/[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 455 by ph10, Sat Sep 26 19:12:32 2009 UTC revision 933 by ph10, Sat Feb 25 12:18:23 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) 1997-2009 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 57  possible. There are also some static sup Line 57  possible. There are also some static sup
57  #undef min  #undef min
58  #undef max  #undef max
59    
60  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
61    of call to match(). We do it this way to save on using another stack variable,
62    as stack usage is to be discouraged. */
63    
64  #define match_condassert     0x01  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
65  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
66    
67  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
68  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 71  defined PCRE_ERROR_xxx codes, which are Line 73  defined PCRE_ERROR_xxx codes, which are
73  /* Special internal returns from the match() function. Make them sufficiently  /* Special internal returns from the match() function. Make them sufficiently
74  negative to avoid the external error codes. */  negative to avoid the external error codes. */
75    
76  #define MATCH_COMMIT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_PRUNE        (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_SKIP         (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_THEN         (-996)  #define MATCH_ONCE         (-996)
80    #define MATCH_PRUNE        (-995)
81    #define MATCH_SKIP         (-994)
82    #define MATCH_SKIP_ARG     (-993)
83    #define MATCH_THEN         (-992)
84    
85  /* 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.
86  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 89  static const char rep_max[] = { 0, 0, 0, Line 95  static const char rep_max[] = { 0, 0, 0,
95    
96    
97    
98  #ifdef DEBUG  #ifdef PCRE_DEBUG
99  /*************************************************  /*************************************************
100  *        Debugging function to print chars       *  *        Debugging function to print chars       *
101  *************************************************/  *************************************************/
# Line 107  Returns:     nothing Line 113  Returns:     nothing
113  */  */
114    
115  static void  static void
116  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)
117  {  {
118  unsigned int c;  unsigned int c;
119  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 128  while (length-- > 0)
128  *          Match a back-reference                *  *          Match a back-reference                *
129  *************************************************/  *************************************************/
130    
131  /* 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
132  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
133    the length passed is zero. Note that in caseless UTF-8 mode, the number of
134    subject bytes matched may be different to the number of reference bytes.
135    
136  Arguments:  Arguments:
137    offset      index into the offset vector    offset      index into the offset vector
138    eptr        points into the subject    eptr        pointer into the subject
139    length      length to be matched    length      length of reference to be matched (number of bytes)
140    md          points to match data block    md          points to match data block
141    ims         the ims flags    caseless    TRUE if caseless
142    
143  Returns:      TRUE if matched  Returns:      >= 0 the number of subject bytes matched
144                  -1 no match
145                  -2 partial match; always given if at end subject
146  */  */
147    
148  static BOOL  static int
149  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,
150    unsigned long int ims)    BOOL caseless)
151  {  {
152  USPTR p = md->start_subject + md->offset_vector[offset];  PCRE_PUCHAR eptr_start = eptr;
153    register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
154    
155  #ifdef DEBUG  #ifdef PCRE_DEBUG
156  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
157    printf("matching subject <null>");    printf("matching subject <null>");
158  else  else
# Line 154  pchars(p, length, FALSE, md); Line 165  pchars(p, length, FALSE, md);
165  printf("\n");  printf("\n");
166  #endif  #endif
167    
168  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible - in that
169    case the length is passed as zero). */
170    
171  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
172    
173  /* 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
174  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
175  ASCII characters. */  ASCII characters. */
176    
177  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
178    {    {
179  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
180  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
181    if (md->utf8)    if (md->utf)
182      {      {
183      USPTR endptr = eptr + length;      /* Match characters up to the end of the reference. NOTE: the number of
184      while (eptr < endptr)      bytes matched may differ, because there are some characters whose upper and
185        lower case versions code as different numbers of bytes. For example, U+023A
186        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
187        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
188        the latter. It is important, therefore, to check the length along the
189        reference, not along the subject (earlier code did this wrong). */
190    
191        PCRE_PUCHAR endptr = p + length;
192        while (p < endptr)
193        {        {
194        int c, d;        int c, d;
195          if (eptr >= md->end_subject) return -2;   /* Partial match */
196        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
197        GETCHARINC(d, p);        GETCHARINC(d, p);
198        if (c != d && c != UCD_OTHERCASE(d)) return FALSE;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
199        }        }
200      }      }
201    else    else
# Line 183  if ((ims & PCRE_CASELESS) != 0) Line 204  if ((ims & PCRE_CASELESS) != 0)
204    
205    /* 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
206    is no UCP support. */    is no UCP support. */
207        {
208    while (length-- > 0)      while (length-- > 0)
209      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }        {
210          if (eptr >= md->end_subject) return -2;   /* Partial match */
211          if (TABLE_GET(*p, md->lcc, *p) != TABLE_GET(*eptr, md->lcc, *eptr)) return -1;
212          p++;
213          eptr++;
214          }
215        }
216    }    }
217    
218  /* 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
219  are in UTF-8 mode. */  are in UTF-8 mode. */
220    
221  else  else
222    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
223      while (length-- > 0)
224        {
225        if (eptr >= md->end_subject) return -2;   /* Partial match */
226        if (*p++ != *eptr++) return -1;
227        }
228      }
229    
230  return TRUE;  return (int)(eptr - eptr_start);
231  }  }
232    
233    
# Line 245  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 278  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
278         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
279         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
280         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
281         RM51,  RM52, RM53, RM54 };         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
282           RM61,  RM62, RM63, RM64, RM65, RM66 };
283    
284  /* 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
285  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
286  actuall used in this definition. */  actually used in this definition. */
287    
288  #ifndef NO_RECURSE  #ifndef NO_RECURSE
289  #define REGISTER register  #define REGISTER register
290    
291  #ifdef DEBUG  #ifdef PCRE_DEBUG
292  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
293    { \    { \
294    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
295    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \
296    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
297    }    }
298  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 267  actuall used in this definition. */ Line 301  actuall used in this definition. */
301    return ra; \    return ra; \
302    }    }
303  #else  #else
304  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
305    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1)
306  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
307  #endif  #endif
308    
# Line 281  argument of match(), which never changes Line 315  argument of match(), which never changes
315    
316  #define REGISTER  #define REGISTER
317    
318  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
319    {\    {\
320    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = frame->Xnextframe;\
321    frame->Xwhere = rw; \    if (newframe == NULL)\
322        {\
323        newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\
324        if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
325        newframe->Xnextframe = NULL;\
326        frame->Xnextframe = newframe;\
327        }\
328      frame->Xwhere = rw;\
329    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
330    newframe->Xecode = rb;\    newframe->Xecode = rb;\
331    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
332    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
333    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
334    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
335    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
336    frame = newframe;\    frame = newframe;\
# Line 303  argument of match(), which never changes Line 342  argument of match(), which never changes
342    
343  #define RRETURN(ra)\  #define RRETURN(ra)\
344    {\    {\
345    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
346    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
   (pcre_stack_free)(newframe);\  
347    if (frame != NULL)\    if (frame != NULL)\
348      {\      {\
349      rrc = ra;\      rrc = ra;\
# Line 319  argument of match(), which never changes Line 357  argument of match(), which never changes
357    
358  typedef struct heapframe {  typedef struct heapframe {
359    struct heapframe *Xprevframe;    struct heapframe *Xprevframe;
360      struct heapframe *Xnextframe;
361    
362    /* Function arguments that may change */    /* Function arguments that may change */
363    
364    USPTR Xeptr;    PCRE_PUCHAR Xeptr;
365    const uschar *Xecode;    const pcre_uchar *Xecode;
366    USPTR Xmstart;    PCRE_PUCHAR Xmstart;
367    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
368    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
369    unsigned int Xrdepth;    unsigned int Xrdepth;
370    
371    /* Function local variables */    /* Function local variables */
372    
373    USPTR Xcallpat;    PCRE_PUCHAR Xcallpat;
374  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
375    USPTR Xcharptr;    PCRE_PUCHAR Xcharptr;
376  #endif  #endif
377    USPTR Xdata;    PCRE_PUCHAR Xdata;
378    USPTR Xnext;    PCRE_PUCHAR Xnext;
379    USPTR Xpp;    PCRE_PUCHAR Xpp;
380    USPTR Xprev;    PCRE_PUCHAR Xprev;
381    USPTR Xsaved_eptr;    PCRE_PUCHAR Xsaved_eptr;
382    
383    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
384    
# Line 349  typedef struct heapframe { Line 386  typedef struct heapframe {
386    BOOL Xcondition;    BOOL Xcondition;
387    BOOL Xprev_is_word;    BOOL Xprev_is_word;
388    
   unsigned long int Xoriginal_ims;  
   
389  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
390    int Xprop_type;    int Xprop_type;
391    int Xprop_value;    int Xprop_value;
392    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
393    int Xoclength;    int Xoclength;
394    uschar Xocchars[8];    pcre_uchar Xocchars[6];
395  #endif  #endif
396    
397    int Xcodelink;    int Xcodelink;
# Line 401  returns a negative (error) response, the Line 433  returns a negative (error) response, the
433  same response. */  same response. */
434    
435  /* 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
436  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
437  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.
438  something has been matched). For hard partial matching, we then return  something has been matched). For hard partial matching, we then return
439  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
440  the subject. */  the subject. */
441    
442  #define CHECK_PARTIAL()\  #define CHECK_PARTIAL()\
443    if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\    if (md->partial != 0 && eptr >= md->end_subject && \
444      {\        eptr > md->start_used_ptr) \
445      md->hitend = TRUE;\      { \
446      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\      md->hitend = TRUE; \
447        if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
448      }      }
449    
450  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
451    if (md->partial && eptr > mstart)\    if (md->partial != 0 && eptr > md->start_used_ptr) \
452      {\      { \
453      md->hitend = TRUE;\      md->hitend = TRUE; \
454      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
455      }      }
456    
457    
458  /* Performance note: It might be tempting to extract commonly used fields from  /* Performance note: It might be tempting to extract commonly used fields from
459  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
460  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
461  made performance worse.  made performance worse.
462    
# Line 434  Arguments: Line 467  Arguments:
467                   by encountering \K)                   by encountering \K)
468     offset_top  current top pointer     offset_top  current top pointer
469     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
470     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
471                   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  
472     rdepth      the recursion depth     rdepth      the recursion depth
473    
474  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
475                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
476                   a negative MATCH_xxx value for PRUNE, SKIP, etc
477                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
478                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
479  */  */
480    
481  static int  static int
482  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode,
483    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
484    int flags, unsigned int rdepth)    unsigned int rdepth)
485  {  {
486  /* 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,
487  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 490  so they can be ordinary variables in all
490  register int  rrc;         /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
491  register int  i;           /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
492  register unsigned int c;   /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
493  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf;         /* Local copy of UTF flag for speed */
494    
495  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
496    BOOL caseless;
497  int condcode;  int condcode;
498    
499  /* 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
500  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
501  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
502  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  whenever RMATCH() does a "recursion". See the macro definitions above. Putting
503    the top-level on the stack rather than malloc-ing them all gives a performance
504    boost in many cases where there is not much "recursion". */
505    
506  #ifdef NO_RECURSE  #ifdef NO_RECURSE
507  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)md->match_frames_base;
 frame->Xprevframe = NULL;            /* Marks the top level */  
508    
509  /* Copy in the original argument variables */  /* Copy in the original argument variables */
510    
# Line 481  frame->Xeptr = eptr; Line 512  frame->Xeptr = eptr;
512  frame->Xecode = ecode;  frame->Xecode = ecode;
513  frame->Xmstart = mstart;  frame->Xmstart = mstart;
514  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
515  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
516  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
517    
518  /* 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 525  HEAP_RECURSE:
525  #define ecode              frame->Xecode  #define ecode              frame->Xecode
526  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
527  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
528  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
529  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
530    
531  /* Ditto for the local variables */  /* Ditto for the local variables */
532    
533  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
534  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
535  #endif  #endif
536  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 520  HEAP_RECURSE: Line 547  HEAP_RECURSE:
547  #define condition          frame->Xcondition  #define condition          frame->Xcondition
548  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
549    
 #define original_ims       frame->Xoriginal_ims  
   
550  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
551  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
552  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
553  #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  
554  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
555  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
556  #endif  #endif
# Line 558  i, and fc and c, can be the same variabl Line 580  i, and fc and c, can be the same variabl
580  #define fi i  #define fi i
581  #define fc c  #define fc c
582    
583    /* Many of the following variables are used only in small blocks of the code.
