/[pcre]/code/branches/pcre16/pcre_exec.c
ViewVC logotype

Diff of /code/branches/pcre16/pcre_exec.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

code/trunk/pcre_exec.c revision 597 by ph10, Mon May 2 17:08:52 2011 UTC code/branches/pcre16/pcre_exec.c revision 801 by ph10, Mon Dec 12 16:23:37 2011 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2010 University of Cambridge             Copyright (c) 1997-2011 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 57  possible. There are also some static sup Line 57  possible. There are also some static sup
57  #undef min  #undef min
58  #undef max  #undef max
59    
60  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
61    of call to match(). We do it this way to save on using another stack variable,
62    as stack usage is to be discouraged. */
63    
64  #define match_condassert     0x01  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
65  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
66    
67  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
68  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 73  negative to avoid the external error cod Line 75  negative to avoid the external error cod
75    
76  #define MATCH_ACCEPT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_COMMIT       (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_PRUNE        (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_SKIP         (-996)  #define MATCH_ONCE         (-996)
80  #define MATCH_SKIP_ARG     (-995)  #define MATCH_PRUNE        (-995)
81  #define MATCH_THEN         (-994)  #define MATCH_SKIP         (-994)
82    #define MATCH_SKIP_ARG     (-993)
83  /* This is a convenience macro for code that occurs many times. */  #define MATCH_THEN         (-992)
   
 #define MRRETURN(ra) \  
   { \  
   md->mark = markptr; \  
   RRETURN(ra); \  
   }  
84    
85  /* Maximum number of ints of offset to save on the stack for recursive calls.  /* Maximum number of ints of offset to save on the stack for recursive calls.
86  If the offset vector is bigger, malloc is used. This should be a multiple of 3,  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
# Line 117  Returns:     nothing Line 113  Returns:     nothing
113  */  */
114    
115  static void  static void
116  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)  pchars(const pcre_uchar *p, int length, BOOL is_subject, match_data *md)
117  {  {
118  unsigned int c;  unsigned int c;
119  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;
# Line 134  while (length-- > 0) Line 130  while (length-- > 0)
130    
131  /* 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
132  negative, so the match always fails. However, in JavaScript compatibility mode,  negative, so the match always fails. However, in JavaScript compatibility mode,
133  the length passed is zero. Note that in caseless UTF-8 mode, the number of  the length passed is zero. Note that in caseless UTF-8 mode, the number of
134  subject bytes matched may be different to the number of reference bytes.  subject bytes matched may be different to the number of reference bytes.
135    
136  Arguments:  Arguments:
# Line 142  Arguments: Line 138  Arguments:
138    eptr        pointer into the subject    eptr        pointer into the subject
139    length      length of reference to be matched (number of bytes)    length      length of reference to be matched (number of bytes)
140    md          points to match data block    md          points to match data block
141    ims         the ims flags    caseless    TRUE if caseless
142    
143  Returns:      < 0 if not matched, otherwise the number of subject bytes matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
144  */  */
145    
146  static int  static int
147  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,
148    unsigned long int ims)    BOOL caseless)
149  {  {
150  USPTR eptr_start = eptr;  PCRE_PUCHAR eptr_start = eptr;
151  register USPTR p = md->start_subject + md->offset_vector[offset];  register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
152    
153  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
154  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 175  if (length < 0) return -1; Line 171  if (length < 0) return -1;
171  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
172  ASCII characters. */  ASCII characters. */
173    
174  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
175    {    {
176  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
177  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
178    if (md->utf8)    if (md->utf)
179      {      {
180      /* 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
181      bytes matched may differ, because there are some characters whose upper and      bytes matched may differ, because there are some characters whose upper and
182      lower case versions code as different numbers of bytes. For example, U+023A      lower case versions code as different numbers of bytes. For example, U+023A
183      (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);      (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
184      a sequence of 3 of the former uses 6 bytes, as does a sequence of two of      a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
185      the latter. It is important, therefore, to check the length along the      the latter. It is important, therefore, to check the length along the
186      reference, not along the subject (earlier code did this wrong). */      reference, not along the subject (earlier code did this wrong). */
187    
188      USPTR endptr = p + length;      PCRE_PUCHAR endptr = p + length;
189      while (p < endptr)      while (p < endptr)
190        {        {
191        int c, d;        int c, d;
# Line 206  if ((ims & PCRE_CASELESS) != 0) Line 202  if ((ims & PCRE_CASELESS) != 0)
202    /* 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
203    is no UCP support. */    is no UCP support. */
204      {      {
205      if (eptr + length > md->end_subject) return -1;      if (eptr + length > md->end_subject) return -1;
206      while (length-- > 0)      while (length-- > 0)
207        { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }        { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
208      }      }
209    }    }
210    
211  /* 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
212  are in UTF-8 mode. */  are in UTF-8 mode. */
213    
214  else  else
215    {    {
216    if (eptr + length > md->end_subject) return -1;    if (eptr + length > md->end_subject) return -1;
217    while (length-- > 0) if (*p++ != *eptr++) return -1;    while (length-- > 0) if (*p++ != *eptr++) return -1;
218    }    }
219    
220  return eptr - eptr_start;  return (int)(eptr - eptr_start);
221  }  }
222    
223    
# Line 273  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 269  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
269         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
270         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
271         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
272         RM61,  RM62 };         RM61,  RM62, RM63, RM64, RM65, RM66 };
273    
274  /* 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
275  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 283  actually used in this definition. */ Line 279  actually used in this definition. */
279  #define REGISTER register  #define REGISTER register
280    
281  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
282  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
283    { \    { \
284    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
285    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \
286    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
287    }    }
288  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 295  actually used in this definition. */ Line 291  actually used in this definition. */
291    return ra; \    return ra; \
292    }    }
293  #else  #else
294  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
295    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1)
296  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
297  #endif  #endif
298    
# Line 309  argument of match(), which never changes Line 305  argument of match(), which never changes
305    
306  #define REGISTER  #define REGISTER
307    
308  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
309    {\    {\
310    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
311    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
# Line 317  argument of match(), which never changes Line 313  argument of match(), which never changes
313    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
314    newframe->Xecode = rb;\    newframe->Xecode = rb;\
315    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
   newframe->Xmarkptr = markptr;\  
316    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
317    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
318    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
319    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
320    frame = newframe;\    frame = newframe;\
# Line 352  typedef struct heapframe { Line 345  typedef struct heapframe {
345    
346    /* Function arguments that may change */    /* Function arguments that may change */
347    
348    USPTR Xeptr;    PCRE_PUCHAR Xeptr;
349    const uschar *Xecode;    const pcre_uchar *Xecode;
350    USPTR Xmstart;    PCRE_PUCHAR Xmstart;
   USPTR Xmarkptr;  
351    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
352    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
353    unsigned int Xrdepth;    unsigned int Xrdepth;
354    
355    /* Function local variables */    /* Function local variables */
356    
357    USPTR Xcallpat;    PCRE_PUCHAR Xcallpat;
358  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
359    USPTR Xcharptr;    PCRE_PUCHAR Xcharptr;
360  #endif  #endif
361    USPTR Xdata;    PCRE_PUCHAR Xdata;
362    USPTR Xnext;    PCRE_PUCHAR Xnext;
363    USPTR Xpp;    PCRE_PUCHAR Xpp;
364    USPTR Xprev;    PCRE_PUCHAR Xprev;
365    USPTR Xsaved_eptr;    PCRE_PUCHAR Xsaved_eptr;
366    
367    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
368    
# Line 380  typedef struct heapframe { Line 370  typedef struct heapframe {
370    BOOL Xcondition;    BOOL Xcondition;
371    BOOL Xprev_is_word;    BOOL Xprev_is_word;
372    
   unsigned long int Xoriginal_ims;  
   
373  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
374    int Xprop_type;    int Xprop_type;
375    int Xprop_value;    int Xprop_value;
376    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
377    int Xoclength;    int Xoclength;
378    uschar Xocchars[8];    pcre_uchar Xocchars[6];
379  #endif  #endif
380    
381    int Xcodelink;    int Xcodelink;
# Line 432  returns a negative (error) response, the Line 417  returns a negative (error) response, the
417  same response. */  same response. */
418    
419  /* 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
420  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
421  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.
422  something has been matched). For hard partial matching, we then return  something has been matched). For hard partial matching, we then return
423  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 443  the subject. */ Line 428  the subject. */
428        eptr > md->start_used_ptr) \        eptr > md->start_used_ptr) \
429      { \      { \
430      md->hitend = TRUE; \      md->hitend = TRUE; \
431      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
432      }      }
433    
434  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
435    if (md->partial != 0 && eptr > md->start_used_ptr) \    if (md->partial != 0 && eptr > md->start_used_ptr) \
436      { \      { \
437      md->hitend = TRUE; \      md->hitend = TRUE; \
438      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \      if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \
439      }      }
440    
441    
442  /* Performance note: It might be tempting to extract commonly used fields from  /* Performance note: It might be tempting to extract commonly used fields from
443  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
444  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
445  made performance worse.  made performance worse.
446    
# Line 464  Arguments: Line 449  Arguments:
449     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
450     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
451                   by encountering \K)                   by encountering \K)
    markptr     pointer to the most recent MARK name, or NULL  
452     offset_top  current top pointer     offset_top  current top pointer
453     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
454     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
455                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
456     rdepth      the recursion depth     rdepth      the recursion depth
457    
458  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
# Line 484  Returns:       MATCH_MATCH if matched Line 463  Returns:       MATCH_MATCH if matched
463  */  */
464    
465  static int  static int
466  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode,
467    const uschar *markptr, int offset_top, match_data *md, unsigned long int ims,    PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb,
468    eptrblock *eptrb, int flags, unsigned int rdepth)    unsigned int rdepth)
469  {  {
470  /* 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,
471  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 495  so they can be ordinary variables in all Line 474  so they can be ordinary variables in all
474  register int  rrc;         /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
475  register int  i;           /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
476  register unsigned int c;   /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
477  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf;         /* Local copy of UTF flag for speed */
478    
479  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
480    BOOL caseless;
481  int condcode;  int condcode;
482    
483  /* 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
# Line 515  frame->Xprevframe = NULL;            /* Line 495  frame->Xprevframe = NULL;            /*
495  frame->Xeptr = eptr;  frame->Xeptr = eptr;
496  frame->Xecode = ecode;  frame->Xecode = ecode;
497  frame->Xmstart = mstart;  frame->Xmstart = mstart;
 frame->Xmarkptr = markptr;  
498  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
499  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
500  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
501    
502  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 531  HEAP_RECURSE: Line 508  HEAP_RECURSE:
508  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
509  #define ecode              frame->Xecode  #define ecode              frame->Xecode
510  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
 #define markptr            frame->Xmarkptr  
511  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
512  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
513  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
514    
515  /* Ditto for the local variables */  /* Ditto for the local variables */
516    
517  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
518  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
519  #endif  #endif
520  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
# Line 557  HEAP_RECURSE: Line 531  HEAP_RECURSE:
531  #define condition          frame->Xcondition  #define condition          frame->Xcondition
532  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
533    
 #define original_ims       frame->Xoriginal_ims  
   
534  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
535  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
536  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
537  #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  
538  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
539  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
540  #endif  #endif
# Line 595  i, and fc and c, can be the same variabl Line 564  i, and fc and c, can be the same variabl
564  #define fi i  #define fi i
565  #define fc c  #define fc c
566    
567    /* Many of the following variables are used only in small blocks of the code.
