/[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 608 by ph10, Sun Jun 12 16:25:55 2011 UTC revision 1271 by ph10, Wed Mar 6 16:50:38 2013 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) 1997-2011 University of Cambridge             Copyright (c) 1997-2013 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  /* Values for setting in md->match_function_type to indicate two special types  /* The md->capture_last field uses the lower 16 bits for the last captured
60  of call to match(). We do it this way to save on using another stack variable,  substring (which can never be greater than 65535) and a bit in the top half
61    to mean "capture vector overflowed". This odd way of doing things was
62    implemented when it was realized that preserving and restoring the overflow bit
63    whenever the last capture number was saved/restored made for a neater
64    interface, and doing it this way saved on (a) another variable, which would
65    have increased the stack frame size (a big NO-NO in PCRE) and (b) another
66    separate set of save/restore instructions. The following defines are used in
67    implementing this. */
68    
69    #define CAPLMASK    0x0000ffff    /* The bits used for last_capture */
70    #define OVFLMASK    0xffff0000    /* The bits used for the overflow flag */
71    #define OVFLBIT     0x00010000    /* The bit that is set for overflow */
72    
73    /* Values for setting in md->match_function_type to indicate two special types
74    of call to match(). We do it this way to save on using another stack variable,
75  as stack usage is to be discouraged. */  as stack usage is to be discouraged. */
76    
77  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
# Line 74  defined PCRE_ERROR_xxx codes, which are Line 87  defined PCRE_ERROR_xxx codes, which are
87  negative to avoid the external error codes. */  negative to avoid the external error codes. */
88    
89  #define MATCH_ACCEPT       (-999)  #define MATCH_ACCEPT       (-999)
90  #define MATCH_COMMIT       (-998)  #define MATCH_KETRPOS      (-998)
91  #define MATCH_KETRPOS      (-997)  #define MATCH_ONCE         (-997)
92  #define MATCH_PRUNE        (-996)  /* The next 5 must be kept together and in sequence so that a test that checks
93  #define MATCH_SKIP         (-995)  for any one of them can use a range. */
94  #define MATCH_SKIP_ARG     (-994)  #define MATCH_COMMIT       (-996)
95  #define MATCH_THEN         (-993)  #define MATCH_PRUNE        (-995)
96    #define MATCH_SKIP         (-994)
97  /* This is a convenience macro for code that occurs many times. */  #define MATCH_SKIP_ARG     (-993)
98    #define MATCH_THEN         (-992)
99  #define MRRETURN(ra) \  #define MATCH_BACKTRACK_MAX MATCH_THEN
100    { \  #define MATCH_BACKTRACK_MIN MATCH_COMMIT
   md->mark = markptr; \  
   RRETURN(ra); \  
   }  
101    
102  /* 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.
103  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 100  because the offset vector is always a mu Line 110  because the offset vector is always a mu
110  static const char rep_min[] = { 0, 0, 1, 1, 0, 0 };  static const char rep_min[] = { 0, 0, 1, 1, 0, 0 };
111  static const char rep_max[] = { 0, 0, 0, 0, 1, 1 };  static const char rep_max[] = { 0, 0, 0, 0, 1, 1 };
112    
   
   
113  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
114  /*************************************************  /*************************************************
115  *        Debugging function to print chars       *  *        Debugging function to print chars       *
# Line 120  Returns:     nothing Line 128  Returns:     nothing
128  */  */
129    
130  static void  static void
131  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)
132  {  {
133  unsigned int c;  pcre_uint32 c;
134    BOOL utf = md->utf;
135  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;
136  while (length-- > 0)  while (length-- > 0)
137    if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c);    if (isprint(c = RAWUCHARINCTEST(p))) printf("%c", (char)c); else printf("\\x{%02x}", c);
138  }  }
139  #endif  #endif
140    
# Line 137  while (length-- > 0) Line 146  while (length-- > 0)
146    
147  /* Normally, if a back reference hasn't been set, the length that is passed is  /* Normally, if a back reference hasn't been set, the length that is passed is
148  negative, so the match always fails. However, in JavaScript compatibility mode,  negative, so the match always fails. However, in JavaScript compatibility mode,
149  the length passed is zero. Note that in caseless UTF-8 mode, the number of  the length passed is zero. Note that in caseless UTF-8 mode, the number of
150  subject bytes matched may be different to the number of reference bytes.  subject bytes matched may be different to the number of reference bytes.
151    
152  Arguments:  Arguments:
# Line 147  Arguments: Line 156  Arguments:
156    md          points to match data block    md          points to match data block
157    caseless    TRUE if caseless    caseless    TRUE if caseless
158    
159  Returns:      < 0 if not matched, otherwise the number of subject bytes matched  Returns:      >= 0 the number of subject bytes matched
160                  -1 no match
161                  -2 partial match; always given if at end subject
162  */  */
163    
164  static int  static int
165  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,
166    BOOL caseless)    BOOL caseless)
167  {  {
168  USPTR eptr_start = eptr;  PCRE_PUCHAR eptr_start = eptr;
169  register USPTR p = md->start_subject + md->offset_vector[offset];  register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
170    #ifdef SUPPORT_UTF
171    BOOL utf = md->utf;
172    #endif
173    
174  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
175  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 170  pchars(p, length, FALSE, md); Line 184  pchars(p, length, FALSE, md);
184  printf("\n");  printf("\n");
185  #endif  #endif
186    
187  /* Always fail if reference not set (and not JavaScript compatible). */  /* Always fail if reference not set (and not JavaScript compatible - in that
188    case the length is passed as zero). */
189    
190  if (length < 0) return -1;  if (length < 0) return -1;
191    
# Line 180  ASCII characters. */ Line 195  ASCII characters. */
195    
196  if (caseless)  if (caseless)
197    {    {
198  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
199  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
200    if (md->utf8)    if (utf)
201      {      {
202      /* Match characters up to the end of the reference. NOTE: the number of      /* Match characters up to the end of the reference. NOTE: the number of
203      bytes matched may differ, because there are some characters whose upper and      data units matched may differ, because in UTF-8 there are some characters
204      lower case versions code as different numbers of bytes. For example, U+023A      whose upper and lower case versions code have different numbers of bytes.
205      (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);      For example, U+023A (2 bytes in UTF-8) is the upper case version of U+2C65
206      a sequence of 3 of the former uses 6 bytes, as does a sequence of two of      (3 bytes in UTF-8); a sequence of 3 of the former uses 6 bytes, as does a
207      the latter. It is important, therefore, to check the length along the      sequence of two of the latter. It is important, therefore, to check the
208      reference, not along the subject (earlier code did this wrong). */      length along the reference, not along the subject (earlier code did this
209        wrong). */
210      USPTR endptr = p + length;  
211        PCRE_PUCHAR endptr = p + length;
212      while (p < endptr)      while (p < endptr)
213        {        {
214        int c, d;        pcre_uint32 c, d;
215        if (eptr >= md->end_subject) return -1;        const ucd_record *ur;
216          if (eptr >= md->end_subject) return -2;   /* Partial match */
217        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
218        GETCHARINC(d, p);        GETCHARINC(d, p);
219        if (c != d && c != UCD_OTHERCASE(d)) return -1;        ur = GET_UCD(d);
220          if (c != d && c != d + ur->other_case)
221            {
222            const pcre_uint32 *pp = PRIV(ucd_caseless_sets) + ur->caseset;
223            for (;;)
224              {
225              if (c < *pp) return -1;
226              if (c == *pp++) break;
227              }
228            }
229        }        }
230      }      }
231    else    else
# Line 209  if (caseless) Line 235  if (caseless)
235    /* 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
236    is no UCP support. */    is no UCP support. */
237      {      {
     if (eptr + length > md->end_subject) return -1;  
238      while (length-- > 0)      while (length-- > 0)
239        { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }        {
240      }        pcre_uint32 cc, cp;
241          if (eptr >= md->end_subject) return -2;   /* Partial match */
242          cc = RAWUCHARTEST(eptr);
243          cp = RAWUCHARTEST(p);
244          if (TABLE_GET(cp, md->lcc, cp) != TABLE_GET(cc, md->lcc, cc)) return -1;
245          p++;
246          eptr++;
247          }
248        }
249    }    }
250    
251  /* 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
252  are in UTF-8 mode. */  are in UTF-8 mode. */
253    
254  else  else
255    {    {
256    if (eptr + length > md->end_subject) return -1;    while (length-- > 0)
257    while (length-- > 0) if (*p++ != *eptr++) return -1;      {
258        if (eptr >= md->end_subject) return -2;   /* Partial match */
259        if (RAWUCHARINCTEST(p) != RAWUCHARINCTEST(eptr)) return -1;
260        }
261    }    }
262    
263  return eptr - eptr_start;  return (int)(eptr - eptr_start);
264  }  }
265    
266    
# Line 276  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 312  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
312         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
313         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
314         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
315         RM61,  RM62, RM63, RM64 };         RM61,  RM62, RM63, RM64, RM65, RM66, RM67, RM68 };
316    
317  /* 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
318  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 289  actually used in this definition. */ Line 325  actually used in this definition. */
325  #define RMATCH(ra,rb,rc,rd,re,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
326    { \    { \
327    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
328    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \
329    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
330    }    }
331  #define RRETURN(ra) \  #define RRETURN(ra) \
332    { \    { \
333    printf("match() returned %d from line %d ", ra, __LINE__); \    printf("match() returned %d from line %d\n", ra, __LINE__); \
334    return ra; \    return ra; \
335    }    }
336  #else  #else
337  #define RMATCH(ra,rb,rc,rd,re,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
338    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1)
339  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
340  #endif  #endif
341    
# Line 314  argument of match(), which never changes Line 350  argument of match(), which never changes
350    
351  #define RMATCH(ra,rb,rc,rd,re,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
352    {\    {\
353    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = frame->Xnextframe;\
354    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    if (newframe == NULL)\
355    frame->Xwhere = rw; \      {\
356        newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\
357        if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
358        newframe->Xnextframe = NULL;\
359        frame->Xnextframe = newframe;\
360        }\
361      frame->Xwhere = rw;\
362    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
363    newframe->Xecode = rb;\    newframe->Xecode = rb;\
364    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
   newframe->Xmarkptr = markptr;\  
365    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
366    newframe->Xeptrb = re;\    newframe->Xeptrb = re;\
367    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
# Line 336  argument of match(), which never changes Line 377  argument of match(), which never changes
377    {\    {\
378    heapframe *oldframe = frame;\    heapframe *oldframe = frame;\
379    frame = oldframe->Xprevframe;\    frame = oldframe->Xprevframe;\
   (pcre_stack_free)(oldframe);\  
380    if (frame != NULL)\    if (frame != NULL)\
381      {\      {\
382      rrc = ra;\      rrc = ra;\
# Line 350  argument of match(), which never changes Line 390  argument of match(), which never changes
390    
391  typedef struct heapframe {  typedef struct heapframe {
392    struct heapframe *Xprevframe;    struct heapframe *Xprevframe;
393      struct heapframe *Xnextframe;
394    
395    /* Function arguments that may change */    /* Function arguments that may change */
396    
397    USPTR Xeptr;    PCRE_PUCHAR Xeptr;
398    const uschar *Xecode;    const pcre_uchar *Xecode;
399    USPTR Xmstart;    PCRE_PUCHAR Xmstart;
   USPTR Xmarkptr;  
400    int Xoffset_top;    int Xoffset_top;
401    eptrblock *Xeptrb;    eptrblock *Xeptrb;
402    unsigned int Xrdepth;    unsigned int Xrdepth;
403    
404    /* Function local variables */    /* Function local variables */
405    
406    USPTR Xcallpat;    PCRE_PUCHAR Xcallpat;
407  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
408    USPTR Xcharptr;    PCRE_PUCHAR Xcharptr;
409  #endif  #endif
410    USPTR Xdata;    PCRE_PUCHAR Xdata;
411    USPTR Xnext;    PCRE_PUCHAR Xnext;
412    USPTR Xpp;    PCRE_PUCHAR Xpp;
413    USPTR Xprev;    PCRE_PUCHAR Xprev;
414    USPTR Xsaved_eptr;    PCRE_PUCHAR Xsaved_eptr;
415    
416    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
417    
# Line 381  typedef struct heapframe { Line 421  typedef struct heapframe {
421    
422  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
423    int Xprop_type;    int Xprop_type;
424    int Xprop_value;    unsigned int Xprop_value;
425    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
426    int Xoclength;    int Xoclength;
427    uschar Xocchars[8];    pcre_uchar Xocchars[6];
428  #endif  #endif
429    
430    int Xcodelink;    int Xcodelink;
# Line 397  typedef struct heapframe { Line 434  typedef struct heapframe {
434    int Xlength;    int Xlength;
435    int Xmax;    int Xmax;
436    int Xmin;    int Xmin;
437    int Xnumber;    unsigned int Xnumber;
438    int Xoffset;    int Xoffset;
439    int Xop;    unsigned int Xop;
440    int Xsave_capture_last;    pcre_int32 Xsave_capture_last;
441    int Xsave_offset1, Xsave_offset2, Xsave_offset3;    int Xsave_offset1, Xsave_offset2, Xsave_offset3;
442    int Xstacksave[REC_STACK_SAVE_MAX];    int Xstacksave[REC_STACK_SAVE_MAX];
443    
# Line 429  returns a negative (error) response, the Line 466  returns a negative (error) response, the
466  same response. */  same response. */
467    
468  /* 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
469  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
470  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.
