/[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 1260 by ph10, Wed Feb 27 15:41:22 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 82  negative to avoid the external error cod Line 95  negative to avoid the external error cod
95  #define MATCH_SKIP_ARG     (-993)  #define MATCH_SKIP_ARG     (-993)
96  #define MATCH_THEN         (-992)  #define MATCH_THEN         (-992)
97    
 /* This is a convenience macro for code that occurs many times. */  
   
 #define MRRETURN(ra) \  
   { \  
   md->mark = markptr; \  
   RRETURN(ra); \  
   }  
   
98  /* 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.
99  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,
100  because the offset vector is always a multiple of 3 long. */  because the offset vector is always a multiple of 3 long. */
# Line 101  because the offset vector is always a mu Line 106  because the offset vector is always a mu
106  static const char rep_min[] = { 0, 0, 1, 1, 0, 0 };  static const char rep_min[] = { 0, 0, 1, 1, 0, 0 };
107  static const char rep_max[] = { 0, 0, 0, 0, 1, 1 };  static const char rep_max[] = { 0, 0, 0, 0, 1, 1 };
108    
   
   
109  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
110  /*************************************************  /*************************************************
111  *        Debugging function to print chars       *  *        Debugging function to print chars       *
# Line 121  Returns:     nothing Line 124  Returns:     nothing
124  */  */
125    
126  static void  static void
127  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)
128  {  {
129  unsigned int c;  pcre_uint32 c;
130    BOOL utf = md->utf;
131  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;
132  while (length-- > 0)  while (length-- > 0)
133    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);
134  }  }
135  #endif  #endif
136    
# Line 138  while (length-- > 0) Line 142  while (length-- > 0)
142    
143  /* 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
144  negative, so the match always fails. However, in JavaScript compatibility mode,  negative, so the match always fails. However, in JavaScript compatibility mode,
145  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
146  subject bytes matched may be different to the number of reference bytes.  subject bytes matched may be different to the number of reference bytes.
147    
148  Arguments:  Arguments:
# Line 148  Arguments: Line 152  Arguments:
152    md          points to match data block    md          points to match data block
153    caseless    TRUE if caseless    caseless    TRUE if caseless
154    
155  Returns:      < 0 if not matched, otherwise the number of subject bytes matched  Returns:      >= 0 the number of subject bytes matched
156                  -1 no match
157                  -2 partial match; always given if at end subject
158  */  */
159    
160  static int  static int
161  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,
162    BOOL caseless)    BOOL caseless)
163  {  {
164  USPTR eptr_start = eptr;  PCRE_PUCHAR eptr_start = eptr;
165  register USPTR p = md->start_subject + md->offset_vector[offset];  register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
166    #ifdef SUPPORT_UTF
167    BOOL utf = md->utf;
168    #endif
169    
170  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
171  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 171  pchars(p, length, FALSE, md); Line 180  pchars(p, length, FALSE, md);
180  printf("\n");  printf("\n");
181  #endif  #endif
182    
183  /* Always fail if reference not set (and not JavaScript compatible). */  /* Always fail if reference not set (and not JavaScript compatible - in that
184    case the length is passed as zero). */
185    
186  if (length < 0) return -1;  if (length < 0) return -1;
187    
# Line 181  ASCII characters. */ Line 191  ASCII characters. */
191    
192  if (caseless)  if (caseless)
193    {    {
194  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
195  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
196    if (md->utf8)    if (utf)
197      {      {
198      /* 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
199      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
200      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.
201      (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
202      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
203      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
204      reference, not along the subject (earlier code did this wrong). */      length along the reference, not along the subject (earlier code did this
205        wrong). */
206      USPTR endptr = p + length;  
207        PCRE_PUCHAR endptr = p + length;
208      while (p < endptr)      while (p < endptr)
209        {        {
210        int c, d;        pcre_uint32 c, d;
211        if (eptr >= md->end_subject) return -1;        const ucd_record *ur;
212          if (eptr >= md->end_subject) return -2;   /* Partial match */
213        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
214        GETCHARINC(d, p);        GETCHARINC(d, p);
215        if (c != d && c != UCD_OTHERCASE(d)) return -1;        ur = GET_UCD(d);
216          if (c != d && c != d + ur->other_case)
217            {
218            const pcre_uint32 *pp = PRIV(ucd_caseless_sets) + ur->caseset;
219            for (;;)
220              {
221              if (c < *pp) return -1;
222              if (c == *pp++) break;
223              }
224            }
225        }        }
226      }      }
227    else    else
# Line 210  if (caseless) Line 231  if (caseless)
231    /* 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
232    is no UCP support. */    is no UCP support. */
233      {      {
     if (eptr + length > md->end_subject) return -1;  
234      while (length-- > 0)      while (length-- > 0)
235        { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }        {
236      }        pcre_uint32 cc, cp;
237          if (eptr >= md->end_subject) return -2;   /* Partial match */
238          cc = RAWUCHARTEST(eptr);
239          cp = RAWUCHARTEST(p);
240          if (TABLE_GET(cp, md->lcc, cp) != TABLE_GET(cc, md->lcc, cc)) return -1;
241          p++;
242          eptr++;
243          }
244        }
245    }    }
246    
247  /* 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
248  are in UTF-8 mode. */  are in UTF-8 mode. */
249    
250  else  else
251    {    {
252    if (eptr + length > md->end_subject) return -1;    while (length-- > 0)
253    while (length-- > 0) if (*p++ != *eptr++) return -1;      {
254        if (eptr >= md->end_subject) return -2;   /* Partial match */
255        if (RAWUCHARINCTEST(p) != RAWUCHARINCTEST(eptr)) return -1;
256        }
257    }    }
258    
259  return eptr - eptr_start;  return (int)(eptr - eptr_start);
260  }  }
261    
262    
# Line 277  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 308  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
308         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
309         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
310         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
311         RM61,  RM62, RM63, RM64, RM65, RM66 };         RM61,  RM62, RM63, RM64, RM65, RM66, RM67, RM68 };
312    
313  /* 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
314  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 321  actually used in this definition. */
321  #define RMATCH(ra,rb,rc,rd,re,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
322    { \    { \
323    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
324    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \
325    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
326    }    }
327  #define RRETURN(ra) \  #define RRETURN(ra) \
328    { \    { \
329    printf("match() returned %d from line %d ", ra, __LINE__); \    printf("match() returned %d from line %d\n", ra, __LINE__); \
330    return ra; \    return ra; \
331    }    }
332  #else  #else
333  #define RMATCH(ra,rb,rc,rd,re,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
334    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1)
335  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
336  #endif  #endif
337    
# Line 315  argument of match(), which never changes Line 346  argument of match(), which never changes
346    
347  #define RMATCH(ra,rb,rc,rd,re,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
348    {\    {\
349    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = frame->Xnextframe;\
350    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    if (newframe == NULL)\
351    frame->Xwhere = rw; \      {\
352        newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\
353        if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
354        newframe->Xnextframe = NULL;\
355        frame->Xnextframe = newframe;\
356        }\
357      frame->Xwhere = rw;\
358    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
359    newframe->Xecode = rb;\    newframe->Xecode = rb;\
360    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
   newframe->Xmarkptr = markptr;\  
361    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
362    newframe->Xeptrb = re;\    newframe->Xeptrb = re;\
363    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
# Line 337  argument of match(), which never changes Line 373  argument of match(), which never changes
373    {\    {\
374    heapframe *oldframe = frame;\    heapframe *oldframe = frame;\
375    frame = oldframe->Xprevframe;\    frame = oldframe->Xprevframe;\
   (pcre_stack_free)(oldframe);\  
376    if (frame != NULL)\    if (frame != NULL)\
377      {\      {\
378      rrc = ra;\      rrc = ra;\
# Line 351  argument of match(), which never changes Line 386  argument of match(), which never changes
386    
387  typedef struct heapframe {  typedef struct heapframe {
388    struct heapframe *Xprevframe;    struct heapframe *Xprevframe;
389      struct heapframe *Xnextframe;
390    
391    /* Function arguments that may change */    /* Function arguments that may change */
392    
393    USPTR Xeptr;    PCRE_PUCHAR Xeptr;
394    const uschar *Xecode;    const pcre_uchar *Xecode;
395    USPTR Xmstart;    PCRE_PUCHAR Xmstart;
   USPTR Xmarkptr;  
396    int Xoffset_top;    int Xoffset_top;
397    eptrblock *Xeptrb;    eptrblock *Xeptrb;
398    unsigned int Xrdepth;    unsigned int Xrdepth;
399    
400    /* Function local variables */    /* Function local variables */
401    
402    USPTR Xcallpat;    PCRE_PUCHAR Xcallpat;
403  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
404    USPTR Xcharptr;    PCRE_PUCHAR Xcharptr;
405  #endif  #endif
406    USPTR Xdata;    PCRE_PUCHAR Xdata;
407    USPTR Xnext;    PCRE_PUCHAR Xnext;
408    USPTR Xpp;    PCRE_PUCHAR Xpp;
409    USPTR Xprev;    PCRE_PUCHAR Xprev;
410    USPTR Xsaved_eptr;    PCRE_PUCHAR Xsaved_eptr;
411    
412    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
413    
# Line 382  typedef struct heapframe { Line 417  typedef struct heapframe {
417    
418  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
419    int Xprop_type;    int Xprop_type;
420    int Xprop_value;    unsigned int Xprop_value;
421    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
422    int Xoclength;    int Xoclength;
423    uschar Xocchars[8];    pcre_uchar Xocchars[6];
424  #endif  #endif
425    
426    int Xcodelink;    int Xcodelink;
# Line 398  typedef struct heapframe { Line 430  typedef struct heapframe {
430    int Xlength;    int Xlength;
431    int Xmax;    int Xmax;
432    int Xmin;    int Xmin;
433    int Xnumber;    unsigned int Xnumber;
434    int Xoffset;    int Xoffset;
435    int Xop;    unsigned int Xop;
436    int Xsave_capture_last;    pcre_int32 Xsave_capture_last;
437    int Xsave_offset1, Xsave_offset2, Xsave_offset3;    int Xsave_offset1, Xsave_offset2, Xsave_offset3;
438    int Xstacksave[REC_STACK_SAVE_MAX];    int Xstacksave[REC_STACK_SAVE_MAX];
439    
# Line 430  returns a negative (error) response, the Line 462  returns a negative (error) response, the
462  same response. */  same response. */
463    
464  /* 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
465  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
466  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.
467  something has been matched). For hard partial matching, we then return  something has been matched). For hard partial matching, we then return
468  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 473  the subject. */
473        eptr > md->start_used_ptr) \        eptr > md->start_used_ptr) \
474      { \      { \
475      md->hitend = TRUE; \      md->hitend = TRUE; \
476      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
477      }      }
478    
479  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
480    if (md->partial != 0 && eptr > md->start_used_ptr) \    if (md->partial != 0 && eptr > md->start_used_ptr) \
481      { \      { \
482      md->hitend = TRUE; \      md->hitend = TRUE; \
483      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
484      }      }
485    
486    
487  /* Performance note: It might be tempting to extract commonly used fields from  /* Performance note: It might be tempting to extract commonly used fields from
488  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
489  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
490  made performance worse.  made performance worse.
