/[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 618 by ph10, Sat Jul 16 17:24:16 2011 UTC revision 1274 by ph10, Fri Mar 8 11:35:41 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_ONCE         (-996)  /* The next 5 must be kept together and in sequence so that a test that checks
93    for any one of them can use a range. */
94    #define MATCH_COMMIT       (-996)
95  #define MATCH_PRUNE        (-995)  #define MATCH_PRUNE        (-995)
96  #define MATCH_SKIP         (-994)  #define MATCH_SKIP         (-994)
97  #define MATCH_SKIP_ARG     (-993)  #define MATCH_SKIP_ARG     (-993)
98  #define MATCH_THEN         (-992)  #define MATCH_THEN         (-992)
99    #define MATCH_BACKTRACK_MAX MATCH_THEN
100  /* This is a convenience macro for code that occurs many times. */  #define MATCH_BACKTRACK_MIN MATCH_COMMIT
   
 #define MRRETURN(ra) \  
   { \  
   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 101  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 121  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 138  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 148  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 171  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 181  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 210  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 277  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, RM65, RM66 };         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 290  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 315  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 337  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 351  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 382  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 398  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 430  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 441  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 462  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 477  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 487  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 523  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 551  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 591  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 623  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 643  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 668  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 681  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 706  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 721  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. When a pattern match ends with a SKIP_ARG for which there was
825        not a matching mark, we have to re-run the match, ignoring the SKIP_ARG
826        that failed and any that preceed it (either they also failed, or were not
827        triggered). To do this, we maintain a count of executed SKIP_ARGs. If a
828        SKIP_ARG gets to top level, the match is re-run with md->ignore_skip_arg
829        set to the count of the one that failed. */
830    
831      case OP_SKIP_ARG:      case OP_SKIP_ARG:
832      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->skip_arg_count++;
833        if (md->skip_arg_count <= md->ignore_skip_arg)
834          {
835          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
836          break;
837          }
838        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
839        eptrb, RM57);        eptrb, RM57);
840      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
841        RRETURN(rrc);        RRETURN(rrc);
842    
843      /* 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
844      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
845      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
846      as PRUNE. */      with md->ignore_skip_arg set to the value of md->skip_arg_count. */
847    
848      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
849      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
850    
851      /* 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
852      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
853      to skip back past alternatives that precede the THEN within the current      match pointer to do this. */
     branch. */  
854    
855      case OP_THEN:      case OP_THEN:
856      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
857        eptrb, RM54);        eptrb, RM54);
858      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
859      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
860      MRRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
861    
862      case OP_THEN_ARG:      case OP_THEN_ARG:
863      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      md->nomatch_mark = ecode + 2;
864        offset_top, md, eptrb, RM58);      md->mark = NULL;    /* In case previously set by assertion */
865        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
866          md, eptrb, RM58);
867        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
868             md->mark == NULL) md->mark = ecode + 2;
869      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
870      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
     md->mark = ecode + LINK_SIZE + 2;  
871      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
872    
873        /* Handle an atomic group that does not contain any capturing parentheses.
874        This can be handled like an assertion. Prior to 8.13, all atomic groups
875        were handled this way. In 8.13, the code was changed as below for ONCE, so
876        that backups pass through the group and thereby reset captured values.
877        However, this uses a lot more stack, so in 8.20, atomic groups that do not
878        contain any captures generate OP_ONCE_NC, which can be handled in the old,
879        less stack intensive way.
880    
881        Check the alternative branches in turn - the matching won't pass the KET
882        for this kind of subpattern. If any one branch matches, we carry on as at
883        the end of a normal bracket, leaving the subject pointer, but resetting
884        the start-of-match value in case it was changed by \K. */
885    
886        case OP_ONCE_NC:
887        prev = ecode;
888        saved_eptr = eptr;
889        save_mark = md->mark;
890        do
891          {
892          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
893          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
894            {
895            mstart = md->start_match_ptr;
896            break;
897            }
898          if (rrc == MATCH_THEN)
899            {
900            next = ecode + GET(ecode,1);
901            if (md->start_match_ptr < next &&
902                (*ecode == OP_ALT || *next == OP_ALT))
903              rrc = MATCH_NOMATCH;
904            }
905    
906          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
907          ecode += GET(ecode,1);
908          md->mark = save_mark;
909          }
910        while (*ecode == OP_ALT);
911    
912        /* If hit the end of the group (which could be repeated), fail */
913    
914        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
915    
916        /* Continue as from after the group, updating the offsets high water
917        mark, since extracts may have been taken. */
918    
919        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
920    
921        offset_top = md->end_offset_top;
922        eptr = md->end_match_ptr;
923    
924        /* For a non-repeating ket, just continue at this level. This also
925        happens for a repeating ket if no characters were matched in the group.
926        This is the forcible breaking of infinite loops as implemented in Perl
927        5.005. */
928    
929        if (*ecode == OP_KET || eptr == saved_eptr)
930          {
931          ecode += 1+LINK_SIZE;
932          break;
933          }
934    
935        /* The repeating kets try the rest of the pattern or restart from the
936        preceding bracket, in the appropriate order. The second "call" of match()
937        uses tail recursion, to avoid using another stack frame. */
938    
939        if (*ecode == OP_KETRMIN)
940          {
941          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
942          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
943          ecode = prev;
944          goto TAIL_RECURSE;
945          }
946        else  /* OP_KETRMAX */
947          {
948          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
949          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
950          ecode += 1 + LINK_SIZE;
951          goto TAIL_RECURSE;
952          }
953        /* Control never gets here */
954    
955      /* Handle a capturing bracket, other than those that are possessive with an      /* Handle a capturing bracket, other than those that are possessive with an
956      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
957      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
958      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
959      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
960      the group. A failure to match might occur after the group has succeeded,      the group. A failure to match might occur after the group has succeeded,
961      if something later on doesn't match. For this reason, we need to restore      if something later on doesn't match. For this reason, we need to restore
962      the working value and also the values of the final offsets, in case they      the working value and also the values of the final offsets, in case they
963      were set by a previous iteration of the same bracket.      were set by a previous iteration of the same bracket.
# Line 822  for (;;) Line 970  for (;;)
970      case OP_SCBRA:      case OP_SCBRA:
971      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
972      offset = number << 1;      offset = number << 1;
973    
974  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
975      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
976      printf("subject=");      printf("subject=");
# Line 836  for (;;) Line 984  for (;;)
984        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
985        save_offset3 = md->offset_vector[md->offset_end - number];        save_offset3 = md->offset_vector[md->offset_end - number];
986        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
987          save_mark = md->mark;
988    
989        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
990        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
# Line 843  for (;;) Line 992  for (;;)
992    
993        for (;;)        for (;;)
994          {          {
995          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
996          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
997            eptrb, RM1);            eptrb, RM1);
998          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
999          if (rrc != MATCH_NOMATCH &&  
1000              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* If we backed up to a THEN, check whether it is within the current
1001            RRETURN(rrc);          branch by comparing the address of the THEN that is passed back with
1002            the end of the branch. If it is within the current branch, and the
1003            branch is one of two or more alternatives (it either starts or ends
1004            with OP_ALT), we have reached the limit of THEN's action, so convert
1005            the return code to NOMATCH, which will cause normal backtracking to
1006            happen from now on. Otherwise, THEN is passed back to an outer
1007            alternative. This implements Perl's treatment of parenthesized groups,
1008            where a group not containing | does not affect the current alternative,
1009            that is, (X) is NOT the same as (X|(*F)). */
1010    
1011            if (rrc == MATCH_THEN)
1012              {
1013              next = ecode + GET(ecode,1);
1014              if (md->start_match_ptr < next &&
1015                  (*ecode == OP_ALT || *next == OP_ALT))
1016                rrc = MATCH_NOMATCH;
1017              }
1018    
1019            /* Anything other than NOMATCH is passed back. */
1020    
1021            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1022          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
1023          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
1024          if (*ecode != OP_ALT) break;          md->mark = save_mark;
1025            if (*ecode != OP_ALT) break;
1026          }          }
1027    
1028        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
1029        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
1030        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
1031        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
   
       /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or  
       MATCH_THEN. */  
1032    
1033        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
1034        RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));  
1035          RRETURN(rrc);
1036        }        }
1037    
1038      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 879  for (;;) Line 1047  for (;;)
1047      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1048    
1049      /* Non-capturing or atomic group, except for possessive with unlimited      /* Non-capturing or atomic group, except for possessive with unlimited
1050      repeat. Loop for all the alternatives. When we get to the final alternative      repeat and ONCE group with no captures. Loop for all the alternatives.
1051      within the brackets, we used to return the result of a recursive call to  
1052      match() whatever happened so it was possible to reduce stack usage by      When we get to the final alternative within the brackets, we used to return
1053      turning this into a tail recursion, except in the case of a possibly empty      the result of a recursive call to match() whatever happened so it was
1054      group. However, now that there is the possiblity of (*THEN) occurring in      possible to reduce stack usage by turning this into a tail recursion,
1055      the final alternative, this optimization is no longer possible.      except in the case of a possibly empty group. However, now that there is
1056        the possiblity of (*THEN) occurring in the final alternative, this
1057      MATCH_ONCE is returned when the end of an atomic group is successfully      optimization is no longer always possible.
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      We can optimize if we know there are no (*THEN)s in the pattern; at present
1060        this is the best that can be done.
