/[pcre]/code/trunk/pcre_exec.c
ViewVC logotype

Diff of /code/trunk/pcre_exec.c

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

revision 530 by ph10, Tue Jun 1 13:42:06 2010 UTC revision 626 by ph10, Wed Jul 20 17:51:54 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    #define MATCH_THEN         (-992)
84    
85  /* This is a convenience macro for code that occurs many times. */  /* This is a convenience macro for code that occurs many times. */
86    
# Line 132  while (length-- > 0) Line 136  while (length-- > 0)
136  *          Match a back-reference                *  *          Match a back-reference                *
137  *************************************************/  *************************************************/
138    
139  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
140  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
141    the length passed is zero. Note that in caseless UTF-8 mode, the number of
142    subject bytes matched may be different to the number of reference bytes.
143    
144  Arguments:  Arguments:
145    offset      index into the offset vector    offset      index into the offset vector
146    eptr        points into the subject    eptr        pointer into the subject
147    length      length to be matched    length      length of reference to be matched (number of bytes)
148    md          points to match data block    md          points to match data block
149    ims         the ims flags    caseless    TRUE if caseless
150    
151  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
152  */  */
153    
154  static BOOL  static int
155  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
156    unsigned long int ims)    BOOL caseless)
157  {  {
158  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
159    register USPTR p = md->start_subject + md->offset_vector[offset];
160    
161  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
162  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 164  pchars(p, length, FALSE, md); Line 171  pchars(p, length, FALSE, md);
171  printf("\n");  printf("\n");
172  #endif  #endif
173    
174  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
175    
176  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
177    
178  /* Separate the caseless case for speed. In UTF-8 mode we can only do this  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
179  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
180  ASCII characters. */  ASCII characters. */
181    
182  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
183    {    {
184  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
185  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
186    if (md->utf8)    if (md->utf8)
187      {      {
188      USPTR endptr = eptr + length;      /* Match characters up to the end of the reference. NOTE: the number of
189      while (eptr < endptr)      bytes matched may differ, because there are some characters whose upper and
190        lower case versions code as different numbers of bytes. For example, U+023A
191        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
192        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
193        the latter. It is important, therefore, to check the length along the
194        reference, not along the subject (earlier code did this wrong). */
195    
196        USPTR endptr = p + length;
197        while (p < endptr)
198        {        {
199        int c, d;        int c, d;
200          if (eptr >= md->end_subject) return -1;
201        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
202        GETCHARINC(d, p);        GETCHARINC(d, p);
203        if (c != d && c != UCD_OTHERCASE(d)) return FALSE;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
204        }        }
205      }      }
206    else    else
# Line 193  if ((ims & PCRE_CASELESS) != 0) Line 209  if ((ims & PCRE_CASELESS) != 0)
209    
210    /* 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
211    is no UCP support. */    is no UCP support. */
212        {
213    while (length-- > 0)      if (eptr + length > md->end_subject) return -1;
214      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }      while (length-- > 0)
215          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
216        }
217    }    }
218    
219  /* 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
220  are in UTF-8 mode. */  are in UTF-8 mode. */
221    
222  else  else
223    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
224      if (eptr + length > md->end_subject) return -1;
225      while (length-- > 0) if (*p++ != *eptr++) return -1;
226      }
227    
228  return TRUE;  return eptr - eptr_start;
229  }  }
230    
231    
# Line 256  enum { RM1=1, RM2, RM3, RM4, RM5, RM Line 277  enum { RM1=1, RM2, RM3, RM4, RM5, RM
277         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
278         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
279         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280         RM61,  RM62 };         RM61,  RM62, RM63 };
281    
282  /* 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
283  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
# Line 266  actually used in this definition. */ Line 287  actually used in this definition. */
287  #define REGISTER register  #define REGISTER register
288    
289  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
290  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
291    { \    { \
292    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
293    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
294    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
295    }    }
296  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 278  actually used in this definition. */ Line 299  actually used in this definition. */
299    return ra; \    return ra; \
300    }    }
301  #else  #else
302  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
303    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
304  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
305  #endif  #endif
306    
# Line 292  argument of match(), which never changes Line 313  argument of match(), which never changes
313    
314  #define REGISTER  #define REGISTER
315    
316  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
317    {\    {\
318    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
319      if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
320    frame->Xwhere = rw; \    frame->Xwhere = rw; \
321    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
322    newframe->Xecode = rb;\    newframe->Xecode = rb;\
323    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
324    newframe->Xmarkptr = markptr;\    newframe->Xmarkptr = markptr;\
325    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
326    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
327    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
328    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
329    frame = newframe;\    frame = newframe;\
# Line 339  typedef struct heapframe { Line 359  typedef struct heapframe {
359    USPTR Xmstart;    USPTR Xmstart;
360    USPTR Xmarkptr;    USPTR Xmarkptr;
361    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
362    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
363    unsigned int Xrdepth;    unsigned int Xrdepth;
364    
365    /* Function local variables */    /* Function local variables */
# Line 362  typedef struct heapframe { Line 380  typedef struct heapframe {
380    BOOL Xcondition;    BOOL Xcondition;
381    BOOL Xprev_is_word;    BOOL Xprev_is_word;
382    
   unsigned long int Xoriginal_ims;  
   
383  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
386    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
387    int Xoclength;    int Xoclength;
388    uschar Xocchars[8];    uschar Xocchars[8];
389  #endif  #endif
# Line 421  immediately. The second one is used when Line 434  immediately. The second one is used when
434  the subject. */  the subject. */
435    
436  #define CHECK_PARTIAL()\  #define CHECK_PARTIAL()\
437    if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\    if (md->partial != 0 && eptr >= md->end_subject && \
438      {\        eptr > md->start_used_ptr) \
439      md->hitend = TRUE;\      { \
440      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\      md->hitend = TRUE; \
441        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
442      }      }
443    
444  #define SCHECK_PARTIAL()\  #define SCHECK_PARTIAL()\
445    if (md->partial != 0 && eptr > mstart)\    if (md->partial != 0 && eptr > md->start_used_ptr) \
446      {\      { \
447      md->hitend = TRUE;\      md->hitend = TRUE; \
448      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\      if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
449      }      }
450    
451    
# Line 448  Arguments: Line 462  Arguments:
462     markptr     pointer to the most recent MARK name, or NULL     markptr     pointer to the most recent MARK name, or NULL
463     offset_top  current top pointer     offset_top  current top pointer
464     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
465     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
466                   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  
467     rdepth      the recursion depth     rdepth      the recursion depth
468    
469  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
# Line 466  Returns: MATCH_MATCH if matched Line 475  Returns: MATCH_MATCH if matched
475    
476  static int  static int
477  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
478    const uschar *markptr, int offset_top, match_data *md, unsigned long int ims,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
479    eptrblock *eptrb, int flags, unsigned int rdepth)    unsigned int rdepth)
480  {  {
481  /* 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,
482  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 479  register unsigned int c; /* Character Line 488  register unsigned int c; /* Character
488  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
489    
490  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
491    BOOL caseless;
492  int condcode;  int condcode;
493    
494  /* 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 487  heap storage. Set up the top-level frame Line 497  heap storage. Set up the top-level frame
497  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
498    
499  #ifdef NO_RECURSE  #ifdef NO_RECURSE
500  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
501    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
502  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
503    
504  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 497  frame->Xecode = ecode; Line 508  frame->Xecode = ecode;
508  frame->Xmstart = mstart;  frame->Xmstart = mstart;
509  frame->Xmarkptr = markptr;  frame->Xmarkptr = markptr;
510  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
511  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
512  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
513    
514  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 513  HEAP_RECURSE: Line 522  HEAP_RECURSE:
522  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
523  #define markptr            frame->Xmarkptr  #define markptr            frame->Xmarkptr
524  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
525  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
526  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
527    
528  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 537  HEAP_RECURSE: Line 544  HEAP_RECURSE:
544  #define condition          frame->Xcondition  #define condition          frame->Xcondition
545  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
546    
 #define original_ims       frame->Xoriginal_ims  
   
547  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
548  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
549  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
550  #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  
551  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
552  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
553  #endif  #endif
# Line 575  i, and fc and c, can be the same variabl Line 577  i, and fc and c, can be the same variabl
577  #define fi i  #define fi i
578  #define fc c  #define fc c
579    
580    /* Many of the following variables are used only in small blocks of the code.
581    My normal style of coding would have declared them within each of those blocks.
582    However, in order to accommodate the version of this code that uses an external
583    "stack" implemented on the heap, it is easier to declare them all here, so the
584    declarations can be cut out in a block. The only declarations within blocks
585    below are for variables that do not have to be preserved over a recursive call
586    to RMATCH(). */
587    
588  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  #ifdef SUPPORT_UTF8
589  const uschar *charptr;             /* in small blocks of the code. My normal */  const uschar *charptr;
590  #endif                             /* style of coding would have declared    */  #endif
591  const uschar *callpat;             /* them within each of those blocks.      */  const uschar *callpat;
592  const uschar *data;                /* However, in order to accommodate the   */  const uschar *data;
593  const uschar *next;                /* version of this code that uses an      */  const uschar *next;
594  USPTR         pp;                  /* external "stack" implemented on the    */  USPTR         pp;
595  const uschar *prev;                /* heap, it is easier to declare them all */  const uschar *prev;
596  USPTR         saved_eptr;          /* here, so the declarations can be cut   */  USPTR         saved_eptr;
597                                     /* out in a block. The only declarations  */  
598  recursion_info new_recursive;      /* within blocks below are for variables  */  recursion_info new_recursive;
599                                     /* that do not have to be preserved over  */  
600  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  BOOL cur_is_word;
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
 unsigned long int original_ims;  
   