491    
# Line 462  Arguments: Line 494  Arguments:
494     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
495     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
496                   by encountering \K)                   by encountering \K)
    markptr     pointer to the most recent MARK name, or NULL  
497     offset_top  current top pointer     offset_top  current top pointer
498     md          pointer to "static" info for the match     md          pointer to "static" info for the match
499     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 508  Returns:       MATCH_MATCH if matched
508  */  */
509    
510  static int  static int
511  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode,
512    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
513    unsigned int rdepth)    unsigned int rdepth)
514  {  {
515  /* 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 518  so they can be ordinary variables in all
518    
519  register int  rrc;         /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
520  register int  i;           /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
521  register unsigned int c;   /* Character values not kept over RMATCH() calls */  register pcre_uint32 c;    /* Character values not kept over RMATCH() calls */
522  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf;         /* Local copy of UTF flag for speed */
523    
524  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
525  BOOL caseless;  BOOL caseless;
526  int condcode;  int condcode;
527    
528  /* 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
529  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
530  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
531  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  whenever RMATCH() does a "recursion". See the macro definitions above. Putting
532    the top-level on the stack rather than malloc-ing them all gives a performance
533    boost in many cases where there is not much "recursion". */
534    
535  #ifdef NO_RECURSE  #ifdef NO_RECURSE
536  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 */  
537    
538  /* Copy in the original argument variables */  /* Copy in the original argument variables */
539    
540  frame->Xeptr = eptr;  frame->Xeptr = eptr;
541  frame->Xecode = ecode;  frame->Xecode = ecode;
542  frame->Xmstart = mstart;  frame->Xmstart = mstart;
 frame->Xmarkptr = markptr;  
543  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
544  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
545  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
# Line 523  HEAP_RECURSE: Line 553  HEAP_RECURSE:
553  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
554  #define ecode              frame->Xecode  #define ecode              frame->Xecode
555  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
 #define markptr            frame->Xmarkptr  
556  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
557  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
558  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
559    
560  /* Ditto for the local variables */  /* Ditto for the local variables */
561    
562  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
563  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
564  #endif  #endif
565  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 551  HEAP_RECURSE: Line 580  HEAP_RECURSE:
580  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
581  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
582  #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  
583  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
584  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
585  #endif  #endif
# Line 591  declarations can be cut out in a block. Line 617  declarations can be cut out in a block.
617  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
618  to RMATCH(). */  to RMATCH(). */
619    
620  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
621  const uschar *charptr;  const pcre_uchar *charptr;
622  #endif  #endif
623  const uschar *callpat;  const pcre_uchar *callpat;
624  const uschar *data;  const pcre_uchar *data;
625  const uschar *next;  const pcre_uchar *next;
626  USPTR         pp;  PCRE_PUCHAR       pp;
627  const uschar *prev;  const pcre_uchar *prev;
628  USPTR         saved_eptr;  PCRE_PUCHAR       saved_eptr;
629    
630  recursion_info new_recursive;  recursion_info new_recursive;
631    
632  BOOL cur_is_word;  BOOL cur_is_word;
633  BOOL condition;  BOOL condition;
634  BOOL prev_is_word;  BOOL prev_is_word;
635    
636  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
637  int prop_type;  int prop_type;
638  int prop_value;  unsigned int prop_value;
639  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
640  int oclength;  int oclength;
641  uschar occhars[8];  pcre_uchar occhars[6];
642  #endif  #endif
643    
644  int codelink;  int codelink;
# Line 623  int ctype; Line 646  int ctype;
646  int length;  int length;
647  int max;  int max;
648  int min;  int min;
649  int number;  unsigned int number;
650  int offset;  int offset;
651  int op;  unsigned int op;
652  int save_capture_last;  pcre_int32 save_capture_last;
653  int save_offset1, save_offset2, save_offset3;  int save_offset1, save_offset2, save_offset3;
654  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
655    
656  eptrblock newptrb;  eptrblock newptrb;
657    
658    /* There is a special fudge for calling match() in a way that causes it to
659    measure the size of its basic stack frame when the stack is being used for
660    recursion. The second argument (ecode) being NULL triggers this behaviour. It
661    cannot normally ever be NULL. The return is the negated value of the frame
662    size. */
663    
664    if (ecode == NULL)
665      {
666      if (rdepth == 0)
667        return match((PCRE_PUCHAR)&rdepth, NULL, NULL, 0, NULL, NULL, 1);
668      else
669        {
670        int len = (char *)&rdepth - (char *)eptr;
671        return (len > 0)? -len : len;
672        }
673      }
674  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
675    
676  /* 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
677  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
678  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
679  the alternative names that are used. */  the alternative names that are used. */
680    
681  #define allow_zero    cur_is_word  #define allow_zero    cur_is_word
# Line 643  the alternative names that are used. */ Line 683  the alternative names that are used. */
683  #define code_offset   codelink  #define code_offset   codelink
684  #define condassert    condition  #define condassert    condition
685  #define matched_once  prev_is_word  #define matched_once  prev_is_word
686    #define foc           number
687    #define save_mark     data
688    
689  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
690  variables. */  variables. */
# Line 668  defined). However, RMATCH isn't like a f Line 710  defined). However, RMATCH isn't like a f
710  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,
711  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
712    
713  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
714  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
715  #else  #else
716  utf8 = FALSE;  utf = FALSE;
717  #endif  #endif
718    
719  /* 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 723  if (md->match_call_count++ >= md->match_
723  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
724    
725  /* 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
726  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
727  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
728  up space on the stack. See also MATCH_CONDASSERT below.  up space on the stack. See also MATCH_CONDASSERT below.
729    
730  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 748  for (;;)
748    {    {
749    minimize = possessive = FALSE;    minimize = possessive = FALSE;
750    op = *ecode;    op = *ecode;
751    
752    switch(op)    switch(op)
753      {      {
754      case OP_MARK:      case OP_MARK:
755      markptr = ecode + 2;      md->nomatch_mark = ecode + 2;
756      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->mark = NULL;    /* In case previously set by assertion */
757        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
758        eptrb, RM55);        eptrb, RM55);
759        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
760             md->mark == NULL) md->mark = ecode + 2;
761    
762      /* 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
763      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 766  for (;;)
766      position and return MATCH_SKIP. Otherwise, pass back the return code      position and return MATCH_SKIP. Otherwise, pass back the return code
767      unaltered. */      unaltered. */
768    
769      if (rrc == MATCH_SKIP_ARG &&      else if (rrc == MATCH_SKIP_ARG &&
770          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)          STRCMP_UC_UC_TEST(ecode + 2, md->start_match_ptr) == 0)
771        {        {
772        md->start_match_ptr = eptr;        md->start_match_ptr = eptr;
773        RRETURN(MATCH_SKIP);        RRETURN(MATCH_SKIP);
774        }        }
   
     if (md->mark == NULL) md->mark = markptr;  
775      RRETURN(rrc);      RRETURN(rrc);
776    
777      case OP_FAIL:      case OP_FAIL:
778      MRRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
779    
780      /* COMMIT overrides PRUNE, SKIP, and THEN */      /* COMMIT overrides PRUNE, SKIP, and THEN */
781    
782      case OP_COMMIT:      case OP_COMMIT:
783      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
784        eptrb, RM52);        eptrb, RM52);
785      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
786          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
787          rrc != MATCH_THEN)          rrc != MATCH_THEN)
788        RRETURN(rrc);        RRETURN(rrc);
789      MRRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
790    
791      /* PRUNE overrides THEN */      /* PRUNE overrides THEN */
792    
793      case OP_PRUNE:      case OP_PRUNE:
794      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
795        eptrb, RM51);        eptrb, RM51);
796      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
797      MRRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
798    
799      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
800      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->nomatch_mark = ecode + 2;
801        md->mark = NULL;    /* In case previously set by assertion */
802        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
803        eptrb, RM56);        eptrb, RM56);
804        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
805             md->mark == NULL) md->mark = ecode + 2;
806      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
     md->mark = ecode + 2;  
807      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
808    
809      /* SKIP overrides PRUNE and THEN */      /* SKIP overrides PRUNE and THEN */
810    
811      case OP_SKIP:      case OP_SKIP:
812      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
813        eptrb, RM53);        eptrb, RM53);
814      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
815        RRETURN(rrc);        RRETURN(rrc);
816      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
817      MRRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
818    
819        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
820        nomatch_mark. There is a flag that disables this opcode when re-matching a
821        pattern that ended with a SKIP for which there was not a matching MARK. */
822    
823      case OP_SKIP_ARG:      case OP_SKIP_ARG:
824      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      if (md->ignore_skip_arg)
825          {
826          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
827          break;
828          }
829        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
830        eptrb, RM57);        eptrb, RM57);
831      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
832        RRETURN(rrc);        RRETURN(rrc);
833    
834      /* 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
835      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
836      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
837      as PRUNE. */      with the md->ignore_skip_arg flag set. */
838    
839      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
840      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
841    
842      /* 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
843      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
844      to skip back past alternatives that precede the THEN within the current      match pointer to do this. */
     branch. */  
845    
846      case OP_THEN:      case OP_THEN:
847      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
848        eptrb, RM54);        eptrb, RM54);
849      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
850      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
851      MRRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
852    
853      case OP_THEN_ARG:      case OP_THEN_ARG:
854      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      md->nomatch_mark = ecode + 2;
855        offset_top, md, eptrb, RM58);      md->mark = NULL;    /* In case previously set by assertion */
856        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
857          md, eptrb, RM58);
858        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
859             md->mark == NULL) md->mark = ecode + 2;
860      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
861      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
     md->mark = ecode + LINK_SIZE + 2;  
862      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
863    
864        /* Handle an atomic group that does not contain any capturing parentheses.
865        This can be handled like an assertion. Prior to 8.13, all atomic groups
866        were handled this way. In 8.13, the code was changed as below for ONCE, so
867        that backups pass through the group and thereby reset captured values.
868        However, this uses a lot more stack, so in 8.20, atomic groups that do not
869        contain any captures generate OP_ONCE_NC, which can be handled in the old,
870        less stack intensive way.
871    
872        Check the alternative branches in turn - the matching won't pass the KET
873        for this kind of subpattern. If any one branch matches, we carry on as at
874        the end of a normal bracket, leaving the subject pointer, but resetting
875        the start-of-match value in case it was changed by \K. */
876    
877        case OP_ONCE_NC:
878        prev = ecode;
879        saved_eptr = eptr;
880        save_mark = md->mark;
881        do
882          {
883          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
884          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
885            {
886            mstart = md->start_match_ptr;
887            break;
888            }
889          if (rrc == MATCH_THEN)
890            {
891            next = ecode + GET(ecode,1);
892            if (md->start_match_ptr < next &&
893                (*ecode == OP_ALT || *next == OP_ALT))
894              rrc = MATCH_NOMATCH;
895            }
896    
897          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
898          ecode += GET(ecode,1);
899          md->mark = save_mark;
900          }
901        while (*ecode == OP_ALT);
902    
903        /* If hit the end of the group (which could be repeated), fail */
904    
905        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
906    
907        /* Continue as from after the group, updating the offsets high water
908        mark, since extracts may have been taken. */
909    
910        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
911    
912        offset_top = md->end_offset_top;
913        eptr = md->end_match_ptr;
914    
915        /* For a non-repeating ket, just continue at this level. This also
916        happens for a repeating ket if no characters were matched in the group.
917        This is the forcible breaking of infinite loops as implemented in Perl
918        5.005. */
919    
920        if (*ecode == OP_KET || eptr == saved_eptr)
921          {
922          ecode += 1+LINK_SIZE;
923          break;
924          }
925    
926        /* The repeating kets try the rest of the pattern or restart from the
927        preceding bracket, in the appropriate order. The second "call" of match()
928        uses tail recursion, to avoid using another stack frame. */
929    
930        if (*ecode == OP_KETRMIN)
931          {
932          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
933          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
934          ecode = prev;
935          goto TAIL_RECURSE;
936          }
937        else  /* OP_KETRMAX */
938          {
939          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
940          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
941          ecode += 1 + LINK_SIZE;
942          goto TAIL_RECURSE;
943          }
944        /* Control never gets here */
945    
946      /* Handle a capturing bracket, other than those that are possessive with an      /* Handle a capturing bracket, other than those that are possessive with an
947      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
948      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
949      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
950      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
951      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,
952      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
953      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
954      were set by a previous iteration of the same bracket.      were set by a previous iteration of the same bracket.
