/[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 482 by ph10, Mon Jan 4 15:55:46 2010 UTC revision 1015 by ph10, Sun Aug 26 16:07:14 2012 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) 1997-2010 University of Cambridge             Copyright (c) 1997-2012 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 37  POSSIBILITY OF SUCH DAMAGE. Line 37  POSSIBILITY OF SUCH DAMAGE.
37  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
38  */  */
39    
   
40  /* This module contains pcre_exec(), the externally visible function that does  /* This module contains pcre_exec(), the externally visible function that does
41  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
42  possible. There are also some static supporting functions. */  possible. There are also some static supporting functions. */
# Line 57  possible. There are also some static sup Line 56  possible. There are also some static sup
56  #undef min  #undef min
57  #undef max  #undef max
58    
59  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
60    of call to match(). We do it this way to save on using another stack variable,
61    as stack usage is to be discouraged. */
62    
63  #define match_condassert     0x01  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
64  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
65    
66  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
67  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 71  defined PCRE_ERROR_xxx codes, which are Line 72  defined PCRE_ERROR_xxx codes, which are
72  /* Special internal returns from the match() function. Make them sufficiently  /* Special internal returns from the match() function. Make them sufficiently
73  negative to avoid the external error codes. */  negative to avoid the external error codes. */
74    
75  #define MATCH_COMMIT       (-999)  #define MATCH_ACCEPT       (-999)
76  #define MATCH_PRUNE        (-998)  #define MATCH_COMMIT       (-998)
77  #define MATCH_SKIP         (-997)  #define MATCH_KETRPOS      (-997)
78  #define MATCH_THEN         (-996)  #define MATCH_ONCE         (-996)
79    #define MATCH_PRUNE        (-995)
80    #define MATCH_SKIP         (-994)
81    #define MATCH_SKIP_ARG     (-993)
82    #define MATCH_THEN         (-992)
83    
84  /* 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.
85  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,
# Line 107  Returns:     nothing Line 112  Returns:     nothing
112  */  */
113    
114  static void  static void
115  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)  pchars(const pcre_uchar *p, int length, BOOL is_subject, match_data *md)
116  {  {
117  unsigned int c;  unsigned int c;
118  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;
# Line 122  while (length-- > 0) Line 127  while (length-- > 0)
127  *          Match a back-reference                *  *          Match a back-reference                *
128  *************************************************/  *************************************************/
129    
130  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
131  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
132    the length passed is zero. Note that in caseless UTF-8 mode, the number of
133    subject bytes matched may be different to the number of reference bytes.
134    
135  Arguments:  Arguments:
136    offset      index into the offset vector    offset      index into the offset vector
137    eptr        points into the subject    eptr        pointer into the subject
138    length      length to be matched    length      length of reference to be matched (number of bytes)
139    md          points to match data block    md          points to match data block
140    ims         the ims flags    caseless    TRUE if caseless
141    
142  Returns:      TRUE if matched  Returns:      >= 0 the number of subject bytes matched
143                  -1 no match
144                  -2 partial match; always given if at end subject
145  */  */
146    
147  static BOOL  static int
148  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register PCRE_PUCHAR eptr, int length, match_data *md,
149    unsigned long int ims)    BOOL caseless)
150  {  {
151  USPTR p = md->start_subject + md->offset_vector[offset];  PCRE_PUCHAR eptr_start = eptr;
152    register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
153    
154  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
155  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 154  pchars(p, length, FALSE, md); Line 164  pchars(p, length, FALSE, md);
164  printf("\n");  printf("\n");
165  #endif  #endif
166    
167  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible - in that
168    case the length is passed as zero). */
169    
170  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
171    
172  /* Separate the caseless case for speed. In UTF-8 mode we can only do this  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
173  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
174  ASCII characters. */  ASCII characters. */
175    
176  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
177    {    {
178  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
179  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
180    if (md->utf8)    if (md->utf)
181      {      {
182      USPTR endptr = eptr + length;      /* Match characters up to the end of the reference. NOTE: the number of
183      while (eptr < endptr)      bytes matched may differ, because there are some characters whose upper and
184        lower case versions code as different numbers of bytes. For example, U+023A
185        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
186        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
187        the latter. It is important, therefore, to check the length along the
188        reference, not along the subject (earlier code did this wrong). */
189    
190        PCRE_PUCHAR endptr = p + length;
191        while (p < endptr)
192        {        {
193        int c, d;        int c, d;
194          if (eptr >= md->end_subject) return -2;   /* Partial match */
195        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
196        GETCHARINC(d, p);        GETCHARINC(d, p);
197        if (c != d && c != UCD_OTHERCASE(d)) return FALSE;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
198        }        }
199      }      }
200    else    else
# Line 183  if ((ims & PCRE_CASELESS) != 0) Line 203  if ((ims & PCRE_CASELESS) != 0)
203    
204    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
205    is no UCP support. */    is no UCP support. */
206        {
207    while (length-- > 0)      while (length-- > 0)
208      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }        {
209          if (eptr >= md->end_subject) return -2;   /* Partial match */
210          if (TABLE_GET(*p, md->lcc, *p) != TABLE_GET(*eptr, md->lcc, *eptr)) return -1;
211          p++;
212          eptr++;
213          }
214        }
215    }    }
216    
217  /* In the caseful case, we can just compare the bytes, whether or not we  /* In the caseful case, we can just compare the bytes, whether or not we
218  are in UTF-8 mode. */  are in UTF-8 mode. */
219    
220  else  else
221    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
222      while (length-- > 0)
223        {
224        if (eptr >= md->end_subject) return -2;   /* Partial match */
225        if (*p++ != *eptr++) return -1;
226        }
227      }
228    
229  return TRUE;  return (int)(eptr - eptr_start);
230  }  }
231    
232    
# Line 245  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 277  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
277         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
278         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
279         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
280         RM51,  RM52, RM53, RM54 };         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
281           RM61,  RM62, RM63, RM64, RM65, RM66 };
282    
283  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
284  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
285  actuall used in this definition. */  actually used in this definition. */
286    
287  #ifndef NO_RECURSE  #ifndef NO_RECURSE
288  #define REGISTER register  #define REGISTER register
289    
290  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
291  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
292    { \    { \
293    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
294    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \
295    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
296    }    }
297  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 267  actuall used in this definition. */ Line 300  actuall used in this definition. */
300    return ra; \    return ra; \
301    }    }
302  #else  #else
303  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
304    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1)
305  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
306  #endif  #endif
307    
# Line 281  argument of match(), which never changes Line 314  argument of match(), which never changes
314    
315  #define REGISTER  #define REGISTER
316    
317  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
318    {\    {\
319    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = frame->Xnextframe;\
320    frame->Xwhere = rw; \    if (newframe == NULL)\
321        {\
322        newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\
323        if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
324        newframe->Xnextframe = NULL;\
325        frame->Xnextframe = newframe;\
326        }\
327      frame->Xwhere = rw;\
328    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
329    newframe->Xecode = rb;\    newframe->Xecode = rb;\
330    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
331    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
332    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
333    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
334    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
335    frame = newframe;\    frame = newframe;\
# Line 303  argument of match(), which never changes Line 341  argument of match(), which never changes
341    
342  #define RRETURN(ra)\  #define RRETURN(ra)\
343    {\    {\
344    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
345    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
   (pcre_stack_free)(newframe);\  
346    if (frame != NULL)\    if (frame != NULL)\
347      {\      {\
348      rrc = ra;\      rrc = ra;\
# Line 319  argument of match(), which never changes Line 356  argument of match(), which never changes
356    
357  typedef struct heapframe {  typedef struct heapframe {
358    struct heapframe *Xprevframe;    struct heapframe *Xprevframe;
359      struct heapframe *Xnextframe;
360    
361    /* Function arguments that may change */    /* Function arguments that may change */
362    
363    USPTR Xeptr;    PCRE_PUCHAR Xeptr;
364    const uschar *Xecode;    const pcre_uchar *Xecode;
365    USPTR Xmstart;    PCRE_PUCHAR Xmstart;
366    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
367    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
368    unsigned int Xrdepth;    unsigned int Xrdepth;
369    
370    /* Function local variables */    /* Function local variables */
371    
372    USPTR Xcallpat;    PCRE_PUCHAR Xcallpat;
373  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
374    USPTR Xcharptr;    PCRE_PUCHAR Xcharptr;
375  #endif  #endif
376    USPTR Xdata;    PCRE_PUCHAR Xdata;
377    USPTR Xnext;    PCRE_PUCHAR Xnext;
378    USPTR Xpp;    PCRE_PUCHAR Xpp;
379    USPTR Xprev;    PCRE_PUCHAR Xprev;
380    USPTR Xsaved_eptr;    PCRE_PUCHAR Xsaved_eptr;
381    
382    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
383    
# Line 349  typedef struct heapframe { Line 385  typedef struct heapframe {
385    BOOL Xcondition;    BOOL Xcondition;
386    BOOL Xprev_is_word;    BOOL Xprev_is_word;
387    
   unsigned long int Xoriginal_ims;  
   
388  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
389    int Xprop_type;    int Xprop_type;
390    int Xprop_value;    int Xprop_value;
391    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
392    int Xoclength;    int Xoclength;
393    uschar Xocchars[8];    pcre_uchar Xocchars[6];
394  #endif  #endif
395    
396    int Xcodelink;    int Xcodelink;
# Line 401  returns a negative (error) response, the Line 432  returns a negative (error) response, the
432  same response. */  same response. */
433    
434  /* These macros pack up tests that are used for partial matching, and which  /* These macros pack up tests that are used for partial matching, and which
435  appears several times in the code. We set the "hit end" flag if the pointer is  appear several times in the code. We set the "hit end" flag if the pointer is
436  at the end of the subject and also past the start of the subject (i.e.  at the end of the subject and also past the start of the subject (i.e.
437  something has been matched). For hard partial matching, we then return  something has been matched). For hard partial matching, we then return
438  immediately. The second one is used when we already know we are past the end of  immediately. The second one is used when we already know we are past the end of
439  the subject. */  the subject. */
440    
441  #define CHECK_PARTIAL()\  #define CHECK_PARTIAL()\
442    if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\    if (md->partial != 0 && eptr >= md->end_subject && \
443      {\        eptr > md->start_used_ptr) \
444      md->hitend = TRUE;\      { \
445      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\      md->hitend = TRUE; \
446        if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
447      }      }
448    
449  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
450    if (md->partial != 0 && eptr > mstart)\    if (md->partial != 0 && eptr > md->start_used_ptr) \
451      {\      { \
452      md->hitend = TRUE;\      md->hitend = TRUE; \
453      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
454      }      }
455    
456    
457  /* Performance note: It might be tempting to extract commonly used fields from  /* Performance note: It might be tempting to extract commonly used fields from
458  the md structure (e.g. utf8, end_subject) into individual variables to improve  the md structure (e.g. utf, end_subject) into individual variables to improve
459  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
460  made performance worse.  made performance worse.
461    
# Line 434  Arguments: Line 466  Arguments:
466                   by encountering \K)                   by encountering \K)
467     offset_top  current top pointer     offset_top  current top pointer
468     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
469     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
470                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
471     rdepth      the recursion depth     rdepth      the recursion depth
472    
473  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
474                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
475                   a negative MATCH_xxx value for PRUNE, SKIP, etc
476                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
477                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
478  */  */
479    
480  static int  static int
481  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode,
482    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
483    int flags, unsigned int rdepth)    unsigned int rdepth)
484  {  {
485  /* 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,
486  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 461  so they can be ordinary variables in all Line 489  so they can be ordinary variables in all
489  register int  rrc;         /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
490  register int  i;           /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
491  register unsigned int c;   /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
492  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf;         /* Local copy of UTF flag for speed */
493    
494  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
495    BOOL caseless;
496  int condcode;  int condcode;
497    
498  /* 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
499  preserved over calls to RMATCH() are part of a "frame" which is obtained from  preserved over calls to RMATCH() are part of a "frame". We set up the top-level
500  heap storage. Set up the top-level frame here; others are obtained from the  frame on the stack here; subsequent instantiations are obtained from the heap
501  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  whenever RMATCH() does a "recursion". See the macro definitions above. Putting
502    the top-level on the stack rather than malloc-ing them all gives a performance
503    boost in many cases where there is not much "recursion". */
504    
505  #ifdef NO_RECURSE  #ifdef NO_RECURSE
506  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)md->match_frames_base;
 frame->Xprevframe = NULL;            /* Marks the top level */  
507    
508  /* Copy in the original argument variables */  /* Copy in the original argument variables */
509    
# Line 481  frame->Xeptr = eptr; Line 511  frame->Xeptr = eptr;
511  frame->Xecode = ecode;  frame->Xecode = ecode;
512  frame->Xmstart = mstart;  frame->Xmstart = mstart;
513  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
514  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
515  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
516    
517  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 496  HEAP_RECURSE: Line 524  HEAP_RECURSE:
524  #define ecode              frame->Xecode  #define ecode              frame->Xecode
525  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
526  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
527  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
528  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
529    
530  /* Ditto for the local variables */  /* Ditto for the local variables */
531    
532  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
533  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
534  #endif  #endif
535  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 520  HEAP_RECURSE: Line 546  HEAP_RECURSE:
546  #define condition          frame->Xcondition  #define condition          frame->Xcondition
547  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
548    
 #define original_ims       frame->Xoriginal_ims  
   
549  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
550  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
551  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
552  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
 #define prop_category      frame->Xprop_category  
 #define prop_chartype      frame->Xprop_chartype  
 #define prop_script        frame->Xprop_script  
553  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
554  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
555  #endif  #endif
# Line 558  i, and fc and c, can be the same variabl Line 579  i, and fc and c, can be the same variabl
579  #define fi i  #define fi i
580  #define fc c  #define fc c
581    
582    /* Many of the following variables are used only in small blocks of the code.
