptr) break;
2858            eptr++;            eptr++;
2859            }            }
2860            if (possessive) continue;
2861    
2862          while (eptr >= pp)          while (eptr >= pp)
2863            {            {
2864            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);
2865            eptr--;            eptr--;
2866            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2867            }            }
# Line 2084  for (;;) Line 2874  for (;;)
2874      checking can be multibyte. */      checking can be multibyte. */
2875    
2876      case OP_NOT:      case OP_NOT:
2877      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2878          {
2879          SCHECK_PARTIAL();
2880          RRETURN(MATCH_NOMATCH);
2881          }
2882      ecode++;      ecode++;
2883      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2884      if ((ims & PCRE_CASELESS) != 0)      if ((ims & PCRE_CASELESS) != 0)
# Line 2121  for (;;) Line 2915  for (;;)
2915      ecode += 3;      ecode += 3;
2916      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
2917    
2918        case OP_NOTPOSSTAR:
2919        possessive = TRUE;
2920        min = 0;
2921        max = INT_MAX;
2922        ecode++;
2923        goto REPEATNOTCHAR;
2924    
2925        case OP_NOTPOSPLUS:
2926        possessive = TRUE;
2927        min = 1;
2928        max = INT_MAX;
2929        ecode++;
2930        goto REPEATNOTCHAR;
2931    
2932        case OP_NOTPOSQUERY:
2933        possessive = TRUE;
2934        min = 0;
2935        max = 1;
2936        ecode++;
2937        goto REPEATNOTCHAR;
2938    
2939        case OP_NOTPOSUPTO:
2940        possessive = TRUE;
2941        min = 0;
2942        max = GET2(ecode, 1);
2943        ecode += 3;
2944        goto REPEATNOTCHAR;
2945    
2946      case OP_NOTSTAR:      case OP_NOTSTAR:
2947      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
2948      case OP_NOTPLUS:      case OP_NOTPLUS:
# Line 2133  for (;;) Line 2955  for (;;)
2955      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2956      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
2957    
2958      /* 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. */  
2959    
2960      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
2961      fc = *ecode++;      fc = *ecode++;
2962    
2963      /* 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 2160  for (;;) Line 2979  for (;;)
2979        /* UTF-8 mode */        /* UTF-8 mode */
2980        if (utf8)        if (utf8)
2981          {          {
2982          register int d;          register unsigned int d;
2983          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2984            {            {
2985              if (eptr >= md->end_subject)
2986                {
2987                SCHECK_PARTIAL();
2988                RRETURN(MATCH_NOMATCH);
2989                }
2990            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
2991            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
2992            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) RRETURN(MATCH_NOMATCH);
# Line 2174  for (;;) Line 2998  for (;;)
2998        /* Not UTF-8 mode */        /* Not UTF-8 mode */
2999          {          {
3000          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3001              {
3002              if (eptr >= md->end_subject)
3003                {
3004                SCHECK_PARTIAL();
3005                RRETURN(MATCH_NOMATCH);
3006                }
3007            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
3008              }
3009          }          }
3010    
3011        if (min == max) continue;        if (min == max) continue;
# Line 2185  for (;;) Line 3016  for (;;)
3016          /* UTF-8 mode */          /* UTF-8 mode */
3017          if (utf8)          if (utf8)
3018            {            {
3019            register int d;            register unsigned int d;
3020            for (fi = min;; fi++)            for (fi = min;; fi++)
3021              {              {
3022              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);
3023              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3024                if (fi >= max) RRETURN(MATCH_NOMATCH);
3025                if (eptr >= md->end_subject)
3026                  {
3027                  SCHECK_PARTIAL();
3028                  RRETURN(MATCH_NOMATCH);
3029                  }
3030              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3031              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3032              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) RRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3033              }              }
3034            }            }
3035          else          else
# Line 2202  for (;;) Line 3038  for (;;)
3038            {            {
3039            for (fi = min;; fi++)            for (fi = min;; fi++)
3040              {              {
3041              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);
3042              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3043              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) RRETURN(MATCH_NOMATCH);
3044                if (eptr >= md->end_subject)
3045                  {
3046                  SCHECK_PARTIAL();
3047                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3048                  }