568    My normal style of coding would have declared them within each of those blocks.
569    However, in order to accommodate the version of this code that uses an external
570    "stack" implemented on the heap, it is easier to declare them all here, so the
571    declarations can be cut out in a block. The only declarations within blocks
572    below are for variables that do not have to be preserved over a recursive call
573    to RMATCH(). */
574    
575    #ifdef SUPPORT_UTF
576    const pcre_uchar *charptr;
577    #endif
578    const pcre_uchar *callpat;
579    const pcre_uchar *data;
580    const pcre_uchar *next;
581    PCRE_PUCHAR       pp;
582    const pcre_uchar *prev;
583    PCRE_PUCHAR       saved_eptr;
584    
585    recursion_info new_recursive;
586    
587  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  BOOL cur_is_word;
 const uschar *charptr;             /* in small blocks of the code. My normal */  
 #endif                             /* style of coding would have declared    */  
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
588  BOOL condition;  BOOL condition;
589  BOOL prev_is_word;  BOOL prev_is_word;
590    
 unsigned long int original_ims;  
   
591  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
592  int prop_type;  int prop_type;
593  int prop_value;  int prop_value;
594  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
595  int oclength;  int oclength;
596  uschar occhars[8];  pcre_uchar occhars[6];
597  #endif  #endif
598    
599  int codelink;  int codelink;
# Line 640  int stacksave[REC_STACK_SAVE_MAX]; Line 611  int stacksave[REC_STACK_SAVE_MAX];
611  eptrblock newptrb;  eptrblock newptrb;
612  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
613    
614    /* To save space on the stack and in the heap frame, I have doubled up on some
615    of the local variables that are used only in localised parts of the code, but
616    still need to be preserved over recursive calls of match(). These macros define
617    the alternative names that are used. */
618    
619    #define allow_zero    cur_is_word
620    #define cbegroup      condition
621    #define code_offset   codelink
622    #define condassert    condition
623    #define matched_once  prev_is_word
624    #define foc           number
625    
626  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
627  variables. */  variables. */
628    
# Line 664  defined). However, RMATCH isn't like a f Line 647  defined). However, RMATCH isn't like a f
647  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,
648  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
649    
650  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
651  utf8 = md->utf8;       /* Local copy of the flag */  utf = md->utf;       /* Local copy of the flag */
652  #else  #else
653  utf8 = FALSE;  utf = FALSE;
654  #endif  #endif
655    
656  /* 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 676  haven't exceeded the recursive call limi Line 659  haven't exceeded the recursive call limi
659  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
660  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
661    
 original_ims = ims;    /* Save for resetting on ')' */  
   
662  /* 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
663  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
664  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
665  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
666  When match() is called in other circumstances, don't add to the chain. The  
667  match_cbegroup flag must NOT be used with tail recursion, because the memory  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
668  block that is used is on the stack, so a new one may be required for each  such remembered pointers, to be checked when we hit the closing ket, in order
669  match(). */  to break infinite loops that match no characters. When match() is called in
670    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
671    NOT be used with tail recursion, because the memory block that is used is on
672    the stack, so a new one may be required for each match(). */
673    
674  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
675    {    {
676    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
677    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
678    eptrb = &newptrb;    eptrb = &newptrb;
679      md->match_function_type = 0;
680    }    }
681    
682  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 704  for (;;) Line 689  for (;;)
689    switch(op)    switch(op)
690      {      {
691      case OP_MARK:      case OP_MARK:
692      markptr = ecode + 2;      md->nomatch_mark = ecode + 2;
693      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->mark = NULL;    /* In case previously set by assertion */
694        ims, eptrb, flags, RM55);      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
695          eptrb, RM55);
696        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
697             md->mark == NULL) md->mark = ecode + 2;
698    
699      /* 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
700      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 715  for (;;) Line 703  for (;;)
703      position and return MATCH_SKIP. Otherwise, pass back the return code      position and return MATCH_SKIP. Otherwise, pass back the return code
704      unaltered. */      unaltered. */
705    
706      if (rrc == MATCH_SKIP_ARG &&      else if (rrc == MATCH_SKIP_ARG &&
707          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)          STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0)
708        {        {
709        md->start_match_ptr = eptr;        md->start_match_ptr = eptr;
710        RRETURN(MATCH_SKIP);        RRETURN(MATCH_SKIP);
711        }        }
   
     if (md->mark == NULL) md->mark = markptr;  
712      RRETURN(rrc);      RRETURN(rrc);
713    
714      case OP_FAIL:      case OP_FAIL:
715      MRRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
716    
717      /* COMMIT overrides PRUNE, SKIP, and THEN */      /* COMMIT overrides PRUNE, SKIP, and THEN */
718    
719      case OP_COMMIT:      case OP_COMMIT:
720      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
721        ims, eptrb, flags, RM52);        eptrb, RM52);
722      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
723          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
724          rrc != MATCH_THEN)          rrc != MATCH_THEN)
725        RRETURN(rrc);        RRETURN(rrc);
726      MRRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
727    
728      /* PRUNE overrides THEN */      /* PRUNE overrides THEN */
729    
730      case OP_PRUNE:      case OP_PRUNE:
731      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
732        ims, eptrb, flags, RM51);        eptrb, RM51);
733      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
734      MRRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
735    
736      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
737      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      md->nomatch_mark = ecode + 2;
738        ims, eptrb, flags, RM56);      md->mark = NULL;    /* In case previously set by assertion */
739        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
740          eptrb, RM56);
741        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
742             md->mark == NULL) md->mark = ecode + 2;
743      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
     md->mark = ecode + 2;  
744      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
745    
746      /* SKIP overrides PRUNE and THEN */      /* SKIP overrides PRUNE and THEN */
747    
748      case OP_SKIP:      case OP_SKIP:
749      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
750        ims, eptrb, flags, RM53);        eptrb, RM53);
751      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
752        RRETURN(rrc);        RRETURN(rrc);
753      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
754      MRRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
755    
756        /* Note that, for Perl compatibility, SKIP with an argument does NOT set
757        nomatch_mark. There is a flag that disables this opcode when re-matching a
758        pattern that ended with a SKIP for which there was not a matching MARK. */
759    
760      case OP_SKIP_ARG:      case OP_SKIP_ARG:
761      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      if (md->ignore_skip_arg)
762        ims, eptrb, flags, RM57);        {
763          ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
764          break;
765          }
766        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
767          eptrb, RM57);
768      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
769        RRETURN(rrc);        RRETURN(rrc);
770    
771      /* 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
772      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
773      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
774      as PRUNE. */      with the md->ignore_skip_arg flag set. */
775    
776      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
777      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
778    
779      /* 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
780      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
781      to skip back past alternatives that precede the THEN within the current      match pointer to do this. */
     branch. */  
782    
783      case OP_THEN:      case OP_THEN:
784      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
785        ims, eptrb, flags, RM54);        eptrb, RM54);
786      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
787      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
788      MRRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
789    
790      case OP_THEN_ARG:      case OP_THEN_ARG:
791      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      md->nomatch_mark = ecode + 2;
792        offset_top, md, ims, eptrb, flags, RM58);      md->mark = NULL;    /* In case previously set by assertion */
793        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top,
794          md, eptrb, RM58);
795        if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
796             md->mark == NULL) md->mark = ecode + 2;
797      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
798      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode;
     md->mark = ecode + LINK_SIZE + 2;  
799      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
800    
801      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle an atomic group that does not contain any capturing parentheses.
802      the current subject position in the working slot at the top of the vector.      This can be handled like an assertion. Prior to 8.13, all atomic groups
803      We mustn't change the current values of the data slot, because they may be      were handled this way. In 8.13, the code was changed as below for ONCE, so
804      set from a previous iteration of this group, and be referred to by a      that backups pass through the group and thereby reset captured values.
805      reference inside the group.      However, this uses a lot more stack, so in 8.20, atomic groups that do not
806        contain any captures generate OP_ONCE_NC, which can be handled in the old,
807      If the bracket fails to match, we need to restore this value and also the      less stack intensive way.
808      values of the final offsets, in case they were set by a previous iteration  
809      of the same bracket.      Check the alternative branches in turn - the matching won't pass the KET
810        for this kind of subpattern. If any one branch matches, we carry on as at
811        the end of a normal bracket, leaving the subject pointer, but resetting
812        the start-of-match value in case it was changed by \K. */
813    
814        case OP_ONCE_NC:
815        prev = ecode;
816        saved_eptr = eptr;
817        do
818          {
819          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
820          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
821            {
822            mstart = md->start_match_ptr;
823            break;
824            }
825          if (rrc == MATCH_THEN)
826            {
827            next = ecode + GET(ecode,1);
828            if (md->start_match_ptr < next &&
829                (*ecode == OP_ALT || *next == OP_ALT))
830              rrc = MATCH_NOMATCH;
831            }
832    
833          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
834          ecode += GET(ecode,1);
835          }
836        while (*ecode == OP_ALT);
837    
838        /* If hit the end of the group (which could be repeated), fail */
839    
840        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
841    
842        /* Continue as from after the group, updating the offsets high water
843        mark, since extracts may have been taken. */
844    
845        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
846    
847        offset_top = md->end_offset_top;
848        eptr = md->end_match_ptr;
849    
850        /* For a non-repeating ket, just continue at this level. This also
851        happens for a repeating ket if no characters were matched in the group.
852        This is the forcible breaking of infinite loops as implemented in Perl
853        5.005. */
854    
855        if (*ecode == OP_KET || eptr == saved_eptr)
856          {
857          ecode += 1+LINK_SIZE;
858          break;
859          }
860    
861        /* The repeating kets try the rest of the pattern or restart from the
862        preceding bracket, in the appropriate order. The second "call" of match()
863        uses tail recursion, to avoid using another stack frame. */
864    
865        if (*ecode == OP_KETRMIN)
866          {
867          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
868          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
869          ecode = prev;
870          goto TAIL_RECURSE;
871          }
872        else  /* OP_KETRMAX */
873          {
874          md->match_function_type = MATCH_CBEGROUP;
875          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
876          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
877          ecode += 1 + LINK_SIZE;
878          goto TAIL_RECURSE;
879          }
880        /* Control never gets here */
881    
882        /* Handle a capturing bracket, other than those that are possessive with an
883        unlimited repeat. If there is space in the offset vector, save the current
884        subject position in the working slot at the top of the vector. We mustn't
885        change the current values of the data slot, because they may be set from a
886        previous iteration of this group, and be referred to by a reference inside
887        the group. A failure to match might occur after the group has succeeded,
888        if something later on doesn't match. For this reason, we need to restore
889        the working value and also the values of the final offsets, in case they
890        were set by a previous iteration of the same bracket.
891    
892      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
893      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 835  for (;;) Line 916  for (;;)
916        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
917          (int)(eptr - md->start_subject);          (int)(eptr - md->start_subject);
918    
919        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
920          {          {
921          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
922            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
923          if (rrc != MATCH_NOMATCH &&            eptrb, RM1);
924              (rrc != MATCH_THEN || md->start_match_ptr != ecode))          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
925            RRETURN(rrc);  
926            /* If we backed up to a THEN, check whether it is within the current
927            branch by comparing the address of the THEN that is passed back with
928            the end of the branch. If it is within the current branch, and the
929            branch is one of two or more alternatives (it either starts or ends
930            with OP_ALT), we have reached the limit of THEN's action, so convert
931            the return code to NOMATCH, which will cause normal backtracking to
932            happen from now on. Otherwise, THEN is passed back to an outer
933            alternative. This implements Perl's treatment of parenthesized groups,
934            where a group not containing | does not affect the current alternative,
935            that is, (X) is NOT the same as (X|(*F)). */
936    
937            if (rrc == MATCH_THEN)
938              {
939              next = ecode + GET(ecode,1);
940              if (md->start_match_ptr < next &&
941                  (*ecode == OP_ALT || *next == OP_ALT))
942                rrc = MATCH_NOMATCH;
943              }
944    
945            /* Anything other than NOMATCH is passed back. */
946    
947            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
948          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
949          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
950            if (*ecode != OP_ALT) break;
951          }          }
       while (*ecode == OP_ALT);  
952    
953        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
954        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
955        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
956        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
957    
958        if (rrc != MATCH_THEN) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
959        RRETURN(MATCH_NOMATCH);  
960          RRETURN(rrc);
961        }        }
962    
963      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 869  for (;;) Line 971  for (;;)
971      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
972      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
973    
974      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
975      final alternative within the brackets, we would return the result of a      repeat and ONCE group with no captures. Loop for all the alternatives.
     recursive call to match() whatever happened. We can reduce stack usage by  
     turning this into a tail recursion, except in the case when match_cbegroup  
     is set.*/  
976    
977        When we get to the final alternative within the brackets, we used to return
978        the result of a recursive call to match() whatever happened so it was
979        possible to reduce stack usage by turning this into a tail recursion,
980        except in the case of a possibly empty group. However, now that there is
981        the possiblity of (*THEN) occurring in the final alternative, this
982        optimization is no longer always possible.