604  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
608  int oclength;  int oclength;
609  uschar occhars[8];  uschar occhars[8];
610  #endif  #endif
# Line 620  int stacksave[REC_STACK_SAVE_MAX]; Line 624  int stacksave[REC_STACK_SAVE_MAX];
624  eptrblock newptrb;  eptrblock newptrb;
625  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
626    
627    /* To save space on the stack and in the heap frame, I have doubled up on some
628    of the local variables that are used only in localised parts of the code, but
629    still need to be preserved over recursive calls of match(). These macros define
630    the alternative names that are used. */
631    
632    #define allow_zero    cur_is_word
633    #define cbegroup      condition
634    #define code_offset   codelink
635    #define condassert    condition
636    #define matched_once  prev_is_word
637    
638  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
639  variables. */  variables. */
640    
# Line 656  haven't exceeded the recursive call limi Line 671  haven't exceeded the recursive call limi
671  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
672  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
673    
 original_ims = ims;    /* Save for resetting on ')' */  
   
674  /* 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
675  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
676  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
677  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
678  When match() is called in other circumstances, don't add to the chain. The  
679  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
680  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
681  match(). */  to break infinite loops that match no characters. When match() is called in
682    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
683    NOT be used with tail recursion, because the memory block that is used is on
684    the stack, so a new one may be required for each match(). */
685    
686  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
687    {    {
688    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
689    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
690    eptrb = &newptrb;    eptrb = &newptrb;
691      md->match_function_type = 0;
692    }    }
693    
694  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 686  for (;;) Line 703  for (;;)
703      case OP_MARK:      case OP_MARK:
704      markptr = ecode + 2;      markptr = ecode + 2;
705      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
706        ims, eptrb, flags, RM55);        eptrb, RM55);
707    
708      /* 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
709      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 708  for (;;) Line 725  for (;;)
725      case OP_FAIL:      case OP_FAIL:
726      MRRETURN(MATCH_NOMATCH);      MRRETURN(MATCH_NOMATCH);
727    
728        /* COMMIT overrides PRUNE, SKIP, and THEN */
729    
730      case OP_COMMIT:      case OP_COMMIT:
731      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
732        ims, eptrb, flags, RM52);        eptrb, RM52);
733      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
734            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
735            rrc != MATCH_THEN)
736          RRETURN(rrc);
737      MRRETURN(MATCH_COMMIT);      MRRETURN(MATCH_COMMIT);
738    
739        /* PRUNE overrides THEN */
740    
741      case OP_PRUNE:      case OP_PRUNE:
742      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
743        ims, eptrb, flags, RM51);        eptrb, RM51);
744      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
745      MRRETURN(MATCH_PRUNE);      MRRETURN(MATCH_PRUNE);
746    
747      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
748      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
749        ims, eptrb, flags, RM56);        eptrb, RM56);
750      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
751      md->mark = ecode + 2;      md->mark = ecode + 2;
752      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
753    
754        /* SKIP overrides PRUNE and THEN */
755    
756      case OP_SKIP:      case OP_SKIP:
757      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
758        ims, eptrb, flags, RM53);        eptrb, RM53);
759      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
760          RRETURN(rrc);
761      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
762      MRRETURN(MATCH_SKIP);      MRRETURN(MATCH_SKIP);
763    
764      case OP_SKIP_ARG:      case OP_SKIP_ARG:
765      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
766        ims, eptrb, flags, RM57);        eptrb, RM57);
767      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769    
770      /* 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
771      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
# Line 747  for (;;) Line 775  for (;;)
775      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
776      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
777    
778        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
779        the alt that is at the start of the current branch. This makes it possible
780        to skip back past alternatives that precede the THEN within the current
781        branch. */
782    
783      case OP_THEN:      case OP_THEN:
784      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_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);
788      MRRETURN(MATCH_THEN);      MRRETURN(MATCH_THEN);
789    
790      case OP_THEN_ARG:      case OP_THEN_ARG:
791      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
792        ims, eptrb, flags, RM58);        offset_top, md, eptrb, RM58);
793      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
794      md->mark = ecode + 2;      md->start_match_ptr = ecode - GET(ecode, 1);
795        md->mark = ecode + LINK_SIZE + 2;
796      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
797    
798      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle a capturing bracket, other than those that are possessive with an
799      the current subject position in the working slot at the top of the vector.      unlimited repeat. If there is space in the offset vector, save the current
800      We mustn't change the current values of the data slot, because they may be      subject position in the working slot at the top of the vector. We mustn't
801      set from a previous iteration of this group, and be referred to by a      change the current values of the data slot, because they may be set from a
802      reference inside the group.      previous iteration of this group, and be referred to by a reference inside
803        the group. A failure to match might occur after the group has succeeded,
804      If the bracket fails to match, we need to restore this value and also the      if something later on doesn't match. For this reason, we need to restore
805      values of the final offsets, in case they were set by a previous iteration      the working value and also the values of the final offsets, in case they
806      of the same bracket.      were set by a previous iteration of the same bracket.
807    
808      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
809      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 794  for (;;) Line 829  for (;;)
829        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
830    
831        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
832        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
833          (int)(eptr - md->start_subject);          (int)(eptr - md->start_subject);
834    
835        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
836          {          {
837            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839            ims, eptrb, flags, RM1);            eptrb, RM1);
840          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841            if (rrc != MATCH_NOMATCH &&
842                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
843              RRETURN(rrc);
844          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
845          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
846            if (*ecode != OP_ALT) break;
847          }          }
       while (*ecode == OP_ALT);  
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854        if (rrc != MATCH_THEN) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855        RRETURN(MATCH_NOMATCH);        MATCH_THEN. */
856    
857          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
858          RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
859        }        }
860    
861      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 829  for (;;) Line 869  for (;;)
869      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
871    
872      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
873      final alternative within the brackets, we would return the result of a      repeat. Loop for all the alternatives. When we get to the final alternative
874      recursive call to match() whatever happened. We can reduce stack usage by      within the brackets, we used to return the result of a recursive call to
875      turning this into a tail recursion, except in the case when match_cbegroup      match() whatever happened so it was possible to reduce stack usage by
876      is set.*/      turning this into a tail recursion, except in the case of a possibly empty
877        group. However, now that there is the possiblity of (*THEN) occurring in
878        the final alternative, this optimization is no longer possible.
879    
880        MATCH_ONCE is returned when the end of an atomic group is successfully
881        reached, but subsequent matching fails. It passes back up the tree (causing
882        captured values to be reset) until the original atomic group level is
883        reached. This is tested by comparing md->once_target with the start of the
884        group. At this point, the return is converted into MATCH_NOMATCH so that
885        previous backup points can be taken. */
886    
887        case OP_ONCE:
888      case OP_BRA:      case OP_BRA:
889      case OP_SBRA:      case OP_SBRA:
890      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
891      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
892      for (;;)      for (;;)
893        {        {
894        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
895          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
896            RM2);
897          if (rrc != MATCH_NOMATCH &&
898              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
899          {          {
900          if (flags == 0)    /* Not a possibly empty group */          if (rrc == MATCH_ONCE)
901            {            {
902            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
903            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
904            goto TAIL_RECURSE;              {
905                while (*scode == OP_ALT) scode += GET(scode, 1);
906                scode -= GET(scode, 1);
907                }
908              if (md->once_target == scode) rrc = MATCH_NOMATCH;
909            }            }
910            RRETURN(rrc);
911            }
912          ecode += GET(ecode, 1);
913          if (*ecode != OP_ALT) break;
914          }
915        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
916        RRETURN(MATCH_NOMATCH);
917    
918          /* Possibly empty group; can't use tail recursion. */      /* Handle possessive capturing brackets with an unlimited repeat. We come
919        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
920        handled similarly to the normal case above. However, the matching is
921        different. The end of these brackets will always be OP_KETRPOS, which
922        returns MATCH_KETRPOS without going further in the pattern. By this means
923        we can handle the group by iteration rather than recursion, thereby
924        reducing the amount of stack needed. */
925    
926        case OP_CBRAPOS:
927        case OP_SCBRAPOS:
928        allow_zero = FALSE;
929    
930          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      POSSESSIVE_CAPTURE:
931            eptrb, flags, RM48);      number = GET2(ecode, 1+LINK_SIZE);
932          if (rrc == MATCH_NOMATCH) md->mark = markptr;      offset = number << 1;
933          RRETURN(rrc);  
934    #ifdef PCRE_DEBUG
935        printf("start possessive bracket %d\n", number);
936        printf("subject=");
937        pchars(eptr, 16, TRUE, md);
938        printf("\n");
939    #endif
940    
941        if (offset < md->offset_max)
942          {
943          matched_once = FALSE;
944          code_offset = ecode - md->start_code;
945    
946          save_offset1 = md->offset_vector[offset];
947          save_offset2 = md->offset_vector[offset+1];
948          save_offset3 = md->offset_vector[md->offset_end - number];
949          save_capture_last = md->capture_last;
950    
951          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
952    
953          /* Each time round the loop, save the current subject position for use
954          when the group matches. For MATCH_MATCH, the group has matched, so we
955          restart it with a new subject starting position, remembering that we had
956          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
957          usual. If we haven't matched any alternatives in any iteration, check to
958          see if a previous iteration matched. If so, the group has matched;
959          continue from afterwards. Otherwise it has failed; restore the previous
960          capture values before returning NOMATCH. */
961    
962          for (;;)
963            {
964            md->offset_vector[md->offset_end - number] =
965              (int)(eptr - md->start_subject);
966            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
967            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
968              eptrb, RM63);
969            if (rrc == MATCH_KETRPOS)
970              {
971              offset_top = md->end_offset_top;
972              eptr = md->end_match_ptr;
973              ecode = md->start_code + code_offset;
974              save_capture_last = md->capture_last;
975              matched_once = TRUE;
976              continue;
977              }
978            if (rrc != MATCH_NOMATCH &&
979                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
980              RRETURN(rrc);
981            md->capture_last = save_capture_last;
982            ecode += GET(ecode, 1);
983            if (*ecode != OP_ALT) break;
984            }
985    
986          if (!matched_once)
987            {
988            md->offset_vector[offset] = save_offset1;
989            md->offset_vector[offset+1] = save_offset2;
990            md->offset_vector[md->offset_end - number] = save_offset3;
991          }          }
992    
993        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
994        otherwise return. */        if (allow_zero || matched_once)
995            {
996            ecode += 1 + LINK_SIZE;
997            break;
998            }
999    
1000        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        RRETURN(MATCH_NOMATCH);
1001          eptrb, flags, RM2);        }
1002        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);  
1003        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1004        as a non-capturing bracket. */
1005    
1006        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1007        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008    
1009        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1010    
1011        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1012        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1013    
1014        /* Non-capturing possessive bracket with unlimited repeat. We come here
1015        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1016        without the capturing complication. It is written out separately for speed
1017        and cleanliness. */
1018    
1019        case OP_BRAPOS:
1020        case OP_SBRAPOS:
1021        allow_zero = FALSE;
1022    
1023        POSSESSIVE_NON_CAPTURE:
1024        matched_once = FALSE;
1025        code_offset = ecode - md->start_code;
1026    
1027        for (;;)
1028          {
1029          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1030          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1031            eptrb, RM48);
1032          if (rrc == MATCH_KETRPOS)
1033            {
1034            offset_top = md->end_offset_top;
1035            eptr = md->end_match_ptr;
1036            ecode = md->start_code + code_offset;
1037            matched_once = TRUE;
1038            continue;
1039            }
1040          if (rrc != MATCH_NOMATCH &&
1041              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1042            RRETURN(rrc);
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045        }        }
1046    
1047        if (matched_once || allow_zero)
1048          {
1049          ecode += 1 + LINK_SIZE;
1050          break;
1051          }
1052        RRETURN(MATCH_NOMATCH);
1053    
1054      /* Control never reaches here. */      /* Control never reaches here. */
1055    
1056      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1057      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
1058      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
1059      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. */  
1060    
1061      case OP_COND:      case OP_COND:
1062      case OP_SCOND:      case OP_SCOND:
1063      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1064    
1065      /* 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
1066      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 1051  for (;;) Line 1235  for (;;)
1235        }        }
1236    
1237      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1238      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
1239      assertion. */      an assertion. */
1240    
1241      else      else
1242        {        {
1243        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1244            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1245        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1246          {          {
1247            if (md->end_offset_top > offset_top)
1248              offset_top = md->end_offset_top;  /* Captures may have happened */
1249          condition = TRUE;          condition = TRUE;
1250          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1251          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1252          }          }
1253        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH &&
1254                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1255          {          {
1256          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1257          }          }
# Line 1076  for (;;) Line 1263  for (;;)
1263        }        }
1264    
1265      /* 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,
1266      we can use tail recursion to avoid using another stack frame, except when      we used to use tail recursion to avoid using another stack frame, except
1267      match_cbegroup is required for an unlimited repeat of a possibly empty      when there was unlimited repeat of a possibly empty group. However, that
1268      group. If the second alternative doesn't exist, we can just plough on. */      strategy no longer works because of the possibilty of (*THEN) being
1269        encountered in the branch. A recursive call to match() is always required,
1270        unless the second alternative doesn't exist, in which case we can just
1271        plough on. */
1272    
1273      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1274        {        {
1275        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1276        if (op == OP_SCOND)        /* Possibly empty group */        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1277          {        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1278          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          rrc = MATCH_NOMATCH;
1279          RRETURN(rrc);        RRETURN(rrc);
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
         goto TAIL_RECURSE;  
         }  
1280        }        }
1281      else                         /* Condition false & no alternative */      else                         /* Condition false & no alternative */
1282        {        {
# Line 1125  for (;;) Line 1309  for (;;)
1309      break;      break;
1310    
1311    
1312      /* 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. */  
1313    
     case OP_ACCEPT:  
1314      case OP_END:      case OP_END:
1315      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1316        {      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;  
       }  
1317    
1318      /* 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
1319      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
1320      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,
1321      if any. */      backtracking will then try other alternatives, if any. */
1322    
1323      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1324          (md->notempty ||           md->recursive == NULL &&
1325            (md->notempty_atstart &&           (md->notempty ||
1326              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1327                 mstart == md->start_subject + md->start_offset)))
1328        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1329    
1330      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1167  for (;;) Line 1339  for (;;)
1339      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1340      MRRETURN(rrc);      MRRETURN(rrc);
1341    
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
   