3049                if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);
3050              }              }
3051            }            }
3052          /* Control never gets here */          /* Control never gets here */
# Line 2221  for (;;) Line 3062  for (;;)
3062          /* UTF-8 mode */          /* UTF-8 mode */
3063          if (utf8)          if (utf8)
3064            {            {
3065            register int d;            register unsigned int d;
3066            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3067              {              {
3068              int len = 1;              int len = 1;
3069              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3070                  {
3071                  SCHECK_PARTIAL();
3072                  break;
3073                  }
3074              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3075              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3076              if (fc == d) break;              if (fc == d) break;
3077              eptr += len;              eptr += len;
3078              }              }
3079            for(;;)          if (possessive) continue;
3080            for(;;)
3081              {              {
3082              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);
3083              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3084              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3085              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2245  for (;;) Line 3091  for (;;)
3091            {            {
3092            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3093              {              {
3094              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3095                  {
3096                  SCHECK_PARTIAL();
3097                  break;
3098                  }
3099                if (fc == md->lcc[*eptr]) break;
3100              eptr++;              eptr++;
3101              }              }
3102              if (possessive) continue;
3103            while (eptr >= pp)            while (eptr >= pp)
3104              {              {
3105              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);
3106              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3107              eptr--;              eptr--;
3108              }              }
# Line 2269  for (;;) Line 3121  for (;;)
3121        /* UTF-8 mode */        /* UTF-8 mode */
3122        if (utf8)        if (utf8)
3123          {          {
3124          register int d;          register unsigned int d;
3125          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3126            {            {
3127              if (eptr >= md->end_subject)
3128                {
3129                SCHECK_PARTIAL();
3130                RRETURN(MATCH_NOMATCH);
3131                }
3132            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3133            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) RRETURN(MATCH_NOMATCH);
3134            }            }
# Line 2281  for (;;) Line 3138  for (;;)
3138        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3139          {          {
3140          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3141              {
3142              if (eptr >= md->end_subject)
3143                {
3144                SCHECK_PARTIAL();
3145                RRETURN(MATCH_NOMATCH);
3146                }
3147            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);
3148              }
3149          }          }
3150    
3151        if (min == max) continue;        if (min == max) continue;
# Line 2292  for (;;) Line 3156  for (;;)
3156          /* UTF-8 mode */          /* UTF-8 mode */
3157          if (utf8)          if (utf8)
3158            {            {
3159            register int d;            register unsigned int d;
3160            for (fi = min;; fi++)            for (fi = min;; fi++)
3161              {              {
3162              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);
3163              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3164              GETCHARINC(d, eptr);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3165              if (fi >= max || eptr >= md->end_subject || fc == d)              if (eptr >= md->end_subject)
3166                  {
3167                  SCHECK_PARTIAL();
3168                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3169                  }
3170                GETCHARINC(d, eptr);
3171                if (fc == d) RRETURN(MATCH_NOMATCH);
3172              }              }
3173            }            }
3174          else          else
# Line 2308  for (;;) Line 3177  for (;;)
3177            {            {
3178            for (fi = min;; fi++)            for (fi = min;; fi++)
3179              {              {
3180              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);
3181              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3182              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) RRETURN(MATCH_NOMATCH);
3183                if (eptr >= md->end_subject)
3184                  {
3185                  SCHECK_PARTIAL();
3186                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3187                  }
3188                if (fc == *eptr++) RRETURN(MATCH_NOMATCH);
3189              }              }
3190            }            }