983    
984        We can optimize if we know there are no (*THEN)s in the pattern; at present
985        this is the best that can be done.
986    
987        MATCH_ONCE is returned when the end of an atomic group is successfully
988        reached, but subsequent matching fails. It passes back up the tree (causing
989        captured values to be reset) until the original atomic group level is
990        reached. This is tested by comparing md->once_target with the start of the
991        group. At this point, the return is converted into MATCH_NOMATCH so that
992        previous backup points can be taken. */
993    
994        case OP_ONCE:
995      case OP_BRA:      case OP_BRA:
996      case OP_SBRA:      case OP_SBRA:
997      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
998      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
999      for (;;)      for (;;)
1000        {        {
1001        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
1002    
1003          /* If this is not a possibly empty group, and there are no (*THEN)s in
1004          the pattern, and this is the final alternative, optimize as described
1005          above. */
1006    
1007          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1008            {
1009            ecode += PRIV(OP_lengths)[*ecode];
1010            goto TAIL_RECURSE;
1011            }
1012    
1013          /* In all other cases, we have to make another call to match(). */
1014    
1015          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1016            RM2);
1017    
1018          /* See comment in the code for capturing groups above about handling
1019          THEN. */
1020    
1021          if (rrc == MATCH_THEN)
1022          {          {
1023          if (flags == 0)    /* Not a possibly empty group */          next = ecode + GET(ecode,1);
1024            if (md->start_match_ptr < next &&
1025                (*ecode == OP_ALT || *next == OP_ALT))
1026              rrc = MATCH_NOMATCH;
1027            }
1028    
1029          if (rrc != MATCH_NOMATCH)
1030            {
1031            if (rrc == MATCH_ONCE)
1032            {            {
1033            ecode += _pcre_OP_lengths[*ecode];            const pcre_uchar *scode = ecode;
1034            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1035            goto TAIL_RECURSE;              {
1036                while (*scode == OP_ALT) scode += GET(scode, 1);
1037                scode -= GET(scode, 1);
1038                }
1039              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1040            }            }
1041            RRETURN(rrc);
1042            }
1043          ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045          }
1046    
1047          /* Possibly empty group; can't use tail recursion. */      RRETURN(MATCH_NOMATCH);
1048    
1049          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* Handle possessive capturing brackets with an unlimited repeat. We come
1050            eptrb, flags, RM48);      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1051          if (rrc == MATCH_NOMATCH) md->mark = markptr;      handled similarly to the normal case above. However, the matching is
1052          RRETURN(rrc);      different. The end of these brackets will always be OP_KETRPOS, which
1053        returns MATCH_KETRPOS without going further in the pattern. By this means
1054        we can handle the group by iteration rather than recursion, thereby
1055        reducing the amount of stack needed. */
1056    
1057        case OP_CBRAPOS:
1058        case OP_SCBRAPOS:
1059        allow_zero = FALSE;
1060    
1061        POSSESSIVE_CAPTURE:
1062        number = GET2(ecode, 1+LINK_SIZE);
1063        offset = number << 1;
1064    
1065    #ifdef PCRE_DEBUG
1066        printf("start possessive bracket %d\n", number);
1067        printf("subject=");
1068        pchars(eptr, 16, TRUE, md);
1069        printf("\n");
1070    #endif
1071    
1072        if (offset < md->offset_max)
1073          {
1074          matched_once = FALSE;
1075          code_offset = (int)(ecode - md->start_code);
1076    
1077          save_offset1 = md->offset_vector[offset];
1078          save_offset2 = md->offset_vector[offset+1];
1079          save_offset3 = md->offset_vector[md->offset_end - number];
1080          save_capture_last = md->capture_last;
1081    
1082          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1083    
1084          /* Each time round the loop, save the current subject position for use
1085          when the group matches. For MATCH_MATCH, the group has matched, so we
1086          restart it with a new subject starting position, remembering that we had
1087          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1088          usual. If we haven't matched any alternatives in any iteration, check to
1089          see if a previous iteration matched. If so, the group has matched;
1090          continue from afterwards. Otherwise it has failed; restore the previous
1091          capture values before returning NOMATCH. */
1092    
1093          for (;;)
1094            {
1095            md->offset_vector[md->offset_end - number] =
1096              (int)(eptr - md->start_subject);
1097            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1098            RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1099              eptrb, RM63);
1100            if (rrc == MATCH_KETRPOS)
1101              {
1102              offset_top = md->end_offset_top;
1103              eptr = md->end_match_ptr;
1104              ecode = md->start_code + code_offset;
1105              save_capture_last = md->capture_last;
1106              matched_once = TRUE;
1107              continue;
1108              }
1109    
1110            /* See comment in the code for capturing groups above about handling
1111            THEN. */
1112    
1113            if (rrc == MATCH_THEN)
1114              {
1115              next = ecode + GET(ecode,1);
1116              if (md->start_match_ptr < next &&
1117                  (*ecode == OP_ALT || *next == OP_ALT))
1118                rrc = MATCH_NOMATCH;
1119              }
1120    
1121            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1122            md->capture_last = save_capture_last;
1123            ecode += GET(ecode, 1);
1124            if (*ecode != OP_ALT) break;
1125          }          }
1126    
1127        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1128        otherwise return. */          {
1129            md->offset_vector[offset] = save_offset1;
1130            md->offset_vector[offset+1] = save_offset2;
1131            md->offset_vector[md->offset_end - number] = save_offset3;
1132            }
1133    
1134        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (allow_zero || matched_once)
1135          eptrb, flags, RM2);          {
1136        if (rrc != MATCH_NOMATCH &&          ecode += 1 + LINK_SIZE;
1137            (rrc != MATCH_THEN || md->start_match_ptr != ecode))          break;
1138          RRETURN(rrc);          }
1139    
1140          RRETURN(MATCH_NOMATCH);
1141          }
1142    
1143        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1144        as a non-capturing bracket. */
1145    
1146        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1147        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1148    
1149        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1150    
1151        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1152        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1153    
1154        /* Non-capturing possessive bracket with unlimited repeat. We come here
1155        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1156        without the capturing complication. It is written out separately for speed
1157        and cleanliness. */
1158    
1159        case OP_BRAPOS:
1160        case OP_SBRAPOS:
1161        allow_zero = FALSE;
1162    
1163        POSSESSIVE_NON_CAPTURE:
1164        matched_once = FALSE;
1165        code_offset = (int)(ecode - md->start_code);
1166    
1167        for (;;)
1168          {
1169          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1170          RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
1171            eptrb, RM48);
1172          if (rrc == MATCH_KETRPOS)
1173            {
1174            offset_top = md->end_offset_top;
1175            eptr = md->end_match_ptr;
1176            ecode = md->start_code + code_offset;
1177            matched_once = TRUE;
1178            continue;
1179            }
1180    
1181          /* See comment in the code for capturing groups above about handling
1182          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        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1194          if (*ecode != OP_ALT) break;
1195        }        }
1196    
1197        if (matched_once || allow_zero)
1198          {
1199          ecode += 1 + LINK_SIZE;
1200          break;
1201          }
1202        RRETURN(MATCH_NOMATCH);
1203    
1204      /* Control never reaches here. */      /* Control never reaches here. */
1205    
1206      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1207      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1208      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1209      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1210    
1211      case OP_COND:      case OP_COND:
1212      case OP_SCOND:      case OP_SCOND:
1213      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1214    
1215      /* Because of the way auto-callout works during compile, a callout item is      /* Because of the way auto-callout works during compile, a callout item is
1216      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 928  for (;;) Line 1220  for (;;)
1220        if (pcre_callout != NULL)        if (pcre_callout != NULL)
1221          {          {
1222          pcre_callout_block cb;          pcre_callout_block cb;
1223          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1224          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1225          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1226          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 940  for (;;) Line 1232  for (;;)
1232          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1233          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1234          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1235          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          cb.mark             = md->nomatch_mark;
1236            if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1237          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1238          }          }
1239        ecode += _pcre_OP_lengths[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1240        }        }
1241    
1242      condcode = ecode[LINK_SIZE+1];      condcode = ecode[LINK_SIZE+1];
# Line 960  for (;;) Line 1253  for (;;)
1253        else        else
1254          {          {
1255          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1256          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1257    
1258          /* 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
1259          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
1260          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
1261          if any one is set. */          if any one is set. */
1262    
1263          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1264            {            {
1265            uschar *slotA = md->name_table;            pcre_uchar *slotA = md->name_table;
1266            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
1267              {              {
1268              if (GET2(slotA, 0) == recno) break;              if (GET2(slotA, 0) == recno) break;
# Line 982  for (;;) Line 1275  for (;;)
1275    
1276            if (i < md->name_count)            if (i < md->name_count)
1277              {              {
1278              uschar *slotB = slotA;              pcre_uchar *slotB = slotA;
1279              while (slotB > md->name_table)              while (slotB > md->name_table)
1280                {                {
1281                slotB -= md->name_entry_size;                slotB -= md->name_entry_size;
1282                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1283                  {                  {
1284                  condition = GET2(slotB, 0) == md->recursive->group_num;                  condition = GET2(slotB, 0) == md->recursive->group_num;
1285                  if (condition) break;                  if (condition) break;
# Line 1002  for (;;) Line 1295  for (;;)
1295                for (i++; i < md->name_count; i++)                for (i++; i < md->name_count; i++)
1296                  {                  {
1297                  slotB += md->name_entry_size;                  slotB += md->name_entry_size;
1298                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                  if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1299                    {                    {
1300                    condition = GET2(slotB, 0) == md->recursive->group_num;                    condition = GET2(slotB, 0) == md->recursive->group_num;
1301                    if (condition) break;                    if (condition) break;
# Line 1015  for (;;) Line 1308  for (;;)
1308    
1309          /* Chose branch according to the condition */          /* Chose branch according to the condition */
1310    
1311          ecode += condition? 3 : GET(ecode, 1);          ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1312          }          }
1313        }        }
1314    
# Line 1032  for (;;) Line 1325  for (;;)
1325        if (!condition && condcode == OP_NCREF)        if (!condition && condcode == OP_NCREF)
1326          {          {
1327          int refno = offset >> 1;          int refno = offset >> 1;
1328          uschar *slotA = md->name_table;          pcre_uchar *slotA = md->name_table;
1329    
1330          for (i = 0; i < md->name_count; i++)          for (i = 0; i < md->name_count; i++)
1331            {            {
# Line 1046  for (;;) Line 1339  for (;;)
1339    
1340          if (i < md->name_count)          if (i < md->name_count)
1341            {            {
1342            uschar *slotB = slotA;            pcre_uchar *slotB = slotA;
1343            while (slotB > md->name_table)            while (slotB > md->name_table)
1344              {              {
1345              slotB -= md->name_entry_size;              slotB -= md->name_entry_size;
1346              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)              if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1347                {                {
1348                offset = GET2(slotB, 0) << 1;                offset = GET2(slotB, 0) << 1;
1349                condition = offset < offset_top &&                condition = offset < offset_top &&
# Line 1068  for (;;) Line 1361  for (;;)
1361              for (i++; i < md->name_count; i++)              for (i++; i < md->name_count; i++)
1362                {                {
1363                slotB += md->name_entry_size;                slotB += md->name_entry_size;
1364                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)                if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)
1365                  {                  {
1366                  offset = GET2(slotB, 0) << 1;                  offset = GET2(slotB, 0) << 1;
1367                  condition = offset < offset_top &&                  condition = offset < offset_top &&
# Line 1083  for (;;) Line 1376  for (;;)
1376    
1377        /* Chose branch according to the condition */        /* Chose branch according to the condition */
1378    
1379        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);
1380        }        }
1381    
1382      else if (condcode == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
# Line 1093  for (;;) Line 1386  for (;;)
1386        }        }
1387    
1388      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1389      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1390      assertion. */      an assertion. */
1391    
1392      else      else
1393        {        {
1394        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1395            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1396        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1397          {          {
1398            if (md->end_offset_top > offset_top)
1399              offset_top = md->end_offset_top;  /* Captures may have happened */
1400          condition = TRUE;          condition = TRUE;
1401          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1402          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1403          }          }
1404        else if (rrc != MATCH_NOMATCH &&  
1405                (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1406          assertion; it is therefore treated as NOMATCH. */
1407    
1408          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1409          {          {
1410          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1411          }          }
# Line 1118  for (;;) Line 1416  for (;;)
1416          }          }
1417        }        }
1418    
1419      /* 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
1420      we can use tail recursion to avoid using another stack frame, except when      use tail recursion to avoid using another stack frame, except when there is
1421      match_cbegroup is required for an unlimited repeat of a possibly empty      unlimited repeat of a possibly empty group. In the latter case, a recursive
1422      group. If the second alternative doesn't exist, we can just plough on. */      call to match() is always required, unless the second alternative doesn't
1423        exist, in which case we can just plough on. Note that, for compatibility
1424        with Perl, the | in a conditional group is NOT treated as creating two
1425        alternatives. If a THEN is encountered in the branch, it propagates out to
1426        the enclosing alternative (unless nested in a deeper set of alternatives,
1427        of course). */
1428    
1429      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1430        {        {
1431        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
1432          {          {
1433          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          ecode += 1 + LINK_SIZE;
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
1434          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1435          }          }
1436    
1437          md->match_function_type = MATCH_CBEGROUP;
1438          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1439          RRETURN(rrc);
1440        }        }
1441      else                         /* Condition false & no alternative */  
1442         /* Condition false & no alternative; continue after the group. */
1443    
1444        else
1445        {        {
1446        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1447        }        }
# Line 1164  for (;;) Line 1468  for (;;)
1468        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1469        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1470        }        }
1471      ecode += 3;      ecode += 1 + IMM2_SIZE;
1472      break;      break;
1473    
1474    
1475      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. */
     recursion, we should restore the offsets appropriately and continue from  
     after the call. */  
1476    
     case OP_ACCEPT:  
1477      case OP_END:      case OP_END:
1478      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1479        {      case OP_ASSERT_ACCEPT:
       recursion_info *rec = md->recursive;  
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
       md->recursive = rec->prevrec;  
       memmove(md->offset_vector, rec->offset_save,  
         rec->saved_max * sizeof(int));  
       offset_top = rec->save_offset_top;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1480    
1481      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is      /* If we have matched an empty string, fail if not in an assertion and not
1482      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1483      the subject. In both cases, backtracking will then try other alternatives,      is set and we have matched at the start of the subject. In both cases,
1484      if any. */      backtracking will then try other alternatives, if any. */
1485    
1486      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1487          (md->notempty ||           md->recursive == NULL &&
1488            (md->notempty_atstart &&           (md->notempty ||
1489              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1490        MRRETURN(MATCH_NOMATCH);               mstart == md->start_subject + md->start_offset)))
1491          RRETURN(MATCH_NOMATCH);
1492    
1493      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1494    
# Line 1205  for (;;) Line 1497  for (;;)
1497      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1498    
1499      /* 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
1500      given as the argument to MRRETURN when the heap is in use. */      given as the argument to RRETURN when the heap is in use. */
1501    
1502      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1503      MRRETURN(rrc);      RRETURN(rrc);
   
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1504    
1505      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1506      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,
1507      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
1508      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
1509      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1510        of a condition, we want to return immediately afterwards. The caller of
1511        this incarnation of the match() function will have set MATCH_CONDASSERT in
1512        md->match_function type, and one of these opcodes will be the first opcode
1513        that is processed. We use a local variable that is preserved over calls to
1514        match() to remember this case. */
1515    
1516      case OP_ASSERT:      case OP_ASSERT:
1517      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1518        if (md->match_function_type == MATCH_CONDASSERT)
1519          {
1520          condassert = TRUE;
1521          md->match_function_type = 0;
1522          }
1523        else condassert = FALSE;
1524    
1525      do      do
1526        {        {
1527        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1528        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1529          {          {
1530          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1531          break;          break;
1532          }          }
1533        if (rrc != MATCH_NOMATCH &&  
1534            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1535          RRETURN(rrc);        as NOMATCH. */
1536    
1537          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1538        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1539        }        }
1540      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1541      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);  
1542        if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1543    
1544      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1545    
1546      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1547    
1548      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1549      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1261  for (;;) Line 1559  for (;;)
1559    
1560      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1561      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1562        if (md->match_function_type == MATCH_CONDASSERT)
1563          {
1564          condassert = TRUE;
1565          md->match_function_type = 0;
1566          }
1567        else condassert = FALSE;
1568    
1569      do      do
1570        {        {
1571        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1572          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);
       if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);  
1573        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1574          {          {
1575          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1576          break;          break;
1577          }          }
1578        if (rrc != MATCH_NOMATCH &&  
1579            (rrc != MATCH_THEN || md->start_match_ptr != ecode))        /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1580          RRETURN(rrc);        as NOMATCH. */
1581    
1582          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1583        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1584        }        }
1585      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1586    
1587      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1588    
1589      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1590      continue;      continue;
# Line 1289  for (;;) Line 1595  for (;;)
1595      back a number of characters, not bytes. */      back a number of characters, not bytes. */
1596    
1597      case OP_REVERSE:      case OP_REVERSE:
1598  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1599      if (utf8)      if (utf)
1600        {        {
1601        i = GET(ecode, 1);        i = GET(ecode, 1);
1602        while (i-- > 0)        while (i-- > 0)
1603          {          {
1604          eptr--;          eptr--;
1605          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1606          BACKCHAR(eptr);          BACKCHAR(eptr);
1607          }          }
1608        }        }
# Line 1307  for (;;) Line 1613  for (;;)
1613    
1614        {        {
1615        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1616        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
1617        }        }
1618    
1619      /* Save the earliest consulted character, then skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
# Line 1324  for (;;) Line 1630  for (;;)
1630      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1631        {        {
1632        pcre_callout_block cb;        pcre_callout_block cb;
1633        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1634        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1635        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1636        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 1336  for (;;) Line 1642  for (;;)
1642        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1643        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1644        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1645        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        cb.mark             = md->nomatch_mark;
1646          if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1647        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1648        }        }
1649      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 1346  for (;;) Line 1653  for (;;)
1653      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
1654      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1655    
1656      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1657      save their starting points and reinstate them after the recursion. However,      re-instated afterwards. We don't know how many are started and not yet
1658      we don't know how many such there are (offset_top records the completed      finished (offset_top records the completed total) so we just have to save
1659      total) so we just have to save all the potential data. There may be up to      all the potential data. There may be up to 65535 such values, which is too
1660      65535 such values, which is too large to put on the stack, but using malloc      large to put on the stack, but using malloc for small numbers seems
1661      for small numbers seems expensive. As a compromise, the stack is used when      expensive. As a compromise, the stack is used when there are no more than
1662      there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc      REC_STACK_SAVE_MAX values to store; otherwise malloc is used.
     is used. A problem is what to do if the malloc fails ... there is no way of  
     returning to the top level with an error. Save the top REC_STACK_SAVE_MAX  
     values on the stack, and accept that the rest may be wrong.  
1663    
1664      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
1665      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
1666      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1667        a lot, so he is not to blame for the current way it works. */
1668    
1669      case OP_RECURSE:      case OP_RECURSE:
1670        {        {
1671          recursion_info *ri;
1672          int recno;
1673    
1674        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1675        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1676          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1677    
1678          /* Check for repeating a recursion without advancing the subject pointer.
1679          This should catch convoluted mutual recursions. (Some simple cases are
1680          caught at compile time.) */
1681    
1682          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1683            if (recno == ri->group_num && eptr == ri->subject_position)
1684              RRETURN(PCRE_ERROR_RECURSELOOP);
1685    
1686        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1687    
1688          new_recursive.group_num = recno;
1689          new_recursive.subject_position = eptr;
1690        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1691        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1692    
1693        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1694    
1695        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1696    
1697        /* Now save the offset data. */        /* Now save the offset data */
1698    
1699        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1700        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1388  for (;;) Line 1705  for (;;)
1705            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1706          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1707          }          }
   
1708        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1709              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1710    
1711        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1712        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1713          might be changed, so reset it before looping. */
1714    
1715        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1716        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1717        do        do
1718          {          {
1719          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1720            md, ims, eptrb, flags, RM6);          RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top,
1721              md, eptrb, RM6);
1722            memcpy(md->offset_vector, new_recursive.offset_save,
1723                new_recursive.saved_max * sizeof(int));
1724            md->recursive = new_recursive.prevrec;
1725          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1726            {            {
1727            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1728            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1729              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1730            MRRETURN(MATCH_MATCH);  
1731              /* Set where we got to in the subject, and reset the start in case
1732              it was changed by \K. This *is* propagated back out of a recursion,
1733              for Perl compatibility. */
1734    
1735              eptr = md->end_match_ptr;
1736              mstart = md->start_match_ptr;
1737              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1738            }            }
1739          else if (rrc != MATCH_NOMATCH &&  
1740                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))          /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1741            as NOMATCH. */
1742    
1743            else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1744            {            {
1745            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1746            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
# Line 1420  for (;;) Line 1749  for (;;)
1749            }            }
1750    
1751          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1752          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1753          }          }
1754        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1430  for (;;) Line 1757  for (;;)
1757        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1758        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1759          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1760        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
       }  
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer, but resetting  
     the start-of-match value in case it was changed by \K. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */  
         {  
         mstart = md->start_match_ptr;  
         break;  
         }  
       if (rrc != MATCH_NOMATCH &&  
           (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
         RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
1761        }        }
1762    
1763      if (*ecode == OP_KETRMIN)      RECURSION_MATCHED:
1764        {      break;
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
1765    
1766      /* An alternation is the end of a branch; scan along to find the end of the      /* An alternation is the end of a branch; scan along to find the end of the
1767      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1529  for (;;) Line 1777  for (;;)
1777      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1778    
1779      case OP_BRAZERO:      case OP_BRAZERO:
1780        {      next = ecode + 1;
1781        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1782        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1783        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1784        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1785      break;      break;
1786    
1787      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1788        {      next = ecode + 1;
1789        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1790        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1791        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1792        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1793      break;      break;
1794    
1795      case OP_SKIPZERO:      case OP_SKIPZERO:
1796        {      next = ecode+1;
1797        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1798        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1799      break;      break;
1800    
1801        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1802        here; just jump to the group, with allow_zero set TRUE. */
1803    
1804        case OP_BRAPOSZERO:
1805        op = *(++ecode);
1806        allow_zero = TRUE;
1807        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1808          goto POSSESSIVE_NON_CAPTURE;
1809    
1810      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1811    
1812      case OP_KET:      case OP_KET:
1813      case OP_KETRMIN:      case OP_KETRMIN:
1814      case OP_KETRMAX:      case OP_KETRMAX:
1815        case OP_KETRPOS:
1816      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1817    
1818      /* 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
1819      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1820      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1821    
1822      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1823        {        {
1824        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1825        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1826        }        }
1827      else saved_eptr = NULL;      else saved_eptr = NULL;
1828    
1829      /* If we are at the end of an assertion group or an atomic group, stop      /* If we are at the end of an assertion group or a non-capturing atomic
1830      matching and return MATCH_MATCH, but record the current high water mark for      group, stop matching and return MATCH_MATCH, but record the current high
1831      use by positive assertions. We also need to record the match start in case      water mark for use by positive assertions. We also need to record the match
1832      it was changed by \K. */      start in case it was changed by \K. */
1833    
1834      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1835          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||           *prev == OP_ONCE_NC)
         *prev == OP_ONCE)  
1836        {        {
1837        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1838        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1839        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1840        MRRETURN(MATCH_MATCH);        RRETURN(MATCH_MATCH);         /* Sets md->mark */
1841        }        }
1842    
1843      /* 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
1844      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1845      bumping the high water mark. Note that whole-pattern recursion is coded as      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1846      a recurse into group 0, so it won't be picked up here. Instead, we catch it      into group 0, so it won't be picked up here. Instead, we catch it when the
1847      when the OP_END is reached. Other recursion is handled here. */      OP_END is reached. Other recursion is handled here. We just have to record
1848        the current subject position and start match pointer and give a MATCH
1849        return. */
1850    
1851      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1852            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1853        {        {
1854        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1855        offset = number << 1;        offset = number << 1;
# Line 1605  for (;;) Line 1859  for (;;)
1859        printf("\n");        printf("\n");
1860  #endif  #endif
1861    
1862          /* Handle a recursively called group. */
1863    
1864          if (md->recursive != NULL && md->recursive->group_num == number)
1865            {
1866            md->end_match_ptr = eptr;
1867            md->start_match_ptr = mstart;
1868            RRETURN(MATCH_MATCH);
1869            }
1870    
1871          /* Deal with capturing */
1872    
1873        md->capture_last = number;        md->capture_last = number;
1874        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1875          {          {
1876            /* If offset is greater than offset_top, it means that we are
1877            "skipping" a capturing group, and that group's offsets must be marked
1878            unset. In earlier versions of PCRE, all the offsets were unset at the
1879            start of matching, but this doesn't work because atomic groups and
1880            assertions can cause a value to be set that should later be unset.