1342      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1343      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,
1344      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
1345      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
1346      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1347        of a condition, we want to return immediately afterwards. The caller of
1348        this incarnation of the match() function will have set MATCH_CONDASSERT in
1349        md->match_function type, and one of these opcodes will be the first opcode
1350        that is processed. We use a local variable that is preserved over calls to
1351        match() to remember this case. */
1352    
1353      case OP_ASSERT:      case OP_ASSERT:
1354      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1355        if (md->match_function_type == MATCH_CONDASSERT)
1356          {
1357          condassert = TRUE;
1358          md->match_function_type = 0;
1359          }
1360        else condassert = FALSE;
1361    
1362      do      do
1363        {        {
1364        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1365        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1366          {          {
1367          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1368          break;          break;
1369          }          }
1370        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH &&
1371              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1372            RRETURN(rrc);
1373        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1374        }        }
1375      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1376    
1377      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1378    
1379      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1380    
1381      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1382    
1383      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1384      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1216  for (;;) Line 1394  for (;;)
1394    
1395      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1396      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1397        if (md->match_function_type == MATCH_CONDASSERT)
1398          {
1399          condassert = TRUE;
1400          md->match_function_type = 0;
1401          }
1402        else condassert = FALSE;
1403    
1404      do      do
1405        {        {
1406        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
         RM5);  
1407        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1408        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1409          {          {
1410          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1411          break;          break;
1412          }          }
1413        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH &&
1414              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1415            RRETURN(rrc);
1416        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1417        }        }
1418      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1419    
1420      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1421    
1422      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1423      continue;      continue;
# Line 1299  for (;;) Line 1485  for (;;)
1485      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
1486      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1487    
1488      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1489      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
1490      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
1491      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
1492      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
1493      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
1494      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.  
1495    
1496      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
1497      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
1498      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1499        a lot, so he is not to blame for the current way it works. */
1500    
1501      case OP_RECURSE:      case OP_RECURSE:
1502        {        {
# Line 1325  for (;;) Line 1509  for (;;)
1509        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1510        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1511    
1512        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1513    
1514        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1515    
1516        /* Now save the offset data. */        /* Now save the offset data */
1517    
1518        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1519        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1341  for (;;) Line 1524  for (;;)
1524            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1525          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1526          }          }
   