1061    
1062        MATCH_ONCE is returned when the end of an atomic group is successfully
1063        reached, but subsequent matching fails. It passes back up the tree (causing
1064        captured values to be reset) until the original atomic group level is
1065      reached. This is tested by comparing md->once_target with the start of the      reached. This is tested by comparing md->once_target with the start of the
1066      group. At this point, the return is converted into MATCH_NOMATCH so that      group. At this point, the return is converted into MATCH_NOMATCH so that
1067      previous backup points can be taken. */      previous backup points can be taken. */
# Line 900  for (;;) Line 1073  for (;;)
1073    
1074      for (;;)      for (;;)
1075        {        {
1076        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA || op == OP_ONCE)
1077        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,          md->match_function_type = MATCH_CBEGROUP;
1078    
1079          /* If this is not a possibly empty group, and there are no (*THEN)s in
1080          the pattern, and this is the final alternative, optimize as described
1081          above. */
1082    
1083          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1084            {
1085            ecode += PRIV(OP_lengths)[*ecode];
1086            goto TAIL_RECURSE;
1087            }
1088    
1089          /* In all other cases, we have to make another call to match(). */
1090    
1091          save_mark = md->mark;
1092          save_capture_last = md->capture_last;
1093          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1094          RM2);          RM2);
1095        if (rrc != MATCH_NOMATCH &&  
1096            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* See comment in the code for capturing groups above about handling
1097          {        THEN. */
1098    
1099          if (rrc == MATCH_THEN)
1100            {
1101            next = ecode + GET(ecode,1);
1102            if (md->start_match_ptr < next &&
1103                (*ecode == OP_ALT || *next == OP_ALT))
1104              rrc = MATCH_NOMATCH;
1105            }
1106    
1107          if (rrc != MATCH_NOMATCH)
1108            {
1109          if (rrc == MATCH_ONCE)          if (rrc == MATCH_ONCE)
1110            {            {
1111            const uschar *scode = ecode;            const pcre_uchar *scode = ecode;
1112            if (*scode != OP_ONCE)           /* If not at start, find it */            if (*scode != OP_ONCE)           /* If not at start, find it */
1113              {              {
1114              while (*scode == OP_ALT) scode += GET(scode, 1);              while (*scode == OP_ALT) scode += GET(scode, 1);
1115              scode -= GET(scode, 1);              scode -= GET(scode, 1);
1116              }              }
1117            if (md->once_target == scode) rrc = MATCH_NOMATCH;            if (md->once_target == scode) rrc = MATCH_NOMATCH;
1118            }            }
1119          RRETURN(rrc);          RRETURN(rrc);
1120          }          }
1121        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1122        if (*ecode != OP_ALT) break;        md->mark = save_mark;
1123          if (*ecode != OP_ALT) break;
1124          md->capture_last = save_capture_last;
1125        }        }
1126      if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;  
1127      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1128    
1129      /* Handle possessive capturing brackets with an unlimited repeat. We come      /* Handle possessive capturing brackets with an unlimited repeat. We come
1130      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
1131      handled similarly to the normal case above. However, the matching is      handled similarly to the normal case above. However, the matching is
1132      different. The end of these brackets will always be OP_KETRPOS, which      different. The end of these brackets will always be OP_KETRPOS, which
1133      returns MATCH_KETRPOS without going further in the pattern. By this means      returns MATCH_KETRPOS without going further in the pattern. By this means
1134      we can handle the group by iteration rather than recursion, thereby      we can handle the group by iteration rather than recursion, thereby
1135      reducing the amount of stack needed. */      reducing the amount of stack needed. */
1136    
1137      case OP_CBRAPOS:      case OP_CBRAPOS:
1138      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1139      allow_zero = FALSE;      allow_zero = FALSE;
1140    
1141      POSSESSIVE_CAPTURE:      POSSESSIVE_CAPTURE:
1142      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
1143      offset = number << 1;      offset = number << 1;
# Line 950  for (;;) Line 1152  for (;;)
1152      if (offset < md->offset_max)      if (offset < md->offset_max)
1153        {        {
1154        matched_once = FALSE;        matched_once = FALSE;
1155        code_offset = ecode - md->start_code;        code_offset = (int)(ecode - md->start_code);
1156    
1157        save_offset1 = md->offset_vector[offset];        save_offset1 = md->offset_vector[offset];
1158        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
# Line 958  for (;;) Line 1160  for (;;)
1160        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
1161    
1162        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1163    
1164        /* Each time round the loop, save the current subject position for use        /* Each time round the loop, save the current subject position for use
1165        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
1166        restart it with a new subject starting position, remembering that we had        restart it with a new subject starting position, remembering that we had
1167        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
1168        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
1169        see if a previous iteration matched. If so, the group has matched;        see if a previous iteration matched. If so, the group has matched;
1170        continue from afterwards. Otherwise it has failed; restore the previous        continue from afterwards. Otherwise it has failed; restore the previous
1171        capture values before returning NOMATCH. */        capture values before returning NOMATCH. */
1172    
1173        for (;;)        for (;;)
1174          {          {
1175          md->offset_vector[md->offset_end - number] =          md->offset_vector[md->offset_end - number] =
1176            (int)(eptr - md->start_subject);            (int)(eptr - md->start_subject);
1177          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1178          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1179            eptrb, RM63);            eptrb, RM63);
1180          if (rrc == MATCH_KETRPOS)          if (rrc == MATCH_KETRPOS)
1181            {            {
1182            offset_top = md->end_offset_top;            offset_top = md->end_offset_top;
1183            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
1184            ecode = md->start_code + code_offset;            ecode = md->start_code + code_offset;
1185            save_capture_last = md->capture_last;            save_capture_last = md->capture_last;
1186            matched_once = TRUE;            matched_once = TRUE;
1187            continue;            continue;
1188            }            }
1189          if (rrc != MATCH_NOMATCH &&  
1190              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* See comment in the code for capturing groups above about handling
1191            RRETURN(rrc);          THEN. */
1192    
1193            if (rrc == MATCH_THEN)
1194              {
1195              next = ecode + GET(ecode,1);
1196              if (md->start_match_ptr < next &&
1197                  (*ecode == OP_ALT || *next == OP_ALT))
1198                rrc = MATCH_NOMATCH;
1199              }
1200    
1201            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1202          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
1203          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
1204          if (*ecode != OP_ALT) break;          if (*ecode != OP_ALT) break;
1205          }          }
1206    
1207        if (!matched_once)        if (!matched_once)
1208          {          {
1209          md->offset_vector[offset] = save_offset1;          md->offset_vector[offset] = save_offset1;
1210          md->offset_vector[offset+1] = save_offset2;          md->offset_vector[offset+1] = save_offset2;
1211          md->offset_vector[md->offset_end - number] = save_offset3;          md->offset_vector[md->offset_end - number] = save_offset3;
1212          }          }
1213    
       if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;  
1214        if (allow_zero || matched_once)        if (allow_zero || matched_once)
1215          {          {
1216          ecode += 1 + LINK_SIZE;          ecode += 1 + LINK_SIZE;
1217          break;          break;
1218          }          }
1219    
1220        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1221        }        }
1222    
1223      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1224      as a non-capturing bracket. */      as a non-capturing bracket. */
1225    
# Line 1020  for (;;) Line 1231  for (;;)
1231      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1232      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1233    
1234      /* Non-capturing possessive bracket with unlimited repeat. We come here      /* Non-capturing possessive bracket with unlimited repeat. We come here
1235      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,
1236      without the capturing complication. It is written out separately for speed      without the capturing complication. It is written out separately for speed
1237      and cleanliness. */      and cleanliness. */
1238    
1239      case OP_BRAPOS:      case OP_BRAPOS:
1240      case OP_SBRAPOS:      case OP_SBRAPOS:
1241      allow_zero = FALSE;      allow_zero = FALSE;
1242    
1243      POSSESSIVE_NON_CAPTURE:      POSSESSIVE_NON_CAPTURE:
1244      matched_once = FALSE;      matched_once = FALSE;
1245      code_offset = ecode - md->start_code;      code_offset = (int)(ecode - md->start_code);
1246        save_capture_last = md->capture_last;
1247    
1248      for (;;)      for (;;)
1249        {        {
1250        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1251        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1252          eptrb, RM48);          eptrb, RM48);
1253        if (rrc == MATCH_KETRPOS)        if (rrc == MATCH_KETRPOS)
1254          {          {
1255          offset_top = md->end_offset_top;          offset_top = md->end_offset_top;
1256          eptr = md->end_match_ptr;          eptr = md->end_match_ptr;
1257          ecode = md->start_code + code_offset;          ecode = md->start_code + code_offset;
1258          matched_once = TRUE;          matched_once = TRUE;
1259          continue;          continue;
1260          }          }
1261        if (rrc != MATCH_NOMATCH &&  
1262            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* See comment in the code for capturing groups above about handling
1263          RRETURN(rrc);        THEN. */
1264    
1265          if (rrc == MATCH_THEN)
1266            {
1267            next = ecode + GET(ecode,1);
1268            if (md->start_match_ptr < next &&
1269                (*ecode == OP_ALT || *next == OP_ALT))
1270              rrc = MATCH_NOMATCH;
1271            }
1272    
1273          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1274        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1275        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1276          md->capture_last = save_capture_last;
1277        }        }
1278    
1279      if (matched_once || allow_zero)      if (matched_once || allow_zero)
1280        {        {
1281        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1282        break;        break;
1283        }        }
1284      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1285    
1286      /* Control never reaches here. */      /* Control never reaches here. */
# Line 1076  for (;;) Line 1299  for (;;)
1299    
1300      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1301        {        {
1302        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1303          {          {
1304          pcre_callout_block cb;          PUBL(callout_block) cb;
1305          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1306          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1307          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1308    #if defined COMPILE_PCRE8
1309          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1310    #elif defined COMPILE_PCRE16
1311            cb.subject          = (PCRE_SPTR16)md->start_subject;
1312    #elif defined COMPILE_PCRE32
1313            cb.subject          = (PCRE_SPTR32)md->start_subject;
1314    #endif
1315          cb.subject_length   = (int)(md->end_subject - md->start_subject);          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1316          cb.start_match      = (int)(mstart - md->start_subject);          cb.start_match      = (int)(mstart - md->start_subject);
1317          cb.current_position = (int)(eptr - md->start_subject);          cb.current_position = (int)(eptr - md->start_subject);
1318          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1319          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1320          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1321          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last & CAPLMASK;
1322            /* Internal change requires this for API compatibility. */
1323            if (cb.capture_last == 0) cb.capture_last = -1;
1324          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1325          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1326            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1327          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1328          }          }
1329        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1330          codelink -= PRIV(OP_lengths)[OP_CALLOUT];
1331        }        }
1332    
1333      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
# Line 1110  for (;;) Line 1343  for (;;)
1343          }          }
1344        else        else
1345          {          {
1346          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          unsigned int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1347          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1348    
1349          /* 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