# Line 822  for (;;) Line 961  for (;;)
961      case OP_SCBRA:      case OP_SCBRA:
962      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
963      offset = number << 1;      offset = number << 1;
964    
965  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
966      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
967      printf("subject=");      printf("subject=");
# Line 836  for (;;) Line 975  for (;;)
975        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
976        save_offset3 = md->offset_vector[md->offset_end - number];        save_offset3 = md->offset_vector[md->offset_end - number];
977        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
978          save_mark = md->mark;
979    
980        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
981        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
# Line 843  for (;;) Line 983  for (;;)
983    
984        for (;;)        for (;;)
985          {          {
986          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
987          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
988            eptrb, RM1);            eptrb, RM1);
989          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
990          if (rrc != MATCH_NOMATCH &&  
991              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* If we backed up to a THEN, check whether it is within the current
992            RRETURN(rrc);          branch by comparing the address of the THEN that is passed back with
993            the end of the branch. If it is within the current branch, and the
994            branch is one of two or more alternatives (it either starts or ends
995            with OP_ALT), we have reached the limit of THEN's action, so convert
996            the return code to NOMATCH, which will cause normal backtracking to
997            happen from now on. Otherwise, THEN is passed back to an outer
998            alternative. This implements Perl's treatment of parenthesized groups,
999            where a group not containing | does not affect the current alternative,
1000            that is, (X) is NOT the same as (X|(*F)). */
1001    
1002            if (rrc == MATCH_THEN)
1003              {
1004              next = ecode + GET(ecode,1);
1005              if (md->start_match_ptr < next &&
1006                  (*ecode == OP_ALT || *next == OP_ALT))
1007                rrc = MATCH_NOMATCH;
1008              }
1009    
1010            /* Anything other than NOMATCH is passed back. */
1011    
1012            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1013          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
1014          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
1015          if (*ecode != OP_ALT) break;          md->mark = save_mark;
1016            if (*ecode != OP_ALT) break;
1017          }          }
1018    
1019        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
1020        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
1021        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
1022        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. */  
1023    
1024        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
1025        RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));  
1026          RRETURN(rrc);
1027        }        }
1028    
1029      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 879  for (;;) Line 1038  for (;;)
1038      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1039    
1040      /* Non-capturing or atomic group, except for possessive with unlimited      /* Non-capturing or atomic group, except for possessive with unlimited
1041      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.
1042      within the brackets, we used to return the result of a recursive call to  
1043      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
1044      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
1045      group. However, now that there is the possiblity of (*THEN) occurring in      possible to reduce stack usage by turning this into a tail recursion,
1046      the final alternative, this optimization is no longer possible.      except in the case of a possibly empty group. However, now that there is
1047        the possiblity of (*THEN) occurring in the final alternative, this
1048      MATCH_ONCE is returned when the end of an atomic group is successfully      optimization is no longer always possible.
1049      reached, but subsequent matching fails. It passes back up the tree (causing  
1050      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
1051        this is the best that can be done.
1052    
1053        MATCH_ONCE is returned when the end of an atomic group is successfully
1054        reached, but subsequent matching fails. It passes back up the tree (causing
1055        captured values to be reset) until the original atomic group level is
1056      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
1057      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
1058      previous backup points can be taken. */      previous backup points can be taken. */
# Line 900  for (;;) Line 1064  for (;;)
1064    
1065      for (;;)      for (;;)
1066        {        {
1067        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA || op == OP_ONCE)
1068        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,          md->match_function_type = MATCH_CBEGROUP;
1069    
1070          /* If this is not a possibly empty group, and there are no (*THEN)s in
1071          the pattern, and this is the final alternative, optimize as described
1072          above. */
1073    
1074          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1075            {
1076            ecode += PRIV(OP_lengths)[*ecode];
1077            goto TAIL_RECURSE;
1078            }
1079    
1080          /* In all other cases, we have to make another call to match(). */
1081    
1082          save_mark = md->mark;
1083          save_capture_last = md->capture_last;
1084          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1085          RM2);          RM2);
1086        if (rrc != MATCH_NOMATCH &&  
1087            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* See comment in the code for capturing groups above about handling
1088          {        THEN. */
1089    
1090          if (rrc == MATCH_THEN)
1091            {
1092            next = ecode + GET(ecode,1);
1093            if (md->start_match_ptr < next &&
1094                (*ecode == OP_ALT || *next == OP_ALT))
1095              rrc = MATCH_NOMATCH;
1096            }
1097    
1098          if (rrc != MATCH_NOMATCH)
1099            {
1100          if (rrc == MATCH_ONCE)          if (rrc == MATCH_ONCE)
1101            {            {
1102            const uschar *scode = ecode;            const pcre_uchar *scode = ecode;
1103            if (*scode != OP_ONCE)           /* If not at start, find it */            if (*scode != OP_ONCE)           /* If not at start, find it */
1104              {              {
1105              while (*scode == OP_ALT) scode += GET(scode, 1);              while (*scode == OP_ALT) scode += GET(scode, 1);
1106              scode -= GET(scode, 1);              scode -= GET(scode, 1);
1107              }              }
1108            if (md->once_target == scode) rrc = MATCH_NOMATCH;            if (md->once_target == scode) rrc = MATCH_NOMATCH;
1109            }            }
1110          RRETURN(rrc);          RRETURN(rrc);
1111          }          }
1112        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1113        if (*ecode != OP_ALT) break;        md->mark = save_mark;
1114          if (*ecode != OP_ALT) break;
1115          md->capture_last = save_capture_last;
1116        }        }
1117      if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;  
1118      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1119    
1120      /* Handle possessive capturing brackets with an unlimited repeat. We come      /* Handle possessive capturing brackets with an unlimited repeat. We come
1121      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
1122      handled similarly to the normal case above. However, the matching is      handled similarly to the normal case above. However, the matching is
1123      different. The end of these brackets will always be OP_KETRPOS, which      different. The end of these brackets will always be OP_KETRPOS, which
1124      returns MATCH_KETRPOS without going further in the pattern. By this means      returns MATCH_KETRPOS without going further in the pattern. By this means
1125      we can handle the group by iteration rather than recursion, thereby      we can handle the group by iteration rather than recursion, thereby
1126      reducing the amount of stack needed. */      reducing the amount of stack needed. */
1127    
1128      case OP_CBRAPOS:      case OP_CBRAPOS:
1129      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1130      allow_zero = FALSE;      allow_zero = FALSE;
1131    
1132      POSSESSIVE_CAPTURE:      POSSESSIVE_CAPTURE:
1133      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
1134      offset = number << 1;      offset = number << 1;
# Line 950  for (;;) Line 1143  for (;;)
1143      if (offset < md->offset_max)      if (offset < md->offset_max)
1144        {        {
1145        matched_once = FALSE;        matched_once = FALSE;
1146        code_offset = ecode - md->start_code;        code_offset = (int)(ecode - md->start_code);
1147    
1148        save_offset1 = md->offset_vector[offset];        save_offset1 = md->offset_vector[offset];
1149        save_offset2 = md->offset_vector[offset+1];        save_offset2 = md->offset_vector[offset+1];
# Line 958  for (;;) Line 1151  for (;;)
1151        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
1152    
1153        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1154    
1155        /* Each time round the loop, save the current subject position for use        /* Each time round the loop, save the current subject position for use
1156        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
1157        restart it with a new subject starting position, remembering that we had        restart it with a new subject starting position, remembering that we had
1158        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
1159        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
1160        see if a previous iteration matched. If so, the group has matched;        see if a previous iteration matched. If so, the group has matched;
1161        continue from afterwards. Otherwise it has failed; restore the previous        continue from afterwards. Otherwise it has failed; restore the previous
1162        capture values before returning NOMATCH. */        capture values before returning NOMATCH. */
1163    
1164        for (;;)        for (;;)
1165          {          {
1166          md->offset_vector[md->offset_end - number] =          md->offset_vector[md->offset_end - number] =
1167            (int)(eptr - md->start_subject);            (int)(eptr - md->start_subject);
1168          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1169          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1170            eptrb, RM63);            eptrb, RM63);
1171          if (rrc == MATCH_KETRPOS)          if (rrc == MATCH_KETRPOS)
1172            {            {
1173            offset_top = md->end_offset_top;            offset_top = md->end_offset_top;
1174            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
1175            ecode = md->start_code + code_offset;            ecode = md->start_code + code_offset;
1176            save_capture_last = md->capture_last;            save_capture_last = md->capture_last;
1177            matched_once = TRUE;            matched_once = TRUE;
1178            continue;            continue;
1179            }            }
1180          if (rrc != MATCH_NOMATCH &&  
1181              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* See comment in the code for capturing groups above about handling
1182            RRETURN(rrc);          THEN. */
1183    
1184            if (rrc == MATCH_THEN)
1185              {
1186              next = ecode + GET(ecode,1);
1187              if (md->start_match_ptr < next &&
1188                  (*ecode == OP_ALT || *next == OP_ALT))
1189                rrc = MATCH_NOMATCH;
1190              }
1191    
1192            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1193          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
1194          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
1195          if (*ecode != OP_ALT) break;          if (*ecode != OP_ALT) break;
1196          }          }
1197    
1198        if (!matched_once)        if (!matched_once)
1199          {          {
1200          md->offset_vector[offset] = save_offset1;          md->offset_vector[offset] = save_offset1;
1201          md->offset_vector[offset+1] = save_offset2;          md->offset_vector[offset+1] = save_offset2;
1202          md->offset_vector[md->offset_end - number] = save_offset3;          md->offset_vector[md->offset_end - number] = save_offset3;
1203          }          }
1204    
       if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;  
1205        if (allow_zero || matched_once)        if (allow_zero || matched_once)
1206          {          {
1207          ecode += 1 + LINK_SIZE;          ecode += 1 + LINK_SIZE;
1208          break;          break;
1209          }          }
1210    
1211        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1212        }        }
1213    
1214      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1215      as a non-capturing bracket. */      as a non-capturing bracket. */
1216    
# Line 1020  for (;;) Line 1222  for (;;)
1222      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1223      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1224    
1225      /* Non-capturing possessive bracket with unlimited repeat. We come here      /* Non-capturing possessive bracket with unlimited repeat. We come here
1226      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,
1227      without the capturing complication. It is written out separately for speed      without the capturing complication. It is written out separately for speed
1228      and cleanliness. */      and cleanliness. */
1229    
1230      case OP_BRAPOS:      case OP_BRAPOS:
1231      case OP_SBRAPOS:      case OP_SBRAPOS:
1232      allow_zero = FALSE;      allow_zero = FALSE;
1233    
1234      POSSESSIVE_NON_CAPTURE:      POSSESSIVE_NON_CAPTURE:
1235      matched_once = FALSE;      matched_once = FALSE;
1236      code_offset = ecode - md->start_code;      code_offset = (int)(ecode - md->start_code);
1237        save_capture_last = md->capture_last;
1238    
1239      for (;;)      for (;;)
1240        {        {
1241        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;        if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1242        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1243          eptrb, RM48);          eptrb, RM48);
1244        if (rrc == MATCH_KETRPOS)        if (rrc == MATCH_KETRPOS)
1245          {          {
1246          offset_top = md->end_offset_top;          offset_top = md->end_offset_top;
1247          eptr = md->end_match_ptr;          eptr = md->end_match_ptr;
1248          ecode = md->start_code + code_offset;          ecode = md->start_code + code_offset;
1249          matched_once = TRUE;          matched_once = TRUE;
1250          continue;          continue;
1251          }          }
1252        if (rrc != MATCH_NOMATCH &&  
1253            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* See comment in the code for capturing groups above about handling
1254          RRETURN(rrc);        THEN. */
1255    
1256          if (rrc == MATCH_THEN)
1257            {
1258            next = ecode + GET(ecode,1);
1259            if (md->start_match_ptr < next &&
1260                (*ecode == OP_ALT || *next == OP_ALT))
1261              rrc = MATCH_NOMATCH;
1262            }
1263    
1264          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1265        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1266        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1267          md->capture_last = save_capture_last;
1268        }        }
1269    
1270      if (matched_once || allow_zero)      if (matched_once || allow_zero)
1271        {        {
1272        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1273        break;        break;
1274        }        }
1275      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
1276    
1277      /* Control never reaches here. */      /* Control never reaches here. */
# Line 1076  for (;;) Line 1290  for (;;)
1290    
1291      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1292        {        {
1293        if (pcre_callout != NULL)        if (PUBL(callout) != NULL)
1294          {          {
1295          pcre_callout_block cb;          PUBL(callout_block) cb;
1296          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1297          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1298          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1299    #if defined COMPILE_PCRE8
1300          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1301    #elif defined COMPILE_PCRE16
1302            cb.subject          = (PCRE_SPTR16)md->start_subject;
1303    #elif defined COMPILE_PCRE32
1304            cb.subject          = (PCRE_SPTR32)md->start_subject;
1305    #endif
1306          cb.subject_length   = (int)(md->end_subject - md->start_subject);          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1307          cb.start_match      = (int)(mstart - md->start_subject);          cb.start_match      = (int)(mstart - md->start_subject);
1308          cb.current_position = (int)(eptr - md->start_subject);          cb.current_position = (int)(eptr - md->start_subject);
1309          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1310          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1311          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1312          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last & CAPLMASK;
1313            /* Internal change requires this for API compatibility. */
1314            if (cb.capture_last == 0) cb.capture_last = -1;
1315          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1316          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1317            if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1318          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1319          }          }
1320        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1321        }        }
1322    
1323      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
# Line 1110  for (;;) Line 1333  for (;;)
1333          }          }
1334        else        else
1335          {          {