1527        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1528              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1529    
1530        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1531        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1532          might be changed, so reset it before looping. */
1533    
1534        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1535        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1536        do        do
1537          {          {
1538            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1539          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1540            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1541            memcpy(md->offset_vector, new_recursive.offset_save,
1542                new_recursive.saved_max * sizeof(int));
1543          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1544            {            {
1545            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1546            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1547            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1548              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1549            MRRETURN(MATCH_MATCH);  
1550              /* Set where we got to in the subject, and reset the start in case
1551              it was changed by \K. This *is* propagated back out of a recursion,
1552              for Perl compatibility. */
1553    
1554              eptr = md->end_match_ptr;
1555              mstart = md->start_match_ptr;
1556              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1557            }            }
1558          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH &&
1559                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1560            {            {
1561            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1562            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
# Line 1372  for (;;) Line 1565  for (;;)
1565            }            }
1566    
1567          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1568          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1569          }          }
1570        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1384  for (;;) Line 1575  for (;;)
1575          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1576        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1577        }        }
     /* 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) 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;  
       }  
1578    
1579      /* The repeating kets try the rest of the pattern or restart from the      RECURSION_MATCHED:
1580      preceding bracket, in the appropriate order. The second "call" of match()      break;
     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));  
       }  
   
     if (*ecode == OP_KETRMIN)  
       {  
       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 */  
1581    
1582      /* 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
1583      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1479  for (;;) Line 1593  for (;;)
1593      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1594    
1595      case OP_BRAZERO:      case OP_BRAZERO:
1596        {      next = ecode + 1;
1597        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1598        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1599        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1600        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1601      break;      break;
1602    
1603      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1604        {      next = ecode + 1;
1605        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1606        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1607        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1608        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1609      break;      break;
1610    
1611      case OP_SKIPZERO:      case OP_SKIPZERO:
1612        {      next = ecode+1;
1613        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1614        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1615      break;      break;
1616    
1617        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1618        here; just jump to the group, with allow_zero set TRUE. */
1619    
1620        case OP_BRAPOSZERO:
1621        op = *(++ecode);
1622        allow_zero = TRUE;
1623        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1624          goto POSSESSIVE_NON_CAPTURE;
1625    
1626      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1627    
1628      case OP_KET:      case OP_KET:
1629      case OP_KETRMIN:      case OP_KETRMIN:
1630      case OP_KETRMAX:      case OP_KETRMAX:
1631        case OP_KETRPOS:
1632      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1633    
1634      /* 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
1635      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1636      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1637    
1638      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1639        {        {
1640        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1641        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1642        }        }
1643      else saved_eptr = NULL;      else saved_eptr = NULL;
1644    
1645      /* 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, stop matching and return
1646      matching and return MATCH_MATCH, but record the current high water mark for      MATCH_MATCH, but record the current high water mark for use by positive
1647      use by positive assertions. We also need to record the match start in case      assertions. We also need to record the match start in case it was changed
1648      it was changed by \K. */      by \K. */
1649    
1650      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1651          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1652        {        {
1653        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1654        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
# Line 1541  for (;;) Line 1658  for (;;)
1658    
1659      /* 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
1660      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1661      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
1662      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
1663      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
1664        the current subject position and start match pointer and give a MATCH
1665        return. */
1666    
1667      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1668            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1669        {        {
1670        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1671        offset = number << 1;        offset = number << 1;
# Line 1555  for (;;) Line 1675  for (;;)
1675        printf("\n");        printf("\n");
1676  #endif  #endif
1677    
1678          /* Handle a recursively called group. */
1679    
1680          if (md->recursive != NULL && md->recursive->group_num == number)
1681            {
1682            md->end_match_ptr = eptr;
1683            md->start_match_ptr = mstart;
1684            RRETURN(MATCH_MATCH);
1685            }
1686    
1687          /* Deal with capturing */
1688    
1689        md->capture_last = number;        md->capture_last = number;
1690        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1691          {          {
1692            /* If offset is greater than offset_top, it means that we are
1693            "skipping" a capturing group, and that group's offsets must be marked
1694            unset. In earlier versions of PCRE, all the offsets were unset at the
1695            start of matching, but this doesn't work because atomic groups and
1696            assertions can cause a value to be set that should later be unset.
1697            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1698            part of the atomic group, but this is not on the final matching path,
1699            so must be unset when 2 is set. (If there is no group 2, there is no
1700            problem, because offset_top will then be 2, indicating no capture.) */
1701    
1702            if (offset > offset_top)
1703              {
1704              register int *iptr = md->offset_vector + offset_top;
1705              register int *iend = md->offset_vector + offset;
1706              while (iptr < iend) *iptr++ = -1;
1707              }
1708    
1709            /* Now make the extraction */
1710    
1711          md->offset_vector[offset] =          md->offset_vector[offset] =
1712            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1713          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1714          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1715          }          }
1716          }
1717    
1718        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1719        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1720        group. This is the forcible breaking of infinite loops as implemented in
1721        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1722        processing the rest of the pattern at a lower level. If this results in a
1723        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1724        bypassing intermediate backup points, but resetting any captures that
1725        happened along the way. */
1726    
1727        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1728          {
1729          if (*prev == OP_ONCE)
1730          {          {
1731          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1732          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1733          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1734          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;  
1735          }          }
1736          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1737          break;
1738        }        }
1739    
1740      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1741      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1742        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. */  
1743    
1744      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRPOS)
1745        {        {
1746        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1747        break;        md->end_offset_top = offset_top;
1748          RRETURN(MATCH_KETRPOS);
1749        }        }
1750    
1751      /* 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
1752      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
1753      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
1754      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
1755        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1756    
1757      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1758        {        {
1759        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1760        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1761        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1762            {
1763            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1764            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1765            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1766            RRETURN(MATCH_ONCE);
1767            }
1768          if (*prev >= OP_SBRA)    /* Could match an empty string */
1769          {          {
1770          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1771            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1772          RRETURN(rrc);          RRETURN(rrc);
1773          }          }
1774        ecode = prev;        ecode = prev;
# Line 1620  for (;;) Line 1776  for (;;)
1776        }        }
1777      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1778        {        {
1779        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1780          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1781          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1782        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1783          if (*prev == OP_ONCE)
1784            {
1785            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1786            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1787            md->once_target = prev;
1788            RRETURN(MATCH_ONCE);
1789            }
1790        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1791        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1792        }        }
1793      /* Control never gets here */      /* Control never gets here */
1794    
1795      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1796    
1797      case OP_CIRC:      case OP_CIRC:
1798      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(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 */  
1799    
1800      /* Start of subject assertion */      /* Start of subject assertion */
1801    
# Line 1649  for (;;) Line 1804  for (;;)
1804      ecode++;      ecode++;
1805      break;      break;
1806    
1807        /* Multiline mode: start of subject unless notbol, or after any newline. */
1808    
1809        case OP_CIRCM:
1810        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1811        if (eptr != md->start_subject &&
1812            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1813          MRRETURN(MATCH_NOMATCH);
1814        ecode++;
1815        break;
1816    
1817      /* Start of match assertion */      /* Start of match assertion */
1818    
1819      case OP_SOM:      case OP_SOM:
# Line 1663  for (;;) Line 1828  for (;;)
1828      ecode++;      ecode++;
1829      break;      break;
1830    
1831      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1832      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1833    
1834      case OP_DOLL:      case OP_DOLLM:
1835      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1836        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
       if (eptr < md->end_subject)  
         { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }  
       else  
         { if (md->noteol) MRRETURN(MATCH_NOMATCH); }  
       ecode++;  
       break;  
       }  
1837      else      else
1838        {        {
1839        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1840        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           MRRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1841        }        }
1842        ecode++;
1843        break;
1844    
1845        /* Not multiline mode: assert before a terminating newline or before end of
1846        subject unless noteol is set. */
1847    
1848        case OP_DOLL:
1849        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1850        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1851    
1852      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1853    
1854      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1855    
1856      case OP_EOD:      case OP_EOD:
1857      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1858        SCHECK_PARTIAL();
1859      ecode++;      ecode++;
1860      break;      break;
1861    
1862      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1863    
1864      case OP_EODN:      case OP_EODN:
1865      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1866        if (eptr < md->end_subject &&
1867          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1868        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1869    
1870        /* Either at end of string or \n before end. */
1871    
1872        SCHECK_PARTIAL();
1873      ecode++;      ecode++;
1874      break;      break;
1875    
# Line 1964  for (;;) Line 2131  for (;;)
2131      switch(c)      switch(c)
2132        {        {
2133        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2134    
2135        case 0x000d:        case 0x000d:
2136        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2137        break;        break;
# Line 2187  for (;;) Line 2355  for (;;)
2355        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2356        }        }
2357      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2358        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2359        while (eptr < md->end_subject)
2360        {        {
2361        int category = UCD_CATEGORY(c);        int len = 1;
2362        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2363        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2364          {        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;  
         }  
2365        }        }
2366      ecode++;      ecode++;
2367      break;      break;
# Line 2216  for (;;) Line 2377  for (;;)
2377      loops). */      loops). */
2378    
2379      case OP_REF:      case OP_REF:
2380        {      case OP_REFI:
2381        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2382        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2383        ecode += 3;
2384    
2385        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2386    
2387        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2388        than the amount of subject left; this ensures that every attempt at a      this ensures that every attempt at a match fails. We can't just fail
2389        match fails. We can't just fail here, because of the possibility of      here, because of the possibility of quantifiers with zero minima.
       quantifiers with zero minima.  
2390    
2391        (b) If the JavaScript compatibility flag is set, set the length to zero      (b) If the JavaScript compatibility flag is set, set the length to zero
2392        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2393    
2394        Otherwise, set the length to the length of what was matched by the      Otherwise, set the length to the length of what was matched by the
2395        referenced subpattern. */      referenced subpattern. */
2396    
2397        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2398          length = (md->jscript_compat)? 0 : (int)(md->end_subject - eptr + 1);        length = (md->jscript_compat)? 0 : -1;
2399        else      else
2400          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2401    
2402        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2403    
2404        switch (*ecode)      switch (*ecode)
2405          {        {
2406          case OP_CRSTAR:        case OP_CRSTAR:
2407          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2408          case OP_CRPLUS:        case OP_CRPLUS:
2409          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2410          case OP_CRQUERY:        case OP_CRQUERY:
2411          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2412          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2413          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2414          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2415          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2416          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2417          break;        break;
2418    
2419          case OP_CRRANGE:        case OP_CRRANGE:
2420          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2421          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2422          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2423          max = GET2(ecode, 3);        max = GET2(ecode, 3);
2424          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2425          ecode += 5;        ecode += 5;
2426          break;        break;
2427    
2428          default:               /* No repeat follows */        default:               /* No repeat follows */
2429          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2430            {          {
2431            CHECK_PARTIAL();          CHECK_PARTIAL();
2432            MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
           }  
         eptr += length;  
         continue;              /* With the main loop */  
2433          }          }
2434          eptr += length;
2435          continue;              /* With the main loop */
2436          }
2437    
2438        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2439        main loop. */      zero, just continue with the main loop. */
2440    
2441        if (length == 0) continue;      if (length == 0) continue;
2442    
2443        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2444        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2445        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2446    
2447        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2448          {
2449          int slength;
2450          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2451          {          {
2452          if (!match_ref(offset, eptr, length, md, ims))          CHECK_PARTIAL();
2453            {          MRRETURN(MATCH_NOMATCH);
           CHECK_PARTIAL();  
           MRRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2454          }          }
2455          eptr += slength;
2456          }
2457    
2458        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2459        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2460    
2461        if (min == max) continue;      if (min == max) continue;
2462    
2463        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2464    
2465        if (minimize)      if (minimize)
2466          {
2467          for (fi = min;; fi++)
2468          {          {
2469          for (fi = min;; fi++)          int slength;
2470            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2471            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2472            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2473            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2474            {            {
2475            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2476            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max) MRRETURN(MATCH_NOMATCH);  
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             MRRETURN(MATCH_NOMATCH);  
             }  
           eptr += length;  
2477            }            }
2478          /* Control never gets here */          eptr += slength;
2479          }          }
2480          /* Control never gets here */
2481          }
2482    
2483        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2484    
2485        else      else
2486          {
2487          pp = eptr;
2488          for (i = min; i < max; i++)
2489          {          {
2490          pp = eptr;          int slength;
2491          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
           {  
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             break;  
             }  
           eptr += length;  
           }  
         while (eptr >= pp)  
2492            {            {
2493            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2494            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2495            }            }
2496          MRRETURN(MATCH_NOMATCH);          eptr += slength;
2497          }          }
2498          while (eptr >= pp)
2499            {
2500            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2501            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2502            eptr -= length;
2503            }
2504          MRRETURN(MATCH_NOMATCH);
2505        }        }
2506      /* Control never gets here */      /* Control never gets here */
2507    
# Line 2444  for (;;) Line 2607  for (;;)
2607            {            {
2608            for (fi = min;; fi++)            for (fi = min;; fi++)
2609              {              {
2610              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2611              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2612              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2613              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2469  for (;;) Line 2632  for (;;)
2632            {            {
2633            for (fi = min;; fi++)            for (fi = min;; fi++)
2634              {              {
2635              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2636              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2637              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2638              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2515  for (;;) Line 2678  for (;;)
2678              }              }
2679            for (;;)            for (;;)
2680              {              {
2681              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2682              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2683              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2684              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2538  for (;;) Line 2701  for (;;)
2701              }              }
2702            while (eptr >= pp)            while (eptr >= pp)
2703              {              {
2704              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2705              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2706              eptr--;              eptr--;
2707              }              }
# Line 2614  for (;;) Line 2777  for (;;)
2777          {          {
2778          for (fi = min;; fi++)          for (fi = min;; fi++)
2779            {            {
2780            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2781            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2782            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2783            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2647  for (;;) Line 2810  for (;;)
2810            }            }
2811          for(;;)          for(;;)
2812            {            {
2813            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2814            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2815            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2816            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2692  for (;;) Line 2855  for (;;)
2855    
2856      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2857    
2858      case OP_CHARNC:      case OP_CHARI:
2859  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2860      if (utf8)      if (utf8)
2861        {        {
# Line 2752  for (;;) Line 2915  for (;;)
2915      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2916    
2917      case OP_EXACT:      case OP_EXACT:
2918        case OP_EXACTI:
2919      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2920      ecode += 3;      ecode += 3;
2921      goto REPEATCHAR;      goto REPEATCHAR;
2922    
2923      case OP_POSUPTO:      case OP_POSUPTO:
2924        case OP_POSUPTOI:
2925      possessive = TRUE;      possessive = TRUE;
2926      /* Fall through */      /* Fall through */
2927    
2928      case OP_UPTO:      case OP_UPTO:
2929        case OP_UPTOI:
2930      case OP_MINUPTO:      case OP_MINUPTO:
2931        case OP_MINUPTOI:
2932      min = 0;      min = 0;
2933      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2934      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2935      ecode += 3;      ecode += 3;
2936      goto REPEATCHAR;      goto REPEATCHAR;
2937    
2938      case OP_POSSTAR:      case OP_POSSTAR:
2939        case OP_POSSTARI:
2940      possessive = TRUE;      possessive = TRUE;
2941      min = 0;      min = 0;
2942      max = INT_MAX;      max = INT_MAX;
# Line 2776  for (;;) Line 2944  for (;;)
2944      goto REPEATCHAR;      goto REPEATCHAR;
2945    
2946      case OP_POSPLUS:      case OP_POSPLUS:
2947        case OP_POSPLUSI:
2948      possessive = TRUE;      possessive = TRUE;
2949      min = 1;      min = 1;
2950      max = INT_MAX;      max = INT_MAX;
# Line 2783  for (;;) Line 2952  for (;;)
2952      goto REPEATCHAR;      goto REPEATCHAR;
2953    
2954      case OP_POSQUERY:      case OP_POSQUERY:
2955        case OP_POSQUERYI:
2956      possessive = TRUE;      possessive = TRUE;
2957      min = 0;      min = 0;
2958      max = 1;      max = 1;
# Line 2790  for (;;) Line 2960  for (;;)
2960      goto REPEATCHAR;      goto REPEATCHAR;
2961    
2962      case OP_STAR:      case OP_STAR:
2963        case OP_STARI:
2964      case OP_MINSTAR:      case OP_MINSTAR:
2965        case OP_MINSTARI:
2966      case OP_PLUS:      case OP_PLUS:
2967        case OP_PLUSI:
2968      case OP_MINPLUS:      case OP_MINPLUS:
2969        case OP_MINPLUSI:
2970      case OP_QUERY:      case OP_QUERY:
2971        case OP_QUERYI:
2972      case OP_MINQUERY:      case OP_MINQUERY:
2973      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
2974        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
2975      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
2976      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2977      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2978      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2820  for (;;) Line 2995  for (;;)
2995          {          {
2996  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2997          unsigned int othercase;          unsigned int othercase;
2998          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
2999              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3000            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3001          else oclength = 0;          else oclength = 0;
# Line 2848  for (;;) Line 3023  for (;;)
3023            {            {
3024            for (fi = min;; fi++)            for (fi = min;; fi++)
3025              {              {
3026              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3027              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3028              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3029              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2890  for (;;) Line 3065  for (;;)
3065    
3066            for(;;)            for(;;)
3067              {              {
3068              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3069              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3070              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3071  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2927  for (;;) Line 3102  for (;;)
3102      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3103        max, eptr));        max, eptr));
3104    
3105      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3106        {        {
3107        fc = md->lcc[fc];        fc = md->lcc[fc];
3108        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2944  for (;;) Line 3119  for (;;)
3119          {          {
3120          for (fi = min;; fi++)          for (fi = min;; fi++)
3121            {            {
3122            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3123            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3124            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3125            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2974  for (;;) Line 3149  for (;;)
3149    
3150          while (eptr >= pp)          while (eptr >= pp)
3151            {            {
3152            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3153            eptr--;            eptr--;
3154            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3155            }            }
# Line 3003  for (;;) Line 3178  for (;;)
3178          {          {
3179          for (fi = min;; fi++)          for (fi = min;; fi++)
3180            {            {
3181            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3182            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3183            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3184            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3032  for (;;) Line 3207  for (;;)
3207    
3208          while (eptr >= pp)          while (eptr >= pp)
3209            {            {
3210            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3211            eptr--;            eptr--;
3212            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3213            }            }
# Line 3045  for (;;) Line 3220  for (;;)
3220      checking can be multibyte. */      checking can be multibyte. */
3221    
3222      case OP_NOT:      case OP_NOT:
3223        case OP_NOTI:
3224      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3225        {        {
3226        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 3052  for (;;) Line 3228  for (;;)
3228        }        }
3229      ecode++;      ecode++;
3230      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3231      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3232        {        {
3233  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3234        if (c < 256)        if (c < 256)
# Line 3060  for (;;) Line 3236  for (;;)
3236        c = md->lcc[c];        c = md->lcc[c];
3237        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3238        }        }
3239      else      else    /* Caseful */
3240        {        {
3241        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3242        }        }
# Line 3074  for (;;) Line 3250  for (;;)
3250      about... */      about... */
3251    
3252      case OP_NOTEXACT:      case OP_NOTEXACT:
3253        case OP_NOTEXACTI:
3254      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3255      ecode += 3;      ecode += 3;
3256      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3257    
3258      case OP_NOTUPTO:      case OP_NOTUPTO:
3259        case OP_NOTUPTOI:
3260      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3261        case OP_NOTMINUPTOI:
3262      min = 0;      min = 0;
3263      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3264      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3265      ecode += 3;      ecode += 3;
3266      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3267    
3268      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3269        case OP_NOTPOSSTARI:
3270      possessive = TRUE;      possessive = TRUE;
3271      min = 0;      min = 0;
3272      max = INT_MAX;      max = INT_MAX;
# Line 3094  for (;;) Line 3274  for (;;)
3274      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3275    
3276      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3277        case OP_NOTPOSPLUSI:
3278      possessive = TRUE;      possessive = TRUE;
3279      min = 1;      min = 1;
3280      max = INT_MAX;      max = INT_MAX;
# Line 3101  for (;;) Line 3282  for (;;)
3282      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3283    
3284      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3285        case OP_NOTPOSQUERYI:
3286      possessive = TRUE;      possessive = TRUE;
3287      min = 0;      min = 0;
3288      max = 1;      max = 1;
# Line 3108  for (;;) Line 3290  for (;;)
3290      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3291    
3292      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3293        case OP_NOTPOSUPTOI:
3294      possessive = TRUE;      possessive = TRUE;
3295      min = 0;      min = 0;
3296      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 3115  for (;;) Line 3298  for (;;)
3298      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3299    
3300      case OP_NOTSTAR:      case OP_NOTSTAR:
3301        case OP_NOTSTARI:
3302      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3303        case OP_NOTMINSTARI:
3304      case OP_NOTPLUS:      case OP_NOTPLUS:
3305        case OP_NOTPLUSI:
3306      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3307        case OP_NOTMINPLUSI:
3308      case OP_NOTQUERY:      case OP_NOTQUERY:
3309        case OP_NOTQUERYI:
3310      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3311      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3312        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3313      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3314      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3315      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3142  for (;;) Line 3331  for (;;)
3331      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,
3332        max, eptr));        max, eptr));
3333    
3334      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3335        {        {
3336        fc = md->lcc[fc];        fc = md->lcc[fc];
3337    
# Line 3190  for (;;) Line 3379  for (;;)
3379            register unsigned int d;            register unsigned int d;
3380            for (fi = min;; fi++)            for (fi = min;; fi++)
3381              {              {
3382              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3383              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3384              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3385              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3209  for (;;) Line 3398  for (;;)
3398            {            {
3399            for (fi = min;; fi++)            for (fi = min;; fi++)
3400              {              {
3401              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3402              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3403              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3404              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3250  for (;;) Line 3439  for (;;)
3439          if (possessive) continue;          if (possessive) continue;
3440          for(;;)          for(;;)
3441              {              {
3442              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3443              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3444              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3445              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3273  for (;;) Line 3462  for (;;)
3462            if (possessive) continue;            if (possessive) continue;
3463            while (eptr >= pp)            while (eptr >= pp)
3464              {              {
3465              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3466              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3467              eptr--;              eptr--;
3468              }              }
# Line 3330  for (;;) Line 3519  for (;;)
3519            register unsigned int d;            register unsigned int d;
3520            for (fi = min;; fi++)            for (fi = min;; fi++)
3521              {              {
3522              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3523              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3524              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3525              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3348  for (;;) Line 3537  for (;;)
3537            {            {
3538            for (fi = min;; fi++)            for (fi = min;; fi++)
3539              {              {
3540              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3541              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3542              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3543              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3388  for (;;) Line 3577  for (;;)
3577            if (possessive) continue;            if (possessive) continue;
3578            for(;;)            for(;;)
3579              {              {
3580              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3581              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3582              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3583              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3411  for (;;) Line 3600  for (;;)
3600            if (possessive) continue;            if (possessive) continue;
3601            while (eptr >= pp)            while (eptr >= pp)
3602              {              {
3603              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3604              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3605              eptr--;              eptr--;
3606              }              }
# Line 3526  for (;;) Line 3715  for (;;)
3715            case PT_LAMP:            case PT_LAMP:
3716            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3717              {              {
3718                int chartype;
3719              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3720                {                {
3721                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3722                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3723                }                }
3724              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3725              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
3726              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
3727                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
3728                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
3729                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3730              }              }
3731            break;            break;
# Line 3549  for (;;) Line 3739  for (;;)
3739                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3740                }                }
3741              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3742              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
3743                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3744              }              }
3745            break;            break;
# Line 3564  for (;;) Line 3753  for (;;)
3753                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3754                }                }
3755              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3756              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
3757                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3758              }              }
3759            break;            break;
# Line 3579  for (;;) Line 3767  for (;;)
3767                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3768                }                }
3769              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3770              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
3771                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3772              }              }
3773            break;            break;
# Line 3588  for (;;) Line 3775  for (;;)
3775            case PT_ALNUM:            case PT_ALNUM:
3776            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3777              {              {
3778                int category;
3779              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3780                {                {
3781                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3782                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3783                }                }
3784              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3785              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3786              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
3787                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3788              }              }
3789            break;            break;
# Line 3610  for (;;) Line 3797  for (;;)
3797                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3798                }                }
3799              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3800              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
3801                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
3802                     == prop_fail_result)                     == prop_fail_result)
3803                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3627  for (;;) Line 3813  for (;;)
3813                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3814                }                }
3815              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3816              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
3817                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3818                     == prop_fail_result)                     == prop_fail_result)
3819                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3638  for (;;) Line 3823  for (;;)
3823            case PT_WORD:            case PT_WORD:
3824            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3825              {              {
3826                int category;
3827              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3828                {                {
3829                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3830                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3831                }                }
3832              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3833              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3834              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
                  c == CHAR_UNDERSCORE)  
3835                     == prop_fail_result)                     == prop_fail_result)
3836                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3837              }              }
# Line 3672  for (;;) Line 3857  for (;;)
3857              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
3858              }              }
3859            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3860            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
3861            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3862              {              {
3863              int len = 1;              int len = 1;
3864              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
3865                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
3866              eptr += len;              eptr += len;
3867              }              }
3868            }            }
# Line 3738  for (;;) Line 3920  for (;;)
3920            switch(c)            switch(c)
3921              {              {
3922              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3923    
3924              case 0x000d:              case 0x000d:
3925              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3926              break;              break;
# Line 4014  for (;;) Line 4197  for (;;)
4197            switch(*eptr++)            switch(*eptr++)
4198              {              {
4199              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4200    
4201              case 0x000d:              case 0x000d:
4202              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4203              break;              break;
4204    
4205              case 0x000a:              case 0x000a:
4206              break;              break;
4207    
# Line 4206  for (;;) Line 4391  for (;;)
4391            case PT_ANY:            case PT_ANY:
4392            for (fi = min;; fi++)            for (fi = min;; fi++)
4393              {              {
4394              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4395              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4396              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4397              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4222  for (;;) Line 4407  for (;;)
4407            case PT_LAMP:            case PT_LAMP:
4408            for (fi = min;; fi++)            for (fi = min;; fi++)
4409              {              {
4410              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              int chartype;
4411                RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4412              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4413              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4414              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4231  for (;;) Line 4417  for (;;)
4417                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4418                }                }
4419              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4420              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4421              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4422                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4423                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4424                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4425              }              }
4426            /* Control never gets here */            /* Control never gets here */
# Line 4242  for (;;) Line 4428  for (;;)
4428            case PT_GC:            case PT_GC:
4429            for (fi = min;; fi++)            for (fi = min;; fi++)
4430              {              {
4431              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4432              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4433              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4434              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4251  for (;;) Line 4437  for (;;)
4437                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4438                }                }
4439              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4440              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4441                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4442              }              }
4443            /* Control never gets here */            /* Control never gets here */
# Line 4260  for (;;) Line 4445  for (;;)
4445            case PT_PC:            case PT_PC:
4446            for (fi = min;; fi++)            for (fi = min;; fi++)
4447              {              {
4448              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
4449              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4450              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4451              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4269  for (;;) Line 4454  for (;;)
4454                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4455                }                }
4456              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4457              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4458                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4459              }              }
4460            /* Control never gets here */            /* Control never gets here */
# Line 4278  for (;;) Line 4462  for (;;)
4462            case PT_SC:            case PT_SC:
4463            for (fi = min;; fi++)            for (fi = min;; fi++)
4464              {              {
4465              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
4466              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4467              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4468              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4287  for (;;) Line 4471  for (;;)
4471                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4472                }                }
4473              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4474              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
4475                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4476              }              }
4477            /* Control never gets here */            /* Control never gets here */
# Line 4296  for (;;) Line 4479  for (;;)
4479            case PT_ALNUM:            case PT_ALNUM:
4480            for (fi = min;; fi++)            for (fi = min;; fi++)
4481              {              {
4482              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59);              int category;
4483                RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4484              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4485              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4486              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4305  for (;;) Line 4489  for (;;)
4489                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4490                }                }
4491              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4492              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4493              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
4494                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4495              }              }
4496            /* Control never gets here */            /* Control never gets here */
# Line 4315  for (;;) Line 4498  for (;;)
4498            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4499            for (fi = min;; fi++)            for (fi = min;; fi++)
4500              {              {
4501              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4502              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4503              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4504              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4324  for (;;) Line 4507  for (;;)
4507                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4508                }                }
4509              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4510              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
4511                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
4512                     == prop_fail_result)                     == prop_fail_result)
4513                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4335  for (;;) Line 4517  for (;;)
4517            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4518            for (fi = min;; fi++)            for (fi = min;; fi++)
4519              {              {
4520              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4521              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4522              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4523              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4344  for (;;) Line 4526  for (;;)
4526                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4527                }                }
4528              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4529              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
4530                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4531                     == prop_fail_result)                     == prop_fail_result)
4532                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4355  for (;;) Line 4536  for (;;)
4536            case PT_WORD:            case PT_WORD:
4537            for (fi = min;; fi++)            for (fi = min;; fi++)
4538              {              {
4539              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62);              int category;
4540                RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4541              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4542              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4543              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4364  for (;;) Line 4546  for (;;)
4546                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4547                }                }
4548              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4549              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4550              if ((prop_category == ucp_L ||              if ((category == ucp_L ||
4551                   prop_category == ucp_N ||                   category == ucp_N ||
4552                   c == CHAR_UNDERSCORE)                   c == CHAR_UNDERSCORE)
4553                     == prop_fail_result)                     == prop_fail_result)
4554                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4387  for (;;) Line 4569  for (;;)
4569          {          {
4570          for (fi = min;; fi++)          for (fi = min;; fi++)
4571            {            {
4572            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4573            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4574            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4575            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4396  for (;;) Line 4578  for (;;)
4578              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4579              }              }
4580            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4581            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4582            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4583              {              {
4584              int len = 1;              int len = 1;
4585              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4586                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4587              eptr += len;              eptr += len;
4588              }              }
4589            }            }
4590          }          }
   