583    My normal style of coding would have declared them within each of those blocks.
584    However, in order to accommodate the version of this code that uses an external
585    "stack" implemented on the heap, it is easier to declare them all here, so the
586    declarations can be cut out in a block. The only declarations within blocks
587    below are for variables that do not have to be preserved over a recursive call
588    to RMATCH(). */
589    
590    #ifdef SUPPORT_UTF
591    const pcre_uchar *charptr;
592    #endif
593    const pcre_uchar *callpat;
594    const pcre_uchar *data;
595    const pcre_uchar *next;
596    PCRE_PUCHAR       pp;
597    const pcre_uchar *prev;
598    PCRE_PUCHAR       saved_eptr;
599    
600  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  recursion_info new_recursive;
601  const uschar *charptr;             /* in small blocks of the code. My normal */  
602  #endif                             /* style of coding would have declared    */  BOOL cur_is_word;
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
603  BOOL condition;  BOOL condition;
604  BOOL prev_is_word;  BOOL prev_is_word;
605    
 unsigned long int original_ims;  
   
606  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
607  int prop_type;  int prop_type;
608  int prop_value;  int prop_value;
609  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
610  int oclength;  int oclength;
611  uschar occhars[8];  pcre_uchar occhars[6];
612  #endif  #endif
613    
614  int codelink;  int codelink;
# Line 601  int save_offset1, save_offset2, save_off Line 624  int save_offset1, save_offset2, save_off
624  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
625    
626  eptrblock newptrb;  eptrblock newptrb;
627    
628    /* There is a special fudge for calling match() in a way that causes it to
629    measure the size of its basic stack frame when the stack is being used for
630    recursion. The second argument (ecode) being NULL triggers this behaviour. It
631    cannot normally ever be NULL. The return is the negated value of the frame
632    size. */
633    
634    if (ecode == NULL)
635      {
636      if (rdepth == 0)
637        return match((PCRE_PUCHAR)&rdepth, NULL, NULL, 0, NULL, NULL, 1);
638      else
639        {
640        int len = (char *)&rdepth - (char *)eptr;
641        return (len > 0)? -len : len;
642        }
643      }
644  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
645    
646    /* To save space on the stack and in the heap frame, I have doubled up on some
647    of the local variables that are used only in localised parts of the code, but
648    still need to be preserved over recursive calls of match(). These macros define
649    the alternative names that are used. */
650    
651    #define allow_zero    cur_is_word
652    #define cbegroup      condition
653    #define code_offset   codelink
654    #define condassert    condition
655    #define matched_once  prev_is_word
656    #define foc           number
657    #define save_mark     data
658    
659  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
660  variables. */  variables. */
661    
# Line 627  defined). However, RMATCH isn't like a f Line 680  defined). However, RMATCH isn't like a f
680  complicated macro. It has to be used in one particular way. This shouldn't,  complicated macro. It has to be used in one particular way. This shouldn't,
681  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
682    
683  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
684  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
685  #else  #else
686  utf8 = FALSE;  utf = FALSE;
687  #endif  #endif
688    
689  /* First check that we haven't called match() too many times, or that we  /* First check that we haven't called match() too many times, or that we
# Line 639  haven't exceeded the recursive call limi Line 692  haven't exceeded the recursive call limi
692  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
693  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
694    
 original_ims = ims;    /* Save for resetting on ')' */  
   
695  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
696  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
697  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
698  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
699  When match() is called in other circumstances, don't add to the chain. The  
700  match_cbegroup flag must NOT be used with tail recursion, because the memory  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
701  block that is used is on the stack, so a new one may be required for each  such remembered pointers, to be checked when we hit the closing ket, in order
702  match(). */  to break infinite loops that match no characters. When match() is called in
703    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
704    NOT be used with tail recursion, because the memory block that is used is on
705    the stack, so a new one may be required for each match(). */
706    
707  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
708    {    {
709    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
710    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
711    eptrb = &newptrb;    eptrb = &newptrb;
712      md->match_function_type = 0;
713    }    }
714    
715  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 666  for (;;) Line 721  for (;;)
721    
722    switch(op)    switch(op)
723      {      {
724        case OP_MARK:
725        md->nomatch_mark = ecode + 2;
726        md->mark = NULL;    /* In case previously set by assertion */
727        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
728          eptrb, RM55);
729        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
730             md->mark == NULL) md->mark = ecode + 2;
731    
732        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
733        argument, and we must check whether that argument matches this MARK's
734        argument. It is passed back in md->start_match_ptr (an overloading of that
735        variable). If it does match, we reset that variable to the current subject
736        position and return MATCH_SKIP. Otherwise, pass back the return code
737        unaltered. */
738    
739        else if (rrc == MATCH_SKIP_ARG &&
740            STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0)
741          {
742          md->start_match_ptr = eptr;
743          RRETURN(MATCH_SKIP);
744          }
745        RRETURN(rrc);
746    
747      case OP_FAIL:      case OP_FAIL:
748      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
749    
750      case OP_PRUNE:      /* COMMIT overrides PRUNE, SKIP, and THEN */
     RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,  
       ims, eptrb, flags, RM51);  
     if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
     RRETURN(MATCH_PRUNE);  
751    
752      case OP_COMMIT:      case OP_COMMIT:
753      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
754        ims, eptrb, flags, RM52);        eptrb, RM52);
755      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
756            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
757            rrc != MATCH_THEN)
758          RRETURN(rrc);
759      RRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
760    
761        /* PRUNE overrides THEN */
762    
763        case OP_PRUNE:
764        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
765          eptrb, RM51);
766        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
767        RRETURN(MATCH_PRUNE);
768    
769        case OP_PRUNE_ARG:
770        md->nomatch_mark = ecode + 2;
771        md->mark = NULL;    /* In case previously set by assertion */
772        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
773          eptrb, RM56);
774        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
775             md->mark == NULL) md->mark = ecode + 2;
776        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
777        RRETURN(MATCH_PRUNE);
778    
779        /* SKIP overrides PRUNE and THEN */
780    
781      case OP_SKIP:      case OP_SKIP:
782      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
783        ims, eptrb, flags, RM53);        eptrb, RM53);
784      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
785          RRETURN(rrc);
786      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
787      RRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
788    
789        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
790        nomatch_mark. There is a flag that disables this opcode when re-matching a
791        pattern that ended with a SKIP for which there was not a matching MARK. */
792    
793        case OP_SKIP_ARG:
794        if (md->ignore_skip_arg)
795          {
796          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
797          break;
798          }
799        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
800          eptrb, RM57);
801        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
802          RRETURN(rrc);
803    
804        /* Pass back the current skip name by overloading md->start_match_ptr and
805        returning the special MATCH_SKIP_ARG return code. This will either be
806        caught by a matching MARK, or get to the top, where it causes a rematch
807        with the md->ignore_skip_arg flag set. */
808    
809        md->start_match_ptr = ecode + 2;
810        RRETURN(MATCH_SKIP_ARG);
811    
812        /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
813        the branch in which it occurs can be determined. Overload the start of
814        match pointer to do this. */
815    
816      case OP_THEN:      case OP_THEN:
817      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
818        ims, eptrb, flags, RM54);        eptrb, RM54);
819      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
820        md->start_match_ptr = ecode;
821      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
822    
823      /* Handle a capturing bracket. If there is space in the offset vector, save      case OP_THEN_ARG:
824      the current subject position in the working slot at the top of the vector.      md->nomatch_mark = ecode + 2;
825      We mustn't change the current values of the data slot, because they may be      md->mark = NULL;    /* In case previously set by assertion */
826      set from a previous iteration of this group, and be referred to by a      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
827      reference inside the group.        md, eptrb, RM58);
828        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
829      If the bracket fails to match, we need to restore this value and also the           md->mark == NULL) md->mark = ecode + 2;
830      values of the final offsets, in case they were set by a previous iteration      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
831      of the same bracket.      md->start_match_ptr = ecode;
832        RRETURN(MATCH_THEN);
833    
834        /* Handle an atomic group that does not contain any capturing parentheses.
835        This can be handled like an assertion. Prior to 8.13, all atomic groups
836        were handled this way. In 8.13, the code was changed as below for ONCE, so
837        that backups pass through the group and thereby reset captured values.
838        However, this uses a lot more stack, so in 8.20, atomic groups that do not
839        contain any captures generate OP_ONCE_NC, which can be handled in the old,
840        less stack intensive way.
841    
842        Check the alternative branches in turn - the matching won't pass the KET
843        for this kind of subpattern. If any one branch matches, we carry on as at
844        the end of a normal bracket, leaving the subject pointer, but resetting
845        the start-of-match value in case it was changed by \K. */
846    
847        case OP_ONCE_NC:
848        prev = ecode;
849        saved_eptr = eptr;
850        save_mark = md->mark;
851        do
852          {
853          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
854          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
855            {
856            mstart = md->start_match_ptr;
857            break;
858            }
859          if (rrc == MATCH_THEN)
860            {
861            next = ecode + GET(ecode,1);
862            if (md->start_match_ptr < next &&
863                (*ecode == OP_ALT || *next == OP_ALT))
864              rrc = MATCH_NOMATCH;
865            }
866    
867          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
868          ecode += GET(ecode,1);
869          md->mark = save_mark;
870          }
871        while (*ecode == OP_ALT);
872    
873        /* If hit the end of the group (which could be repeated), fail */
874    
875        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
876    
877        /* Continue as from after the group, updating the offsets high water
878        mark, since extracts may have been taken. */
879    
880        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
881    
882        offset_top = md->end_offset_top;
883        eptr = md->end_match_ptr;
884    
885        /* For a non-repeating ket, just continue at this level. This also
886        happens for a repeating ket if no characters were matched in the group.
887        This is the forcible breaking of infinite loops as implemented in Perl
888        5.005. */
889    
890        if (*ecode == OP_KET || eptr == saved_eptr)
891          {
892          ecode += 1+LINK_SIZE;
893          break;
894          }
895    
896        /* The repeating kets try the rest of the pattern or restart from the
897        preceding bracket, in the appropriate order. The second "call" of match()
898        uses tail recursion, to avoid using another stack frame. */
899    
900        if (*ecode == OP_KETRMIN)
901          {
902          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
903          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
904          ecode = prev;
905          goto TAIL_RECURSE;
906          }
907        else  /* OP_KETRMAX */
908          {
909          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
910          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
911          ecode += 1 + LINK_SIZE;
912          goto TAIL_RECURSE;
913          }
914        /* Control never gets here */
915    
916        /* Handle a capturing bracket, other than those that are possessive with an
917        unlimited repeat. If there is space in the offset vector, save the current
918        subject position in the working slot at the top of the vector. We mustn't
919        change the current values of the data slot, because they may be set from a
920        previous iteration of this group, and be referred to by a reference inside
921        the group. A failure to match might occur after the group has succeeded,
922        if something later on doesn't match. For this reason, we need to restore
923        the working value and also the values of the final offsets, in case they
924        were set by a previous iteration of the same bracket.