584    My normal style of coding would have declared them within each of those blocks.
585    However, in order to accommodate the version of this code that uses an external
586    "stack" implemented on the heap, it is easier to declare them all here, so the
587    declarations can be cut out in a block. The only declarations within blocks
588    below are for variables that do not have to be preserved over a recursive call
589    to RMATCH(). */
590    
591    #ifdef SUPPORT_UTF
592    const pcre_uchar *charptr;
593    #endif
594    const pcre_uchar *callpat;
595    const pcre_uchar *data;
596    const pcre_uchar *next;
597    PCRE_PUCHAR       pp;
598    const pcre_uchar *prev;
599    PCRE_PUCHAR       saved_eptr;
600    
601  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  recursion_info new_recursive;
602  const uschar *charptr;             /* in small blocks of the code. My normal */  
603  #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().          */  
604  BOOL condition;  BOOL condition;
605  BOOL prev_is_word;  BOOL prev_is_word;
606    
 unsigned long int original_ims;  
   
607  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
608  int prop_type;  int prop_type;
609  int prop_value;  int prop_value;
610  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
611  int oclength;  int oclength;
612  uschar occhars[8];  pcre_uchar occhars[6];
613  #endif  #endif
614    
615  int codelink;  int codelink;
# Line 601  int save_offset1, save_offset2, save_off Line 625  int save_offset1, save_offset2, save_off
625  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
626    
627  eptrblock newptrb;  eptrblock newptrb;
628    
629    /* There is a special fudge for calling match() in a way that causes it to
630    measure the size of its basic stack frame when the stack is being used for
631    recursion. The second argument (ecode) being NULL triggers this behaviour. It
632    cannot normally ever be NULL. The return is the negated value of the frame
633    size. */
634    
635    if (ecode == NULL)
636      {
637      if (rdepth == 0)
638        return match((PCRE_PUCHAR)&rdepth, NULL, NULL, 0, NULL, NULL, 1);
639      else
640        {
641        int len = (char *)&rdepth - (char *)eptr;
642        return (len > 0)? -len : len;
643        }
644      }
645  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
646    
647    /* To save space on the stack and in the heap frame, I have doubled up on some
648    of the local variables that are used only in localised parts of the code, but
649    still need to be preserved over recursive calls of match(). These macros define
650    the alternative names that are used. */
651    
652    #define allow_zero    cur_is_word
653    #define cbegroup      condition
654    #define code_offset   codelink
655    #define condassert    condition
656    #define matched_once  prev_is_word
657    #define foc           number
658    #define save_mark     data
659    
660  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
661  variables. */  variables. */
662    
# Line 622  TAIL_RECURSE: Line 676  TAIL_RECURSE:
676  /* OK, now we can get on with the real code of the function. Recursive calls  /* OK, now we can get on with the real code of the function. Recursive calls
677  are specified by the macro RMATCH and RRETURN is used to return. When  are specified by the macro RMATCH and RRETURN is used to return. When
678  NO_RECURSE is *not* defined, these just turn into a recursive call to match()  NO_RECURSE is *not* defined, these just turn into a recursive call to match()
679  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
680  defined). However, RMATCH isn't like a function call because it's quite a  defined). However, RMATCH isn't like a function call because it's quite a
681  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,
682  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
683    
684  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
685  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
686  #else  #else
687  utf8 = FALSE;  utf = FALSE;
688  #endif  #endif
689    
690  /* 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 693  haven't exceeded the recursive call limi
693  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
694  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
695    
 original_ims = ims;    /* Save for resetting on ')' */  
   
696  /* 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
697  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
698  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
699  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
700  When match() is called in other circumstances, don't add to the chain. The  
701  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
702  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
703  match(). */  to break infinite loops that match no characters. When match() is called in
704    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
705    NOT be used with tail recursion, because the memory block that is used is on
706    the stack, so a new one may be required for each match(). */
707    
708  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
709    {    {
710    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
711    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
712    eptrb = &newptrb;    eptrb = &newptrb;
713      md->match_function_type = 0;
714    }    }
715    
716  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 666  for (;;) Line 722  for (;;)
722    
723    switch(op)    switch(op)
724      {      {
725        case OP_MARK:
726        md->nomatch_mark = ecode + 2;
727        md->mark = NULL;    /* In case previously set by assertion */
728        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
729          eptrb, RM55);
730        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
731             md->mark == NULL) md->mark = ecode + 2;
732    
733        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
734        argument, and we must check whether that argument matches this MARK's
735        argument. It is passed back in md->start_match_ptr (an overloading of that
736        variable). If it does match, we reset that variable to the current subject
737        position and return MATCH_SKIP. Otherwise, pass back the return code
738        unaltered. */
739    
740        else if (rrc == MATCH_SKIP_ARG &&
741            STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0)
742          {
743          md->start_match_ptr = eptr;
744          RRETURN(MATCH_SKIP);
745          }
746        RRETURN(rrc);
747    
748      case OP_FAIL:      case OP_FAIL:
749      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
750    
751      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);  
752    
753      case OP_COMMIT:      case OP_COMMIT:
754      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
755        ims, eptrb, flags, RM52);        eptrb, RM52);
756      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
757            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
758            rrc != MATCH_THEN)
759          RRETURN(rrc);
760      RRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
761    
762        /* PRUNE overrides THEN */
763    
764        case OP_PRUNE:
765        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
766          eptrb, RM51);
767        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
768        RRETURN(MATCH_PRUNE);
769    
770        case OP_PRUNE_ARG:
771        md->nomatch_mark = ecode + 2;
772        md->mark = NULL;    /* In case previously set by assertion */
773        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
774          eptrb, RM56);
775        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
776             md->mark == NULL) md->mark = ecode + 2;
777        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
778        RRETURN(MATCH_PRUNE);
779    
780        /* SKIP overrides PRUNE and THEN */
781    
782      case OP_SKIP:      case OP_SKIP:
783      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
784        ims, eptrb, flags, RM53);        eptrb, RM53);
785      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
786          RRETURN(rrc);
787      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
788      RRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
789    
790        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
791        nomatch_mark. There is a flag that disables this opcode when re-matching a
792        pattern that ended with a SKIP for which there was not a matching MARK. */
793    
794        case OP_SKIP_ARG:
795        if (md->ignore_skip_arg)
796          {
797          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
798          break;
799          }
800        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
801          eptrb, RM57);
802        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
803          RRETURN(rrc);
804    
805        /* Pass back the current skip name by overloading md->start_match_ptr and
806        returning the special MATCH_SKIP_ARG return code. This will either be
807        caught by a matching MARK, or get to the top, where it causes a rematch
808        with the md->ignore_skip_arg flag set. */
809    
810        md->start_match_ptr = ecode + 2;
811        RRETURN(MATCH_SKIP_ARG);
812    
813        /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
814        the branch in which it occurs can be determined. Overload the start of
815        match pointer to do this. */
816    
817      case OP_THEN:      case OP_THEN:
818      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
819        ims, eptrb, flags, RM54);        eptrb, RM54);
820      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
821        md->start_match_ptr = ecode;
822      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
823    
824      /* Handle a capturing bracket. If there is space in the offset vector, save      case OP_THEN_ARG:
825      the current subject position in the working slot at the top of the vector.      md->nomatch_mark = ecode + 2;
826      We mustn't change the current values of the data slot, because they may be      md->mark = NULL;    /* In case previously set by assertion */
827      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,
828      reference inside the group.        md, eptrb, RM58);
829        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
830      If the bracket fails to match, we need to restore this value and also the           md->mark == NULL) md->mark = ecode + 2;
831      values of the final offsets, in case they were set by a previous iteration      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
832      of the same bracket.      md->start_match_ptr = ecode;
833        RRETURN(MATCH_THEN);
834    
835        /* Handle an atomic group that does not contain any capturing parentheses.
836        This can be handled like an assertion. Prior to 8.13, all atomic groups
837        were handled this way. In 8.13, the code was changed as below for ONCE, so
838        that backups pass through the group and thereby reset captured values.
839        However, this uses a lot more stack, so in 8.20, atomic groups that do not
840        contain any captures generate OP_ONCE_NC, which can be handled in the old,
841        less stack intensive way.
842    
843        Check the alternative branches in turn - the matching won't pass the KET
844        for this kind of subpattern. If any one branch matches, we carry on as at
845        the end of a normal bracket, leaving the subject pointer, but resetting
846        the start-of-match value in case it was changed by \K. */
847    
848        case OP_ONCE_NC:
849        prev = ecode;
850        saved_eptr = eptr;
851        save_mark = md->mark;
852        do
853          {
854          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
855          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
856            {
857            mstart = md->start_match_ptr;
858            break;
859            }
860          if (rrc == MATCH_THEN)
861            {
862            next = ecode + GET(ecode,1);
863            if (md->start_match_ptr < next &&
864                (*ecode == OP_ALT || *next == OP_ALT))
865              rrc = MATCH_NOMATCH;
866            }
867    
868          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
869          ecode += GET(ecode,1);
870          md->mark = save_mark;
871          }
872        while (*ecode == OP_ALT);
873    
874        /* If hit the end of the group (which could be repeated), fail */
875    
876        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
877    
878        /* Continue as from after the group, updating the offsets high water
879        mark, since extracts may have been taken. */
880    
881        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
882    
883        offset_top = md->end_offset_top;
884        eptr = md->end_match_ptr;
885    
886        /* For a non-repeating ket, just continue at this level. This also
887        happens for a repeating ket if no characters were matched in the group.
888        This is the forcible breaking of infinite loops as implemented in Perl
889        5.005. */
890    
891        if (*ecode == OP_KET || eptr == saved_eptr)
892          {
893          ecode += 1+LINK_SIZE;
894          break;
895          }
896    
897        /* The repeating kets try the rest of the pattern or restart from the
898        preceding bracket, in the appropriate order. The second "call" of match()
899        uses tail recursion, to avoid using another stack frame. */
900    
901        if (*ecode == OP_KETRMIN)
902          {
903          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
904          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
905          ecode = prev;
906          goto TAIL_RECURSE;
907          }
908        else  /* OP_KETRMAX */
909          {
910          md->match_function_type = MATCH_CBEGROUP;
911          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
912          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
913          ecode += 1 + LINK_SIZE;
914          goto TAIL_RECURSE;
915          }
916        /* Control never gets here */
917    
918        /* Handle a capturing bracket, other than those that are possessive with an
919        unlimited repeat. If there is space in the offset vector, save the current
920        subject position in the working slot at the top of the vector. We mustn't
921        change the current values of the data slot, because they may be set from a
922        previous iteration of this group, and be referred to by a reference inside
923        the group. A failure to match might occur after the group has succeeded,
924        if something later on doesn't match. For this reason, we need to restore
925        the working value and also the values of the final offsets, in case they
926        were set by a previous iteration of the same bracket.
927    
928      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
929      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 713  for (;;) Line 934  for (;;)
934      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
935      offset = number << 1;      offset = number << 1;
936    
937  #ifdef DEBUG  #ifdef PCRE_DEBUG
938      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
939      printf("subject=");      printf("subject=");
940      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 726  for (;;) Line 947  for (;;)
947        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
948        save_offset3 = md->offset_vector[md->offset_end - number];        save_offset3 = md->offset_vector[md->offset_end - number];
949        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
950          save_mark = md->mark;
951    
952        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
953        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
954            (int)(eptr - md->start_subject);
955    
956        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
957          {          {
958          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
959            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
960          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);            eptrb, RM1);
961            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
962    
963            /* If we backed up to a THEN, check whether it is within the current
964            branch by comparing the address of the THEN that is passed back with
965            the end of the branch. If it is within the current branch, and the
966            branch is one of two or more alternatives (it either starts or ends
967            with OP_ALT), we have reached the limit of THEN's action, so convert
968            the return code to NOMATCH, which will cause normal backtracking to
969            happen from now on. Otherwise, THEN is passed back to an outer
970            alternative. This implements Perl's treatment of parenthesized groups,
971            where a group not containing | does not affect the current alternative,
972            that is, (X) is NOT the same as (X|(*F)). */
973    
974            if (rrc == MATCH_THEN)
975              {
976              next = ecode + GET(ecode,1);
977              if (md->start_match_ptr < next &&
978                  (*ecode == OP_ALT || *next == OP_ALT))
979                rrc = MATCH_NOMATCH;
980              }
981    
982            /* Anything other than NOMATCH is passed back. */
983    
984            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
985          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
986          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
987            md->mark = save_mark;
988            if (*ecode != OP_ALT) break;
989          }          }
       while (*ecode == OP_ALT);  
990    
991        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
992        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
993        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
994        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
995    
996        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
997    
998          RRETURN(rrc);
999        }        }
1000    
1001      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 761  for (;;) Line 1009  for (;;)
1009      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1010      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1011    
1012      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
1013      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.*/  
1014    
1015        When we get to the final alternative within the brackets, we used to return
1016        the result of a recursive call to match() whatever happened so it was
1017        possible to reduce stack usage by turning this into a tail recursion,
1018        except in the case of a possibly empty group. However, now that there is
1019        the possiblity of (*THEN) occurring in the final alternative, this
1020        optimization is no longer always possible.