1336          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          unsigned int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1337          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1338    
1339          /* 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
1340          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
1341          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
1342          if any one is set. */          if any one is set. */
1343    
1344          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1345            {            {
1346            uschar *slotA = md->name_table;            pcre_uchar *slotA = md->name_table;
1347            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
1348              {              {
1349              if (GET2(slotA, 0) == recno) break;              if (GET2(slotA, 0) == recno) break;
# Line 1133  for (;;) Line 1356  for (;;)
1356    
1357            if (i < md->name_count)            if (i < md->name_count)
1358              {              {
1359              uschar *slotB = slotA;              pcre_uchar *slotB = slotA;
1360              while (slotB > md->name_table)              while (slotB > md->name_table)
1361                {                {
1362                slotB -= md->name_entry_size;                slotB -= md->name_entry_size;
1363                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1364                  {                  {
1365                  condition = GET2(slotB, 0) == md->recursive->group_num;                  condition = GET2(slotB, 0) == md->recursive->group_num;
1366                  if (condition) break;                  if (condition) break;
# Line 1153  for (;;) Line 1376  for (;;)
1376                for (i++; i < md->name_count; i++)                for (i++; i < md->name_count; i++)
1377                  {                  {
1378                  slotB += md->name_entry_size;                  slotB += md->name_entry_size;
1379                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1380                    {                    {
1381                    condition = GET2(slotB, 0) == md->recursive->group_num;                    condition = GET2(slotB, 0) == md->recursive->group_num;
1382                    if (condition) break;                    if (condition) break;
# Line 1166  for (;;) Line 1389  for (;;)
1389    
1390          /* Chose branch according to the condition */          /* Chose branch according to the condition */
1391    
1392          ecode += condition? 3 : GET(ecode, 1);          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1393          }          }
1394        }        }
1395    
# Line 1182  for (;;) Line 1405  for (;;)
1405    
1406        if (!condition && condcode == OP_NCREF)        if (!condition && condcode == OP_NCREF)
1407          {          {
1408          int refno = offset >> 1;          unsigned int refno = offset >> 1;
1409          uschar *slotA = md->name_table;          pcre_uchar *slotA = md->name_table;
1410    
1411          for (i = 0; i < md->name_count; i++)          for (i = 0; i < md->name_count; i++)
1412            {            {
# Line 1197  for (;;) Line 1420  for (;;)
1420    
1421          if (i < md->name_count)          if (i < md->name_count)
1422            {            {
1423            uschar *slotB = slotA;            pcre_uchar *slotB = slotA;
1424            while (slotB > md->name_table)            while (slotB > md->name_table)
1425              {              {
1426              slotB -= md->name_entry_size;              slotB -= md->name_entry_size;
1427              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)              if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1428                {                {
1429                offset = GET2(slotB, 0) << 1;                offset = GET2(slotB, 0) << 1;
1430                condition = offset < offset_top &&                condition = offset < offset_top &&
# Line 1219  for (;;) Line 1442  for (;;)
1442              for (i++; i < md->name_count; i++)              for (i++; i < md->name_count; i++)
1443                {                {
1444                slotB += md->name_entry_size;                slotB += md->name_entry_size;
1445                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1446                  {                  {
1447                  offset = GET2(slotB, 0) << 1;                  offset = GET2(slotB, 0) << 1;
1448                  condition = offset < offset_top &&                  condition = offset < offset_top &&
# Line 1234  for (;;) Line 1457  for (;;)
1457    
1458        /* Chose branch according to the condition */        /* Chose branch according to the condition */
1459    
1460        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1461        }        }
1462    
1463      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 1249  for (;;) Line 1472  for (;;)
1472    
1473      else      else
1474        {        {
1475        md->match_function_type = MATCH_CONDASSERT;        md->match_function_type = MATCH_CONDASSERT;
1476        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1477        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1478          {          {
1479            if (md->end_offset_top > offset_top)
1480              offset_top = md->end_offset_top;  /* Captures may have happened */
1481          condition = TRUE;          condition = TRUE;
1482          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1483          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1484          }          }
1485        else if (rrc != MATCH_NOMATCH &&  
1486                (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1487          assertion; it is therefore treated as NOMATCH. */
1488    
1489          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1490          {          {
1491          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1492          }          }
# Line 1269  for (;;) Line 1497  for (;;)
1497          }          }
1498        }        }
1499    
1500      /* 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
1501      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
1502      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
1503      strategy no longer works because of the possibilty of (*THEN) being      call to match() is always required, unless the second alternative doesn't
1504      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
1505      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
1506      plough on. */      alternatives. If a THEN is encountered in the branch, it propagates out to
1507        the enclosing alternative (unless nested in a deeper set of alternatives,
1508        of course). */
1509    
1510      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1511        {        {
1512        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;        if (op != OP_SCOND)
1513            {
1514            ecode += 1 + LINK_SIZE;
1515            goto TAIL_RECURSE;
1516            }
1517    
1518          md->match_function_type = MATCH_CBEGROUP;
1519        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;  
1520        RRETURN(rrc);        RRETURN(rrc);
1521        }        }
1522      else                         /* Condition false & no alternative */  
1523         /* Condition false & no alternative; continue after the group. */
1524    
1525        else
1526        {        {
1527        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1528        }        }
# Line 1296  for (;;) Line 1533  for (;;)
1533      to close any currently open capturing brackets. */      to close any currently open capturing brackets. */
1534    
1535      case OP_CLOSE:      case OP_CLOSE:
1536      number = GET2(ecode, 1);      number = GET2(ecode, 1);   /* Must be less than 65536 */
1537      offset = number << 1;      offset = number << 1;
1538    
1539  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 1304  for (;;) Line 1541  for (;;)
1541        printf("\n");        printf("\n");
1542  #endif  #endif
1543    
1544      md->capture_last = number;      md->capture_last = (md->capture_last & OVFLMASK) | number;
1545      if (offset >= md->offset_max) md->offset_overflow = TRUE; else      if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else
1546        {        {
1547        md->offset_vector[offset] =        md->offset_vector[offset] =
1548          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1549        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1550        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1551        }        }
1552      ecode += 3;      ecode += 1 + IMM2_SIZE;
1553      break;      break;
1554    
1555    
1556      /* 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. */  
1557    
     case OP_ACCEPT:  
     case OP_ASSERT_ACCEPT:  
1558      case OP_END:      case OP_END:
1559        case OP_ACCEPT:
1560  /*      case OP_ASSERT_ACCEPT:
     if (md->recursive != NULL)  
       {  
       recursion_info *rec = md->recursive;  
   
       md->recursive = rec->prevrec;  
1561    
1562        memmove(md->offset_vector, rec->offset_save,      /* If we have matched an empty string, fail if not in an assertion and not
1563          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  
1564      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,
1565      backtracking will then try other alternatives, if any. */      backtracking will then try other alternatives, if any. */
1566    
1567  /*    else */ if (eptr == mstart && op != OP_ASSERT_ACCEPT &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
   
1568           md->recursive == NULL &&           md->recursive == NULL &&
1569             (md->notempty ||
1570          (md->notempty ||             (md->notempty_atstart &&
1571            (md->notempty_atstart &&               mstart == md->start_subject + md->start_offset)))
1572              mstart == md->start_subject + md->start_offset)))        RRETURN(MATCH_NOMATCH);
       MRRETURN(MATCH_NOMATCH);  
1573    
1574      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1575    
1576      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1577      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1578      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1579    
1580      /* 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
1581      given as the argument to MRRETURN when the heap is in use. */      given as the argument to RRETURN when the heap is in use. */
1582    
1583      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1584      MRRETURN(rrc);      RRETURN(rrc);
1585    
1586      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1587      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,
1588      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
1589      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
1590      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
1591      of a condition, we want to return immediately afterwards. The caller of      of a condition, we want to return immediately afterwards. The caller of
1592      this incarnation of the match() function will have set MATCH_CONDASSERT in      this incarnation of the match() function will have set MATCH_CONDASSERT in
1593      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
1594      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
1595      match() to remember this case. */      match() to remember this case. */
1596    
1597      case OP_ASSERT:      case OP_ASSERT:
1598      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1599        save_mark = md->mark;
1600      if (md->match_function_type == MATCH_CONDASSERT)      if (md->match_function_type == MATCH_CONDASSERT)
1601        {        {
1602        condassert = TRUE;        condassert = TRUE;
1603        md->match_function_type = 0;        md->match_function_type = 0;
1604        }        }
1605      else condassert = FALSE;      else condassert = FALSE;
1606    
1607      do      do
1608        {        {
1609        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 1612  for (;;)
1612          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1613          break;          break;
1614          }          }
1615        if (rrc != MATCH_NOMATCH &&        md->mark = save_mark;
1616            (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
1617          RRETURN(rrc);        /* A COMMIT failure must fail the entire assertion, without trying any
1618          subsequent branches. */
1619    
1620          if (rrc == MATCH_COMMIT) RRETURN(MATCH_NOMATCH);
1621    
1622          /* PCRE does not allow THEN to escape beyond an assertion; it
1623          is treated as NOMATCH. */
1624    
1625          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1626        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1627        }        }
1628      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1629    
1630      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1631    
1632      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1633    
# Line 1422  for (;;) Line 1647  for (;;)
1647    
1648      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1649      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1650        save_mark = md->mark;
1651      if (md->match_function_type == MATCH_CONDASSERT)      if (md->match_function_type == MATCH_CONDASSERT)
1652        {        {
1653        condassert = TRUE;        condassert = TRUE;
1654        md->match_function_type = 0;        md->match_function_type = 0;
1655        }        }
1656      else condassert = FALSE;      else condassert = FALSE;
1657    
1658      do      do
1659        {        {
1660        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1661        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        md->mark = save_mark;
1662          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
1663        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1664          {          {
1665          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1666          break;          break;
1667          }          }
1668        if (rrc != MATCH_NOMATCH &&  
1669            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1670          RRETURN(rrc);        as NOMATCH. */
1671    
1672          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1673        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1674        }        }
1675      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1676    
1677      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1678    
1679      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1680      continue;      continue;
1681    
# Line 1456  for (;;) Line 1685  for (;;)
1685      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1686    
1687      case OP_REVERSE:      case OP_REVERSE:
1688  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1689      if (utf8)      if (utf)
1690        {        {
1691        i = GET(ecode, 1);        i = GET(ecode, 1);
1692        while (i-- > 0)        while (i-- > 0)
1693          {          {
1694          eptr--;          eptr--;
1695          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1696          BACKCHAR(eptr);          BACKCHAR(eptr);
1697          }          }
1698        }        }
# Line 1474  for (;;) Line 1703  for (;;)
1703    
1704        {        {
1705        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1706        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1707        }        }
1708    
1709      /* 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 1717  for (;;)
1717      function is able to force a failure. */      function is able to force a failure. */
1718    
1719      case OP_CALLOUT:      case OP_CALLOUT:
1720      if (pcre_callout != NULL)      if (PUBL(callout) != NULL)
1721        {        {
1722        pcre_callout_block cb;        PUBL(callout_block) cb;
1723        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1724        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1725        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1726    #if defined COMPILE_PCRE8
1727        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1728    #elif defined COMPILE_PCRE16
1729          cb.subject          = (PCRE_SPTR16)md->start_subject;
1730    #elif defined COMPILE_PCRE32
1731          cb.subject          = (PCRE_SPTR32)md->start_subject;
1732    #endif
1733        cb.subject_length   = (int)(md->end_subject - md->start_subject);        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1734        cb.start_match      = (int)(mstart - md->start_subject);        cb.start_match      = (int)(mstart - md->start_subject);
1735        cb.current_position = (int)(eptr - md->start_subject);        cb.current_position = (int)(eptr - md->start_subject);
1736        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1737        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1738        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1739        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last & CAPLMASK;
1740          /* Internal change requires this for API compatibility. */
1741          if (cb.capture_last == 0) cb.capture_last = -1;
1742        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1743        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1744          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1745        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1746        }        }
1747      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1512  for (;;) Line 1750  for (;;)
1750      /* Recursion either matches the current regex, or some subexpression. The      /* Recursion either matches the current regex, or some subexpression. The
1751      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
1752      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1753    
1754      The state of the capturing groups is preserved over recursion, and      The state of the capturing groups is preserved over recursion, and
1755      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
1756      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
1757      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
1758      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 1761  for (;;)
1761    
1762      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
1763      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
1764      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
1765      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. */
1766    
1767      case OP_RECURSE:      case OP_RECURSE:
1768        {        {
1769          recursion_info *ri;
1770          unsigned int recno;
1771    
1772        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1773        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1774          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1775    
1776          /* Check for repeating a recursion without advancing the subject pointer.