471  something has been matched). For hard partial matching, we then return  something has been matched). For hard partial matching, we then return
472  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 440  the subject. */ Line 477  the subject. */
477        eptr > md->start_used_ptr) \        eptr > md->start_used_ptr) \
478      { \      { \
479      md->hitend = TRUE; \      md->hitend = TRUE; \
480      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
481      }      }
482    
483  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
484    if (md->partial != 0 && eptr > md->start_used_ptr) \    if (md->partial != 0 && eptr > md->start_used_ptr) \
485      { \      { \
486      md->hitend = TRUE; \      md->hitend = TRUE; \
487      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
488      }      }
489    
490    
491  /* Performance note: It might be tempting to extract commonly used fields from  /* Performance note: It might be tempting to extract commonly used fields from
492  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
493  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
494  made performance worse.  made performance worse.
495    
# Line 461  Arguments: Line 498  Arguments:
498     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
499     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
500                   by encountering \K)                   by encountering \K)
    markptr     pointer to the most recent MARK name, or NULL  
501     offset_top  current top pointer     offset_top  current top pointer
502     md          pointer to "static" info for the match     md          pointer to "static" info for the match
503     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
# Line 476  Returns:       MATCH_MATCH if matched Line 512  Returns:       MATCH_MATCH if matched
512  */  */
513    
514  static int  static int
515  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode,
516    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
517    unsigned int rdepth)    unsigned int rdepth)
518  {  {
519  /* 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,
# Line 486  so they can be ordinary variables in all Line 522  so they can be ordinary variables in all
522    
523  register int  rrc;         /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
524  register int  i;           /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
525  register unsigned int c;   /* Character values not kept over RMATCH() calls */  register pcre_uint32 c;    /* Character values not kept over RMATCH() calls */
526  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf;         /* Local copy of UTF flag for speed */
527    
528  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
529  BOOL caseless;  BOOL caseless;
530  int condcode;  int condcode;
531    
532  /* 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
533  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
534  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
535  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  whenever RMATCH() does a "recursion". See the macro definitions above. Putting
536    the top-level on the stack rather than malloc-ing them all gives a performance
537    boost in many cases where there is not much "recursion". */
538    
539  #ifdef NO_RECURSE  #ifdef NO_RECURSE
540  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 */  
541    
542  /* Copy in the original argument variables */  /* Copy in the original argument variables */
543    
544  frame->Xeptr = eptr;  frame->Xeptr = eptr;
545  frame->Xecode = ecode;  frame->Xecode = ecode;
546  frame->Xmstart = mstart;  frame->Xmstart = mstart;
 frame->Xmarkptr = markptr;  
547  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
548  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
549  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
# Line 522  HEAP_RECURSE: Line 557  HEAP_RECURSE:
557  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
558  #define ecode              frame->Xecode  #define ecode              frame->Xecode
559  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
 #define markptr            frame->Xmarkptr  
560  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
561  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
562  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
563    
564  /* Ditto for the local variables */  /* Ditto for the local variables */
565    
566  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
567  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
568  #endif  #endif
569  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 550  HEAP_RECURSE: Line 584  HEAP_RECURSE:
584  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
585  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
586  #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  
587  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
588  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
589  #endif  #endif
# Line 590  declarations can be cut out in a block. Line 621  declarations can be cut out in a block.
621  below are for variables that do not have to be preserved over a recursive call  below are for variables that do not have to be preserved over a recursive call
622  to RMATCH(). */  to RMATCH(). */
623    
624  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
625  const uschar *charptr;  const pcre_uchar *charptr;
626  #endif  #endif
627  const uschar *callpat;  const pcre_uchar *callpat;
628  const uschar *data;  const pcre_uchar *data;
629  const uschar *next;  const pcre_uchar *next;
630  USPTR         pp;  PCRE_PUCHAR       pp;
631  const uschar *prev;  const pcre_uchar *prev;
632  USPTR         saved_eptr;  PCRE_PUCHAR       saved_eptr;
633    
634  recursion_info new_recursive;  recursion_info new_recursive;
635    
636  BOOL cur_is_word;  BOOL cur_is_word;
637  BOOL condition;  BOOL condition;
638  BOOL prev_is_word;  BOOL prev_is_word;
639    
640  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
641  int prop_type;  int prop_type;
642  int prop_value;  unsigned int prop_value;
643  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
644  int oclength;  int oclength;
645  uschar occhars[8];  pcre_uchar occhars[6];
646  #endif  #endif
647    
648  int codelink;  int codelink;
# Line 622  int ctype; Line 650  int ctype;
650  int length;  int length;
651  int max;  int max;
652  int min;  int min;
653  int number;  unsigned int number;
654  int offset;  int offset;
655  int op;  unsigned int op;
656  int save_capture_last;  pcre_int32 save_capture_last;
657  int save_offset1, save_offset2, save_offset3;  int save_offset1, save_offset2, save_offset3;
658  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
659    
660  eptrblock newptrb;  eptrblock newptrb;
661    
662    /* There is a special fudge for calling match() in a way that causes it to
663    measure the size of its basic stack frame when the stack is being used for
664    recursion. The second argument (ecode) being NULL triggers this behaviour. It
665    cannot normally ever be NULL. The return is the negated value of the frame
666    size. */
667    
668    if (ecode == NULL)
669      {
670      if (rdepth == 0)
671        return match((PCRE_PUCHAR)&rdepth, NULL, NULL, 0, NULL, NULL, 1);
672      else
673        {
674        int len = (char *)&rdepth - (char *)eptr;
675        return (len > 0)? -len : len;
676        }
677      }
678  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
679    
680  /* To save space on the stack and in the heap frame, I have doubled up on some  /* To save space on the stack and in the heap frame, I have doubled up on some
681  of the local variables that are used only in localised parts of the code, but  of the local variables that are used only in localised parts of the code, but
682  still need to be preserved over recursive calls of match(). These macros define  still need to be preserved over recursive calls of match(). These macros define
683  the alternative names that are used. */  the alternative names that are used. */
684    
685  #define allow_zero    cur_is_word  #define allow_zero    cur_is_word
# Line 642  the alternative names that are used. */ Line 687  the alternative names that are used. */
687  #define code_offset   codelink  #define code_offset   codelink
688  #define condassert    condition  #define condassert    condition
689  #define matched_once  prev_is_word  #define matched_once  prev_is_word
690    #define foc           number
691    #define save_mark     data
692    
693  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
694  variables. */  variables. */
# Line 667  defined). However, RMATCH isn't like a f Line 714  defined). However, RMATCH isn't like a f
714  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,
715  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
716    
717  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
718  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
719  #else  #else
720  utf8 = FALSE;  utf = FALSE;
721  #endif  #endif
722    
723  /* 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 680  if (md->match_call_count++ >= md->match_ Line 727  if (md->match_call_count++ >= md->match_
727  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
728    
729  /* 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
730  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
731  done this way to save having to use another function argument, which would take  done this way to save having to use another function argument, which would take
732  up space on the stack. See also MATCH_CONDASSERT below.  up space on the stack. See also MATCH_CONDASSERT below.
733    
734  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
# Line 705  for (;;) Line 752  for (;;)
752    {    {
753    minimize = possessive = FALSE;    minimize = possessive = FALSE;
754    op = *ecode;    op = *ecode;
755    
756    switch(op)    switch(op)
757      {      {
758      case OP_MARK:      case OP_MARK:
759      markptr = ecode + 2;      md->nomatch_mark = ecode + 2;
760      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->mark = NULL;    /* In case previously set by assertion */
761        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
762        eptrb, RM55);        eptrb, RM55);
763        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
764             md->mark == NULL) md->mark = ecode + 2;
765    
766      /* 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
767      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 720  for (;;) Line 770  for (;;)
770      position and return MATCH_SKIP. Otherwise, pass back the return code      position and return MATCH_SKIP. Otherwise, pass back the return code
771      unaltered. */      unaltered. */
772    
773      if (rrc == MATCH_SKIP_ARG &&      else if (rrc == MATCH_SKIP_ARG &&
774          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)          STRCMP_UC_UC_TEST(ecode + 2, md->start_match_ptr) == 0)
775        {        {
776        md->start_match_ptr = eptr;        md->start_match_ptr = eptr;
777        RRETURN(MATCH_SKIP);        RRETURN(MATCH_SKIP);
778        }        }
   
     if (md->mark == NULL) md->mark = markptr;  
779      RRETURN(rrc);      RRETURN(rrc);
780    
781      case OP_FAIL:      case OP_FAIL:
782      MRRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
783    
784      /* COMMIT overrides PRUNE, SKIP, and THEN */      /* COMMIT overrides PRUNE, SKIP, and THEN */
785    
786      case OP_COMMIT:      case OP_COMMIT:
787      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
788        eptrb, RM52);        eptrb, RM52);
789      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
790          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
791          rrc != MATCH_THEN)          rrc != MATCH_THEN)
792        RRETURN(rrc);        RRETURN(rrc);
793      MRRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
794    
795      /* PRUNE overrides THEN */      /* PRUNE overrides THEN */
796    
797      case OP_PRUNE:      case OP_PRUNE:
798      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
799        eptrb, RM51);        eptrb, RM51);
800      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
801      MRRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
802    
803      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
804      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->nomatch_mark = ecode + 2;
805        md->mark = NULL;    /* In case previously set by assertion */
806        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
807        eptrb, RM56);        eptrb, RM56);
808        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
809             md->mark == NULL) md->mark = ecode + 2;
810      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
     md->mark = ecode + 2;  
811      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
812    
813      /* SKIP overrides PRUNE and THEN */      /* SKIP overrides PRUNE and THEN */
814    
815      case OP_SKIP:      case OP_SKIP:
816      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
817        eptrb, RM53);        eptrb, RM53);
818      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
819        RRETURN(rrc);        RRETURN(rrc);
820      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
821      MRRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
822    
823        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
824        nomatch_mark. There is a flag that disables this opcode when re-matching a
825        pattern that ended with a SKIP for which there was not a matching MARK. */
826    
827      case OP_SKIP_ARG:      case OP_SKIP_ARG:
828      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      if (md->ignore_skip_arg)
829          {
830          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
831          break;
832          }
833        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
834        eptrb, RM57);        eptrb, RM57);
835      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
836        RRETURN(rrc);        RRETURN(rrc);
837    
838      /* 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
839      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
840      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
841      as PRUNE. */      with the md->ignore_skip_arg flag set. */
842    
843      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
844      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
845    
846      /* 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
847      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
848      to skip back past alternatives that precede the THEN within the current      match pointer to do this. */
     branch. */  
849    
850      case OP_THEN:      case OP_THEN:
851      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
852        eptrb, RM54);        eptrb, RM54);
853      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
854      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
855      MRRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
856    
857      case OP_THEN_ARG:      case OP_THEN_ARG:
858      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      md->nomatch_mark = ecode + 2;
859        offset_top, md, eptrb, RM58);      md->mark = NULL;    /* In case previously set by assertion */
860        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
861          md, eptrb, RM58);
862        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
863             md->mark == NULL) md->mark = ecode + 2;
864      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
865      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
     md->mark = ecode + LINK_SIZE + 2;  
866      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
867    
868        /* Handle an atomic group that does not contain any capturing parentheses.
869        This can be handled like an assertion. Prior to 8.13, all atomic groups
870        were handled this way. In 8.13, the code was changed as below for ONCE, so
871        that backups pass through the group and thereby reset captured values.
872        However, this uses a lot more stack, so in 8.20, atomic groups that do not
873        contain any captures generate OP_ONCE_NC, which can be handled in the old,
874        less stack intensive way.
875    
876        Check the alternative branches in turn - the matching won't pass the KET
877        for this kind of subpattern. If any one branch matches, we carry on as at
878        the end of a normal bracket, leaving the subject pointer, but resetting
879        the start-of-match value in case it was changed by \K. */
880    
881        case OP_ONCE_NC:
882        prev = ecode;
883        saved_eptr = eptr;
884        save_mark = md->mark;
885        do
886          {
887          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
888          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
889            {
890            mstart = md->start_match_ptr;
891            break;
892            }
893          if (rrc == MATCH_THEN)
894            {
895            next = ecode + GET(ecode,1);
896            if (md->start_match_ptr < next &&
897                (*ecode == OP_ALT || *next == OP_ALT))
898              rrc = MATCH_NOMATCH;
899            }
900    
901          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
902          ecode += GET(ecode,1);
903          md->mark = save_mark;
904          }
905        while (*ecode == OP_ALT);
906    
907        /* If hit the end of the group (which could be repeated), fail */
908    
909        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
910    
911        /* Continue as from after the group, updating the offsets high water
912        mark, since extracts may have been taken. */
913    
914        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
915    
916        offset_top = md->end_offset_top;
917        eptr = md->end_match_ptr;
918    
919        /* For a non-repeating ket, just continue at this level. This also
920        happens for a repeating ket if no characters were matched in the group.