925    
926      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
927      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 726  for (;;) Line 945  for (;;)
945        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
946        save_offset3 = md->offset_vector[md->offset_end - number];        save_offset3 = md->offset_vector[md->offset_end - number];
947        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
948          save_mark = md->mark;
949    
950        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
951        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
952            (int)(eptr - md->start_subject);
953    
954        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
955          {          {
956          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
957            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
958          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);            eptrb, RM1);
959            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
960    
961            /* If we backed up to a THEN, check whether it is within the current
962            branch by comparing the address of the THEN that is passed back with
963            the end of the branch. If it is within the current branch, and the
964            branch is one of two or more alternatives (it either starts or ends
965            with OP_ALT), we have reached the limit of THEN's action, so convert
966            the return code to NOMATCH, which will cause normal backtracking to
967            happen from now on. Otherwise, THEN is passed back to an outer
968            alternative. This implements Perl's treatment of parenthesized groups,
969            where a group not containing | does not affect the current alternative,
970            that is, (X) is NOT the same as (X|(*F)). */
971    
972            if (rrc == MATCH_THEN)
973              {
974              next = ecode + GET(ecode,1);
975              if (md->start_match_ptr < next &&
976                  (*ecode == OP_ALT || *next == OP_ALT))
977                rrc = MATCH_NOMATCH;
978              }
979    
980            /* Anything other than NOMATCH is passed back. */
981    
982            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
983          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
984          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
985            md->mark = save_mark;
986            if (*ecode != OP_ALT) break;
987          }          }
       while (*ecode == OP_ALT);  
988    
989        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
990        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
991        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
992        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
993    
994        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
995    
996          RRETURN(rrc);
997        }        }
998    
999      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 761  for (;;) Line 1007  for (;;)
1007      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1009    
1010      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
1011      final alternative within the brackets, we would return the result of a      repeat and ONCE group with no captures. Loop for all the alternatives.
     recursive call to match() whatever happened. We can reduce stack usage by  
     turning this into a tail recursion, except in the case when match_cbegroup  
     is set.*/  
1012    
1013        When we get to the final alternative within the brackets, we used to return
1014        the result of a recursive call to match() whatever happened so it was
1015        possible to reduce stack usage by turning this into a tail recursion,
1016        except in the case of a possibly empty group. However, now that there is
1017        the possiblity of (*THEN) occurring in the final alternative, this
1018        optimization is no longer always possible.
1019    
1020        We can optimize if we know there are no (*THEN)s in the pattern; at present
1021        this is the best that can be done.
1022    
1023        MATCH_ONCE is returned when the end of an atomic group is successfully
1024        reached, but subsequent matching fails. It passes back up the tree (causing
1025        captured values to be reset) until the original atomic group level is
1026        reached. This is tested by comparing md->once_target with the start of the
1027        group. At this point, the return is converted into MATCH_NOMATCH so that
1028        previous backup points can be taken. */
1029    
1030        case OP_ONCE:
1031      case OP_BRA:      case OP_BRA:
1032      case OP_SBRA:      case OP_SBRA:
1033      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
1034      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
1035      for (;;)      for (;;)
1036        {        {
1037        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE)
1038            md->match_function_type = MATCH_CBEGROUP;
1039    
1040          /* If this is not a possibly empty group, and there are no (*THEN)s in
1041          the pattern, and this is the final alternative, optimize as described
1042          above. */
1043    
1044          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1045            {
1046            ecode += PRIV(OP_lengths)[*ecode];
1047            goto TAIL_RECURSE;
1048            }
1049    
1050          /* In all other cases, we have to make another call to match(). */
1051    
1052          save_mark = md->mark;
1053          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1054            RM2);
1055    
1056          /* See comment in the code for capturing groups above about handling
1057          THEN. */
1058    
1059          if (rrc == MATCH_THEN)
1060            {
1061            next = ecode + GET(ecode,1);
1062            if (md->start_match_ptr < next &&
1063                (*ecode == OP_ALT || *next == OP_ALT))
1064              rrc = MATCH_NOMATCH;
1065            }
1066    
1067          if (rrc != MATCH_NOMATCH)
1068          {          {
1069          if (flags == 0)    /* Not a possibly empty group */          if (rrc == MATCH_ONCE)
1070            {            {
1071            ecode += _pcre_OP_lengths[*ecode];            const pcre_uchar *scode = ecode;
1072            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1073            goto TAIL_RECURSE;              {
1074                while (*scode == OP_ALT) scode += GET(scode, 1);
1075                scode -= GET(scode, 1);
1076                }
1077              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1078            }            }
1079            RRETURN(rrc);
1080            }
1081          ecode += GET(ecode, 1);
1082          md->mark = save_mark;
1083          if (*ecode != OP_ALT) break;
1084          }
1085    
1086        RRETURN(MATCH_NOMATCH);
1087    
1088          /* Possibly empty group; can't use tail recursion. */      /* Handle possessive capturing brackets with an unlimited repeat. We come
1089        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1090        handled similarly to the normal case above. However, the matching is
1091        different. The end of these brackets will always be OP_KETRPOS, which
1092        returns MATCH_KETRPOS without going further in the pattern. By this means
1093        we can handle the group by iteration rather than recursion, thereby
1094        reducing the amount of stack needed. */
1095    
1096        case OP_CBRAPOS:
1097        case OP_SCBRAPOS:
1098        allow_zero = FALSE;
1099    
1100          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      POSSESSIVE_CAPTURE:
1101            eptrb, flags, RM48);      number = GET2(ecode, 1+LINK_SIZE);
1102          RRETURN(rrc);      offset = number << 1;
1103    
1104    #ifdef PCRE_DEBUG
1105        printf("start possessive bracket %d\n", number);
1106        printf("subject=");
1107        pchars(eptr, 16, TRUE, md);
1108        printf("\n");
1109    #endif
1110    
1111        if (offset < md->offset_max)
1112          {
1113          matched_once = FALSE;
1114          code_offset = (int)(ecode - md->start_code);
1115    
1116          save_offset1 = md->offset_vector[offset];
1117          save_offset2 = md->offset_vector[offset+1];
1118          save_offset3 = md->offset_vector[md->offset_end - number];
1119          save_capture_last = md->capture_last;
1120    
1121          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1122    
1123          /* Each time round the loop, save the current subject position for use
1124          when the group matches. For MATCH_MATCH, the group has matched, so we
1125          restart it with a new subject starting position, remembering that we had
1126          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1127          usual. If we haven't matched any alternatives in any iteration, check to
1128          see if a previous iteration matched. If so, the group has matched;
1129          continue from afterwards. Otherwise it has failed; restore the previous
1130          capture values before returning NOMATCH. */
1131    
1132          for (;;)
1133            {
1134            md->offset_vector[md->offset_end - number] =
1135              (int)(eptr - md->start_subject);
1136            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1137            RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1138              eptrb, RM63);
1139            if (rrc == MATCH_KETRPOS)
1140              {
1141              offset_top = md->end_offset_top;
1142              eptr = md->end_match_ptr;
1143              ecode = md->start_code + code_offset;
1144              save_capture_last = md->capture_last;
1145              matched_once = TRUE;
1146              continue;
1147              }
1148    
1149            /* See comment in the code for capturing groups above about handling
1150            THEN. */
1151    
1152            if (rrc == MATCH_THEN)
1153              {
1154              next = ecode + GET(ecode,1);
1155              if (md->start_match_ptr < next &&
1156                  (*ecode == OP_ALT || *next == OP_ALT))
1157                rrc = MATCH_NOMATCH;
1158              }
1159    
1160            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1161            md->capture_last = save_capture_last;
1162            ecode += GET(ecode, 1);
1163            if (*ecode != OP_ALT) break;
1164          }          }
1165    
1166        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1167        otherwise return. */          {
1168            md->offset_vector[offset] = save_offset1;
1169            md->offset_vector[offset+1] = save_offset2;
1170            md->offset_vector[md->offset_end - number] = save_offset3;
1171            }
1172    
1173        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (allow_zero || matched_once)
1174          eptrb, flags, RM2);          {
1175        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          ecode += 1 + LINK_SIZE;
1176            break;
1177            }
1178    
1179          RRETURN(MATCH_NOMATCH);
1180          }
1181    
1182        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1183        as a non-capturing bracket. */
1184    
1185        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1186        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1187    
1188        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1189    
1190        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1191        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1192    
1193        /* Non-capturing possessive bracket with unlimited repeat. We come here
1194        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1195        without the capturing complication. It is written out separately for speed
1196        and cleanliness. */
1197    
1198        case OP_BRAPOS:
1199        case OP_SBRAPOS:
1200        allow_zero = FALSE;
1201    
1202        POSSESSIVE_NON_CAPTURE:
1203        matched_once = FALSE;
1204        code_offset = (int)(ecode - md->start_code);
1205    
1206        for (;;)
1207          {
1208          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1209          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1210            eptrb, RM48);
1211          if (rrc == MATCH_KETRPOS)
1212            {
1213            offset_top = md->end_offset_top;
1214            eptr = md->end_match_ptr;
1215            ecode = md->start_code + code_offset;
1216            matched_once = TRUE;
1217            continue;
1218            }
1219    
1220          /* See comment in the code for capturing groups above about handling
1221          THEN. */
1222    
1223          if (rrc == MATCH_THEN)
1224            {
1225            next = ecode + GET(ecode,1);
1226            if (md->start_match_ptr < next &&
1227                (*ecode == OP_ALT || *next == OP_ALT))
1228              rrc = MATCH_NOMATCH;
1229            }
1230    
1231          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1232        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1233          if (*ecode != OP_ALT) break;
1234        }        }
1235    
1236        if (matched_once || allow_zero)
1237          {
1238          ecode += 1 + LINK_SIZE;
1239          break;
1240          }
1241        RRETURN(MATCH_NOMATCH);
1242    
1243      /* Control never reaches here. */      /* Control never reaches here. */
1244    
1245      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1246      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
1247      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
1248      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1249    
1250      case OP_COND:      case OP_COND:
1251      case OP_SCOND:      case OP_SCOND:
1252      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1253    
1254      /* Because of the way auto-callout works during compile, a callout item is      /* Because of the way auto-callout works during compile, a callout item is
1255      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
1256    
1257      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1258        {        {
1259        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1260          {          {
1261          pcre_callout_block cb;          PUBL(callout_block) cb;
1262          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1263          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1264          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1265    #ifdef COMPILE_PCRE8
1266          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1267          cb.subject_length   = md->end_subject - md->start_subject;  #else
1268          cb.start_match      = mstart - md->start_subject;          cb.subject          = (PCRE_SPTR16)md->start_subject;
1269          cb.current_position = eptr - md->start_subject;  #endif
1270            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1271            cb.start_match      = (int)(mstart - md->start_subject);
1272            cb.current_position = (int)(eptr - md->start_subject);
1273          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1274          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1275          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1276          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1277          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1278          if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1279            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1280          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1281          }          }
1282        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1283        }        }
1284    
1285      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
# Line 849  for (;;) Line 1296  for (;;)
1296        else        else
1297          {          {
1298          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1299          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1300    
1301          /* If the test is for recursion into a specific subpattern, and it is          /* If the test is for recursion into a specific subpattern, and it is
1302          false, but the test was set up by name, scan the table to see if the          false, but the test was set up by name, scan the table to see if the
1303          name refers to any other numbers, and test them. The condition is true          name refers to any other numbers, and test them. The condition is true
1304          if any one is set. */          if any one is set. */
1305    
1306          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1307            {            {
1308            uschar *slotA = md->name_table;            pcre_uchar *slotA = md->name_table;
1309            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
1310              {              {
1311              if (GET2(slotA, 0) == recno) break;              if (GET2(slotA, 0) == recno) break;
# Line 871  for (;;) Line 1318  for (;;)
1318    
1319            if (i < md->name_count)            if (i < md->name_count)
1320              {              {
1321              uschar *slotB = slotA;              pcre_uchar *slotB = slotA;
1322              while (slotB > md->name_table)              while (slotB > md->name_table)
1323                {                {
1324                slotB -= md->name_entry_size;                slotB -= md->name_entry_size;
1325                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1326                  {                  {
1327                  condition = GET2(slotB, 0) == md->recursive->group_num;                  condition = GET2(slotB, 0) == md->recursive->group_num;
1328                  if (condition) break;                  if (condition) break;
# Line 891  for (;;) Line 1338  for (;;)
1338                for (i++; i < md->name_count; i++)                for (i++; i < md->name_count; i++)
1339                  {                  {
1340                  slotB += md->name_entry_size;                  slotB += md->name_entry_size;
1341                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1342                    {                    {
1343                    condition = GET2(slotB, 0) == md->recursive->group_num;                    condition = GET2(slotB, 0) == md->recursive->group_num;
1344                    if (condition) break;                    if (condition) break;
# Line 904  for (;;) Line 1351  for (;;)
1351    
1352          /* Chose branch according to the condition */          /* Chose branch according to the condition */
1353    
1354          ecode += condition? 3 : GET(ecode, 1);          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1355          }          }
1356        }        }
1357    
# Line 921  for (;;) Line 1368  for (;;)
1368        if (!condition && condcode == OP_NCREF)        if (!condition && condcode == OP_NCREF)
1369          {          {
1370          int refno = offset >> 1;          int refno = offset >> 1;
1371          uschar *slotA = md->name_table;          pcre_uchar *slotA = md->name_table;
1372    
1373          for (i = 0; i < md->name_count; i++)          for (i = 0; i < md->name_count; i++)
1374            {            {
# Line 935  for (;;) Line 1382  for (;;)
1382    
1383          if (i < md->name_count)          if (i < md->name_count)
1384            {            {
1385            uschar *slotB = slotA;            pcre_uchar *slotB = slotA;
1386            while (slotB > md->name_table)            while (slotB > md->name_table)
1387              {              {
1388              slotB -= md->name_entry_size;              slotB -= md->name_entry_size;
1389              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)              if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1390                {                {
1391                offset = GET2(slotB, 0) << 1;                offset = GET2(slotB, 0) << 1;
1392                condition = offset < offset_top &&                condition = offset < offset_top &&
# Line 957  for (;;) Line 1404  for (;;)
1404              for (i++; i < md->name_count; i++)              for (i++; i < md->name_count; i++)
1405                {                {
1406                slotB += md->name_entry_size;                slotB += md->name_entry_size;
1407                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1408                  {                  {
1409                  offset = GET2(slotB, 0) << 1;                  offset = GET2(slotB, 0) << 1;
1410                  condition = offset < offset_top &&                  condition = offset < offset_top &&
# Line 972  for (;;) Line 1419  for (;;)
1419    
1420        /* Chose branch according to the condition */        /* Chose branch according to the condition */
1421    
1422        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1423        }        }
1424    
1425      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 982  for (;;) Line 1429  for (;;)
1429        }        }
1430    
1431      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1432      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1433      assertion. */      an assertion. */
1434    
1435      else      else
1436        {        {
1437        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1438            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1439        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1440          {          {
1441            if (md->end_offset_top > offset_top)
1442              offset_top = md->end_offset_top;  /* Captures may have happened */
1443          condition = TRUE;          condition = TRUE;
1444          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1445          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1446          }          }
1447    
1448          /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1449          assertion; it is therefore treated as NOMATCH. */
1450    
1451        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1452          {          {
1453          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
# Line 1006  for (;;) Line 1459  for (;;)
1459          }          }
1460        }        }
1461    
1462      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one, can
1463      we can use tail recursion to avoid using another stack frame, except when      use tail recursion to avoid using another stack frame, except when there is
1464      match_cbegroup is required for an unlimited repeat of a possibly empty      unlimited repeat of a possibly empty group. In the latter case, a recursive
1465      group. If the second alternative doesn't exist, we can just plough on. */      call to match() is always required, unless the second alternative doesn't
1466        exist, in which case we can just plough on. Note that, for compatibility