1881            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1882            part of the atomic group, but this is not on the final matching path,
1883            so must be unset when 2 is set. (If there is no group 2, there is no
1884            problem, because offset_top will then be 2, indicating no capture.) */
1885    
1886            if (offset > offset_top)
1887              {
1888              register int *iptr = md->offset_vector + offset_top;
1889              register int *iend = md->offset_vector + offset;
1890              while (iptr < iend) *iptr++ = -1;
1891              }
1892    
1893            /* Now make the extraction */
1894    
1895          md->offset_vector[offset] =          md->offset_vector[offset] =
1896            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1897          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1898          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1899          }          }
1900          }
1901    
1902        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1903        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1904        group. This is the forcible breaking of infinite loops as implemented in
1905        Perl 5.005. For a non-repeating atomic group that includes captures,
1906        establish a backup point by processing the rest of the pattern at a lower
1907        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1908        original OP_ONCE level, thereby bypassing intermediate backup points, but
1909        resetting any captures that happened along the way. */
1910    
1911        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1912          {
1913          if (*prev == OP_ONCE)
1914          {          {
1915          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1916          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1917          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1918          memcpy(md->offset_vector, rec->offset_save,          RRETURN(MATCH_ONCE);
           rec->saved_max * sizeof(int));  
         offset_top = rec->save_offset_top;  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1919          }          }
1920          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1921          break;
1922        }        }
1923    
1924      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1925      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1926        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
1927    
1928      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRPOS)
1929        {        {
1930        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1931        break;        md->end_offset_top = offset_top;
1932          RRETURN(MATCH_KETRPOS);
1933        }        }
1934    
1935      /* The repeating kets try the rest of the pattern or restart from the      /* The normal repeating kets try the rest of the pattern or restart from
1936      preceding bracket, in the appropriate order. In the second case, we can use      the preceding bracket, in the appropriate order. In the second case, we can
1937      tail recursion to avoid using another stack frame, unless we have an      use tail recursion to avoid using another stack frame, unless we have an
1938      unlimited repeat of a group that can match an empty string. */      an atomic group or an unlimited repeat of a group that can match an empty
1939        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1940    
1941      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1942        {        {
1943        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1944        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1945        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1946            {
1947            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1948            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1949            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1950            RRETURN(MATCH_ONCE);
1951            }
1952          if (*prev >= OP_SBRA)    /* Could match an empty string */
1953          {          {
1954          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1955            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1956          RRETURN(rrc);          RRETURN(rrc);
1957          }          }
1958        ecode = prev;        ecode = prev;
# Line 1670  for (;;) Line 1960  for (;;)
1960        }        }
1961      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1962        {        {
1963        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1964          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1965          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1966        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1967          if (*prev == OP_ONCE)
1968            {
1969            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1970            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1971            md->once_target = prev;
1972            RRETURN(MATCH_ONCE);
1973            }
1974        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1975        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1976        }        }
1977      /* Control never gets here */      /* Control never gets here */
1978    
1979      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1980    
1981      case OP_CIRC:      case OP_CIRC:
1982      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
     if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         MRRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
1983    
1984      /* Start of subject assertion */      /* Start of subject assertion */
1985    
1986      case OP_SOD:      case OP_SOD:
1987      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);
1988        ecode++;
1989        break;
1990    
1991        /* Multiline mode: start of subject unless notbol, or after any newline. */
1992    
1993        case OP_CIRCM:
1994        if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
1995        if (eptr != md->start_subject &&
1996            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1997          RRETURN(MATCH_NOMATCH);
1998      ecode++;      ecode++;
1999      break;      break;
2000    
2001      /* Start of match assertion */      /* Start of match assertion */
2002    
2003      case OP_SOM:      case OP_SOM:
2004      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);
2005      ecode++;      ecode++;
2006      break;      break;
2007    
# Line 1713  for (;;) Line 2012  for (;;)
2012      ecode++;      ecode++;
2013      break;      break;
2014    
2015      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2016      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2017    
2018      case OP_DOLL:      case OP_DOLLM:
2019      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2020        {        { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }
2021        if (eptr < md->end_subject)      else
         { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }  
       else  
         {  
         if (md->noteol) MRRETURN(MATCH_NOMATCH);  
         SCHECK_PARTIAL();  
         }  
       ecode++;  
       break;  
       }  
     else  /* Not multiline */  
2022        {        {
2023        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
2024        if (!md->endonly) goto ASSERT_NL_OR_EOS;        SCHECK_PARTIAL();
2025        }        }
2026        ecode++;
2027        break;
2028    
2029        /* Not multiline mode: assert before a terminating newline or before end of
2030        subject unless noteol is set. */
2031    
2032        case OP_DOLL:
2033        if (md->noteol) RRETURN(MATCH_NOMATCH);
2034        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2035    
2036      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2037    
2038      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2039    
2040      case OP_EOD:      case OP_EOD:
2041      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
2042      SCHECK_PARTIAL();      SCHECK_PARTIAL();
2043      ecode++;      ecode++;
2044      break;      break;
# Line 1751  for (;;) Line 2049  for (;;)
2049      ASSERT_NL_OR_EOS:      ASSERT_NL_OR_EOS:
2050      if (eptr < md->end_subject &&      if (eptr < md->end_subject &&
2051          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2052        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2053    
2054      /* Either at end of string or \n before end. */      /* Either at end of string or \n before end. */
2055    
# Line 1770  for (;;) Line 2068  for (;;)
2068        be "non-word" characters. Remember the earliest consulted character for        be "non-word" characters. Remember the earliest consulted character for
2069        partial matching. */        partial matching. */
2070    
2071  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2072        if (utf8)        if (utf)
2073          {          {
2074          /* Get status of previous character */          /* Get status of previous character */
2075    
2076          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2077            {            {
2078            USPTR lastptr = eptr - 1;            PCRE_PUCHAR lastptr = eptr - 1;
2079            while((*lastptr & 0xc0) == 0x80) lastptr--;            BACKCHAR(lastptr);
2080            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2081            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2082  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 1873  for (;;) Line 2171  for (;;)
2171    
2172        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2173             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2174          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2175        }        }
2176      break;      break;
2177    
2178      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2179    
2180      case OP_ANY:      case OP_ANY:
2181      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
2182      /* Fall through */      /* Fall through */
2183    
2184      case OP_ALLANY:      case OP_ALLANY:
2185      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2186        {        {                            /* not be updated before SCHECK_PARTIAL. */
2187        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2188        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2189        }        }
2190      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      eptr++;
2191    #ifdef SUPPORT_UTF
2192        if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++);
2193    #endif
2194      ecode++;      ecode++;
2195      break;      break;
2196    
# Line 1897  for (;;) Line 2198  for (;;)
2198      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2199    
2200      case OP_ANYBYTE:      case OP_ANYBYTE:
2201      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2202        {        {                            /* not be updated before SCHECK_PARTIAL. */
2203        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2204        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2205        }        }
2206        eptr++;
2207      ecode++;      ecode++;
2208      break;      break;
2209    
# Line 1909  for (;;) Line 2211  for (;;)
2211      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2212        {        {
2213        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2214        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2215        }        }
2216      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2217      if (      if (
2218  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2219         c < 256 &&         c < 256 &&
2220  #endif  #endif
2221         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2222         )         )
2223        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2224      ecode++;      ecode++;
2225      break;      break;
2226    
# Line 1926  for (;;) Line 2228  for (;;)
2228      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2229        {        {
2230        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2231        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2232        }        }
2233      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2234      if (      if (
2235  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2236         c >= 256 ||         c > 255 ||
2237  #endif  #endif
2238         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2239         )         )
2240        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2241      ecode++;      ecode++;
2242      break;      break;
2243    
# Line 1943  for (;;) Line 2245  for (;;)
2245      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2246        {        {
2247        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2248        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2249        }        }
2250      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2251      if (      if (
2252  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2253         c < 256 &&         c < 256 &&
2254  #endif  #endif
2255         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2256         )         )
2257        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2258      ecode++;      ecode++;
2259      break;      break;
2260    
# Line 1960  for (;;) Line 2262  for (;;)
2262      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2263        {        {
2264        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2265        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2266        }        }
2267      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2268      if (      if (
2269  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2270         c >= 256 ||         c > 255 ||
2271  #endif  #endif
2272         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2273         )         )
2274        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2275      ecode++;      ecode++;
2276      break;      break;
2277    
# Line 1977  for (;;) Line 2279  for (;;)
2279      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2280        {        {
2281        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2282        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2283        }        }
2284      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2285      if (      if (
2286  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2287         c < 256 &&         c < 256 &&
2288  #endif  #endif
2289         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2290         )         )
2291        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2292      ecode++;      ecode++;
2293      break;      break;
2294    
# Line 1994  for (;;) Line 2296  for (;;)
2296      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2297        {        {
2298        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2299        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2300        }        }
2301      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2302      if (      if (
2303  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
2304         c >= 256 ||         c > 255 ||
2305  #endif  #endif
2306         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2307         )         )
2308        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2309      ecode++;      ecode++;
2310      break;      break;
2311    
# Line 2011  for (;;) Line 2313  for (;;)
2313      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2314        {        {
2315        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2316        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2317        }        }
2318      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2319      switch(c)      switch(c)
2320        {        {
2321        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2322    
2323        case 0x000d:        case 0x000d:
2324        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2325        break;        break;
# Line 2029  for (;;) Line 2332  for (;;)
2332        case 0x0085:        case 0x0085:
2333        case 0x2028:        case 0x2028:
2334        case 0x2029:        case 0x2029:
2335        if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);        if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
2336        break;        break;
2337        }        }
2338      ecode++;      ecode++;
# Line 2039  for (;;) Line 2342  for (;;)
2342      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2343        {        {
2344        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2345        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2346        }        }
2347      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2348      switch(c)      switch(c)
# Line 2064  for (;;) Line 2367  for (;;)
2367        case 0x202f:    /* NARROW NO-BREAK SPACE */        case 0x202f:    /* NARROW NO-BREAK SPACE */
2368        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2369        case 0x3000:    /* IDEOGRAPHIC SPACE */        case 0x3000:    /* IDEOGRAPHIC SPACE */
2370        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2371        }        }
2372      ecode++;      ecode++;
2373      break;      break;
# Line 2073  for (;;) Line 2376  for (;;)
2376      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2377        {        {
2378        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2379        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2380        }        }
2381      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2382      switch(c)      switch(c)
2383        {        {
2384        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2385        case 0x09:      /* HT */        case 0x09:      /* HT */
2386        case 0x20:      /* SPACE */        case 0x20:      /* SPACE */
2387        case 0xa0:      /* NBSP */        case 0xa0:      /* NBSP */
# Line 2107  for (;;) Line 2410  for (;;)
2410      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2411        {        {
2412        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2413        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2414        }        }
2415      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2416      switch(c)      switch(c)
# Line 2120  for (;;) Line 2423  for (;;)
2423        case 0x85:      /* NEL */        case 0x85:      /* NEL */
2424        case 0x2028:    /* LINE SEPARATOR */        case 0x2028:    /* LINE SEPARATOR */
2425        case 0x2029:    /* PARAGRAPH SEPARATOR */        case 0x2029:    /* PARAGRAPH SEPARATOR */
2426        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2427        }        }
2428      ecode++;      ecode++;
2429      break;      break;