1350          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
1351          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
1352          if any one is set. */          if any one is set. */
1353    
1354          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1355            {            {
1356            uschar *slotA = md->name_table;            pcre_uchar *slotA = md->name_table;
1357            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
1358              {              {
1359              if (GET2(slotA, 0) == recno) break;              if (GET2(slotA, 0) == recno) break;
# Line 1133  for (;;) Line 1366  for (;;)
1366    
1367            if (i < md->name_count)            if (i < md->name_count)
1368              {              {
1369              uschar *slotB = slotA;              pcre_uchar *slotB = slotA;
1370              while (slotB > md->name_table)              while (slotB > md->name_table)
1371                {                {
1372                slotB -= md->name_entry_size;                slotB -= md->name_entry_size;
1373                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1374                  {                  {
1375                  condition = GET2(slotB, 0) == md->recursive->group_num;                  condition = GET2(slotB, 0) == md->recursive->group_num;
1376                  if (condition) break;                  if (condition) break;
# Line 1153  for (;;) Line 1386  for (;;)
1386                for (i++; i < md->name_count; i++)                for (i++; i < md->name_count; i++)
1387                  {                  {
1388                  slotB += md->name_entry_size;                  slotB += md->name_entry_size;
1389                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1390                    {                    {
1391                    condition = GET2(slotB, 0) == md->recursive->group_num;                    condition = GET2(slotB, 0) == md->recursive->group_num;
1392                    if (condition) break;                    if (condition) break;
# Line 1166  for (;;) Line 1399  for (;;)
1399    
1400          /* Chose branch according to the condition */          /* Chose branch according to the condition */
1401    
1402          ecode += condition? 3 : GET(ecode, 1);          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1403          }          }
1404        }        }
1405    
# Line 1182  for (;;) Line 1415  for (;;)
1415    
1416        if (!condition && condcode == OP_NCREF)        if (!condition && condcode == OP_NCREF)
1417          {          {
1418          int refno = offset >> 1;          unsigned int refno = offset >> 1;
1419          uschar *slotA = md->name_table;          pcre_uchar *slotA = md->name_table;
1420    
1421          for (i = 0; i < md->name_count; i++)          for (i = 0; i < md->name_count; i++)
1422            {            {
# Line 1197  for (;;) Line 1430  for (;;)
1430    
1431          if (i < md->name_count)          if (i < md->name_count)
1432            {            {
1433            uschar *slotB = slotA;            pcre_uchar *slotB = slotA;
1434            while (slotB > md->name_table)            while (slotB > md->name_table)
1435              {              {
1436              slotB -= md->name_entry_size;              slotB -= md->name_entry_size;
1437              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)              if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1438                {                {
1439                offset = GET2(slotB, 0) << 1;                offset = GET2(slotB, 0) << 1;
1440                condition = offset < offset_top &&                condition = offset < offset_top &&
# Line 1219  for (;;) Line 1452  for (;;)
1452              for (i++; i < md->name_count; i++)              for (i++; i < md->name_count; i++)
1453                {                {
1454                slotB += md->name_entry_size;                slotB += md->name_entry_size;
1455                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1456                  {                  {
1457                  offset = GET2(slotB, 0) << 1;                  offset = GET2(slotB, 0) << 1;
1458                  condition = offset < offset_top &&                  condition = offset < offset_top &&
# Line 1234  for (;;) Line 1467  for (;;)
1467    
1468        /* Chose branch according to the condition */        /* Chose branch according to the condition */
1469    
1470        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1471        }        }
1472    
1473      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 1249  for (;;) Line 1482  for (;;)
1482    
1483      else      else
1484        {        {
1485        md->match_function_type = MATCH_CONDASSERT;        md->match_function_type = MATCH_CONDASSERT;
1486        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1487        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1488          {          {
1489            if (md->end_offset_top > offset_top)
1490              offset_top = md->end_offset_top;  /* Captures may have happened */
1491          condition = TRUE;          condition = TRUE;
1492          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1493          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1494          }          }
1495        else if (rrc != MATCH_NOMATCH &&  
1496                (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1497          assertion; it is therefore treated as NOMATCH. */
1498    
1499          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1500          {          {
1501          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1502          }          }
# Line 1269  for (;;) Line 1507  for (;;)
1507          }          }
1508        }        }
1509    
1510      /* 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
1511      we used to use tail recursion to avoid using another stack frame, except      use tail recursion to avoid using another stack frame, except when there is
1512      when there was unlimited repeat of a possibly empty group. However, that      unlimited repeat of a possibly empty group. In the latter case, a recursive
1513      strategy no longer works because of the possibilty of (*THEN) being      call to match() is always required, unless the second alternative doesn't
1514      encountered in the branch. A recursive call to match() is always required,      exist, in which case we can just plough on. Note that, for compatibility
1515      unless the second alternative doesn't exist, in which case we can just      with Perl, the | in a conditional group is NOT treated as creating two
1516      plough on. */      alternatives. If a THEN is encountered in the branch, it propagates out to
1517        the enclosing alternative (unless nested in a deeper set of alternatives,
1518        of course). */
1519    
1520      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1521        {        {
1522        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;        if (op != OP_SCOND)
1523            {
1524            ecode += 1 + LINK_SIZE;
1525            goto TAIL_RECURSE;
1526            }
1527    
1528          md->match_function_type = MATCH_CBEGROUP;
1529        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
       if (rrc == MATCH_THEN && md->start_match_ptr == ecode)  
         rrc = MATCH_NOMATCH;  
1530        RRETURN(rrc);        RRETURN(rrc);
1531        }        }
1532      else                         /* Condition false & no alternative */  
1533         /* Condition false & no alternative; continue after the group. */
1534    
1535        else
1536        {        {
1537        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1538        }        }
# Line 1296  for (;;) Line 1543  for (;;)
1543      to close any currently open capturing brackets. */      to close any currently open capturing brackets. */
1544    
1545      case OP_CLOSE:      case OP_CLOSE:
1546      number = GET2(ecode, 1);      number = GET2(ecode, 1);   /* Must be less than 65536 */
1547      offset = number << 1;      offset = number << 1;
1548    
1549  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 1304  for (;;) Line 1551  for (;;)
1551        printf("\n");        printf("\n");
1552  #endif  #endif
1553    
1554      md->capture_last = number;      md->capture_last = (md->capture_last & OVFLMASK) | number;
1555      if (offset >= md->offset_max) md->offset_overflow = TRUE; else      if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else
1556        {        {
1557        md->offset_vector[offset] =        md->offset_vector[offset] =
1558          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1559        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1560        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1561        }        }
1562      ecode += 3;      ecode += 1 + IMM2_SIZE;
1563      break;      break;
1564    
1565    
1566      /* 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. */  
1567    
     case OP_ACCEPT:  
     case OP_ASSERT_ACCEPT:  
1568      case OP_END:      case OP_END:
1569        case OP_ACCEPT:
1570  /*      case OP_ASSERT_ACCEPT:
     if (md->recursive != NULL)  
       {  
       recursion_info *rec = md->recursive;  
   
       md->recursive = rec->prevrec;  
1571    
1572        memmove(md->offset_vector, rec->offset_save,      /* If we have matched an empty string, fail if not in an assertion and not
1573          rec->saved_max * sizeof(int));      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
       offset_top = rec->save_offset_top;  
       if (rec->group_num == 0)  
         {  
         ecode = rec->after_call;  
         break;  
         }  
       }  
 */  
     /* Otherwise, if we have matched an empty string, fail if not in an  
     assertion and if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART  
1574      is set and we have matched at the start of the subject. In both cases,      is set and we have matched at the start of the subject. In both cases,
1575      backtracking will then try other alternatives, if any. */      backtracking will then try other alternatives, if any. */
1576    
1577  /*    else */ if (eptr == mstart && op != OP_ASSERT_ACCEPT &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
   
1578           md->recursive == NULL &&           md->recursive == NULL &&
1579             (md->notempty ||
1580          (md->notempty ||             (md->notempty_atstart &&
1581            (md->notempty_atstart &&               mstart == md->start_subject + md->start_offset)))
1582              mstart == md->start_subject + md->start_offset)))        RRETURN(MATCH_NOMATCH);
       MRRETURN(MATCH_NOMATCH);  
1583    
1584      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1585    
1586      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1587      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1588      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1589    
1590      /* 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
1591      given as the argument to MRRETURN when the heap is in use. */      given as the argument to RRETURN when the heap is in use. */
1592    
1593      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1594      MRRETURN(rrc);      RRETURN(rrc);
1595    
1596      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1597      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,
1598      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
1599      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
1600      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
1601      of a condition, we want to return immediately afterwards. The caller of      of a condition, we want to return immediately afterwards. The caller of
1602      this incarnation of the match() function will have set MATCH_CONDASSERT in      this incarnation of the match() function will have set MATCH_CONDASSERT in
1603      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
1604      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
1605      match() to remember this case. */      match() to remember this case. */
1606    
1607      case OP_ASSERT:      case OP_ASSERT:
1608      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1609        save_mark = md->mark;
1610      if (md->match_function_type == MATCH_CONDASSERT)      if (md->match_function_type == MATCH_CONDASSERT)
1611        {        {
1612        condassert = TRUE;        condassert = TRUE;
1613        md->match_function_type = 0;        md->match_function_type = 0;
1614        }        }
1615      else condassert = FALSE;      else condassert = FALSE;
1616    
1617      do      do
1618        {        {
1619        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
# Line 1395  for (;;) Line 1622  for (;;)
1622          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1623          break;          break;
1624          }          }
1625        if (rrc != MATCH_NOMATCH &&        md->mark = save_mark;
1626            (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
1627          RRETURN(rrc);        /* A COMMIT failure must fail the entire assertion, without trying any
1628          subsequent branches. */
1629    
1630          if (rrc == MATCH_COMMIT) RRETURN(MATCH_NOMATCH);
1631    
1632          /* PCRE does not allow THEN to escape beyond an assertion; it
1633          is treated as NOMATCH. */
1634    
1635          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1636        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1637        }        }
1638      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1639    
1640      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1641    
1642      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1643    
# Line 1422  for (;;) Line 1657  for (;;)
1657    
1658      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1659      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1660        save_mark = md->mark;
1661      if (md->match_function_type == MATCH_CONDASSERT)      if (md->match_function_type == MATCH_CONDASSERT)
1662        {        {
1663        condassert = TRUE;        condassert = TRUE;
1664        md->match_function_type = 0;        md->match_function_type = 0;
1665        }        }
1666      else condassert = FALSE;      else condassert = FALSE;
1667    
1668      do      do
1669        {        {
1670        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1671        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        md->mark = save_mark;
1672          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1673        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1674          {          {