1467        with Perl, the | in a conditional group is NOT treated as creating two
1468        alternatives. If a THEN is encountered in the branch, it propagates out to
1469        the enclosing alternative (unless nested in a deeper set of alternatives,
1470        of course). */
1471    
1472      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1473        {        {
1474        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
1475          {          {
1476          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          ecode += 1 + LINK_SIZE;
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
1477          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1478          }          }
1479    
1480          md->match_function_type = MATCH_CBEGROUP;
1481          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1482          RRETURN(rrc);
1483        }        }
1484      else                         /* Condition false & no alternative */  
1485         /* Condition false & no alternative; continue after the group. */
1486    
1487        else
1488        {        {
1489        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1490        }        }
# Line 1049  for (;;) Line 1508  for (;;)
1508        {        {
1509        md->offset_vector[offset] =        md->offset_vector[offset] =
1510          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1511        md->offset_vector[offset+1] = eptr - md->start_subject;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1512        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1513        }        }
1514      ecode += 3;      ecode += 1 + IMM2_SIZE;
1515      break;      break;
1516    
1517    
1518      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. */
     recursion, we should restore the offsets appropriately and continue from  
     after the call. */  
1519    
     case OP_ACCEPT:  
1520      case OP_END:      case OP_END:
1521      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1522        {      case OP_ASSERT_ACCEPT:
       recursion_info *rec = md->recursive;  
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
       md->recursive = rec->prevrec;  
       memmove(md->offset_vector, rec->offset_save,  
         rec->saved_max * sizeof(int));  
       offset_top = rec->save_offset_top;  
       mstart = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1523    
1524      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is      /* If we have matched an empty string, fail if not in an assertion and not
1525      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1526      the subject. In both cases, backtracking will then try other alternatives,      is set and we have matched at the start of the subject. In both cases,
1527      if any. */      backtracking will then try other alternatives, if any. */
1528    
1529      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1530          (md->notempty ||           md->recursive == NULL &&
1531            (md->notempty_atstart &&           (md->notempty ||
1532              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1533                 mstart == md->start_subject + md->start_offset)))
1534        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1535    
1536      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1092  for (;;) Line 1538  for (;;)
1538      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1539      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1540      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
     RRETURN(MATCH_MATCH);  
1541    
1542      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1543        given as the argument to RRETURN when the heap is in use. */
1544    
1545      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1546      ims = ecode[1];      RRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1547    
1548      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1549      matching won't pass the KET for an assertion. If any one branch matches,      matching won't pass the KET for an assertion. If any one branch matches,
1550      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
1551      start of each branch to move the current point backwards, so the code at      start of each branch to move the current point backwards, so the code at
1552      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1553        of a condition, we want to return immediately afterwards. The caller of
1554        this incarnation of the match() function will have set MATCH_CONDASSERT in
1555        md->match_function type, and one of these opcodes will be the first opcode
1556        that is processed. We use a local variable that is preserved over calls to
1557        match() to remember this case. */
1558    
1559      case OP_ASSERT:      case OP_ASSERT:
1560      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1561        save_mark = md->mark;
1562        if (md->match_function_type == MATCH_CONDASSERT)
1563          {
1564          condassert = TRUE;
1565          md->match_function_type = 0;
1566          }
1567        else condassert = FALSE;
1568    
1569      do      do
1570        {        {
1571        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1572          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1573        if (rrc == MATCH_MATCH) break;          {
1574            mstart = md->start_match_ptr;   /* In case \K reset it */
1575            break;
1576            }
1577          md->mark = save_mark;
1578    
1579          /* A COMMIT failure must fail the entire assertion, without trying any
1580          subsequent branches. */
1581    
1582          if (rrc == MATCH_COMMIT) RRETURN(MATCH_NOMATCH);
1583    
1584          /* PCRE does not allow THEN to escape beyond an assertion; it
1585          is treated as NOMATCH. */
1586    
1587        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1588        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1589        }        }
1590      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1591    
1592      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1593    
1594      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1595    
1596      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1597    
1598      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1599      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1139  for (;;) Line 1609  for (;;)
1609    
1610      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1611      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1612        save_mark = md->mark;
1613        if (md->match_function_type == MATCH_CONDASSERT)
1614          {
1615          condassert = TRUE;
1616          md->match_function_type = 0;
1617          }
1618        else condassert = FALSE;
1619    
1620      do      do
1621        {        {
1622        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1623          RM5);        md->mark = save_mark;
1624        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1625        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1626          {          {
1627          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1628          break;          break;
1629          }          }
1630    
1631          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1632          as NOMATCH. */
1633    
1634        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1635        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1636        }        }
1637      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1638    
1639      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1640    
1641      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1642      continue;      continue;
# Line 1165  for (;;) Line 1647  for (;;)
1647      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1648    
1649      case OP_REVERSE:      case OP_REVERSE:
1650  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1651      if (utf8)      if (utf)
1652        {        {
1653        i = GET(ecode, 1);        i = GET(ecode, 1);
1654        while (i-- > 0)        while (i-- > 0)
# Line 1197  for (;;) Line 1679  for (;;)
1679      function is able to force a failure. */      function is able to force a failure. */
1680    
1681      case OP_CALLOUT:      case OP_CALLOUT:
1682      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1683        {        {
1684        pcre_callout_block cb;        PUBL(callout_block) cb;
1685        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1686        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1687        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1688    #ifdef COMPILE_PCRE8
1689        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1690        cb.subject_length   = md->end_subject - md->start_subject;  #else
1691        cb.start_match      = mstart - md->start_subject;        cb.subject          = (PCRE_SPTR16)md->start_subject;
1692        cb.current_position = eptr - md->start_subject;  #endif
1693          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1694          cb.start_match      = (int)(mstart - md->start_subject);
1695          cb.current_position = (int)(eptr - md->start_subject);
1696        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1697        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1698        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1699        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1700        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1701        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1702          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1703        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1704        }        }
1705      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1222  for (;;) Line 1709  for (;;)
1709      offset data is the offset to the starting bracket from the start of the      offset data is the offset to the starting bracket from the start of the
1710      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1711    
1712      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1713      save their starting points and reinstate them after the recursion. However,      re-instated afterwards. We don't know how many are started and not yet
1714      we don't know how many such there are (offset_top records the completed      finished (offset_top records the completed total) so we just have to save
1715      total) so we just have to save all the potential data. There may be up to      all the potential data. There may be up to 65535 such values, which is too
1716      65535 such values, which is too large to put on the stack, but using malloc      large to put on the stack, but using malloc for small numbers seems
1717      for small numbers seems expensive. As a compromise, the stack is used when      expensive. As a compromise, the stack is used when there are no more than
1718      there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc      REC_STACK_SAVE_MAX values to store; otherwise malloc is used.
     is used. A problem is what to do if the malloc fails ... there is no way of  
     returning to the top level with an error. Save the top REC_STACK_SAVE_MAX  
     values on the stack, and accept that the rest may be wrong.  
1719    
1720      There are also other values that have to be saved. We use a chained      There are also other values that have to be saved. We use a chained
1721      sequence of blocks that actually live on the stack. Thanks to Robin Houston      sequence of blocks that actually live on the stack. Thanks to Robin Houston
1722      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1723        a lot, so he is not to blame for the current way it works. */
1724    
1725      case OP_RECURSE:      case OP_RECURSE:
1726        {        {
1727          recursion_info *ri;
1728          int recno;
1729    
1730        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1731        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1732          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1733    
1734          /* Check for repeating a recursion without advancing the subject pointer.
1735          This should catch convoluted mutual recursions. (Some simple cases are
1736          caught at compile time.) */
1737    
1738          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1739            if (recno == ri->group_num && eptr == ri->subject_position)
1740              RRETURN(PCRE_ERROR_RECURSELOOP);
1741    
1742        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1743    
1744          new_recursive.group_num = recno;
1745          new_recursive.subject_position = eptr;
1746        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1747        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1748    
1749        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1750    
1751        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1752    
1753        /* Now save the offset data. */        /* Now save the offset data */
1754    
1755        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1756        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1261  for (;;) Line 1758  for (;;)
1758        else        else
1759          {          {
1760          new_recursive.offset_save =          new_recursive.offset_save =
1761            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1762          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1763          }          }
   
1764        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1765              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = mstart;  
       new_recursive.save_offset_top = offset_top;  
       mstart = eptr;  
1766    
1767        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1768        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1769          might be changed, so reset it before looping. */
1770    
1771        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1772        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1773        do        do
1774          {          {
1775          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1776            md, ims, eptrb, flags, RM6);          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1777          if (rrc == MATCH_MATCH)            md, eptrb, RM6);
1778            memcpy(md->offset_vector, new_recursive.offset_save,
1779                new_recursive.saved_max * sizeof(int));
1780            md->recursive = new_recursive.prevrec;
1781            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1782            {            {
1783            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1784            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1785              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1786            RRETURN(MATCH_MATCH);  
1787              /* Set where we got to in the subject, and reset the start in case
1788              it was changed by \K. This *is* propagated back out of a recursion,
1789              for Perl compatibility. */
1790    
1791              eptr = md->end_match_ptr;
1792              mstart = md->start_match_ptr;
1793              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1794            }            }
1795          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)  
1796            /* PCRE does not allow THEN or COMMIT to escape beyond a recursion; it
1797            is treated as NOMATCH. */
1798    
1799            else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN &&
1800                     rrc != MATCH_COMMIT)
1801            {            {
1802            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1803            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1804              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1805            RRETURN(rrc);            RRETURN(rrc);
1806            }            }
1807    
1808          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1809          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1810          }          }
1811        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1306  for (;;) Line 1813  for (;;)
1813        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1814        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1815        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1816          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1817        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1818        }        }
     /* Control never reaches here */  
1819    
1820      /* "Once" brackets are like assertion brackets except that after a match,      RECURSION_MATCHED:
1821      the point in the subject string is not moved back. Thus there can never be      break;
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer. */  
1822    
1823      case OP_ONCE:      /* An alternation is the end of a branch; scan along to find the end of the
1824      prev = ecode;      bracketed group and go to there. */
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH) break;  
       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
   
     /* An alternation is the end of a branch; scan along to find the end of the  
     bracketed group and go to there. */  
1825    
1826      case OP_ALT:      case OP_ALT:
1827      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
# Line 1399  for (;;) Line 1834  for (;;)
1834      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1835    
1836      case OP_BRAZERO:      case OP_BRAZERO:
1837        {      next = ecode + 1;
1838        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1839        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1840        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1841        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1842      break;      break;
1843    
1844      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1845        {      next = ecode + 1;
1846        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1847        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1848        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1849        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1850      break;      break;
1851    
1852      case OP_SKIPZERO:      case OP_SKIPZERO:
1853        {      next = ecode+1;
1854        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1855        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1856      break;      break;
1857    
1858        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1859        here; just jump to the group, with allow_zero set TRUE. */
1860    
1861        case OP_BRAPOSZERO:
1862        op = *(++ecode);
1863        allow_zero = TRUE;
1864        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1865          goto POSSESSIVE_NON_CAPTURE;
1866    
1867      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1868    
1869      case OP_KET:      case OP_KET:
1870      case OP_KETRMIN:      case OP_KETRMIN:
1871      case OP_KETRMAX:      case OP_KETRMAX:
1872        case OP_KETRPOS:
1873      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1874    
1875      /* If this was a group that remembered the subject start, in order to break      /* If this was a group that remembered the subject start, in order to break
1876      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1877      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1878    
1879      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1880        {        {
1881        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1882        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1883        }        }
1884      else saved_eptr = NULL;      else saved_eptr = NULL;
1885    
1886      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group or a non-capturing atomic
1887      MATCH_MATCH, but record the current high water mark for use by positive      group, stop matching and return MATCH_MATCH, but record the current high
1888      assertions. Do this also for the "once" (atomic) groups. */      water mark for use by positive assertions. We also need to record the match
1889        start in case it was changed by \K. */
1890      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||  
1891          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1892          *prev == OP_ONCE)           *prev == OP_ONCE_NC)
1893        {        {
1894        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1895        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1896        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1897          RRETURN(MATCH_MATCH);         /* Sets md->mark */
1898        }        }
1899    
1900      /* For capturing groups we have to check the group number back at the start      /* For capturing groups we have to check the group number back at the start
1901      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1902      bumping the high water mark. Note that whole-pattern recursion is coded as      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1903      a recurse into group 0, so it won't be picked up here. Instead, we catch it      into group 0, so it won't be picked up here. Instead, we catch it when the
1904      when the OP_END is reached. Other recursion is handled here. */      OP_END is reached. Other recursion is handled here. We just have to record
1905        the current subject position and start match pointer and give a MATCH
1906        return. */
1907    
1908      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1909            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1910        {        {
1911        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1912        offset = number << 1;        offset = number << 1;
# Line 1473  for (;;) Line 1916  for (;;)
1916        printf("\n");        printf("\n");
1917  #endif  #endif
1918    
1919          /* Handle a recursively called group. */
1920    
1921          if (md->recursive != NULL && md->recursive->group_num == number)
1922            {
1923            md->end_match_ptr = eptr;
1924            md->start_match_ptr = mstart;
1925            RRETURN(MATCH_MATCH);
1926            }
1927    
1928          /* Deal with capturing */
1929    
1930        md->capture_last = number;        md->capture_last = number;
1931        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1932          {          {
1933            /* If offset is greater than offset_top, it means that we are
1934            "skipping" a capturing group, and that group's offsets must be marked
1935            unset. In earlier versions of PCRE, all the offsets were unset at the
1936            start of matching, but this doesn't work because atomic groups and
1937            assertions can cause a value to be set that should later be unset.
