/[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 569 by ph10, Sun Nov 7 16:14:50 2010 UTC revision 1033 by ph10, Mon Sep 10 11:02:48 2012 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2010 University of Cambridge             Copyright (c) 1997-2012 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 37  POSSIBILITY OF SUCH DAMAGE. Line 37  POSSIBILITY OF SUCH DAMAGE.
37  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
38  */  */
39    
   
40  /* This module contains pcre_exec(), the externally visible function that does  /* This module contains pcre_exec(), the externally visible function that does
41  pattern matching using an NFA algorithm, trying to mimic Perl as closely as  pattern matching using an NFA algorithm, trying to mimic Perl as closely as
42  possible. There are also some static supporting functions. */  possible. There are also some static supporting functions. */
# Line 57  possible. There are also some static sup Line 56  possible. There are also some static sup
56  #undef min  #undef min
57  #undef max  #undef max
58    
59  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
60    of call to match(). We do it this way to save on using another stack variable,
61    as stack usage is to be discouraged. */
62    
63  #define match_condassert     0x01  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
64  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
65    
66  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
67  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 73  negative to avoid the external error cod Line 74  negative to avoid the external error cod
74    
75  #define MATCH_ACCEPT       (-999)  #define MATCH_ACCEPT       (-999)
76  #define MATCH_COMMIT       (-998)  #define MATCH_COMMIT       (-998)
77  #define MATCH_PRUNE        (-997)  #define MATCH_KETRPOS      (-997)
78  #define MATCH_SKIP         (-996)  #define MATCH_ONCE         (-996)
79  #define MATCH_SKIP_ARG     (-995)  #define MATCH_PRUNE        (-995)
80  #define MATCH_THEN         (-994)  #define MATCH_SKIP         (-994)
81    #define MATCH_SKIP_ARG     (-993)
82  /* This is a convenience macro for code that occurs many times. */  #define MATCH_THEN         (-992)
   
 #define MRRETURN(ra) \  
   { \  
   md->mark = markptr; \  
   RRETURN(ra); \  
   }  
83    
84  /* Maximum number of ints of offset to save on the stack for recursive calls.  /* Maximum number of ints of offset to save on the stack for recursive calls.
85  If the offset vector is bigger, malloc is used. This should be a multiple of 3,  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
# Line 117  Returns:     nothing Line 112  Returns:     nothing
112  */  */
113    
114  static void  static void
115  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)  pchars(const pcre_uchar *p, int length, BOOL is_subject, match_data *md)
116  {  {
117  unsigned int c;  unsigned int c;
118  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;
# Line 132  while (length-- > 0) Line 127  while (length-- > 0)
127  *          Match a back-reference                *  *          Match a back-reference                *
128  *************************************************/  *************************************************/
129    
130  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
131  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
132    the length passed is zero. Note that in caseless UTF-8 mode, the number of
133    subject bytes matched may be different to the number of reference bytes.
134    
135  Arguments:  Arguments:
136    offset      index into the offset vector    offset      index into the offset vector
137    eptr        points into the subject    eptr        pointer into the subject
138    length      length to be matched    length      length of reference to be matched (number of bytes)
139    md          points to match data block    md          points to match data block
140    ims         the ims flags    caseless    TRUE if caseless
141    
142  Returns:      TRUE if matched  Returns:      >= 0 the number of subject bytes matched
143                  -1 no match
144                  -2 partial match; always given if at end subject
145  */  */
146    
147  static BOOL  static int
148  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register PCRE_PUCHAR eptr, int length, match_data *md,
149    unsigned long int ims)    BOOL caseless)
150  {  {
151  USPTR p = md->start_subject + md->offset_vector[offset];  PCRE_PUCHAR eptr_start = eptr;
152    register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
153    
154  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
155  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 164  pchars(p, length, FALSE, md); Line 164  pchars(p, length, FALSE, md);
164  printf("\n");  printf("\n");
165  #endif  #endif
166    
167  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible - in that
168    case the length is passed as zero). */
169    
170  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
171    
172  /* Separate the caseless case for speed. In UTF-8 mode we can only do this  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
173  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
174  ASCII characters. */  ASCII characters. */
175    
176  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
177    {    {
178  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
179  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
180    if (md->utf8)    if (md->utf)
181      {      {
182      USPTR endptr = eptr + length;      /* Match characters up to the end of the reference. NOTE: the number of
183      while (eptr < endptr)      bytes matched may differ, because there are some characters whose upper and
184        lower case versions code as different numbers of bytes. For example, U+023A
185        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
186        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
187        the latter. It is important, therefore, to check the length along the
188        reference, not along the subject (earlier code did this wrong). */
189    
190        PCRE_PUCHAR endptr = p + length;
191        while (p < endptr)
192        {        {
193        int c, d;        int c, d;
194          if (eptr >= md->end_subject) return -2;   /* Partial match */
195        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
196        GETCHARINC(d, p);        GETCHARINC(d, p);
197        if (c != d && c != UCD_OTHERCASE(d)) return FALSE;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
198        }        }
199      }      }
200    else    else
# Line 193  if ((ims & PCRE_CASELESS) != 0) Line 203  if ((ims & PCRE_CASELESS) != 0)
203    
204    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
205    is no UCP support. */    is no UCP support. */
206        {
207    while (length-- > 0)      while (length-- > 0)
208      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }        {
209          if (eptr >= md->end_subject) return -2;   /* Partial match */
210          if (TABLE_GET(*p, md->lcc, *p) != TABLE_GET(*eptr, md->lcc, *eptr)) return -1;
211          p++;
212          eptr++;
213          }
214        }
215    }    }
216    
217  /* In the caseful case, we can just compare the bytes, whether or not we  /* In the caseful case, we can just compare the bytes, whether or not we
218  are in UTF-8 mode. */  are in UTF-8 mode. */
219    
220  else  else
221    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
222      while (length-- > 0)
223        {
224        if (eptr >= md->end_subject) return -2;   /* Partial match */
225        if (*p++ != *eptr++) return -1;
226        }
227      }
228    
229  return TRUE;  return (int)(eptr - eptr_start);
230  }  }
231    
232    
# Line 256  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 278  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
278         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
279         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
280         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
281         RM61,  RM62 };         RM61,  RM62, RM63, RM64, RM65, RM66 };
282    
283  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
284  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
# Line 266  actually used in this definition. */ Line 288  actually used in this definition. */
288  #define REGISTER register  #define REGISTER register
289    
290  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
291  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
292    { \    { \
293    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
294    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \
295    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
296    }    }
297  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 278  actually used in this definition. */ Line 300  actually used in this definition. */
300    return ra; \    return ra; \
301    }    }
302  #else  #else
303  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
304    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1)
305  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
306  #endif  #endif
307    
# Line 292  argument of match(), which never changes Line 314  argument of match(), which never changes
314    
315  #define REGISTER  #define REGISTER
316    
317  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
318    {\    {\
319    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = frame->Xnextframe;\
320    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    if (newframe == NULL)\
321    frame->Xwhere = rw; \      {\
322        newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\
323        if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
324        newframe->Xnextframe = NULL;\
325        frame->Xnextframe = newframe;\
326        }\
327      frame->Xwhere = rw;\
328    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
329    newframe->Xecode = rb;\    newframe->Xecode = rb;\
330    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
   newframe->Xmarkptr = markptr;\  
331    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
332    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
333    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
334    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
335    frame = newframe;\    frame = newframe;\
# Line 318  argument of match(), which never changes Line 343  argument of match(), which never changes
343    {\    {\
344    heapframe *oldframe = frame;\    heapframe *oldframe = frame;\
345    frame = oldframe->Xprevframe;\    frame = oldframe->Xprevframe;\
   (pcre_stack_free)(oldframe);\  
346    if (frame != NULL)\    if (frame != NULL)\
347      {\      {\
348      rrc = ra;\      rrc = ra;\
# Line 332  argument of match(), which never changes Line 356  argument of match(), which never changes
356    
357  typedef struct heapframe {  typedef struct heapframe {
358    struct heapframe *Xprevframe;    struct heapframe *Xprevframe;
359      struct heapframe *Xnextframe;
360    
361    /* Function arguments that may change */    /* Function arguments that may change */
362    
363    USPTR Xeptr;    PCRE_PUCHAR Xeptr;
364    const uschar *Xecode;    const pcre_uchar *Xecode;
365    USPTR Xmstart;    PCRE_PUCHAR Xmstart;
   USPTR Xmarkptr;  
366    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
367    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
368    unsigned int Xrdepth;    unsigned int Xrdepth;
369    
370    /* Function local variables */    /* Function local variables */
371    
372    USPTR Xcallpat;    PCRE_PUCHAR Xcallpat;
373  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
374    USPTR Xcharptr;    PCRE_PUCHAR Xcharptr;
375  #endif  #endif
376    USPTR Xdata;    PCRE_PUCHAR Xdata;
377    USPTR Xnext;    PCRE_PUCHAR Xnext;
378    USPTR Xpp;    PCRE_PUCHAR Xpp;
379    USPTR Xprev;    PCRE_PUCHAR Xprev;
380    USPTR Xsaved_eptr;    PCRE_PUCHAR Xsaved_eptr;
381    
382    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
383    
# Line 363  typedef struct heapframe { Line 385  typedef struct heapframe {
385    BOOL Xcondition;    BOOL Xcondition;
386    BOOL Xprev_is_word;    BOOL Xprev_is_word;
387    
   unsigned long int Xoriginal_ims;  
   
388  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
389    int Xprop_type;    int Xprop_type;
390    int Xprop_value;    int Xprop_value;
391    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
392    int Xoclength;    int Xoclength;
393    uschar Xocchars[8];    pcre_uchar Xocchars[6];
394  #endif  #endif
395    
396    int Xcodelink;    int Xcodelink;
# Line 415  returns a negative (error) response, the Line 432  returns a negative (error) response, the
432  same response. */  same response. */
433    
434  /* These macros pack up tests that are used for partial matching, and which  /* These macros pack up tests that are used for partial matching, and which
435  appears several times in the code. We set the "hit end" flag if the pointer is  appear several times in the code. We set the "hit end" flag if the pointer is
436  at the end of the subject and also past the start of the subject (i.e.  at the end of the subject and also past the start of the subject (i.e.
437  something has been matched). For hard partial matching, we then return  something has been matched). For hard partial matching, we then return
438  immediately. The second one is used when we already know we are past the end of  immediately. The second one is used when we already know we are past the end of
# Line 426  the subject. */ Line 443  the subject. */
443        eptr > md->start_used_ptr) \        eptr > md->start_used_ptr) \
444      { \      { \
445      md->hitend = TRUE; \      md->hitend = TRUE; \
446      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
447      }      }
448    
449  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
450    if (md->partial != 0 && eptr > md->start_used_ptr) \    if (md->partial != 0 && eptr > md->start_used_ptr) \
451      { \      { \
452      md->hitend = TRUE; \      md->hitend = TRUE; \
453      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
454      }      }
455    
456    
457  /* Performance note: It might be tempting to extract commonly used fields from  /* Performance note: It might be tempting to extract commonly used fields from
458  the md structure (e.g. utf8, end_subject) into individual variables to improve  the md structure (e.g. utf, end_subject) into individual variables to improve
459  performance. Tests using gcc on a SPARC disproved this; in the first case, it  performance. Tests using gcc on a SPARC disproved this; in the first case, it
460  made performance worse.  made performance worse.
461    
# Line 447  Arguments: Line 464  Arguments:
464     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
465     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
466                   by encountering \K)                   by encountering \K)
    markptr     pointer to the most recent MARK name, or NULL  
467     offset_top  current top pointer     offset_top  current top pointer
468     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
469     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
470                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
471     rdepth      the recursion depth     rdepth      the recursion depth
472    
473  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
# Line 467  Returns:       MATCH_MATCH if matched Line 478  Returns:       MATCH_MATCH if matched
478  */  */
479    
480  static int  static int
481  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode,
482    const uschar *markptr, int offset_top, match_data *md, unsigned long int ims,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
483    eptrblock *eptrb, int flags, unsigned int rdepth)    unsigned int rdepth)
484  {  {
485  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
486  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 478  so they can be ordinary variables in all Line 489  so they can be ordinary variables in all
489  register int  rrc;         /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
490  register int  i;           /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
491  register unsigned int c;   /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
492  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf;         /* Local copy of UTF flag for speed */
493    
494  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
495    BOOL caseless;
496  int condcode;  int condcode;
497    
498  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
499  preserved over calls to RMATCH() are part of a "frame" which is obtained from  preserved over calls to RMATCH() are part of a "frame". We set up the top-level
500  heap storage. Set up the top-level frame here; others are obtained from the  frame on the stack here; subsequent instantiations are obtained from the heap
501  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  whenever RMATCH() does a "recursion". See the macro definitions above. Putting
502    the top-level on the stack rather than malloc-ing them all gives a performance
503    boost in many cases where there is not much "recursion". */
504    
505  #ifdef NO_RECURSE  #ifdef NO_RECURSE
506  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)md->match_frames_base;
 if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);  
 frame->Xprevframe = NULL;            /* Marks the top level */  
507    
508  /* Copy in the original argument variables */  /* Copy in the original argument variables */
509    
510  frame->Xeptr = eptr;  frame->Xeptr = eptr;
511  frame->Xecode = ecode;  frame->Xecode = ecode;
512  frame->Xmstart = mstart;  frame->Xmstart = mstart;
 frame->Xmarkptr = markptr;  
513  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
514  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
515  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
516    
517  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 514  HEAP_RECURSE: Line 523  HEAP_RECURSE:
523  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
524  #define ecode              frame->Xecode  #define ecode              frame->Xecode
525  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
 #define markptr            frame->Xmarkptr  
526  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
527  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
528  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
529    
530  /* Ditto for the local variables */  /* Ditto for the local variables */
531    
532  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
533  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
534  #endif  #endif
535  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 540  HEAP_RECURSE: Line 546  HEAP_RECURSE:
546  #define condition          frame->Xcondition  #define condition          frame->Xcondition
547  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
548    
 #define original_ims       frame->Xoriginal_ims  
   
549  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
550  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
551  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
552  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
 #define prop_category      frame->Xprop_category  
 #define prop_chartype      frame->Xprop_chartype  
 #define prop_script        frame->Xprop_script  
553  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
554  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
555  #endif  #endif
# Line 578  i, and fc and c, can be the same variabl Line 579  i, and fc and c, can be the same variabl
579  #define fi i  #define fi i
580  #define fc c  #define fc c
581    
582    /* Many of the following variables are used only in small blocks of the code.