4591        else        else
4592  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4593    
# Line 4419  for (;;) Line 4597  for (;;)
4597          {          {
4598          for (fi = min;; fi++)          for (fi = min;; fi++)
4599            {            {
4600            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4601            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4602            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4603            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4582  for (;;) Line 4760  for (;;)
4760          {          {
4761          for (fi = min;; fi++)          for (fi = min;; fi++)
4762            {            {
4763            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4764            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4765            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4766            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4730  for (;;) Line 4908  for (;;)
4908            case PT_LAMP:            case PT_LAMP:
4909            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4910              {              {
4911                int chartype;
4912              int len = 1;              int len = 1;
4913              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4914                {                {
# Line 4737  for (;;) Line 4916  for (;;)
4916                break;                break;
4917                }                }
4918              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4919              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4920              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4921                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4922                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4923                break;                break;
4924              eptr+= len;              eptr+= len;
4925              }              }
# Line 4756  for (;;) Line 4935  for (;;)
4935                break;                break;
4936                }                }
4937              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4938              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
4939              eptr+= len;              eptr+= len;
4940              }              }
4941            break;            break;
# Line 4773  for (;;) Line 4950  for (;;)
4950                break;                break;
4951                }                }
4952              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4953              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
4954              eptr+= len;              eptr+= len;
4955              }              }
4956            break;            break;
# Line 4790  for (;;) Line 4965  for (;;)
4965                break;                break;
4966                }                }
4967              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4968              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
             if ((prop_script == prop_value) == prop_fail_result)  
               break;  
4969              eptr+= len;              eptr+= len;
4970              }              }
4971            break;            break;
# Line 4800  for (;;) Line 4973  for (;;)
4973            case PT_ALNUM:            case PT_ALNUM:
4974            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4975              {              {
4976                int category;
4977              int len = 1;              int len = 1;
4978              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4979                {                {
# Line 4807  for (;;) Line 4981  for (;;)
4981                break;                break;
4982                }                }
4983              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4984              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4985              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                  == prop_fail_result)  
4986                break;                break;
4987              eptr+= len;              eptr+= len;
4988              }              }
# Line 4825  for (;;) Line 4998  for (;;)
4998                break;                break;
4999                }                }
5000              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5001              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
5002                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
5003                   == prop_fail_result)                   == prop_fail_result)
5004                break;                break;
# Line 4844  for (;;) Line 5016  for (;;)
5016                break;                break;
5017                }                }
5018              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5019              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
5020                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5021                   == prop_fail_result)                   == prop_fail_result)
5022                break;                break;
# Line 4856  for (;;) Line 5027  for (;;)
5027            case PT_WORD:            case PT_WORD:
5028            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5029              {              {
5030                int category;
5031              int len = 1;              int len = 1;
5032              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5033                {                {
# Line 4863  for (;;) Line 5035  for (;;)
5035                break;                break;
5036                }                }
5037              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5038              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5039              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N ||
5040                   c == CHAR_UNDERSCORE) == prop_fail_result)                   c == CHAR_UNDERSCORE) == prop_fail_result)
5041                break;                break;
5042              eptr+= len;              eptr+= len;
# Line 4880  for (;;) Line 5052  for (;;)
5052          if (possessive) continue;          if (possessive) continue;
5053          for(;;)          for(;;)
5054            {            {
5055            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5056            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5057            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5058            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4894  for (;;) Line 5066  for (;;)
5066          {          {
5067          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5068            {            {
5069              int len = 1;
5070            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5071              {              {
5072              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5073              break;              break;
5074              }              }
5075            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5076            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5077            if (prop_category == ucp_M) break;            eptr += len;
5078            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5079              {              {
5080              int len = 1;              len = 1;
5081              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5082                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5083              eptr += len;              eptr += len;
5084              }              }
5085            }            }
# Line 4921  for (;;) Line 5090  for (;;)
5090    
5091          for(;;)          for(;;)
5092            {            {
5093            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5094            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5095            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5096            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5097              {              {
             int len = 1;  
5098              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5099                {                {
5100                BACKCHAR(eptr);                BACKCHAR(eptr);
5101                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5102                }                }
5103              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5104              eptr--;              eptr--;
5105              }              }
5106            }            }
# Line 5205  for (;;) Line 5372  for (;;)
5372            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5373            }            }
5374    
5375          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5376            done (no backing up). Otherwise, match at this position; anything other
5377            than no match is immediately returned. For nomatch, back up one
5378            character, unless we are matching \R and the last thing matched was
5379            \r\n, in which case, back up two bytes. */
5380    
5381          if (possessive) continue;          if (possessive) continue;
5382          for(;;)          for(;;)
5383            {            {
5384            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5385            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5386            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5387            BACKCHAR(eptr);            BACKCHAR(eptr);
5388              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5389                  eptr[-1] == '\r') eptr--;
5390            }            }
5391          }          }
5392        else        else
# Line 5412  for (;;) Line 5585  for (;;)
5585            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5586            }            }
5587    
5588          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5589            done (no backing up). Otherwise, match at this position; anything other
5590            than no match is immediately returned. For nomatch, back up one
5591            character (byte), unless we are matching \R and the last thing matched
5592            was \r\n, in which case, back up two bytes. */
5593    
5594          if (possessive) continue;          if (possessive) continue;
5595          while (eptr >= pp)          while (eptr >= pp)
5596            {            {
5597            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5598            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5599              eptr--;
5600              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5601                  eptr[-1] == '\r') eptr--;
5602            }            }
5603          }          }
5604    
# Line 5458  switch (frame->Xwhere) Line 5637  switch (frame->Xwhere)
5637    LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)    LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)
5638    LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)    LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)
5639    LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)    LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)
5640    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63)
5641  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5642    LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)    LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)
5643    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
# Line 5487  Undefine all the macros that were define Line 5666  Undefine all the macros that were define
5666  #undef ecode  #undef ecode
5667  #undef mstart  #undef mstart
5668  #undef offset_top  #undef offset_top
 #undef ims  
5669  #undef eptrb  #undef eptrb
5670  #undef flags  #undef flags
5671    
# Line 5505  Undefine all the macros that were define Line 5683  Undefine all the macros that were define
5683  #undef condition  #undef condition
5684  #undef prev_is_word  #undef prev_is_word
5685    
 #undef original_ims  
   