3191          /* Control never gets here */          /* Control never gets here */
# Line 2327  for (;;) Line 3201  for (;;)
3201          /* UTF-8 mode */          /* UTF-8 mode */
3202          if (utf8)          if (utf8)
3203            {            {
3204            register int d;            register unsigned int d;
3205            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3206              {              {
3207              int len = 1;              int len = 1;
3208              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3209                  {
3210                  SCHECK_PARTIAL();
3211                  break;
3212                  }
3213              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3214              if (fc == d) break;              if (fc == d) break;
3215              eptr += len;              eptr += len;
3216              }              }
3217              if (possessive) continue;
3218            for(;;)            for(;;)
3219              {              {
3220              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);
3221              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3222              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3223              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2350  for (;;) Line 3229  for (;;)
3229            {            {
3230            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3231              {              {
3232              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3233                  {
3234                  SCHECK_PARTIAL();
3235                  break;
3236                  }
3237                if (fc == *eptr) break;
3238              eptr++;              eptr++;
3239              }              }
3240              if (possessive) continue;
3241            while (eptr >= pp)            while (eptr >= pp)
3242              {              {
3243              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);
3244              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3245              eptr--;              eptr--;
3246              }              }
# Line 2384  for (;;) Line 3269  for (;;)
3269      ecode += 3;      ecode += 3;
3270      goto REPEATTYPE;      goto REPEATTYPE;
3271    
3272        case OP_TYPEPOSSTAR:
3273        possessive = TRUE;
3274        min = 0;
3275        max = INT_MAX;
3276        ecode++;
3277        goto REPEATTYPE;
3278    
3279        case OP_TYPEPOSPLUS:
3280        possessive = TRUE;
3281        min = 1;
3282        max = INT_MAX;
3283        ecode++;
3284        goto REPEATTYPE;
3285    
3286        case OP_TYPEPOSQUERY:
3287        possessive = TRUE;
3288        min = 0;
3289        max = 1;
3290        ecode++;
3291        goto REPEATTYPE;
3292    
3293        case OP_TYPEPOSUPTO:
3294        possessive = TRUE;
3295        min = 0;
3296        max = GET2(ecode, 1);
3297        ecode += 3;
3298        goto REPEATTYPE;
3299    
3300      case OP_TYPESTAR:      case OP_TYPESTAR:
3301      case OP_TYPEMINSTAR:      case OP_TYPEMINSTAR:
3302      case OP_TYPEPLUS:      case OP_TYPEPLUS:
# Line 2408  for (;;) Line 3321  for (;;)
3321        {        {
3322        prop_fail_result = ctype == OP_NOTPROP;        prop_fail_result = ctype == OP_NOTPROP;
3323        prop_type = *ecode++;        prop_type = *ecode++;
3324        if (prop_type >= 128)        prop_value = *ecode++;
         {  
         prop_test_against = prop_type - 128;  
         prop_test_variable = &prop_category;  
         }  
       else  
         {  
         prop_test_against = prop_type;  
         prop_test_variable = &prop_chartype;  
         }  
3325        }        }
3326      else prop_type = -1;      else prop_type = -1;
3327  #endif  #endif
3328    
3329      /* First, ensure the minimum number of matches are present. Use inline      /* First, ensure the minimum number of matches are present. Use inline
3330      code for maximizing the speed, and do the type test once at the start      code for maximizing the speed, and do the type test once at the start
3331      (i.e. keep it out of the loop). Also we can test that there are at least      (i.e. keep it out of the loop). Separate the UTF-8 code completely as that
     the minimum number of bytes before we start. This isn't as effective in  
     UTF-8 mode, but it does no harm. Separate the UTF-8 code completely as that  
3332      is tidier. Also separate the UCP code, which can be the same for both UTF-8      is tidier. Also separate the UCP code, which can be the same for both UTF-8
3333      and single-bytes. */      and single-bytes. */
3334    
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3335      if (min > 0)      if (min > 0)
3336        {        {
3337  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3338        if (prop_type > 0)        if (prop_type >= 0)
3339          {          {
3340          for (i = 1; i <= min; i++)          switch(prop_type)
3341            {            {
3342            GETCHARINC(c, eptr);            case PT_ANY:
3343            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            if (prop_fail_result) RRETURN(MATCH_NOMATCH);
3344            if ((*prop_test_variable == prop_test_against) == prop_fail_result)            for (i = 1; i <= min; i++)
3345              RRETURN(MATCH_NOMATCH);              {
3346                if (eptr >= md->end_subject)
3347                  {