583    My normal style of coding would have declared them within each of those blocks.
584    However, in order to accommodate the version of this code that uses an external
585    "stack" implemented on the heap, it is easier to declare them all here, so the
586    declarations can be cut out in a block. The only declarations within blocks
587    below are for variables that do not have to be preserved over a recursive call
588    to RMATCH(). */
589    
590    #ifdef SUPPORT_UTF
591    const pcre_uchar *charptr;
592    #endif
593    const pcre_uchar *callpat;
594    const pcre_uchar *data;
595    const pcre_uchar *next;
596    PCRE_PUCHAR       pp;
597    const pcre_uchar *prev;
598    PCRE_PUCHAR       saved_eptr;
599    
600  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  recursion_info new_recursive;
601  const uschar *charptr;             /* in small blocks of the code. My normal */  
602  #endif                             /* style of coding would have declared    */  BOOL cur_is_word;
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
603  BOOL condition;  BOOL condition;
604  BOOL prev_is_word;  BOOL prev_is_word;
605    
 unsigned long int original_ims;  
   
606  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
607  int prop_type;  int prop_type;
608  int prop_value;  int prop_value;
609  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
610  int oclength;  int oclength;
611  uschar occhars[8];  pcre_uchar occhars[6];
612  #endif  #endif
613    
614  int codelink;  int codelink;
# Line 621  int save_offset1, save_offset2, save_off Line 624  int save_offset1, save_offset2, save_off
624  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
625    
626  eptrblock newptrb;  eptrblock newptrb;
627    
628    /* There is a special fudge for calling match() in a way that causes it to
629    measure the size of its basic stack frame when the stack is being used for
630    recursion. The second argument (ecode) being NULL triggers this behaviour. It
631    cannot normally ever be NULL. The return is the negated value of the frame
632    size. */
633    
634    if (ecode == NULL)
635      {
636      if (rdepth == 0)
637        return match((PCRE_PUCHAR)&rdepth, NULL, NULL, 0, NULL, NULL, 1);
638      else
639        {
640        int len = (char *)&rdepth - (char *)eptr;
641        return (len > 0)? -len : len;
642        }
643      }
644  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
645    
646    /* To save space on the stack and in the heap frame, I have doubled up on some
647    of the local variables that are used only in localised parts of the code, but
648    still need to be preserved over recursive calls of match(). These macros define
649    the alternative names that are used. */
650    
651    #define allow_zero    cur_is_word
652    #define cbegroup      condition
653    #define code_offset   codelink
654    #define condassert    condition
655    #define matched_once  prev_is_word
656    #define foc           number
657    #define save_mark     data
658    
659  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
660  variables. */  variables. */
661    
# Line 647  defined). However, RMATCH isn't like a f Line 680  defined). However, RMATCH isn't like a f
680  complicated macro. It has to be used in one particular way. This shouldn't,  complicated macro. It has to be used in one particular way. This shouldn't,
681  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
682    
683  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
684  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
685  #else  #else
686  utf8 = FALSE;  utf = FALSE;
687  #endif  #endif
688    
689  /* First check that we haven't called match() too many times, or that we  /* First check that we haven't called match() too many times, or that we
# Line 659  haven't exceeded the recursive call limi Line 692  haven't exceeded the recursive call limi
692  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
693  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
694    
 original_ims = ims;    /* Save for resetting on ')' */  
   
695  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
696  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
697  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
698  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
699  When match() is called in other circumstances, don't add to the chain. The  
700  match_cbegroup flag must NOT be used with tail recursion, because the memory  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
701  block that is used is on the stack, so a new one may be required for each  such remembered pointers, to be checked when we hit the closing ket, in order
702  match(). */  to break infinite loops that match no characters. When match() is called in
703    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
704    NOT be used with tail recursion, because the memory block that is used is on
705    the stack, so a new one may be required for each match(). */
706    
707  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
708    {    {
709    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
710    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
711    eptrb = &newptrb;    eptrb = &newptrb;
712      md->match_function_type = 0;
713    }    }
714    
715  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 687  for (;;) Line 722  for (;;)
722    switch(op)    switch(op)
723      {      {
724      case OP_MARK:      case OP_MARK:
725      markptr = ecode + 2;      md->nomatch_mark = ecode + 2;
726      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->mark = NULL;    /* In case previously set by assertion */
727        ims, eptrb, flags, RM55);      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
728          eptrb, RM55);
729        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
730             md->mark == NULL) md->mark = ecode + 2;
731    
732      /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an      /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
733      argument, and we must check whether that argument matches this MARK's      argument, and we must check whether that argument matches this MARK's
# Line 698  for (;;) Line 736  for (;;)
736      position and return MATCH_SKIP. Otherwise, pass back the return code      position and return MATCH_SKIP. Otherwise, pass back the return code
737      unaltered. */      unaltered. */
738    
739      if (rrc == MATCH_SKIP_ARG &&      else if (rrc == MATCH_SKIP_ARG &&
740          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)          STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0)
741        {        {
742        md->start_match_ptr = eptr;        md->start_match_ptr = eptr;
743        RRETURN(MATCH_SKIP);        RRETURN(MATCH_SKIP);
744        }        }
   
     if (md->mark == NULL) md->mark = markptr;  
745      RRETURN(rrc);      RRETURN(rrc);
746    
747      case OP_FAIL:      case OP_FAIL:
748      MRRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
749    
750      /* COMMIT overrides PRUNE, SKIP, and THEN */      /* COMMIT overrides PRUNE, SKIP, and THEN */
751    
752      case OP_COMMIT:      case OP_COMMIT:
753      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
754        ims, eptrb, flags, RM52);        eptrb, RM52);
755      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
756          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
757          rrc != MATCH_THEN)          rrc != MATCH_THEN)
758        RRETURN(rrc);        RRETURN(rrc);
759      MRRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
760    
761      /* PRUNE overrides THEN */      /* PRUNE overrides THEN */
762    
763      case OP_PRUNE:      case OP_PRUNE:
764      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
765        ims, eptrb, flags, RM51);        eptrb, RM51);
766      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
767      MRRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
768    
769      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
770      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->nomatch_mark = ecode + 2;
771        ims, eptrb, flags, RM56);      md->mark = NULL;    /* In case previously set by assertion */
772        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
773          eptrb, RM56);
774        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
775             md->mark == NULL) md->mark = ecode + 2;
776      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
     md->mark = ecode + 2;  
777      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
778    
779      /* SKIP overrides PRUNE and THEN */      /* SKIP overrides PRUNE and THEN */
780    
781      case OP_SKIP:      case OP_SKIP:
782      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
783        ims, eptrb, flags, RM53);        eptrb, RM53);
784      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
785        RRETURN(rrc);        RRETURN(rrc);
786      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
787      MRRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
788    
789        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
790        nomatch_mark. There is a flag that disables this opcode when re-matching a
791        pattern that ended with a SKIP for which there was not a matching MARK. */
792    
793      case OP_SKIP_ARG:      case OP_SKIP_ARG:
794      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      if (md->ignore_skip_arg)
795        ims, eptrb, flags, RM57);        {
796          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
797          break;
798          }
799        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
800          eptrb, RM57);
801      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
802        RRETURN(rrc);        RRETURN(rrc);
803    
804      /* Pass back the current skip name by overloading md->start_match_ptr and      /* Pass back the current skip name by overloading md->start_match_ptr and
805      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
806      caught by a matching MARK, or get to the top, where it is treated the same      caught by a matching MARK, or get to the top, where it causes a rematch
807      as PRUNE. */      with the md->ignore_skip_arg flag set. */
808    
809      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
810      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
811    
812      /* For THEN (and THEN_ARG) we pass back the address of the bracket or      /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
813      the alt that is at the start of the current branch. This makes it possible      the branch in which it occurs can be determined. Overload the start of
814      to skip back past alternatives that precede the THEN within the current      match pointer to do this. */
     branch. */  
815    
816      case OP_THEN:      case OP_THEN:
817      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
818        ims, eptrb, flags, RM54);        eptrb, RM54);
819      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
820      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
821      MRRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
822    
823      case OP_THEN_ARG:      case OP_THEN_ARG:
824      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      md->nomatch_mark = ecode + 2;
825        offset_top, md, ims, eptrb, flags, RM58);      md->mark = NULL;    /* In case previously set by assertion */
826        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
827          md, eptrb, RM58);
828        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
829             md->mark == NULL) md->mark = ecode + 2;
830      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
831      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
     md->mark = ecode + LINK_SIZE + 2;  
832      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
833    
834      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle an atomic group that does not contain any capturing parentheses.
835      the current subject position in the working slot at the top of the vector.      This can be handled like an assertion. Prior to 8.13, all atomic groups
836      We mustn't change the current values of the data slot, because they may be      were handled this way. In 8.13, the code was changed as below for ONCE, so
837      set from a previous iteration of this group, and be referred to by a      that backups pass through the group and thereby reset captured values.
838      reference inside the group.      However, this uses a lot more stack, so in 8.20, atomic groups that do not
839        contain any captures generate OP_ONCE_NC, which can be handled in the old,
840      If the bracket fails to match, we need to restore this value and also the      less stack intensive way.
841      values of the final offsets, in case they were set by a previous iteration  
842      of the same bracket.      Check the alternative branches in turn - the matching won't pass the KET
843        for this kind of subpattern. If any one branch matches, we carry on as at
844        the end of a normal bracket, leaving the subject pointer, but resetting
845        the start-of-match value in case it was changed by \K. */
846    
847        case OP_ONCE_NC:
848        prev = ecode;
849        saved_eptr = eptr;
850        save_mark = md->mark;
851        do
852          {
853          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
854          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
855            {
856            mstart = md->start_match_ptr;
857            break;
858            }
859          if (rrc == MATCH_THEN)
860            {
861            next = ecode + GET(ecode,1);
862            if (md->start_match_ptr < next &&
863                (*ecode == OP_ALT || *next == OP_ALT))
864              rrc = MATCH_NOMATCH;
865            }
866    
867          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
868          ecode += GET(ecode,1);
869          md->mark = save_mark;
870          }
871        while (*ecode == OP_ALT);
872    
873        /* If hit the end of the group (which could be repeated), fail */
874    
875        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
876    
877        /* Continue as from after the group, updating the offsets high water
878        mark, since extracts may have been taken. */
879    
880        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
881    
882        offset_top = md->end_offset_top;
883        eptr = md->end_match_ptr;
884    
885        /* For a non-repeating ket, just continue at this level. This also
886        happens for a repeating ket if no characters were matched in the group.
887        This is the forcible breaking of infinite loops as implemented in Perl
888        5.005. */
889    
890        if (*ecode == OP_KET || eptr == saved_eptr)
891          {
892          ecode += 1+LINK_SIZE;
893          break;
894          }
895    
896        /* The repeating kets try the rest of the pattern or restart from the
897        preceding bracket, in the appropriate order. The second "call" of match()
898        uses tail recursion, to avoid using another stack frame. */
899    
900        if (*ecode == OP_KETRMIN)
901          {
902          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
903          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
904          ecode = prev;
905          goto TAIL_RECURSE;
906          }
907        else  /* OP_KETRMAX */
908          {
909          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
910          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
911          ecode += 1 + LINK_SIZE;
912          goto TAIL_RECURSE;
913          }
914        /* Control never gets here */
915    
916        /* Handle a capturing bracket, other than those that are possessive with an
917        unlimited repeat. If there is space in the offset vector, save the current
918        subject position in the working slot at the top of the vector. We mustn't
919        change the current values of the data slot, because they may be set from a
920        previous iteration of this group, and be referred to by a reference inside
921        the group. A failure to match might occur after the group has succeeded,
922        if something later on doesn't match. For this reason, we need to restore
923        the working value and also the values of the final offsets, in case they
924        were set by a previous iteration of the same bracket.