5686  #undef ctype  #undef ctype
5687  #undef length  #undef length
5688  #undef max  #undef max
# Line 5563  pcre_exec(const pcre *argument_re, const Line 5739  pcre_exec(const pcre *argument_re, const
5739    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5740    int offsetcount)    int offsetcount)
5741  {  {
5742  int rc, resetcount, ocount;  int rc, ocount;
5743  int first_byte = -1;  int first_byte = -1;
5744  int req_byte = -1;  int req_byte = -1;
5745  int req_byte2 = -1;  int req_byte2 = -1;
5746  int newline;  int newline;
 unsigned long int ims;  
5747  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5748  BOOL anchored;  BOOL anchored;
5749  BOOL startline;  BOOL startline;
# Line 5598  if ((options & ~PUBLIC_EXEC_OPTIONS) != Line 5773  if ((options & ~PUBLIC_EXEC_OPTIONS) !=
5773  if (re == NULL || subject == NULL ||  if (re == NULL || subject == NULL ||
5774     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;
5775  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
5776    if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
5777    
5778  /* This information is for finding all the numbers associated with a given  /* This information is for finding all the numbers associated with a given
5779  name, for condition testing. */  name, for condition testing. */
# Line 5671  utf8 = md->utf8 = (re->options & PCRE_UT Line 5847  utf8 = md->utf8 = (re->options & PCRE_UT
5847  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
5848  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
5849    
5850    /* Some options are unpacked into BOOL variables in the hope that testing
5851    them will be faster than individual option bits. */
5852    
5853  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
5854  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
5855  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
5856  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
5857  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5858                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5859    
5860    
5861  md->hitend = FALSE;  md->hitend = FALSE;
5862  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
5863    
# Line 5758  defined (though never set). So there's n Line 5939  defined (though never set). So there's n
5939  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
5940    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
5941    
5942  /* Check a UTF-8 string if required. Unfortunately there's no way of passing  /* Check a UTF-8 string if required. Pass back the character offset and error
5943  back the character offset. */  code for an invalid string if a results vector is available. */
5944    
5945  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5946  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5947    {    {
5948    if (_pcre_valid_utf8((USPTR)subject, length) >= 0)    int erroroffset;
5949      return PCRE_ERROR_BADUTF8;    int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5950    if (start_offset > 0 && start_offset < length)    if (errorcode != 0)
5951      {      {
5952      int tb = ((USPTR)subject)[start_offset];      if (offsetcount >= 2)
     if (tb > 127)  
5953        {        {
5954        tb &= 0xc0;        offsets[0] = erroroffset;
5955        if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET;        offsets[1] = errorcode;
5956        }        }
5957        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5958          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5959      }      }
   }  
 #endif  
5960    
5961  /* The ims options can vary during the matching as a result of the presence    /* Check that a start_offset points to the start of a UTF-8 character. */
 of (?ims) items in the pattern. They are kept in a local variable so that  
 restoring at the exit of a group is easy. */  
5962    
5963  ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL);    if (start_offset > 0 && start_offset < length &&
5964          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
5965        return PCRE_ERROR_BADUTF8_OFFSET;
5966      }
5967    #endif
5968    
5969  /* If the expression has got more back references than the offsets supplied can  /* If the expression has got more back references than the offsets supplied can
5970  hold, we get a temporary chunk of working store to use during the matching.  hold, we get a temporary chunk of working store to use during the matching.
# Line 5806  md->offset_max = (2*ocount)/3; Line 5988  md->offset_max = (2*ocount)/3;
5988  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
5989  md->capture_last = -1;  md->capture_last = -1;
5990    
 /* Compute the minimum number of offsets that we need to reset each time. Doing  
 this makes a huge difference to execution time when there aren't many brackets  
 in the pattern. */  
   