3348                  SCHECK_PARTIAL();
3349                  RRETURN(MATCH_NOMATCH);
3350                  }
3351                GETCHARINCTEST(c, eptr);
3352                }
3353              break;
3354    
3355              case PT_LAMP:
3356              for (i = 1; i <= min; i++)
3357                {
3358                if (eptr >= md->end_subject)
3359                  {
3360                  SCHECK_PARTIAL();
3361                  RRETURN(MATCH_NOMATCH);
3362                  }
3363                GETCHARINCTEST(c, eptr);
3364                prop_chartype = UCD_CHARTYPE(c);
3365                if ((prop_chartype == ucp_Lu ||
3366                     prop_chartype == ucp_Ll ||
3367                     prop_chartype == ucp_Lt) == prop_fail_result)
3368                  RRETURN(MATCH_NOMATCH);
3369                }
3370              break;
3371    
3372              case PT_GC:
3373              for (i = 1; i <= min; i++)
3374                {
3375                if (eptr >= md->end_subject)
3376                  {
3377                  SCHECK_PARTIAL();
3378                  RRETURN(MATCH_NOMATCH);
3379                  }
3380                GETCHARINCTEST(c, eptr);
3381                prop_category = UCD_CATEGORY(c);
3382                if ((prop_category == prop_value) == prop_fail_result)
3383                  RRETURN(MATCH_NOMATCH);
3384                }
3385              break;
3386    
3387              case PT_PC:
3388              for (i = 1; i <= min; i++)
3389                {
3390                if (eptr >= md->end_subject)
3391                  {
3392                  SCHECK_PARTIAL();
3393                  RRETURN(MATCH_NOMATCH);
3394                  }
3395                GETCHARINCTEST(c, eptr);
3396                prop_chartype = UCD_CHARTYPE(c);
3397                if ((prop_chartype == prop_value) == prop_fail_result)
3398                  RRETURN(MATCH_NOMATCH);
3399                }
3400              break;
3401    
3402              case PT_SC:
3403              for (i = 1; i <= min; i++)
3404                {
3405                if (eptr >= md->end_subject)
3406                  {
3407                  SCHECK_PARTIAL();
3408                  RRETURN(MATCH_NOMATCH);
3409                  }
3410                GETCHARINCTEST(c, eptr);
3411                prop_script = UCD_SCRIPT(c);
3412                if ((prop_script == prop_value) == prop_fail_result)
3413                  RRETURN(MATCH_NOMATCH);
3414                }
3415              break;
3416    
3417              default:
3418              RRETURN(PCRE_ERROR_INTERNAL);
3419            }            }
3420          }          }
3421    
# Line 2452  for (;;) Line 3426  for (;;)
3426          {          {
3427          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3428            {            {
3429              if (eptr >= md->end_subject)
3430                {
3431                SCHECK_PARTIAL();
3432                RRETURN(MATCH_NOMATCH);
3433                }
3434            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3435            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            prop_category = UCD_CATEGORY(c);
3436            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);
3437            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3438              {              {
3439              int len = 1;              int len = 1;
3440              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr;
3441                {                else { GETCHARLEN(c, eptr, len); }
3442                GETCHARLEN(c, eptr, len);              prop_category = UCD_CATEGORY(c);
               }  
             prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);  
3443              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
3444              eptr += len;              eptr += len;
3445              }              }
# Line 2480  for (;;) Line 3457  for (;;)
3457          case OP_ANY:          case OP_ANY:
3458          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3459            {            {
3460            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject)
3461               (*eptr++ == NEWLINE && (ims & PCRE_DOTALL) == 0))              {
3462                SCHECK_PARTIAL();
3463                RRETURN(MATCH_NOMATCH);
3464                }
3465              if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
3466              eptr++;
3467              while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
3468              }
3469            break;
3470    
3471            case OP_ALLANY:
3472            for (i = 1; i <= min; i++)
3473              {
3474              if (eptr >= md->end_subject)
3475                {
3476                SCHECK_PARTIAL();
3477              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3478                }
3479              eptr++;
3480            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
3481            }            }
3482          break;          break;
3483    
3484          case OP_ANYBYTE:          case OP_ANYBYTE:
3485            if (eptr > md->end_subject - min) RRETURN(MATCH_NOMATCH);
3486          eptr += min;          eptr += min;
3487          break;          break;
3488    
3489            case OP_ANYNL:
3490            for (i = 1; i <= min; i++)
3491              {
3492              if (eptr >= md->end_subject)
3493                {
3494                SCHECK_PARTIAL();