925    
926      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
927      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 813  for (;;) Line 945  for (;;)
945        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
946        save_offset3 = md->offset_vector[md->offset_end - number];        save_offset3 = md->offset_vector[md->offset_end - number];
947        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
948          save_mark = md->mark;
949    
950        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
951        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
952          (int)(eptr - md->start_subject);          (int)(eptr - md->start_subject);
953    
954        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
955          {          {
956          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
957            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
958          if (rrc != MATCH_NOMATCH &&            eptrb, RM1);
959              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
960            RRETURN(rrc);  
961            /* If we backed up to a THEN, check whether it is within the current
962            branch by comparing the address of the THEN that is passed back with
963            the end of the branch. If it is within the current branch, and the
964            branch is one of two or more alternatives (it either starts or ends
965            with OP_ALT), we have reached the limit of THEN's action, so convert
966            the return code to NOMATCH, which will cause normal backtracking to
967            happen from now on. Otherwise, THEN is passed back to an outer
968            alternative. This implements Perl's treatment of parenthesized groups,
969            where a group not containing | does not affect the current alternative,
970            that is, (X) is NOT the same as (X|(*F)). */
971    
972            if (rrc == MATCH_THEN)
973              {
974              next = ecode + GET(ecode,1);
975              if (md->start_match_ptr < next &&
976                  (*ecode == OP_ALT || *next == OP_ALT))
977                rrc = MATCH_NOMATCH;
978              }
979    
980            /* Anything other than NOMATCH is passed back. */
981    
982            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
983          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
984          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
985            md->mark = save_mark;
986            if (*ecode != OP_ALT) break;
987          }          }
       while (*ecode == OP_ALT);  
988    
989        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
990        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
991        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
992        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
993    
994        if (rrc != MATCH_THEN) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
995        RRETURN(MATCH_NOMATCH);  
996          RRETURN(rrc);
997        }        }
998    
999      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 852  for (;;) Line 1007  for (;;)
1007      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1009    
1010      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
1011      final alternative within the brackets, we would return the result of a      repeat and ONCE group with no captures. Loop for all the alternatives.
1012      recursive call to match() whatever happened. We can reduce stack usage by  
1013      turning this into a tail recursion, except in the case when match_cbegroup      When we get to the final alternative within the brackets, we used to return
1014      is set.*/      the result of a recursive call to match() whatever happened so it was
1015        possible to reduce stack usage by turning this into a tail recursion,
1016        except in the case of a possibly empty group. However, now that there is
1017        the possiblity of (*THEN) occurring in the final alternative, this
1018        optimization is no longer always possible.
1019    
1020        We can optimize if we know there are no (*THEN)s in the pattern; at present
1021        this is the best that can be done.
1022    
1023        MATCH_ONCE is returned when the end of an atomic group is successfully
1024        reached, but subsequent matching fails. It passes back up the tree (causing
1025        captured values to be reset) until the original atomic group level is
1026        reached. This is tested by comparing md->once_target with the start of the
1027        group. At this point, the return is converted into MATCH_NOMATCH so that
1028        previous backup points can be taken. */
1029    
1030        case OP_ONCE:
1031      case OP_BRA:      case OP_BRA:
1032      case OP_SBRA:      case OP_SBRA:
1033      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
1034      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
1035      for (;;)      for (;;)
1036        {        {
1037        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE)
1038            md->match_function_type = MATCH_CBEGROUP;
1039    
1040          /* If this is not a possibly empty group, and there are no (*THEN)s in
1041          the pattern, and this is the final alternative, optimize as described
1042          above. */
1043    
1044          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1045          {          {
1046          if (flags == 0)    /* Not a possibly empty group */          ecode += PRIV(OP_lengths)[*ecode];
1047            goto TAIL_RECURSE;
1048            }
1049    
1050          /* In all other cases, we have to make another call to match(). */
1051    
1052          save_mark = md->mark;
1053          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1054            RM2);
1055    
1056          /* See comment in the code for capturing groups above about handling
1057          THEN. */
1058    
1059          if (rrc == MATCH_THEN)
1060            {
1061            next = ecode + GET(ecode,1);
1062            if (md->start_match_ptr < next &&
1063                (*ecode == OP_ALT || *next == OP_ALT))
1064              rrc = MATCH_NOMATCH;
1065            }
1066    
1067          if (rrc != MATCH_NOMATCH)
1068            {
1069            if (rrc == MATCH_ONCE)
1070            {            {
1071            ecode += _pcre_OP_lengths[*ecode];            const pcre_uchar *scode = ecode;
1072            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1073            goto TAIL_RECURSE;              {
1074                while (*scode == OP_ALT) scode += GET(scode, 1);
1075                scode -= GET(scode, 1);
1076                }
1077              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1078            }            }
1079            RRETURN(rrc);
1080            }
1081          ecode += GET(ecode, 1);
1082          md->mark = save_mark;
1083          if (*ecode != OP_ALT) break;
1084          }
1085    
1086          /* Possibly empty group; can't use tail recursion. */      RRETURN(MATCH_NOMATCH);
1087    
1088          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* Handle possessive capturing brackets with an unlimited repeat. We come
1089            eptrb, flags, RM48);      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1090          if (rrc == MATCH_NOMATCH) md->mark = markptr;      handled similarly to the normal case above. However, the matching is
1091          RRETURN(rrc);      different. The end of these brackets will always be OP_KETRPOS, which
1092        returns MATCH_KETRPOS without going further in the pattern. By this means
1093        we can handle the group by iteration rather than recursion, thereby
1094        reducing the amount of stack needed. */
1095    
1096        case OP_CBRAPOS:
1097        case OP_SCBRAPOS:
1098        allow_zero = FALSE;
1099    
1100        POSSESSIVE_CAPTURE:
1101        number = GET2(ecode, 1+LINK_SIZE);
1102        offset = number << 1;
1103    
1104    #ifdef PCRE_DEBUG
1105        printf("start possessive bracket %d\n", number);
1106        printf("subject=");
1107        pchars(eptr, 16, TRUE, md);
1108        printf("\n");
1109    #endif
1110    
1111        if (offset < md->offset_max)
1112          {
1113          matched_once = FALSE;
1114          code_offset = (int)(ecode - md->start_code);
1115    
1116          save_offset1 = md->offset_vector[offset];
1117          save_offset2 = md->offset_vector[offset+1];
1118          save_offset3 = md->offset_vector[md->offset_end - number];
1119          save_capture_last = md->capture_last;
1120    
1121          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1122    
1123          /* Each time round the loop, save the current subject position for use
1124          when the group matches. For MATCH_MATCH, the group has matched, so we
1125          restart it with a new subject starting position, remembering that we had
1126          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1127          usual. If we haven't matched any alternatives in any iteration, check to
1128          see if a previous iteration matched. If so, the group has matched;
1129          continue from afterwards. Otherwise it has failed; restore the previous
1130          capture values before returning NOMATCH. */
1131    
1132          for (;;)
1133            {
1134            md->offset_vector[md->offset_end - number] =
1135              (int)(eptr - md->start_subject);
1136            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1137            RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1138              eptrb, RM63);
1139            if (rrc == MATCH_KETRPOS)
1140              {
1141              offset_top = md->end_offset_top;
1142              eptr = md->end_match_ptr;
1143              ecode = md->start_code + code_offset;
1144              save_capture_last = md->capture_last;
1145              matched_once = TRUE;
1146              continue;
1147              }
1148    
1149            /* See comment in the code for capturing groups above about handling
1150            THEN. */
1151    
1152            if (rrc == MATCH_THEN)
1153              {
1154              next = ecode + GET(ecode,1);
1155              if (md->start_match_ptr < next &&
1156                  (*ecode == OP_ALT || *next == OP_ALT))
1157                rrc = MATCH_NOMATCH;
1158              }
1159    
1160            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1161            md->capture_last = save_capture_last;
1162            ecode += GET(ecode, 1);
1163            if (*ecode != OP_ALT) break;
1164          }          }
1165    
1166        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1167        otherwise return. */          {
1168            md->offset_vector[offset] = save_offset1;
1169            md->offset_vector[offset+1] = save_offset2;
1170            md->offset_vector[md->offset_end - number] = save_offset3;
1171            }
1172    
1173        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (allow_zero || matched_once)
1174          eptrb, flags, RM2);          {
1175        if (rrc != MATCH_NOMATCH &&          ecode += 1 + LINK_SIZE;
1176            (rrc != MATCH_THEN || md->start_match_ptr != ecode))          break;
1177          RRETURN(rrc);          }
1178    
1179          RRETURN(MATCH_NOMATCH);
1180          }
1181    
1182        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1183        as a non-capturing bracket. */
1184    
1185        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1186        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1187    
1188        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1189    
1190        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1191        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1192    
1193        /* Non-capturing possessive bracket with unlimited repeat. We come here
1194        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1195        without the capturing complication. It is written out separately for speed
1196        and cleanliness. */
1197    
1198        case OP_BRAPOS:
1199        case OP_SBRAPOS:
1200        allow_zero = FALSE;
1201    
1202        POSSESSIVE_NON_CAPTURE:
1203        matched_once = FALSE;
1204        code_offset = (int)(ecode - md->start_code);
1205    
1206        for (;;)
1207          {
1208          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1209          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1210            eptrb, RM48);
1211          if (rrc == MATCH_KETRPOS)
1212            {
1213            offset_top = md->end_offset_top;
1214            eptr = md->end_match_ptr;
1215            ecode = md->start_code + code_offset;
1216            matched_once = TRUE;
1217            continue;
1218            }
1219    
1220          /* See comment in the code for capturing groups above about handling
1221          THEN. */
1222    
1223          if (rrc == MATCH_THEN)
1224            {
1225            next = ecode + GET(ecode,1);
1226            if (md->start_match_ptr < next &&
1227                (*ecode == OP_ALT || *next == OP_ALT))
1228              rrc = MATCH_NOMATCH;
1229            }
1230    
1231          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1232        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1233          if (*ecode != OP_ALT) break;
1234        }        }
1235    
1236        if (matched_once || allow_zero)
1237          {
1238          ecode += 1 + LINK_SIZE;
1239          break;
1240          }
1241        RRETURN(MATCH_NOMATCH);
1242    
1243      /* Control never reaches here. */      /* Control never reaches here. */
1244    
1245      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1246      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1247      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1248      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1249    
1250      case OP_COND:      case OP_COND:
1251      case OP_SCOND:      case OP_SCOND:
1252      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1253    
1254      /* Because of the way auto-callout works during compile, a callout item is      /* Because of the way auto-callout works during compile, a callout item is
1255      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
1256    
1257      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1258        {        {
1259        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1260          {          {
1261          pcre_callout_block cb;          PUBL(callout_block) cb;
1262          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1263          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1264          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1265    #ifdef COMPILE_PCRE8
1266          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1267    #else
1268            cb.subject          = (PCRE_SPTR16)md->start_subject;
1269    #endif
1270          cb.subject_length   = (int)(md->end_subject - md->start_subject);          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1271          cb.start_match      = (int)(mstart - md->start_subject);          cb.start_match      = (int)(mstart - md->start_subject);
1272          cb.current_position = (int)(eptr - md->start_subject);          cb.current_position = (int)(eptr - md->start_subject);
# Line 923  for (;;) Line 1275  for (;;)
1275          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1276          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1277          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1278          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1279            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1280          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1281          }          }
1282        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1283        }        }
1284    
1285      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
# Line 943  for (;;) Line 1296  for (;;)
1296        else        else
1297          {          {
1298          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1299          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1300    
1301          /* If the test is for recursion into a specific subpattern, and it is          /* If the test is for recursion into a specific subpattern, and it is
1302          false, but the test was set up by name, scan the table to see if the          false, but the test was set up by name, scan the table to see if the
1303          name refers to any other numbers, and test them. The condition is true          name refers to any other numbers, and test them. The condition is true
1304          if any one is set. */          if any one is set. */
1305    
1306          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1307            {            {
1308            uschar *slotA = md->name_table;            pcre_uchar *slotA = md->name_table;
1309            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
1310              {              {
1311              if (GET2(slotA, 0) == recno) break;              if (GET2(slotA, 0) == recno) break;
# Line 965  for (;;) Line 1318  for (;;)
1318    
1319            if (i < md->name_count)            if (i < md->name_count)
1320              {              {
1321              uschar *slotB = slotA;              pcre_uchar *slotB = slotA;
1322              while (slotB > md->name_table)              while (slotB > md->name_table)
1323                {                {
1324                slotB -= md->name_entry_size;                slotB -= md->name_entry_size;
1325                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1326                  {                  {
1327                  condition = GET2(slotB, 0) == md->recursive->group_num;                  condition = GET2(slotB, 0) == md->recursive->group_num;
1328                  if (condition) break;                  if (condition) break;