1675          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1676          break;          break;
1677          }          }
1678        if (rrc != MATCH_NOMATCH &&  
1679            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1680          RRETURN(rrc);        as NOMATCH. */
1681    
1682          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1683        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1684        }        }
1685      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1686    
1687      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1688    
1689      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1690      continue;      continue;
1691    
# Line 1456  for (;;) Line 1695  for (;;)
1695      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1696    
1697      case OP_REVERSE:      case OP_REVERSE:
1698  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1699      if (utf8)      if (utf)
1700        {        {
1701        i = GET(ecode, 1);        i = GET(ecode, 1);
1702        while (i-- > 0)        while (i-- > 0)
1703          {          {
1704          eptr--;          eptr--;
1705          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1706          BACKCHAR(eptr);          BACKCHAR(eptr);
1707          }          }
1708        }        }
# Line 1474  for (;;) Line 1713  for (;;)
1713    
1714        {        {
1715        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1716        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1717        }        }
1718    
1719      /* Save the earliest consulted character, then skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
# Line 1488  for (;;) Line 1727  for (;;)
1727      function is able to force a failure. */      function is able to force a failure. */
1728    
1729      case OP_CALLOUT:      case OP_CALLOUT:
1730      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1731        {        {
1732        pcre_callout_block cb;        PUBL(callout_block) cb;
1733        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1734        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1735        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1736    #if defined COMPILE_PCRE8
1737        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1738    #elif defined COMPILE_PCRE16
1739          cb.subject          = (PCRE_SPTR16)md->start_subject;
1740    #elif defined COMPILE_PCRE32
1741          cb.subject          = (PCRE_SPTR32)md->start_subject;
1742    #endif
1743        cb.subject_length   = (int)(md->end_subject - md->start_subject);        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1744        cb.start_match      = (int)(mstart - md->start_subject);        cb.start_match      = (int)(mstart - md->start_subject);
1745        cb.current_position = (int)(eptr - md->start_subject);        cb.current_position = (int)(eptr - md->start_subject);
1746        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1747        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1748        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1749        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last & CAPLMASK;
1750          /* Internal change requires this for API compatibility. */
1751          if (cb.capture_last == 0) cb.capture_last = -1;
1752        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1753        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1754          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1755        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1756        }        }
1757      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1512  for (;;) Line 1760  for (;;)
1760      /* Recursion either matches the current regex, or some subexpression. The      /* Recursion either matches the current regex, or some subexpression. The
1761      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
1762      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1763    
1764      The state of the capturing groups is preserved over recursion, and      The state of the capturing groups is preserved over recursion, and
1765      re-instated afterwards. We don't know how many are started and not yet      re-instated afterwards. We don't know how many are started and not yet
1766      finished (offset_top records the completed total) so we just have to save      finished (offset_top records the completed total) so we just have to save
1767      all the potential data. There may be up to 65535 such values, which is too      all the potential data. There may be up to 65535 such values, which is too
1768      large to put on the stack, but using malloc for small numbers seems      large to put on the stack, but using malloc for small numbers seems
# Line 1523  for (;;) Line 1771  for (;;)
1771    
1772      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
1773      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
1774      for the original version of this logic. It has, however, been hacked around      for the original version of this logic. It has, however, been hacked around
1775      a lot, so he is not to blame for the current way it works. */      a lot, so he is not to blame for the current way it works. */
1776    
1777      case OP_RECURSE:      case OP_RECURSE:
1778        {        {
1779          recursion_info *ri;
1780          unsigned int recno;
1781    
1782        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1783        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1784          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1785    
1786          /* Check for repeating a recursion without advancing the subject pointer.
1787          This should catch convoluted mutual recursions. (Some simple cases are
1788          caught at compile time.) */
1789    
1790          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1791            if (recno == ri->group_num && eptr == ri->subject_position)
1792              RRETURN(PCRE_ERROR_RECURSELOOP);
1793    
1794        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1795    
1796          new_recursive.group_num = recno;
1797          new_recursive.saved_capture_last = md->capture_last;
1798          new_recursive.subject_position = eptr;
1799        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1800        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1801    
# Line 1549  for (;;) Line 1811  for (;;)
1811        else        else
1812          {          {
1813          new_recursive.offset_save =          new_recursive.offset_save =
1814            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1815          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1816          }          }
1817        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1818              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1819    
1820        /* OK, now we can do the recursion. After processing each alternative,        /* OK, now we can do the recursion. After processing each alternative,
1821        restore the offset data. If there were nested recursions, md->recursive        restore the offset data and the last captured value. If there were nested
1822        might be changed, so reset it before looping. */        recursions, md->recursive might be changed, so reset it before looping.
1823          */
1824    
1825        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1826        cbegroup = (*callpat >= OP_SBRA);        cbegroup = (*callpat >= OP_SBRA);
1827        do        do
1828          {          {
1829          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1830          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1831            md, eptrb, RM6);            md, eptrb, RM6);
1832          memcpy(md->offset_vector, new_recursive.offset_save,          memcpy(md->offset_vector, new_recursive.offset_save,
1833              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1834            md->capture_last = new_recursive.saved_capture_last;
1835            md->recursive = new_recursive.prevrec;
1836          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1837            {            {
1838            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1839            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1840              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1841    
1842            /* Set where we got to in the subject, and reset the start in case            /* Set where we got to in the subject, and reset the start in case
1843            it was changed by \K. This *is* propagated back out of a recursion,            it was changed by \K. This *is* propagated back out of a recursion,
1844            for Perl compatibility. */            for Perl compatibility. */
1845    
1846            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
1847            mstart = md->start_match_ptr;            mstart = md->start_match_ptr;
1848            goto RECURSION_MATCHED;        /* Exit loop; end processing */            goto RECURSION_MATCHED;        /* Exit loop; end processing */
1849            }            }
1850          else if (rrc != MATCH_NOMATCH &&  
1851                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* PCRE does not allow THEN, SKIP, PRUNE or COMMIT to escape beyond a
1852            recursion; they are treated as NOMATCH. These codes are defined in a
1853            range that can be tested for. Any other return code is an error. */
1854    
1855            else if (rrc != MATCH_NOMATCH &&
1856                     (rrc < MATCH_BACKTRACK_MIN || rrc > MATCH_BACKTRACK_MAX))
1857            {            {
1858            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1859            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1860              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1861            RRETURN(rrc);            RRETURN(rrc);
1862            }            }
1863    
# Line 1600  for (;;) Line 1869  for (;;)
1869        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1870        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1871        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1872          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1873        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1874        }        }
1875    
1876      RECURSION_MATCHED:      RECURSION_MATCHED:
1877      break;      break;
1878    
# Line 1619  for (;;) Line 1888  for (;;)
1888      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
1889      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
1890      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1891    
1892      case OP_BRAZERO:      case OP_BRAZERO:
1893      next = ecode + 1;      next = ecode + 1;
1894      RMATCH(eptr, next, offset_top, md, eptrb, RM10);      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
# Line 1627  for (;;) Line 1896  for (;;)
1896      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
1897      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1898      break;      break;
1899    
1900      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1901      next = ecode + 1;      next = ecode + 1;
1902      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
# Line 1641  for (;;) Line 1910  for (;;)
1910      do next += GET(next,1); while (*next == OP_ALT);      do next += GET(next,1); while (*next == OP_ALT);
1911      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1912      break;      break;
1913    
1914      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1915      here; just jump to the group, with allow_zero set TRUE. */      here; just jump to the group, with allow_zero set TRUE. */
1916    
1917      case OP_BRAPOSZERO:      case OP_BRAPOSZERO:
1918      op = *(++ecode);      op = *(++ecode);
1919      allow_zero = TRUE;      allow_zero = TRUE;
1920      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1921        goto POSSESSIVE_NON_CAPTURE;        goto POSSESSIVE_NON_CAPTURE;
# Line 1656  for (;;) Line 1925  for (;;)
1925      case OP_KET:      case OP_KET:
1926      case OP_KETRMIN:      case OP_KETRMIN:
1927      case OP_KETRMAX:      case OP_KETRMAX:
1928      case OP_KETRPOS:      case OP_KETRPOS:
1929      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1930    
1931      /* 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
1932      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1933      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
# Line 1670  for (;;) Line 1939  for (;;)
1939        }        }
1940      else saved_eptr = NULL;      else saved_eptr = NULL;
1941    
1942      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group or a non-capturing atomic
1943      MATCH_MATCH, but record the current high water mark for use by positive      group, stop matching and return MATCH_MATCH, but record the current high
1944      assertions. We also need to record the match start in case it was changed      water mark for use by positive assertions. We also need to record the match
1945      by \K. */      start in case it was changed by \K. */
1946    
1947      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1948          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)           *prev == OP_ONCE_NC)
1949        {        {
1950        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1951        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1952        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1953        MRRETURN(MATCH_MATCH);        RRETURN(MATCH_MATCH);         /* Sets md->mark */
1954        }        }
1955    
1956      /* 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
# Line 1714  for (;;) Line 1983  for (;;)
1983    
1984        /* Deal with capturing */        /* Deal with capturing */
1985    
1986        md->capture_last = number;        md->capture_last = (md->capture_last & OVFLMASK) | number;
1987        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else
1988          {          {
1989          /* If offset is greater than offset_top, it means that we are          /* If offset is greater than offset_top, it means that we are
1990          "skipping" a capturing group, and that group's offsets must be marked          "skipping" a capturing group, and that group's offsets must be marked
1991          unset. In earlier versions of PCRE, all the offsets were unset at the          unset. In earlier versions of PCRE, all the offsets were unset at the
1992          start of matching, but this doesn't work because atomic groups and          start of matching, but this doesn't work because atomic groups and
1993          assertions can cause a value to be set that should later be unset.          assertions can cause a value to be set that should later be unset.