1777          This should catch convoluted mutual recursions. (Some simple cases are
1778          caught at compile time.) */
1779    
1780          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1781            if (recno == ri->group_num && eptr == ri->subject_position)
1782              RRETURN(PCRE_ERROR_RECURSELOOP);
1783    
1784        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1785    
1786          new_recursive.group_num = recno;
1787          new_recursive.saved_capture_last = md->capture_last;
1788          new_recursive.subject_position = eptr;
1789        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1790        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1791    
# Line 1549  for (;;) Line 1801  for (;;)
1801        else        else
1802          {          {
1803          new_recursive.offset_save =          new_recursive.offset_save =
1804            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int));
1805          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1806          }          }
1807        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1808              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1809    
1810        /* OK, now we can do the recursion. After processing each alternative,        /* OK, now we can do the recursion. After processing each alternative,
1811        restore the offset data. If there were nested recursions, md->recursive        restore the offset data and the last captured value. If there were nested
1812        might be changed, so reset it before looping. */        recursions, md->recursive might be changed, so reset it before looping.
1813          */
1814    
1815        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1816        cbegroup = (*callpat >= OP_SBRA);        cbegroup = (*callpat >= OP_SBRA);
1817        do        do
1818          {          {
1819          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1820          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1821            md, eptrb, RM6);            md, eptrb, RM6);
1822          memcpy(md->offset_vector, new_recursive.offset_save,          memcpy(md->offset_vector, new_recursive.offset_save,
1823              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1824            md->capture_last = new_recursive.saved_capture_last;
1825            md->recursive = new_recursive.prevrec;
1826          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1827            {            {
1828            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1829            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1830              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1831    
1832            /* 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
1833            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,
1834            for Perl compatibility. */            for Perl compatibility. */
1835    
1836            eptr = md->end_match_ptr;            eptr = md->end_match_ptr;
1837            mstart = md->start_match_ptr;            mstart = md->start_match_ptr;
1838            goto RECURSION_MATCHED;        /* Exit loop; end processing */            goto RECURSION_MATCHED;        /* Exit loop; end processing */
1839            }            }
1840          else if (rrc != MATCH_NOMATCH &&  
1841                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* PCRE does not allow THEN or COMMIT to escape beyond a recursion; it
1842            is treated as NOMATCH. */
1843    
1844            else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN &&
1845                     rrc != MATCH_COMMIT)
1846            {            {
1847            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1848            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1849              (pcre_free)(new_recursive.offset_save);              (PUBL(free))(new_recursive.offset_save);
1850            RRETURN(rrc);            RRETURN(rrc);
1851            }            }
1852    
# Line 1600  for (;;) Line 1858  for (;;)
1858        DPRINTF(("Recursion didn't match\n"));        DPRINTF(("Recursion didn't match\n"));
1859        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1860        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1861          (pcre_free)(new_recursive.offset_save);          (PUBL(free))(new_recursive.offset_save);
1862        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
1863        }        }
1864    
1865      RECURSION_MATCHED:      RECURSION_MATCHED:
1866      break;      break;
1867    
# Line 1619  for (;;) Line 1877  for (;;)
1877      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
1878      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
1879      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1880    
1881      case OP_BRAZERO:      case OP_BRAZERO:
1882      next = ecode + 1;      next = ecode + 1;
1883      RMATCH(eptr, next, offset_top, md, eptrb, RM10);      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
# Line 1627  for (;;) Line 1885  for (;;)
1885      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
1886      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1887      break;      break;
1888    
1889      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1890      next = ecode + 1;      next = ecode + 1;
1891      do next += GET(next, 1); while (*next == OP_ALT);      do next += GET(next, 1); while (*next == OP_ALT);
# Line 1641  for (;;) Line 1899  for (;;)
1899      do next += GET(next,1); while (*next == OP_ALT);      do next += GET(next,1); while (*next == OP_ALT);
1900      ecode = next + 1 + LINK_SIZE;      ecode = next + 1 + LINK_SIZE;
1901      break;      break;
1902    
1903      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1904      here; just jump to the group, with allow_zero set TRUE. */      here; just jump to the group, with allow_zero set TRUE. */
1905    
1906      case OP_BRAPOSZERO:      case OP_BRAPOSZERO:
1907      op = *(++ecode);      op = *(++ecode);
1908      allow_zero = TRUE;      allow_zero = TRUE;
1909      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1910        goto POSSESSIVE_NON_CAPTURE;        goto POSSESSIVE_NON_CAPTURE;
# Line 1656  for (;;) Line 1914  for (;;)
1914      case OP_KET:      case OP_KET:
1915      case OP_KETRMIN:      case OP_KETRMIN:
1916      case OP_KETRMAX:      case OP_KETRMAX:
1917      case OP_KETRPOS:      case OP_KETRPOS:
1918      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1919    
1920      /* 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
1921      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1922      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
# Line 1670  for (;;) Line 1928  for (;;)
1928        }        }
1929      else saved_eptr = NULL;      else saved_eptr = NULL;
1930    
1931      /* 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
1932      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
1933      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
1934      by \K. */      start in case it was changed by \K. */
1935    
1936      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1937          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)           *prev == OP_ONCE_NC)
1938        {        {
1939        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1940        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1941        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1942        MRRETURN(MATCH_MATCH);        RRETURN(MATCH_MATCH);         /* Sets md->mark */
1943        }        }
1944    
1945      /* 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 1972  for (;;)
1972    
1973        /* Deal with capturing */        /* Deal with capturing */
1974    
1975        md->capture_last = number;        md->capture_last = (md->capture_last & OVFLMASK) | number;
1976        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else
1977          {          {
1978          /* If offset is greater than offset_top, it means that we are          /* If offset is greater than offset_top, it means that we are
1979          "skipping" a capturing group, and that group's offsets must be marked          "skipping" a capturing group, and that group's offsets must be marked
1980          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
1981          start of matching, but this doesn't work because atomic groups and          start of matching, but this doesn't work because atomic groups and
1982          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.
1983          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
1984          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,
1985          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
1986          problem, because offset_top will then be 2, indicating no capture.) */          problem, because offset_top will then be 2, indicating no capture.) */
1987    
1988          if (offset > offset_top)          if (offset > offset_top)
1989            {            {
1990            register int *iptr = md->offset_vector + offset_top;            register int *iptr = md->offset_vector + offset_top;
1991            register int *iend = md->offset_vector + offset;            register int *iend = md->offset_vector + offset;
1992            while (iptr < iend) *iptr++ = -1;            while (iptr < iend) *iptr++ = -1;
1993            }            }
1994    
1995          /* Now make the extraction */          /* Now make the extraction */
1996    
1997          md->offset_vector[offset] =          md->offset_vector[offset] =
# Line 1746  for (;;) Line 2004  for (;;)
2004      /* For an ordinary non-repeating ket, just continue at this level. This      /* For an ordinary non-repeating ket, just continue at this level. This
2005      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
2006      group. This is the forcible breaking of infinite loops as implemented in      group. This is the forcible breaking of infinite loops as implemented in
2007      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,
2008      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
2009      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
2010      bypassing intermediate backup points, but resetting any captures that      original OP_ONCE level, thereby bypassing intermediate backup points, but
2011      happened along the way. */      resetting any captures that happened along the way. */
2012    
2013      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
2014        {        {
# Line 1759  for (;;) Line 2017  for (;;)
2017          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
2018          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2019          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2020          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
2021          }          }
2022        ecode += 1 + LINK_SIZE;    /* Carry on at this level */        ecode += 1 + LINK_SIZE;    /* Carry on at this level */
2023        break;        break;
2024        }        }
2025    
2026      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
2027      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
2028      at a time from the outer level, thus saving stack. */      at a time from the outer level, thus saving stack. */
2029    
2030      if (*ecode == OP_KETRPOS)      if (*ecode == OP_KETRPOS)
2031        {        {
2032        md->end_match_ptr = eptr;        md->end_match_ptr = eptr;
2033        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
2034        RRETURN(MATCH_KETRPOS);        RRETURN(MATCH_KETRPOS);
2035        }        }
2036    
2037      /* 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
2038      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 2042  for (;;)
2042    
2043      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
2044        {        {
2045        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
2046        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2047        if (*prev == OP_ONCE)        if (*prev == OP_ONCE)
2048          {          {
2049          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
2050          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2051          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
2052          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
2053          }          }
2054        if (*prev >= OP_SBRA)    /* Could match an empty string */        if (*prev >= OP_SBRA)    /* Could match an empty string */
2055          {          {
         md->match_function_type = MATCH_CBEGROUP;  
2056          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
2057          RRETURN(rrc);          RRETURN(rrc);
2058          }          }
# Line 1804  for (;;) Line 2061  for (;;)
2061        }        }
2062      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
2063        {        {
       if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;  
2064        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);        RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
2065        if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;        if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
2066        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2067        if (*prev == OP_ONCE)        if (*prev == OP_ONCE)
2068          {          {
2069          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
2070          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2071          md->once_target = prev;          md->once_target = prev;
2072          RRETURN(MATCH_ONCE);          RRETURN(MATCH_ONCE);
2073          }          }
2074        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
2075        goto TAIL_RECURSE;        goto TAIL_RECURSE;
2076        }        }
# Line 1823  for (;;) Line 2079  for (;;)
2079      /* Not multiline mode: start of subject assertion, unless notbol. */      /* Not multiline mode: start of subject assertion, unless notbol. */
2080    
2081      case OP_CIRC:      case OP_CIRC:
2082      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2083    
2084      /* Start of subject assertion */      /* Start of subject assertion */
2085    
2086      case OP_SOD:      case OP_SOD:
2087      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);
2088      ecode++;      ecode++;
2089      break;      break;
2090    
2091      /* Multiline mode: start of subject unless notbol, or after any newline. */      /* Multiline mode: start of subject unless notbol, or after any newline. */
2092    
2093      case OP_CIRCM:      case OP_CIRCM:
2094      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
2095      if (eptr != md->start_subject &&      if (eptr != md->start_subject &&
2096          (eptr == md->end_subject || !WAS_NEWLINE(eptr)))          (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
2097        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2098      ecode++;      ecode++;
2099      break;      break;
2100    
2101      /* Start of match assertion */      /* Start of match assertion */
2102    
2103      case OP_SOM:      case OP_SOM:
2104      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);
2105      ecode++;      ecode++;
2106      break;      break;
2107    
# Line 1861  for (;;) Line 2117  for (;;)
2117    
2118      case OP_DOLLM:      case OP_DOLLM:
2119      if (eptr < md->end_subject)      if (eptr < md->end_subject)
2120        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }        {
2121          if (!IS_NEWLINE(eptr))
2122            {
2123            if (md->partial != 0 &&
2124                eptr + 1 >= md->end_subject &&
2125                NLBLOCK->nltype == NLTYPE_FIXED &&
2126                NLBLOCK->nllen == 2 &&
2127                RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2128              {
2129              md->hitend = TRUE;
2130              if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2131              }
2132            RRETURN(MATCH_NOMATCH);
2133            }
2134          }
2135      else      else
2136        {        {
2137        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2138        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2139        }        }
2140      ecode++;      ecode++;
2141      break;      break;
2142    
2143      /* Not multiline mode: assert before a terminating newline or before end of      /* Not multiline mode: assert before a terminating newline or before end of
2144      subject unless noteol is set. */      subject unless noteol is set. */
2145    
2146      case OP_DOLL:      case OP_DOLL:
2147      if (md->noteol) MRRETURN(MATCH_NOMATCH);      if (md->noteol) RRETURN(MATCH_NOMATCH);
2148      if (!md->endonly) goto ASSERT_NL_OR_EOS;      if (!md->endonly) goto ASSERT_NL_OR_EOS;
2149    
2150      /* ... else fall through for endonly */      /* ... else fall through for endonly */
# Line 1882  for (;;) Line 2152  for (;;)
2152      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2153    
2154      case OP_EOD:      case OP_EOD:
2155      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2156      SCHECK_PARTIAL();      SCHECK_PARTIAL();
2157      ecode++;      ecode++;
2158      break;      break;
# Line 1893  for (;;) Line 2163  for (;;)
2163      ASSERT_NL_OR_EOS:      ASSERT_NL_OR_EOS:
2164      if (eptr < md->end_subject &&      if (eptr < md->end_subject &&
2165          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2166        MRRETURN(MATCH_NOMATCH);        {
2167          if (md->partial != 0 &&
2168              eptr + 1 >= md->end_subject &&
2169              NLBLOCK->nltype == NLTYPE_FIXED &&
2170              NLBLOCK->nllen == 2 &&
2171              RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2172            {
2173            md->hitend = TRUE;
2174            if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2175            }
2176          RRETURN(MATCH_NOMATCH);
2177          }
2178    
2179      /* Either at end of string or \n before end. */      /* Either at end of string or \n before end. */
2180    
# Line 1912  for (;;) Line 2193  for (;;)
2193        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2194        partial matching. */        partial matching. */
2195    
2196  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2197        if (utf8)        if (utf)
2198          {          {
2199          /* Get status of previous character */          /* Get status of previous character */
2200    
2201          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2202            {            {