1938            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1939            part of the atomic group, but this is not on the final matching path,
1940            so must be unset when 2 is set. (If there is no group 2, there is no
1941            problem, because offset_top will then be 2, indicating no capture.) */
1942    
1943            if (offset > offset_top)
1944              {
1945              register int *iptr = md->offset_vector + offset_top;
1946              register int *iend = md->offset_vector + offset;
1947              while (iptr < iend) *iptr++ = -1;
1948              }
1949    
1950            /* Now make the extraction */
1951    
1952          md->offset_vector[offset] =          md->offset_vector[offset] =
1953            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1954          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1955          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1956          }          }
1957          }
1958    
1959        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1960        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1961        group. This is the forcible breaking of infinite loops as implemented in
1962        Perl 5.005. For a non-repeating atomic group that includes captures,
1963        establish a backup point by processing the rest of the pattern at a lower
1964        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1965        original OP_ONCE level, thereby bypassing intermediate backup points, but
1966        resetting any captures that happened along the way. */
1967    
1968        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1969          {
1970          if (*prev == OP_ONCE)
1971          {          {
1972          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1973          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1974          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1975          mstart = rec->save_start;          RRETURN(MATCH_ONCE);
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         offset_top = rec->save_offset_top;  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1976          }          }
1977          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1978          break;
1979        }        }
1980    
1981      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1982      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1983        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
1984    
1985      /* For a non-repeating ket, just continue at this level. This also      if (*ecode == OP_KETRPOS)
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
1986        {        {
1987        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1988        break;        md->end_offset_top = offset_top;
1989          RRETURN(MATCH_KETRPOS);
1990        }        }
1991    
1992      /* The repeating kets try the rest of the pattern or restart from the      /* The normal repeating kets try the rest of the pattern or restart from
1993      preceding bracket, in the appropriate order. In the second case, we can use      the preceding bracket, in the appropriate order. In the second case, we can
1994      tail recursion to avoid using another stack frame, unless we have an      use tail recursion to avoid using another stack frame, unless we have an
1995      unlimited repeat of a group that can match an empty string. */      an atomic group or an unlimited repeat of a group that can match an empty
1996        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1997    
1998      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1999        {        {
2000        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
2001        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2002        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
2003            {
2004            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
2005            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2006            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2007            RRETURN(MATCH_ONCE);
2008            }
2009          if (*prev >= OP_SBRA)    /* Could match an empty string */
2010          {          {
2011          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
2012          RRETURN(rrc);          RRETURN(rrc);
2013          }          }
2014        ecode = prev;        ecode = prev;
# Line 1539  for (;;) Line 2016  for (;;)
2016        }        }
2017      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
2018        {        {
2019        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
2020          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
2021        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2022          if (*prev == OP_ONCE)
2023            {
2024            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
2025            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2026            md->once_target = prev;
2027            RRETURN(MATCH_ONCE);
2028            }
2029        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
2030        goto TAIL_RECURSE;        goto TAIL_RECURSE;
2031        }        }
2032      /* Control never gets here */      /* Control never gets here */
2033    
2034      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
2035    
2036      case OP_CIRC:      case OP_CIRC:
2037      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
     if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         RRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
2038    
2039      /* Start of subject assertion */      /* Start of subject assertion */
2040    
# Line 1568  for (;;) Line 2043  for (;;)
2043      ecode++;      ecode++;
2044      break;      break;
2045    
2046        /* Multiline mode: start of subject unless notbol, or after any newline. */
2047    
2048        case OP_CIRCM:
2049        if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2050        if (eptr != md->start_subject &&
2051            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
2052          RRETURN(MATCH_NOMATCH);
2053        ecode++;
2054        break;
2055    
2056      /* Start of match assertion */      /* Start of match assertion */
2057    
2058      case OP_SOM:      case OP_SOM:
# Line 1582  for (;;) Line 2067  for (;;)
2067      ecode++;      ecode++;
2068      break;      break;
2069    
2070      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2071      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2072    
2073      case OP_DOLL:      case OP_DOLLM:
2074      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2075        {        {
2076        if (eptr < md->end_subject)        if (!IS_NEWLINE(eptr))
2077          { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }          {
2078        else          if (md->partial != 0 &&
2079          { if (md->noteol) RRETURN(MATCH_NOMATCH); }              eptr + 1 >= md->end_subject &&
2080        ecode++;              NLBLOCK->nltype == NLTYPE_FIXED &&
2081        break;              NLBLOCK->nllen == 2 &&
2082                *eptr == NLBLOCK->nl[0])
2083              {
2084              md->hitend = TRUE;
2085              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2086              }
2087            RRETURN(MATCH_NOMATCH);
2088            }
2089        }        }
2090      else      else
2091        {        {
2092        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2093        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
2094        }        }
2095        ecode++;
2096        break;
2097    
2098        /* Not multiline mode: assert before a terminating newline or before end of
2099        subject unless noteol is set. */
2100    
2101        case OP_DOLL:
2102        if (md->noteol) RRETURN(MATCH_NOMATCH);
2103        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2104    
2105      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2106    
2107      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2108    
2109      case OP_EOD:      case OP_EOD:
2110      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2111        SCHECK_PARTIAL();
2112      ecode++;      ecode++;
2113      break;      break;
2114    
2115      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
2116    
2117      case OP_EODN:      case OP_EODN:
2118      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
2119        if (eptr < md->end_subject &&
2120          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2121          {
2122          if (md->partial != 0 &&
2123              eptr + 1 >= md->end_subject &&
2124              NLBLOCK->nltype == NLTYPE_FIXED &&
2125              NLBLOCK->nllen == 2 &&
2126              *eptr == NLBLOCK->nl[0])
2127            {
2128            md->hitend = TRUE;
2129            if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2130            }
2131        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2132          }
2133    
2134        /* Either at end of string or \n before end. */
2135    
2136        SCHECK_PARTIAL();
2137      ecode++;      ecode++;
2138      break;      break;
2139    
# Line 1636  for (;;) Line 2148  for (;;)
2148        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2149        partial matching. */        partial matching. */
2150    
2151  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2152        if (utf8)        if (utf)
2153          {          {
2154            /* Get status of previous character */
2155    
2156          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2157            {            {
2158            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2159            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2160            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2161            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2162    #ifdef SUPPORT_UCP
2163              if (md->use_ucp)
2164                {
2165                if (c == '_') prev_is_word = TRUE; else
2166                  {
2167                  int cat = UCD_CATEGORY(c);
2168                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2169                  }
2170                }
2171              else
2172    #endif
2173            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2174            }            }
2175    
2176            /* Get status of next character */
2177    
2178          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2179            {            {
2180            SCHECK_PARTIAL();            SCHECK_PARTIAL();
# Line 1655  for (;;) Line 2183  for (;;)
2183          else          else
2184            {            {
2185            GETCHAR(c, eptr);            GETCHAR(c, eptr);
2186    #ifdef SUPPORT_UCP
2187              if (md->use_ucp)
2188                {
2189                if (c == '_') cur_is_word = TRUE; else
2190                  {
2191                  int cat = UCD_CATEGORY(c);
2192                  cur_is_word = (cat == ucp_L || cat == ucp_N);
2193                  }
2194                }
2195              else
2196    #endif
2197            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2198            }            }
2199          }          }
2200        else        else
2201  #endif  #endif
2202    
2203        /* Not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
2204          consistency with the behaviour of \w we do use it in this case. */
2205    
2206          {          {
2207            /* Get status of previous character */
2208    
2209          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2210            {            {
2211            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
2212            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);  #ifdef SUPPORT_UCP
2213              if (md->use_ucp)
2214                {
2215                c = eptr[-1];
2216                if (c == '_') prev_is_word = TRUE; else
2217                  {
2218                  int cat = UCD_CATEGORY(c);
2219                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2220                  }
2221                }
2222              else
2223    #endif
2224              prev_is_word = MAX_255(eptr[-1])
2225                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2226            }            }
2227    
2228            /* Get status of next character */
2229    
2230          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2231            {            {
2232            SCHECK_PARTIAL();            SCHECK_PARTIAL();
2233            cur_is_word = FALSE;            cur_is_word = FALSE;
2234            }            }
2235          else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          else
2236    #ifdef SUPPORT_UCP
2237            if (md->use_ucp)
2238              {
2239              c = *eptr;
2240              if (c == '_') cur_is_word = TRUE; else
2241                {
2242                int cat = UCD_CATEGORY(c);
2243                cur_is_word = (cat == ucp_L || cat == ucp_N);
2244                }
2245              }
2246            else
2247    #endif
2248            cur_is_word = MAX_255(*eptr)
2249              && ((md->ctypes[*eptr] & ctype_word) != 0);
2250          }          }
2251    
2252        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
# Line 1685  for (;;) Line 2257  for (;;)
2257        }        }
2258      break;      break;
2259    
2260      /* Match a single character type; inline for speed */      /* Match any single character type except newline; have to take care with
2261        CRLF newlines and partial matching. */
2262    
2263      case OP_ANY:      case OP_ANY:
2264      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
2265        if (md->partial != 0 &&
2266            eptr + 1 >= md->end_subject &&
2267            NLBLOCK->nltype == NLTYPE_FIXED &&
2268            NLBLOCK->nllen == 2 &&
2269            *eptr == NLBLOCK->nl[0])
2270          {
2271          md->hitend = TRUE;
2272          if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2273          }
2274    
2275      /* Fall through */      /* Fall through */
2276    
2277        /* Match any single character whatsoever. */
2278    
2279      case OP_ALLANY:      case OP_ALLANY:
2280      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2281        {        {                            /* not be updated before SCHECK_PARTIAL. */
2282        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2283        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2284        }        }
2285      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2286    #ifdef SUPPORT_UTF
2287        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2288    #endif
2289      ecode++;      ecode++;
2290      break;      break;
2291    
# Line 1705  for (;;) Line 2293  for (;;)
2293      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2294    
2295      case OP_ANYBYTE:      case OP_ANYBYTE:
2296      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2297        {        {                            /* not be updated before SCHECK_PARTIAL. */
2298        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2299        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2300        }        }
2301        eptr++;
2302      ecode++;      ecode++;
2303      break;      break;
2304    
# Line 1721  for (;;) Line 2310  for (;;)
2310        }        }
2311      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2312      if (      if (
2313  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2314         c < 256 &&         c < 256 &&
2315  #endif  #endif
2316         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
# Line 1738  for (;;) Line 2327  for (;;)
2327        }        }
2328      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2329      if (      if (
2330  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2331         c >= 256 ||         c > 255 ||
2332  #endif  #endif
2333         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2334         )         )
# Line 1755  for (;;) Line 2344  for (;;)
2344        }        }
2345      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2346      if (      if (
2347  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2348         c < 256 &&         c < 256 &&
2349  #endif  #endif
2350         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
# Line 1772  for (;;) Line 2361  for (;;)
2361        }        }
2362      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2363      if (      if (
2364  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2365         c >= 256 ||         c > 255 ||
2366  #endif  #endif
2367         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2368         )         )
# Line 1789  for (;;) Line 2378  for (;;)
2378        }        }
2379      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2380      if (      if (
2381  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2382         c < 256 &&         c < 256 &&
2383  #endif  #endif
2384         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
# Line 1806  for (;;) Line 2395  for (;;)
2395        }        }
2396      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2397      if (      if (
2398  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2399         c >= 256 ||         c > 255 ||
2400  #endif  #endif
2401         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2402         )         )
# Line 1825  for (;;) Line 2414  for (;;)
2414      switch(c)      switch(c)
2415        {        {
2416        default: RRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2417    
2418        case 0x000d:        case 0x000d:
2419        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr >= md->end_subject)
2420            {
2421            SCHECK_PARTIAL();
2422            }
2423          else if (*eptr == 0x0a) eptr++;
2424        break;        break;
2425    
2426        case 0x000a:        case 0x000a:
# Line 1981  for (;;) Line 2575  for (;;)
2575               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2576               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2577            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2578           break;          break;
2579    
2580          case PT_GC:          case PT_GC:
2581          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2582            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2583          break;          break;
2584    
# Line 1998  for (;;) Line 2592  for (;;)
2592            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2593          break;          break;
2594    
2595            /* These are specials */
2596    
2597            case PT_ALNUM:
2598            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2599                 PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2600              RRETURN(MATCH_NOMATCH);
2601            break;
2602    
2603            case PT_SPACE:    /* Perl space */
2604            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2605                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2606                   == (op == OP_NOTPROP))
2607              RRETURN(MATCH_NOMATCH);
2608            break;
2609    
2610            case PT_PXSPACE:  /* POSIX space */
2611            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2612                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2613                 c == CHAR_FF || c == CHAR_CR)
2614                   == (op == OP_NOTPROP))
2615              RRETURN(MATCH_NOMATCH);
2616            break;
2617    
2618            case PT_WORD:
2619            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2620                 PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2621                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2622              RRETURN(MATCH_NOMATCH);
2623            break;
2624    
2625            /* This should never occur */
2626    
2627          default:          default:
2628          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2629          }          }
# Line 2015  for (;;) Line 2641  for (;;)
2641        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2642        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2643        }        }
2644      GETCHARINCTEST(c, eptr);      else
2645        {        {
2646        int category = UCD_CATEGORY(c);        int lgb, rgb;
2647        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        GETCHARINCTEST(c, eptr);
2648          lgb = UCD_GRAPHBREAK(c);
2649        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2650          {          {
2651          int len = 1;          int len = 1;
2652          if (!utf8) c = *eptr; else          if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2653            {          rgb = UCD_GRAPHBREAK(c);
2654            GETCHARLEN(c, eptr, len);          if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
2655            }          lgb = rgb;
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
2656          eptr += len;          eptr += len;
2657          }          }
2658        }        }
2659        CHECK_PARTIAL();
2660      ecode++;      ecode++;
2661      break;      break;
2662  #endif  #endif
# Line 2045  for (;;) Line 2671  for (;;)
2671      loops). */      loops). */
2672    
2673      case OP_REF:      case OP_REF:
2674        {      case OP_REFI:
2675        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2676        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2677        ecode += 1 + IMM2_SIZE;
2678    
2679        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2680    
2681        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2682        than the amount of subject left; this ensures that every attempt at a      this ensures that every attempt at a match fails. We can't just fail
2683        match fails. We can't just fail here, because of the possibility of      here, because of the possibility of quantifiers with zero minima.