3495                RRETURN(MATCH_NOMATCH);
3496                }
3497              GETCHARINC(c, eptr);
3498              switch(c)
3499                {
3500                default: RRETURN(MATCH_NOMATCH);
3501                case 0x000d:
3502                if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3503                break;
3504    
3505                case 0x000a:
3506                break;
3507    
3508                case 0x000b:
3509                case 0x000c:
3510                case 0x0085:
3511                case 0x2028:
3512                case 0x2029:
3513                if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
3514                break;
3515                }
3516              }
3517            break;
3518    
3519            case OP_NOT_HSPACE:
3520            for (i = 1; i <= min; i++)
3521              {
3522              if (eptr >= md->end_subject)
3523                {
3524                SCHECK_PARTIAL();
3525                RRETURN(MATCH_NOMATCH);
3526                }
3527              GETCHARINC(c, eptr);
3528              switch(c)
3529                {
3530                default: break;
3531                case 0x09:      /* HT */
3532                case 0x20:      /* SPACE */
3533                case 0xa0:      /* NBSP */
3534                case 0x1680:    /* OGHAM SPACE MARK */
3535                case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3536                case 0x2000:    /* EN QUAD */
3537                case 0x2001:    /* EM QUAD */
3538                case 0x2002:    /* EN SPACE */
3539                case 0x2003:    /* EM SPACE */
3540                case 0x2004:    /* THREE-PER-EM SPACE */
3541                case 0x2005:    /* FOUR-PER-EM SPACE */
3542                case 0x2006:    /* SIX-PER-EM SPACE */
3543                case 0x2007:    /* FIGURE SPACE */
3544                case 0x2008:    /* PUNCTUATION SPACE */
3545                case 0x2009:    /* THIN SPACE */
3546                case 0x200A:    /* HAIR SPACE */
3547                case 0x202f:    /* NARROW NO-BREAK SPACE */
3548                case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3549                case 0x3000:    /* IDEOGRAPHIC SPACE */
3550                RRETURN(MATCH_NOMATCH);
3551                }
3552              }
3553            break;
3554    
3555            case OP_HSPACE:
3556            for (i = 1; i <= min; i++)
3557              {
3558              if (eptr >= md->end_subject)
3559                {
3560                SCHECK_PARTIAL();
3561                RRETURN(MATCH_NOMATCH);
3562                }
3563              GETCHARINC(c, eptr);
3564              switch(c)
3565                {
3566                default: RRETURN(MATCH_NOMATCH);
3567                case 0x09:      /* HT */
3568                case 0x20:      /* SPACE */
3569                case 0xa0:      /* NBSP */
3570                case 0x1680:    /* OGHAM SPACE MARK */
3571                case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3572                case 0x2000:    /* EN QUAD */
3573                case 0x2001:    /* EM QUAD */
3574                case 0x2002:    /* EN SPACE */
3575                case 0x2003:    /* EM SPACE */
3576                case 0x2004:    /* THREE-PER-EM SPACE */
3577                case 0x2005:    /* FOUR-PER-EM SPACE */
3578                case 0x2006:    /* SIX-PER-EM SPACE */
3579                case 0x2007:    /* FIGURE SPACE */
3580                case 0x2008:    /* PUNCTUATION SPACE */
3581                case 0x2009:    /* THIN SPACE */
3582                case 0x200A:    /* HAIR SPACE */
3583                case 0x202f:    /* NARROW NO-BREAK SPACE */
3584                case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3585                case 0x3000:    /* IDEOGRAPHIC SPACE */
3586                break;
3587                }
3588              }
3589            break;
3590    
3591            case OP_NOT_VSPACE:
3592            for (i = 1; i <= min; i++)
3593              {
3594              if (eptr >= md->end_subject)
3595                {
3596                SCHECK_PARTIAL();
3597                RRETURN(MATCH_NOMATCH);
3598                }
3599              GETCHARINC(c, eptr);
3600              switch(c)
3601                {
3602                default: break;
3603                case 0x0a:      /* LF */
3604                case 0x0b:      /* VT */
3605                case 0x0c:      /* FF */
3606                case 0x0d:      /* CR */
3607                case 0x85:      /* NEL */
3608                case 0x2028:    /* LINE SEPARATOR */
3609                case 0x2029:    /* PARAGRAPH SEPARATOR */
3610                RRETURN(MATCH_NOMATCH);
3611                }
3612              }
3613            break;
3614    
3615            case OP_VSPACE:
3616            for (i = 1; i <= min; i++)
3617              {
3618              if (eptr >= md->end_subject)
3619                {
3620                SCHECK_PARTIAL();
3621                RRETURN(MATCH_NOMATCH);
3622                }
3623              GETCHARINC(c, eptr);
3624              switch(c)
3625                {
3626                default: RRETURN(MATCH_NOMATCH);
3627                case 0x0a:      /* LF */
3628                case 0x0b:      /* VT */
3629                case 0x0c:      /* FF */
3630                case 0x0d:      /* CR */
3631                case 0x85:      /* NEL */