# Line 2129  for (;;) Line 2432  for (;;)
2432      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2433        {        {
2434        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2435        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2436        }        }
2437      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2438      switch(c)      switch(c)
2439        {        {
2440        default: MRRETURN(MATCH_NOMATCH);        default: RRETURN(MATCH_NOMATCH);
2441        case 0x0a:      /* LF */        case 0x0a:      /* LF */
2442        case 0x0b:      /* VT */        case 0x0b:      /* VT */
2443        case 0x0c:      /* FF */        case 0x0c:      /* FF */
# Line 2156  for (;;) Line 2459  for (;;)
2459      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2460        {        {
2461        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2462        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2463        }        }
2464      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2465        {        {
# Line 2165  for (;;) Line 2468  for (;;)
2468        switch(ecode[1])        switch(ecode[1])
2469          {          {
2470          case PT_ANY:          case PT_ANY:
2471          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2472          break;          break;
2473    
2474          case PT_LAMP:          case PT_LAMP:
2475          if ((prop->chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2476               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2477               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2478            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2479          break;          break;
2480    
2481          case PT_GC:          case PT_GC:
2482          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP))
2483            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2484          break;          break;
2485    
2486          case PT_PC:          case PT_PC:
2487          if ((ecode[2] != prop->chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2488            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2489          break;          break;
2490    
2491          case PT_SC:          case PT_SC:
2492          if ((ecode[2] != prop->script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2493            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2494          break;          break;
2495    
2496          /* These are specials */          /* These are specials */
2497    
2498          case PT_ALNUM:          case PT_ALNUM:
2499          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2500               _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))               PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2501            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2502          break;          break;
2503    
2504          case PT_SPACE:    /* Perl space */          case PT_SPACE:    /* Perl space */
2505          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2506               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2507                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2508            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2509          break;          break;
2510    
2511          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2512          if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2513               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2514               c == CHAR_FF || c == CHAR_CR)               c == CHAR_FF || c == CHAR_CR)
2515                 == (op == OP_NOTPROP))                 == (op == OP_NOTPROP))
2516            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2517          break;          break;
2518    
2519          case PT_WORD:          case PT_WORD:
2520          if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2521               _pcre_ucp_gentype[prop->chartype] == ucp_N ||               PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2522               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))               c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2523            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2524          break;          break;
2525    
2526          /* This should never occur */          /* This should never occur */
# Line 2237  for (;;) Line 2540  for (;;)
2540      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
2541        {        {
2542        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2543        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2544        }        }
2545      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2546        if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH);
2547        while (eptr < md->end_subject)
2548        {        {
2549        int category = UCD_CATEGORY(c);        int len = 1;
2550        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2551        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2552          {        eptr += len;
         int len = 1;  
         if (!utf8) c = *eptr; else  
           {  
           GETCHARLEN(c, eptr, len);  
           }  
         category = UCD_CATEGORY(c);  
         if (category != ucp_M) break;  
         eptr += len;  
         }  
2553        }        }
2554      ecode++;      ecode++;
2555      break;      break;
# Line 2269  for (;;) Line 2565  for (;;)
2565      loops). */      loops). */
2566    
2567      case OP_REF:      case OP_REF:
2568        case OP_REFI:
2569        caseless = op == OP_REFI;
2570      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2571      ecode += 3;      ecode += 1 + IMM2_SIZE;
2572    
2573      /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2574    
# Line 2310  for (;;) Line 2608  for (;;)
2608        case OP_CRMINRANGE:        case OP_CRMINRANGE:
2609        minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2610        min = GET2(ecode, 1);        min = GET2(ecode, 1);
2611        max = GET2(ecode, 3);        max = GET2(ecode, 1 + IMM2_SIZE);
2612        if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2613        ecode += 5;        ecode += 1 + 2 * IMM2_SIZE;
2614        break;        break;
2615    
2616        default:               /* No repeat follows */        default:               /* No repeat follows */
2617        if ((length = match_ref(offset, eptr, length, md, ims)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2618          {          {
2619          CHECK_PARTIAL();          CHECK_PARTIAL();
2620          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2621          }          }
2622        eptr += length;        eptr += length;
2623        continue;              /* With the main loop */        continue;              /* With the main loop */
# Line 2336  for (;;) Line 2634  for (;;)
2634    
2635      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2636        {        {
2637        int slength;        int slength;
2638        if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2639          {          {
2640          CHECK_PARTIAL();          CHECK_PARTIAL();
2641          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2642          }          }
2643        eptr += slength;        eptr += slength;
2644        }        }
# Line 2356  for (;;) Line 2654  for (;;)
2654        {        {
2655        for (fi = min;; fi++)        for (fi = min;; fi++)
2656          {          {
2657          int slength;          int slength;
2658          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2659          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2660          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) RRETURN(MATCH_NOMATCH);
2661          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2662            {            {
2663            CHECK_PARTIAL();            CHECK_PARTIAL();
2664            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2665            }            }
2666          eptr += slength;          eptr += slength;
2667          }          }
# Line 2377  for (;;) Line 2675  for (;;)
2675        pp = eptr;        pp = eptr;
2676        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2677          {          {
2678          int slength;          int slength;
2679          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2680            {            {
2681            CHECK_PARTIAL();            CHECK_PARTIAL();
2682            break;            break;
# Line 2387  for (;;) Line 2685  for (;;)
2685          }          }
2686        while (eptr >= pp)        while (eptr >= pp)
2687          {          {
2688          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2689          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2690          eptr -= length;          eptr -= length;
2691          }          }
2692        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
2693        }        }
2694      /* Control never gets here */      /* Control never gets here */
2695    
# Line 2409  for (;;) Line 2707  for (;;)
2707      case OP_NCLASS:      case OP_NCLASS:
2708      case OP_CLASS:      case OP_CLASS:
2709        {        {
2710          /* The data variable is saved across frames, so the byte map needs to
2711          be stored there. */
2712    #define BYTE_MAP ((pcre_uint8 *)data)
2713        data = ecode + 1;                /* Save for matching */        data = ecode + 1;                /* Save for matching */
2714        ecode += 33;                     /* Advance past the item */        ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */
2715    
2716        switch (*ecode)        switch (*ecode)
2717          {          {
# Line 2431  for (;;) Line 2732  for (;;)
2732          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2733          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2734          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2735          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2736          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2737          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2738          break;          break;
2739    
2740          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2443  for (;;) Line 2744  for (;;)
2744    
2745        /* First, ensure the minimum number of matches are present. */        /* First, ensure the minimum number of matches are present. */
2746    
2747  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2748        /* UTF-8 mode */        if (utf)
       if (utf8)  
2749          {          {
2750          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2751            {            {
2752            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2753              {              {
2754              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2755              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2756              }              }
2757            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2758            if (c > 255)            if (c > 255)
2759              {              {
2760              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2761              }              }
2762            else            else
2763              {              if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
             if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
             }  
2764            }            }
2765          }          }
2766        else        else
2767  #endif  #endif
2768        /* Not UTF-8 mode */        /* Not UTF mode */
2769          {          {
2770          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2771            {            {
2772            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2773              {              {
2774              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2775              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2776              }              }
2777            c = *eptr++;            c = *eptr++;
2778            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2779              if (c > 255)
2780                {
2781                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2782                }
2783              else
2784    #endif
2785                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2786            }            }
2787          }          }
2788    
# Line 2491  for (;;) Line 2796  for (;;)
2796    
2797        if (minimize)        if (minimize)
2798          {          {
2799  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2800          /* UTF-8 mode */          if (utf)
         if (utf8)  
2801            {            {
2802            for (fi = min;; fi++)            for (fi = min;; fi++)
2803              {              {
2804              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2805              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2806              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2807              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2808                {                {
2809                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2810                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2811                }                }
2812              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2813              if (c > 255)              if (c > 255)
2814                {                {
2815                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2816                }                }
2817              else              else
2818                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
               if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  
               }  
2819              }              }
2820            }            }
2821          else          else
2822  #endif  #endif
2823          /* Not UTF-8 mode */          /* Not UTF mode */
2824            {            {
2825            for (fi = min;; fi++)            for (fi = min;; fi++)
2826              {              {
2827              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2828              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2829              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
2830              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
2831                {                {
2832                SCHECK_PARTIAL();                SCHECK_PARTIAL();
2833                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
2834                }                }
2835              c = *eptr++;              c = *eptr++;
2836              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);  #ifndef COMPILE_PCRE8
2837                if (c > 255)
2838                  {
2839                  if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);
2840                  }
2841                else
2842    #endif
2843                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);
2844              }              }
2845            }            }
2846          /* Control never gets here */          /* Control never gets here */
# Line 2543  for (;;) Line 2852  for (;;)
2852          {          {
2853          pp = eptr;          pp = eptr;
2854    
2855  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2856          /* UTF-8 mode */          if (utf)
         if (utf8)  
2857            {            {
2858            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2859              {              {
# Line 2561  for (;;) Line 2869  for (;;)
2869                if (op == OP_CLASS) break;                if (op == OP_CLASS) break;
2870                }                }
2871              else              else
2872                {                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
               if ((data[c/8] & (1 << (c&7))) == 0) break;  
               }  
2873              eptr += len;              eptr += len;
2874              }              }
2875            for (;;)            for (;;)
2876              {              {
2877              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2878              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2879              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2880              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2576  for (;;) Line 2882  for (;;)
2882            }            }
2883          else          else
2884  #endif  #endif
2885            /* Not UTF-8 mode */            /* Not UTF mode */
2886            {            {
2887            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2888              {              {
# Line 2586  for (;;) Line 2892  for (;;)
2892                break;                break;
2893                }                }
2894              c = *eptr;              c = *eptr;
2895              if ((data[c/8] & (1 << (c&7))) == 0) break;  #ifndef COMPILE_PCRE8
2896                if (c > 255)
2897                  {
2898                  if (op == OP_CLASS) break;
2899                  }
2900                else
2901    #endif
2902                  if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
2903              eptr++;              eptr++;
2904              }              }
2905            while (eptr >= pp)            while (eptr >= pp)
2906              {              {
2907              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2908              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2909              eptr--;              eptr--;
2910              }              }
2911            }            }
2912    
2913          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2914          }          }
2915    #undef BYTE_MAP
2916        }        }
2917      /* Control never gets here */      /* Control never gets here */
2918    
# Line 2607  for (;;) Line 2921  for (;;)
2921      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
2922      mode, because Unicode properties are supported in non-UTF-8 mode. */      mode, because Unicode properties are supported in non-UTF-8 mode. */
2923    
2924  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2925      case OP_XCLASS:      case OP_XCLASS:
2926        {        {
2927        data = ecode + 1 + LINK_SIZE;                /* Save for matching */        data = ecode + 1 + LINK_SIZE;                /* Save for matching */
# Line 2632  for (;;) Line 2946  for (;;)
2946          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2947          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2948          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2949          max = GET2(ecode, 3);          max = GET2(ecode, 1 + IMM2_SIZE);
2950          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