2203            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2204            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2205            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2206            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2207  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 1985  for (;;) Line 2266  for (;;)
2266              }              }
2267            else            else
2268  #endif  #endif
2269            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);            prev_is_word = MAX_255(eptr[-1])
2270                && ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2271            }            }
2272    
2273          /* Get status of next character */          /* Get status of next character */
# Line 2008  for (;;) Line 2290  for (;;)
2290            }            }
2291          else          else
2292  #endif  #endif
2293          cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          cur_is_word = MAX_255(*eptr)
2294              && ((md->ctypes[*eptr] & ctype_word) != 0);
2295          }          }
2296    
2297        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2298    
2299        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2300             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2301          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2302        }        }
2303      break;      break;
2304    
2305      /* Match a single character type; inline for speed */      /* Match any single character type except newline; have to take care with
2306        CRLF newlines and partial matching. */
2307    
2308      case OP_ANY:      case OP_ANY:
2309      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
2310        if (md->partial != 0 &&
2311            eptr + 1 >= md->end_subject &&
2312            NLBLOCK->nltype == NLTYPE_FIXED &&
2313            NLBLOCK->nllen == 2 &&
2314            RAWUCHARTEST(eptr) == NLBLOCK->nl[0])
2315          {
2316          md->hitend = TRUE;
2317          if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2318          }
2319    
2320      /* Fall through */      /* Fall through */
2321    
2322        /* Match any single character whatsoever. */
2323    
2324      case OP_ALLANY:      case OP_ALLANY:
2325      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2326        {        {                            /* not be updated before SCHECK_PARTIAL. */
2327        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2328        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2329        }        }
2330      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2331    #ifdef SUPPORT_UTF
2332        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2333    #endif
2334      ecode++;      ecode++;
2335      break;      break;
2336    
# Line 2039  for (;;) Line 2338  for (;;)
2338      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2339    
2340      case OP_ANYBYTE:      case OP_ANYBYTE:
2341      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2342        {        {                            /* not be updated before SCHECK_PARTIAL. */
2343        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2344        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2345        }        }
2346        eptr++;
2347      ecode++;      ecode++;
2348      break;      break;
2349    
# Line 2051  for (;;) Line 2351  for (;;)
2351      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2352        {        {
2353        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2354        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2355        }        }
2356      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2357      if (      if (
2358  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2359         c < 256 &&         c < 256 &&
2360  #endif  #endif
2361         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2362         )         )
2363        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2364      ecode++;      ecode++;
2365      break;      break;
2366    
# Line 2068  for (;;) Line 2368  for (;;)
2368      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2369        {        {
2370        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2371        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2372        }        }
2373      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2374      if (      if (
2375  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2376         c >= 256 ||         c > 255 ||
2377  #endif  #endif
2378         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2379         )         )
2380        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2381      ecode++;      ecode++;
2382      break;      break;
2383    
# Line 2085  for (;;) Line 2385  for (;;)
2385      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2386        {        {
2387        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2388        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2389        }        }
2390      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2391      if (      if (
2392  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2393         c < 256 &&         c < 256 &&
2394  #endif  #endif
2395         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2396         )         )
2397        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2398      ecode++;      ecode++;
2399      break;      break;
2400    
# Line 2102  for (;;) Line 2402  for (;;)
2402      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2403        {        {
2404        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2405        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2406        }        }
2407      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2408      if (      if (
2409  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2410         c >= 256 ||         c > 255 ||
2411  #endif  #endif
2412         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2413         )         )
2414        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2415      ecode++;      ecode++;
2416      break;      break;
2417    
# Line 2119  for (;;) Line 2419  for (;;)
2419      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2420        {        {
2421        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2422        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2423        }        }
2424      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2425      if (      if (
2426  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2427         c < 256 &&         c < 256 &&
2428  #endif  #endif
2429         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2430         )         )
2431        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2432      ecode++;      ecode++;
2433      break;      break;
2434    
# Line 2136  for (;;) Line 2436  for (;;)
2436      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2437        {        {
2438        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2439        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2440        }        }
2441      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2442      if (      if (
2443  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2444         c >= 256 ||         c > 255 ||
2445  #endif  #endif
2446         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2447         )         )
2448        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2449      ecode++;      ecode++;
2450      break;      break;
2451    
# Line 2153  for (;;) Line 2453  for (;;)
2453      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2454        {        {
2455        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2456        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2457        }        }
2458      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2459      switch(c)      switch(c)
2460        {        {
2461        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2462    
2463        case 0x000d:        case CHAR_CR:
2464        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr >= md->end_subject)
2465            {
2466            SCHECK_PARTIAL();
2467            }
2468          else if (RAWUCHARTEST(eptr) == CHAR_LF) eptr++;
2469        break;        break;
2470    
2471        case 0x000a:        case CHAR_LF:
2472        break;        break;
2473    
2474        case 0x000b:        case CHAR_VT:
2475        case 0x000c:        case CHAR_FF:
2476        case 0x0085:        case CHAR_NEL:
2477    #ifndef EBCDIC
2478        case 0x2028:        case 0x2028:
2479        case 0x2029:        case 0x2029:
2480        if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);  #endif  /* Not EBCDIC */
2481          if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
2482        break;        break;
2483        }        }
2484      ecode++;      ecode++;
# Line 2182  for (;;) Line 2488  for (;;)
2488      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2489        {        {
2490        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2491        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2492        }        }
2493      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2494      switch(c)      switch(c)
2495        {        {
2496          HSPACE_CASES: RRETURN(MATCH_NOMATCH);  /* Byte and multibyte cases */
2497        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);  
2498        }        }
2499      ecode++;      ecode++;
2500      break;      break;
# Line 2216  for (;;) Line 2503  for (;;)
2503      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2504        {        {
2505        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2506        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2507        }        }
2508      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2509      switch(c)      switch(c)
2510        {        {
2511        default: MRRETURN(MATCH_NOMATCH);        HSPACE_CASES: break;  /* Byte and multibyte cases */
2512        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;  
2513        }        }
2514      ecode++;      ecode++;
2515      break;      break;
# Line 2250  for (;;) Line 2518  for (;;)
2518      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2519        {        {
2520        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2521        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2522        }        }
2523      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2524      switch(c)      switch(c)
2525        {        {
2526          VSPACE_CASES: RRETURN(MATCH_NOMATCH);
2527        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);  
2528        }        }
2529      ecode++;      ecode++;
2530      break;      break;
# Line 2272  for (;;) Line 2533  for (;;)
2533      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2534        {        {
2535        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2536        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2537        }        }
2538      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2539      switch(c)      switch(c)
2540        {        {
2541        default: MRRETURN(MATCH_NOMATCH);        VSPACE_CASES: break;
2542        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;  
2543        }        }
2544      ecode++;      ecode++;
2545      break;      break;
# Line 2299  for (;;) Line 2553  for (;;)
2553      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2554        {        {
2555        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2556        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2557        }        }
2558      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2559        {        {
2560          const pcre_uint32 *cp;
2561        const ucd_record *prop = GET_UCD(c);        const ucd_record *prop = GET_UCD(c);
2562    
2563        switch(ecode[1])        switch(ecode[1])
2564          {          {
2565          case PT_ANY:          case PT_ANY:
2566          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2567          break;          break;
2568    
2569          case PT_LAMP:          case PT_LAMP:
2570          if ((prop->chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2571               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2572               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2573            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2574          break;          break;
2575    
2576          case PT_GC:          case PT_GC:
2577          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2578            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2579          break;          break;
2580    
2581          case PT_PC:          case PT_PC:
2582          if ((ecode[2] != prop->chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2583            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2584          break;          break;
2585    
2586          case PT_SC:          case PT_SC:
2587          if ((ecode[2] != prop->script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2588            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2589          break;          break;
2590    
2591          /* These are specials */          /* These are specials */
2592    
2593          case PT_ALNUM:          case PT_ALNUM:
2594          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2595               _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))               PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2596            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2597          break;          break;
2598    
2599          case PT_SPACE:    /* Perl space */          case PT_SPACE:    /* Perl space */
2600          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2601               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2602                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2603            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2604          break;          break;
2605    
2606          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2607          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2608               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2609               c == CHAR_FF || c == CHAR_CR)               c == CHAR_FF || c == CHAR_CR)
2610                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2611            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2612          break;          break;
2613    
2614          case PT_WORD:          case PT_WORD:
2615          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2616               _pcre_ucp_gentype[prop->chartype] == ucp_N ||               PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2617               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2618            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2619            break;
2620    
2621            case PT_CLIST:
2622            cp = PRIV(ucd_caseless_sets) + ecode[2];
2623            for (;;)
2624              {
2625              if (c < *cp)
2626                { if (op == OP_PROP) { RRETURN(MATCH_NOMATCH); } else break; }
2627              if (c == *cp++)
2628                { if (op == OP_PROP) break; else { RRETURN(MATCH_NOMATCH); } }
2629              }
2630          break;          break;
2631    
2632            case PT_UCNC:
2633            if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
2634                 c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
2635                 c >= 0xe000) == (op == OP_NOTPROP))
2636              RRETURN(MATCH_NOMATCH);
2637            break;
2638    
2639          /* This should never occur */          /* This should never occur */
2640    
# Line 2380  for (;;) Line 2653  for (;;)
2653      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2654        {        {
2655        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2656        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2657        }        }
2658      GETCHARINCTEST(c, eptr);      else
2659        {        {
2660        int category = UCD_CATEGORY(c);        int lgb, rgb;
2661        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        GETCHARINCTEST(c, eptr);
2662          lgb = UCD_GRAPHBREAK(c);
2663        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2664          {          {
2665          int len = 1;          int len = 1;
2666          if (!utf8) c = *eptr; else          if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2667            {          rgb = UCD_GRAPHBREAK(c);
2668            GETCHARLEN(c, eptr, len);          if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
2669            }          lgb = rgb;
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
2670          eptr += len;          eptr += len;
2671          }          }
2672        }        }
2673        CHECK_PARTIAL();
2674      ecode++;      ecode++;
2675      break;      break;
2676  #endif  #endif  /* SUPPORT_UCP */
2677    
2678    
2679      /* 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 2685  for (;;)
2685      loops). */      loops). */
2686    
2687      case OP_REF:      case OP_REF:
2688      case OP_REFI:      case OP_REFI:
2689      caseless = op == OP_REFI;      caseless = op == OP_REFI;
2690      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2691      ecode += 3;      ecode += 1 + IMM2_SIZE;
2692    
2693      /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2694    
# Line 2455  for (;;) Line 2728  for (;;)
2728        case OP_CRMINRANGE:        case OP_CRMINRANGE:
2729        minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2730        min = GET2(ecode, 1);        min = GET2(ecode, 1);
2731        max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2732        if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2733        ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2734        break;        break;
2735    
2736        default:               /* No repeat follows */        default:               /* No repeat follows */
2737        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2738          {          {
2739            if (length == -2) eptr = md->end_subject;   /* Partial match */
2740          CHECK_PARTIAL();          CHECK_PARTIAL();
2741          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2742          }          }
2743        eptr += length;        eptr += length;
2744        continue;              /* With the main loop */        continue;              /* With the main loop */
2745        }        }
2746    
2747      /* Handle repeated back references. If the length of the reference is      /* Handle repeated back references. If the length of the reference is