1994          Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as          Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1995          part of the atomic group, but this is not on the final matching path,          part of the atomic group, but this is not on the final matching path,
1996          so must be unset when 2 is set. (If there is no group 2, there is no          so must be unset when 2 is set. (If there is no group 2, there is no
1997          problem, because offset_top will then be 2, indicating no capture.) */          problem, because offset_top will then be 2, indicating no capture.) */
1998    
1999          if (offset > offset_top)          if (offset > offset_top)
2000            {            {
2001            register int *iptr = md->offset_vector + offset_top;            register int *iptr = md->offset_vector + offset_top;
2002            register int *iend = md->offset_vector + offset;            register int *iend = md->offset_vector + offset;
2003            while (iptr < iend) *iptr++ = -1;            while (iptr < iend) *iptr++ = -1;
2004            }            }
2005    
2006          /* Now make the extraction */          /* Now make the extraction */
2007    
2008          md->offset_vector[offset] =          md->offset_vector[offset] =
# Line 1746  for (;;) Line 2015  for (;;)
2015      /* For an ordinary non-repeating ket, just continue at this level. This      /* For an ordinary non-repeating ket, just continue at this level. This
2016      also happens for a repeating ket if no characters were matched in the      also happens for a repeating ket if no characters were matched in the
2017      group. This is the forcible breaking of infinite loops as implemented in      group. This is the forcible breaking of infinite loops as implemented in
2018      Perl 5.005. For a non-repeating atomic group, establish a backup point by      Perl 5.005. For a non-repeating atomic group that includes captures,
2019      processing the rest of the pattern at a lower level. If this results in a      establish a backup point by processing the rest of the pattern at a lower
2020      NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby      level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
2021      bypassing intermediate backup points, but resetting any captures that      original OP_ONCE level, thereby bypassing intermediate backup points, but
2022      happened along the way. */      resetting any captures that happened along the way. */
2023    
2024      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
2025        {        {
# Line 1759  for (;;) Line 2028  for (;;)
2028          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
2029          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2030          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2031          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
2032          }          }
2033        ecode += 1 + LINK_SIZE;    /* Carry on at this level */        ecode += 1 + LINK_SIZE;    /* Carry on at this level */
2034        break;        break;
2035        }        }
2036    
2037      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
2038      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
2039      at a time from the outer level, thus saving stack. */      at a time from the outer level, thus saving stack. */
2040    
2041      if (*ecode == OP_KETRPOS)      if (*ecode == OP_KETRPOS)
2042        {        {
2043        md->end_match_ptr = eptr;        md->end_match_ptr = eptr;
2044        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
2045        RRETURN(MATCH_KETRPOS);        RRETURN(MATCH_KETRPOS);
2046        }        }
2047    
2048      /* 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
2049      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
# Line 1784  for (;;) Line 2053  for (;;)
2053    
2054      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
2055        {        {
2056        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
2057        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2058        if (*prev == OP_ONCE)        if (*prev == OP_ONCE)
2059          {          {
2060          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
2061          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2062          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2063          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
2064          }          }
2065        if (*prev >= OP_SBRA)    /* Could match an empty string */        if (*prev >= OP_SBRA)    /* Could match an empty string */
2066          {          {
         md->match_function_type = MATCH_CBEGROUP;  
2067          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
2068          RRETURN(rrc);          RRETURN(rrc);
2069          }          }
# Line 1804  for (;;) Line 2072  for (;;)
2072        }        }
2073      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
2074        {        {
       if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;  
2075        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
2076        if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;        if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
2077        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2078        if (*prev == OP_ONCE)        if (*prev == OP_ONCE)
2079          {          {
2080          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
2081          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2082          md->once_target = prev;          md->once_target = prev;
2083          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
2084          }          }
2085        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
2086        goto TAIL_RECURSE;        goto TAIL_RECURSE;
2087        }        }
# Line 1823  for (;;) Line 2090  for (;;)
2090      /* Not multiline mode: start of subject assertion, unless notbol. */      /* Not multiline mode: start of subject assertion, unless notbol. */
2091    
2092      case OP_CIRC:      case OP_CIRC:
2093      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2094    
2095      /* Start of subject assertion */      /* Start of subject assertion */
2096    
2097      case OP_SOD:      case OP_SOD:
2098      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);
2099      ecode++;      ecode++;
2100      break;      break;
2101    
2102      /* Multiline mode: start of subject unless notbol, or after any newline. */      /* Multiline mode: start of subject unless notbol, or after any newline. */
2103    
2104      case OP_CIRCM:      case OP_CIRCM:
2105      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2106      if (eptr != md->start_subject &&      if (eptr != md->start_subject &&
2107          (eptr == md->end_subject || !WAS_NEWLINE(eptr)))          (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
2108        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2109      ecode++;      ecode++;
2110      break;      break;
2111    
2112      /* Start of match assertion */      /* Start of match assertion */
2113    
2114      case OP_SOM:      case OP_SOM:
2115      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);
2116      ecode++;      ecode++;
2117      break;      break;
2118    
# Line 1861  for (;;) Line 2128  for (;;)
2128    
2129      case OP_DOLLM:      case OP_DOLLM:
2130      if (eptr < md->end_subject)      if (eptr < md->end_subject)
2131        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }        {
2132          if (!IS_NEWLINE(eptr))
2133            {
2134            if (md->partial != 0 &&
2135                eptr + 1 >= md->end_subject &&
2136                NLBLOCK->nltype == NLTYPE_FIXED &&
2137                NLBLOCK->nllen == 2 &&
2138                RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2139              {
2140              md->hitend = TRUE;
2141              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2142              }
2143            RRETURN(MATCH_NOMATCH);
2144            }
2145          }
2146      else      else
2147        {        {
2148        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2149        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2150        }        }
2151      ecode++;      ecode++;
2152      break;      break;
2153    
2154      /* Not multiline mode: assert before a terminating newline or before end of      /* Not multiline mode: assert before a terminating newline or before end of
2155      subject unless noteol is set. */      subject unless noteol is set. */
2156    
2157      case OP_DOLL:      case OP_DOLL:
2158      if (md->noteol) MRRETURN(MATCH_NOMATCH);      if (md->noteol) RRETURN(MATCH_NOMATCH);
2159      if (!md->endonly) goto ASSERT_NL_OR_EOS;      if (!md->endonly) goto ASSERT_NL_OR_EOS;
2160    
2161      /* ... else fall through for endonly */      /* ... else fall through for endonly */
# Line 1882  for (;;) Line 2163  for (;;)
2163      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2164    
2165      case OP_EOD:      case OP_EOD:
2166      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2167      SCHECK_PARTIAL();      SCHECK_PARTIAL();
2168      ecode++;      ecode++;
2169      break;      break;
# Line 1893  for (;;) Line 2174  for (;;)
2174      ASSERT_NL_OR_EOS:      ASSERT_NL_OR_EOS:
2175      if (eptr < md->end_subject &&      if (eptr < md->end_subject &&
2176          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2177        MRRETURN(MATCH_NOMATCH);        {
2178          if (md->partial != 0 &&
2179              eptr + 1 >= md->end_subject &&
2180              NLBLOCK->nltype == NLTYPE_FIXED &&
2181              NLBLOCK->nllen == 2 &&
2182              RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2183            {
2184            md->hitend = TRUE;
2185            if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2186            }
2187          RRETURN(MATCH_NOMATCH);
2188          }
2189    
2190      /* Either at end of string or \n before end. */      /* Either at end of string or \n before end. */
2191    
# Line 1912  for (;;) Line 2204  for (;;)
2204        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2205        partial matching. */        partial matching. */
2206    
2207  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2208        if (utf8)        if (utf)
2209          {          {
2210          /* Get status of previous character */          /* Get status of previous character */
2211    
2212          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2213            {            {
2214            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2215            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2216            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2217            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2218  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 1985  for (;;) Line 2277  for (;;)
2277              }              }
2278            else            else
2279  #endif  #endif
2280            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);            prev_is_word = MAX_255(eptr[-1])
2281                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2282            }            }
2283    
2284          /* Get status of next character */          /* Get status of next character */
# Line 2008  for (;;) Line 2301  for (;;)
2301            }            }
2302          else          else
2303  #endif  #endif
2304          cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          cur_is_word = MAX_255(*eptr)
2305              && ((md->ctypes[*eptr] & ctype_word) != 0);
2306          }          }
2307    
2308        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2309    
2310        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2311             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2312          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2313        }        }
2314      break;      break;
2315    
2316      /* Match a single character type; inline for speed */      /* Match any single character type except newline; have to take care with
2317        CRLF newlines and partial matching. */
2318    
2319      case OP_ANY:      case OP_ANY:
2320      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
2321        if (md->partial != 0 &&
2322            eptr + 1 >= md->end_subject &&
2323            NLBLOCK->nltype == NLTYPE_FIXED &&
2324            NLBLOCK->nllen == 2 &&
2325            RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2326          {
2327          md->hitend = TRUE;
2328          if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2329          }
2330    
2331      /* Fall through */      /* Fall through */
2332    
2333        /* Match any single character whatsoever. */
2334    
2335      case OP_ALLANY:      case OP_ALLANY:
2336      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2337        {        {                            /* not be updated before SCHECK_PARTIAL. */
2338        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2339        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2340        }        }
2341      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2342    #ifdef SUPPORT_UTF
2343        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2344    #endif
2345      ecode++;      ecode++;
2346      break;      break;
2347    
# Line 2039  for (;;) Line 2349  for (;;)
2349      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2350    
2351      case OP_ANYBYTE:      case OP_ANYBYTE:
2352      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2353        {        {                            /* not be updated before SCHECK_PARTIAL. */
2354        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2355        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2356        }        }
2357        eptr++;
2358      ecode++;      ecode++;
2359      break;      break;
2360    
# Line 2051  for (;;) Line 2362  for (;;)
2362      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2363        {        {
2364        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2365        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2366        }        }
2367      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2368      if (      if (
2369  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2370         c < 256 &&         c < 256 &&
2371  #endif  #endif
2372         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2373         )         )
2374        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2375      ecode++;      ecode++;
2376      break;      break;
2377    
# Line 2068  for (;;) Line 2379  for (;;)
2379      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2380        {        {
2381        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2382        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2383        }        }
2384      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2385      if (      if (
2386  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2387         c >= 256 ||         c > 255 ||
2388  #endif  #endif
2389         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2390         )         )
2391        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2392      ecode++;      ecode++;
2393      break;      break;
2394    
# Line 2085  for (;;) Line 2396  for (;;)
2396      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2397        {        {
2398        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2399        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2400        }        }
2401      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2402      if (      if (
2403  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2404         c < 256 &&         c < 256 &&
2405  #endif  #endif
2406         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2407         )         )
2408        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2409      ecode++;      ecode++;
2410      break;      break;
2411    
# Line 2102  for (;;) Line 2413  for (;;)
2413      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2414        {        {
2415        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2416        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2417        }        }
2418      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2419      if (      if (
2420  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2421         c >= 256 ||         c > 255 ||