       quantifiers with zero minima.  
2684    
2685        (b) If the JavaScript compatibility flag is set, set the length to zero      (b) If the JavaScript compatibility flag is set, set the length to zero
2686        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2687    
2688        Otherwise, set the length to the length of what was matched by the      Otherwise, set the length to the length of what was matched by the
2689        referenced subpattern. */      referenced subpattern. */
2690    
2691        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2692          length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;        length = (md->jscript_compat)? 0 : -1;
2693        else      else
2694          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2695    
2696        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2697    
2698        switch (*ecode)      switch (*ecode)
2699          {        {
2700          case OP_CRSTAR:        case OP_CRSTAR:
2701          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2702          case OP_CRPLUS:        case OP_CRPLUS:
2703          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2704          case OP_CRQUERY:        case OP_CRQUERY:
2705          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2706          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2707          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2708          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2709          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2710          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2711          break;        break;
2712    
2713          case OP_CRRANGE:        case OP_CRRANGE:
2714          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2715          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2716          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2717          max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2718          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2719          ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2720          break;        break;
2721    
2722          default:               /* No repeat follows */        default:               /* No repeat follows */
2723          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2724            {          {
2725            CHECK_PARTIAL();          if (length == -2) eptr = md->end_subject;   /* Partial match */
2726            RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2727            }          RRETURN(MATCH_NOMATCH);
         eptr += length;  
         continue;              /* With the main loop */  
2728          }          }
2729          eptr += length;
2730          continue;              /* With the main loop */
2731          }
2732    
2733        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2734        main loop. */      zero, just continue with the main loop. If the length is negative, it
2735        means the reference is unset in non-Java-compatible mode. If the minimum is
2736        zero, we can continue at the same level without recursion. For any other
2737        minimum, carrying on will result in NOMATCH. */
2738    
2739        if (length == 0) continue;      if (length == 0) continue;
2740        if (length < 0 && min == 0) continue;
2741    
2742        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2743        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2744        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2745    
2746        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2747          {
2748          int slength;
2749          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2750          {          {
2751          if (!match_ref(offset, eptr, length, md, ims))          if (slength == -2) eptr = md->end_subject;   /* Partial match */
2752            {          CHECK_PARTIAL();
2753            CHECK_PARTIAL();          RRETURN(MATCH_NOMATCH);
           RRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2754          }          }
2755          eptr += slength;
2756          }
2757    
2758        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2759        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2760    
2761        if (min == max) continue;      if (min == max) continue;
2762    
2763        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2764    
2765        if (minimize)      if (minimize)
2766          {
2767          for (fi = min;; fi++)
2768          {          {
2769          for (fi = min;; fi++)          int slength;
2770            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2771            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2772            if (fi >= max) RRETURN(MATCH_NOMATCH);
2773            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2774            {            {
2775            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            if (slength == -2) eptr = md->end_subject;   /* Partial match */
2776            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            CHECK_PARTIAL();
2777            if (fi >= max) RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             RRETURN(MATCH_NOMATCH);  
             }  
           eptr += length;  
2778            }            }
2779          /* Control never gets here */          eptr += slength;
2780          }          }
2781          /* Control never gets here */
2782          }
2783    
2784        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2785    
2786        else      else
2787          {
2788          pp = eptr;
2789          for (i = min; i < max; i++)
2790          {          {
2791          pp = eptr;          int slength;
2792          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2793            {            {
2794            if (!match_ref(offset, eptr, length, md, ims))            /* Can't use CHECK_PARTIAL because we don't want to update eptr in
2795              the soft partial matching case. */
2796    
2797              if (slength == -2 && md->partial != 0 &&
2798                  md->end_subject > md->start_used_ptr)
2799              {              {
2800              CHECK_PARTIAL();              md->hitend = TRUE;
2801              break;              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2802              }              }
2803            eptr += length;            break;
           }  
         while (eptr >= pp)  
           {  
           RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);  
           if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
           eptr -= length;  
2804            }            }
2805          RRETURN(MATCH_NOMATCH);          eptr += slength;
2806            }
2807    
2808          while (eptr >= pp)
2809            {
2810            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2811            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2812            eptr -= length;
2813          }          }
2814          RRETURN(MATCH_NOMATCH);
2815        }        }
2816      /* Control never gets here */      /* Control never gets here */
2817    
# Line 2185  for (;;) Line 2829  for (;;)
2829      case OP_NCLASS:      case OP_NCLASS:
2830      case OP_CLASS:      case OP_CLASS:
2831        {        {
2832          /* The data variable is saved across frames, so the byte map needs to
2833          be stored there. */
2834    #define BYTE_MAP ((pcre_uint8 *)data)
2835        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2836        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2837    
2838        switch (*ecode)        switch (*ecode)
2839          {          {
# Line 2207  for (;;) Line 2854  for (;;)
2854          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2855          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2856          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2857          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2858          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2859          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2860          break;          break;
2861    
2862          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2219  for (;;) Line 2866  for (;;)
2866    
2867        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2868    
2869  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2870        /* UTF-8 mode */        if (utf)
       if (utf8)  
2871          {          {
2872          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2873            {            {
# Line 2236  for (;;) Line 2882  for (;;)
2882              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2883              }              }
2884            else            else
2885              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  
             }  
2886            }            }
2887          }          }
2888        else        else
2889  #endif  #endif
2890        /* Not UTF-8 mode */        /* Not UTF mode */
2891          {          {
2892          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2893            {            {
# Line 2253  for (;;) Line 2897  for (;;)
2897              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2898              }              }
2899            c = *eptr++;            c = *eptr++;
2900            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2901              if (c > 255)
2902                {
2903                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2904                }
2905              else
2906    #endif
2907                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2908            }            }
2909          }          }
2910    
# Line 2267  for (;;) Line 2918  for (;;)
2918    
2919        if (minimize)        if (minimize)
2920          {          {
2921  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2922          /* UTF-8 mode */          if (utf)
         if (utf8)  
2923            {            {
2924            for (fi = min;; fi++)            for (fi = min;; fi++)
2925              {              {
2926              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2927              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2928              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2929              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2287  for (;;) Line 2937  for (;;)
2937                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2938                }                }
2939              else              else
2940                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  
               }  
2941              }              }
2942            }            }
2943          else          else
2944  #endif  #endif
2945          /* Not UTF-8 mode */          /* Not UTF mode */
2946            {            {
2947            for (fi = min;; fi++)            for (fi = min;; fi++)
2948              {              {
2949              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2950              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2951              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2952              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2307  for (;;) Line 2955  for (;;)
2955                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2956                }                }
2957              c = *eptr++;              c = *eptr++;
2958              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2959                if (c > 255)
2960                  {
2961                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2962                  }
2963                else
2964    #endif
2965                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2966              }              }
2967            }            }
2968          /* Control never gets here */          /* Control never gets here */
# Line 2319  for (;;) Line 2974  for (;;)
2974          {          {
2975          pp = eptr;          pp = eptr;
2976    
2977  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2978          /* UTF-8 mode */          if (utf)
         if (utf8)  
2979            {            {
2980            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2981              {              {
# Line 2337  for (;;) Line 2991  for (;;)
2991                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
2992                }                }
2993              else              else
2994                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
2995              eptr += len;              eptr += len;
2996              }              }
2997            for (;;)            for (;;)
2998              {              {
2999              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
3000              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3001              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3002              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2352  for (;;) Line 3004  for (;;)
3004            }            }
3005          else          else
3006  #endif  #endif
3007            /* Not UTF-8 mode */            /* Not UTF mode */
3008            {            {
3009            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3010              {              {
# Line 2362  for (;;) Line 3014  for (;;)
3014                break;                break;
3015                }                }
3016              c = *eptr;              c = *eptr;
3017              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
3018                if (c > 255)
3019                  {
3020                  if (op == OP_CLASS) break;
3021                  }
3022                else
3023    #endif
3024                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3025              eptr++;              eptr++;
3026              }              }
3027            while (eptr >= pp)            while (eptr >= pp)
3028              {              {
3029              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
3030              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3031              eptr--;              eptr--;
3032              }              }
# Line 2375  for (;;) Line 3034  for (;;)
3034    
3035          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3036          }          }
3037    #undef BYTE_MAP
3038        }        }
3039      /* Control never gets here */      /* Control never gets here */
3040    
# Line 2383  for (;;) Line 3043  for (;;)
3043      when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8      when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8
3044      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
3045    
3046  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3047      case OP_XCLASS:      case OP_XCLASS:
3048        {        {
3049        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2408  for (;;) Line 3068  for (;;)
3068          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3069          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3070          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3071          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
3072          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
3073          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
3074          break;          break;
3075    
3076          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2428  for (;;) Line 3088  for (;;)
3088            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3089            }            }
3090          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
3091          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3092          }          }
3093    
3094        /* If max == min we can continue with the main loop without the        /* If max == min we can continue with the main loop without the
# Line 2443  for (;;) Line 3103  for (;;)
3103          {          {
3104          for (fi = min;; fi++)          for (fi = min;; fi++)
3105            {            {
3106            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
3107            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3108            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3109            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2452  for (;;) Line 3112  for (;;)
3112              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3113              }              }
3114            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3115            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3116            }            }
3117          /* Control never gets here */          /* Control never gets here */
3118          }          }
# Line 2470  for (;;) Line 3130  for (;;)
3130              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3131              break;              break;
3132              }              }
3133    #ifdef SUPPORT_UTF
3134            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3135            if (!_pcre_xclass(c, data)) break;  #else
3136              c = *eptr;
3137    #endif
3138              if (!PRIV(xclass)(c, data, utf)) break;
3139            eptr += len;            eptr += len;
3140            }            }
3141          for(;;)          for(;;)
3142            {            {
3143            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3144            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3145            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3146            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3147              if (utf) BACKCHAR(eptr);
3148    #endif
3149            }            }
3150          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3151          }          }
# Line 2491  for (;;) Line 3157  for (;;)
3157      /* Match a single character, casefully */      /* Match a single character, casefully */
3158    
3159      case OP_CHAR:      case OP_CHAR:
3160  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3161      if (utf8)      if (utf)
3162        {        {
3163        length = 1;        length = 1;
3164        ecode++;        ecode++;