1021    
1022        We can optimize if we know there are no (*THEN)s in the pattern; at present
1023        this is the best that can be done.
1024    
1025        MATCH_ONCE is returned when the end of an atomic group is successfully
1026        reached, but subsequent matching fails. It passes back up the tree (causing
1027        captured values to be reset) until the original atomic group level is
1028        reached. This is tested by comparing md->once_target with the start of the
1029        group. At this point, the return is converted into MATCH_NOMATCH so that
1030        previous backup points can be taken. */
1031    
1032        case OP_ONCE:
1033      case OP_BRA:      case OP_BRA:
1034      case OP_SBRA:      case OP_SBRA:
1035      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
1036      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
1037      for (;;)      for (;;)
1038        {        {
1039        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
1040    
1041          /* If this is not a possibly empty group, and there are no (*THEN)s in
1042          the pattern, and this is the final alternative, optimize as described
1043          above. */
1044    
1045          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1046            {
1047            ecode += PRIV(OP_lengths)[*ecode];
1048            goto TAIL_RECURSE;
1049            }
1050    
1051          /* In all other cases, we have to make another call to match(). */
1052    
1053          save_mark = md->mark;
1054          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1055            RM2);
1056    
1057          /* See comment in the code for capturing groups above about handling
1058          THEN. */
1059    
1060          if (rrc == MATCH_THEN)
1061          {          {
1062          if (flags == 0)    /* Not a possibly empty group */          next = ecode + GET(ecode,1);
1063            if (md->start_match_ptr < next &&
1064                (*ecode == OP_ALT || *next == OP_ALT))
1065              rrc = MATCH_NOMATCH;
1066            }
1067    
1068          if (rrc != MATCH_NOMATCH)
1069            {
1070            if (rrc == MATCH_ONCE)
1071            {            {
1072            ecode += _pcre_OP_lengths[*ecode];            const pcre_uchar *scode = ecode;
1073            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1074            goto TAIL_RECURSE;              {
1075                while (*scode == OP_ALT) scode += GET(scode, 1);
1076                scode -= GET(scode, 1);
1077                }
1078              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1079            }            }
1080            RRETURN(rrc);
1081            }
1082          ecode += GET(ecode, 1);
1083          md->mark = save_mark;
1084          if (*ecode != OP_ALT) break;
1085          }
1086    
1087          /* Possibly empty group; can't use tail recursion. */      RRETURN(MATCH_NOMATCH);
1088    
1089          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* Handle possessive capturing brackets with an unlimited repeat. We come
1090            eptrb, flags, RM48);      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1091          RRETURN(rrc);      handled similarly to the normal case above. However, the matching is
1092        different. The end of these brackets will always be OP_KETRPOS, which
1093        returns MATCH_KETRPOS without going further in the pattern. By this means
1094        we can handle the group by iteration rather than recursion, thereby
1095        reducing the amount of stack needed. */
1096    
1097        case OP_CBRAPOS:
1098        case OP_SCBRAPOS:
1099        allow_zero = FALSE;
1100    
1101        POSSESSIVE_CAPTURE:
1102        number = GET2(ecode, 1+LINK_SIZE);
1103        offset = number << 1;
1104    
1105    #ifdef PCRE_DEBUG
1106        printf("start possessive bracket %d\n", number);
1107        printf("subject=");
1108        pchars(eptr, 16, TRUE, md);
1109        printf("\n");
1110    #endif
1111    
1112        if (offset < md->offset_max)
1113          {
1114          matched_once = FALSE;
1115          code_offset = (int)(ecode - md->start_code);
1116    
1117          save_offset1 = md->offset_vector[offset];
1118          save_offset2 = md->offset_vector[offset+1];
1119          save_offset3 = md->offset_vector[md->offset_end - number];
1120          save_capture_last = md->capture_last;
1121    
1122          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1123    
1124          /* Each time round the loop, save the current subject position for use
1125          when the group matches. For MATCH_MATCH, the group has matched, so we
1126          restart it with a new subject starting position, remembering that we had
1127          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1128          usual. If we haven't matched any alternatives in any iteration, check to
1129          see if a previous iteration matched. If so, the group has matched;
1130          continue from afterwards. Otherwise it has failed; restore the previous
1131          capture values before returning NOMATCH. */
1132    
1133          for (;;)
1134            {
1135            md->offset_vector[md->offset_end - number] =
1136              (int)(eptr - md->start_subject);
1137            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1138            RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1139              eptrb, RM63);
1140            if (rrc == MATCH_KETRPOS)
1141              {
1142              offset_top = md->end_offset_top;
1143              eptr = md->end_match_ptr;
1144              ecode = md->start_code + code_offset;
1145              save_capture_last = md->capture_last;
1146              matched_once = TRUE;
1147              continue;
1148              }
1149    
1150            /* See comment in the code for capturing groups above about handling
1151            THEN. */
1152    
1153            if (rrc == MATCH_THEN)
1154              {
1155              next = ecode + GET(ecode,1);
1156              if (md->start_match_ptr < next &&
1157                  (*ecode == OP_ALT || *next == OP_ALT))
1158                rrc = MATCH_NOMATCH;
1159              }
1160    
1161            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1162            md->capture_last = save_capture_last;
1163            ecode += GET(ecode, 1);
1164            if (*ecode != OP_ALT) break;
1165          }          }
1166    
1167        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1168        otherwise return. */          {
1169            md->offset_vector[offset] = save_offset1;
1170            md->offset_vector[offset+1] = save_offset2;
1171            md->offset_vector[md->offset_end - number] = save_offset3;
1172            }
1173    
1174        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (allow_zero || matched_once)
1175          eptrb, flags, RM2);          {
1176        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          ecode += 1 + LINK_SIZE;
1177            break;
1178            }
1179    
1180          RRETURN(MATCH_NOMATCH);
1181          }
1182    
1183        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1184        as a non-capturing bracket. */
1185    
1186        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1187        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1188    
1189        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1190    
1191        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1192        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1193    
1194        /* Non-capturing possessive bracket with unlimited repeat. We come here
1195        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1196        without the capturing complication. It is written out separately for speed
1197        and cleanliness. */
1198    
1199        case OP_BRAPOS:
1200        case OP_SBRAPOS:
1201        allow_zero = FALSE;
1202    
1203        POSSESSIVE_NON_CAPTURE:
1204        matched_once = FALSE;
1205        code_offset = (int)(ecode - md->start_code);
1206    
1207        for (;;)
1208          {
1209          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1210          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1211            eptrb, RM48);
1212          if (rrc == MATCH_KETRPOS)
1213            {
1214            offset_top = md->end_offset_top;
1215            eptr = md->end_match_ptr;
1216            ecode = md->start_code + code_offset;
1217            matched_once = TRUE;
1218            continue;
1219            }
1220    
1221          /* See comment in the code for capturing groups above about handling
1222          THEN. */
1223    
1224          if (rrc == MATCH_THEN)
1225            {
1226            next = ecode + GET(ecode,1);
1227            if (md->start_match_ptr < next &&
1228                (*ecode == OP_ALT || *next == OP_ALT))
1229              rrc = MATCH_NOMATCH;
1230            }
1231    
1232          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1233        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1234          if (*ecode != OP_ALT) break;
1235          }
1236    
1237        if (matched_once || allow_zero)
1238          {
1239          ecode += 1 + LINK_SIZE;
1240          break;
1241        }        }
1242        RRETURN(MATCH_NOMATCH);
1243    
1244      /* Control never reaches here. */      /* Control never reaches here. */
1245    
1246      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1247      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
1248      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
1249      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. */  
1250    
1251      case OP_COND:      case OP_COND:
1252      case OP_SCOND:      case OP_SCOND:
1253      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1254    
1255      /* 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
1256      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
1257    
1258      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1259        {        {
1260        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1261          {          {
1262          pcre_callout_block cb;          PUBL(callout_block) cb;
1263          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1264          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1265          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1266    #ifdef COMPILE_PCRE8
1267          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1268          cb.subject_length   = md->end_subject - md->start_subject;  #else
1269          cb.start_match      = mstart - md->start_subject;          cb.subject          = (PCRE_SPTR16)md->start_subject;
1270          cb.current_position = eptr - md->start_subject;  #endif
1271            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1272            cb.start_match      = (int)(mstart - md->start_subject);
1273            cb.current_position = (int)(eptr - md->start_subject);
1274          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1275          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1276          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1277          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1278          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1279          if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1280            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1281          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1282          }          }
1283        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1284        }        }
1285    
1286      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
1287    
1288      /* Now see what the actual condition is */      /* Now see what the actual condition is */
1289    
1290      if (condcode == OP_RREF)         /* Recursion test */      if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1291        {        {
1292        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (md->recursive == NULL)                /* Not recursing => FALSE */
1293        condition = md->recursive != NULL &&          {
1294          (offset == RREF_ANY || offset == md->recursive->group_num);          condition = FALSE;
1295        ecode += condition? 3 : GET(ecode, 1);          ecode += GET(ecode, 1);
1296            }
1297          else
1298            {
1299            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1300            condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1301    
1302            /* If the test is for recursion into a specific subpattern, and it is
1303            false, but the test was set up by name, scan the table to see if the
1304            name refers to any other numbers, and test them. The condition is true
1305            if any one is set. */
1306    
1307            if (!condition && condcode == OP_NRREF)
1308              {
1309              pcre_uchar *slotA = md->name_table;
1310              for (i = 0; i < md->name_count; i++)
1311                {
1312                if (GET2(slotA, 0) == recno) break;
1313                slotA += md->name_entry_size;
1314                }
1315    
1316              /* Found a name for the number - there can be only one; duplicate
1317              names for different numbers are allowed, but not vice versa. First
1318              scan down for duplicates. */
1319    
1320              if (i < md->name_count)
1321                {
1322                pcre_uchar *slotB = slotA;
1323                while (slotB > md->name_table)
1324                  {
1325                  slotB -= md->name_entry_size;
1326                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1327                    {
1328                    condition = GET2(slotB, 0) == md->recursive->group_num;
1329                    if (condition) break;
1330                    }
1331                  else break;
1332                  }
1333    
1334                /* Scan up for duplicates */
1335    
1336                if (!condition)
1337                  {
1338                  slotB = slotA;
1339                  for (i++; i < md->name_count; i++)
1340                    {
1341                    slotB += md->name_entry_size;
1342                    if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1343                      {
1344                      condition = GET2(slotB, 0) == md->recursive->group_num;
1345                      if (condition) break;
1346                      }
1347                    else break;
1348                    }
1349                  }
1350                }
1351              }
1352    
1353            /* Chose branch according to the condition */
1354    
1355            ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1356            }
1357        }        }
1358    
1359      else if (condcode == OP_CREF)    /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1360        {        {
1361        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1362        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1363        ecode += condition? 3 : GET(ecode, 1);  
1364          /* If the numbered capture is unset, but the reference was by name,
1365          scan the table to see if the name refers to any other numbers, and test
1366          them. The condition is true if any one is set. This is tediously similar
1367          to the code above, but not close enough to try to amalgamate. */
1368    
1369          if (!condition && condcode == OP_NCREF)
1370            {
1371            int refno = offset >> 1;
1372            pcre_uchar *slotA = md->name_table;
1373    
1374            for (i = 0; i < md->name_count; i++)
1375              {
1376              if (GET2(slotA, 0) == refno) break;
1377              slotA += md->name_entry_size;
1378              }
1379    
1380            /* Found a name for the number - there can be only one; duplicate names
1381            for different numbers are allowed, but not vice versa. First scan down
1382            for duplicates. */
1383    
1384            if (i < md->name_count)
1385              {
1386              pcre_uchar *slotB = slotA;
1387              while (slotB > md->name_table)
1388                {
1389                slotB -= md->name_entry_size;
1390                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1391                  {
1392                  offset = GET2(slotB, 0) << 1;
1393                  condition = offset < offset_top &&
1394                    md->offset_vector[offset] >= 0;
1395                  if (condition) break;
1396                  }
1397                else break;
1398                }
1399    
1400              /* Scan up for duplicates */
1401    
1402              if (!condition)
1403                {
1404                slotB = slotA;
1405                for (i++; i < md->name_count; i++)
1406                  {
1407                  slotB += md->name_entry_size;
1408                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1409                    {
1410                    offset = GET2(slotB, 0) << 1;
1411                    condition = offset < offset_top &&
1412                      md->offset_vector[offset] >= 0;
1413                    if (condition) break;
1414                    }
1415                  else break;
1416                  }
1417                }
1418              }
1419            }
1420    
1421          /* Chose branch according to the condition */
1422    
1423          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1424        }        }
1425    
1426      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 861  for (;;) Line 1430  for (;;)
1430        }        }
1431    
1432      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1433      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
1434      assertion. */      an assertion. */
1435    
1436      else      else
1437        {        {
1438        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1439            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1440        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1441          {          {
1442            if (md->end_offset_top > offset_top)
1443              offset_top = md->end_offset_top;  /* Captures may have happened */
1444          condition = TRUE;          condition = TRUE;
1445          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1446          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1447          }          }
1448    
1449          /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1450          assertion; it is therefore treated as NOMATCH. */
1451    
1452        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1453          {          {
1454          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
# Line 885  for (;;) Line 1460  for (;;)
1460          }          }
1461        }        }
1462    
1463      /* 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
1464      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
1465      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
1466      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
1467        exist, in which case we can just plough on. Note that, for compatibility
1468        with Perl, the | in a conditional group is NOT treated as creating two
1469        alternatives. If a THEN is encountered in the branch, it propagates out to
1470        the enclosing alternative (unless nested in a deeper set of alternatives,
1471        of course). */
1472    
1473      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1474        {        {
1475        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
1476          {          {
1477          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          ecode += 1 + LINK_SIZE;
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
1478          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1479          }          }
1480    
1481          md->match_function_type = MATCH_CBEGROUP;
1482          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1483          RRETURN(rrc);