2422  #endif  #endif
2423         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2424         )         )
2425        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2426      ecode++;      ecode++;
2427      break;      break;
2428    
# Line 2119  for (;;) Line 2430  for (;;)
2430      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2431        {        {
2432        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2433        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2434        }        }
2435      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2436      if (      if (
2437  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2438         c < 256 &&         c < 256 &&
2439  #endif  #endif
2440         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2441         )         )
2442        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2443      ecode++;      ecode++;
2444      break;      break;
2445    
# Line 2136  for (;;) Line 2447  for (;;)
2447      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2448        {        {
2449        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2450        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2451        }        }
2452      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2453      if (      if (
2454  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2455         c >= 256 ||         c > 255 ||
2456  #endif  #endif
2457         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2458         )         )
2459        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2460      ecode++;      ecode++;
2461      break;      break;
2462    
# Line 2153  for (;;) Line 2464  for (;;)
2464      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2465        {        {
2466        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2467        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2468        }        }
2469      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2470      switch(c)      switch(c)
2471        {        {
2472        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2473    
2474        case 0x000d:        case CHAR_CR:
2475        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr >= md->end_subject)
2476            {
2477            SCHECK_PARTIAL();
2478            }
2479          else if (RAWUCHARTEST(eptr) == CHAR_LF) eptr++;
2480        break;        break;
2481    
2482        case 0x000a:        case CHAR_LF:
2483        break;        break;
2484    
2485        case 0x000b:        case CHAR_VT:
2486        case 0x000c:        case CHAR_FF:
2487        case 0x0085:        case CHAR_NEL:
2488    #ifndef EBCDIC
2489        case 0x2028:        case 0x2028:
2490        case 0x2029:        case 0x2029:
2491        if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);  #endif  /* Not EBCDIC */
2492          if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
2493        break;        break;
2494        }        }
2495      ecode++;      ecode++;
# Line 2182  for (;;) Line 2499  for (;;)
2499      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2500        {        {
2501        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2502        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2503        }        }
2504      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2505      switch(c)      switch(c)
2506        {        {
2507          HSPACE_CASES: RRETURN(MATCH_NOMATCH);  /* Byte and multibyte cases */
2508        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);  
2509        }        }
2510      ecode++;      ecode++;
2511      break;      break;
# Line 2216  for (;;) Line 2514  for (;;)
2514      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2515        {        {
2516        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2517        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2518        }        }
2519      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2520      switch(c)      switch(c)
2521        {        {
2522        default: MRRETURN(MATCH_NOMATCH);        HSPACE_CASES: break;  /* Byte and multibyte cases */
2523        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;  
2524        }        }
2525      ecode++;      ecode++;
2526      break;      break;
# Line 2250  for (;;) Line 2529  for (;;)
2529      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2530        {        {
2531        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2532        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2533        }        }
2534      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2535      switch(c)      switch(c)
2536        {        {
2537          VSPACE_CASES: RRETURN(MATCH_NOMATCH);
2538        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);  
2539        }        }
2540      ecode++;      ecode++;
2541      break;      break;
# Line 2272  for (;;) Line 2544  for (;;)
2544      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2545        {        {
2546        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2547        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2548        }        }
2549      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2550      switch(c)      switch(c)
2551        {        {
2552        default: MRRETURN(MATCH_NOMATCH);        VSPACE_CASES: break;
2553        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;  
2554        }        }
2555      ecode++;      ecode++;
2556      break;      break;
# Line 2299  for (;;) Line 2564  for (;;)
2564      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2565        {        {
2566        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2567        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2568        }        }
2569      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2570        {        {
2571          const pcre_uint32 *cp;
2572        const ucd_record *prop = GET_UCD(c);        const ucd_record *prop = GET_UCD(c);
2573    
2574        switch(ecode[1])        switch(ecode[1])
2575          {          {
2576          case PT_ANY:          case PT_ANY:
2577          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2578          break;          break;
2579    
2580          case PT_LAMP:          case PT_LAMP:
2581          if ((prop->chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2582               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2583               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2584            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2585          break;          break;
2586    
2587          case PT_GC:          case PT_GC:
2588          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2589            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2590          break;          break;
2591    
2592          case PT_PC:          case PT_PC:
2593          if ((ecode[2] != prop->chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2594            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2595          break;          break;
2596    
2597          case PT_SC:          case PT_SC:
2598          if ((ecode[2] != prop->script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2599            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2600          break;          break;
2601    
2602          /* These are specials */          /* These are specials */
2603    
2604          case PT_ALNUM:          case PT_ALNUM:
2605          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2606               _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))               PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2607            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2608          break;          break;
2609    
2610          case PT_SPACE:    /* Perl space */          case PT_SPACE:    /* Perl space */
2611          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2612               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2613                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2614            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2615          break;          break;
2616    
2617          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2618          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2619               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2620               c == CHAR_FF || c == CHAR_CR)               c == CHAR_FF || c == CHAR_CR)
2621                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2622            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2623          break;          break;
2624    
2625          case PT_WORD:          case PT_WORD:
2626          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2627               _pcre_ucp_gentype[prop->chartype] == ucp_N ||               PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2628               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2629            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2630          break;          break;
2631    
2632          /* This should never occur */          case PT_CLIST:
2633            cp = PRIV(ucd_caseless_sets) + ecode[2];
2634          default:          for (;;)
2635              {
2636              if (c < *cp)
2637                { if (op == OP_PROP) { RRETURN(MATCH_NOMATCH); } else break; }
2638              if (c == *cp++)
2639                { if (op == OP_PROP) break; else { RRETURN(MATCH_NOMATCH); } }
2640              }
2641            break;
2642    
2643            case PT_UCNC:
2644            if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
2645                 c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
2646                 c >= 0xe000) == (op == OP_NOTPROP))
2647              RRETURN(MATCH_NOMATCH);
2648            break;
2649    
2650            /* This should never occur */
2651    
2652            default:
2653          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2654          }          }
2655    
# Line 2380  for (;;) Line 2664  for (;;)
2664      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2665        {        {
2666        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2667        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2668        }        }
2669      GETCHARINCTEST(c, eptr);      else
2670        {        {
2671        int category = UCD_CATEGORY(c);        int lgb, rgb;
2672        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        GETCHARINCTEST(c, eptr);
2673          lgb = UCD_GRAPHBREAK(c);
2674        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2675          {          {
2676          int len = 1;          int len = 1;
2677          if (!utf8) c = *eptr; else          if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2678            {          rgb = UCD_GRAPHBREAK(c);
2679            GETCHARLEN(c, eptr, len);          if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
2680            }          lgb = rgb;
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
2681          eptr += len;          eptr += len;
2682          }          }
2683        }        }
2684        CHECK_PARTIAL();
2685      ecode++;      ecode++;
2686      break;      break;
2687  #endif  #endif  /* SUPPORT_UCP */
2688    
2689    
2690      /* 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 2412  for (;;) Line 2696  for (;;)
2696      loops). */      loops). */
2697    
2698      case OP_REF:      case OP_REF:
2699      case OP_REFI:      case OP_REFI:
2700      caseless = op == OP_REFI;      caseless = op == OP_REFI;
2701      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2702      ecode += 3;      ecode += 1 + IMM2_SIZE;
2703    
2704      /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2705    
# Line 2455  for (;;) Line 2739  for (;;)
2739        case OP_CRMINRANGE:        case OP_CRMINRANGE:
2740        minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2741        min = GET2(ecode, 1);        min = GET2(ecode, 1);
2742        max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2743        if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2744        ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2745        break;        break;
2746    
2747        default:               /* No repeat follows */        default:               /* No repeat follows */
2748        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2749          {          {
2750            if (length == -2) eptr = md->end_subject;   /* Partial match */
2751          CHECK_PARTIAL();          CHECK_PARTIAL();
2752          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2753          }          }
2754        eptr += length;        eptr += length;
2755        continue;              /* With the main loop */        continue;              /* With the main loop */
2756        }        }
2757    
2758      /* Handle repeated back references. If the length of the reference is      /* Handle repeated back references. If the length of the reference is
2759      zero, just continue with the main loop. */      zero, just continue with the main loop. If the length is negative, it
2760        means the reference is unset in non-Java-compatible mode. If the minimum is
2761        zero, we can continue at the same level without recursion. For any other
2762        minimum, carrying on will result in NOMATCH. */
2763    
2764      if (length == 0) continue;      if (length == 0) continue;
2765        if (length < 0 && min == 0) continue;
2766    
2767      /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2768      the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
# Line 2481  for (;;) Line 2770  for (;;)
2770    
2771      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2772        {        {
2773        int slength;        int slength;
2774        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2775          {          {
2776            if (slength == -2) eptr = md->end_subject;   /* Partial match */
2777          CHECK_PARTIAL();          CHECK_PARTIAL();
2778          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2779          }          }
2780        eptr += slength;        eptr += slength;
2781        }        }
# Line 2501  for (;;) Line 2791  for (;;)
2791        {        {
2792        for (fi = min;; fi++)        for (fi = min;; fi++)
2793          {          {
2794          int slength;          int slength;
2795          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2796          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2797          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) RRETURN(MATCH_NOMATCH);
2798          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2799            {            {
2800              if (slength == -2) eptr = md->end_subject;   /* Partial match */
2801            CHECK_PARTIAL();            CHECK_PARTIAL();
2802            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2803            }            }
2804          eptr += slength;          eptr += slength;
2805          }          }
# Line 2522  for (;;) Line 2813  for (;;)
2813        pp = eptr;        pp = eptr;
2814        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2815          {          {
2816          int slength;          int slength;
2817          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2818            {            {
2819            CHECK_PARTIAL();            /* Can't use CHECK_PARTIAL because we don't want to update eptr in
2820              the soft partial matching case. */
2821    
2822              if (slength == -2 && md->partial != 0 &&
2823                  md->end_subject > md->start_used_ptr)
2824                {
2825                md->hitend = TRUE;
2826                if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2827                }
2828            break;            break;
2829            }            }
2830          eptr += slength;          eptr += slength;
2831          }          }
2832    
2833        while (eptr >= pp)        while (eptr >= pp)
2834          {          {
2835          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2836          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2837          eptr -= length;          eptr -= length;
2838          }          }
2839        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2840        }        }
2841      /* Control never gets here */      /* Control never gets here */
2842    