921        This is the forcible breaking of infinite loops as implemented in Perl
922        5.005. */
923    
924        if (*ecode == OP_KET || eptr == saved_eptr)
925          {
926          ecode += 1+LINK_SIZE;
927          break;
928          }
929    
930        /* The repeating kets try the rest of the pattern or restart from the
931        preceding bracket, in the appropriate order. The second "call" of match()
932        uses tail recursion, to avoid using another stack frame. */
933    
934        if (*ecode == OP_KETRMIN)
935          {
936          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
937          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
938          ecode = prev;
939          goto TAIL_RECURSE;
940          }
941        else  /* OP_KETRMAX */
942          {
943          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
944          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
945          ecode += 1 + LINK_SIZE;
946          goto TAIL_RECURSE;
947          }
948        /* Control never gets here */
949    
950      /* Handle a capturing bracket, other than those that are possessive with an      /* Handle a capturing bracket, other than those that are possessive with an
951      unlimited repeat. If there is space in the offset vector, save the current      unlimited repeat. If there is space in the offset vector, save the current
952      subject position in the working slot at the top of the vector. We mustn't      subject position in the working slot at the top of the vector. We mustn't
953      change the current values of the data slot, because they may be set from a      change the current values of the data slot, because they may be set from a
954      previous iteration of this group, and be referred to by a reference inside      previous iteration of this group, and be referred to by a reference inside
955      the group. If we fail to match, we need to restore this value and also the      the group. A failure to match might occur after the group has succeeded,
956      values of the final offsets, in case they were set by a previous iteration      if something later on doesn't match. For this reason, we need to restore
957      of the same bracket.      the working value and also the values of the final offsets, in case they
958        were set by a previous iteration of the same bracket.
959    
960      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
961      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 820  for (;;) Line 965  for (;;)
965      case OP_SCBRA:      case OP_SCBRA:
966      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
967      offset = number << 1;      offset = number << 1;
968    
969  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
970      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
971      printf("subject=");      printf("subject=");
# Line 834  for (;;) Line 979  for (;;)
979        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
980        save_offset3 = md->offset_vector[md->offset_end - number];        save_offset3 = md->offset_vector[md->offset_end - number];
981        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
982          save_mark = md->mark;
983    
984        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
985        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
# Line 841  for (;;) Line 987  for (;;)
987    
988        for (;;)        for (;;)
989          {          {
990          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
991          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
992            eptrb, RM1);            eptrb, RM1);
993          if (rrc != MATCH_NOMATCH &&          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
994              (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
995            RRETURN(rrc);          /* If we backed up to a THEN, check whether it is within the current
996            branch by comparing the address of the THEN that is passed back with
997            the end of the branch. If it is within the current branch, and the
998            branch is one of two or more alternatives (it either starts or ends
999            with OP_ALT), we have reached the limit of THEN's action, so convert
1000            the return code to NOMATCH, which will cause normal backtracking to
1001            happen from now on. Otherwise, THEN is passed back to an outer
1002            alternative. This implements Perl's treatment of parenthesized groups,
1003            where a group not containing | does not affect the current alternative,
1004            that is, (X) is NOT the same as (X|(*F)). */
1005    
1006            if (rrc == MATCH_THEN)
1007              {
1008              next = ecode + GET(ecode,1);
1009              if (md->start_match_ptr < next &&
1010                  (*ecode == OP_ALT || *next == OP_ALT))
1011                rrc = MATCH_NOMATCH;
1012              }
1013    
1014            /* Anything other than NOMATCH is passed back. */
1015    
1016            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1017          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
1018          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
1019          if (*ecode != OP_ALT) break;          md->mark = save_mark;
1020            if (*ecode != OP_ALT) break;
1021          }          }
1022    
1023        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
1024        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
1025        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
1026        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
1027    
1028        if (rrc != MATCH_THEN) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
1029        RRETURN(MATCH_NOMATCH);  
1030          RRETURN(rrc);
1031        }        }
1032    
1033      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 873  for (;;) Line 1041  for (;;)
1041      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1042      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1043    
1044      /* Non-capturing bracket, except for possessive with unlimited repeat. Loop      /* Non-capturing or atomic group, except for possessive with unlimited
1045      for all the alternatives. When we get to the final alternative within the      repeat and ONCE group with no captures. Loop for all the alternatives.
1046      brackets, we would return the result of a recursive call to match()  
1047      whatever happened. We can reduce stack usage by turning this into a tail      When we get to the final alternative within the brackets, we used to return
1048      recursion, except in the case of a possibly empty group.*/      the result of a recursive call to match() whatever happened so it was
1049        possible to reduce stack usage by turning this into a tail recursion,
1050        except in the case of a possibly empty group. However, now that there is
1051        the possiblity of (*THEN) occurring in the final alternative, this
1052        optimization is no longer always possible.
1053    
1054        We can optimize if we know there are no (*THEN)s in the pattern; at present
1055        this is the best that can be done.
1056    
1057        MATCH_ONCE is returned when the end of an atomic group is successfully
1058        reached, but subsequent matching fails. It passes back up the tree (causing
1059        captured values to be reset) until the original atomic group level is
1060        reached. This is tested by comparing md->once_target with the start of the
1061        group. At this point, the return is converted into MATCH_NOMATCH so that
1062        previous backup points can be taken. */
1063    
1064        case OP_ONCE:
1065      case OP_BRA:      case OP_BRA:
1066      case OP_SBRA:      case OP_SBRA:
1067      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
1068    
1069      for (;;)      for (;;)
1070        {        {
1071        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE)
1072            md->match_function_type = MATCH_CBEGROUP;
1073    
1074          /* If this is not a possibly empty group, and there are no (*THEN)s in
1075          the pattern, and this is the final alternative, optimize as described
1076          above. */
1077    
1078          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1079          {          {
1080          if (op >= OP_SBRA)   /* Possibly empty group */          ecode += PRIV(OP_lengths)[*ecode];
           {  
           md->match_function_type = MATCH_CBEGROUP;  
           RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,  
             RM48);  
           if (rrc == MATCH_NOMATCH) md->mark = markptr;  
           RRETURN(rrc);  
           }  
         /* Not a possibly empty group; use tail recursion */  
         ecode += _pcre_OP_lengths[*ecode];  
         DPRINTF(("bracket 0 tail recursion\n"));  
1081          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1082          }          }
1083    
1084        /* For non-final alternatives, continue the loop for a NOMATCH result;        /* In all other cases, we have to make another call to match(). */
       otherwise return. */  
1085    
1086        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        save_mark = md->mark;
1087        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,        save_capture_last = md->capture_last;
1088          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1089          RM2);          RM2);
1090        if (rrc != MATCH_NOMATCH &&  
1091            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* See comment in the code for capturing groups above about handling
1092          THEN. */
1093    
1094          if (rrc == MATCH_THEN)
1095            {
1096            next = ecode + GET(ecode,1);
1097            if (md->start_match_ptr < next &&
1098                (*ecode == OP_ALT || *next == OP_ALT))
1099              rrc = MATCH_NOMATCH;
1100            }
1101    
1102          if (rrc != MATCH_NOMATCH)
1103            {
1104            if (rrc == MATCH_ONCE)
1105              {
1106              const pcre_uchar *scode = ecode;
1107              if (*scode != OP_ONCE)           /* If not at start, find it */
1108                {
1109                while (*scode == OP_ALT) scode += GET(scode, 1);
1110                scode -= GET(scode, 1);
1111                }
1112              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1113              }
1114          RRETURN(rrc);          RRETURN(rrc);
1115            }
1116        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1117          md->mark = save_mark;
1118          if (*ecode != OP_ALT) break;
1119          md->capture_last = save_capture_last;
1120        }        }
     /* Control never reaches here. */  
1121    
1122      /* Handle possessive capturing brackets with an unlimited repeat. We come      RRETURN(MATCH_NOMATCH);
1123    
1124        /* Handle possessive capturing brackets with an unlimited repeat. We come
1125      here from BRAZERO with allow_zero set TRUE. The offset_vector values are      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1126      handled similarly to the normal case above. However, the matching is      handled similarly to the normal case above. However, the matching is
1127      different. The end of these brackets will always be OP_KETRPOS, which      different. The end of these brackets will always be OP_KETRPOS, which
1128      returns MATCH_KETRPOS without going further in the pattern. By this means      returns MATCH_KETRPOS without going further in the pattern. By this means
1129      we can handle the group by iteration rather than recursion, thereby      we can handle the group by iteration rather than recursion, thereby
1130      reducing the amount of stack needed. */      reducing the amount of stack needed. */
1131    
1132      case OP_CBRAPOS:      case OP_CBRAPOS:
1133      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1134      allow_zero = FALSE;      allow_zero = FALSE;
1135    
1136      POSSESSIVE_CAPTURE:      POSSESSIVE_CAPTURE:
1137      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
1138      offset = number << 1;      offset = number << 1;
# Line 939  for (;;) Line 1147  for (;;)
1147      if (offset < md->offset_max)      if (offset < md->offset_max)
1148        {        {
1149        matched_once = FALSE;        matched_once = FALSE;
1150        code_offset = ecode - md->start_code;        code_offset = (int)(ecode - md->start_code);
1151    
1152        save_offset1 = md->offset_vector[offset];        save_offset1 = md->offset_vector[offset];
1153        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
# Line 947  for (;;) Line 1155  for (;;)
1155        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
1156    
1157        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1158    
1159        /* Each time round the loop, save the current subject position for use        /* Each time round the loop, save the current subject position for use
1160        when the group matches. For MATCH_MATCH, the group has matched, so we        when the group matches. For MATCH_MATCH, the group has matched, so we
1161        restart it with a new subject starting position, remembering that we had        restart it with a new subject starting position, remembering that we had
1162        at least one match. For MATCH_NOMATCH, carry on with the alternatives, as        at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1163        usual. If we haven't matched any alternatives in any iteration, check to        usual. If we haven't matched any alternatives in any iteration, check to
1164        see if a previous iteration matched. If so, the group has matched;        see if a previous iteration matched. If so, the group has matched;
1165        continue from afterwards. Otherwise it has failed; restore the previous        continue from afterwards. Otherwise it has failed; restore the previous
1166        capture values before returning NOMATCH. */        capture values before returning NOMATCH. */
1167    
1168        for (;;)        for (;;)
1169          {          {
1170          md->offset_vector[md->offset_end - number] =          md->offset_vector[md->offset_end - number] =
1171            (int)(eptr - md->start_subject);            (int)(eptr - md->start_subject);
1172          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1173          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1174            eptrb, RM63);            eptrb, RM63);
1175          if (rrc == MATCH_KETRPOS)          if (rrc == MATCH_KETRPOS)
1176            {            {
1177            offset_top = md->end_offset_top;            offset_top = md->end_offset_top;
1178            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
1179            ecode = md->start_code + code_offset;            ecode = md->start_code + code_offset;
1180            save_capture_last = md->capture_last;            save_capture_last = md->capture_last;
1181            matched_once = TRUE;            matched_once = TRUE;
1182            continue;            continue;
1183            }            }
1184          if (rrc != MATCH_NOMATCH &&  
1185              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* See comment in the code for capturing groups above about handling
1186            RRETURN(rrc);          THEN. */
1187    
1188            if (rrc == MATCH_THEN)
1189              {
1190              next = ecode + GET(ecode,1);
1191              if (md->start_match_ptr < next &&
1192                  (*ecode == OP_ALT || *next == OP_ALT))
1193                rrc = MATCH_NOMATCH;
1194              }
1195    
1196            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1197          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
1198          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
1199          if (*ecode != OP_ALT) break;          if (*ecode != OP_ALT) break;
1200          }          }
1201    
1202        if (!matched_once)        if (!matched_once)
1203          {          {
1204          md->offset_vector[offset] = save_offset1;          md->offset_vector[offset] = save_offset1;
1205          md->offset_vector[offset+1] = save_offset2;          md->offset_vector[offset+1] = save_offset2;
1206          md->offset_vector[md->offset_end - number] = save_offset3;          md->offset_vector[md->offset_end - number] = save_offset3;
1207          }          }
1208    
       if (rrc != MATCH_THEN) md->mark = markptr;  
1209        if (allow_zero || matched_once)        if (allow_zero || matched_once)
1210          {          {
1211          ecode += 1 + LINK_SIZE;          ecode += 1 + LINK_SIZE;
1212          break;          break;
1213          }          }
1214    
1215        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1216        }        }
1217    
1218      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1219      as a non-capturing bracket. */      as a non-capturing bracket. */
1220    
# Line 1009  for (;;) Line 1226  for (;;)
1226      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1227      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1228    
1229      /* Non-capturing possessive bracket with unlimited repeat. We come here      /* Non-capturing possessive bracket with unlimited repeat. We come here
1230      from BRAZERO with allow_zero = TRUE. The code is similar to the above,      from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1231      without the capturing complication. It is written out separately for speed      without the capturing complication. It is written out separately for speed
1232      and cleanliness. */      and cleanliness. */
1233    
1234      case OP_BRAPOS:      case OP_BRAPOS:
1235      case OP_SBRAPOS:      case OP_SBRAPOS:
1236      allow_zero = FALSE;      allow_zero = FALSE;
1237    
1238      POSSESSIVE_NON_CAPTURE:      POSSESSIVE_NON_CAPTURE:
1239      matched_once = FALSE;      matched_once = FALSE;
1240      code_offset = ecode - md->start_code;      code_offset = (int)(ecode - md->start_code);
1241        save_capture_last = md->capture_last;
1242    
1243      for (;;)      for (;;)
1244        {        {
1245        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1246        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1247          eptrb, RM64);          eptrb, RM48);
1248        if (rrc == MATCH_KETRPOS)        if (rrc == MATCH_KETRPOS)
1249          {          {
1250            offset_top = md->end_offset_top;
1251          eptr = md->end_match_ptr;          eptr = md->end_match_ptr;
1252          ecode = md->start_code + code_offset;          ecode = md->start_code + code_offset;
1253          matched_once = TRUE;          matched_once = TRUE;
1254          continue;          continue;
1255          }          }
1256        if (rrc != MATCH_NOMATCH &&  
1257            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* See comment in the code for capturing groups above about handling
1258          RRETURN(rrc);        THEN. */
1259    
1260          if (rrc == MATCH_THEN)
1261            {
1262            next = ecode + GET(ecode,1);
1263            if (md->start_match_ptr < next &&
1264                (*ecode == OP_ALT || *next == OP_ALT))
1265              rrc = MATCH_NOMATCH;
1266            }
1267    
1268          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1269        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1270        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1271          md->capture_last = save_capture_last;
1272        }        }
1273    
1274      if (matched_once || allow_zero)      if (matched_once || allow_zero)
1275        {        {
1276        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1277        break;        break;
1278        }        }
1279      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1280    
1281      /* Control never reaches here. */      /* Control never reaches here. */
# Line 1053  for (;;) Line 1283  for (;;)
1283      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1284      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
1285      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
1286      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. */  
1287    
1288      case OP_COND:      case OP_COND:
1289      case OP_SCOND:      case OP_SCOND:
# Line 1065  for (;;) Line 1294  for (;;)
1294    
1295      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1296        {        {
1297        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1298          {          {
1299          pcre_callout_block cb;          PUBL(callout_block) cb;
1300          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1301          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1302          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1303    #if defined COMPILE_PCRE8
1304          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1305    #elif defined COMPILE_PCRE16