1484        }        }
1485      else                         /* Condition false & no alternative */  
1486         /* Condition false & no alternative; continue after the group. */
1487    
1488        else
1489        {        {
1490        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1491        }        }
1492      break;      break;
1493    
1494    
1495      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1496      to close any currently open capturing brackets. */      to close any currently open capturing brackets. */
1497    
1498      case OP_CLOSE:      case OP_CLOSE:
1499      number = GET2(ecode, 1);      number = GET2(ecode, 1);
1500      offset = number << 1;      offset = number << 1;
1501    
1502  #ifdef DEBUG  #ifdef PCRE_DEBUG
1503        printf("end bracket %d at *ACCEPT", number);        printf("end bracket %d at *ACCEPT", number);
1504        printf("\n");        printf("\n");
1505  #endif  #endif
# Line 928  for (;;) Line 1509  for (;;)
1509        {        {
1510        md->offset_vector[offset] =        md->offset_vector[offset] =
1511          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1512        md->offset_vector[offset+1] = eptr - md->start_subject;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1513        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1514        }        }
1515      ecode += 3;      ecode += 1 + IMM2_SIZE;
1516      break;      break;
1517    
1518    
1519      /* 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. */  
1520    
     case OP_ACCEPT:  
1521      case OP_END:      case OP_END:
1522      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1523        {      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->offset_top;  
       mstart = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1524    
1525      /* 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
1526      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
1527      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,
1528      if any. */      backtracking will then try other alternatives, if any. */
1529    
1530      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1531          (md->notempty ||           md->recursive == NULL &&
1532            (md->notempty_atstart &&           (md->notempty ||
1533              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1534                 mstart == md->start_subject + md->start_offset)))
1535        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1536    
1537      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 971  for (;;) Line 1539  for (;;)
1539      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1540      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1541      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);  
1542    
1543      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1544        given as the argument to RRETURN when the heap is in use. */
1545    
1546      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1547      ims = ecode[1];      RRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1548    
1549      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1550      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,
1551      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
1552      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
1553      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1554        of a condition, we want to return immediately afterwards. The caller of
1555        this incarnation of the match() function will have set MATCH_CONDASSERT in
1556        md->match_function type, and one of these opcodes will be the first opcode
1557        that is processed. We use a local variable that is preserved over calls to
1558        match() to remember this case. */
1559    
1560      case OP_ASSERT:      case OP_ASSERT:
1561      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1562        save_mark = md->mark;
1563        if (md->match_function_type == MATCH_CONDASSERT)
1564          {
1565          condassert = TRUE;
1566          md->match_function_type = 0;
1567          }
1568        else condassert = FALSE;
1569    
1570      do      do
1571        {        {
1572        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1573          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1574        if (rrc == MATCH_MATCH) break;          {
1575            mstart = md->start_match_ptr;   /* In case \K reset it */
1576            break;
1577            }
1578    
1579          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1580          as NOMATCH. */
1581    
1582        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1583        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1584          md->mark = save_mark;
1585        }        }
1586      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1587    
1588      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1589    
1590      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1591    
1592      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1593    
1594      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1595      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1012  for (;;) Line 1599  for (;;)
1599      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1600      continue;      continue;
1601    
1602      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1603        PRUNE, or COMMIT means we must assume failure without checking subsequent
1604        branches. */
1605    
1606      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1607      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1608        save_mark = md->mark;
1609        if (md->match_function_type == MATCH_CONDASSERT)
1610          {
1611          condassert = TRUE;
1612          md->match_function_type = 0;
1613          }
1614        else condassert = FALSE;
1615    
1616      do      do
1617        {        {
1618        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1619          RM5);        md->mark = save_mark;
1620        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1621          if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1622            {
1623            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1624            break;
1625            }
1626    
1627          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1628          as NOMATCH. */
1629    
1630        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1631        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1632        }        }
1633      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1634    
1635      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1636    
1637      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1638      continue;      continue;
# Line 1037  for (;;) Line 1643  for (;;)
1643      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1644    
1645      case OP_REVERSE:      case OP_REVERSE:
1646  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1647      if (utf8)      if (utf)
1648        {        {
1649        i = GET(ecode, 1);        i = GET(ecode, 1);
1650        while (i-- > 0)        while (i-- > 0)
# Line 1069  for (;;) Line 1675  for (;;)
1675      function is able to force a failure. */      function is able to force a failure. */
1676    
1677      case OP_CALLOUT:      case OP_CALLOUT:
1678      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1679        {        {
1680        pcre_callout_block cb;        PUBL(callout_block) cb;
1681        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1682        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1683        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1684    #ifdef COMPILE_PCRE8
1685        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1686        cb.subject_length   = md->end_subject - md->start_subject;  #else
1687        cb.start_match      = mstart - md->start_subject;        cb.subject          = (PCRE_SPTR16)md->start_subject;
1688        cb.current_position = eptr - md->start_subject;  #endif
1689          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1690          cb.start_match      = (int)(mstart - md->start_subject);
1691          cb.current_position = (int)(eptr - md->start_subject);
1692        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1693        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1694        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1695        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1696        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1697        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1698          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1699        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1700        }        }
1701      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1094  for (;;) Line 1705  for (;;)
1705      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
1706      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1707    
1708      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1709      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
1710      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
1711      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
1712      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
1713      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
1714      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.  
1715    
1716      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
1717      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
1718      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1719        a lot, so he is not to blame for the current way it works. */
1720    
1721      case OP_RECURSE:      case OP_RECURSE:
1722        {        {
1723          recursion_info *ri;
1724          int recno;
1725    
1726        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1727        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1728          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1729    
1730          /* Check for repeating a recursion without advancing the subject pointer.
1731          This should catch convoluted mutual recursions. (Some simple cases are
1732          caught at compile time.) */
1733    
1734          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1735            if (recno == ri->group_num && eptr == ri->subject_position)
1736              RRETURN(PCRE_ERROR_RECURSELOOP);
1737    
1738        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1739    
1740          new_recursive.group_num = recno;
1741          new_recursive.subject_position = eptr;
1742        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1743        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1744    
1745        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1746    
1747        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1748    
1749        /* Now save the offset data. */        /* Now save the offset data */
1750    
1751        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1752        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1133  for (;;) Line 1754  for (;;)
1754        else        else
1755          {          {
1756          new_recursive.offset_save =          new_recursive.offset_save =
1757            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1758          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1759          }          }
   
1760        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1761              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = mstart;  
       new_recursive.offset_top = offset_top;  
       mstart = eptr;  
1762    
1763        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1764        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1765          might be changed, so reset it before looping. */
1766    
1767        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1768        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1769        do        do
1770          {          {
1771          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1772            md, ims, eptrb, flags, RM6);          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1773          if (rrc == MATCH_MATCH)            md, eptrb, RM6);
1774            memcpy(md->offset_vector, new_recursive.offset_save,
1775                new_recursive.saved_max * sizeof(int));
1776            md->recursive = new_recursive.prevrec;
1777            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1778            {            {
1779            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1780            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1781              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1782            RRETURN(MATCH_MATCH);  
1783              /* Set where we got to in the subject, and reset the start in case
1784              it was changed by \K. This *is* propagated back out of a recursion,
1785              for Perl compatibility. */
1786    
1787              eptr = md->end_match_ptr;
1788              mstart = md->start_match_ptr;
1789              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1790            }            }
1791    
1792            /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1793            as NOMATCH. */
1794    
1795          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1796            {            {
1797            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1798            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1799              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1800            RRETURN(rrc);            RRETURN(rrc);
1801            }            }
1802    
1803          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1804          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1805          }          }
1806        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1178  for (;;) Line 1808  for (;;)
1808        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1809        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1810        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1811          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1812        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1813        }        }
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     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. */  
   
     case OP_ONCE:  
     prev = ecode;  
     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));  
       }  
1814    
1815      if (*ecode == OP_KETRMIN)      RECURSION_MATCHED:
1816        {      break;
       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 */  
1817    
1818      /* An alternation is the end of a branch; scan along to find the end of the      /* An alternation is the end of a branch; scan along to find the end of the
1819      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1271  for (;;) Line 1829  for (;;)
1829      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1830    
1831      case OP_BRAZERO:      case OP_BRAZERO:
1832        {      next = ecode + 1;
1833        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1834        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1835        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1836        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1837      break;      break;
1838    
1839      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1840        {      next = ecode + 1;
1841        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1842        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1843        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1844        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1845      break;      break;
1846    
1847      case OP_SKIPZERO:      case OP_SKIPZERO:
1848        {      next = ecode+1;
1849        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1850        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1851      break;      break;
1852    
1853        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1854        here; just jump to the group, with allow_zero set TRUE. */
1855    
1856        case OP_BRAPOSZERO:
1857        op = *(++ecode);
1858        allow_zero = TRUE;
1859        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1860          goto POSSESSIVE_NON_CAPTURE;
1861    
1862      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1863    
1864      case OP_KET:      case OP_KET:
1865      case OP_KETRMIN:      case OP_KETRMIN:
1866      case OP_KETRMAX:      case OP_KETRMAX:
1867        case OP_KETRPOS:
1868      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1869    
1870      /* 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
1871      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1872      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1873    
1874      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1875        {        {
1876        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1877        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1878        }        }
1879      else saved_eptr = NULL;      else saved_eptr = NULL;
1880    
1881      /* 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
1882      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
1883      assertions. Do this also for the "once" (atomic) groups. */      water mark for use by positive assertions. We also need to record the match
1884        start in case it was changed by \K. */
1885      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||  
1886          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1887          *prev == OP_ONCE)           *prev == OP_ONCE_NC)
1888        {        {
1889        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1890        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1891        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1892          RRETURN(MATCH_MATCH);         /* Sets md->mark */
1893        }        }
1894    
1895      /* 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
1896      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1897      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
1898      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
1899      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
1900        the current subject position and start match pointer and give a MATCH
1901        return. */
1902    
1903      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1904            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1905        {        {
1906        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1907        offset = number << 1;        offset = number << 1;
1908    
1909  #ifdef DEBUG  #ifdef PCRE_DEBUG
1910        printf("end bracket %d", number);        printf("end bracket %d", number);
1911        printf("\n");        printf("\n");
1912  #endif  #endif
1913    
1914          /* Handle a recursively called group. */
1915    
1916          if (md->recursive != NULL && md->recursive->group_num == number)
1917            {
1918            md->end_match_ptr = eptr;
1919            md->start_match_ptr = mstart;
1920            RRETURN(MATCH_MATCH);
1921            }
1922    
1923          /* Deal with capturing */
1924    
1925        md->capture_last = number;        md->capture_last = number;
1926        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1927          {          {
1928            /* If offset is greater than offset_top, it means that we are
1929            "skipping" a capturing group, and that group's offsets must be marked
1930            unset. In earlier versions of PCRE, all the offsets were unset at the
1931            start of matching, but this doesn't work because atomic groups and
1932            assertions can cause a value to be set that should later be unset.