# Line 985  for (;;) Line 1338  for (;;)
1338                for (i++; i < md->name_count; i++)                for (i++; i < md->name_count; i++)
1339                  {                  {
1340                  slotB += md->name_entry_size;                  slotB += md->name_entry_size;
1341                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1342                    {                    {
1343                    condition = GET2(slotB, 0) == md->recursive->group_num;                    condition = GET2(slotB, 0) == md->recursive->group_num;
1344                    if (condition) break;                    if (condition) break;
# Line 998  for (;;) Line 1351  for (;;)
1351    
1352          /* Chose branch according to the condition */          /* Chose branch according to the condition */
1353    
1354          ecode += condition? 3 : GET(ecode, 1);          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1355          }          }
1356        }        }
1357    
# Line 1015  for (;;) Line 1368  for (;;)
1368        if (!condition && condcode == OP_NCREF)        if (!condition && condcode == OP_NCREF)
1369          {          {
1370          int refno = offset >> 1;          int refno = offset >> 1;
1371          uschar *slotA = md->name_table;          pcre_uchar *slotA = md->name_table;
1372    
1373          for (i = 0; i < md->name_count; i++)          for (i = 0; i < md->name_count; i++)
1374            {            {
# Line 1029  for (;;) Line 1382  for (;;)
1382    
1383          if (i < md->name_count)          if (i < md->name_count)
1384            {            {
1385            uschar *slotB = slotA;            pcre_uchar *slotB = slotA;
1386            while (slotB > md->name_table)            while (slotB > md->name_table)
1387              {              {
1388              slotB -= md->name_entry_size;              slotB -= md->name_entry_size;
1389              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)              if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1390                {                {
1391                offset = GET2(slotB, 0) << 1;                offset = GET2(slotB, 0) << 1;
1392                condition = offset < offset_top &&                condition = offset < offset_top &&
# Line 1051  for (;;) Line 1404  for (;;)
1404              for (i++; i < md->name_count; i++)              for (i++; i < md->name_count; i++)
1405                {                {
1406                slotB += md->name_entry_size;                slotB += md->name_entry_size;
1407                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1408                  {                  {
1409                  offset = GET2(slotB, 0) << 1;                  offset = GET2(slotB, 0) << 1;
1410                  condition = offset < offset_top &&                  condition = offset < offset_top &&
# Line 1066  for (;;) Line 1419  for (;;)
1419    
1420        /* Chose branch according to the condition */        /* Chose branch according to the condition */
1421    
1422        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1423        }        }
1424    
1425      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 1076  for (;;) Line 1429  for (;;)
1429        }        }
1430    
1431      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1432      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1433      assertion. */      an assertion. */
1434    
1435      else      else
1436        {        {
1437        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1438            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1439        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1440          {          {
1441            if (md->end_offset_top > offset_top)
1442              offset_top = md->end_offset_top;  /* Captures may have happened */
1443          condition = TRUE;          condition = TRUE;
1444          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1445          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1446          }          }
1447        else if (rrc != MATCH_NOMATCH &&  
1448                (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1449          assertion; it is therefore treated as NOMATCH. */
1450    
1451          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1452          {          {
1453          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1454          }          }
# Line 1101  for (;;) Line 1459  for (;;)
1459          }          }
1460        }        }
1461    
1462      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one, can
1463      we can use tail recursion to avoid using another stack frame, except when      use tail recursion to avoid using another stack frame, except when there is
1464      match_cbegroup is required for an unlimited repeat of a possibly empty      unlimited repeat of a possibly empty group. In the latter case, a recursive
1465      group. If the second alternative doesn't exist, we can just plough on. */      call to match() is always required, unless the second alternative doesn't
1466        exist, in which case we can just plough on. Note that, for compatibility
1467        with Perl, the | in a conditional group is NOT treated as creating two
1468        alternatives. If a THEN is encountered in the branch, it propagates out to
1469        the enclosing alternative (unless nested in a deeper set of alternatives,
1470        of course). */
1471    
1472      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1473        {        {
1474        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
         {  
         RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);  
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
1475          {          {
1476          flags = 0;          ecode += 1 + LINK_SIZE;
1477          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1478          }          }
1479    
1480          md->match_function_type = MATCH_CBEGROUP;
1481          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1482          RRETURN(rrc);
1483        }        }
1484      else                         /* Condition false & no alternative */  
1485         /* Condition false & no alternative; continue after the group. */
1486    
1487        else
1488        {        {
1489        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1490        }        }
# Line 1147  for (;;) Line 1511  for (;;)
1511        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1512        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1513        }        }
1514      ecode += 3;      ecode += 1 + IMM2_SIZE;
1515      break;      break;
1516    
1517    
1518      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. */
     recursion, we should restore the offsets appropriately and continue from  
     after the call. */  
1519    
     case OP_ACCEPT:  
1520      case OP_END:      case OP_END:
1521      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1522        {      case OP_ASSERT_ACCEPT:
       recursion_info *rec = md->recursive;  
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
       md->recursive = rec->prevrec;  
       memmove(md->offset_vector, rec->offset_save,  
         rec->saved_max * sizeof(int));  
       offset_top = rec->save_offset_top;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1523    
1524      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is      /* If we have matched an empty string, fail if not in an assertion and not
1525      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1526      the subject. In both cases, backtracking will then try other alternatives,      is set and we have matched at the start of the subject. In both cases,
1527      if any. */      backtracking will then try other alternatives, if any. */
1528    
1529      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1530          (md->notempty ||           md->recursive == NULL &&
1531            (md->notempty_atstart &&           (md->notempty ||
1532              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1533        MRRETURN(MATCH_NOMATCH);               mstart == md->start_subject + md->start_offset)))
1534          RRETURN(MATCH_NOMATCH);
1535    
1536      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1537    
# Line 1188  for (;;) Line 1540  for (;;)
1540      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1541    
1542      /* For some reason, the macros don't work properly if an expression is      /* For some reason, the macros don't work properly if an expression is
1543      given as the argument to MRRETURN when the heap is in use. */      given as the argument to RRETURN when the heap is in use. */
1544    
1545      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1546      MRRETURN(rrc);      RRETURN(rrc);
   
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1547    
1548      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1549      matching won't pass the KET for an assertion. If any one branch matches,      matching won't pass the KET for an assertion. If any one branch matches,
1550      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
1551      start of each branch to move the current point backwards, so the code at      start of each branch to move the current point backwards, so the code at
1552      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1553        of a condition, we want to return immediately afterwards. The caller of
1554        this incarnation of the match() function will have set MATCH_CONDASSERT in
1555        md->match_function type, and one of these opcodes will be the first opcode
1556        that is processed. We use a local variable that is preserved over calls to
1557        match() to remember this case. */
1558    
1559      case OP_ASSERT:      case OP_ASSERT:
1560      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1561        save_mark = md->mark;
1562        if (md->match_function_type == MATCH_CONDASSERT)
1563          {
1564          condassert = TRUE;
1565          md->match_function_type = 0;
1566          }
1567        else condassert = FALSE;
1568    
1569      do      do
1570        {        {
1571        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1572        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1573          {          {
1574          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1575          break;          break;
1576          }          }
1577        if (rrc != MATCH_NOMATCH &&        md->mark = save_mark;
1578            (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
1579          RRETURN(rrc);        /* A COMMIT failure must fail the entire assertion, without trying any
1580          subsequent branches. */
1581    
1582          if (rrc == MATCH_COMMIT) RRETURN(MATCH_NOMATCH);
1583    
1584          /* PCRE does not allow THEN to escape beyond an assertion; it
1585          is treated as NOMATCH. */
1586    
1587          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1588        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1589        }        }
1590      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1591      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);  
1592        if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1593    
1594      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1595    
1596      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1597    
1598      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1599      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1244  for (;;) Line 1609  for (;;)
1609    
1610      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1611      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1612        save_mark = md->mark;
1613        if (md->match_function_type == MATCH_CONDASSERT)
1614          {
1615          condassert = TRUE;
1616          md->match_function_type = 0;
1617          }
1618        else condassert = FALSE;
1619    
1620      do      do
1621        {        {
1622        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1623          RM5);        md->mark = save_mark;
1624        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1625        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1626          {          {
1627          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1628          break;          break;
1629          }          }
1630        if (rrc != MATCH_NOMATCH &&  
1631            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1632          RRETURN(rrc);        as NOMATCH. */
1633    
1634          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1635        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1636        }        }
1637      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1638    
1639      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1640    
1641      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1642      continue;      continue;
# Line 1272  for (;;) Line 1647  for (;;)
1647      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1648    
1649      case OP_REVERSE:      case OP_REVERSE:
1650  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1651      if (utf8)      if (utf)
1652        {        {
1653        i = GET(ecode, 1);        i = GET(ecode, 1);
1654        while (i-- > 0)        while (i-- > 0)
1655          {          {
1656          eptr--;          eptr--;
1657          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1658          BACKCHAR(eptr);          BACKCHAR(eptr);
1659          }          }
1660        }        }
# Line 1290  for (;;) Line 1665  for (;;)
1665    
1666        {        {
1667        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1668        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1669        }        }
1670    
1671      /* Save the earliest consulted character, then skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
# Line 1304  for (;;) Line 1679  for (;;)
1679      function is able to force a failure. */      function is able to force a failure. */
1680    
1681      case OP_CALLOUT:      case OP_CALLOUT:
1682      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1683        {        {
1684        pcre_callout_block cb;        PUBL(callout_block) cb;
1685        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1686        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1687        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1688    #ifdef COMPILE_PCRE8
1689        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1690    #else
1691          cb.subject          = (PCRE_SPTR16)md->start_subject;
1692    #endif
1693        cb.subject_length   = (int)(md->end_subject - md->start_subject);        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1694        cb.start_match      = (int)(mstart - md->start_subject);        cb.start_match      = (int)(mstart - md->start_subject);
1695        cb.current_position = (int)(eptr - md->start_subject);        cb.current_position = (int)(eptr - md->start_subject);
# Line 1319  for (;;) Line 1698  for (;;)
1698        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1699        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1700        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1701        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1702          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1703        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1704        }        }
1705      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1329  for (;;) Line 1709  for (;;)
1709      offset data is the offset to the starting bracket from the start of the      offset data is the offset to the starting bracket from the start of the
1710      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1711    
1712      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1713      save their starting points and reinstate them after the recursion. However,      re-instated afterwards. We don't know how many are started and not yet
1714      we don't know how many such there are (offset_top records the completed      finished (offset_top records the completed total) so we just have to save
1715      total) so we just have to save all the potential data. There may be up to      all the potential data. There may be up to 65535 such values, which is too
1716      65535 such values, which is too large to put on the stack, but using malloc      large to put on the stack, but using malloc for small numbers seems
1717      for small numbers seems expensive. As a compromise, the stack is used when      expensive. As a compromise, the stack is used when there are no more than
1718      there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc      REC_STACK_SAVE_MAX values to store; otherwise malloc is used.
     is used. A problem is what to do if the malloc fails ... there is no way of  
     returning to the top level with an error. Save the top REC_STACK_SAVE_MAX  
     values on the stack, and accept that the rest may be wrong.  
1719    
1720      There are also other values that have to be saved. We use a chained      There are also other values that have to be saved. We use a chained
1721      sequence of blocks that actually live on the stack. Thanks to Robin Houston      sequence of blocks that actually live on the stack. Thanks to Robin Houston
1722      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1723        a lot, so he is not to blame for the current way it works. */
1724    
1725      case OP_RECURSE:      case OP_RECURSE:
1726        {        {
1727          recursion_info *ri;
1728          int recno;
1729    
1730        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1731        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1732          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1733    
1734          /* Check for repeating a recursion without advancing the subject pointer.