# Line 2506  for (;;) Line 3172  for (;;)
3172        }        }
3173      else      else
3174  #endif  #endif
3175        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3176        {        {
3177        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3178          {          {
# Line 2519  for (;;) Line 3184  for (;;)
3184        }        }
3185      break;      break;
3186    
3187      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3188        subject, give up immediately. */
3189    
3190        case OP_CHARI:
3191        if (eptr >= md->end_subject)
3192          {
3193          SCHECK_PARTIAL();
3194          RRETURN(MATCH_NOMATCH);
3195          }
3196    
3197      case OP_CHARNC:  #ifdef SUPPORT_UTF
3198  #ifdef SUPPORT_UTF8      if (utf)
     if (utf8)  
3199        {        {
3200        length = 1;        length = 1;
3201        ecode++;        ecode++;
3202        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3203    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         RRETURN(MATCH_NOMATCH);  
         }  
   
3204        /* 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
3205        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3206          fast lookup table. We know that there is at least one byte left in the
3207          subject. */
3208    
3209        if (fc < 128)        if (fc < 128)
3210          {          {
3211          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[fc]
3212                != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3213            ecode++;
3214            eptr++;
3215          }          }
3216    
3217        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3218          use the value of "length" to check for sufficient bytes left, because the
3219          other case of the character may have more or fewer bytes.  */
3220    
3221        else        else
3222          {          {
# Line 2564  for (;;) Line 3237  for (;;)
3237          }          }
3238        }        }
3239      else      else
3240  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3241    
3242      /* Non-UTF-8 mode */      /* Not UTF mode */
3243        {        {
3244        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3245          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3246          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         RRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);  
3247        ecode += 2;        ecode += 2;
3248        }        }
3249      break;      break;
# Line 2581  for (;;) Line 3251  for (;;)
3251      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3252    
3253      case OP_EXACT:      case OP_EXACT:
3254        case OP_EXACTI:
3255      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3256      ecode += 3;      ecode += 1 + IMM2_SIZE;
3257      goto REPEATCHAR;      goto REPEATCHAR;
3258    
3259      case OP_POSUPTO:      case OP_POSUPTO:
3260        case OP_POSUPTOI:
3261      possessive = TRUE;      possessive = TRUE;
3262      /* Fall through */      /* Fall through */
3263    
3264      case OP_UPTO:      case OP_UPTO:
3265        case OP_UPTOI:
3266      case OP_MINUPTO:      case OP_MINUPTO:
3267        case OP_MINUPTOI:
3268      min = 0;      min = 0;
3269      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3270      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3271      ecode += 3;      ecode += 1 + IMM2_SIZE;
3272      goto REPEATCHAR;      goto REPEATCHAR;
3273    
3274      case OP_POSSTAR:      case OP_POSSTAR:
3275        case OP_POSSTARI:
3276      possessive = TRUE;      possessive = TRUE;
3277      min = 0;      min = 0;
3278      max = INT_MAX;      max = INT_MAX;
# Line 2605  for (;;) Line 3280  for (;;)
3280      goto REPEATCHAR;      goto REPEATCHAR;
3281    
3282      case OP_POSPLUS:      case OP_POSPLUS:
3283        case OP_POSPLUSI:
3284      possessive = TRUE;      possessive = TRUE;
3285      min = 1;      min = 1;
3286      max = INT_MAX;      max = INT_MAX;
# Line 2612  for (;;) Line 3288  for (;;)
3288      goto REPEATCHAR;      goto REPEATCHAR;
3289    
3290      case OP_POSQUERY:      case OP_POSQUERY:
3291        case OP_POSQUERYI:
3292      possessive = TRUE;      possessive = TRUE;
3293      min = 0;      min = 0;
3294      max = 1;      max = 1;
# Line 2619  for (;;) Line 3296  for (;;)
3296      goto REPEATCHAR;      goto REPEATCHAR;
3297    
3298      case OP_STAR:      case OP_STAR:
3299        case OP_STARI:
3300      case OP_MINSTAR:      case OP_MINSTAR:
3301        case OP_MINSTARI:
3302      case OP_PLUS:      case OP_PLUS:
3303        case OP_PLUSI:
3304      case OP_MINPLUS:      case OP_MINPLUS:
3305        case OP_MINPLUSI:
3306      case OP_QUERY:      case OP_QUERY:
3307        case OP_QUERYI:
3308      case OP_MINQUERY:      case OP_MINQUERY:
3309      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3310        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3311      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3312      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3313      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3314      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2634  for (;;) Line 3316  for (;;)
3316      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3317    
3318      REPEATCHAR:      REPEATCHAR:
3319  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3320      if (utf8)      if (utf)
3321        {        {
3322        length = 1;        length = 1;
3323        charptr = ecode;        charptr = ecode;
# Line 2649  for (;;) Line 3331  for (;;)
3331          {          {
3332  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3333          unsigned int othercase;          unsigned int othercase;
3334          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3335              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3336            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3337          else oclength = 0;          else oclength = 0;
3338  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3339    
3340          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3341            {            {
3342            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3343              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3344  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3345            else if (oclength > 0 &&            else if (oclength > 0 &&
3346                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3347                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3348  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3349            else            else
3350              {              {
# Line 2677  for (;;) Line 3359  for (;;)
3359            {            {
3360            for (fi = min;; fi++)            for (fi = min;; fi++)
3361              {              {
3362              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3363              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3364              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3365              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3366                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3367  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3368              else if (oclength > 0 &&              else if (oclength > 0 &&
3369                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3370                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3371  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3372              else              else
3373                {                {
# Line 2702  for (;;) Line 3384  for (;;)
3384            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3385              {              {
3386              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3387                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3388  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3389              else if (oclength > 0 &&              else if (oclength > 0 &&
3390                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3391                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3392  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3393              else              else
3394                {                {
# Line 2719  for (;;) Line 3401  for (;;)
3401    
3402            for(;;)            for(;;)
3403              {              {
3404              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3405              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3406              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3407  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2738  for (;;) Line 3420  for (;;)
3420        value of fc will always be < 128. */        value of fc will always be < 128. */
3421        }        }
3422      else      else
3423  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3424          /* When not in UTF-8 mode, load a single-byte character. */
3425      /* When not in UTF-8 mode, load a single-byte character. */        fc = *ecode++;
   
     fc = *ecode++;  
3426    
3427      /* 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 one character, though we may
3428      may not be in UTF-8 mode. The code is duplicated for the caseless and      or may not be in UTF mode. The code is duplicated for the caseless and
3429      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3430      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3431      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 2754  for (;;) Line 3434  for (;;)
3434      maximizing, find the maximum number of characters and work backwards. */      maximizing, find the maximum number of characters and work backwards. */
3435    
3436      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3437        max, eptr));        max, (char *)eptr));
3438    
3439      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3440        {        {
3441        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3442          /* fc must be < 128 if UTF is enabled. */
3443          foc = md->fcc[fc];
3444    #else
3445    #ifdef SUPPORT_UTF
3446    #ifdef SUPPORT_UCP
3447          if (utf && fc > 127)
3448            foc = UCD_OTHERCASE(fc);
3449    #else
3450          if (utf && fc > 127)
3451            foc = fc;
3452    #endif /* SUPPORT_UCP */
3453          else
3454    #endif /* SUPPORT_UTF */
3455            foc = TABLE_GET(fc, md->fcc, fc);
3456    #endif /* COMPILE_PCRE8 */
3457    
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)
# Line 2766  for (;;) Line 3462  for (;;)
3462            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3463            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3464            }            }
3465          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3466            eptr++;
3467          }          }
3468        if (min == max) continue;        if (min == max) continue;
3469        if (minimize)        if (minimize)
3470          {          {
3471          for (fi = min;; fi++)          for (fi = min;; fi++)
3472            {            {
3473            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3474            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3475            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3476            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2781  for (;;) Line 3478  for (;;)
3478              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3479              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3480              }              }
3481            if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3482              eptr++;
3483            }            }
3484          /* Control never gets here */          /* Control never gets here */
3485          }          }
# Line 2795  for (;;) Line 3493  for (;;)
3493              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3494              break;              break;
3495              }              }
3496            if (fc != md->lcc[*eptr]) break;            if (fc != *eptr && foc != *eptr) break;
3497            eptr++;            eptr++;
3498            }            }
3499    
# Line 2803  for (;;) Line 3501  for (;;)
3501    
3502          while (eptr >= pp)          while (eptr >= pp)
3503            {            {
3504            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3505            eptr--;            eptr--;
3506            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3507            }            }
# Line 2832  for (;;) Line 3530  for (;;)
3530          {          {
3531          for (fi = min;; fi++)          for (fi = min;; fi++)
3532            {            {
3533            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3534            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3535            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3536            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2861  for (;;) Line 3559  for (;;)
3559    
3560          while (eptr >= pp)          while (eptr >= pp)
3561            {            {
3562            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3563            eptr--;            eptr--;
3564            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3565            }            }
# Line 2874  for (;;) Line 3572  for (;;)
3572      checking can be multibyte. */      checking can be multibyte. */
3573    
3574      case OP_NOT:      case OP_NOT:
3575        case OP_NOTI:
3576      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3577        {        {
3578        SCHECK_PARTIAL();        SCHECK_PARTIAL();
3579        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
3580        }        }
3581      ecode++;  #ifdef SUPPORT_UTF
3582      GETCHARINCTEST(c, eptr);      if (utf)
     if ((ims & PCRE_CASELESS) != 0)  
3583        {        {
3584  #ifdef SUPPORT_UTF8        register unsigned int ch, och;
3585        if (c < 256)  
3586  #endif        ecode++;
3587        c = md->lcc[c];        GETCHARINC(ch, ecode);
3588        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        GETCHARINC(c, eptr);
3589    
3590          if (op == OP_NOT)
3591            {
3592            if (ch == c) RRETURN(MATCH_NOMATCH);
3593            }
3594          else
3595            {
3596    #ifdef SUPPORT_UCP
3597            if (ch > 127)
3598              och = UCD_OTHERCASE(ch);
3599    #else
3600            if (ch > 127)
3601              och = ch;
3602    #endif /* SUPPORT_UCP */
3603            else
3604              och = TABLE_GET(ch, md->fcc, ch);
3605            if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3606            }
3607        }        }
3608      else      else
3609    #endif
3610        {        {
3611        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        register unsigned int ch = ecode[1];
3612          c = *eptr++;
3613          if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c))
3614            RRETURN(MATCH_NOMATCH);
3615          ecode += 2;
3616        }        }
3617      break;      break;
3618    
# Line 2903  for (;;) Line 3624  for (;;)
3624      about... */      about... */
3625    
3626      case OP_NOTEXACT:      case OP_NOTEXACT:
3627        case OP_NOTEXACTI:
3628      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3629      ecode += 3;      ecode += 1 + IMM2_SIZE;
3630      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3631    
3632      case OP_NOTUPTO:      case OP_NOTUPTO:
3633        case OP_NOTUPTOI:
3634      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3635        case OP_NOTMINUPTOI:
3636      min = 0;      min = 0;
3637      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3638      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3639      ecode += 3;      ecode += 1 + IMM2_SIZE;
3640      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3641    
3642      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3643        case OP_NOTPOSSTARI:
3644      possessive = TRUE;      possessive = TRUE;
3645      min = 0;      min = 0;
3646      max = INT_MAX;      max = INT_MAX;
# Line 2923  for (;;) Line 3648  for (;;)
3648      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3649    
3650      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3651        case OP_NOTPOSPLUSI:
3652      possessive = TRUE;      possessive = TRUE;
3653      min = 1;      min = 1;
3654      max = INT_MAX;      max = INT_MAX;
# Line 2930  for (;;) Line 3656  for (;;)
3656      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3657    
3658      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3659        case OP_NOTPOSQUERYI:
3660      possessive = TRUE;      possessive = TRUE;
3661      min = 0;      min = 0;
3662      max = 1;      max = 1;
# Line 2937  for (;;) Line 3664  for (;;)
3664      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3665    
3666      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3667        case OP_NOTPOSUPTOI:
3668      possessive = TRUE;      possessive = TRUE;
3669      min = 0;      min = 0;
3670      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3671      ecode += 3;      ecode += 1 + IMM2_SIZE;
3672      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3673    
3674      case OP_NOTSTAR:      case OP_NOTSTAR:
3675        case OP_NOTSTARI:
3676      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3677        case OP_NOTMINSTARI:
3678      case OP_NOTPLUS:      case OP_NOTPLUS:
3679        case OP_NOTPLUSI:
3680      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3681        case OP_NOTMINPLUSI:
3682      case OP_NOTQUERY:      case OP_NOTQUERY:
3683        case OP_NOTQUERYI:
3684      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3685      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3686        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3687      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3688      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3689      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 2958  for (;;) Line 3692  for (;;)
3692      /* Common code for all repeated single-byte matches. */      /* Common code for all repeated single-byte matches. */