1306            cb.subject          = (PCRE_SPTR16)md->start_subject;
1307    #elif defined COMPILE_PCRE32
1308            cb.subject          = (PCRE_SPTR32)md->start_subject;
1309    #endif
1310          cb.subject_length   = (int)(md->end_subject - md->start_subject);          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1311          cb.start_match      = (int)(mstart - md->start_subject);          cb.start_match      = (int)(mstart - md->start_subject);
1312          cb.current_position = (int)(eptr - md->start_subject);          cb.current_position = (int)(eptr - md->start_subject);
1313          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1314          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1315          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1316          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last & CAPLMASK;
1317            /* Internal change requires this for API compatibility. */
1318            if (cb.capture_last == 0) cb.capture_last = -1;
1319          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1320          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1321            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1322          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1323          }          }
1324        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1325          codelink -= PRIV(OP_lengths)[OP_CALLOUT];
1326        }        }
1327    
1328      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
# Line 1099  for (;;) Line 1338  for (;;)
1338          }          }
1339        else        else
1340          {          {
1341          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          unsigned int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1342          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1343    
1344          /* 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
1345          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
1346          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
1347          if any one is set. */          if any one is set. */
1348    
1349          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1350            {            {
1351            uschar *slotA = md->name_table;            pcre_uchar *slotA = md->name_table;
1352            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
1353              {              {
1354              if (GET2(slotA, 0) == recno) break;              if (GET2(slotA, 0) == recno) break;
# Line 1122  for (;;) Line 1361  for (;;)
1361    
1362            if (i < md->name_count)            if (i < md->name_count)
1363              {              {
1364              uschar *slotB = slotA;              pcre_uchar *slotB = slotA;
1365              while (slotB > md->name_table)              while (slotB > md->name_table)
1366                {                {
1367                slotB -= md->name_entry_size;                slotB -= md->name_entry_size;
1368                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1369                  {                  {
1370                  condition = GET2(slotB, 0) == md->recursive->group_num;                  condition = GET2(slotB, 0) == md->recursive->group_num;
1371                  if (condition) break;                  if (condition) break;
# Line 1142  for (;;) Line 1381  for (;;)
1381                for (i++; i < md->name_count; i++)                for (i++; i < md->name_count; i++)
1382                  {                  {
1383                  slotB += md->name_entry_size;                  slotB += md->name_entry_size;
1384                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1385                    {                    {
1386                    condition = GET2(slotB, 0) == md->recursive->group_num;                    condition = GET2(slotB, 0) == md->recursive->group_num;
1387                    if (condition) break;                    if (condition) break;
# Line 1155  for (;;) Line 1394  for (;;)
1394    
1395          /* Chose branch according to the condition */          /* Chose branch according to the condition */
1396    
1397          ecode += condition? 3 : GET(ecode, 1);          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1398          }          }
1399        }        }
1400    
# Line 1171  for (;;) Line 1410  for (;;)
1410    
1411        if (!condition && condcode == OP_NCREF)        if (!condition && condcode == OP_NCREF)
1412          {          {
1413          int refno = offset >> 1;          unsigned int refno = offset >> 1;
1414          uschar *slotA = md->name_table;          pcre_uchar *slotA = md->name_table;
1415    
1416          for (i = 0; i < md->name_count; i++)          for (i = 0; i < md->name_count; i++)
1417            {            {
# Line 1186  for (;;) Line 1425  for (;;)
1425    
1426          if (i < md->name_count)          if (i < md->name_count)
1427            {            {
1428            uschar *slotB = slotA;            pcre_uchar *slotB = slotA;
1429            while (slotB > md->name_table)            while (slotB > md->name_table)
1430              {              {
1431              slotB -= md->name_entry_size;              slotB -= md->name_entry_size;
1432              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)              if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1433                {                {
1434                offset = GET2(slotB, 0) << 1;                offset = GET2(slotB, 0) << 1;
1435                condition = offset < offset_top &&                condition = offset < offset_top &&
# Line 1208  for (;;) Line 1447  for (;;)
1447              for (i++; i < md->name_count; i++)              for (i++; i < md->name_count; i++)
1448                {                {
1449                slotB += md->name_entry_size;                slotB += md->name_entry_size;
1450                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1451                  {                  {
1452                  offset = GET2(slotB, 0) << 1;                  offset = GET2(slotB, 0) << 1;
1453                  condition = offset < offset_top &&                  condition = offset < offset_top &&
# Line 1223  for (;;) Line 1462  for (;;)
1462    
1463        /* Chose branch according to the condition */        /* Chose branch according to the condition */
1464    
1465        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1466        }        }
1467    
1468      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 1238  for (;;) Line 1477  for (;;)
1477    
1478      else      else
1479        {        {
1480        md->match_function_type = MATCH_CONDASSERT;        md->match_function_type = MATCH_CONDASSERT;
1481        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1482        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1483          {          {
1484            if (md->end_offset_top > offset_top)
1485              offset_top = md->end_offset_top;  /* Captures may have happened */
1486          condition = TRUE;          condition = TRUE;
1487          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1488          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1489          }          }
1490        else if (rrc != MATCH_NOMATCH &&  
1491                (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1492          assertion; it is therefore treated as NOMATCH. */
1493    
1494          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1495          {          {
1496          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1497          }          }
# Line 1258  for (;;) Line 1502  for (;;)
1502          }          }
1503        }        }
1504    
1505      /* 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
1506      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
1507      we have an unlimited repeat of a possibly empty group. If the second      unlimited repeat of a possibly empty group. In the latter case, a recursive
1508      alternative doesn't exist, we can just plough on. */      call to match() is always required, unless the second alternative doesn't
1509        exist, in which case we can just plough on. Note that, for compatibility
1510        with Perl, the | in a conditional group is NOT treated as creating two
1511        alternatives. If a THEN is encountered in the branch, it propagates out to
1512        the enclosing alternative (unless nested in a deeper set of alternatives,
1513        of course). */
1514    
1515      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1516        {        {
1517        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
1518          {          {
1519          md->match_function_type = MATCH_CBEGROUP;          ecode += 1 + LINK_SIZE;
1520          RMATCH(eptr, ecode, offset_top, md, eptrb, RM49);          goto TAIL_RECURSE;
         RRETURN(rrc);  
1521          }          }
1522        else goto TAIL_RECURSE;  
1523          md->match_function_type = MATCH_CBEGROUP;
1524          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1525          RRETURN(rrc);
1526        }        }
1527      else                         /* Condition false & no alternative */  
1528         /* Condition false & no alternative; continue after the group. */
1529    
1530        else
1531        {        {
1532        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1533        }        }
# Line 1285  for (;;) Line 1538  for (;;)
1538      to close any currently open capturing brackets. */      to close any currently open capturing brackets. */
1539    
1540      case OP_CLOSE:      case OP_CLOSE:
1541      number = GET2(ecode, 1);      number = GET2(ecode, 1);   /* Must be less than 65536 */
1542      offset = number << 1;      offset = number << 1;
1543    
1544  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 1293  for (;;) Line 1546  for (;;)
1546        printf("\n");        printf("\n");
1547  #endif  #endif
1548    
1549      md->capture_last = number;      md->capture_last = (md->capture_last & OVFLMASK) | number;
1550      if (offset >= md->offset_max) md->offset_overflow = TRUE; else      if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else
1551        {        {
1552        md->offset_vector[offset] =        md->offset_vector[offset] =
1553          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1554        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1555        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1556        }        }
1557      ecode += 3;      ecode += 1 + IMM2_SIZE;
1558      break;      break;
1559    
1560    
1561      /* End of the pattern, either real or forced. If we are in a recursion, we      /* End of the pattern, either real or forced. */
     should restore the offsets appropriately, and if it's a top-level  
     recursion, continue from after the call. */  
1562    
     case OP_ACCEPT:  
1563      case OP_END:      case OP_END:
1564      if (md->recursive != NULL)      case OP_ACCEPT:
1565        {      case OP_ASSERT_ACCEPT:
1566        recursion_info *rec = md->recursive;  
1567        md->recursive = rec->prevrec;      /* If we have matched an empty string, fail if not in an assertion and not
1568        memmove(md->offset_vector, rec->offset_save,      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1569          rec->saved_max * sizeof(int));      is set and we have matched at the start of the subject. In both cases,
1570        offset_top = rec->save_offset_top;      backtracking will then try other alternatives, if any. */
1571        if (rec->group_num == 0)  
1572          {      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1573          ecode = rec->after_call;           md->recursive == NULL &&
1574          break;           (md->notempty ||
1575          }             (md->notempty_atstart &&
1576        }               mstart == md->start_subject + md->start_offset)))
1577          RRETURN(MATCH_NOMATCH);
     /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is  
     set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of  
     the subject. In both cases, backtracking will then try other alternatives,  
     if any. */  
   
     else if (eptr == mstart &&  
         (md->notempty ||  
           (md->notempty_atstart &&  
             mstart == md->start_subject + md->start_offset)))  
       MRRETURN(MATCH_NOMATCH);  
1578    
1579      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1580    
1581      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1582      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1583      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1584    
1585      /* 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
1586      given as the argument to MRRETURN when the heap is in use. */      given as the argument to RRETURN when the heap is in use. */
1587    
1588      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1589      MRRETURN(rrc);      RRETURN(rrc);
1590    
1591      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1592      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,
1593      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
1594      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
1595      this level is identical to the lookahead case. When the assertion is part      this level is identical to the lookahead case. When the assertion is part
1596      of a condition, we want to return immediately afterwards. The caller of      of a condition, we want to return immediately afterwards. The caller of
1597      this incarnation of the match() function will have set MATCH_CONDASSERT in      this incarnation of the match() function will have set MATCH_CONDASSERT in
1598      md->match_function type, and one of these opcodes will be the first opcode      md->match_function type, and one of these opcodes will be the first opcode
1599      that is processed. We use a local variable that is preserved over calls to      that is processed. We use a local variable that is preserved over calls to
1600      match() to remember this case. */      match() to remember this case. */
1601    
1602      case OP_ASSERT:      case OP_ASSERT:
1603      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1604        save_mark = md->mark;
1605      if (md->match_function_type == MATCH_CONDASSERT)      if (md->match_function_type == MATCH_CONDASSERT)
1606        {        {
1607        condassert = TRUE;        condassert = TRUE;
1608        md->match_function_type = 0;        md->match_function_type = 0;
1609        }        }
1610      else condassert = FALSE;      else condassert = FALSE;
1611    
1612      do      do
1613        {        {
1614        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
# Line 1376  for (;;) Line 1617  for (;;)
1617          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1618          break;          break;
1619          }          }
1620        if (rrc != MATCH_NOMATCH &&        md->mark = save_mark;
1621            (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
1622          RRETURN(rrc);        /* A COMMIT failure must fail the entire assertion, without trying any
1623          subsequent branches. */
1624    
1625          if (rrc == MATCH_COMMIT) RRETURN(MATCH_NOMATCH);
1626    
1627          /* PCRE does not allow THEN to escape beyond an assertion; it
1628          is treated as NOMATCH. */
1629    
1630          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1631        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1632        }        }
1633      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1634    
1635      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1636    
1637      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1638    
# Line 1403  for (;;) Line 1652  for (;;)
1652    
1653      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1654      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1655        save_mark = md->mark;
1656      if (md->match_function_type == MATCH_CONDASSERT)      if (md->match_function_type == MATCH_CONDASSERT)
1657        {        {
1658        condassert = TRUE;        condassert = TRUE;
1659        md->match_function_type = 0;        md->match_function_type = 0;
1660        }        }
1661      else condassert = FALSE;      else condassert = FALSE;
1662    
1663      do      do
1664        {        {
1665        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1666        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        md->mark = save_mark;
1667          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1668        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1669          {          {
1670          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1671          break;          break;
1672          }          }
1673        if (rrc != MATCH_NOMATCH &&  
1674            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1675          RRETURN(rrc);        as NOMATCH. */
1676    
1677          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1678        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1679        }        }
1680      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1681    
1682      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1683    
1684      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1685      continue;      continue;
1686    
# Line 1437  for (;;) Line 1690  for (;;)
1690      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1691    
1692      case OP_REVERSE:      case OP_REVERSE:
1693  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1694      if (utf8)      if (utf)
1695        {        {
1696        i = GET(ecode, 1);        i = GET(ecode, 1);
1697        while (i-- > 0)        while (i-- > 0)
1698          {          {
1699          eptr--;          eptr--;
1700          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1701          BACKCHAR(eptr);          BACKCHAR(eptr);
1702          }          }
1703        }        }
# Line 1455  for (;;) Line 1708  for (;;)
1708    
1709        {        {
1710        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1711        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1712        }        }
1713    
1714      /* Save the earliest consulted character, then skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
# Line 1469  for (;;) Line 1722  for (;;)
1722      function is able to force a failure. */      function is able to force a failure. */
1723    
1724      case OP_CALLOUT:      case OP_CALLOUT:
1725      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1726        {        {
1727        pcre_callout_block cb;        PUBL(callout_block) cb;
1728        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1729        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1730        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1731    #if defined COMPILE_PCRE8
1732        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1733    #elif defined COMPILE_PCRE16
1734          cb.subject          = (PCRE_SPTR16)md->start_subject;
1735    #elif defined COMPILE_PCRE32
1736          cb.subject          = (PCRE_SPTR32)md->start_subject;
1737    #endif
1738        cb.subject_length   = (int)(md->end_subject - md->start_subject);        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1739        cb.start_match      = (int)(mstart - md->start_subject);        cb.start_match      = (int)(mstart - md->start_subject);
1740        cb.current_position = (int)(eptr - md->start_subject);        cb.current_position = (int)(eptr - md->start_subject);
1741        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1742        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1743        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1744        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last & CAPLMASK;
1745          /* Internal change requires this for API compatibility. */
1746          if (cb.capture_last == 0) cb.capture_last = -1;
1747        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1748        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1749          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1750        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1751        }        }
1752      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1494  for (;;) Line 1756  for (;;)
1756      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
1757      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1758    
1759      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1760      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
1761      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
1762      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
1763      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
1764      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
1765      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.  