1933            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1934            part of the atomic group, but this is not on the final matching path,
1935            so must be unset when 2 is set. (If there is no group 2, there is no
1936            problem, because offset_top will then be 2, indicating no capture.) */
1937    
1938            if (offset > offset_top)
1939              {
1940              register int *iptr = md->offset_vector + offset_top;
1941              register int *iend = md->offset_vector + offset;
1942              while (iptr < iend) *iptr++ = -1;
1943              }
1944    
1945            /* Now make the extraction */
1946    
1947          md->offset_vector[offset] =          md->offset_vector[offset] =
1948            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1949          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1950          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1951          }          }
1952          }
1953    
1954        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1955        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1956        group. This is the forcible breaking of infinite loops as implemented in
1957        Perl 5.005. For a non-repeating atomic group that includes captures,
1958        establish a backup point by processing the rest of the pattern at a lower
1959        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1960        original OP_ONCE level, thereby bypassing intermediate backup points, but
1961        resetting any captures that happened along the way. */
1962    
1963        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1964          {
1965          if (*prev == OP_ONCE)
1966          {          {
1967          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1968          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1969          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1970          mstart = rec->save_start;          RRETURN(MATCH_ONCE);
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         offset_top = rec->offset_top;  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1971          }          }
1972          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1973          break;
1974        }        }
1975    
1976      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1977      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1978        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
1979    
1980      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRPOS)
1981        {        {
1982        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1983        break;        md->end_offset_top = offset_top;
1984          RRETURN(MATCH_KETRPOS);
1985        }        }
1986    
1987      /* 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
1988      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
1989      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
1990      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
1991        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1992    
1993      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1994        {        {
1995        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1996        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1997        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1998            {
1999            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
2000            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2001            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2002            RRETURN(MATCH_ONCE);
2003            }
2004          if (*prev >= OP_SBRA)    /* Could match an empty string */
2005          {          {
2006          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
2007            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
2008          RRETURN(rrc);          RRETURN(rrc);
2009          }          }
2010        ecode = prev;        ecode = prev;
# Line 1411  for (;;) Line 2012  for (;;)
2012        }        }
2013      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
2014        {        {
2015        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
2016          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
2017          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
2018        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2019          if (*prev == OP_ONCE)
2020            {
2021            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
2022            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2023            md->once_target = prev;
2024            RRETURN(MATCH_ONCE);
2025            }
2026        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
2027        goto TAIL_RECURSE;        goto TAIL_RECURSE;
2028        }        }
2029      /* Control never gets here */      /* Control never gets here */
2030    
2031      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
2032    
2033      case OP_CIRC:      case OP_CIRC:
2034      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 */  
2035    
2036      /* Start of subject assertion */      /* Start of subject assertion */
2037    
# Line 1440  for (;;) Line 2040  for (;;)
2040      ecode++;      ecode++;
2041      break;      break;
2042    
2043        /* Multiline mode: start of subject unless notbol, or after any newline. */
2044    
2045        case OP_CIRCM:
2046        if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2047        if (eptr != md->start_subject &&
2048            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
2049          RRETURN(MATCH_NOMATCH);
2050        ecode++;
2051        break;
2052    
2053      /* Start of match assertion */      /* Start of match assertion */
2054    
2055      case OP_SOM:      case OP_SOM:
# Line 1454  for (;;) Line 2064  for (;;)
2064      ecode++;      ecode++;
2065      break;      break;
2066    
2067      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2068      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2069    
2070      case OP_DOLL:      case OP_DOLLM:
2071      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2072        {        {
2073        if (eptr < md->end_subject)        if (!IS_NEWLINE(eptr))
2074          { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }          {
2075        else          if (md->partial != 0 &&
2076          { if (md->noteol) RRETURN(MATCH_NOMATCH); }              eptr + 1 >= md->end_subject &&
2077        ecode++;              NLBLOCK->nltype == NLTYPE_FIXED &&
2078        break;              NLBLOCK->nllen == 2 &&
2079                *eptr == NLBLOCK->nl[0])
2080              {
2081              md->hitend = TRUE;
2082              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2083              }
2084            RRETURN(MATCH_NOMATCH);
2085            }
2086        }        }
2087      else      else
2088        {        {
2089        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2090        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
2091        }        }
2092        ecode++;
2093        break;
2094    
2095        /* Not multiline mode: assert before a terminating newline or before end of
2096        subject unless noteol is set. */
2097    
2098        case OP_DOLL:
2099        if (md->noteol) RRETURN(MATCH_NOMATCH);
2100        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2101    
2102      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2103    
2104      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2105    
2106      case OP_EOD:      case OP_EOD:
2107      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2108        SCHECK_PARTIAL();
2109      ecode++;      ecode++;
2110      break;      break;
2111    
2112      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
2113    
2114      case OP_EODN:      case OP_EODN:
2115      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
2116        if (eptr < md->end_subject &&
2117          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2118          {
2119          if (md->partial != 0 &&
2120              eptr + 1 >= md->end_subject &&
2121              NLBLOCK->nltype == NLTYPE_FIXED &&
2122              NLBLOCK->nllen == 2 &&
2123              *eptr == NLBLOCK->nl[0])
2124            {
2125            md->hitend = TRUE;
2126            if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2127            }
2128        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2129          }
2130    
2131        /* Either at end of string or \n before end. */
2132    
2133        SCHECK_PARTIAL();
2134      ecode++;      ecode++;
2135      break;      break;
2136    
# Line 1508  for (;;) Line 2145  for (;;)
2145        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2146        partial matching. */        partial matching. */
2147    
2148  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2149        if (utf8)        if (utf)
2150          {          {
2151            /* Get status of previous character */
2152    
2153          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2154            {            {
2155            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2156            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2157            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2158            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2159    #ifdef SUPPORT_UCP
2160              if (md->use_ucp)
2161                {
2162                if (c == '_') prev_is_word = TRUE; else
2163                  {
2164                  int cat = UCD_CATEGORY(c);
2165                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2166                  }
2167                }
2168              else
2169    #endif
2170            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2171            }            }
2172    
2173            /* Get status of next character */
2174    
2175          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2176            {            {
2177            SCHECK_PARTIAL();            SCHECK_PARTIAL();
# Line 1527  for (;;) Line 2180  for (;;)
2180          else          else
2181            {            {
2182            GETCHAR(c, eptr);            GETCHAR(c, eptr);
2183    #ifdef SUPPORT_UCP
2184              if (md->use_ucp)
2185                {
2186                if (c == '_') cur_is_word = TRUE; else
2187                  {
2188                  int cat = UCD_CATEGORY(c);
2189                  cur_is_word = (cat == ucp_L || cat == ucp_N);
2190                  }
2191                }
2192              else
2193    #endif
2194            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2195            }            }
2196          }          }
2197        else        else
2198  #endif  #endif
2199    
2200        /* Not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
2201          consistency with the behaviour of \w we do use it in this case. */
2202    
2203          {          {
2204            /* Get status of previous character */
2205    
2206          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2207            {            {
2208            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
2209            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);  #ifdef SUPPORT_UCP
2210              if (md->use_ucp)
2211                {
2212                c = eptr[-1];
2213                if (c == '_') prev_is_word = TRUE; else
2214                  {
2215                  int cat = UCD_CATEGORY(c);
2216                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2217                  }
2218                }
2219              else
2220    #endif
2221              prev_is_word = MAX_255(eptr[-1])
2222                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2223            }            }
2224    
2225            /* Get status of next character */
2226    
2227          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2228            {            {
2229            SCHECK_PARTIAL();            SCHECK_PARTIAL();
2230            cur_is_word = FALSE;            cur_is_word = FALSE;
2231            }            }
2232          else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          else
2233    #ifdef SUPPORT_UCP
2234            if (md->use_ucp)
2235              {
2236              c = *eptr;
2237              if (c == '_') cur_is_word = TRUE; else
2238                {
2239                int cat = UCD_CATEGORY(c);
2240                cur_is_word = (cat == ucp_L || cat == ucp_N);
2241                }
2242              }
2243            else
2244    #endif
2245            cur_is_word = MAX_255(*eptr)
2246              && ((md->ctypes[*eptr] & ctype_word) != 0);
2247          }          }
2248    
2249        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
# Line 1557  for (;;) Line 2254  for (;;)
2254        }        }
2255      break;      break;
2256    
2257      /* Match a single character type; inline for speed */      /* Match any single character type except newline; have to take care with
2258        CRLF newlines and partial matching. */
2259    
2260      case OP_ANY:      case OP_ANY:
2261      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
2262        if (md->partial != 0 &&
2263            eptr + 1 >= md->end_subject &&
2264            NLBLOCK->nltype == NLTYPE_FIXED &&
2265            NLBLOCK->nllen == 2 &&
2266            *eptr == NLBLOCK->nl[0])
2267          {
2268          md->hitend = TRUE;
2269          if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2270          }
2271    
2272      /* Fall through */      /* Fall through */
2273    
2274        /* Match any single character whatsoever. */
2275    
2276      case OP_ALLANY:      case OP_ALLANY:
2277      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2278        {        {                            /* not be updated before SCHECK_PARTIAL. */
2279        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2280        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2281        }        }
2282      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2283    #ifdef SUPPORT_UTF
2284        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2285    #endif
2286      ecode++;      ecode++;
2287      break;      break;
2288    
# Line 1577  for (;;) Line 2290  for (;;)
2290      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2291    
2292      case OP_ANYBYTE:      case OP_ANYBYTE:
2293      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2294        {        {                            /* not be updated before SCHECK_PARTIAL. */
2295        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2296        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2297        }        }
2298        eptr++;
2299      ecode++;      ecode++;
2300      break;      break;
2301    
# Line 1593  for (;;) Line 2307  for (;;)
2307        }        }
2308      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2309      if (      if (
2310  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2311         c < 256 &&         c < 256 &&
2312  #endif  #endif
2313         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
# Line 1610  for (;;) Line 2324  for (;;)
2324        }        }
2325      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2326      if (      if (
2327  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2328         c >= 256 ||         c > 255 ||
2329  #endif  #endif
2330         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2331         )         )
# Line 1627  for (;;) Line 2341  for (;;)
2341        }        }
2342      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2343      if (      if (
2344  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2345         c < 256 &&         c < 256 &&
2346  #endif  #endif
2347         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
# Line 1644  for (;;) Line 2358  for (;;)
2358        }        }
2359      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2360      if (      if (
2361  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2362         c >= 256 ||         c > 255 ||
2363  #endif  #endif
2364         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2365         )         )
# Line 1661  for (;;) Line 2375  for (;;)
2375        }        }
2376      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2377      if (      if (
2378  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2379         c < 256 &&         c < 256 &&
2380  #endif  #endif
2381         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
# Line 1678  for (;;) Line 2392  for (;;)
2392        }        }
2393      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2394      if (      if (
2395  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2396         c >= 256 ||         c > 255 ||
2397  #endif  #endif
2398         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2399         )         )
# Line 1697  for (;;) Line 2411  for (;;)
2411      switch(c)      switch(c)
2412        {        {
2413        default: RRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2414    
2415        case 0x000d:        case 0x000d:
2416        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr >= md->end_subject)
2417            {
2418            SCHECK_PARTIAL();
2419            }
2420          else if (*eptr == 0x0a) eptr++;
2421        break;        break;
2422    
2423        case 0x000a:        case 0x000a:
# Line 1853  for (;;) Line 2572  for (;;)
2572               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2573               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2574            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2575           break;          break;
2576    
2577          case PT_GC:          case PT_GC:
2578          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2579            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2580          break;          break;
2581    
# Line 1870  for (;;) Line 2589  for (;;)
2589            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2590          break;          break;
2591    
2592            /* These are specials */
2593    
2594            case PT_ALNUM:
2595            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2596                 PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2597              RRETURN(MATCH_NOMATCH);
2598            break;
2599    
2600            case PT_SPACE:    /* Perl space */
2601            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2602                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2603                   == (op == OP_NOTPROP))
2604              RRETURN(MATCH_NOMATCH);
2605            break;
2606    
2607            case PT_PXSPACE:  /* POSIX space */
2608            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2609                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2610                 c == CHAR_FF || c == CHAR_CR)
2611                   == (op == OP_NOTPROP))
2612              RRETURN(MATCH_NOMATCH);
2613            break;
2614    
2615            case PT_WORD:
2616            if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2617                 PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2618                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2619              RRETURN(MATCH_NOMATCH);
2620            break;
2621    
2622            /* This should never occur */
2623    
2624          default:          default:
2625          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2626          }          }
# Line 1888  for (;;) Line 2639  for (;;)
2639        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2640        }        }
2641      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2642        if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH);
2643        while (eptr < md->end_subject)
2644        {        {
2645        int category = UCD_CATEGORY(c);        int len = 1;
2646        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2647        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2648          {        eptr += len;
         int len = 1;  
         if (!utf8) c = *eptr; else  
           {  
           GETCHARLEN(c, eptr, len);  
           }  
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
         eptr += len;  
         }  
2649        }        }
2650        CHECK_PARTIAL();
2651      ecode++;      ecode++;
2652      break;      break;
2653  #endif  #endif
# Line 1917  for (;;) Line 2662  for (;;)
2662      loops). */      loops). */
2663    
2664      case OP_REF:      case OP_REF:
2665        {      case OP_REFI:
2666        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2667        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2668        ecode += 1 + IMM2_SIZE;
2669    
2670        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2671    
2672        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2673        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
2674        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.  