1735          This should catch convoluted mutual recursions. (Some simple cases are
1736          caught at compile time.) */
1737    
1738          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1739            if (recno == ri->group_num && eptr == ri->subject_position)
1740              RRETURN(PCRE_ERROR_RECURSELOOP);
1741    
1742        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1743    
1744          new_recursive.group_num = recno;
1745          new_recursive.subject_position = eptr;
1746        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1747        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1748    
1749        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1750    
1751        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1752    
1753        /* Now save the offset data. */        /* Now save the offset data */
1754    
1755        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1756        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1368  for (;;) Line 1758  for (;;)
1758        else        else
1759          {          {
1760          new_recursive.offset_save =          new_recursive.offset_save =
1761            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1762          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1763          }          }
   
1764        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1765              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1766    
1767        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1768        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1769          might be changed, so reset it before looping. */
1770    
1771        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1772        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1773        do        do
1774          {          {
1775          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1776            md, ims, eptrb, flags, RM6);          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1777              md, eptrb, RM6);
1778            memcpy(md->offset_vector, new_recursive.offset_save,
1779                new_recursive.saved_max * sizeof(int));
1780            md->recursive = new_recursive.prevrec;
1781          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1782            {            {
1783            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1784            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1785              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1786            MRRETURN(MATCH_MATCH);  
1787              /* Set where we got to in the subject, and reset the start in case
1788              it was changed by \K. This *is* propagated back out of a recursion,
1789              for Perl compatibility. */
1790    
1791              eptr = md->end_match_ptr;
1792              mstart = md->start_match_ptr;
1793              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1794            }            }
1795          else if (rrc != MATCH_NOMATCH &&  
1796                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* PCRE does not allow THEN or COMMIT to escape beyond a recursion; it
1797            is treated as NOMATCH. */
1798    
1799            else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN &&
1800                     rrc != MATCH_COMMIT)
1801            {            {
1802            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1803            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1804              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1805            RRETURN(rrc);            RRETURN(rrc);
1806            }            }
1807    
1808          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1809          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1810          }          }
1811        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1412  for (;;) Line 1813  for (;;)
1813        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1814        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1815        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1816          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1817        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
       }  
     /* 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, but resetting  
     the start-of-match value in case it was changed by \K. */  
   
     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)  /* Note: _not_ MATCH_ACCEPT */  
         {  
         mstart = md->start_match_ptr;  
         break;  
         }  
       if (rrc != MATCH_NOMATCH &&  
           (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
         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));  
1818        }        }
1819    
1820      if (*ecode == OP_KETRMIN)      RECURSION_MATCHED:
1821        {      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 */  
1822    
1823      /* 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
1824      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1512  for (;;) Line 1834  for (;;)
1834      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1835    
1836      case OP_BRAZERO:      case OP_BRAZERO:
1837        {      next = ecode + 1;
1838        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1839        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1840        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1841        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1842      break;      break;
1843    
1844      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1845        {      next = ecode + 1;
1846        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1847        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1848        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1849        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1850      break;      break;
1851    
1852      case OP_SKIPZERO:      case OP_SKIPZERO:
1853        {      next = ecode+1;
1854        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1855        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1856      break;      break;
1857    
1858        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1859        here; just jump to the group, with allow_zero set TRUE. */
1860    
1861        case OP_BRAPOSZERO:
1862        op = *(++ecode);
1863        allow_zero = TRUE;
1864        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1865          goto POSSESSIVE_NON_CAPTURE;
1866    
1867      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1868    
1869      case OP_KET:      case OP_KET:
1870      case OP_KETRMIN:      case OP_KETRMIN:
1871      case OP_KETRMAX:      case OP_KETRMAX:
1872        case OP_KETRPOS:
1873      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1874    
1875      /* If this was a group that remembered the subject start, in order to break      /* If this was a group that remembered the subject start, in order to break
1876      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1877      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1878    
1879      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1880        {        {
1881        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1882        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1883        }        }
1884      else saved_eptr = NULL;      else saved_eptr = NULL;
1885    
1886      /* If we are at the end of an assertion group or an atomic group, stop      /* If we are at the end of an assertion group or a non-capturing atomic
1887      matching and return MATCH_MATCH, but record the current high water mark for      group, stop matching and return MATCH_MATCH, but record the current high
1888      use by positive assertions. We also need to record the match start in case      water mark for use by positive assertions. We also need to record the match
1889      it was changed by \K. */      start in case it was changed by \K. */
1890    
1891      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1892          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||           *prev == OP_ONCE_NC)
         *prev == OP_ONCE)  
1893        {        {
1894        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1895        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1896        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1897        MRRETURN(MATCH_MATCH);        RRETURN(MATCH_MATCH);         /* Sets md->mark */
1898        }        }
1899    
1900      /* For capturing groups we have to check the group number back at the start      /* For capturing groups we have to check the group number back at the start
1901      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1902      bumping the high water mark. Note that whole-pattern recursion is coded as      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1903      a recurse into group 0, so it won't be picked up here. Instead, we catch it      into group 0, so it won't be picked up here. Instead, we catch it when the
1904      when the OP_END is reached. Other recursion is handled here. */      OP_END is reached. Other recursion is handled here. We just have to record
1905        the current subject position and start match pointer and give a MATCH
1906        return. */
1907    
1908      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1909            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1910        {        {
1911        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1912        offset = number << 1;        offset = number << 1;
# Line 1588  for (;;) Line 1916  for (;;)
1916        printf("\n");        printf("\n");
1917  #endif  #endif
1918    
1919          /* Handle a recursively called group. */
1920    
1921          if (md->recursive != NULL && md->recursive->group_num == number)
1922            {
1923            md->end_match_ptr = eptr;
1924            md->start_match_ptr = mstart;
1925            RRETURN(MATCH_MATCH);
1926            }
1927    
1928          /* Deal with capturing */
1929    
1930        md->capture_last = number;        md->capture_last = number;
1931        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1932          {          {
1933            /* If offset is greater than offset_top, it means that we are
1934            "skipping" a capturing group, and that group's offsets must be marked
1935            unset. In earlier versions of PCRE, all the offsets were unset at the
1936            start of matching, but this doesn't work because atomic groups and
1937            assertions can cause a value to be set that should later be unset.
1938            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1939            part of the atomic group, but this is not on the final matching path,
1940            so must be unset when 2 is set. (If there is no group 2, there is no
1941            problem, because offset_top will then be 2, indicating no capture.) */
1942    
1943            if (offset > offset_top)
1944              {
1945              register int *iptr = md->offset_vector + offset_top;
1946              register int *iend = md->offset_vector + offset;
1947              while (iptr < iend) *iptr++ = -1;
1948              }
1949    
1950            /* Now make the extraction */
1951    
1952          md->offset_vector[offset] =          md->offset_vector[offset] =
1953            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1954          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1955          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1956          }          }
   
       /* Handle a recursively called group. Restore the offsets  
       appropriately and continue from after the call. */  
   
       if (md->recursive != NULL && md->recursive->group_num == number)  
         {  
         recursion_info *rec = md->recursive;  
         DPRINTF(("Recursion (%d) succeeded - continuing\n", number));  
         md->recursive = rec->prevrec;  
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         offset_top = rec->save_offset_top;  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
         }  
1957        }        }
1958    
1959      /* For both capturing and non-capturing groups, reset the value of the ims      /* For an ordinary non-repeating ket, just continue at this level. This
1960      flags, in case they got changed during the group. */      also happens for a repeating ket if no characters were matched in the
1961        group. This is the forcible breaking of infinite loops as implemented in
1962      ims = original_ims;      Perl 5.005. For a non-repeating atomic group that includes captures,
1963      DPRINTF(("ims reset to %02lx\n", ims));      establish a backup point by processing the rest of the pattern at a lower
1964        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1965      /* For a non-repeating ket, just continue at this level. This also      original OP_ONCE level, thereby bypassing intermediate backup points, but
1966      happens for a repeating ket if no characters were matched in the group.      resetting any captures that happened along the way. */
     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. */  
1967    
1968      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1969        {        {
1970        ecode += 1 + LINK_SIZE;        if (*prev == OP_ONCE)
1971            {
1972            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1973            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1974            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1975            RRETURN(MATCH_ONCE);
1976            }
1977          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1978        break;        break;
1979        }        }
1980    
1981      /* The repeating kets try the rest of the pattern or restart from the      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1982      preceding bracket, in the appropriate order. In the second case, we can use      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1983      tail recursion to avoid using another stack frame, unless we have an      at a time from the outer level, thus saving stack. */
1984      unlimited repeat of a group that can match an empty string. */  
1985        if (*ecode == OP_KETRPOS)
1986          {
1987          md->end_match_ptr = eptr;
1988          md->end_offset_top = offset_top;
1989          RRETURN(MATCH_KETRPOS);
1990          }
1991    
1992      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;      /* The normal repeating kets try the rest of the pattern or restart from
1993        the preceding bracket, in the appropriate order. In the second case, we can
1994        use tail recursion to avoid using another stack frame, unless we have an
1995        an atomic group or an unlimited repeat of a group that can match an empty
1996        string. */
1997    
1998      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1999        {        {
2000        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
2001        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2002        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
2003            {
2004            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
2005            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2006            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2007            RRETURN(MATCH_ONCE);
2008            }
2009          if (*prev >= OP_SBRA)    /* Could match an empty string */
2010          {          {
2011          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
2012          RRETURN(rrc);          RRETURN(rrc);
2013          }          }
2014        ecode = prev;        ecode = prev;
# Line 1653  for (;;) Line 2016  for (;;)
2016        }        }
2017      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
2018        {        {
2019        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
2020          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
2021        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2022          if (*prev == OP_ONCE)
2023            {
2024            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
2025            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2026            md->once_target = prev;
2027            RRETURN(MATCH_ONCE);
2028            }
2029        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
2030        goto TAIL_RECURSE;        goto TAIL_RECURSE;
2031        }        }
2032      /* Control never gets here */      /* Control never gets here */
2033    
2034      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
2035    
2036      case OP_CIRC:      case OP_CIRC:
2037      if (md->notbol && eptr == md->start_subject) MRRETURN(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)))  
         MRRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
2038    
2039      /* Start of subject assertion */      /* Start of subject assertion */
2040    
2041      case OP_SOD:      case OP_SOD:
2042      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);
2043        ecode++;
2044        break;
2045    
2046        /* Multiline mode: start of subject unless notbol, or after any newline. */
2047    
2048        case OP_CIRCM:
2049        if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2050        if (eptr != md->start_subject &&
2051            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
2052          RRETURN(MATCH_NOMATCH);
2053      ecode++;      ecode++;
2054      break;      break;
2055    
2056      /* Start of match assertion */      /* Start of match assertion */
2057    
2058      case OP_SOM:      case OP_SOM:
2059      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);
2060      ecode++;      ecode++;
2061      break;      break;
2062    
# Line 1696  for (;;) Line 2067  for (;;)
2067      ecode++;      ecode++;
2068      break;      break;
2069    
2070      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2071      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2072    
2073      case OP_DOLL:      case OP_DOLLM:
2074      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2075        {        {
2076        if (eptr < md->end_subject)        if (!IS_NEWLINE(eptr))
2077          { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }          {
2078        else          if (md->partial != 0 &&
2079          {              eptr + 1 >= md->end_subject &&
2080          if (md->noteol) MRRETURN(MATCH_NOMATCH);              NLBLOCK->nltype == NLTYPE_FIXED &&
2081          SCHECK_PARTIAL();              NLBLOCK->nllen == 2 &&
2082                *eptr == NLBLOCK->nl[0])
2083              {
2084              md->hitend = TRUE;
2085              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2086              }
2087            RRETURN(MATCH_NOMATCH);
2088          }          }
       ecode++;  
       break;  
2089        }        }
2090      else  /* Not multiline */      else
2091        {        {
2092        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2093        if (!md->endonly) goto ASSERT_NL_OR_EOS;        SCHECK_PARTIAL();
2094        }        }
2095        ecode++;
2096        break;
2097    
2098        /* Not multiline mode: assert before a terminating newline or before end of
2099        subject unless noteol is set. */
2100    
2101        case OP_DOLL:
2102        if (md->noteol) RRETURN(MATCH_NOMATCH);
2103        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2104    
2105      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2106    
2107      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2108    
2109      case OP_EOD:      case OP_EOD:
2110      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2111      SCHECK_PARTIAL();      SCHECK_PARTIAL();
2112      ecode++;      ecode++;
2113      break;      break;
# Line 1734  for (;;) Line 2118  for (;;)
2118      ASSERT_NL_OR_EOS:      ASSERT_NL_OR_EOS:
2119      if (eptr < md->end_subject &&      if (eptr < md->end_subject &&
2120          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2121        MRRETURN(MATCH_NOMATCH);        {
2122          if (md->partial != 0 &&
2123              eptr + 1 >= md->end_subject &&
2124              NLBLOCK->nltype == NLTYPE_FIXED &&
2125              NLBLOCK->nllen == 2 &&