1766    
1767      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
1768      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
1769      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1770        a lot, so he is not to blame for the current way it works. */
1771    
1772      case OP_RECURSE:      case OP_RECURSE:
1773        {        {
1774          recursion_info *ri;
1775          unsigned int recno;
1776    
1777        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1778        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1779          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1780    
1781          /* Check for repeating a recursion without advancing the subject pointer.
1782          This should catch convoluted mutual recursions. (Some simple cases are
1783          caught at compile time.) */
1784    
1785          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1786            if (recno == ri->group_num && eptr == ri->subject_position)
1787              RRETURN(PCRE_ERROR_RECURSELOOP);
1788    
1789        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1790    
1791          new_recursive.group_num = recno;
1792          new_recursive.saved_capture_last = md->capture_last;
1793          new_recursive.subject_position = eptr;
1794        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1795        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1796    
1797        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1798    
1799        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1800    
1801        /* Now save the offset data. */        /* Now save the offset data */
1802    
1803        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1804        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1533  for (;;) Line 1806  for (;;)
1806        else        else
1807          {          {
1808          new_recursive.offset_save =          new_recursive.offset_save =
1809            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1810          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1811          }          }
   
1812        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1813              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1814        new_recursive.save_offset_top = offset_top;  
1815          /* OK, now we can do the recursion. After processing each alternative,
1816        /* OK, now we can do the recursion. For each top-level alternative we        restore the offset data and the last captured value. If there were nested
1817        restore the offset and recursion data. */        recursions, md->recursive might be changed, so reset it before looping.
1818          */
1819    
1820        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1821        cbegroup = (*callpat >= OP_SBRA);        cbegroup = (*callpat >= OP_SBRA);
1822        do        do
1823          {          {
1824          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1825          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1826            md, eptrb, RM6);            md, eptrb, RM6);
1827            memcpy(md->offset_vector, new_recursive.offset_save,
1828                new_recursive.saved_max * sizeof(int));
1829            md->capture_last = new_recursive.saved_capture_last;
1830            md->recursive = new_recursive.prevrec;
1831          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1832            {            {
1833            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1834            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1835              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1836            MRRETURN(MATCH_MATCH);  
1837              /* Set where we got to in the subject, and reset the start in case
1838              it was changed by \K. This *is* propagated back out of a recursion,
1839              for Perl compatibility. */
1840    
1841              eptr = md->end_match_ptr;
1842              mstart = md->start_match_ptr;
1843              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1844            }            }
1845          else if (rrc != MATCH_NOMATCH &&  
1846                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* PCRE does not allow THEN, SKIP, PRUNE or COMMIT to escape beyond a
1847            recursion; they are treated as NOMATCH. These codes are defined in a
1848            range that can be tested for. Any other return code is an error. */
1849    
1850            else if (rrc != MATCH_NOMATCH &&
1851                     (rrc < MATCH_BACKTRACK_MIN || rrc > MATCH_BACKTRACK_MAX))
1852            {            {
1853            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1854            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1855              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1856            RRETURN(rrc);            RRETURN(rrc);
1857            }            }
1858    
1859          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1860          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1861          }          }
1862        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1578  for (;;) Line 1864  for (;;)
1864        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1865        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1866        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1867          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1868        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, eptrb, 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;  
1869        }        }
1870    
1871      /* The repeating kets try the rest of the pattern or restart from the      RECURSION_MATCHED:
1872      preceding bracket, in the appropriate order. The second "call" of match()      break;
     uses tail recursion, to avoid using another stack frame. */  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       md->match_function_type = MATCH_CBEGROUP;  
       RMATCH(eptr, prev, offset_top, md, eptrb, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
1873    
1874      /* 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
1875      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1667  for (;;) Line 1883  for (;;)
1883      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1884      with fixed upper repeat limits are compiled as a number of copies, with the      with fixed upper repeat limits are compiled as a number of copies, with the
1885      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1886    
1887      case OP_BRAZERO:      case OP_BRAZERO:
1888      next = ecode + 1;      next = ecode + 1;
1889      RMATCH(eptr, next, offset_top, md, eptrb, RM10);      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
# Line 1675  for (;;) Line 1891  for (;;)
1891      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
1892      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1893      break;      break;
1894    
1895      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1896      next = ecode + 1;      next = ecode + 1;
1897      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
# Line 1689  for (;;) Line 1905  for (;;)
1905      do next += GET(next,1); while (*next == OP_ALT);      do next += GET(next,1); while (*next == OP_ALT);
1906      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1907      break;      break;
1908    
1909      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1910      here; just jump to the group, with allow_zero set TRUE. */      here; just jump to the group, with allow_zero set TRUE. */
1911    
1912      case OP_BRAPOSZERO:      case OP_BRAPOSZERO:
1913      op = *(++ecode);      op = *(++ecode);
1914      allow_zero = TRUE;      allow_zero = TRUE;
1915      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1916        goto POSSESSIVE_NON_CAPTURE;        goto POSSESSIVE_NON_CAPTURE;
# Line 1704  for (;;) Line 1920  for (;;)
1920      case OP_KET:      case OP_KET:
1921      case OP_KETRMIN:      case OP_KETRMIN:
1922      case OP_KETRMAX:      case OP_KETRMAX:
1923      case OP_KETRPOS:      case OP_KETRPOS:
1924      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1925    
1926      /* 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
1927      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1928      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1929    
1930      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1931        {        {
1932        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1933        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1934        }        }
1935      else saved_eptr = NULL;      else saved_eptr = NULL;
1936    
1937      /* 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
1938      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
1939      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
1940      it was changed by \K. */      start in case it was changed by \K. */
1941    
1942      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1943          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||           *prev == OP_ONCE_NC)
         *prev == OP_ONCE)  
1944        {        {
1945        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1946        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1947        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1948        MRRETURN(MATCH_MATCH);        RRETURN(MATCH_MATCH);         /* Sets md->mark */
1949        }        }
1950    
1951      /* 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
1952      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1953      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
1954      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
1955      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
1956        the current subject position and start match pointer and give a MATCH
1957        return. */
1958    
1959      if (*prev == OP_CBRA || *prev == OP_SCBRA ||      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1960          *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)          *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
# Line 1750  for (;;) Line 1967  for (;;)
1967        printf("\n");        printf("\n");
1968  #endif  #endif
1969    
1970        md->capture_last = number;        /* Handle a recursively called group. */
1971        if (offset >= md->offset_max) md->offset_overflow = TRUE; else  
1972          if (md->recursive != NULL && md->recursive->group_num == number)
1973          {          {
1974          md->offset_vector[offset] =          md->end_match_ptr = eptr;
1975            md->offset_vector[md->offset_end - number];          md->start_match_ptr = mstart;
1976          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          RRETURN(MATCH_MATCH);
         if (offset_top <= offset) offset_top = offset + 2;  
1977          }          }
1978    
1979        /* Handle a recursively called group. Restore the offsets        /* Deal with capturing */
       appropriately and continue from after the call. */  
1980    
1981        if (md->recursive != NULL && md->recursive->group_num == number)        md->capture_last = (md->capture_last & OVFLMASK) | number;
1982          if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else
1983          {          {
1984          recursion_info *rec = md->recursive;          /* If offset is greater than offset_top, it means that we are
1985          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          "skipping" a capturing group, and that group's offsets must be marked
1986          md->recursive = rec->prevrec;          unset. In earlier versions of PCRE, all the offsets were unset at the
1987          memcpy(md->offset_vector, rec->offset_save,          start of matching, but this doesn't work because atomic groups and
1988            rec->saved_max * sizeof(int));          assertions can cause a value to be set that should later be unset.
1989          offset_top = rec->save_offset_top;          Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1990          ecode = rec->after_call;          part of the atomic group, but this is not on the final matching path,
1991          break;          so must be unset when 2 is set. (If there is no group 2, there is no
1992            problem, because offset_top will then be 2, indicating no capture.) */
1993    
1994            if (offset > offset_top)
1995              {
1996              register int *iptr = md->offset_vector + offset_top;
1997              register int *iend = md->offset_vector + offset;
1998              while (iptr < iend) *iptr++ = -1;
1999              }
2000    
2001            /* Now make the extraction */
2002    
2003            md->offset_vector[offset] =
2004              md->offset_vector[md->offset_end - number];
2005            md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
2006            if (offset_top <= offset) offset_top = offset + 2;
2007          }          }
2008        }        }
2009    
2010      /* For a non-repeating ket, just continue at this level. This also      /* For an ordinary non-repeating ket, just continue at this level. This
2011      happens for a repeating ket if no characters were matched in the group.      also happens for a repeating ket if no characters were matched in the
2012      This is the forcible breaking of infinite loops as implemented in Perl      group. This is the forcible breaking of infinite loops as implemented in
2013      5.005. If there is an options reset, it will get obeyed in the normal      Perl 5.005. For a non-repeating atomic group that includes captures,
2014      course of events. */      establish a backup point by processing the rest of the pattern at a lower
2015        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
2016        original OP_ONCE level, thereby bypassing intermediate backup points, but
2017        resetting any captures that happened along the way. */
2018    
2019      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
2020        {        {
2021        ecode += 1 + LINK_SIZE;        if (*prev == OP_ONCE)
2022            {
2023            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
2024            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2025            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2026            RRETURN(MATCH_ONCE);
2027            }
2028          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
2029        break;        break;
2030        }        }
2031    
2032      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
2033      and return the MATCH_KETRPOS. This makes it possible to do the repeats one      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
2034      at a time from the outer level, thus saving stack. */      at a time from the outer level, thus saving stack. */
2035    
2036      if (*ecode == OP_KETRPOS)      if (*ecode == OP_KETRPOS)
2037        {        {
2038        md->end_match_ptr = eptr;        md->end_match_ptr = eptr;
2039        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
2040        RRETURN(MATCH_KETRPOS);        RRETURN(MATCH_KETRPOS);
2041        }        }
2042    
2043      /* The normal repeating kets try the rest of the pattern or restart from      /* The normal repeating kets try the rest of the pattern or restart from
2044      the preceding bracket, in the appropriate order. In the second case, we can      the preceding bracket, in the appropriate order. In the second case, we can
2045      use tail recursion to avoid using another stack frame, unless we have an      use tail recursion to avoid using another stack frame, unless we have an
2046      unlimited repeat of a group that can match an empty string. */      an atomic group or an unlimited repeat of a group that can match an empty
2047        string. */
2048    
2049      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
2050        {        {
2051        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
2052        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2053          if (*prev == OP_ONCE)
2054            {
2055            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
2056            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2057            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2058            RRETURN(MATCH_ONCE);
2059            }
2060        if (*prev >= OP_SBRA)    /* Could match an empty string */        if (*prev >= OP_SBRA)    /* Could match an empty string */
2061          {          {
         md->match_function_type = MATCH_CBEGROUP;  
2062          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
2063          RRETURN(rrc);          RRETURN(rrc);
2064          }          }
# Line 1818  for (;;) Line 2067  for (;;)
2067        }        }
2068      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
2069        {        {
       if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;  
2070        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
2071          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
2072        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2073          if (*prev == OP_ONCE)
2074            {
2075            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
2076            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2077            md->once_target = prev;
2078            RRETURN(MATCH_ONCE);
2079            }
2080        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
2081        goto TAIL_RECURSE;        goto TAIL_RECURSE;
2082        }        }
# Line 1829  for (;;) Line 2085  for (;;)
2085      /* Not multiline mode: start of subject assertion, unless notbol. */      /* Not multiline mode: start of subject assertion, unless notbol. */
2086    
2087      case OP_CIRC:      case OP_CIRC:
2088      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2089    
2090      /* Start of subject assertion */      /* Start of subject assertion */
2091    
2092      case OP_SOD:      case OP_SOD:
2093      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);
2094      ecode++;      ecode++;
2095      break;      break;
2096    
2097      /* Multiline mode: start of subject unless notbol, or after any newline. */      /* Multiline mode: start of subject unless notbol, or after any newline. */
2098    
2099      case OP_CIRCM:      case OP_CIRCM:
2100      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2101      if (eptr != md->start_subject &&      if (eptr != md->start_subject &&
2102          (eptr == md->end_subject || !WAS_NEWLINE(eptr)))          (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
2103        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2104      ecode++;      ecode++;
2105      break;      break;
2106    
2107      /* Start of match assertion */      /* Start of match assertion */
2108    
2109      case OP_SOM:      case OP_SOM:
2110      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);
2111      ecode++;      ecode++;
2112      break;      break;
2113    
# Line 1867  for (;;) Line 2123  for (;;)
2123    
2124      case OP_DOLLM:      case OP_DOLLM:
2125      if (eptr < md->end_subject)      if (eptr < md->end_subject)
2126        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }        {
2127          if (!IS_NEWLINE(eptr))
2128            {
2129            if (md->partial != 0 &&
2130                eptr + 1 >= md->end_subject &&
2131                NLBLOCK->nltype == NLTYPE_FIXED &&
2132                NLBLOCK->nllen == 2 &&
2133                RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2134              {
2135              md->hitend = TRUE;
2136              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2137              }
2138            RRETURN(MATCH_NOMATCH);
2139            }
2140          }
2141      else      else
2142        {        {
2143        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2144        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2145        }        }
2146      ecode++;      ecode++;
2147      break;      break;
2148    
2149      /* Not multiline mode: assert before a terminating newline or before end of      /* Not multiline mode: assert before a terminating newline or before end of