3693    
3694      REPEATNOTCHAR:      REPEATNOTCHAR:
3695      fc = *ecode++;      GETCHARINCTEST(fc, ecode);
3696    
3697      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
3698      since matching characters is likely to be quite common. First, ensure the      since matching characters is likely to be quite common. First, ensure the
# Line 2969  for (;;) Line 3703  for (;;)
3703      characters and work backwards. */      characters and work backwards. */
3704    
3705      DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3706        max, eptr));        max, (char *)eptr));
3707    
3708      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3709        {        {
3710        fc = md->lcc[fc];  #ifdef SUPPORT_UTF
3711    #ifdef SUPPORT_UCP
3712          if (utf && fc > 127)
3713            foc = UCD_OTHERCASE(fc);
3714    #else
3715          if (utf && fc > 127)
3716            foc = fc;
3717    #endif /* SUPPORT_UCP */
3718          else
3719    #endif /* SUPPORT_UTF */
3720            foc = TABLE_GET(fc, md->fcc, fc);
3721    
3722  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3723        /* UTF-8 mode */        if (utf)
       if (utf8)  
3724          {          {
3725          register unsigned int d;          register unsigned int d;
3726          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 2988  for (;;) Line 3731  for (;;)
3731              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3732              }              }
3733            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3734            if (d < 256) d = md->lcc[d];            if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) RRETURN(MATCH_NOMATCH);  
3735            }            }
3736          }          }
3737        else        else
3738  #endif  #endif
3739          /* Not UTF mode */
       /* Not UTF-8 mode */  
3740          {          {
3741          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3742            {            {
# Line 3004  for (;;) Line 3745  for (;;)
3745              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3746              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3747              }              }
3748            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3749              eptr++;
3750            }            }
3751          }          }
3752    
# Line 3012  for (;;) Line 3754  for (;;)
3754    
3755        if (minimize)        if (minimize)
3756          {          {
3757  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3758          /* UTF-8 mode */          if (utf)
         if (utf8)  
3759            {            {
3760            register unsigned int d;            register unsigned int d;
3761            for (fi = min;; fi++)            for (fi = min;; fi++)
3762              {              {
3763              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3764              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3765              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3766              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3028  for (;;) Line 3769  for (;;)
3769                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3770                }                }
3771              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3772              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
             if (fc == d) RRETURN(MATCH_NOMATCH);  
3773              }              }
3774            }            }
3775          else          else
3776  #endif  #endif
3777          /* Not UTF-8 mode */          /* Not UTF mode */
3778            {            {
3779            for (fi = min;; fi++)            for (fi = min;; fi++)
3780              {              {
3781              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3782              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3783              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3784              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3046  for (;;) Line 3786  for (;;)
3786                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3787                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3788                }                }
3789              if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);              if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3790                eptr++;
3791              }              }
3792            }            }
3793          /* Control never gets here */          /* Control never gets here */
# Line 3058  for (;;) Line 3799  for (;;)
3799          {          {
3800          pp = eptr;          pp = eptr;
3801    
3802  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3803          /* UTF-8 mode */          if (utf)
         if (utf8)  
3804            {            {
3805            register unsigned int d;            register unsigned int d;
3806            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3072  for (;;) Line 3812  for (;;)
3812                break;                break;
3813                }                }
3814              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3815              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) break;
             if (fc == d) break;  
3816              eptr += len;              eptr += len;
3817              }              }
3818          if (possessive) continue;            if (possessive) continue;
3819          for(;;)            for(;;)
3820              {              {
3821              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3822              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3823              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3824              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3087  for (;;) Line 3826  for (;;)
3826            }            }
3827          else          else
3828  #endif  #endif
3829          /* Not UTF-8 mode */          /* Not UTF mode */
3830            {            {
3831            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3832              {              {
# Line 3096  for (;;) Line 3835  for (;;)
3835                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3836                break;                break;
3837                }                }
3838              if (fc == md->lcc[*eptr]) break;              if (fc == *eptr || foc == *eptr) break;
3839              eptr++;              eptr++;
3840              }              }
3841            if (possessive) continue;            if (possessive) continue;
3842            while (eptr >= pp)            while (eptr >= pp)
3843              {              {
3844              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3845              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3846              eptr--;              eptr--;
3847              }              }
# Line 3117  for (;;) Line 3856  for (;;)
3856    
3857      else      else
3858        {        {
3859  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3860        /* UTF-8 mode */        if (utf)
       if (utf8)  
3861          {          {
3862          register unsigned int d;          register unsigned int d;
3863          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3135  for (;;) Line 3873  for (;;)
3873          }          }
3874        else        else
3875  #endif  #endif
3876        /* Not UTF-8 mode */        /* Not UTF mode */
3877          {          {
3878          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3879            {            {
# Line 3152  for (;;) Line 3890  for (;;)
3890    
3891        if (minimize)        if (minimize)
3892          {          {
3893  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3894          /* UTF-8 mode */          if (utf)
         if (utf8)  
3895            {            {
3896            register unsigned int d;            register unsigned int d;
3897            for (fi = min;; fi++)            for (fi = min;; fi++)
3898              {              {
3899              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3900              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3901              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3902              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3173  for (;;) Line 3910  for (;;)
3910            }            }
3911          else          else
3912  #endif  #endif
3913          /* Not UTF-8 mode */          /* Not UTF mode */
3914            {            {
3915            for (fi = min;; fi++)            for (fi = min;; fi++)
3916              {              {
3917              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3918              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3919              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3920              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3197  for (;;) Line 3934  for (;;)
3934          {          {
3935          pp = eptr;          pp = eptr;
3936    
3937  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3938          /* UTF-8 mode */          if (utf)
         if (utf8)  
3939            {            {
3940            register unsigned int d;            register unsigned int d;
3941            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3217  for (;;) Line 3953  for (;;)
3953            if (possessive) continue;            if (possessive) continue;
3954            for(;;)            for(;;)
3955              {              {
3956              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3957              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3958              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3959              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3225  for (;;) Line 3961  for (;;)
3961            }            }
3962          else          else
3963  #endif  #endif
3964          /* Not UTF-8 mode */          /* Not UTF mode */
3965            {            {
3966            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3967              {              {
# Line 3240  for (;;) Line 3976  for (;;)
3976            if (possessive) continue;            if (possessive) continue;
3977            while (eptr >= pp)            while (eptr >= pp)
3978              {              {
3979              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3980              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3981              eptr--;              eptr--;
3982              }              }
# Line 3258  for (;;) Line 3994  for (;;)
3994      case OP_TYPEEXACT:      case OP_TYPEEXACT:
3995      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3996      minimize = TRUE;      minimize = TRUE;
3997      ecode += 3;      ecode += 1 + IMM2_SIZE;
3998      goto REPEATTYPE;      goto REPEATTYPE;
3999    
4000      case OP_TYPEUPTO:      case OP_TYPEUPTO:
# Line 3266  for (;;) Line 4002  for (;;)
4002      min = 0;      min = 0;
4003      max = GET2(ecode, 1);      max = GET2(ecode, 1);
4004      minimize = *ecode == OP_TYPEMINUPTO;      minimize = *ecode == OP_TYPEMINUPTO;
4005      ecode += 3;      ecode += 1 + IMM2_SIZE;
4006      goto REPEATTYPE;      goto REPEATTYPE;
4007    
4008      case OP_TYPEPOSSTAR:      case OP_TYPEPOSSTAR:
# Line 3294  for (;;) Line 4030  for (;;)
4030      possessive = TRUE;      possessive = TRUE;
4031      min = 0;      min = 0;
4032      max = GET2(ecode, 1);      max = GET2(ecode, 1);
4033      ecode += 3;      ecode += 1 + IMM2_SIZE;
4034      goto REPEATTYPE;      goto REPEATTYPE;
4035    
4036      case OP_TYPESTAR:      case OP_TYPESTAR:
# Line 3355  for (;;) Line 4091  for (;;)
4091            case PT_LAMP:            case PT_LAMP:
4092            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4093              {              {
4094                int chartype;
4095              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4096                {                {
4097                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4098                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4099                }                }
4100              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4101              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4102              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4103                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4104                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4105                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4106              }              }
4107            break;            break;
# Line 3378  for (;;) Line 4115  for (;;)
4115                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4116                }                }
4117              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4118              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4119                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4120              }              }
4121            break;            break;
# Line 3393  for (;;) Line 4129  for (;;)
4129                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4130                }                }
4131              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4132              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4133                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4134              }              }
4135            break;            break;
# Line 3408  for (;;) Line 4143  for (;;)
4143                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4144                }                }
4145              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4146              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
4147              if ((prop_script == prop_value) == prop_fail_result)                RRETURN(MATCH_NOMATCH);
4148                }
4149              break;
4150    
4151              case PT_ALNUM:
4152              for (i = 1; i <= min; i++)
4153                {
4154                int category;
4155                if (eptr >= md->end_subject)
4156                  {
4157                  SCHECK_PARTIAL();
4158                  RRETURN(MATCH_NOMATCH);
4159                  }
4160                GETCHARINCTEST(c, eptr);
4161                category = UCD_CATEGORY(c);
4162                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
4163                  RRETURN(MATCH_NOMATCH);
4164                }
4165              break;
4166    
4167              case PT_SPACE:    /* Perl space */
4168              for (i = 1; i <= min; i++)
4169                {
4170                if (eptr >= md->end_subject)
4171                  {
4172                  SCHECK_PARTIAL();
4173                  RRETURN(MATCH_NOMATCH);
4174                  }
4175                GETCHARINCTEST(c, eptr);
4176                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4177                     c == CHAR_FF || c == CHAR_CR)
4178                       == prop_fail_result)
4179                  RRETURN(MATCH_NOMATCH);
4180                }
4181              break;
4182    
4183              case PT_PXSPACE:  /* POSIX space */
4184              for (i = 1; i <= min; i++)
4185                {
4186                if (eptr >= md->end_subject)
4187                  {
4188                  SCHECK_PARTIAL();
4189                  RRETURN(MATCH_NOMATCH);
4190                  }
4191                GETCHARINCTEST(c, eptr);
4192                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4193                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4194                       == prop_fail_result)
4195                  RRETURN(MATCH_NOMATCH);
4196                }
4197              break;
4198    
4199              case PT_WORD:
4200              for (i = 1; i <= min; i++)
4201                {
4202                int category;
4203                if (eptr >= md->end_subject)
4204                  {
4205                  SCHECK_PARTIAL();
4206                  RRETURN(MATCH_NOMATCH);
4207                  }
4208                GETCHARINCTEST(c, eptr);
4209                category = UCD_CATEGORY(c);
4210                if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
4211                       == prop_fail_result)
4212                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4213              }              }
4214            break;            break;
4215    
4216              /* This should not occur */
4217    
4218            default:            default:
4219            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
4220            }            }
# Line 3431  for (;;) Line 4232  for (;;)
4232              SCHECK_PARTIAL();              SCHECK_PARTIAL();
4233              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
4234              }              }
4235            GETCHARINCTEST(c, eptr);            else
4236            prop_category = UCD_CATEGORY(c);              {
4237            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);              int lgb, rgb;
4238            while (eptr < md->end_subject)              GETCHARINCTEST(c, eptr);
4239              {              lgb = UCD_GRAPHBREAK(c);
4240              int len = 1;             while (eptr < md->end_subject)
4241              if (!utf8) c = *eptr;                {
4242                else { GETCHARLEN(c, eptr, len); }                int len = 1;
4243              prop_category = UCD_CATEGORY(c);                if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4244              if (prop_category != ucp_M) break;                rgb = UCD_GRAPHBREAK(c);
4245              eptr += len;                if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
4246                  lgb = rgb;
4247                  eptr += len;
4248                  }
4249              }              }
4250              CHECK_PARTIAL();
4251            }            }
4252          }          }
4253    
# Line 3451  for (;;) Line 4256  for (;;)
4256    
4257  /* Handle all other cases when the coding is UTF-8 */  /* Handle all other cases when the coding is UTF-8 */
4258    
4259  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4260