2126              *eptr == NLBLOCK->nl[0])
2127            {
2128            md->hitend = TRUE;
2129            if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2130            }
2131          RRETURN(MATCH_NOMATCH);
2132          }
2133    
2134      /* Either at end of string or \n before end. */      /* Either at end of string or \n before end. */
2135    
2136      SCHECK_PARTIAL();      SCHECK_PARTIAL();
2137      ecode++;      ecode++;
2138      break;      break;
# Line 1753  for (;;) Line 2148  for (;;)
2148        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2149        partial matching. */        partial matching. */
2150    
2151  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2152        if (utf8)        if (utf)
2153          {          {
2154          /* Get status of previous character */          /* Get status of previous character */
2155    
2156          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2157            {            {
2158            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2159            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2160            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2161            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2162  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 1826  for (;;) Line 2221  for (;;)
2221              }              }
2222            else            else
2223  #endif  #endif
2224            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);            prev_is_word = MAX_255(eptr[-1])
2225                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2226            }            }
2227    
2228          /* Get status of next character */          /* Get status of next character */
# Line 1849  for (;;) Line 2245  for (;;)
2245            }            }
2246          else          else
2247  #endif  #endif
2248          cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          cur_is_word = MAX_255(*eptr)
2249              && ((md->ctypes[*eptr] & ctype_word) != 0);
2250          }          }
2251    
2252        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2253    
2254        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2255             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2256          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2257        }        }
2258      break;      break;
2259    
2260      /* Match a single character type; inline for speed */      /* Match any single character type except newline; have to take care with
2261        CRLF newlines and partial matching. */
2262    
2263      case OP_ANY:      case OP_ANY:
2264      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
2265        if (md->partial != 0 &&
2266            eptr + 1 >= md->end_subject &&
2267            NLBLOCK->nltype == NLTYPE_FIXED &&
2268            NLBLOCK->nllen == 2 &&
2269            *eptr == NLBLOCK->nl[0])
2270          {
2271          md->hitend = TRUE;
2272          if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2273          }
2274    
2275      /* Fall through */      /* Fall through */
2276    
2277        /* Match any single character whatsoever. */
2278    
2279      case OP_ALLANY:      case OP_ALLANY:
2280      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2281        {        {                            /* not be updated before SCHECK_PARTIAL. */
2282        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2283        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2284        }        }
2285      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2286    #ifdef SUPPORT_UTF
2287        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2288    #endif
2289      ecode++;      ecode++;
2290      break;      break;
2291    
# Line 1880  for (;;) Line 2293  for (;;)
2293      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2294    
2295      case OP_ANYBYTE:      case OP_ANYBYTE:
2296      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2297        {        {                            /* not be updated before SCHECK_PARTIAL. */
2298        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2299        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2300        }        }
2301        eptr++;
2302      ecode++;      ecode++;
2303      break;      break;
2304    
# Line 1892  for (;;) Line 2306  for (;;)
2306      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2307        {        {
2308        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2309        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2310        }        }
2311      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2312      if (      if (
2313  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2314         c < 256 &&         c < 256 &&
2315  #endif  #endif
2316         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2317         )         )
2318        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2319      ecode++;      ecode++;
2320      break;      break;
2321    
# Line 1909  for (;;) Line 2323  for (;;)
2323      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2324        {        {
2325        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2326        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2327        }        }
2328      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2329      if (      if (
2330  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2331         c >= 256 ||         c > 255 ||
2332  #endif  #endif
2333         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2334         )         )
2335        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2336      ecode++;      ecode++;
2337      break;      break;
2338    
# Line 1926  for (;;) Line 2340  for (;;)
2340      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2341        {        {
2342        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2343        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2344        }        }
2345      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2346      if (      if (
2347  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2348         c < 256 &&         c < 256 &&
2349  #endif  #endif
2350         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2351         )         )
2352        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2353      ecode++;      ecode++;
2354      break;      break;
2355    
# Line 1943  for (;;) Line 2357  for (;;)
2357      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2358        {        {
2359        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2360        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2361        }        }
2362      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2363      if (      if (
2364  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2365         c >= 256 ||         c > 255 ||
2366  #endif  #endif
2367         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2368         )         )
2369        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2370      ecode++;      ecode++;
2371      break;      break;
2372    
# Line 1960  for (;;) Line 2374  for (;;)
2374      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2375        {        {
2376        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2377        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2378        }        }
2379      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2380      if (      if (
2381  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2382         c < 256 &&         c < 256 &&
2383  #endif  #endif
2384         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2385         )         )
2386        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2387      ecode++;      ecode++;
2388      break;      break;
2389    
# Line 1977  for (;;) Line 2391  for (;;)
2391      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2392        {        {
2393        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2394        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2395        }        }
2396      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2397      if (      if (
2398  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2399         c >= 256 ||         c > 255 ||
2400  #endif  #endif
2401         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2402         )         )
2403        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2404      ecode++;      ecode++;
2405      break;      break;
2406    
# Line 1994  for (;;) Line 2408  for (;;)
2408      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2409        {        {
2410        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2411        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2412        }        }
2413      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2414      switch(c)      switch(c)
2415        {        {
2416        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2417        case 0x000d:  
2418        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        case CHAR_CR:
2419          if (eptr >= md->end_subject)
2420            {
2421            SCHECK_PARTIAL();
2422            }
2423          else if (*eptr == CHAR_LF) eptr++;
2424        break;        break;
2425    
2426        case 0x000a:        case CHAR_LF:
2427        break;        break;
2428    
2429        case 0x000b:        case CHAR_VT:
2430        case 0x000c:        case CHAR_FF:
2431        case 0x0085:        case CHAR_NEL:
2432    #ifndef EBCDIC
2433        case 0x2028:        case 0x2028:
2434        case 0x2029:        case 0x2029:
2435        if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);  #endif  /* Not EBCDIC */
2436          if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
2437        break;        break;
2438        }        }
2439      ecode++;      ecode++;
# Line 2022  for (;;) Line 2443  for (;;)
2443      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2444        {        {
2445        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2446        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2447        }        }
2448      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2449      switch(c)      switch(c)
2450        {        {
2451        default: break;        default: break;
2452        case 0x09:      /* HT */        case CHAR_HT:
2453        case 0x20:      /* SPACE */        case CHAR_SPACE:
2454    #ifndef EBCDIC
2455        case 0xa0:      /* NBSP */        case 0xa0:      /* NBSP */
2456        case 0x1680:    /* OGHAM SPACE MARK */        case 0x1680:    /* OGHAM SPACE MARK */
2457        case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */        case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
# Line 2047  for (;;) Line 2469  for (;;)
2469        case 0x202f:    /* NARROW NO-BREAK SPACE */        case 0x202f:    /* NARROW NO-BREAK SPACE */
2470        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2471        case 0x3000:    /* IDEOGRAPHIC SPACE */        case 0x3000:    /* IDEOGRAPHIC SPACE */
2472        MRRETURN(MATCH_NOMATCH);  #endif  /* Not EBCDIC */
2473          RRETURN(MATCH_NOMATCH);
2474        }        }
2475      ecode++;      ecode++;
2476      break;      break;
# Line 2056  for (;;) Line 2479  for (;;)
2479      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2480        {        {
2481        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2482        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2483        }        }
2484      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2485      switch(c)      switch(c)
2486        {        {
2487        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2488        case 0x09:      /* HT */        case CHAR_HT:
2489        case 0x20:      /* SPACE */        case CHAR_SPACE:
2490    #ifndef EBCDIC
2491        case 0xa0:      /* NBSP */        case 0xa0:      /* NBSP */
2492        case 0x1680:    /* OGHAM SPACE MARK */        case 0x1680:    /* OGHAM SPACE MARK */
2493        case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */        case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
# Line 2081  for (;;) Line 2505  for (;;)
2505        case 0x202f:    /* NARROW NO-BREAK SPACE */        case 0x202f:    /* NARROW NO-BREAK SPACE */
2506        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2507        case 0x3000:    /* IDEOGRAPHIC SPACE */        case 0x3000:    /* IDEOGRAPHIC SPACE */
2508    #endif  /* Not EBCDIC */
2509        break;        break;
2510        }        }
2511      ecode++;      ecode++;
# Line 2090  for (;;) Line 2515  for (;;)
2515      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2516        {        {
2517        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2518        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2519        }        }
2520      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2521      switch(c)      switch(c)
2522        {        {
2523        default: break;        default: break;
2524        case 0x0a:      /* LF */        case CHAR_LF:
2525        case 0x0b:      /* VT */        case CHAR_VT:
2526        case 0x0c:      /* FF */        case CHAR_FF:
2527        case 0x0d:      /* CR */        case CHAR_CR:
2528        case 0x85:      /* NEL */        case CHAR_NEL:
2529    #ifndef EBCDIC
2530        case 0x2028:    /* LINE SEPARATOR */        case 0x2028:    /* LINE SEPARATOR */
2531        case 0x2029:    /* PARAGRAPH SEPARATOR */        case 0x2029:    /* PARAGRAPH SEPARATOR */
2532        MRRETURN(MATCH_NOMATCH);  #endif  /* Not EBCDIC */
2533          RRETURN(MATCH_NOMATCH);
2534        }        }
2535      ecode++;      ecode++;
2536      break;      break;
# Line 2112  for (;;) Line 2539  for (;;)
2539      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2540        {        {
2541        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2542        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2543        }        }
2544      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2545      switch(c)      switch(c)
2546        {        {
2547        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2548        case 0x0a:      /* LF */        case CHAR_LF:
2549        case 0x0b:      /* VT */        case CHAR_VT:
2550        case 0x0c:      /* FF */        case CHAR_FF:
2551        case 0x0d:      /* CR */        case CHAR_CR:
2552        case 0x85:      /* NEL */        case CHAR_NEL:
2553    #ifndef EBCDIC
2554        case 0x2028:    /* LINE SEPARATOR */        case 0x2028:    /* LINE SEPARATOR */
2555        case 0x2029:    /* PARAGRAPH SEPARATOR */        case 0x2029:    /* PARAGRAPH SEPARATOR */
2556    #endif  /* Not EBCDIC */
2557        break;        break;
2558        }        }
2559      ecode++;      ecode++;
# Line 2139  for (;;) Line 2568  for (;;)
2568      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2569        {        {
2570        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2571        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2572        }        }
2573      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2574        {        {
# Line 2148  for (;;) Line 2577  for (;;)
2577        switch(ecode[1])        switch(ecode[1])
2578          {          {
2579          case PT_ANY:          case PT_ANY:
2580          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2581          break;          break;
2582    
2583          case PT_LAMP:          case PT_LAMP:
2584          if ((prop->chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2585               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2586               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2587            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2588          break;          break;
2589    
2590          case PT_GC:          case PT_GC:
2591          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2592            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2593          break;          break;
2594    
2595          case PT_PC:          case PT_PC:
2596          if ((ecode[2] != prop->chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2597            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2598          break;          break;
2599    
2600          case PT_SC:          case PT_SC:
2601          if ((ecode[2] != prop->script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2602            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2603          break;          break;
2604    
2605          /* These are specials */          /* These are specials */
2606    
2607          case PT_ALNUM:          case PT_ALNUM:
2608          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2609               _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))               PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2610            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2611          break;          break;
2612    
2613          case PT_SPACE:    /* Perl space */          case PT_SPACE:    /* Perl space */
2614          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2615               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2616                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2617            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2618          break;          break;
2619    
2620          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2621          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2622               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2623               c == CHAR_FF || c == CHAR_CR)               c == CHAR_FF || c == CHAR_CR)
2624                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2625            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2626          break;          break;
2627    
2628          case PT_WORD:          case PT_WORD:
2629          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2630               _pcre_ucp_gentype[prop->chartype] == ucp_N ||               PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2631               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2632            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2633          break;          break;
2634    
2635          /* This should never occur */          /* This should never occur */
# Line 2220  for (;;) Line 2649  for (;;)
2649      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2650        {        {
2651        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2652        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2653        }        }
2654      GETCHARINCTEST(c, eptr);      else
2655        {        {
2656        int category = UCD_CATEGORY(c);        int lgb, rgb;
2657        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        GETCHARINCTEST(c, eptr);
2658          lgb = UCD_GRAPHBREAK(c);
2659        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2660          {          {
2661          int len = 1;          int len = 1;
2662          if (!utf8) c = *eptr; else          if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2663            {          rgb = UCD_GRAPHBREAK(c);
2664            GETCHARLEN(c, eptr, len);          if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
2665            }          lgb = rgb;
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
2666          eptr += len;          eptr += len;
2667          }          }
2668        }        }
2669        CHECK_PARTIAL();
2670      ecode++;      ecode++;
2671      break;      break;
2672  #endif  #endif
# Line 2252  for (;;) Line 2681  for (;;)
2681      loops). */      loops). */
2682    
2683      case OP_REF:      case OP_REF:
2684        {      case OP_REFI:
2685        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2686        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2687        ecode += 1 + IMM2_SIZE;
2688    
2689        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2690    
2691        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2692        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
2693        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.  