3632                case 0x2028:    /* LINE SEPARATOR */
3633                case 0x2029:    /* PARAGRAPH SEPARATOR */
3634                break;
3635                }
3636              }
3637            break;
3638    
3639          case OP_NOT_DIGIT:          case OP_NOT_DIGIT:
3640          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3641            {            {
3642            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3643                {
3644                SCHECK_PARTIAL();
3645                RRETURN(MATCH_NOMATCH);
3646                }
3647            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
3648            if (c < 128 && (md->ctypes[c] & ctype_digit) != 0)            if (c < 128 && (md->ctypes[c] & ctype_digit) != 0)
3649              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
# Line 2504  for (;;) Line 3653  for (;;)
3653          case OP_DIGIT:          case OP_DIGIT:
3654          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3655            {            {
3656            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject)
3657               *eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0)              {
3658                SCHECK_PARTIAL();
3659                RRETURN(MATCH_NOMATCH);
3660                }
3661              if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0)
3662              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3663            /* No need to skip more bytes - we know it's a 1-byte character */            /* No need to skip more bytes - we know it's a 1-byte character */
3664            }            }
# Line 2514  for (;;) Line 3667  for (;;)
3667          case OP_NOT_WHITESPACE:          case OP_NOT_WHITESPACE:
3668          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3669            {            {
3670            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject)
3671               (*eptr < 128 && (md->ctypes[*eptr++] & ctype_space) != 0))              {
3672                SCHECK_PARTIAL();
3673              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3674            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;              }
3675              if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0)
3676                RRETURN(MATCH_NOMATCH);
3677              while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
3678            }            }
3679          break;          break;
3680    
3681          case OP_WHITESPACE:          case OP_WHITESPACE:
3682          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3683            {            {
3684            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject)
3685               *eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0)              {
3686                SCHECK_PARTIAL();
3687                RRETURN(MATCH_NOMATCH);
3688                }
3689              if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0)
3690              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3691            /* No need to skip more bytes - we know it's a 1-byte character */            /* No need to skip more bytes - we know it's a 1-byte character */
3692            }            }
# Line 2535  for (;;) Line 3696  for (;;)
3696          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3697            {            {
3698            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject ||
3699               (*eptr < 128 && (md->ctypes[*eptr++] & ctype_word) != 0))               (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0))
3700              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3701            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
3702            }            }
3703          break;          break;
3704    
3705          case OP_WORDCHAR:          case OP_WORDCHAR:
3706          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3707            {            {
3708            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject)
3709               *eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0)              {
3710                SCHECK_PARTIAL();
3711                RRETURN(MATCH_NOMATCH);
3712                }
3713              if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0)
3714              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3715            /* No need to skip more bytes - we know it's a 1-byte character */            /* No need to skip more bytes - we know it's a 1-byte character */
3716            }            }
# Line 2564  for (;;) Line 3729  for (;;)
3729        switch(ctype)        switch(ctype)
3730          {          {
3731          case OP_ANY:          case OP_ANY:
3732          if ((ims & PCRE_DOTALL) == 0)          for (i = 1; i <= min; i++)
3733            {            {
3734            for (i = 1; i <= min; i++)            if (eptr >= md->end_subject)
3735              if (*eptr++ == NEWLINE) RRETURN(MATCH_NOMATCH);              {
3736                SCHECK_PARTIAL();
3737                RRETURN(MATCH_NOMATCH);
3738                }
3739              if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
3740              eptr++;
3741            }            }
         else eptr += min;  
3742          break;          break;
3743    
3744          case OP_ANYBYTE:          case OP_ALLANY:
3745            if (eptr > md->end_subject - min)
3746              {