2951          ecode += 5;          ecode += 1 + 2 * IMM2_SIZE;
2952          break;          break;
2953    
2954          default:               /* No repeat follows */          default:               /* No repeat follows */
# Line 2649  for (;;) Line 2963  for (;;)
2963          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2964            {            {
2965            SCHECK_PARTIAL();            SCHECK_PARTIAL();
2966            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2967            }            }
2968          GETCHARINCTEST(c, eptr);          GETCHARINCTEST(c, eptr);
2969          if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);          if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
2970          }          }
2971    
2972        /* 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 2667  for (;;) Line 2981  for (;;)
2981          {          {
2982          for (fi = min;; fi++)          for (fi = min;; fi++)
2983            {            {
2984            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2985            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2986            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
2987            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
2988              {              {
2989              SCHECK_PARTIAL();              SCHECK_PARTIAL();
2990              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2991              }              }
2992            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
2993            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);            if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH);
2994            }            }
2995          /* Control never gets here */          /* Control never gets here */
2996          }          }
# Line 2694  for (;;) Line 3008  for (;;)
3008              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3009              break;              break;
3010              }              }
3011    #ifdef SUPPORT_UTF
3012            GETCHARLENTEST(c, eptr, len);            GETCHARLENTEST(c, eptr, len);
3013            if (!_pcre_xclass(c, data)) break;  #else
3014              c = *eptr;
3015    #endif
3016              if (!PRIV(xclass)(c, data, utf)) break;
3017            eptr += len;            eptr += len;
3018            }            }
3019          for(;;)          for(;;)
3020            {            {
3021            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3022            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3023            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3024            if (utf8) BACKCHAR(eptr);  #ifdef SUPPORT_UTF
3025              if (utf) BACKCHAR(eptr);
3026    #endif
3027            }            }
3028          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3029          }          }
3030    
3031        /* Control never gets here */        /* Control never gets here */
# Line 2715  for (;;) Line 3035  for (;;)
3035      /* Match a single character, casefully */      /* Match a single character, casefully */
3036    
3037      case OP_CHAR:      case OP_CHAR:
3038  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3039      if (utf8)      if (utf)
3040        {        {
3041        length = 1;        length = 1;
3042        ecode++;        ecode++;
# Line 2724  for (;;) Line 3044  for (;;)
3044        if (length > md->end_subject - eptr)        if (length > md->end_subject - eptr)
3045          {          {
3046          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */          CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
3047          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3048          }          }
3049        while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);
3050        }        }
3051      else      else
3052  #endif  #endif
3053        /* Not UTF mode */
     /* Non-UTF-8 mode */  
3054        {        {
3055        if (md->end_subject - eptr < 1)        if (md->end_subject - eptr < 1)
3056          {          {
3057          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
3058          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3059          }          }
3060        if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);
3061        ecode += 2;        ecode += 2;
3062        }        }
3063      break;      break;
3064    
3065      /* Match a single character, caselessly */      /* Match a single character, caselessly. If we are at the end of the
3066        subject, give up immediately. */
3067    
3068        case OP_CHARI:
3069        if (eptr >= md->end_subject)
3070          {
3071          SCHECK_PARTIAL();
3072          RRETURN(MATCH_NOMATCH);
3073          }
3074    
3075      case OP_CHARNC:  #ifdef SUPPORT_UTF
3076  #ifdef SUPPORT_UTF8      if (utf)
     if (utf8)  
3077        {        {
3078        length = 1;        length = 1;
3079        ecode++;        ecode++;
3080        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
3081    
       if (length > md->end_subject - eptr)  
         {  
         CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */  
         MRRETURN(MATCH_NOMATCH);  
         }  
   
3082        /* 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
3083        can use the fast lookup table. */        we know that its other case must also be one byte long, so we can use the
3084          fast lookup table. We know that there is at least one byte left in the
3085          subject. */
3086    
3087        if (fc < 128)        if (fc < 128)
3088          {          {
3089          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          if (md->lcc[fc]
3090                != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3091            ecode++;
3092            eptr++;
3093          }          }
3094    
3095        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character. Note that we cannot
3096          use the value of "length" to check for sufficient bytes left, because the
3097          other case of the character may have more or fewer bytes.  */
3098    
3099        else        else
3100          {          {
# Line 2783  for (;;) Line 3110  for (;;)
3110  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3111            if (dc != UCD_OTHERCASE(fc))            if (dc != UCD_OTHERCASE(fc))
3112  #endif  #endif
3113              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3114            }            }
3115          }          }
3116        }        }
3117      else      else
3118  #endif   /* SUPPORT_UTF8 */  #endif   /* SUPPORT_UTF */
3119    
3120      /* Non-UTF-8 mode */      /* Not UTF mode */
3121        {        {
3122        if (md->end_subject - eptr < 1)        if (TABLE_GET(ecode[1], md->lcc, ecode[1])
3123          {            != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH);
3124          SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */        eptr++;
         MRRETURN(MATCH_NOMATCH);  
         }  
       if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);  
3125        ecode += 2;        ecode += 2;
3126        }        }
3127      break;      break;
# Line 2805  for (;;) Line 3129  for (;;)
3129      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3130    
3131      case OP_EXACT:      case OP_EXACT:
3132        case OP_EXACTI:
3133      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3134      ecode += 3;      ecode += 1 + IMM2_SIZE;
3135      goto REPEATCHAR;      goto REPEATCHAR;
3136    
3137      case OP_POSUPTO:      case OP_POSUPTO:
3138        case OP_POSUPTOI:
3139      possessive = TRUE;      possessive = TRUE;
3140      /* Fall through */      /* Fall through */
3141    
3142      case OP_UPTO:      case OP_UPTO:
3143        case OP_UPTOI:
3144      case OP_MINUPTO:      case OP_MINUPTO:
3145        case OP_MINUPTOI:
3146      min = 0;      min = 0;
3147      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3148      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3149      ecode += 3;      ecode += 1 + IMM2_SIZE;
3150      goto REPEATCHAR;      goto REPEATCHAR;
3151    
3152      case OP_POSSTAR:      case OP_POSSTAR:
3153        case OP_POSSTARI:
3154      possessive = TRUE;      possessive = TRUE;
3155      min = 0;      min = 0;
3156      max = INT_MAX;      max = INT_MAX;
# Line 2829  for (;;) Line 3158  for (;;)
3158      goto REPEATCHAR;      goto REPEATCHAR;
3159    
3160      case OP_POSPLUS:      case OP_POSPLUS:
3161        case OP_POSPLUSI:
3162      possessive = TRUE;      possessive = TRUE;
3163      min = 1;      min = 1;
3164      max = INT_MAX;      max = INT_MAX;
# Line 2836  for (;;) Line 3166  for (;;)
3166      goto REPEATCHAR;      goto REPEATCHAR;
3167    
3168      case OP_POSQUERY:      case OP_POSQUERY:
3169        case OP_POSQUERYI:
3170      possessive = TRUE;      possessive = TRUE;
3171      min = 0;      min = 0;
3172      max = 1;      max = 1;
# Line 2843  for (;;) Line 3174  for (;;)
3174      goto REPEATCHAR;      goto REPEATCHAR;
3175    
3176      case OP_STAR:      case OP_STAR:
3177        case OP_STARI:
3178      case OP_MINSTAR:      case OP_MINSTAR:
3179        case OP_MINSTARI:
3180      case OP_PLUS:      case OP_PLUS:
3181        case OP_PLUSI:
3182      case OP_MINPLUS:      case OP_MINPLUS:
3183        case OP_MINPLUSI:
3184      case OP_QUERY:      case OP_QUERY:
3185        case OP_QUERYI:
3186      case OP_MINQUERY:      case OP_MINQUERY:
3187      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3188        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3189      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3190      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3191      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3192      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2858  for (;;) Line 3194  for (;;)
3194      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. */
3195    
3196      REPEATCHAR:      REPEATCHAR:
3197  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3198      if (utf8)      if (utf)
3199        {        {
3200        length = 1;        length = 1;
3201        charptr = ecode;        charptr = ecode;
# Line 2873  for (;;) Line 3209  for (;;)
3209          {          {
3210  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3211          unsigned int othercase;          unsigned int othercase;
3212          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3213              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3214            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = PRIV(ord2utf)(othercase, occhars);
3215          else oclength = 0;          else oclength = 0;
3216  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3217    
3218          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3219            {            {
3220            if (eptr <= md->end_subject - length &&            if (eptr <= md->end_subject - length &&
3221              memcmp(eptr, charptr, length) == 0) eptr += length;              memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3222  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3223            else if (oclength > 0 &&            else if (oclength > 0 &&
3224                     eptr <= md->end_subject - oclength &&                     eptr <= md->end_subject - oclength &&
3225                     memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                     memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3226  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3227            else            else
3228              {              {
3229              CHECK_PARTIAL();              CHECK_PARTIAL();
3230              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3231              }              }
3232            }            }
3233    
# Line 2901  for (;;) Line 3237  for (;;)
3237            {            {
3238            for (fi = min;; fi++)            for (fi = min;; fi++)
3239              {              {
3240              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3241              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3242              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3243              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3244                memcmp(eptr, charptr, length) == 0) eptr += length;                memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3245  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3246              else if (oclength > 0 &&              else if (oclength > 0 &&
3247                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3248                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3249  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3250              else              else
3251                {                {
3252                CHECK_PARTIAL();                CHECK_PARTIAL();
3253                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3254                }                }
3255              }              }
3256            /* Control never gets here */            /* Control never gets here */
# Line 2926  for (;;) Line 3262  for (;;)
3262            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3263              {              {
3264              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
3265                  memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length;
3266  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3267              else if (oclength > 0 &&              else if (oclength > 0 &&
3268                       eptr <= md->end_subject - oclength &&                       eptr <= md->end_subject - oclength &&
3269                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;                       memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength;
3270  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3271              else              else
3272                {                {
# Line 2943  for (;;) Line 3279  for (;;)
3279    
3280            for(;;)            for(;;)
3281              {              {
3282              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3283              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3284              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { RRETURN(MATCH_NOMATCH); }
3285  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3286              eptr--;              eptr--;
3287              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2962  for (;;) Line 3298  for (;;)
3298        value of fc will always be < 128. */        value of fc will always be < 128. */
3299        }        }
3300      else      else
3301  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3302          /* When not in UTF-8 mode, load a single-byte character. */
3303      /* When not in UTF-8 mode, load a single-byte character. */        fc = *ecode++;
   
     fc = *ecode++;  
3304    
3305      /* 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
3306      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
3307      caseful cases, for speed, since matching characters is likely to be quite      caseful cases, for speed, since matching characters is likely to be quite
3308      common. First, ensure the minimum number of matches are present. If min =      common. First, ensure the minimum number of matches are present. If min =
3309      max, continue at the same level without recursing. Otherwise, if      max, continue at the same level without recursing. Otherwise, if
# Line 2980  for (;;) Line 3314  for (;;)
3314      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3315        max, eptr));        max, eptr));
3316    
3317      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3318        {        {
3319        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3320          /* fc must be < 128 if UTF is enabled. */
3321          foc = md->fcc[fc];
3322    #else
3323    #ifdef SUPPORT_UTF
3324    #ifdef SUPPORT_UCP
3325          if (utf && fc > 127)
3326            foc = UCD_OTHERCASE(fc);
3327    #else
3328          if (utf && fc > 127)
3329            foc = fc;
3330    #endif /* SUPPORT_UCP */
3331          else
3332    #endif /* SUPPORT_UTF */
3333            foc = TABLE_GET(fc, md->fcc, fc);
3334    #endif /* COMPILE_PCRE8 */
3335    
3336        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3337          {          {
3338          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3339            {            {
3340            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3341            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3342            }            }
3343          if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);          if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3344            eptr++;
3345          }          }
3346        if (min == max) continue;        if (min == max) continue;
3347        if (minimize)        if (minimize)
3348          {          {
3349          for (fi = min;; fi++)          for (fi = min;; fi++)
3350            {            {
3351            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3352            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3353            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3354            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3355              {              {
3356              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3357              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3358              }              }
3359            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH);
3360              eptr++;
3361            }            }
3362          /* Control never gets here */          /* Control never gets here */
3363          }          }
# Line 3019  for (;;) Line 3371  for (;;)
3371              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3372              break;              break;
3373              }              }
3374            if (fc != md->lcc[*eptr]) break;            if (fc != *eptr && foc != *eptr) break;
3375            eptr++;            eptr++;
3376            }            }
3377    
# Line 3027  for (;;) Line 3379  for (;;)
3379    
3380          while (eptr >= pp)          while (eptr >= pp)
3381            {            {
3382            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3383            eptr--;            eptr--;
3384            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3385            }            }
3386          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3387          }          }
3388        /* Control never gets here */        /* Control never gets here */
3389        }        }
# Line 3045  for (;;) Line 3397  for (;;)
3397          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