# Line 2554  for (;;) Line 2854  for (;;)
2854      case OP_NCLASS:      case OP_NCLASS:
2855      case OP_CLASS:      case OP_CLASS:
2856        {        {
2857          /* The data variable is saved across frames, so the byte map needs to
2858          be stored there. */
2859    #define BYTE_MAP ((pcre_uint8 *)data)
2860        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2861        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2862    
2863        switch (*ecode)        switch (*ecode)
2864          {          {
# Line 2576  for (;;) Line 2879  for (;;)
2879          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2880          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2881          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2882          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2883          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2884          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2885          break;          break;
2886    
2887          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2588  for (;;) Line 2891  for (;;)
2891    
2892        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2893    
2894  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2895        /* UTF-8 mode */        if (utf)
       if (utf8)  
2896          {          {
2897          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2898            {            {
2899            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2900              {              {
2901              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2902              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2903              }              }
2904            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2905            if (c > 255)            if (c > 255)
2906              {              {
2907              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2908              }              }
2909            else            else
2910              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
             }  
2911            }            }
2912          }          }
2913        else        else
2914  #endif  #endif
2915        /* Not UTF-8 mode */        /* Not UTF mode */
2916          {          {
2917          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2918            {            {
2919            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2920              {              {
2921              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2922              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2923              }              }
2924            c = *eptr++;            c = *eptr++;
2925            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2926              if (c > 255)
2927                {
2928                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2929                }
2930              else
2931    #endif
2932                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2933            }            }
2934          }          }
2935    
# Line 2636  for (;;) Line 2943  for (;;)
2943    
2944        if (minimize)        if (minimize)
2945          {          {
2946  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2947          /* UTF-8 mode */          if (utf)
         if (utf8)  
2948            {            {
2949            for (fi = min;; fi++)            for (fi = min;; fi++)
2950              {              {
2951              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2952              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2953              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2954              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2955                {                {
2956                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2957                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2958                }                }
2959              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2960              if (c > 255)              if (c > 255)
2961                {                {
2962                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2963                }                }
2964              else              else
2965                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
               }  
2966              }              }
2967            }            }
2968          else          else
2969  #endif  #endif
2970          /* Not UTF-8 mode */          /* Not UTF mode */
2971            {            {
2972            for (fi = min;; fi++)            for (fi = min;; fi++)
2973              {              {
2974              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2975              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2976              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2977              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2978                {                {
2979                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2980                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2981                }                }
2982              c = *eptr++;              c = *eptr++;
2983              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2984                if (c > 255)
2985                  {
2986                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2987                  }
2988                else
2989    #endif
2990                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2991              }              }
2992            }            }
2993          /* Control never gets here */          /* Control never gets here */
# Line 2688  for (;;) Line 2999  for (;;)
2999          {          {
3000          pp = eptr;          pp = eptr;
3001    
3002  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3003          /* UTF-8 mode */          if (utf)
         if (utf8)  
3004            {            {
3005            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3006              {              {
# Line 2706  for (;;) Line 3016  for (;;)
3016                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
3017                }                }
3018              else              else
3019                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
3020              eptr += len;              eptr += len;
3021              }              }
3022            for (;;)            for (;;)
# Line 2721  for (;;) Line 3029  for (;;)
3029            }            }
3030          else          else
3031  #endif  #endif
3032            /* Not UTF-8 mode */            /* Not UTF mode */
3033            {            {
3034            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3035              {              {
# Line 2731  for (;;) Line 3039  for (;;)
3039                break;                break;
3040                }                }
3041              c = *eptr;              c = *eptr;
3042              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
3043                if (c > 255)
3044                  {
3045                  if (op == OP_CLASS) break;
3046                  }
3047                else
3048    #endif
3049                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3050              eptr++;              eptr++;
3051              }              }
3052            while (eptr >= pp)            while (eptr >= pp)
# Line 2742  for (;;) Line 3057  for (;;)
3057              }              }
3058            }            }
3059    
3060          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3061          }          }
3062    #undef BYTE_MAP
3063        }        }
3064      /* Control never gets here */      /* Control never gets here */
3065    
# Line 2752  for (;;) Line 3068  for (;;)
3068      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
3069      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
3070    
3071  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3072      case OP_XCLASS:      case OP_XCLASS:
3073        {        {
3074        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2777  for (;;) Line 3093  for (;;)
3093          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3094          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3095          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3096          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
3097          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
3098          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
3099          break;          break;
3100    
3101          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2794  for (;;) Line 3110  for (;;)
3110          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3111            {            {
3112            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3113            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3114            }            }
3115          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
3116          if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3117          }          }
3118    
3119        /* 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 2814  for (;;) Line 3130  for (;;)
3130            {            {
3131            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
3132            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3133            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3134            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3135              {              {
3136              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3137              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3138              }              }
3139            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3140            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3141            }            }
3142          /* Control never gets here */          /* Control never gets here */
3143          }          }
# Line 2839  for (;;) Line 3155  for (;;)
3155              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3156              break;              break;
3157              }              }
3158    #ifdef SUPPORT_UTF
3159            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3160            if (!_pcre_xclass(c, data)) break;  #else
3161              c = *eptr;
3162    #endif
3163              if (!PRIV(xclass)(c, data, utf)) break;
3164            eptr += len;            eptr += len;
3165            }            }
3166          for(;;)          for(;;)
# Line 2848  for (;;) Line 3168  for (;;)
3168            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3169            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3170            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3171            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3172              if (utf) BACKCHAR(eptr);
3173    #endif
3174            }            }
3175          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3176          }          }
3177    
3178        /* Control never gets here */        /* Control never gets here */
# Line 2860  for (;;) Line 3182  for (;;)
3182      /* Match a single character, casefully */      /* Match a single character, casefully */
3183    
3184      case OP_CHAR:      case OP_CHAR:
3185  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3186      if (utf8)      if (utf)
3187        {        {
3188        length = 1;        length = 1;
3189        ecode++;        ecode++;
# Line 2869  for (;;) Line 3191  for (;;)
3191        if (length > md->end_subject - eptr)        if (length > md->end_subject - eptr)
3192          {          {
3193          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
3194          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3195          }          }
3196        while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);        while (length-- > 0) if (*ecode++ != RAWUCHARINC(eptr)) RRETURN(MATCH_NOMATCH);
3197        }        }
3198      else      else
3199  #endif  #endif
3200        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3201        {        {
3202        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3203          {          {
3204          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
3205          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3206          }          }
3207        if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
3208        ecode += 2;        ecode += 2;
3209        }        }
3210      break;      break;
3211    
3212      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3213        subject, give up immediately. */
3214    
3215      case OP_CHARI:      case OP_CHARI:
3216  #ifdef SUPPORT_UTF8      if (eptr >= md->end_subject)
3217      if (utf8)        {
3218          SCHECK_PARTIAL();
3219          RRETURN(MATCH_NOMATCH);
3220          }
3221    
3222    #ifdef SUPPORT_UTF
3223        if (utf)
3224        {        {
3225        length = 1;        length = 1;
3226        ecode++;        ecode++;
3227        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3228    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         MRRETURN(MATCH_NOMATCH);  
         }  
   
3229        /* 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
3230        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3231          fast lookup table. We know that there is at least one byte left in the
3232          subject. */
3233    
3234        if (fc < 128)        if (fc < 128)
3235          {          {
3236          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          pcre_uint32 cc = RAWUCHAR(eptr);
3237            if (md->lcc[fc] != TABLE_GET(cc, md->lcc, cc)) RRETURN(MATCH_NOMATCH);
3238            ecode++;
3239            eptr++;
3240          }          }
3241    
3242        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3243          use the value of "length" to check for sufficient bytes left, because the
3244          other case of the character may have more or fewer bytes.  */
3245    
3246        else        else
3247          {          {
3248          unsigned int dc;          pcre_uint32 dc;
3249          GETCHARINC(dc, eptr);          GETCHARINC(dc, eptr);
3250          ecode += length;          ecode += length;
3251    
# Line 2928  for (;;) Line 3257  for (;;)
3257  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3258            if (dc != UCD_OTHERCASE(fc))            if (dc != UCD_OTHERCASE(fc))
3259  #endif  #endif
3260              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3261            }            }
3262          }          }
3263        }        }
3264      else      else
3265  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3266    
3267      /* Non-UTF-8 mode */      /* Not UTF mode */
3268        {        {
3269        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3270          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3271          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         MRRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);  
3272        ecode += 2;        ecode += 2;
3273        }        }
3274      break;      break;
# Line 2952  for (;;) Line 3278  for (;;)
3278      case OP_EXACT:      case OP_EXACT:
3279      case OP_EXACTI:      case OP_EXACTI:
3280      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3281      ecode += 3;      ecode += 1 + IMM2_SIZE;
3282      goto REPEATCHAR;      goto REPEATCHAR;
3283    
3284      case OP_POSUPTO:      case OP_POSUPTO:
# Line 2967  for (;;) Line 3293  for (;;)
3293      min = 0;      min = 0;
3294      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3295      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3296      ecode += 3;      ecode += 1 + IMM2_SIZE;
3297      goto REPEATCHAR;      goto REPEATCHAR;
3298    
3299      case OP_POSSTAR:      case OP_POSSTAR:
# Line 3015  for (;;) Line 3341  for (;;)
3341      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3342    
3343      REPEATCHAR:      REPEATCHAR:
3344  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3345      if (utf8)      if (utf)
3346        {        {
3347        length = 1;        length = 1;
3348        charptr = ecode;        charptr = ecode;
# Line 3029  for (;;) Line 3355  for (;;)
3355        if (length > 1)        if (length > 1)
3356          {          {
3357  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3358          unsigned int othercase;          pcre_uint32 othercase;