2748      zero, just continue with the main loop. */      zero, just continue with the main loop. If the length is negative, it
2749        means the reference is unset in non-Java-compatible mode. If the minimum is
2750        zero, we can continue at the same level without recursion. For any other
2751        minimum, carrying on will result in NOMATCH. */
2752    
2753      if (length == 0) continue;      if (length == 0) continue;
2754        if (length < 0 && min == 0) continue;
2755    
2756      /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2757      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 2759  for (;;)
2759    
2760      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2761        {        {
2762        int slength;        int slength;
2763        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2764          {          {
2765            if (slength == -2) eptr = md->end_subject;   /* Partial match */
2766          CHECK_PARTIAL();          CHECK_PARTIAL();
2767          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2768          }          }
2769        eptr += slength;        eptr += slength;
2770        }        }
# Line 2501  for (;;) Line 2780  for (;;)
2780        {        {
2781        for (fi = min;; fi++)        for (fi = min;; fi++)
2782          {          {
2783          int slength;          int slength;
2784          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2785          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2786          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) RRETURN(MATCH_NOMATCH);
2787          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2788            {            {
2789              if (slength == -2) eptr = md->end_subject;   /* Partial match */
2790            CHECK_PARTIAL();            CHECK_PARTIAL();
2791            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2792            }            }
2793          eptr += slength;          eptr += slength;
2794          }          }
# Line 2522  for (;;) Line 2802  for (;;)
2802        pp = eptr;        pp = eptr;
2803        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2804          {          {
2805          int slength;          int slength;
2806          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2807            {            {
2808            CHECK_PARTIAL();            /* Can't use CHECK_PARTIAL because we don't want to update eptr in
2809              the soft partial matching case. */
2810    
2811              if (slength == -2 && md->partial != 0 &&
2812                  md->end_subject > md->start_used_ptr)
2813                {
2814                md->hitend = TRUE;
2815                if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);
2816                }
2817            break;            break;
2818            }            }
2819          eptr += slength;          eptr += slength;
2820          }          }
2821    
2822        while (eptr >= pp)        while (eptr >= pp)
2823          {          {
2824          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2825          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2826          eptr -= length;          eptr -= length;
2827          }          }
2828        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2829        }        }
2830      /* Control never gets here */      /* Control never gets here */
2831    
# Line 2554  for (;;) Line 2843  for (;;)
2843      case OP_NCLASS:      case OP_NCLASS:
2844      case OP_CLASS:      case OP_CLASS:
2845        {        {
2846          /* The data variable is saved across frames, so the byte map needs to
2847          be stored there. */
2848    #define BYTE_MAP ((pcre_uint8 *)data)
2849        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2850        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2851    
2852        switch (*ecode)        switch (*ecode)
2853          {          {
# Line 2576  for (;;) Line 2868  for (;;)
2868          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2869          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2870          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2871          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2872          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2873          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2874          break;          break;
2875    
2876          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2588  for (;;) Line 2880  for (;;)
2880    
2881        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2882    
2883  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2884        /* UTF-8 mode */        if (utf)
       if (utf8)  
2885          {          {
2886          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2887            {            {
2888            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2889              {              {
2890              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2891              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2892              }              }
2893            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2894            if (c > 255)            if (c > 255)
2895              {              {
2896              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2897              }              }
2898            else            else
2899              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
             }  
2900            }            }
2901          }          }
2902        else        else
2903  #endif  #endif
2904        /* Not UTF-8 mode */        /* Not UTF mode */
2905          {          {
2906          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2907            {            {
2908            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2909              {              {
2910              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2911              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2912              }              }
2913            c = *eptr++;            c = *eptr++;
2914            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2915              if (c > 255)
2916                {
2917                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2918                }
2919              else
2920    #endif
2921                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2922            }            }
2923          }          }
2924    
# Line 2636  for (;;) Line 2932  for (;;)
2932    
2933        if (minimize)        if (minimize)
2934          {          {
2935  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2936          /* UTF-8 mode */          if (utf)
         if (utf8)  
2937            {            {
2938            for (fi = min;; fi++)            for (fi = min;; fi++)
2939              {              {
2940              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2941              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2942              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2943              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2944                {                {
2945                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2946                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2947                }                }
2948              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2949              if (c > 255)              if (c > 255)
2950                {                {
2951                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2952                }                }
2953              else              else
2954                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
               }  
2955              }              }
2956            }            }
2957          else          else
2958  #endif  #endif
2959          /* Not UTF-8 mode */          /* Not UTF mode */
2960            {            {
2961            for (fi = min;; fi++)            for (fi = min;; fi++)
2962              {              {
2963              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2964              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2965              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2966              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2967                {                {
2968                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2969                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2970                }                }
2971              c = *eptr++;              c = *eptr++;
2972              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2973                if (c > 255)
2974                  {
2975                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2976                  }
2977                else
2978    #endif
2979                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2980              }              }
2981            }            }
2982          /* Control never gets here */          /* Control never gets here */
# Line 2688  for (;;) Line 2988  for (;;)
2988          {          {
2989          pp = eptr;          pp = eptr;
2990    
2991  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2992          /* UTF-8 mode */          if (utf)
         if (utf8)  
2993            {            {
2994            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2995              {              {
# Line 2706  for (;;) Line 3005  for (;;)
3005                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
3006                }                }
3007              else              else
3008                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
3009              eptr += len;              eptr += len;
3010              }              }
3011            for (;;)            for (;;)
# Line 2721  for (;;) Line 3018  for (;;)
3018            }            }
3019          else          else
3020  #endif  #endif
3021            /* Not UTF-8 mode */            /* Not UTF mode */
3022            {            {
3023            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3024              {              {
# Line 2731  for (;;) Line 3028  for (;;)
3028                break;                break;
3029                }                }
3030              c = *eptr;              c = *eptr;
3031              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
3032                if (c > 255)
3033                  {
3034                  if (op == OP_CLASS) break;
3035                  }
3036                else
3037    #endif
3038                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3039              eptr++;              eptr++;
3040              }              }
3041            while (eptr >= pp)            while (eptr >= pp)
# Line 2742  for (;;) Line 3046  for (;;)
3046              }              }
3047            }            }
3048    
3049          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3050          }          }
3051    #undef BYTE_MAP
3052        }        }
3053      /* Control never gets here */      /* Control never gets here */
3054    
# Line 2752  for (;;) Line 3057  for (;;)
3057      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
3058      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
3059    
3060  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3061      case OP_XCLASS:      case OP_XCLASS:
3062        {        {
3063        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2777  for (;;) Line 3082  for (;;)
3082          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3083          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3084          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3085          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
3086          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
3087          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
3088          break;          break;
3089    
3090          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2794  for (;;) Line 3099  for (;;)
3099          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3100            {            {
3101            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3102            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3103            }            }
3104          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
3105          if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3106          }          }
3107    
3108        /* 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 3119  for (;;)
3119            {            {
3120            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
3121            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3122            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3123            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3124              {              {
3125              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3126              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3127              }              }
3128            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3129            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
3130            }            }
3131          /* Control never gets here */          /* Control never gets here */
3132          }          }
# Line 2839  for (;;) Line 3144  for (;;)
3144              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3145              break;              break;
3146              }              }
3147    #ifdef SUPPORT_UTF
3148            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3149            if (!_pcre_xclass(c, data)) break;  #else
3150              c = *eptr;
3151    #endif
3152              if (!PRIV(xclass)(c, data, utf)) break;
3153            eptr += len;            eptr += len;
3154            }            }
3155          for(;;)          for(;;)
# Line 2848  for (;;) Line 3157  for (;;)
3157            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3158            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3159            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3160            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3161              if (utf) BACKCHAR(eptr);
3162    #endif
3163            }            }
3164          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3165          }          }
3166    
3167        /* Control never gets here */        /* Control never gets here */
# Line 2860  for (;;) Line 3171  for (;;)
3171      /* Match a single character, casefully */      /* Match a single character, casefully */
3172    
3173      case OP_CHAR:      case OP_CHAR:
3174  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3175      if (utf8)      if (utf)
3176        {        {
3177        length = 1;        length = 1;
3178        ecode++;        ecode++;
# Line 2869  for (;;) Line 3180  for (;;)
3180        if (length > md->end_subject - eptr)        if (length > md->end_subject - eptr)
3181          {          {
3182          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
3183          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3184          }          }
3185        while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);        while (length-- > 0) if (*ecode++ != RAWUCHARINC(eptr)) RRETURN(MATCH_NOMATCH);
3186        }        }
3187      else      else
3188  #endif  #endif
3189        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3190        {        {
3191        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3192          {          {
3193          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
3194          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3195          }          }
3196        if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
3197        ecode += 2;        ecode += 2;
3198        }        }
3199      break;      break;
3200    
3201      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3202        subject, give up immediately. */
3203    
3204      case OP_CHARI:      case OP_CHARI:
3205  #ifdef SUPPORT_UTF8      if (eptr >= md->end_subject)
3206      if (utf8)        {
3207          SCHECK_PARTIAL();
3208          RRETURN(MATCH_NOMATCH);
3209          }
3210    
3211    #ifdef SUPPORT_UTF
3212        if (utf)
3213        {        {
3214        length = 1;        length = 1;
3215        ecode++;        ecode++;
3216        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3217    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         MRRETURN(MATCH_NOMATCH);  
         }  
   
3218        /* 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
3219        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3220          fast lookup table. We know that there is at least one byte left in the
3221          subject. */
3222    
3223        if (fc < 128)        if (fc < 128)
3224          {          {
3225          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          pcre_uint32 cc = RAWUCHAR(eptr);
3226            if (md->lcc[fc] != TABLE_GET(cc, md->lcc, cc)) RRETURN(MATCH_NOMATCH);
3227            ecode++;
3228            eptr++;
3229          }          }
3230    
3231        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3232          use the value of "length" to check for sufficient bytes left, because the
3233          other case of the character may have more or fewer bytes.  */
3234    
3235        else        else
3236          {          {
3237          unsigned int dc;          pcre_uint32 dc;
3238          GETCHARINC(dc, eptr);          GETCHARINC(dc, eptr);
3239          ecode += length;          ecode += length;
3240    
# Line 2928  for (;;) Line 3246  for (;;)
3246  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3247            if (dc != UCD_OTHERCASE(fc))            if (dc != UCD_OTHERCASE(fc))
3248  #endif  #endif
3249              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3250            }            }
3251          }          }
3252        }        }
3253      else      else
3254  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3255    
3256      /* Non-UTF-8 mode */      /* Not UTF mode */
3257        {        {
3258        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3259          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3260          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         MRRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);  
3261        ecode += 2;        ecode += 2;
3262        }        }
3263      break;      break;