3398            {            {
3399            SCHECK_PARTIAL();            SCHECK_PARTIAL();
3400            MRRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
3401            }            }
3402          if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);          if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
3403          }          }
3404    
3405        if (min == max) continue;        if (min == max) continue;
# Line 3056  for (;;) Line 3408  for (;;)
3408          {          {
3409          for (fi = min;; fi++)          for (fi = min;; fi++)
3410            {            {
3411            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3412            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3413            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) RRETURN(MATCH_NOMATCH);
3414            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3415              {              {
3416              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3417              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3418              }              }
3419            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);            if (fc != *eptr++) RRETURN(MATCH_NOMATCH);
3420            }            }
3421          /* Control never gets here */          /* Control never gets here */
3422          }          }
# Line 3085  for (;;) Line 3437  for (;;)
3437    
3438          while (eptr >= pp)          while (eptr >= pp)
3439            {            {
3440            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3441            eptr--;            eptr--;
3442            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3443            }            }
3444          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3445          }          }
3446        }        }
3447      /* Control never gets here */      /* Control never gets here */
# Line 3098  for (;;) Line 3450  for (;;)
3450      checking can be multibyte. */      checking can be multibyte. */
3451    
3452      case OP_NOT:      case OP_NOT:
3453        case OP_NOTI:
3454      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3455        {        {
3456        SCHECK_PARTIAL();        SCHECK_PARTIAL();
3457        MRRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
3458        }        }
3459      ecode++;      ecode++;
3460      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3461      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3462        {        {
3463  #ifdef SUPPORT_UTF8        register int ch, och;
3464        if (c < 256)        ch = *ecode++;
3465  #endif  #ifdef COMPILE_PCRE8
3466        c = md->lcc[c];        /* ch must be < 128 if UTF is enabled. */
3467        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        och = md->fcc[ch];
3468    #else
3469    #ifdef SUPPORT_UTF
3470    #ifdef SUPPORT_UCP
3471          if (utf && ch > 127)
3472            och = UCD_OTHERCASE(ch);
3473    #else
3474          if (utf && ch > 127)
3475            och = ch;
3476    #endif /* SUPPORT_UCP */
3477          else
3478    #endif /* SUPPORT_UTF */
3479            och = TABLE_GET(ch, md->fcc, ch);
3480    #endif /* COMPILE_PCRE8 */
3481          if (ch == c || och == c) RRETURN(MATCH_NOMATCH);
3482        }        }
3483      else      else    /* Caseful */
3484        {        {
3485        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);
3486        }        }
3487      break;      break;
3488    
# Line 3127  for (;;) Line 3494  for (;;)
3494      about... */      about... */
3495    
3496      case OP_NOTEXACT:      case OP_NOTEXACT:
3497        case OP_NOTEXACTI:
3498      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3499      ecode += 3;      ecode += 1 + IMM2_SIZE;
3500      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3501    
3502      case OP_NOTUPTO:      case OP_NOTUPTO:
3503        case OP_NOTUPTOI:
3504      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3505        case OP_NOTMINUPTOI:
3506      min = 0;      min = 0;
3507      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3508      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3509      ecode += 3;      ecode += 1 + IMM2_SIZE;
3510      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3511    
3512      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3513        case OP_NOTPOSSTARI:
3514      possessive = TRUE;      possessive = TRUE;
3515      min = 0;      min = 0;
3516      max = INT_MAX;      max = INT_MAX;
# Line 3147  for (;;) Line 3518  for (;;)
3518      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3519    
3520      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3521        case OP_NOTPOSPLUSI:
3522      possessive = TRUE;      possessive = TRUE;
3523      min = 1;      min = 1;
3524      max = INT_MAX;      max = INT_MAX;
# Line 3154  for (;;) Line 3526  for (;;)
3526      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3527    
3528      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3529        case OP_NOTPOSQUERYI:
3530      possessive = TRUE;      possessive = TRUE;
3531      min = 0;      min = 0;
3532      max = 1;      max = 1;
# Line 3161  for (;;) Line 3534  for (;;)
3534      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3535    
3536      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3537        case OP_NOTPOSUPTOI:
3538      possessive = TRUE;      possessive = TRUE;
3539      min = 0;      min = 0;
3540      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3541      ecode += 3;      ecode += 1 + IMM2_SIZE;
3542      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3543    
3544      case OP_NOTSTAR:      case OP_NOTSTAR:
3545        case OP_NOTSTARI:
3546      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3547        case OP_NOTMINSTARI:
3548      case OP_NOTPLUS:      case OP_NOTPLUS:
3549        case OP_NOTPLUSI:
3550      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3551        case OP_NOTMINPLUSI:
3552      case OP_NOTQUERY:      case OP_NOTQUERY:
3553        case OP_NOTQUERYI:
3554      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3555      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3556        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3557      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3558      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3559      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3195  for (;;) Line 3575  for (;;)
3575      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,
3576        max, eptr));        max, eptr));
3577    
3578      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3579        {        {
3580        fc = md->lcc[fc];  #ifdef COMPILE_PCRE8
3581          /* fc must be < 128 if UTF is enabled. */
3582          foc = md->fcc[fc];
3583    #else
3584    #ifdef SUPPORT_UTF
3585    #ifdef SUPPORT_UCP
3586          if (utf && fc > 127)
3587            foc = UCD_OTHERCASE(fc);
3588    #else
3589          if (utf && fc > 127)
3590            foc = fc;
3591    #endif /* SUPPORT_UCP */
3592          else
3593    #endif /* SUPPORT_UTF */
3594            foc = TABLE_GET(fc, md->fcc, fc);
3595    #endif /* COMPILE_PCRE8 */
3596    
3597  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3598        /* UTF-8 mode */        if (utf)
       if (utf8)  
3599          {          {
3600          register unsigned int d;          register unsigned int d;
3601          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3209  for (;;) Line 3603  for (;;)
3603            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3604              {              {
3605              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3606              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3607              }              }
3608            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3609            if (d < 256) d = md->lcc[d];            if (fc == d || foc == d) RRETURN(MATCH_NOMATCH);
           if (fc == d) MRRETURN(MATCH_NOMATCH);  
3610            }            }
3611          }          }
3612        else        else
3613  #endif  #endif
3614          /* Not UTF mode */
       /* Not UTF-8 mode */  
3615          {          {
3616          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3617            {            {
3618            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3619              {              {
3620              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3621              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3622              }              }
3623            if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);            if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3624              eptr++;
3625            }            }
3626          }          }
3627    
# Line 3236  for (;;) Line 3629  for (;;)
3629    
3630        if (minimize)        if (minimize)
3631          {          {
3632  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3633          /* UTF-8 mode */          if (utf)
         if (utf8)  
3634            {            {
3635            register unsigned int d;            register unsigned int d;
3636            for (fi = min;; fi++)            for (fi = min;; fi++)
3637              {              {
3638              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3639              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3640              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3641              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3642                {                {
3643                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3644                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3645                }                }
3646              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3647              if (d < 256) d = md->lcc[d];              if (fc == d || foc == d) RRETURN(MATCH_NOMATCH);
             if (fc == d) MRRETURN(MATCH_NOMATCH);  
3648              }              }
3649            }            }
3650          else          else
3651  #endif  #endif
3652          /* Not UTF-8 mode */          /* Not UTF mode */
3653            {            {
3654            for (fi = min;; fi++)            for (fi = min;; fi++)
3655              {              {
3656              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3657              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3658              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3659              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3660                {                {
3661                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3662                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3663                }                }
3664              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);              if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH);
3665                eptr++;
3666              }              }
3667            }            }
3668          /* Control never gets here */          /* Control never gets here */
# Line 3282  for (;;) Line 3674  for (;;)
3674          {          {
3675          pp = eptr;          pp = eptr;
3676    
3677  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3678          /* UTF-8 mode */          if (utf)
         if (utf8)  
3679            {            {
3680            register unsigned int d;            register unsigned int d;
3681            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3296  for (;;) Line 3687  for (;;)
3687                break;                break;
3688                }                }
3689              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3690              if (d < 256) d = md->lcc[d];              if (fc == d || foc == d) break;
             if (fc == d) break;  
3691              eptr += len;              eptr += len;
3692              }              }
3693          if (possessive) continue;          if (possessive) continue;
3694          for(;;)          for(;;)
3695              {              {
3696              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3697              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3698              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3699              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3311  for (;;) Line 3701  for (;;)
3701            }            }
3702          else          else
3703  #endif  #endif
3704          /* Not UTF-8 mode */          /* Not UTF mode */
3705            {            {
3706            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3707              {              {
# Line 3320  for (;;) Line 3710  for (;;)
3710                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3711                break;                break;
3712                }                }
3713              if (fc == md->lcc[*eptr]) break;              if (fc == *eptr || foc == *eptr) break;
3714              eptr++;              eptr++;
3715              }              }
3716            if (possessive) continue;            if (possessive) continue;
3717            while (eptr >= pp)            while (eptr >= pp)
3718              {              {
3719              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3720              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3721              eptr--;              eptr--;
3722              }              }
3723            }            }
3724    
3725          MRRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
3726          }          }
3727        /* Control never gets here */        /* Control never gets here */
3728        }        }
# Line 3341  for (;;) Line 3731  for (;;)
3731    
3732      else      else
3733        {        {
3734  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3735        /* UTF-8 mode */        if (utf)
       if (utf8)  
3736          {          {
3737          register unsigned int d;          register unsigned int d;
3738          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3351  for (;;) Line 3740  for (;;)
3740            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3741              {              {
3742              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3743              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3744              }              }
3745            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3746            if (fc == d) MRRETURN(MATCH_NOMATCH);            if (fc == d) RRETURN(MATCH_NOMATCH);
3747            }            }
3748          }          }
3749        else        else
3750  #endif  #endif
3751        /* Not UTF-8 mode */        /* Not UTF mode */
3752          {          {
3753          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3754            {            {
3755            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
3756              {              {
3757              SCHECK_PARTIAL();              SCHECK_PARTIAL();
3758              MRRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3759              }              }
3760            if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);
3761            }            }
3762          }          }
3763    
# Line 3376  for (;;) Line 3765  for (;;)
3765    
3766        if (minimize)        if (minimize)
3767          {          {
3768  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3769          /* UTF-8 mode */          if (utf)
         if (utf8)  
3770            {            {
3771            register unsigned int d;            register unsigned int d;
3772            for (fi = min;; fi++)            for (fi = min;; fi++)
3773              {              {
3774              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3775              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3776              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3777              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3778                {                {
3779                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3780                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3781                }                }
3782              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3783              if (fc == d) MRRETURN(MATCH_NOMATCH);              if (fc == d) RRETURN(MATCH_NOMATCH);
3784              }              }
3785            }            }
3786          else          else
3787  #endif  #endif
3788          /* Not UTF-8 mode */          /* Not UTF mode */
3789            {            {
3790            for (fi = min;; fi++)            for (fi = min;; fi++)
3791              {              {
3792              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3793              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3794              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) RRETURN(MATCH_NOMATCH);
3795              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3796                {                {
3797                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3798                MRRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
3799                }                }
3800              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);              if (fc == *eptr++) RRETURN(MATCH_NOMATCH);
3801              }              }
3802            }            }
3803          /* Control never gets here */          /* Control never gets here */
# Line 3421  for (;;) Line 3809  for (;;)
3809          {          {
3810          pp = eptr;          pp = eptr;
3811    
3812  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3813          /* UTF-8 mode */          if (utf)
         if (utf8)  
3814            {            {
3815            register unsigned int d;            register unsigned int d;
3816            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3441  for (;;) Line 3828  for (;;)
3828            if (possessive) continue;            if (possessive) continue;
3829            for(;;)            for(;;)
3830              {              {
3831              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3832              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3833              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3834              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3449  for (;;) Line 3836  for (;;)
3836            }            }
3837          else          else
3838  #endif  #endif
3839          /* Not UTF-8 mode */          /* Not UTF mode */
3840            {            {
3841            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3842              {              {
# Line 3464  for (;;) Line 3851  for (;;)
3851            if (possessive) continue;