2150      subject unless noteol is set. */      subject unless noteol is set. */
2151    
2152      case OP_DOLL:      case OP_DOLL:
2153      if (md->noteol) MRRETURN(MATCH_NOMATCH);      if (md->noteol) RRETURN(MATCH_NOMATCH);
2154      if (!md->endonly) goto ASSERT_NL_OR_EOS;      if (!md->endonly) goto ASSERT_NL_OR_EOS;
2155    
2156      /* ... else fall through for endonly */      /* ... else fall through for endonly */
# Line 1888  for (;;) Line 2158  for (;;)
2158      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2159    
2160      case OP_EOD:      case OP_EOD:
2161      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2162      SCHECK_PARTIAL();      SCHECK_PARTIAL();
2163      ecode++;      ecode++;
2164      break;      break;
# Line 1899  for (;;) Line 2169  for (;;)
2169      ASSERT_NL_OR_EOS:      ASSERT_NL_OR_EOS:
2170      if (eptr < md->end_subject &&      if (eptr < md->end_subject &&
2171          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2172        MRRETURN(MATCH_NOMATCH);        {
2173          if (md->partial != 0 &&
2174              eptr + 1 >= md->end_subject &&
2175              NLBLOCK->nltype == NLTYPE_FIXED &&
2176              NLBLOCK->nllen == 2 &&
2177              RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2178            {
2179            md->hitend = TRUE;
2180            if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2181            }
2182          RRETURN(MATCH_NOMATCH);
2183          }
2184    
2185      /* Either at end of string or \n before end. */      /* Either at end of string or \n before end. */
2186    
# Line 1918  for (;;) Line 2199  for (;;)
2199        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2200        partial matching. */        partial matching. */
2201    
2202  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2203        if (utf8)        if (utf)
2204          {          {
2205          /* Get status of previous character */          /* Get status of previous character */
2206    
2207          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2208            {            {
2209            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2210            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2211            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2212            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2213  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 1991  for (;;) Line 2272  for (;;)
2272              }              }
2273            else            else
2274  #endif  #endif
2275            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);            prev_is_word = MAX_255(eptr[-1])
2276                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2277            }            }
2278    
2279          /* Get status of next character */          /* Get status of next character */
# Line 2014  for (;;) Line 2296  for (;;)
2296            }            }
2297          else          else
2298  #endif  #endif
2299          cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          cur_is_word = MAX_255(*eptr)
2300              && ((md->ctypes[*eptr] & ctype_word) != 0);
2301          }          }
2302    
2303        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2304    
2305        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2306             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2307          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2308        }        }
2309      break;      break;
2310    
2311      /* Match a single character type; inline for speed */      /* Match any single character type except newline; have to take care with
2312        CRLF newlines and partial matching. */
2313    
2314      case OP_ANY:      case OP_ANY:
2315      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
2316        if (md->partial != 0 &&
2317            eptr + 1 >= md->end_subject &&
2318            NLBLOCK->nltype == NLTYPE_FIXED &&
2319            NLBLOCK->nllen == 2 &&
2320            RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2321          {
2322          md->hitend = TRUE;
2323          if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2324          }
2325    
2326      /* Fall through */      /* Fall through */
2327    
2328        /* Match any single character whatsoever. */
2329    
2330      case OP_ALLANY:      case OP_ALLANY:
2331      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2332        {        {                            /* not be updated before SCHECK_PARTIAL. */
2333        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2334        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2335        }        }
2336      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2337    #ifdef SUPPORT_UTF
2338        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2339    #endif
2340      ecode++;      ecode++;
2341      break;      break;
2342    
# Line 2045  for (;;) Line 2344  for (;;)
2344      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2345    
2346      case OP_ANYBYTE:      case OP_ANYBYTE:
2347      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2348        {        {                            /* not be updated before SCHECK_PARTIAL. */
2349        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2350        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2351        }        }
2352        eptr++;
2353      ecode++;      ecode++;
2354      break;      break;
2355    
# Line 2057  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 < 256 &&
2366  #endif  #endif
2367         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2368         )         )
2369        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2370      ecode++;      ecode++;
2371      break;      break;
2372    
# Line 2074  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 > 255 ||
2383  #endif  #endif
2384         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2385         )         )
2386        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2387      ecode++;      ecode++;
2388      break;      break;
2389    
# Line 2091  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 < 256 &&
2400  #endif  #endif
2401         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2402         )         )
2403        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2404      ecode++;      ecode++;
2405      break;      break;
2406    
# Line 2108  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      if (      if (
2415  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2416         c >= 256 ||         c > 255 ||
2417  #endif  #endif
2418         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2419         )         )
2420        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2421      ecode++;      ecode++;
2422      break;      break;
2423    
# Line 2125  for (;;) Line 2425  for (;;)
2425      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2426        {        {
2427        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2428        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2429        }        }
2430      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2431      if (      if (
2432  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2433         c < 256 &&         c < 256 &&
2434  #endif  #endif
2435         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2436         )         )
2437        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2438      ecode++;      ecode++;
2439      break;      break;
2440    
# Line 2142  for (;;) Line 2442  for (;;)
2442      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2443        {        {
2444        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2445        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2446        }        }
2447      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2448      if (      if (
2449  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2450         c >= 256 ||         c > 255 ||
2451  #endif  #endif
2452         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2453         )         )
2454        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2455      ecode++;      ecode++;
2456      break;      break;
2457    
# Line 2159  for (;;) Line 2459  for (;;)
2459      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2460        {        {
2461        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2462        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2463        }        }
2464      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2465      switch(c)      switch(c)
2466        {        {
2467        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2468    
2469        case 0x000d:        case CHAR_CR:
2470        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr >= md->end_subject)
2471            {
2472            SCHECK_PARTIAL();
2473            }
2474          else if (RAWUCHARTEST(eptr) == CHAR_LF) eptr++;
2475        break;        break;
2476    
2477        case 0x000a:        case CHAR_LF:
2478        break;        break;
2479    
2480        case 0x000b:        case CHAR_VT:
2481        case 0x000c:        case CHAR_FF:
2482        case 0x0085:        case CHAR_NEL:
2483    #ifndef EBCDIC
2484        case 0x2028:        case 0x2028:
2485        case 0x2029:        case 0x2029:
2486        if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);  #endif  /* Not EBCDIC */
2487          if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
2488        break;        break;
2489        }        }
2490      ecode++;      ecode++;
# Line 2188  for (;;) Line 2494  for (;;)
2494      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2495        {        {
2496        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2497        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2498        }        }
2499      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2500      switch(c)      switch(c)
2501        {        {
2502          HSPACE_CASES: RRETURN(MATCH_NOMATCH);  /* Byte and multibyte cases */
2503        default: break;        default: break;
       case 0x09:      /* HT */  
       case 0x20:      /* SPACE */  
       case 0xa0:      /* NBSP */  
       case 0x1680:    /* OGHAM SPACE MARK */  
       case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */  
       case 0x2000:    /* EN QUAD */  
       case 0x2001:    /* EM QUAD */  
       case 0x2002:    /* EN SPACE */  
       case 0x2003:    /* EM SPACE */  
       case 0x2004:    /* THREE-PER-EM SPACE */  
       case 0x2005:    /* FOUR-PER-EM SPACE */  
       case 0x2006:    /* SIX-PER-EM SPACE */  
       case 0x2007:    /* FIGURE SPACE */  
       case 0x2008:    /* PUNCTUATION SPACE */  
       case 0x2009:    /* THIN SPACE */  
       case 0x200A:    /* HAIR SPACE */  
       case 0x202f:    /* NARROW NO-BREAK SPACE */  
       case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */  
       case 0x3000:    /* IDEOGRAPHIC SPACE */  
       MRRETURN(MATCH_NOMATCH);  
2504        }        }
2505      ecode++;      ecode++;
2506      break;      break;
# Line 2222  for (;;) Line 2509  for (;;)
2509      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2510        {        {
2511        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2512        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2513        }        }
2514      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2515      switch(c)      switch(c)
2516        {        {
2517        default: MRRETURN(MATCH_NOMATCH);        HSPACE_CASES: break;  /* Byte and multibyte cases */
2518        case 0x09:      /* HT */        default: RRETURN(MATCH_NOMATCH);
       case 0x20:      /* SPACE */  
       case 0xa0:      /* NBSP */  
       case 0x1680:    /* OGHAM SPACE MARK */  
       case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */  
       case 0x2000:    /* EN QUAD */  
       case 0x2001:    /* EM QUAD */  
       case 0x2002:    /* EN SPACE */  
       case 0x2003:    /* EM SPACE */  
       case 0x2004:    /* THREE-PER-EM SPACE */  
       case 0x2005:    /* FOUR-PER-EM SPACE */  
       case 0x2006:    /* SIX-PER-EM SPACE */  
       case 0x2007:    /* FIGURE SPACE */  
       case 0x2008:    /* PUNCTUATION SPACE */  
       case 0x2009:    /* THIN SPACE */  
       case 0x200A:    /* HAIR SPACE */  
       case 0x202f:    /* NARROW NO-BREAK SPACE */  
       case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */  
       case 0x3000:    /* IDEOGRAPHIC SPACE */  
       break;  
2519        }        }
2520      ecode++;      ecode++;
2521      break;      break;
# Line 2256  for (;;) Line 2524  for (;;)
2524      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2525        {        {
2526        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2527        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2528        }        }
2529      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2530      switch(c)      switch(c)
2531        {        {
2532          VSPACE_CASES: RRETURN(MATCH_NOMATCH);
2533        default: break;        default: break;
       case 0x0a:      /* LF */  
       case 0x0b:      /* VT */  
       case 0x0c:      /* FF */  
       case 0x0d:      /* CR */  
       case 0x85:      /* NEL */  
       case 0x2028:    /* LINE SEPARATOR */  
       case 0x2029:    /* PARAGRAPH SEPARATOR */  
       MRRETURN(MATCH_NOMATCH);  
2534        }        }
2535      ecode++;      ecode++;
2536      break;      break;
# Line 2278  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);        VSPACE_CASES: break;
2548        case 0x0a:      /* LF */        default: RRETURN(MATCH_NOMATCH);
       case 0x0b:      /* VT */  
       case 0x0c:      /* FF */  
       case 0x0d:      /* CR */  
       case 0x85:      /* NEL */  
       case 0x2028:    /* LINE SEPARATOR */  
       case 0x2029:    /* PARAGRAPH SEPARATOR */  
       break;  
2549        }        }
2550      ecode++;      ecode++;
2551      break;      break;
# Line 2305  for (;;) Line 2559  for (;;)
2559      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2560        {        {
2561        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2562        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2563        }        }
2564      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2565        {        {
2566          const pcre_uint32 *cp;
2567        const ucd_record *prop = GET_UCD(c);        const ucd_record *prop = GET_UCD(c);
2568    
2569        switch(ecode[1])        switch(ecode[1])
2570          {          {
2571          case PT_ANY:          case PT_ANY:
2572          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2573          break;          break;
2574    
2575          case PT_LAMP:          case PT_LAMP:
2576          if ((prop->chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2577               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2578               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2579            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2580          break;          break;
2581    
2582          case PT_GC:          case PT_GC:
2583          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2584            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2585          break;          break;
2586    
2587          case PT_PC:          case PT_PC:
2588          if ((ecode[2] != prop->chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2589            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2590          break;          break;
2591    
2592          case PT_SC:          case PT_SC:
2593          if ((ecode[2] != prop->script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2594            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2595          break;          break;
2596    
2597          /* These are specials */          /* These are specials */
2598    
2599          case PT_ALNUM:          case PT_ALNUM:
2600          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2601               _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))               PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2602            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2603          break;          break;
2604    
2605          case PT_SPACE:    /* Perl space */          case PT_SPACE:    /* Perl space */
2606          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2607               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2608                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2609            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2610          break;          break;
2611    
2612          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2613          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2614               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2615               c == CHAR_FF || c == CHAR_CR)               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_WORD:          case PT_WORD:
2621          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2622               _pcre_ucp_gentype[prop->chartype] == ucp_N ||               PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2623               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2624            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2625            break;
2626    
2627            case PT_CLIST:
2628            cp = PRIV(ucd_caseless_sets) + ecode[2];
2629            for (;;)
2630              {
2631              if (c < *cp)
2632                { if (op == OP_PROP) { RRETURN(MATCH_NOMATCH); } else break; }
2633              if (c == *cp++)
2634                { if (op == OP_PROP) break; else { RRETURN(MATCH_NOMATCH); } }
2635              }
2636            break;
2637    
2638            case PT_UCNC:
2639            if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
2640                 c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
2641                 c >= 0xe000) == (op == OP_NOTPROP))
2642              RRETURN(MATCH_NOMATCH);
2643          break;          break;
2644    
2645          /* This should never occur */          /* This should never occur */
# Line 2386  for (;;) Line 2659  for (;;)
2659      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2660        {        {
2661        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2662        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2663        }        }
2664      GETCHARINCTEST(c, eptr);      else
2665        {        {
2666        int category = UCD_CATEGORY(c);        int lgb, rgb;
2667        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        GETCHARINCTEST(c, eptr);
2668          lgb = UCD_GRAPHBREAK(c);
2669        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2670          {          {
2671          int len = 1;          int len = 1;
2672          if (!utf8) c = *eptr; else          if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2673            {          rgb = UCD_GRAPHBREAK(c);
2674            GETCHARLEN(c, eptr, len);          if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
2675            }          lgb = rgb;
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
2676          eptr += len;          eptr += len;
2677          }          }
2678        }        }
2679        CHECK_PARTIAL();
2680      ecode++;      ecode++;
2681      break;      break;
2682  #endif  #endif  /* SUPPORT_UCP */
2683    
2684    
2685      /* Match a back reference, possibly repeatedly. Look past the end of the      /* Match a back reference, possibly repeatedly. Look past the end of the
# Line 2418  for (;;) Line 2691  for (;;)
2691      loops). */      loops). */
2692    
2693      case OP_REF:      case OP_REF:
2694      case OP_REFI:      case OP_REFI:
2695      caseless = op == OP_REFI;      caseless = op == OP_REFI;
2696      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2697      ecode += 3;      ecode += 1 + IMM2_SIZE;
2698    
2699      /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2700    
# Line 2461  for (;;) Line 2734  for (;;)
2734        case OP_CRMINRANGE:        case OP_CRMINRANGE:
2735        minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2736        min = GET2(ecode, 1);        min = GET2(ecode, 1);
2737        max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2738        if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2739        ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2740        break;        break;
2741    
2742        default:               /* No repeat follows */        default:               /* No repeat follows */
2743        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2744          {          {
2745            if (length == -2) eptr = md->end_subject;   /* Partial match */
2746          CHECK_PARTIAL();          CHECK_PARTIAL();
2747          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2748          }          }
2749        eptr += length;        eptr += length;
2750        continue;              /* With the main loop */        continue;              /* With the main loop */
2751        }        }
2752    
2753      /* Handle repeated back references. If the length of the reference is      /* Handle repeated back references. If the length of the reference is
2754      zero, just continue with the main loop. */      zero, just continue with the main loop. If the length is negative, it
2755        means the reference is unset in non-Java-compatible mode. If the minimum is
2756        zero, we can continue at the same level without recursion. For any other
2757        minimum, carrying on will result in NOMATCH. */
2758    
2759      if (length == 0) continue;      if (length == 0) continue;
2760        if (length < 0 && min == 0) continue;
2761    
2762      /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2763      the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
# Line 2487  for (;;) Line 2765  for (;;)
2765    
2766      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2767        {        {
2768        int slength;        int slength;
2769        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2770          {          {
2771            if (slength == -2) eptr = md->end_subject;   /* Partial match */
2772          CHECK_PARTIAL();          CHECK_PARTIAL();
2773          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2774          }          }
2775        eptr += slength;        eptr += slength;
2776        }        }
# Line 2507  for (;;) Line 2786  for (;;)
2786        {        {
2787        for (fi = min;; fi++)        for (fi = min;; fi++)
2788          {          {
2789          int slength;          int slength;
2790          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2791          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2792          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) RRETURN(MATCH_NOMATCH);
2793          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2794            {            {
2795              if (slength == -2) eptr = md->end_subject;   /* Partial match */
2796            CHECK_PARTIAL();            CHECK_PARTIAL();
2797            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2798            }            }
2799          eptr += slength;          eptr += slength;
2800          }          }
# Line 2528  for (;;) Line 2808  for (;;)
2808        pp = eptr;        pp = eptr;
2809        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2810          {          {
2811          int slength;          int slength;
2812          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2813            {            {
2814            CHECK_PARTIAL();            /* Can't use CHECK_PARTIAL because we don't want to update eptr in
2815              the soft partial matching case. */
2816    
2817              if (slength == -2 && md->partial != 0 &&
2818                  md->end_subject > md->start_used_ptr)
2819                {
2820                md->hitend = TRUE;
2821                if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2822                }
2823            break;            break;
2824            }            }
2825          eptr += slength;          eptr += slength;
2826          }          }
2827    
2828        while (eptr >= pp)        while (eptr >= pp)
2829          {          {
2830          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2831          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2832          eptr -= length;          eptr -= length;
2833          }          }
2834        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2835        }        }
2836      /* Control never gets here */      /* Control never gets here */
2837    
# Line 2560  for (;;) Line 2849  for (;;)
2849      case OP_NCLASS:      case OP_NCLASS:
2850      case OP_CLASS:      case OP_CLASS:
2851        {        {
2852          /* The data variable is saved across frames, so the byte map needs to
2853          be stored there. */
2854    #define BYTE_MAP ((pcre_uint8 *)data)
2855        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2856        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2857    
2858        switch (*ecode)        switch (*ecode)
2859          {          {
# Line 2582  for (;;) Line 2874  for (;;)
2874          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2875          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2876          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2877          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2878          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2879          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2880          break;          break;
2881    
2882          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2594  for (;;) Line 2886  for (;;)
2886    
2887        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2888    
2889  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2890        /* UTF-8 mode */        if (utf)
       if (utf8)  
2891          {          {
2892          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2893            {            {
2894            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2895              {              {
2896              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2897              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2898              }              }
2899            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2900            if (c > 255)            if (c > 255)
2901              {              {
2902              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2903              }              }
2904            else            else
2905              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
             }  
2906            }            }
2907          }          }
2908        else        else
2909  #endif  #endif
2910        /* Not UTF-8 mode */        /* Not UTF mode */
2911          {          {
2912          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2913            {            {
2914            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2915              {              {
2916              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2917              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2918              }              }
2919            c = *eptr++;            c = *eptr++;
2920            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2921              if (c > 255)
2922                {
2923                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2924                }
2925              else
2926    #endif
2927                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2928            }            }
2929          }          }
2930    
# Line 2642  for (;;) Line 2938  for (;;)
2938    
2939        if (minimize)        if (minimize)
2940          {          {
2941  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2942          /* UTF-8 mode */          if (utf)
         if (utf8)  
2943            {            {
2944            for (fi = min;; fi++)            for (fi = min;; fi++)
2945              {              {
2946              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2947              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2948              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2949              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2950                {                {
2951                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2952                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2953                }                }
2954              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2955              if (c > 255)              if (c > 255)
2956                {                {
2957                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2958                }                }
2959              else              else
2960                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
               }  
2961              }              }
2962            }            }
2963          else          else
2964  #endif  #endif
2965          /* Not UTF-8 mode */          /* Not UTF mode */
2966            {            {
2967            for (fi = min;; fi++)            for (fi = min;; fi++)
2968              {              {
2969              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2970              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2971              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2972              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2973                {                {
2974                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2975                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2976                }                }
2977              c = *eptr++;              c = *eptr++;
2978              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2979                if (c > 255)
2980                  {
2981                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2982                  }
2983                else
2984    #endif
2985                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2986              }              }
2987            }            }
2988          /* Control never gets here */          /* Control never gets here */
# Line 2694  for (;;) Line 2994  for (;;)
2994          {          {
2995          pp = eptr;          pp = eptr;
2996    
2997  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2998          /* UTF-8 mode */          if (utf)
         if (utf8)  
2999            {            {
3000            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3001              {              {
# Line 2712  for (;;) Line 3011  for (;;)
3011                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
3012                }                }
3013              else              else
3014                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
3015              eptr += len;              eptr += len;
3016              }              }
3017            for (;;)            for (;;)
# Line 2727  for (;;) Line 3024  for (;;)
3024            }            }
3025          else          else
3026  #endif  #endif
3027            /* Not UTF-8 mode */            /* Not UTF mode */
3028            {            {
3029            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3030              {              {
# Line 2737  for (;;) Line 3034  for (;;)
3034                break;                break;
3035                }                }
3036              c = *eptr;              c = *eptr;
3037              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
3038                if (c > 255)
3039                  {
3040                  if (op == OP_CLASS) break;
3041                  }
3042                else
3043    #endif
3044                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3045              eptr++;              eptr++;
3046              }              }
3047            while (eptr >= pp)            while (eptr >= pp)
# Line 2748  for (;;) Line 3052  for (;;)
3052              }              }
3053            }            }
3054    
3055          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3056          }          }
3057    #undef BYTE_MAP
3058        }        }
3059      /* Control never gets here */      /* Control never gets here */
3060    
# Line 2758  for (;;) Line 3063  for (;;)
3063      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
3064      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
3065    
3066  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3067      case OP_XCLASS:      case OP_XCLASS:
3068        {        {
3069        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2783  for (;;) Line 3088  for (;;)
3088          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3089          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3090          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3091          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
3092          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
3093          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
3094          break;          break;
3095    
3096          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2800  for (;;) Line 3105  for (;;)
3105          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3106            {            {
3107            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3108            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3109            }            }
3110          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
3111          if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3112          }          }
3113    
3114        /* 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 2820  for (;;) Line 3125  for (;;)
3125            {            {
3126            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
3127            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3128            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3129            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3130              {              {
3131              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3132              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3133              }              }
3134            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3135            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3136            }            }
3137          /* Control never gets here */          /* Control never gets here */
3138          }          }
# Line 2845  for (;;) Line 3150  for (;;)
3150              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3151              break;              break;
3152              }              }
3153    #ifdef SUPPORT_UTF
3154            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3155            if (!_pcre_xclass(c, data)) break;  #else
3156              c = *eptr;
3157    #endif
3158              if (!PRIV(xclass)(c, data, utf)) break;
3159            eptr += len;            eptr += len;
3160            }            }
3161          for(;;)          for(;;)
# Line 2854  for (;;) Line 3163  for (;;)
3163            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3164            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3165            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3166            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3167              if (utf) BACKCHAR(eptr);
3168    #endif
3169            }            }
3170          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3171          }          }
3172    
3173        /* Control never gets here */        /* Control never gets here */
# Line 2866  for (;;) Line 3177  for (;;)
3177      /* Match a single character, casefully */      /* Match a single character, casefully */
3178    
3179      case OP_CHAR:      case OP_CHAR:
3180  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3181      if (utf8)      if (utf)
3182        {        {
3183        length = 1;        length = 1;
3184        ecode++;        ecode++;
# Line 2875  for (;;) Line 3186  for (;;)
3186        if (length > md->end_subject - eptr)        if (length > md->end_subject - eptr)
3187          {          {
3188          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
3189          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3190          }          }
3191        while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);        while (length-- > 0) if (*ecode++ != RAWUCHARINC(eptr)) RRETURN(MATCH_NOMATCH);
3192        }        }
3193      else      else
3194  #endif  #endif
3195        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3196        {        {
3197        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3198          {          {
3199          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
3200          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3201          }          }
3202        if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
3203        ecode += 2;        ecode += 2;
3204        }        }
3205      break;      break;
3206    
3207      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3208        subject, give up immediately. */
3209    
3210      case OP_CHARI:      case OP_CHARI:
3211  #ifdef SUPPORT_UTF8      if (eptr >= md->end_subject)
3212      if (utf8)        {
3213          SCHECK_PARTIAL();
3214          RRETURN(MATCH_NOMATCH);
3215          }
3216    
3217    #ifdef SUPPORT_UTF
3218        if (utf)
3219        {        {
3220        length = 1;        length = 1;
3221        ecode++;        ecode++;
3222        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3223    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         MRRETURN(MATCH_NOMATCH);  
         }  
   
3224        /* 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
3225        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3226          fast lookup table. We know that there is at least one byte left in the
3227          subject. */
3228    
3229        if (fc < 128)        if (fc < 128)
3230          {          {
3231          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          pcre_uint32 cc = RAWUCHAR(eptr);
3232            if (md->lcc[fc] != TABLE_GET(cc, md->lcc, cc)) RRETURN(MATCH_NOMATCH);
3233            ecode++;
3234            eptr++;
3235          }          }
3236    
3237        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3238          use the value of "length" to check for sufficient bytes left, because the
3239          other case of the character may have more or fewer bytes.  */
3240    
3241        else        else
3242          {          {
3243          unsigned int dc;          pcre_uint32 dc;
3244          GETCHARINC(dc, eptr);          GETCHARINC(dc, eptr);
3245          ecode += length;          ecode += length;
3246    
# Line 2934  for (;;) Line 3252  for (;;)
3252  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3253            if (dc != UCD_OTHERCASE(fc))            if (dc != UCD_OTHERCASE(fc))
3254  #endif  #endif
3255              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3256            }            }
3257          }          }
3258        }        }
3259      else      else
3260  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3261    
3262      /* Non-UTF-8 mode */      /* Not UTF mode */
3263        {        {
3264        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3265          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3266          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         MRRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);  
3267        ecode += 2;        ecode += 2;
3268        }        }
3269      break;      break;
# Line 2958  for (;;) Line 3273  for (;;)
3273      case OP_EXACT:      case OP_EXACT:
3274      case OP_EXACTI:      case OP_EXACTI:
3275      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3276      ecode += 3;      ecode += 1 + IMM2_SIZE;
3277      goto REPEATCHAR;      goto REPEATCHAR;
3278    