# Line 2952  for (;;) Line 3267  for (;;)
3267      case OP_EXACT:      case OP_EXACT:
3268      case OP_EXACTI:      case OP_EXACTI:
3269      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3270      ecode += 3;      ecode += 1 + IMM2_SIZE;
3271      goto REPEATCHAR;      goto REPEATCHAR;
3272    
3273      case OP_POSUPTO:      case OP_POSUPTO:
# Line 2967  for (;;) Line 3282  for (;;)
3282      min = 0;      min = 0;
3283      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3284      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3285      ecode += 3;      ecode += 1 + IMM2_SIZE;
3286      goto REPEATCHAR;      goto REPEATCHAR;
3287    
3288      case OP_POSSTAR:      case OP_POSSTAR:
# Line 3015  for (;;) Line 3330  for (;;)
3330      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3331    
3332      REPEATCHAR:      REPEATCHAR:
3333  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3334      if (utf8)      if (utf)
3335        {        {
3336        length = 1;        length = 1;
3337        charptr = ecode;        charptr = ecode;
# Line 3029  for (;;) Line 3344  for (;;)
3344        if (length > 1)        if (length > 1)
3345          {          {
3346  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3347          unsigned int othercase;          pcre_uint32 othercase;
3348          if (op >= OP_STARI &&     /* Caseless */          if (op >= OP_STARI &&     /* Caseless */
3349              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3350            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3351          else oclength = 0;          else oclength = 0;
3352  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3353    
3354          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3355            {            {
3356            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3357              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3358  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3359            else if (oclength > 0 &&            else if (oclength > 0 &&
3360                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3361                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3362  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3363            else            else
3364              {              {
3365              CHECK_PARTIAL();              CHECK_PARTIAL();
3366              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3367              }              }
3368            }            }
3369    
# Line 3060  for (;;) Line 3375  for (;;)
3375              {              {
3376              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3377              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3378              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3379              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3380                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3381  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3382              else if (oclength > 0 &&              else if (oclength > 0 &&
3383                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3384                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3385  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3386              else              else
3387                {                {
3388                CHECK_PARTIAL();                CHECK_PARTIAL();
3389                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3390                }                }
3391              }              }
3392            /* Control never gets here */            /* Control never gets here */
# Line 3083  for (;;) Line 3398  for (;;)
3398            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3399              {              {
3400              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3401                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3402  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3403              else if (oclength > 0 &&              else if (oclength > 0 &&
3404                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3405                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3406  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3407              else              else
3408                {                {
# Line 3102  for (;;) Line 3417  for (;;)
3417              {              {
3418              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3419              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3420              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3421  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3422              eptr--;              eptr--;
3423              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3119  for (;;) Line 3434  for (;;)
3434        value of fc will always be < 128. */        value of fc will always be < 128. */
3435        }        }
3436      else      else
3437  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3438          /* When not in UTF-8 mode, load a single-byte character. */
3439      /* When not in UTF-8 mode, load a single-byte character. */        fc = *ecode++;
   
     fc = *ecode++;  
3440    
3441      /* 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
3442      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
3443      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3444      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3445      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 3135  for (;;) Line 3448  for (;;)
3448      maximizing, find the maximum number of characters and work backwards. */      maximizing, find the maximum number of characters and work backwards. */
3449    
3450      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3451        max, eptr));        max, (char *)eptr));
3452    
3453      if (op >= OP_STARI)  /* Caseless */      if (op >= OP_STARI)  /* Caseless */
3454        {        {
3455        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3456          /* fc must be < 128 if UTF is enabled. */
3457          foc = md->fcc[fc];
3458    #else
3459    #ifdef SUPPORT_UTF
3460    #ifdef SUPPORT_UCP
3461          if (utf && fc > 127)
3462            foc = UCD_OTHERCASE(fc);
3463    #else
3464          if (utf && fc > 127)
3465            foc = fc;
3466    #endif /* SUPPORT_UCP */
3467          else
3468    #endif /* SUPPORT_UTF */
3469            foc = TABLE_GET(fc, md->fcc, fc);
3470    #endif /* COMPILE_PCRE8 */
3471    
3472        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3473          {          {
3474            pcre_uint32 cc;                 /* Faster than pcre_uchar */
3475          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3476            {            {
3477            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3478            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3479            }            }
3480          if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          cc = RAWUCHARTEST(eptr);
3481            if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH);
3482            eptr++;
3483          }          }
3484        if (min == max) continue;        if (min == max) continue;
3485        if (minimize)        if (minimize)
3486          {          {
3487          for (fi = min;; fi++)          for (fi = min;; fi++)
3488            {            {
3489              pcre_uint32 cc;               /* Faster than pcre_uchar */
3490            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3491            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3492            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3493            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3494              {              {
3495              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3496              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3497              }              }
3498            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            cc = RAWUCHARTEST(eptr);
3499              if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH);
3500              eptr++;
3501            }            }
3502          /* Control never gets here */          /* Control never gets here */
3503          }          }
# Line 3171  for (;;) Line 3506  for (;;)
3506          pp = eptr;          pp = eptr;
3507          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3508            {            {
3509              pcre_uint32 cc;               /* Faster than pcre_uchar */
3510            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3511              {              {
3512              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3513              break;              break;
3514              }              }
3515            if (fc != md->lcc[*eptr]) break;            cc = RAWUCHARTEST(eptr);
3516              if (fc != cc && foc != cc) break;
3517            eptr++;            eptr++;
3518            }            }
3519    
# Line 3188  for (;;) Line 3525  for (;;)
3525            eptr--;            eptr--;
3526            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3527            }            }
3528          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3529          }          }
3530        /* Control never gets here */        /* Control never gets here */
3531        }        }
# Line 3202  for (;;) Line 3539  for (;;)
3539          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3540            {            {
3541            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3542            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3543            }            }
3544          if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);          if (fc != RAWUCHARINCTEST(eptr)) RRETURN(MATCH_NOMATCH);
3545          }          }
3546    
3547        if (min == max) continue;        if (min == max) continue;
# Line 3215  for (;;) Line 3552  for (;;)
3552            {            {
3553            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3554            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3555            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3556            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3557              {              {
3558              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3559              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3560              }              }
3561            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);            if (fc != RAWUCHARINCTEST(eptr)) RRETURN(MATCH_NOMATCH);
3562            }            }
3563          /* Control never gets here */          /* Control never gets here */
3564          }          }
# Line 3235  for (;;) Line 3572  for (;;)
3572              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3573              break;              break;
3574              }              }
3575            if (fc != *eptr) break;            if (fc != RAWUCHARTEST(eptr)) break;
3576            eptr++;            eptr++;
3577            }            }
3578          if (possessive) continue;          if (possessive) continue;
# Line 3246  for (;;) Line 3583  for (;;)
3583            eptr--;            eptr--;
3584            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3585            }            }
3586          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3587          }          }
3588        }        }
3589      /* Control never gets here */      /* Control never gets here */
# Line 3255  for (;;) Line 3592  for (;;)
3592      checking can be multibyte. */      checking can be multibyte. */
3593    
3594      case OP_NOT:      case OP_NOT:
3595      case OP_NOTI:      case OP_NOTI:
3596      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3597        {        {
3598        SCHECK_PARTIAL();        SCHECK_PARTIAL();
3599        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
3600        }        }
3601      ecode++;  #ifdef SUPPORT_UTF
3602      GETCHARINCTEST(c, eptr);      if (utf)
     if (op == OP_NOTI)         /* The caseless case */  
3603        {        {
3604  #ifdef SUPPORT_UTF8        register pcre_uint32 ch, och;
3605        if (c < 256)  
3606  #endif        ecode++;
3607        c = md->lcc[c];        GETCHARINC(ch, ecode);
3608        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        GETCHARINC(c, eptr);
3609    
3610          if (op == OP_NOT)
3611            {
3612            if (ch == c) RRETURN(MATCH_NOMATCH);
3613            }
3614          else
3615            {
3616    #ifdef SUPPORT_UCP
3617            if (ch > 127)
3618              och = UCD_OTHERCASE(ch);
3619    #else
3620            if (ch > 127)
3621              och = ch;
3622    #endif /* SUPPORT_UCP */
3623            else
3624              och = TABLE_GET(ch, md->fcc, ch);
3625            if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3626            }
3627        }        }
3628      else    /* Caseful */      else
3629    #endif
3630        {        {
3631        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        register pcre_uint32 ch = ecode[1];
3632          c = *eptr++;
3633          if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c))
3634            RRETURN(MATCH_NOMATCH);
3635          ecode += 2;
3636        }        }
3637      break;      break;
3638    
# Line 3287  for (;;) Line 3646  for (;;)
3646      case OP_NOTEXACT:      case OP_NOTEXACT:
3647      case OP_NOTEXACTI:      case OP_NOTEXACTI:
3648      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3649      ecode += 3;      ecode += 1 + IMM2_SIZE;
3650      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3651    
3652      case OP_NOTUPTO:      case OP_NOTUPTO:
# Line 3297  for (;;) Line 3656  for (;;)
3656      min = 0;      min = 0;
3657      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3658      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3659      ecode += 3;      ecode += 1 + IMM2_SIZE;
3660      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3661    
3662      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
# Line 3329  for (;;) Line 3688  for (;;)
3688      possessive = TRUE;      possessive = TRUE;
3689      min = 0;      min = 0;
3690      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3691      ecode += 3;      ecode += 1 + IMM2_SIZE;
3692      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3693    
3694      case OP_NOTSTAR:      case OP_NOTSTAR:
# Line 3353  for (;;) Line 3712  for (;;)
3712      /* Common code for all repeated single-byte matches. */      /* Common code for all repeated single-byte matches. */
3713    
3714      REPEATNOTCHAR:      REPEATNOTCHAR:
3715      fc = *ecode++;      GETCHARINCTEST(fc, ecode);
3716    
3717      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
3718      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 3723  for (;;)
3723      characters and work backwards. */      characters and work backwards. */
3724    
3725      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,
3726        max, eptr));        max, (char *)eptr));
3727    
3728      if (op >= OP_NOTSTARI)     /* Caseless */      if (op >= OP_NOTSTARI)     /* Caseless */
3729        {        {
3730        fc = md->lcc[fc];  #ifdef SUPPORT_UTF
3731    #ifdef SUPPORT_UCP
3732          if (utf && fc > 127)
3733            foc = UCD_OTHERCASE(fc);
3734    #else
3735          if (utf && fc > 127)
3736            foc = fc;
3737    #endif /* SUPPORT_UCP */
3738          else
3739    #endif /* SUPPORT_UTF */
3740            foc = TABLE_GET(fc, md->fcc, fc);
3741    
3742  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3743        /* UTF-8 mode */        if (utf)
       if (utf8)  
3744          {          {
3745          register unsigned int d;          register pcre_uint32 d;
3746          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3747            {            {
3748            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3749              {              {
3750              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3751              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3752              }              }
3753            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3754            if (d < 256) d = md->lcc[d];            if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) MRRETURN(MATCH_NOMATCH);  
3755            }            }
3756          }          }
3757        else        else
3758  #endif  #endif
3759          /* Not UTF mode */
       /* Not UTF-8 mode */  
3760          {          {
3761          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3762            {            {
3763            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3764              {              {
3765              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3766              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3767              }              }
3768            if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3769              eptr++;
3770            }            }
3771          }          }
3772    
# Line 3407  for (;;) Line 3774  for (;;)
3774    
3775        if (minimize)        if (minimize)
3776          {          {
3777  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3778          /* UTF-8 mode */          if (utf)
         if (utf8)  
3779            {            {
3780            register unsigned int d;            register pcre_uint32 d;
3781            for (fi = min;; fi++)            for (fi = min;; fi++)
3782              {              {
3783              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3784              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3785              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3786              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3787                {                {
3788                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3789                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3790                }                }
3791              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3792              if (d < 256) d = md->lcc[d];              if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH);
             if (fc == d) MRRETURN(MATCH_NOMATCH);  
3793              }              }
3794            }            }
3795          else          else
3796  #endif  #endif
3797          /* Not UTF-8 mode */          /* Not UTF mode */
3798            {            {
3799            for (fi = min;; fi++)            for (fi = min;; fi++)
3800              {              {