2675    
2676        (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
2677        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2678    
2679        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
2680        referenced subpattern. */      referenced subpattern. */
2681    
2682        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2683          length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;        length = (md->jscript_compat)? 0 : -1;
2684        else      else
2685          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2686    
2687        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2688    
2689        switch (*ecode)      switch (*ecode)
2690          {
2691          case OP_CRSTAR:
2692          case OP_CRMINSTAR:
2693          case OP_CRPLUS:
2694          case OP_CRMINPLUS:
2695          case OP_CRQUERY:
2696          case OP_CRMINQUERY:
2697          c = *ecode++ - OP_CRSTAR;
2698          minimize = (c & 1) != 0;
2699          min = rep_min[c];                 /* Pick up values from tables; */
2700          max = rep_max[c];                 /* zero for max => infinity */
2701          if (max == 0) max = INT_MAX;
2702          break;
2703    
2704          case OP_CRRANGE:
2705          case OP_CRMINRANGE:
2706          minimize = (*ecode == OP_CRMINRANGE);
2707          min = GET2(ecode, 1);
2708          max = GET2(ecode, 1 + IMM2_SIZE);
2709          if (max == 0) max = INT_MAX;
2710          ecode += 1 + 2 * IMM2_SIZE;
2711          break;
2712    
2713          default:               /* No repeat follows */
2714          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2715          {          {
2716          case OP_CRSTAR:          if (length == -2) eptr = md->end_subject;   /* Partial match */
2717          case OP_CRMINSTAR:          CHECK_PARTIAL();
2718          case OP_CRPLUS:          RRETURN(MATCH_NOMATCH);
2719          case OP_CRMINPLUS:          }
2720          case OP_CRQUERY:        eptr += length;
2721          case OP_CRMINQUERY:        continue;              /* With the main loop */
2722          c = *ecode++ - OP_CRSTAR;        }
         minimize = (c & 1) != 0;  
         min = rep_min[c];                 /* Pick up values from tables; */  
         max = rep_max[c];                 /* zero for max => infinity */  
         if (max == 0) max = INT_MAX;  
         break;  
2723    
2724          case OP_CRRANGE:      /* Handle repeated back references. If the length of the reference is
2725          case OP_CRMINRANGE:      zero, just continue with the main loop. If the length is negative, it
2726          minimize = (*ecode == OP_CRMINRANGE);      means the reference is unset in non-Java-compatible mode. If the minimum is
2727          min = GET2(ecode, 1);      zero, we can continue at the same level without recursion. For any other
2728          max = GET2(ecode, 3);      minimum, carrying on will result in NOMATCH. */
         if (max == 0) max = INT_MAX;  
         ecode += 5;  
         break;  
2729    
2730          default:               /* No repeat follows */      if (length == 0) continue;
2731          if (!match_ref(offset, eptr, length, md, ims))      if (length < 0 && min == 0) continue;
2732            {  
2733            CHECK_PARTIAL();      /* First, ensure the minimum number of matches are present. We get back
2734            RRETURN(MATCH_NOMATCH);      the length of the reference string explicitly rather than passing the
2735            }      address of eptr, so that eptr can be a register variable. */
2736          eptr += length;  
2737          continue;              /* With the main loop */      for (i = 1; i <= min; i++)
2738          {
2739          int slength;
2740          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2741            {
2742            if (slength == -2) eptr = md->end_subject;   /* Partial match */
2743            CHECK_PARTIAL();
2744            RRETURN(MATCH_NOMATCH);
2745          }          }
2746          eptr += slength;
2747          }
2748    
2749        /* If the length of the reference is zero, just continue with the      /* If min = max, continue at the same level without recursion.
2750        main loop. */      They are not both allowed to be zero. */
2751    
2752        if (length == 0) continue;      if (min == max) continue;
2753    
2754        /* First, ensure the minimum number of matches are present. We get back      /* If minimizing, keep trying and advancing the pointer */
       the length of the reference string explicitly rather than passing the  
       address of eptr, so that eptr can be a register variable. */  
2755    
2756        for (i = 1; i <= min; i++)      if (minimize)
2757          {
2758          for (fi = min;; fi++)
2759          {          {
2760          if (!match_ref(offset, eptr, length, md, ims))          int slength;
2761            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2762            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2763            if (fi >= max) RRETURN(MATCH_NOMATCH);
2764            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2765            {            {
2766              if (slength == -2) eptr = md->end_subject;   /* Partial match */
2767            CHECK_PARTIAL();            CHECK_PARTIAL();
2768            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2769            }            }
2770          eptr += length;          eptr += slength;
2771          }          }
2772          /* Control never gets here */
2773          }
2774    
2775        /* If min = max, continue at the same level without recursion.      /* If maximizing, find the longest string and work backwards */
       They are not both allowed to be zero. */  
   
       if (min == max) continue;  
   
       /* If minimizing, keep trying and advancing the pointer */  
2776    
2777        if (minimize)      else
2778          {
2779          pp = eptr;
2780          for (i = min; i < max; i++)
2781          {          {
2782          for (fi = min;; fi++)          int slength;
2783            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2784            {            {
2785            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            /* Can't use CHECK_PARTIAL because we don't want to update eptr in
2786            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            the soft partial matching case. */
2787            if (fi >= max) RRETURN(MATCH_NOMATCH);  
2788            if (!match_ref(offset, eptr, length, md, ims))            if (slength == -2 && md->partial != 0 &&
2789                  md->end_subject > md->start_used_ptr)
2790              {              {
2791              CHECK_PARTIAL();              md->hitend = TRUE;
2792              RRETURN(MATCH_NOMATCH);              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2793              }              }
2794            eptr += length;            break;
2795            }            }
2796          /* Control never gets here */          eptr += slength;
2797          }          }
2798    
2799        /* If maximizing, find the longest string and work backwards */        while (eptr >= pp)
   
       else  
2800          {          {
2801          pp = eptr;          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2802          for (i = min; i < max; i++)          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2803            {          eptr -= length;
           if (!match_ref(offset, eptr, length, md, ims)) break;  
           eptr += length;  
           }  
         while (eptr >= pp)  
           {  
           RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);  
           if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
           eptr -= length;  
           }  
         RRETURN(MATCH_NOMATCH);  
2804          }          }
2805          RRETURN(MATCH_NOMATCH);
2806        }        }
2807      /* Control never gets here */      /* Control never gets here */
2808    
# Line 2053  for (;;) Line 2820  for (;;)
2820      case OP_NCLASS:      case OP_NCLASS:
2821      case OP_CLASS:      case OP_CLASS:
2822        {        {
2823          /* The data variable is saved across frames, so the byte map needs to
2824          be stored there. */
2825    #define BYTE_MAP ((pcre_uint8 *)data)
2826        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2827        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2828    
2829        switch (*ecode)        switch (*ecode)
2830          {          {
# Line 2075  for (;;) Line 2845  for (;;)
2845          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2846          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2847          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2848          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2849          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2850          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2851          break;          break;
2852    
2853          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2087  for (;;) Line 2857  for (;;)
2857    
2858        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2859    
2860  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2861        /* UTF-8 mode */        if (utf)
       if (utf8)  
2862          {          {
2863          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2864            {            {
# Line 2104  for (;;) Line 2873  for (;;)
2873              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2874              }              }
2875            else            else
2876              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  
             }  
2877            }            }
2878          }          }
2879        else        else
2880  #endif  #endif
2881        /* Not UTF-8 mode */        /* Not UTF mode */
2882          {          {
2883          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2884            {            {
# Line 2121  for (;;) Line 2888  for (;;)
2888              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2889              }              }
2890            c = *eptr++;            c = *eptr++;
2891            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2892              if (c > 255)
2893                {
2894                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2895                }
2896              else
2897    #endif
2898                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2899            }            }
2900          }          }
2901    
# Line 2135  for (;;) Line 2909  for (;;)
2909    
2910        if (minimize)        if (minimize)
2911          {          {
2912  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2913          /* UTF-8 mode */          if (utf)
         if (utf8)  
2914            {            {
2915            for (fi = min;; fi++)            for (fi = min;; fi++)
2916              {              {
2917              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2918              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2919              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2920              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2155  for (;;) Line 2928  for (;;)
2928                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2929                }                }
2930              else              else
2931                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  
               }  
2932              }              }
2933            }            }
2934          else          else
2935  #endif  #endif
2936          /* Not UTF-8 mode */          /* Not UTF mode */
2937            {            {
2938            for (fi = min;; fi++)            for (fi = min;; fi++)
2939              {              {
2940              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2941              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2942              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2943              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2175  for (;;) Line 2946  for (;;)
2946                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2947                }                }
2948              c = *eptr++;              c = *eptr++;
2949              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2950                if (c > 255)
2951                  {
2952                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2953                  }
2954                else
2955    #endif
2956                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2957              }              }
2958            }            }
2959          /* Control never gets here */          /* Control never gets here */
# Line 2187  for (;;) Line 2965  for (;;)
2965          {          {
2966          pp = eptr;          pp = eptr;
2967    
2968  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2969          /* UTF-8 mode */          if (utf)
         if (utf8)  
2970            {            {
2971            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2972              {              {
2973              int len = 1;              int len = 1;
2974              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2975                  {
2976                  SCHECK_PARTIAL();
2977                  break;
2978                  }
2979              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2980              if (c > 255)              if (c > 255)
2981                {                {
2982                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
2983                }                }
2984              else              else
2985                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
2986              eptr += len;              eptr += len;
2987              }              }
2988            for (;;)            for (;;)
2989              {              {
2990              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2991              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2992              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2993              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2216  for (;;) Line 2995  for (;;)
2995            }            }
2996          else          else
2997  #endif  #endif
2998            /* Not UTF-8 mode */            /* Not UTF mode */
2999            {            {
3000            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3001              {              {
3002              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3003                  {
3004                  SCHECK_PARTIAL();
3005                  break;
3006                  }
3007              c = *eptr;              c = *eptr;
3008              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
3009                if (c > 255)
3010                  {
3011                  if (op == OP_CLASS) break;
3012                  }
3013                else
3014    #endif
3015                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3016              eptr++;              eptr++;
3017              }              }
3018            while (eptr >= pp)            while (eptr >= pp)
3019              {              {
3020              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
3021              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3022              eptr--;              eptr--;
3023              }              }
# Line 2235  for (;;) Line 3025  for (;;)
3025    
3026          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3027          }          }
3028    #undef BYTE_MAP
3029        }        }
3030      /* Control never gets here */      /* Control never gets here */
3031    
# Line 2243  for (;;) Line 3034  for (;;)
3034      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
3035      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
3036    
3037  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3038      case OP_XCLASS:      case OP_XCLASS:
3039        {        {
3040        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2268  for (;;) Line 3059  for (;;)
3059          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3060          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3061          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3062          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
3063          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
3064          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
3065          break;          break;
3066    
3067          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2288  for (;;) Line 3079  for (;;)
3079            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3080            }            }
3081          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
3082          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3083          }          }
3084    
3085        /* 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 2303  for (;;) Line 3094  for (;;)
3094          {          {
3095          for (fi = min;; fi++)          for (fi = min;; fi++)
3096            {            {
3097            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
3098            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3099            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3100            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2312  for (;;) Line 3103  for (;;)