3359          if (op >= OP_STARI &&     /* Caseless */          if (op >= OP_STARI &&     /* Caseless */
3360              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3361            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3362          else oclength = 0;          else oclength = 0;
3363  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3364    
3365          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3366            {            {
3367            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3368              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3369  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3370            else if (oclength > 0 &&            else if (oclength > 0 &&
3371                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3372                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3373  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3374            else            else
3375              {              {
3376              CHECK_PARTIAL();              CHECK_PARTIAL();
3377              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3378              }              }
3379            }            }
3380    
# Line 3060  for (;;) Line 3386  for (;;)
3386              {              {
3387              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3388              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3389              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3390              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3391                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3392  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3393              else if (oclength > 0 &&              else if (oclength > 0 &&
3394                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3395                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3396  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3397              else              else
3398                {                {
3399                CHECK_PARTIAL();                CHECK_PARTIAL();
3400                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3401                }                }
3402              }              }
3403            /* Control never gets here */            /* Control never gets here */
# Line 3083  for (;;) Line 3409  for (;;)
3409            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3410              {              {
3411              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3412                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3413  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3414              else if (oclength > 0 &&              else if (oclength > 0 &&
3415                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3416                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3417  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3418              else              else
3419                {                {
# Line 3102  for (;;) Line 3428  for (;;)
3428              {              {
3429              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3430              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3431              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3432  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3433              eptr--;              eptr--;
3434              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3119  for (;;) Line 3445  for (;;)
3445        value of fc will always be < 128. */        value of fc will always be < 128. */
3446        }        }
3447      else      else
3448  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3449          /* When not in UTF-8 mode, load a single-byte character. */
3450      /* When not in UTF-8 mode, load a single-byte character. */        fc = *ecode++;
   
     fc = *ecode++;  
3451    
3452      /* The value of fc at this point is always less than 256, though we may or      /* The value of fc at this point is always one character, though we may
3453      may not be in UTF-8 mode. The code is duplicated for the caseless and      or may not be in UTF mode. The code is duplicated for the caseless and
3454      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3455      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3456      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 3135  for (;;) Line 3459  for (;;)
3459      maximizing, find the maximum number of characters and work backwards. */      maximizing, find the maximum number of characters and work backwards. */
3460    
3461      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3462        max, eptr));        max, (char *)eptr));
3463    
3464      if (op >= OP_STARI)  /* Caseless */      if (op >= OP_STARI)  /* Caseless */
3465        {        {
3466        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3467          /* fc must be < 128 if UTF is enabled. */
3468          foc = md->fcc[fc];
3469    #else
3470    #ifdef SUPPORT_UTF
3471    #ifdef SUPPORT_UCP
3472          if (utf && fc > 127)
3473            foc = UCD_OTHERCASE(fc);
3474    #else
3475          if (utf && fc > 127)
3476            foc = fc;
3477    #endif /* SUPPORT_UCP */
3478          else
3479    #endif /* SUPPORT_UTF */
3480            foc = TABLE_GET(fc, md->fcc, fc);
3481    #endif /* COMPILE_PCRE8 */
3482    
3483        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3484          {          {
3485            pcre_uint32 cc;                 /* Faster than pcre_uchar */
3486          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3487            {            {
3488            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3489            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3490            }            }
3491          if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          cc = RAWUCHARTEST(eptr);
3492            if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH);
3493            eptr++;
3494          }          }
3495        if (min == max) continue;        if (min == max) continue;
3496        if (minimize)        if (minimize)
3497          {          {
3498          for (fi = min;; fi++)          for (fi = min;; fi++)
3499            {            {
3500              pcre_uint32 cc;               /* Faster than pcre_uchar */
3501            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3502            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3503            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3504            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3505              {              {
3506              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3507              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3508              }              }
3509            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            cc = RAWUCHARTEST(eptr);
3510              if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH);
3511              eptr++;
3512            }            }
3513          /* Control never gets here */          /* Control never gets here */
3514          }          }
# Line 3171  for (;;) Line 3517  for (;;)
3517          pp = eptr;          pp = eptr;
3518          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3519            {            {
3520              pcre_uint32 cc;               /* Faster than pcre_uchar */
3521            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3522              {              {
3523              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3524              break;              break;
3525              }              }
3526            if (fc != md->lcc[*eptr]) break;            cc = RAWUCHARTEST(eptr);
3527              if (fc != cc && foc != cc) break;
3528            eptr++;            eptr++;
3529            }            }
3530    
# Line 3188  for (;;) Line 3536  for (;;)
3536            eptr--;            eptr--;
3537            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3538            }            }
3539          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3540          }          }
3541        /* Control never gets here */        /* Control never gets here */
3542        }        }
# Line 3202  for (;;) Line 3550  for (;;)
3550          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3551            {            {
3552            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3553            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3554            }            }
3555          if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);          if (fc != RAWUCHARINCTEST(eptr)) RRETURN(MATCH_NOMATCH);
3556          }          }
3557    
3558        if (min == max) continue;        if (min == max) continue;
# Line 3215  for (;;) Line 3563  for (;;)
3563            {            {
3564            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3565            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3566            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3567            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3568              {              {
3569              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3570              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3571              }              }
3572            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);            if (fc != RAWUCHARINCTEST(eptr)) RRETURN(MATCH_NOMATCH);
3573            }            }
3574          /* Control never gets here */          /* Control never gets here */
3575          }          }
# Line 3235  for (;;) Line 3583  for (;;)
3583              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3584              break;              break;
3585              }              }
3586            if (fc != *eptr) break;            if (fc != RAWUCHARTEST(eptr)) break;
3587            eptr++;            eptr++;
3588            }            }
3589          if (possessive) continue;          if (possessive) continue;
# Line 3246  for (;;) Line 3594  for (;;)
3594            eptr--;            eptr--;
3595            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3596            }            }
3597          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3598          }          }
3599        }        }
3600      /* Control never gets here */      /* Control never gets here */
# Line 3255  for (;;) Line 3603  for (;;)
3603      checking can be multibyte. */      checking can be multibyte. */
3604    
3605      case OP_NOT:      case OP_NOT:
3606      case OP_NOTI:      case OP_NOTI:
3607      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3608        {        {
3609        SCHECK_PARTIAL();        SCHECK_PARTIAL();
3610        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
3611        }        }
3612      ecode++;  #ifdef SUPPORT_UTF
3613      GETCHARINCTEST(c, eptr);      if (utf)
     if (op == OP_NOTI)         /* The caseless case */  
3614        {        {
3615  #ifdef SUPPORT_UTF8        register pcre_uint32 ch, och;
3616        if (c < 256)  
3617  #endif        ecode++;
3618        c = md->lcc[c];        GETCHARINC(ch, ecode);
3619        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        GETCHARINC(c, eptr);
3620    
3621          if (op == OP_NOT)
3622            {
3623            if (ch == c) RRETURN(MATCH_NOMATCH);
3624            }
3625          else
3626            {
3627    #ifdef SUPPORT_UCP
3628            if (ch > 127)
3629              och = UCD_OTHERCASE(ch);
3630    #else
3631            if (ch > 127)
3632              och = ch;
3633    #endif /* SUPPORT_UCP */
3634            else
3635              och = TABLE_GET(ch, md->fcc, ch);
3636            if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3637            }
3638        }        }
3639      else    /* Caseful */      else
3640    #endif
3641        {        {
3642        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        register pcre_uint32 ch = ecode[1];
3643          c = *eptr++;
3644          if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c))
3645            RRETURN(MATCH_NOMATCH);
3646          ecode += 2;
3647        }        }
3648      break;      break;
3649    
# Line 3287  for (;;) Line 3657  for (;;)
3657      case OP_NOTEXACT:      case OP_NOTEXACT:
3658      case OP_NOTEXACTI:      case OP_NOTEXACTI:
3659      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3660      ecode += 3;      ecode += 1 + IMM2_SIZE;
3661      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3662    
3663      case OP_NOTUPTO:      case OP_NOTUPTO:
# Line 3297  for (;;) Line 3667  for (;;)
3667      min = 0;      min = 0;
3668      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3669      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3670      ecode += 3;      ecode += 1 + IMM2_SIZE;
3671      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3672    
3673      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
# Line 3329  for (;;) Line 3699  for (;;)
3699      possessive = TRUE;      possessive = TRUE;
3700      min = 0;      min = 0;
3701      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3702      ecode += 3;      ecode += 1 + IMM2_SIZE;
3703      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3704    
3705      case OP_NOTSTAR:      case OP_NOTSTAR:
# Line 3353  for (;;) Line 3723  for (;;)
3723      /* Common code for all repeated single-byte matches. */      /* Common code for all repeated single-byte matches. */
3724    
3725      REPEATNOTCHAR:      REPEATNOTCHAR:
3726      fc = *ecode++;      GETCHARINCTEST(fc, ecode);
3727    
3728      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
3729      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 3364  for (;;) Line 3734  for (;;)
3734      characters and work backwards. */      characters and work backwards. */
3735    
3736      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,
3737        max, eptr));        max, (char *)eptr));
3738    
3739      if (op >= OP_NOTSTARI)     /* Caseless */      if (op >= OP_NOTSTARI)     /* Caseless */
3740        {        {
3741        fc = md->lcc[fc];  #ifdef SUPPORT_UTF
3742    #ifdef SUPPORT_UCP
3743          if (utf && fc > 127)
3744            foc = UCD_OTHERCASE(fc);
3745    #else
3746          if (utf && fc > 127)
3747            foc = fc;
3748    #endif /* SUPPORT_UCP */
3749          else
3750    #endif /* SUPPORT_UTF */
3751            foc = TABLE_GET(fc, md->fcc, fc);
3752    
3753  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3754        /* UTF-8 mode */        if (utf)
       if (utf8)  
3755          {          {
3756          register unsigned int d;          register pcre_uint32 d;
3757          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3758            {            {
3759            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3760              {              {
3761              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3762              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3763              }              }
3764            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3765            if (d < 256) d = md->lcc[d];            if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) MRRETURN(MATCH_NOMATCH);  
3766            }            }
3767          }          }
3768        else        else
3769  #endif  #endif
3770          /* Not UTF mode */
       /* Not UTF-8 mode */  
3771          {          {
3772          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3773            {            {
3774            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3775              {              {
3776              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3777              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3778              }              }
3779            if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3780              eptr++;
3781            }            }
3782          }          }
3783    
# Line 3407  for (;;) Line 3785  for (;;)
3785    
3786        if (minimize)        if (minimize)
3787          {          {
3788  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3789          /* UTF-8 mode */          if (utf)
         if (utf8)  
3790            {            {
3791            register unsigned int d;            register pcre_uint32 d;
3792            for (fi = min;; fi++)