2694    
2695        (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
2696        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2697    
2698        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
2699        referenced subpattern. */      referenced subpattern. */
2700    
2701        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2702          length = (md->jscript_compat)? 0 : (int)(md->end_subject - eptr + 1);        length = (md->jscript_compat)? 0 : -1;
2703        else      else
2704          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2705    
2706        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2707    
2708        switch (*ecode)      switch (*ecode)
2709          {        {
2710          case OP_CRSTAR:        case OP_CRSTAR:
2711          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2712          case OP_CRPLUS:        case OP_CRPLUS:
2713          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2714          case OP_CRQUERY:        case OP_CRQUERY:
2715          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2716          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2717          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2718          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2719          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2720          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2721          break;        break;
2722    
2723          case OP_CRRANGE:        case OP_CRRANGE:
2724          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2725          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2726          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2727          max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2728          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2729          ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2730          break;        break;
2731    
2732          default:               /* No repeat follows */        default:               /* No repeat follows */
2733          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2734            {          {
2735            CHECK_PARTIAL();          if (length == -2) eptr = md->end_subject;   /* Partial match */
2736            MRRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2737            }          RRETURN(MATCH_NOMATCH);
         eptr += length;  
         continue;              /* With the main loop */  
2738          }          }
2739          eptr += length;
2740          continue;              /* With the main loop */
2741          }
2742    
2743        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2744        main loop. */      zero, just continue with the main loop. If the length is negative, it
2745        means the reference is unset in non-Java-compatible mode. If the minimum is
2746        zero, we can continue at the same level without recursion. For any other
2747        minimum, carrying on will result in NOMATCH. */
2748    
2749        if (length == 0) continue;      if (length == 0) continue;
2750        if (length < 0 && min == 0) continue;
2751    
2752        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2753        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2754        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2755    
2756        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2757          {
2758          int slength;
2759          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2760          {          {
2761          if (!match_ref(offset, eptr, length, md, ims))          if (slength == -2) eptr = md->end_subject;   /* Partial match */
2762            {          CHECK_PARTIAL();
2763            CHECK_PARTIAL();          RRETURN(MATCH_NOMATCH);
           MRRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2764          }          }
2765          eptr += slength;
2766          }
2767    
2768        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2769        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2770    
2771        if (min == max) continue;      if (min == max) continue;
2772    
2773        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2774    
2775        if (minimize)      if (minimize)
2776          {
2777          for (fi = min;; fi++)
2778          {          {
2779          for (fi = min;; fi++)          int slength;
2780            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2781            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2782            if (fi >= max) RRETURN(MATCH_NOMATCH);
2783            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2784            {            {
2785            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            if (slength == -2) eptr = md->end_subject;   /* Partial match */
2786            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            CHECK_PARTIAL();
2787            if (fi >= max) MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             MRRETURN(MATCH_NOMATCH);  
             }  
           eptr += length;  
2788            }            }
2789          /* Control never gets here */          eptr += slength;
2790          }          }
2791          /* Control never gets here */
2792          }
2793    
2794        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2795    
2796        else      else
2797          {
2798          pp = eptr;
2799          for (i = min; i < max; i++)
2800          {          {
2801          pp = eptr;          int slength;
2802          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2803            {            {
2804            if (!match_ref(offset, eptr, length, md, ims))            /* Can't use CHECK_PARTIAL because we don't want to update eptr in
2805              the soft partial matching case. */
2806    
2807              if (slength == -2 && md->partial != 0 &&
2808                  md->end_subject > md->start_used_ptr)
2809              {              {
2810              CHECK_PARTIAL();              md->hitend = TRUE;
2811              break;              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2812              }              }
2813            eptr += length;            break;
           }  
         while (eptr >= pp)  
           {  
           RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);  
           if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
           eptr -= length;  
2814            }            }
2815          MRRETURN(MATCH_NOMATCH);          eptr += slength;
2816          }          }
2817    
2818          while (eptr >= pp)
2819            {
2820            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2821            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2822            eptr -= length;
2823            }
2824          RRETURN(MATCH_NOMATCH);
2825        }        }
2826      /* Control never gets here */      /* Control never gets here */
2827    
# Line 2392  for (;;) Line 2839  for (;;)
2839      case OP_NCLASS:      case OP_NCLASS:
2840      case OP_CLASS:      case OP_CLASS:
2841        {        {
2842          /* The data variable is saved across frames, so the byte map needs to
2843          be stored there. */
2844    #define BYTE_MAP ((pcre_uint8 *)data)
2845        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2846        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2847    
2848        switch (*ecode)        switch (*ecode)
2849          {          {
# Line 2414  for (;;) Line 2864  for (;;)
2864          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2865          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2866          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2867          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2868          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2869          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2870          break;          break;
2871    
2872          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2426  for (;;) Line 2876  for (;;)
2876    
2877        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2878    
2879  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2880        /* UTF-8 mode */        if (utf)
       if (utf8)  
2881          {          {
2882          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2883            {            {
2884            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2885              {              {
2886              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2887              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2888              }              }
2889            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2890            if (c > 255)            if (c > 255)
2891              {              {
2892              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2893              }              }
2894            else            else
2895              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
             }  
2896            }            }
2897          }          }
2898        else        else
2899  #endif  #endif
2900        /* Not UTF-8 mode */        /* Not UTF mode */
2901          {          {
2902          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2903            {            {
2904            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2905              {              {
2906              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2907              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2908              }              }
2909            c = *eptr++;            c = *eptr++;
2910            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2911              if (c > 255)
2912                {
2913                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2914                }
2915              else
2916    #endif
2917                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2918            }            }
2919          }          }
2920    
# Line 2474  for (;;) Line 2928  for (;;)
2928    
2929        if (minimize)        if (minimize)
2930          {          {
2931  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2932          /* UTF-8 mode */          if (utf)
         if (utf8)  
2933            {            {
2934            for (fi = min;; fi++)            for (fi = min;; fi++)
2935              {              {
2936              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2937              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2938              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2939              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2940                {                {
2941                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2942                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2943                }                }
2944              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2945              if (c > 255)              if (c > 255)
2946                {                {
2947                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2948                }                }
2949              else              else
2950                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
               }  
2951              }              }
2952            }            }
2953          else          else
2954  #endif  #endif
2955          /* Not UTF-8 mode */          /* Not UTF mode */
2956            {            {
2957            for (fi = min;; fi++)            for (fi = min;; fi++)
2958              {              {
2959              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2960              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2961              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2962              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2963                {                {
2964                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2965                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2966                }                }
2967              c = *eptr++;              c = *eptr++;
2968              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2969                if (c > 255)
2970                  {
2971                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2972                  }
2973                else
2974    #endif
2975                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2976              }              }
2977            }            }
2978          /* Control never gets here */          /* Control never gets here */
# Line 2526  for (;;) Line 2984  for (;;)
2984          {          {
2985          pp = eptr;          pp = eptr;
2986    
2987  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2988          /* UTF-8 mode */          if (utf)
         if (utf8)  
2989            {            {
2990            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2991              {              {
# Line 2544  for (;;) Line 3001  for (;;)
3001                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
3002                }                }
3003              else              else
3004                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
3005              eptr += len;              eptr += len;
3006              }              }
3007            for (;;)            for (;;)
3008              {              {
3009              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
3010              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3011              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3012              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2559  for (;;) Line 3014  for (;;)
3014            }            }
3015          else          else
3016  #endif  #endif
3017            /* Not UTF-8 mode */            /* Not UTF mode */
3018            {            {
3019            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3020              {              {
# Line 2569  for (;;) Line 3024  for (;;)
3024                break;                break;
3025                }                }
3026              c = *eptr;              c = *eptr;
3027              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
3028                if (c > 255)
3029                  {
3030                  if (op == OP_CLASS) break;
3031                  }
3032                else
3033    #endif
3034                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3035              eptr++;              eptr++;
3036              }              }
3037            while (eptr >= pp)            while (eptr >= pp)
3038              {              {
3039              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
3040              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3041              eptr--;              eptr--;
3042              }              }
3043            }            }
3044    
3045          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3046          }          }
3047    #undef BYTE_MAP
3048        }        }
3049      /* Control never gets here */      /* Control never gets here */
3050    
# Line 2590  for (;;) Line 3053  for (;;)
3053      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
3054      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
3055    
3056  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3057      case OP_XCLASS:      case OP_XCLASS:
3058        {        {
3059        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2615  for (;;) Line 3078  for (;;)
3078          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3079          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3080          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3081          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
3082          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
3083          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
3084          break;          break;
3085    
3086          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2632  for (;;) Line 3095  for (;;)
3095          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3096            {            {
3097            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3098            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3099            }            }
3100          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
3101          if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3102          }          }
3103    
3104        /* 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 2650  for (;;) Line 3113  for (;;)
3113          {          {
3114          for (fi = min;; fi++)          for (fi = min;; fi++)
3115            {            {
3116            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
3117            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3118            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3119            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3120              {              {
3121              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3122              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3123              }              }
3124            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3125            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3126            }            }
3127          /* Control never gets here */          /* Control never gets here */
3128          }          }
# Line 2677  for (;;) Line 3140  for (;;)
3140              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3141              break;              break;
3142              }              }
3143    #ifdef SUPPORT_UTF
3144            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3145            if (!_pcre_xclass(c, data)) break;  #else
3146              c = *eptr;
3147    #endif
3148              if (!PRIV(xclass)(c, data, utf)) break;
3149            eptr += len;            eptr += len;
3150            }            }
3151          for(;;)          for(;;)
3152            {            {
3153            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3154            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3155            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3156            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3157              if (utf) BACKCHAR(eptr);
3158    #endif
3159            }            }
3160          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3161          }          }
3162    
3163        /* Control never gets here */        /* Control never gets here */
# Line 2698  for (;;) Line 3167  for (;;)
3167      /* Match a single character, casefully */      /* Match a single character, casefully */
3168    
3169      case OP_CHAR:      case OP_CHAR:
3170  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3171      if (utf8)      if (utf)
3172        {        {
3173        length = 1;        length = 1;
3174        ecode++;        ecode++;
# Line 2707  for (;;) Line 3176  for (;;)
3176        if (length > md->end_subject - eptr)        if (length > md->end_subject - eptr)
3177          {          {
3178          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
3179          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3180          }          }
3181        while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);
3182        }        }
3183      else      else
3184  #endif  #endif
3185        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3186        {        {
3187        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3188          {          {
3189          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
3190          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3191          }          }
3192        if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
3193        ecode += 2;        ecode += 2;
3194        }        }
3195      break;      break;
3196    
3197      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3198        subject, give up immediately. */
3199    
3200        case OP_CHARI:
3201        if (eptr >= md->end_subject)
3202          {
3203          SCHECK_PARTIAL();
3204          RRETURN(MATCH_NOMATCH);
3205          }
3206    
3207      case OP_CHARNC:  #ifdef SUPPORT_UTF
3208  #ifdef SUPPORT_UTF8      if (utf)
     if (utf8)  
3209        {        {
3210        length = 1;        length = 1;
3211        ecode++;        ecode++;
3212        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3213    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         MRRETURN(MATCH_NOMATCH);  
         }  
   
3214        /* 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
3215        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3216          fast lookup table. We know that there is at least one byte left in the
3217          subject. */
3218    
3219        if (fc < 128)        if (fc < 128)
3220          {          {
3221          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          if (md->lcc[fc]
3222                != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3223            ecode++;
3224            eptr++;
3225          }          }
3226    
3227        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3228          use the value of "length" to check for sufficient bytes left, because the
3229          other case of the character may have more or fewer bytes.  */
3230    
3231        else        else
3232          {          {
# Line 2766  for (;;) Line 3242  for (;;)
3242  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3243            if (dc != UCD_OTHERCASE(fc))            if (dc != UCD_OTHERCASE(fc))
3244  #endif  #endif
3245              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3246            }            }
3247          }          }
3248        }        }
3249      else      else
3250  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3251    
3252      /* Non-UTF-8 mode */      /* Not UTF mode */
3253        {        {
3254        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3255          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3256          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         MRRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);  
3257        ecode += 2;        ecode += 2;
3258        }        }
3259      break;      break;
# Line 2788  for (;;) Line 3261  for (;;)
3261      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3262    
3263      case OP_EXACT:      case OP_EXACT:
3264        case OP_EXACTI:
3265      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3266      ecode += 3;      ecode += 1 + IMM2_SIZE;
3267      goto REPEATCHAR;      goto REPEATCHAR;
3268    
3269      case OP_POSUPTO:      case OP_POSUPTO:
3270        case OP_POSUPTOI:
3271      possessive = TRUE;      possessive = TRUE;
3272      /* Fall through */      /* Fall through */
3273    
3274      case OP_UPTO:      case OP_UPTO:
3275        case OP_UPTOI:
3276      case OP_MINUPTO:      case OP_MINUPTO:
3277        case OP_MINUPTOI:
3278      min = 0;      min = 0;
3279      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3280      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3281      ecode += 3;      ecode += 1 + IMM2_SIZE;
3282      goto REPEATCHAR;      goto REPEATCHAR;
3283    
3284      case OP_POSSTAR:      case OP_POSSTAR:
3285        case OP_POSSTARI:
3286      possessive = TRUE;      possessive = TRUE;
3287      min = 0;      min = 0;
3288      max = INT_MAX;      max = INT_MAX;
# Line 2812  for (;;) Line 3290  for (;;)
3290      goto REPEATCHAR;      goto REPEATCHAR;
3291    
3292      case OP_POSPLUS:      case OP_POSPLUS:
3293        case OP_POSPLUSI:
3294      possessive = TRUE;      possessive = TRUE;
3295      min = 1;      min = 1;
3296      max = INT_MAX;      max = INT_MAX;