3103              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3104              }              }
3105            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3106            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3107            }            }
3108          /* Control never gets here */          /* Control never gets here */
3109          }          }
# Line 2325  for (;;) Line 3116  for (;;)
3116          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3117            {            {
3118            int len = 1;            int len = 1;
3119            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
3120                {
3121                SCHECK_PARTIAL();
3122                break;
3123                }
3124    #ifdef SUPPORT_UTF
3125            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3126            if (!_pcre_xclass(c, data)) break;  #else
3127              c = *eptr;
3128    #endif
3129              if (!PRIV(xclass)(c, data, utf)) break;
3130            eptr += len;            eptr += len;
3131            }            }
3132          for(;;)          for(;;)
3133            {            {
3134            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3135            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3136            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3137            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3138              if (utf) BACKCHAR(eptr);
3139    #endif
3140            }            }
3141          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3142          }          }
# Line 2347  for (;;) Line 3148  for (;;)
3148      /* Match a single character, casefully */      /* Match a single character, casefully */
3149    
3150      case OP_CHAR:      case OP_CHAR:
3151  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3152      if (utf8)      if (utf)
3153        {        {
3154        length = 1;        length = 1;
3155        ecode++;        ecode++;
# Line 2362  for (;;) Line 3163  for (;;)
3163        }        }
3164      else      else
3165  #endif  #endif
3166        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3167        {        {
3168        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3169          {          {
# Line 2375  for (;;) Line 3175  for (;;)
3175        }        }
3176      break;      break;
3177    
3178      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3179        subject, give up immediately. */
3180    
3181        case OP_CHARI:
3182        if (eptr >= md->end_subject)
3183          {
3184          SCHECK_PARTIAL();
3185          RRETURN(MATCH_NOMATCH);
3186          }
3187    
3188      case OP_CHARNC:  #ifdef SUPPORT_UTF
3189  #ifdef SUPPORT_UTF8      if (utf)
     if (utf8)  
3190        {        {
3191        length = 1;        length = 1;
3192        ecode++;        ecode++;
3193        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3194    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         RRETURN(MATCH_NOMATCH);  
         }  
   
3195        /* 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
3196        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3197          fast lookup table. We know that there is at least one byte left in the
3198          subject. */
3199    
3200        if (fc < 128)        if (fc < 128)
3201          {          {
3202          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[fc]
3203                != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3204            ecode++;
3205            eptr++;
3206          }          }
3207    
3208        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3209          use the value of "length" to check for sufficient bytes left, because the
3210          other case of the character may have more or fewer bytes.  */
3211    
3212        else        else
3213          {          {
# Line 2420  for (;;) Line 3228  for (;;)
3228          }          }
3229        }        }
3230      else      else
3231  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3232    
3233      /* Non-UTF-8 mode */      /* Not UTF mode */
3234        {        {
3235        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3236          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3237          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         RRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);  
3238        ecode += 2;        ecode += 2;
3239        }        }
3240      break;      break;
# Line 2437  for (;;) Line 3242  for (;;)
3242      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3243    
3244      case OP_EXACT:      case OP_EXACT:
3245        case OP_EXACTI:
3246      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3247      ecode += 3;      ecode += 1 + IMM2_SIZE;
3248      goto REPEATCHAR;      goto REPEATCHAR;
3249    
3250      case OP_POSUPTO:      case OP_POSUPTO:
3251        case OP_POSUPTOI:
3252      possessive = TRUE;      possessive = TRUE;
3253      /* Fall through */      /* Fall through */
3254    
3255      case OP_UPTO:      case OP_UPTO:
3256        case OP_UPTOI:
3257      case OP_MINUPTO:      case OP_MINUPTO:
3258        case OP_MINUPTOI:
3259      min = 0;      min = 0;
3260      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3261      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3262      ecode += 3;      ecode += 1 + IMM2_SIZE;
3263      goto REPEATCHAR;      goto REPEATCHAR;
3264    
3265      case OP_POSSTAR:      case OP_POSSTAR:
3266        case OP_POSSTARI:
3267      possessive = TRUE;      possessive = TRUE;
3268      min = 0;      min = 0;
3269      max = INT_MAX;      max = INT_MAX;
# Line 2461  for (;;) Line 3271  for (;;)
3271      goto REPEATCHAR;      goto REPEATCHAR;
3272    
3273      case OP_POSPLUS:      case OP_POSPLUS:
3274        case OP_POSPLUSI:
3275      possessive = TRUE;      possessive = TRUE;
3276      min = 1;      min = 1;
3277      max = INT_MAX;      max = INT_MAX;
# Line 2468  for (;;) Line 3279  for (;;)
3279      goto REPEATCHAR;      goto REPEATCHAR;
3280    
3281      case OP_POSQUERY:      case OP_POSQUERY:
3282        case OP_POSQUERYI:
3283      possessive = TRUE;      possessive = TRUE;
3284      min = 0;      min = 0;
3285      max = 1;      max = 1;
# Line 2475  for (;;) Line 3287  for (;;)
3287      goto REPEATCHAR;      goto REPEATCHAR;
3288    
3289      case OP_STAR:      case OP_STAR:
3290        case OP_STARI:
3291      case OP_MINSTAR:      case OP_MINSTAR:
3292        case OP_MINSTARI:
3293      case OP_PLUS:      case OP_PLUS:
3294        case OP_PLUSI:
3295      case OP_MINPLUS:      case OP_MINPLUS:
3296        case OP_MINPLUSI:
3297      case OP_QUERY:      case OP_QUERY:
3298        case OP_QUERYI:
3299      case OP_MINQUERY:      case OP_MINQUERY:
3300      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3301        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3302      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3303      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3304      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3305      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2490  for (;;) Line 3307  for (;;)
3307      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3308    
3309      REPEATCHAR:      REPEATCHAR:
3310  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3311      if (utf8)      if (utf)
3312        {        {
3313        length = 1;        length = 1;
3314        charptr = ecode;        charptr = ecode;
# Line 2505  for (;;) Line 3322  for (;;)
3322          {          {
3323  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3324          unsigned int othercase;          unsigned int othercase;
3325          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3326              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3327            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3328          else oclength = 0;          else oclength = 0;
3329  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3330    
3331          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3332            {            {
3333            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3334              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3335  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3336            else if (oclength > 0 &&            else if (oclength > 0 &&
3337                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3338                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3339  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3340            else            else
3341              {              {
# Line 2533  for (;;) Line 3350  for (;;)
3350            {            {
3351            for (fi = min;; fi++)            for (fi = min;; fi++)
3352              {              {
3353              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3354              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3355              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3356              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3357                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3358  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3359              else if (oclength > 0 &&              else if (oclength > 0 &&
3360                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3361                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3362  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3363              else              else
3364                {                {
# Line 2558  for (;;) Line 3375  for (;;)
3375            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3376              {              {
3377              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3378                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3379  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3380              else if (oclength > 0 &&              else if (oclength > 0 &&
3381                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3382                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3383  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3384              else break;              else
3385                  {
3386                  CHECK_PARTIAL();
3387                  break;
3388                  }
3389              }              }
3390    
3391            if (possessive) continue;            if (possessive) continue;
3392    
3393            for(;;)            for(;;)
3394              {              {
3395              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3396              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3397              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3398  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2590  for (;;) Line 3411  for (;;)
3411        value of fc will always be < 128. */        value of fc will always be < 128. */
3412        }        }
3413      else      else
3414  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3415          /* When not in UTF-8 mode, load a single-byte character. */
3416          fc = *ecode++;
3417    
3418      /* When not in UTF-8 mode, load a single-byte character. */      /* The value of fc at this point is always one character, though we may
3419        or may not be in UTF mode. The code is duplicated for the caseless and
     fc = *ecode++;  
   
     /* The value of fc at this point is always less than 256, though we may or  
     may not be in UTF-8 mode. The code is duplicated for the caseless and  
3420      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3421      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3422      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 2608  for (;;) Line 3427  for (;;)
3427      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3428        max, eptr));        max, eptr));
3429    
3430      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3431        {        {
3432        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3433          /* fc must be < 128 if UTF is enabled. */
3434          foc = md->fcc[fc];
3435    #else
3436    #ifdef SUPPORT_UTF
3437    #ifdef SUPPORT_UCP
3438          if (utf && fc > 127)
3439            foc = UCD_OTHERCASE(fc);
3440    #else
3441          if (utf && fc > 127)
3442            foc = fc;
3443    #endif /* SUPPORT_UCP */
3444          else
3445    #endif /* SUPPORT_UTF */
3446            foc = TABLE_GET(fc, md->fcc, fc);
3447    #endif /* COMPILE_PCRE8 */
3448    
3449        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3450          {          {
3451          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
# Line 2618  for (;;) Line 3453  for (;;)
3453            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3454            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3455            }            }
3456          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3457            eptr++;
3458          }          }
3459        if (min == max) continue;        if (min == max) continue;
3460        if (minimize)        if (minimize)
3461          {          {
3462          for (fi = min;; fi++)          for (fi = min;; fi++)
3463            {            {
3464            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3465            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3466            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3467            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2633  for (;;) Line 3469  for (;;)
3469              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3470              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3471              }              }
3472            if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3473              eptr++;
3474            }            }
3475          /* Control never gets here */          /* Control never gets here */
3476          }          }
# Line 2642  for (;;) Line 3479  for (;;)
3479          pp = eptr;          pp = eptr;
3480          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3481            {            {
3482            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3483                {
3484                SCHECK_PARTIAL();
3485                break;
3486                }
3487              if (fc != *eptr && foc != *eptr) break;
3488            eptr++;            eptr++;
3489            }            }
3490    
# Line 2650  for (;;) Line 3492  for (;;)
3492    
3493          while (eptr >= pp)          while (eptr >= pp)
3494            {            {
3495            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3496            eptr--;            eptr--;
3497            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3498            }            }
# Line 2679  for (;;) Line 3521  for (;;)
3521          {          {
3522          for (fi = min;; fi++)          for (fi = min;; fi++)
3523            {            {
3524            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3525            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3526            if (fi >= max) RRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3527            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2696  for (;;) Line 3538  for (;;)
3538          pp = eptr;          pp = eptr;
3539          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3540            {            {
3541            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3542                {
3543                SCHECK_PARTIAL();
3544                break;
3545                }
3546              if (fc != *eptr) break;
3547            eptr++;            eptr++;
3548            }            }
3549          if (possessive) continue;          if (possessive) continue;
3550    
3551          while (eptr >= pp)          while (eptr >= pp)
3552            {            {
3553            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3554            eptr--;            eptr--;
3555            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3556            }            }
# Line 2716  for (;;) Line 3563  for (;;)
3563      checking can be multibyte. */      checking can be multibyte. */
3564    
3565      case OP_NOT:      case OP_NOT:
3566        case OP_NOTI:
3567      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3568        {        {
3569        SCHECK_PARTIAL();        SCHECK_PARTIAL();
3570        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
3571        }        }
3572      ecode++;  #ifdef SUPPORT_UTF
3573      GETCHARINCTEST(c, eptr);      if (utf)
     if ((ims & PCRE_CASELESS) != 0)  
3574        {        {
3575  #ifdef SUPPORT_UTF8        register unsigned int ch, och;
3576        if (c < 256)  
3577  #endif        ecode++;
3578        c = md->lcc[c];        GETCHARINC(ch, ecode);
3579        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        GETCHARINC(c, eptr);
3580    
3581          if (op == OP_NOT)
3582            {
3583            if (ch == c) RRETURN(MATCH_NOMATCH);
3584            }
3585          else
3586            {
3587    #ifdef SUPPORT_UCP
3588            if (ch > 127)
3589              och = UCD_OTHERCASE(ch);
3590    #else
3591            if (ch > 127)
3592              och = ch;
3593    #endif /* SUPPORT_UCP */