3747              SCHECK_PARTIAL();
3748              RRETURN(MATCH_NOMATCH);
3749              }
3750          eptr += min;          eptr += min;
3751          break;          break;
3752    
3753          case OP_NOT_DIGIT:          case OP_ANYBYTE:
3754          for (i = 1; i <= min; i++)          if (eptr > md->end_subject - min)
3755            if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH);            {
3756          break;            SCHECK_PARTIAL();
3757              RRETURN(MATCH_NOMATCH);
3758              }
3759            eptr += min;
3760            break;
3761    
3762            case OP_ANYNL:
3763            for (i = 1; i <= min; i++)
3764              {
3765              if (eptr >= md->end_subject)
3766                {
3767                SCHECK_PARTIAL();
3768                RRETURN(MATCH_NOMATCH);
3769                }
3770              switch(*eptr++)
3771                {
3772                default: RRETURN(MATCH_NOMATCH);
3773                case 0x000d:
3774                if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3775                break;
3776                case 0x000a:
3777                break;
3778    
3779                case 0x000b:
3780                case 0x000c:
3781                case 0x0085:
3782                if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
3783                break;
3784                }
3785              }
3786            break;
3787    
3788            case OP_NOT_HSPACE:
3789            for (i = 1; i <= min; i++)
3790              {
3791              if (eptr >= md->end_subject)
3792                {
3793                SCHECK_PARTIAL();
3794                RRETURN(MATCH_NOMATCH);
3795                }
3796              switch(*eptr++)
3797                {
3798                default: break;
3799                case 0x09:      /* HT */
3800                case 0x20:      /* SPACE */
3801                case 0xa0:      /* NBSP */
3802                RRETURN(MATCH_NOMATCH);
3803                }
3804              }
3805            break;
3806    
3807            case OP_HSPACE:
3808            for (i = 1; i <= min; i++)
3809              {
3810              if (eptr >= md->end_subject)
3811                {
3812                SCHECK_PARTIAL();
3813                RRETURN(MATCH_NOMATCH);
3814                }
3815              switch(*eptr++)
3816                {
3817                default: RRETURN(MATCH_NOMATCH);
3818                case 0x09:      /* HT */
3819                case 0x20:      /* SPACE */
3820                case 0xa0:      /* NBSP */
3821                break;
3822                }
3823              }
3824            break;
3825    
3826            case OP_NOT_VSPACE:
3827            for (i = 1; i <= min; i++)
3828              {
3829              if (eptr >= md->end_subject)
3830                {
3831                SCHECK_PARTIAL();
3832                RRETURN(MATCH_NOMATCH);
3833                }
3834              switch(*eptr++)
3835                {
3836                default: break;
3837                case 0x0a:      /* LF */
3838                case 0x0b:      /* VT */
3839                case 0x0c:      /* FF */
3840                case 0x0d:      /* CR */
3841                case 0x85:      /* NEL */
3842                RRETURN(MATCH_NOMATCH);
3843                }
3844              }
3845            break;
3846    
3847            case OP_VSPACE:
3848            for (i = 1; i <= min; i++)
3849              {
3850              if (eptr >= md->end_subject)
3851                {
3852                SCHECK_PARTIAL();
3853                RRETURN(MATCH_NOMATCH);
3854                }
3855              switch(*eptr++)
3856                {
3857                default: RRETURN(MATCH_NOMATCH);
3858                case 0x0a:      /* LF */
3859                case 0x0b:      /* VT */
3860                case 0x0c:      /* FF */
3861                case 0x0d:      /* CR */
3862                case 0x85:      /* NEL */
3863                break;
3864                }
3865              }
3866            break;
3867    
3868            case OP_NOT_DIGIT:
3869            for (i = 1; i <= min; i++)
3870              {
3871              if (eptr >= md->end_subject)
3872                {
3873                SCHECK_PARTIAL();
3874                RRETURN(MATCH_NOMATCH);
3875                }
3876              if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH);
3877              }
3878            break;
3879    
3880          case OP_DIGIT:          case OP_DIGIT:
3881          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3882              {
3883              if (eptr >= md->end_subject)
3884                {
3885                SCHECK_PARTIAL();
3886                RRETURN(MATCH_NOMATCH);
3887                }
3888            if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH);            if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH);
3889              }
3890          break;          break;
3891    
3892          case OP_NOT_WHITESPACE:          case OP_NOT_WHITESPACE:
3893          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3894              {
3895              if (eptr >= md->end_subject)
3896                {
3897                SCHECK_PARTIAL();
3898                RRETURN(MATCH_NOMATCH);
3899                }
3900            if ((md->ctypes[*eptr++] & ctype_space) != 0) RRETURN(MATCH_NOMATCH);            if ((md->ctypes[