3279      case OP_POSUPTO:      case OP_POSUPTO:
# Line 2973  for (;;) Line 3288  for (;;)
3288      min = 0;      min = 0;
3289      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3290      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3291      ecode += 3;      ecode += 1 + IMM2_SIZE;
3292      goto REPEATCHAR;      goto REPEATCHAR;
3293    
3294      case OP_POSSTAR:      case OP_POSSTAR:
# Line 3021  for (;;) Line 3336  for (;;)
3336      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3337    
3338      REPEATCHAR:      REPEATCHAR:
3339  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3340      if (utf8)      if (utf)
3341        {        {
3342        length = 1;        length = 1;
3343        charptr = ecode;        charptr = ecode;
# Line 3035  for (;;) Line 3350  for (;;)
3350        if (length > 1)        if (length > 1)
3351          {          {
3352  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3353          unsigned int othercase;          pcre_uint32 othercase;
3354          if (op >= OP_STARI &&     /* Caseless */          if (op >= OP_STARI &&     /* Caseless */
3355              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3356            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3357          else oclength = 0;          else oclength = 0;
3358  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3359    
3360          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3361            {            {
3362            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3363              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3364  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3365            else if (oclength > 0 &&            else if (oclength > 0 &&
3366                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3367                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3368  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3369            else            else
3370              {              {
3371              CHECK_PARTIAL();              CHECK_PARTIAL();
3372              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3373              }              }
3374            }            }
3375    
# Line 3066  for (;;) Line 3381  for (;;)
3381              {              {
3382              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3383              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3384              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3385              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3386                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3387  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3388              else if (oclength > 0 &&              else if (oclength > 0 &&
3389                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3390                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3391  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3392              else              else
3393                {                {
3394                CHECK_PARTIAL();                CHECK_PARTIAL();
3395                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3396                }                }
3397              }              }
3398            /* Control never gets here */            /* Control never gets here */
# Line 3089  for (;;) Line 3404  for (;;)
3404            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3405              {              {
3406              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3407                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3408  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3409              else if (oclength > 0 &&              else if (oclength > 0 &&
3410                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3411                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3412  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3413              else              else
3414                {                {
# Line 3108  for (;;) Line 3423  for (;;)
3423              {              {
3424              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3425              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3426              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3427  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3428              eptr--;              eptr--;
3429              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3125  for (;;) Line 3440  for (;;)
3440        value of fc will always be < 128. */        value of fc will always be < 128. */
3441        }        }
3442      else      else
3443  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3444          /* When not in UTF-8 mode, load a single-byte character. */
3445      /* When not in UTF-8 mode, load a single-byte character. */        fc = *ecode++;
3446    
3447      fc = *ecode++;      /* The value of fc at this point is always one character, though we may
3448        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  
3449      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3450      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3451      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 3141  for (;;) Line 3454  for (;;)
3454      maximizing, find the maximum number of characters and work backwards. */      maximizing, find the maximum number of characters and work backwards. */
3455    
3456      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3457        max, eptr));        max, (char *)eptr));
3458    
3459      if (op >= OP_STARI)  /* Caseless */      if (op >= OP_STARI)  /* Caseless */
3460        {        {
3461        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3462          /* fc must be < 128 if UTF is enabled. */
3463          foc = md->fcc[fc];
3464    #else
3465    #ifdef SUPPORT_UTF
3466    #ifdef SUPPORT_UCP
3467          if (utf && fc > 127)
3468            foc = UCD_OTHERCASE(fc);
3469    #else
3470          if (utf && fc > 127)
3471            foc = fc;
3472    #endif /* SUPPORT_UCP */
3473          else
3474    #endif /* SUPPORT_UTF */
3475            foc = TABLE_GET(fc, md->fcc, fc);
3476    #endif /* COMPILE_PCRE8 */
3477    
3478        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3479          {          {
3480            pcre_uint32 cc;                 /* Faster than pcre_uchar */
3481          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3482            {            {
3483            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3484            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3485            }            }
3486          if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          cc = RAWUCHARTEST(eptr);
3487            if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH);
3488            eptr++;
3489          }          }
3490        if (min == max) continue;        if (min == max) continue;
3491        if (minimize)        if (minimize)
3492          {          {
3493          for (fi = min;; fi++)          for (fi = min;; fi++)
3494            {            {
3495              pcre_uint32 cc;               /* Faster than pcre_uchar */
3496            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3497            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3498            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3499            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3500              {              {
3501              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3502              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3503              }              }
3504            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            cc = RAWUCHARTEST(eptr);
3505              if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH);
3506              eptr++;
3507            }            }
3508          /* Control never gets here */          /* Control never gets here */
3509          }          }
# Line 3177  for (;;) Line 3512  for (;;)
3512          pp = eptr;          pp = eptr;
3513          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3514            {            {
3515              pcre_uint32 cc;               /* Faster than pcre_uchar */
3516            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3517              {              {
3518              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3519              break;              break;
3520              }              }
3521            if (fc != md->lcc[*eptr]) break;            cc = RAWUCHARTEST(eptr);
3522              if (fc != cc && foc != cc) break;
3523            eptr++;            eptr++;
3524            }            }
3525    
# Line 3194  for (;;) Line 3531  for (;;)
3531            eptr--;            eptr--;
3532            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3533            }            }
3534          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3535          }          }
3536        /* Control never gets here */        /* Control never gets here */
3537        }        }
# Line 3208  for (;;) Line 3545  for (;;)
3545          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3546            {            {
3547            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3548            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3549            }            }
3550          if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);          if (fc != RAWUCHARINCTEST(eptr)) RRETURN(MATCH_NOMATCH);
3551          }          }
3552    
3553        if (min == max) continue;        if (min == max) continue;
# Line 3221  for (;;) Line 3558  for (;;)
3558            {            {
3559            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3560            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3561            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3562            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3563              {              {
3564              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3565              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3566              }              }
3567            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);            if (fc != RAWUCHARINCTEST(eptr)) RRETURN(MATCH_NOMATCH);
3568            }            }
3569          /* Control never gets here */          /* Control never gets here */
3570          }          }
# Line 3241  for (;;) Line 3578  for (;;)
3578              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3579              break;              break;
3580              }              }
3581            if (fc != *eptr) break;            if (fc != RAWUCHARTEST(eptr)) break;
3582            eptr++;            eptr++;
3583            }            }
3584          if (possessive) continue;          if (possessive) continue;
# Line 3252  for (;;) Line 3589  for (;;)
3589            eptr--;            eptr--;
3590            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3591            }            }
3592          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3593          }          }
3594        }        }
3595      /* Control never gets here */      /* Control never gets here */
# Line 3261  for (;;) Line 3598  for (;;)
3598      checking can be multibyte. */      checking can be multibyte. */
3599    
3600      case OP_NOT:      case OP_NOT:
3601      case OP_NOTI:      case OP_NOTI:
3602      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3603        {        {
3604        SCHECK_PARTIAL();        SCHECK_PARTIAL();
3605        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
3606        }        }
3607      ecode++;  #ifdef SUPPORT_UTF
3608      GETCHARINCTEST(c, eptr);      if (utf)
     if (op == OP_NOTI)         /* The caseless case */  
3609        {        {
3610  #ifdef SUPPORT_UTF8        register pcre_uint32 ch, och;
3611        if (c < 256)  
3612  #endif        ecode++;
3613        c = md->lcc[c];        GETCHARINC(ch, ecode);
3614        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        GETCHARINC(c, eptr);
3615    
3616          if (op == OP_NOT)
3617            {
3618            if (ch == c) RRETURN(MATCH_NOMATCH);
3619            }
3620          else
3621            {
3622    #ifdef SUPPORT_UCP
3623            if (ch > 127)
3624              och = UCD_OTHERCASE(ch);
3625    #else
3626            if (ch > 127)
3627              och = ch;
3628    #endif /* SUPPORT_UCP */
3629            else
3630              och = TABLE_GET(ch, md->fcc, ch);
3631            if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3632            }
3633        }        }
3634      else    /* Caseful */      else
3635    #endif
3636        {        {
3637        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        register pcre_uint32 ch = ecode[1];
3638          c = *eptr++;
3639          if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c))
3640            RRETURN(MATCH_NOMATCH);
3641          ecode += 2;
3642        }        }
3643      break;      break;
3644    
# Line 3293  for (;;) Line 3652  for (;;)
3652      case OP_NOTEXACT:      case OP_NOTEXACT:
3653      case OP_NOTEXACTI:      case OP_NOTEXACTI:
3654      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3655      ecode += 3;      ecode += 1 + IMM2_SIZE;
3656      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3657    
3658      case OP_NOTUPTO:      case OP_NOTUPTO:
# Line 3303  for (;;) Line 3662  for (;;)
3662      min = 0;      min = 0;
3663      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3664      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3665      ecode += 3;      ecode += 1 + IMM2_SIZE;
3666      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3667    
3668      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
# Line 3335  for (;;) Line 3694  for (;;)
3694      possessive = TRUE;      possessive = TRUE;
3695      min = 0;      min = 0;
3696      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3697      ecode += 3;      ecode += 1 + IMM2_SIZE;
3698      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3699    
3700      case OP_NOTSTAR:      case OP_NOTSTAR:
# Line 3359  for (;;) Line 3718  for (;;)
3718      /* Common code for all repeated single-byte matches. */      /* Common code for all repeated single-byte matches. */
3719    
3720      REPEATNOTCHAR:      REPEATNOTCHAR:
3721      fc = *ecode++;      GETCHARINCTEST(fc, ecode);
3722    
3723      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
3724      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 3370  for (;;) Line 3729  for (;;)
3729      characters and work backwards. */      characters and work backwards. */
3730    
3731      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,
3732        max, eptr));        max, (char *)eptr));
3733    
3734      if (op >= OP_NOTSTARI)     /* Caseless */      if (op >= OP_NOTSTARI)     /* Caseless */
3735        {        {
3736        fc = md->lcc[fc];  #ifdef SUPPORT_UTF
3737    #ifdef SUPPORT_UCP
3738          if (utf && fc > 127)
3739            foc = UCD_OTHERCASE(fc);
3740    #else
3741          if (utf && fc > 127)
3742            foc = fc;
3743    #endif /* SUPPORT_UCP */
3744          else
3745    #endif /* SUPPORT_UTF */
3746            foc = TABLE_GET(fc, md->fcc, fc);
3747    
3748  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3749        /* UTF-8 mode */        if (utf)
       if (utf8)  
3750          {          {
3751          register unsigned int d;          register pcre_uint32 d;
3752          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3753            {            {
3754            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3755              {              {
3756              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3757              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3758              }              }
3759            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3760            if (d < 256) d = md->lcc[d];            if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) MRRETURN(MATCH_NOMATCH);  
3761            }            }
3762          }          }
3763        else        else
3764  #endif  #endif
3765          /* Not UTF mode */
       /* Not UTF-8 mode */  
3766          {          {
3767          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3768            {            {
3769            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3770              {              {
3771              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3772              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3773              }              }
3774            if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3775              eptr++;
3776            }            }
3777          }          }
3778    
# Line 3413  for (;;) Line 3780  for (;;)
3780    
3781        if (minimize)        if (minimize)
3782          {          {
3783  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3784          /* UTF-8 mode */          if (utf)
         if (utf8)  
3785            {            {
3786            register unsigned int d;            register pcre_uint32 d;
3787            for (fi = min;; fi++)            for (fi = min;; fi++)
3788              {              {
3789              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3790              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3791              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3792              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3793                {                {
3794                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3795                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3796                }                }
3797              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3798              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
             if (fc == d) MRRETURN(MATCH_NOMATCH);  
3799              }              }
3800            }            }
3801          else          else
3802  #endif  #endif
3803          /* Not UTF-8 mode */          /* Not UTF mode */
3804            {            {
3805            for (fi = min;; fi++)            for (fi = min;; fi++)
3806              {              {
3807              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3808              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3809              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3810              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3811                {                {
3812                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3813                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3814                }                }
3815              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);              if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3816                eptr++;
3817              }              }
3818            }            }
3819          /* Control never gets here */          /* Control never gets here */
# Line 3459  for (;;) Line 3825  for (;;)
3825          {          {
3826          pp = eptr;          pp = eptr;
3827    
3828  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3829          /* UTF-8 mode */          if (utf)
         if (utf8)  
3830            {            {
3831            register unsigned int d;            register pcre_uint32 d;
3832            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3833              {              {
3834              int len = 1;              int len = 1;
# Line 3473  for (;;) Line 3838  for (;;)
3838                break;                break;
3839                }                }
3840              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3841              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) break;
             if (fc == d) break;  
3842              eptr += len;              eptr += len;
3843              }              }
3844          if (possessive) continue;            if (possessive) continue;