3594            else
3595              och = TABLE_GET(ch, md->fcc, ch);
3596            if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3597            }
3598        }        }
3599      else      else
3600    #endif
3601        {        {
3602        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        register unsigned int ch = ecode[1];
3603          c = *eptr++;
3604          if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c))
3605            RRETURN(MATCH_NOMATCH);
3606          ecode += 2;
3607        }        }
3608      break;      break;
3609    
# Line 2745  for (;;) Line 3615  for (;;)
3615      about... */      about... */
3616    
3617      case OP_NOTEXACT:      case OP_NOTEXACT:
3618        case OP_NOTEXACTI:
3619      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3620      ecode += 3;      ecode += 1 + IMM2_SIZE;
3621      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3622    
3623      case OP_NOTUPTO:      case OP_NOTUPTO:
3624        case OP_NOTUPTOI:
3625      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3626        case OP_NOTMINUPTOI:
3627      min = 0;      min = 0;
3628      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3629      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3630      ecode += 3;      ecode += 1 + IMM2_SIZE;
3631      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3632    
3633      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3634        case OP_NOTPOSSTARI:
3635      possessive = TRUE;      possessive = TRUE;
3636      min = 0;      min = 0;
3637      max = INT_MAX;      max = INT_MAX;
# Line 2765  for (;;) Line 3639  for (;;)
3639      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3640    
3641      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3642        case OP_NOTPOSPLUSI:
3643      possessive = TRUE;      possessive = TRUE;
3644      min = 1;      min = 1;
3645      max = INT_MAX;      max = INT_MAX;
# Line 2772  for (;;) Line 3647  for (;;)
3647      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3648    
3649      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3650        case OP_NOTPOSQUERYI:
3651      possessive = TRUE;      possessive = TRUE;
3652      min = 0;      min = 0;
3653      max = 1;      max = 1;
# Line 2779  for (;;) Line 3655  for (;;)
3655      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3656    
3657      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3658        case OP_NOTPOSUPTOI:
3659      possessive = TRUE;      possessive = TRUE;
3660      min = 0;      min = 0;
3661      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3662      ecode += 3;      ecode += 1 + IMM2_SIZE;
3663      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3664    
3665      case OP_NOTSTAR:      case OP_NOTSTAR:
3666        case OP_NOTSTARI:
3667      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3668        case OP_NOTMINSTARI:
3669      case OP_NOTPLUS:      case OP_NOTPLUS:
3670        case OP_NOTPLUSI:
3671      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3672        case OP_NOTMINPLUSI:
3673      case OP_NOTQUERY:      case OP_NOTQUERY:
3674        case OP_NOTQUERYI:
3675      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3676      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3677        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3678      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3679      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3680      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 2800  for (;;) Line 3683  for (;;)
3683      /* Common code for all repeated single-byte matches. */      /* Common code for all repeated single-byte matches. */
3684    
3685      REPEATNOTCHAR:      REPEATNOTCHAR:
3686      fc = *ecode++;      GETCHARINCTEST(fc, ecode);
3687    
3688      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
3689      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 2813  for (;;) Line 3696  for (;;)
3696      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,
3697        max, eptr));        max, eptr));
3698    
3699      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3700        {        {
3701        fc = md->lcc[fc];  #ifdef SUPPORT_UTF
3702    #ifdef SUPPORT_UCP
3703          if (utf && fc > 127)
3704            foc = UCD_OTHERCASE(fc);
3705    #else
3706          if (utf && fc > 127)
3707            foc = fc;
3708    #endif /* SUPPORT_UCP */
3709          else
3710    #endif /* SUPPORT_UTF */
3711            foc = TABLE_GET(fc, md->fcc, fc);
3712    
3713  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3714        /* UTF-8 mode */        if (utf)
       if (utf8)  
3715          {          {
3716          register unsigned int d;          register unsigned int d;
3717          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 2830  for (;;) Line 3722  for (;;)
3722              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3723              }              }
3724            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3725            if (d < 256) d = md->lcc[d];            if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) RRETURN(MATCH_NOMATCH);  
3726            }            }
3727          }          }
3728        else        else
3729  #endif  #endif
3730          /* Not UTF mode */
       /* Not UTF-8 mode */  
3731          {          {
3732          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3733            {            {
# Line 2846  for (;;) Line 3736  for (;;)
3736              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3737              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3738              }              }
3739            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3740              eptr++;
3741            }            }
3742          }          }
3743    
# Line 2854  for (;;) Line 3745  for (;;)
3745    
3746        if (minimize)        if (minimize)
3747          {          {
3748  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3749          /* UTF-8 mode */          if (utf)
         if (utf8)  
3750            {            {
3751            register unsigned int d;            register unsigned int d;
3752            for (fi = min;; fi++)            for (fi = min;; fi++)
3753              {              {
3754              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3755              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3756              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3757              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2870  for (;;) Line 3760  for (;;)
3760                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3761                }                }
3762              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3763              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
             if (fc == d) RRETURN(MATCH_NOMATCH);  
3764              }              }
3765            }            }
3766          else          else
3767  #endif  #endif
3768          /* Not UTF-8 mode */          /* Not UTF mode */
3769            {            {
3770            for (fi = min;; fi++)            for (fi = min;; fi++)
3771              {              {
3772              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3773              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3774              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3775              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2888  for (;;) Line 3777  for (;;)
3777                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3778                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3779                }                }
3780              if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);              if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3781                eptr++;
3782              }              }
3783            }            }
3784          /* Control never gets here */          /* Control never gets here */
# Line 2900  for (;;) Line 3790  for (;;)
3790          {          {
3791          pp = eptr;          pp = eptr;
3792    
3793  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3794          /* UTF-8 mode */          if (utf)
         if (utf8)  
3795            {            {
3796            register unsigned int d;            register unsigned int d;
3797            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3798              {              {
3799              int len = 1;              int len = 1;
3800              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3801                  {
3802                  SCHECK_PARTIAL();
3803                  break;
3804                  }
3805              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3806              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) break;
             if (fc == d) break;  
3807              eptr += len;              eptr += len;
3808              }              }
3809          if (possessive) continue;            if (possessive) continue;
3810          for(;;)            for(;;)
3811              {              {
3812              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3813              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3814              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3815              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2925  for (;;) Line 3817  for (;;)
3817            }            }
3818          else          else
3819  #endif  #endif
3820          /* Not UTF-8 mode */          /* Not UTF mode */
3821            {            {
3822            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3823              {              {
3824              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3825                  {
3826                  SCHECK_PARTIAL();
3827                  break;
3828                  }
3829                if (fc == *eptr || foc == *eptr) break;
3830              eptr++;              eptr++;
3831              }              }
3832            if (possessive) continue;            if (possessive) continue;
3833            while (eptr >= pp)            while (eptr >= pp)
3834              {              {
3835              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3836              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3837              eptr--;              eptr--;
3838              }              }
# Line 2950  for (;;) Line 3847  for (;;)
3847    
3848      else      else
3849        {        {
3850  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3851        /* UTF-8 mode */        if (utf)
       if (utf8)  
3852          {          {
3853          register unsigned int d;          register unsigned int d;
3854          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 2968  for (;;) Line 3864  for (;;)
3864          }          }
3865        else        else
3866  #endif  #endif
3867        /* Not UTF-8 mode */        /* Not UTF mode */
3868          {          {
3869          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3870            {            {
# Line 2985  for (;;) Line 3881  for (;;)
3881    
3882        if (minimize)        if (minimize)
3883          {          {
3884  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3885          /* UTF-8 mode */          if (utf)
         if (utf8)  
3886            {            {
3887            register unsigned int d;            register unsigned int d;
3888            for (fi = min;; fi++)            for (fi = min;; fi++)
3889              {              {
3890              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3891              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3892              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3893              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3006  for (;;) Line 3901  for (;;)
3901            }            }
3902          else          else
3903  #endif  #endif
3904          /* Not UTF-8 mode */          /* Not UTF mode */
3905            {            {
3906            for (fi = min;; fi++)            for (fi = min;; fi++)
3907              {              {
3908              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3909              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3910              if (fi >= max) RRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3911              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3030  for (;;) Line 3925  for (;;)
3925          {          {
3926          pp = eptr;          pp = eptr;
3927    
3928  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3929          /* UTF-8 mode */          if (utf)
         if (utf8)  
3930            {            {
3931            register unsigned int d;            register unsigned int d;
3932            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3933              {              {
3934              int len = 1;              int len = 1;
3935              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3936                  {
3937                  SCHECK_PARTIAL();
3938                  break;
3939                  }
3940              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3941              if (fc == d) break;              if (fc == d) break;
3942              eptr += len;              eptr += len;
# Line 3046  for (;;) Line 3944  for (;;)
3944            if (possessive) continue;            if (possessive) continue;
3945            for(;;)            for(;;)
3946              {              {
3947              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3948              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3949              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3950              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3054  for (;;) Line 3952  for (;;)
3952            }            }
3953          else          else
3954  #endif  #endif
3955          /* Not UTF-8 mode */          /* Not UTF mode */
3956            {            {
3957            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3958              {              {
3959              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3960                  {
3961                  SCHECK_PARTIAL();
3962                  break;
3963                  }
3964                if (fc == *eptr) break;
3965              eptr++;              eptr++;
3966              }              }
3967            if (possessive) continue;            if (possessive) continue;
3968            while (eptr >= pp)            while (eptr >= pp)
3969              {              {
3970              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3971              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3972              eptr--;              eptr--;
3973              }              }
# Line 3082  for (;;) Line 3985  for (;;)
3985      case OP_TYPEEXACT:      case OP_TYPEEXACT:
3986      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3987      minimize = TRUE;      minimize = TRUE;
3988      ecode += 3;      ecode += 1 + IMM2_SIZE;
3989      goto REPEATTYPE;      goto REPEATTYPE;
3990    
3991      case OP_TYPEUPTO:      case OP_TYPEUPTO:
# Line 3090  for (;;) Line 3993  for (;;)
3993      min = 0;      min = 0;
3994      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3995      minimize = *ecode == OP_TYPEMINUPTO;      minimize = *ecode == OP_TYPEMINUPTO;
3996      ecode += 3;      ecode += 1 + IMM2_SIZE;
3997      goto REPEATTYPE;      goto REPEATTYPE;
3998    
3999      case OP_TYPEPOSSTAR:      case OP_TYPEPOSSTAR:
# Line 3118  for (;;) Line 4021  for (;;)
4021      possessive = TRUE;      possessive = TRUE;
4022      min = 0;      min = 0;
4023      max = GET2(ecode, 1);      max = GET2(ecode, 1);
4024      ecode += 3;      ecode += 1 + IMM2_SIZE;
4025      goto REPEATTYPE;      goto REPEATTYPE;
4026    
4027      case OP_TYPESTAR:      case OP_TYPESTAR:
# Line 3179  for (;;) Line 4082  for (;;)
4082            case PT_LAMP:            case PT_LAMP:
4083            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4084              {              {
4085                int chartype;
4086              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4087                {                {
4088                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4089                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4090                }                }
4091              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4092              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4093              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4094                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4095                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4096                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4097              }              }
4098            break;            break;
# Line 3202  for (;;) Line 4106  for (;;)
4106                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4107                }                }
4108              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4109              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4110                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4111              }              }
4112            break;            break;