# Line 2819  for (;;) Line 3298  for (;;)
3298      goto REPEATCHAR;      goto REPEATCHAR;
3299    
3300      case OP_POSQUERY:      case OP_POSQUERY:
3301        case OP_POSQUERYI:
3302      possessive = TRUE;      possessive = TRUE;
3303      min = 0;      min = 0;
3304      max = 1;      max = 1;
# Line 2826  for (;;) Line 3306  for (;;)
3306      goto REPEATCHAR;      goto REPEATCHAR;
3307    
3308      case OP_STAR:      case OP_STAR:
3309        case OP_STARI:
3310      case OP_MINSTAR:      case OP_MINSTAR:
3311        case OP_MINSTARI:
3312      case OP_PLUS:      case OP_PLUS:
3313        case OP_PLUSI:
3314      case OP_MINPLUS:      case OP_MINPLUS:
3315        case OP_MINPLUSI:
3316      case OP_QUERY:      case OP_QUERY:
3317        case OP_QUERYI:
3318      case OP_MINQUERY:      case OP_MINQUERY:
3319      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3320        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3321      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3322      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3323      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3324      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2841  for (;;) Line 3326  for (;;)
3326      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3327    
3328      REPEATCHAR:      REPEATCHAR:
3329  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3330      if (utf8)      if (utf)
3331        {        {
3332        length = 1;        length = 1;
3333        charptr = ecode;        charptr = ecode;
# Line 2856  for (;;) Line 3341  for (;;)
3341          {          {
3342  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3343          unsigned int othercase;          unsigned int othercase;
3344          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3345              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3346            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3347          else oclength = 0;          else oclength = 0;
3348  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3349    
3350          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3351            {            {
3352            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3353              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3354  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3355            else if (oclength > 0 &&            else if (oclength > 0 &&
3356                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3357                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3358  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3359            else            else
3360              {              {
3361              CHECK_PARTIAL();              CHECK_PARTIAL();
3362              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3363              }              }
3364            }            }
3365    
# Line 2884  for (;;) Line 3369  for (;;)
3369            {            {
3370            for (fi = min;; fi++)            for (fi = min;; fi++)
3371              {              {
3372              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3373              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3374              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3375              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3376                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3377  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3378              else if (oclength > 0 &&              else if (oclength > 0 &&
3379                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3380                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3381  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3382              else              else
3383                {                {
3384                CHECK_PARTIAL();                CHECK_PARTIAL();
3385                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3386                }                }
3387              }              }
3388            /* Control never gets here */            /* Control never gets here */
# Line 2909  for (;;) Line 3394  for (;;)
3394            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3395              {              {
3396              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3397                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3398  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3399              else if (oclength > 0 &&              else if (oclength > 0 &&
3400                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3401                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3402  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3403              else              else
3404                {                {
# Line 2926  for (;;) Line 3411  for (;;)
3411    
3412            for(;;)            for(;;)
3413              {              {
3414              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3415              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3416              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3417  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3418              eptr--;              eptr--;
3419              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2945  for (;;) Line 3430  for (;;)
3430        value of fc will always be < 128. */        value of fc will always be < 128. */
3431        }        }
3432      else      else
3433  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3434          /* When not in UTF-8 mode, load a single-byte character. */
3435      /* When not in UTF-8 mode, load a single-byte character. */        fc = *ecode++;
3436    
3437      fc = *ecode++;      /* The value of fc at this point is always one character, though we may
3438        or may not be in UTF mode. The code is duplicated for the caseless and
     /* 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  
3439      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3440      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3441      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 2961  for (;;) Line 3444  for (;;)
3444      maximizing, find the maximum number of characters and work backwards. */      maximizing, find the maximum number of characters and work backwards. */
3445    
3446      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3447        max, eptr));        max, (char *)eptr));
3448    
3449      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3450        {        {
3451        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3452          /* fc must be < 128 if UTF is enabled. */
3453          foc = md->fcc[fc];
3454    #else
3455    #ifdef SUPPORT_UTF
3456    #ifdef SUPPORT_UCP
3457          if (utf && fc > 127)
3458            foc = UCD_OTHERCASE(fc);
3459    #else
3460          if (utf && fc > 127)
3461            foc = fc;
3462    #endif /* SUPPORT_UCP */
3463          else
3464    #endif /* SUPPORT_UTF */
3465            foc = TABLE_GET(fc, md->fcc, fc);
3466    #endif /* COMPILE_PCRE8 */
3467    
3468        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3469          {          {
3470          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3471            {            {
3472            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3473            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3474            }            }
3475          if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3476            eptr++;
3477          }          }
3478        if (min == max) continue;        if (min == max) continue;
3479        if (minimize)        if (minimize)
3480          {          {
3481          for (fi = min;; fi++)          for (fi = min;; fi++)
3482            {            {
3483            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3484            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3485            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3486            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3487              {              {
3488              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3489              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3490              }              }
3491            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3492              eptr++;
3493            }            }
3494          /* Control never gets here */          /* Control never gets here */
3495          }          }
# Line 3002  for (;;) Line 3503  for (;;)
3503              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3504              break;              break;
3505              }              }
3506            if (fc != md->lcc[*eptr]) break;            if (fc != *eptr && foc != *eptr) break;
3507            eptr++;            eptr++;
3508            }            }
3509    
# Line 3010  for (;;) Line 3511  for (;;)
3511    
3512          while (eptr >= pp)          while (eptr >= pp)
3513            {            {
3514            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3515            eptr--;            eptr--;
3516            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3517            }            }
3518          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3519          }          }
3520        /* Control never gets here */        /* Control never gets here */
3521        }        }
# Line 3028  for (;;) Line 3529  for (;;)
3529          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3530            {            {
3531            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3532            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3533            }            }
3534          if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);          if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
3535          }          }
3536    
3537        if (min == max) continue;        if (min == max) continue;
# Line 3039  for (;;) Line 3540  for (;;)
3540          {          {
3541          for (fi = min;; fi++)          for (fi = min;; fi++)
3542            {            {
3543            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3544            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3545            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3546            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3547              {              {
3548              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3549              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3550              }              }
3551            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);            if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
3552            }            }
3553          /* Control never gets here */          /* Control never gets here */
3554          }          }
# Line 3068  for (;;) Line 3569  for (;;)
3569    
3570          while (eptr >= pp)          while (eptr >= pp)
3571            {            {
3572            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3573            eptr--;            eptr--;
3574            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3575            }            }
3576          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3577          }          }
3578        }        }
3579      /* Control never gets here */      /* Control never gets here */
# Line 3081  for (;;) Line 3582  for (;;)
3582      checking can be multibyte. */      checking can be multibyte. */
3583    
3584      case OP_NOT:      case OP_NOT:
3585        case OP_NOTI:
3586      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3587        {        {
3588        SCHECK_PARTIAL();        SCHECK_PARTIAL();
3589        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
3590        }        }
3591      ecode++;  #ifdef SUPPORT_UTF
3592      GETCHARINCTEST(c, eptr);      if (utf)
     if ((ims & PCRE_CASELESS) != 0)  
3593        {        {
3594  #ifdef SUPPORT_UTF8        register unsigned int ch, och;
3595        if (c < 256)  
3596  #endif        ecode++;
3597        c = md->lcc[c];        GETCHARINC(ch, ecode);
3598        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        GETCHARINC(c, eptr);
3599    
3600          if (op == OP_NOT)
3601            {
3602            if (ch == c) RRETURN(MATCH_NOMATCH);
3603            }
3604          else
3605            {
3606    #ifdef SUPPORT_UCP
3607            if (ch > 127)
3608              och = UCD_OTHERCASE(ch);
3609    #else
3610            if (ch > 127)
3611              och = ch;
3612    #endif /* SUPPORT_UCP */
3613            else
3614              och = TABLE_GET(ch, md->fcc, ch);
3615            if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3616            }
3617        }        }
3618      else      else
3619    #endif
3620        {        {
3621        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        register unsigned int ch = ecode[1];
3622          c = *eptr++;
3623          if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c))
3624            RRETURN(MATCH_NOMATCH);
3625          ecode += 2;
3626        }        }
3627      break;      break;
3628    
# Line 3110  for (;;) Line 3634  for (;;)
3634      about... */      about... */
3635    
3636      case OP_NOTEXACT:      case OP_NOTEXACT:
3637        case OP_NOTEXACTI:
3638      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3639      ecode += 3;      ecode += 1 + IMM2_SIZE;
3640      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3641    
3642      case OP_NOTUPTO:      case OP_NOTUPTO:
3643        case OP_NOTUPTOI:
3644      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3645        case OP_NOTMINUPTOI:
3646      min = 0;      min = 0;
3647      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3648      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3649      ecode += 3;      ecode += 1 + IMM2_SIZE;
3650      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3651    
3652      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3653        case OP_NOTPOSSTARI:
3654      possessive = TRUE;      possessive = TRUE;
3655      min = 0;      min = 0;
3656      max = INT_MAX;      max = INT_MAX;
# Line 3130  for (;;) Line 3658  for (;;)
3658      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3659    
3660      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3661        case OP_NOTPOSPLUSI:
3662      possessive = TRUE;      possessive = TRUE;
3663      min = 1;      min = 1;
3664      max = INT_MAX;      max = INT_MAX;
# Line 3137  for (;;) Line 3666  for (;;)
3666      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3667    
3668      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3669        case OP_NOTPOSQUERYI:
3670      possessive = TRUE;      possessive = TRUE;
3671      min = 0;      min = 0;
3672      max = 1;      max = 1;
# Line 3144  for (;;) Line 3674  for (;;)
3674      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3675    
3676      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3677        case OP_NOTPOSUPTOI:
3678      possessive = TRUE;      possessive = TRUE;
3679      min = 0;      min = 0;
3680      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3681      ecode += 3;      ecode += 1 + IMM2_SIZE;
3682      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3683    
3684      case OP_NOTSTAR:      case OP_NOTSTAR:
3685        case OP_NOTSTARI:
3686      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3687        case OP_NOTMINSTARI:
3688      case OP_NOTPLUS:      case OP_NOTPLUS:
3689        case OP_NOTPLUSI:
3690      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3691        case OP_NOTMINPLUSI:
3692      case OP_NOTQUERY:      case OP_NOTQUERY:
3693        case OP_NOTQUERYI:
3694      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3695      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3696        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3697      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3698      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3699      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3165  for (;;) Line 3702  for (;;)
3702      /* Common code for all repeated single-byte matches. */      /* Common code for all repeated single-byte matches. */
3703    
3704      REPEATNOTCHAR:      REPEATNOTCHAR:
3705      fc = *ecode++;      GETCHARINCTEST(fc, ecode);
3706    
3707      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
3708      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 3176  for (;;) Line 3713  for (;;)
3713      characters and work backwards. */      characters and work backwards. */
3714    
3715      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,
3716        max, eptr));        max, (char *)eptr));
3717    
3718      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3719        {        {
3720        fc = md->lcc[fc];  #ifdef SUPPORT_UTF
3721    #ifdef SUPPORT_UCP
3722          if (utf && fc > 127)
3723            foc = UCD_OTHERCASE(fc);
3724    #else
3725          if (utf && fc > 127)
3726            foc = fc;
3727    #endif /* SUPPORT_UCP */
3728          else
3729    #endif /* SUPPORT_UTF */
3730            foc = TABLE_GET(fc, md->fcc, fc);
3731    
3732  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3733        /* UTF-8 mode */        if (utf)
       if (utf8)  
3734          {          {
3735          register unsigned int d;          register unsigned int d;
3736          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3192  for (;;) Line 3738  for (;;)
3738            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3739              {              {
3740              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3741              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3742              }              }
3743            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3744            if (d < 256) d = md->lcc[d];            if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) MRRETURN(MATCH_NOMATCH);  
3745            }            }
3746          }          }
3747        else        else
3748  #endif  #endif
3749          /* Not UTF mode */
       /* Not UTF-8 mode */  
3750          {          {
3751          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3752            {            {
3753            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3754              {              {
3755              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3756              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3757              }              }
3758            if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3759              eptr++;
3760            }            }
3761          }          }
3762    
# Line 3219  for (;;) Line 3764  for (;;)
3764    
3765        if (minimize)        if (minimize)
3766          {          {
3767  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3768          /* UTF-8 mode */          if (utf)
         if (utf8)  
3769            {            {
3770            register unsigned int d;            register unsigned int d;
3771            for (fi = min;; fi++)            for (fi = min;; fi++)
3772              {              {
3773              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3774              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3775              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3776              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3777                {                {
3778                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3779                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3780                }                }
3781              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3782              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
             if (fc == d) MRRETURN(MATCH_NOMATCH);  
3783              }              }
3784            }            }
3785          else          else
3786  #endif  #endif
3787          /* Not UTF-8 mode */          /* Not UTF mode */
3788            {            {
3789            for (fi = min;; fi++)            for (fi = min;; fi++)
3790              {              {
3791              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3792              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3793              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3794              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3795                {                {
3796                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3797                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3798                }                }
3799              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);              if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3800                eptr++;
3801              }              }
3802            }            }
3803          /* Control never gets here */          /* Control never gets here */
# Line 3265  for (;;) Line 3809  for (;;)
3809          {          {
3810          pp = eptr;          pp = eptr;
3811    
3812  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3813          /* UTF-8 mode */          if (utf)
         if (utf8)  
3814            {            {
3815            register unsigned int d;            register unsigned int d;
3816            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3279  for (;;) Line 3822  for (;;)
3822                break;                break;
3823                }                }
3824              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3825              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) break;
             if (fc == d) break;  
3826              eptr += len;              eptr += len;
3827              }              }
3828          if (possessive) continue;            if (possessive) continue;
3829          for(;;)            for(;;)
3830              {              {
3831              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3832              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3833              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3834              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3294  for (;;) Line 3836  for (;;)
3836            }            }
3837          else          else
3838  #endif  #endif
3839          /* Not UTF-8 mode */          /* Not UTF mode */
3840            {            {
3841            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3842              {              {
# Line 3303  for (;;) Line 3845  for (;;)
3845                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3846                break;                break;
3847                }                }
3848              if (fc == md->lcc[*eptr]) break;              if (fc == *eptr || foc == *eptr) break;
3849              eptr++;              eptr++;
3850              }              }
3851            if (possessive) continue;            if (possessive) continue;
3852            while (eptr >= pp)            while (eptr >= pp)
3853              {              {
3854              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3855              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3856              eptr--;              eptr--;
3857              }              }
3858            }            }
3859    
3860          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3861          }          }
3862        /* Control never gets here */        /* Control never gets here */
3863        }        }
# Line 3324  for (;;) Line 3866  for (;;)
3866    
3867      else      else
3868        {        {
3869  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3870        /* UTF-8 mode */        if (utf)
       if (utf8)  
3871          {          {
3872          register unsigned int d;          register