 resetcount = 2 + re->top_bracket * 2;  
 if (resetcount > offsetcount) resetcount = ocount;  
   
5991  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
5992  never be used unless previously set, but they get saved and restored, and so we  never be used unless previously set, but they get saved and restored, and so we
5993  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
5994    offsets for the matched string. This is really just for tidiness with callouts,
5995    in case they inspect these fields. */
5996    
5997  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
5998    {    {
5999    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6000    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6001      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6002    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6003      md->offset_vector[0] = md->offset_vector[1] = -1;
6004    }    }
6005    
6006  /* Set up the first character to match, if available. The first_byte value is  /* Set up the first character to match, if available. The first_byte value is
# Line 5855  if ((re->flags & PCRE_REQCHSET) != 0) Line 6034  if ((re->flags & PCRE_REQCHSET) != 0)
6034    }    }
6035    
6036    
6037    
6038    
6039  /* ==========================================================================*/  /* ==========================================================================*/
6040    
6041  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 5865  for(;;) Line 6046  for(;;)
6046    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6047    USPTR new_start_match;    USPTR new_start_match;
6048    
   /* Reset the maximum number of extractions we might see. */  
   
   if (md->offset_vector != NULL)  
     {  
     register int *iptr = md->offset_vector;  
     register int *iend = iptr + resetcount;  
     while (iptr < iend) *iptr++ = -1;  
     }  
   
6049    /* If firstline is TRUE, the start of the match is constrained to the first    /* If firstline is TRUE, the start of the match is constrained to the first
6050    line of a multiline string. That is, the match must be before or at the first    line of a multiline string. That is, the match must be before or at the first
6051    newline. Implement this by temporarily adjusting end_subject so that we stop    newline. Implement this by temporarily adjusting end_subject so that we stop
# Line 5901  for(;;) Line 6073  for(;;)
6073    /* There are some optimizations that avoid running the match if a known    /* There are some optimizations that avoid running the match if a known
6074    starting point is not found, or if a known later character is not present.    starting point is not found, or if a known later character is not present.
6075    However, there is an option that disables these, for testing and for ensuring    However, there is an option that disables these, for testing and for ensuring
6076    that all callouts do actually occur. */    that all callouts do actually occur. The option can be set in the regex by
6077      (*NO_START_OPT) or passed in match-time options. */
6078    
6079    if ((options & PCRE_NO_START_OPTIMIZE) == 0)    if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0)
6080      {      {
6081      /* Advance to a unique first byte if there is one. */      /* Advance to a unique first byte if there is one. */
6082    
# Line 5957  for(;;) Line 6130  for(;;)
6130        while (start_match < end_subject)        while (start_match < end_subject)
6131          {          {
6132          register unsigned int c = *start_match;          register unsigned int c = *start_match;
6133          if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++;          if ((start_bits[c/8] & (1 << (c&7))) == 0)
6134            else break;            {
6135              start_match++;
6136    #ifdef SUPPORT_UTF8
6137              if (utf8)
6138                while(start_match < end_subject && (*start_match & 0xc0) == 0x80)
6139                  start_match++;
6140    #endif
6141              }
6142            else break;
6143          }          }
6144        }        }
6145      }   /* Starting optimizations */      }   /* Starting optimizations */
# Line 6053  for(;;) Line 6234  for(;;)
6234    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6235    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6236    md->match_call_count = 0;    md->match_call_count = 0;
6237    rc = match(start_match, md->start_code, start_match, NULL, 2, md, ims, NULL,    md->match_function_type = 0;
6238      0, 0);    md->end_offset_top = 0;
6239      rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6240    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
6241    
6242    switch(rc)    switch(rc)
6243      {      {
6244      /* NOMATCH and PRUNE advance by one character. If MATCH_SKIP_ARG reaches      /* SKIP passes back the next starting point explicitly, but if it is the
6245      this level it means that a MARK that matched the SKIP's arg was not found.      same as the match we have just done, treat it as NOMATCH. */
6246      We treat this as NOMATCH. THEN at this level acts exactly like PRUNE. */  
6247        case MATCH_SKIP:
6248        if (md->start_match_ptr != start_match)
6249          {
6250          new_start_match = md->start_match_ptr;
6251          break;
6252          }
6253        /* Fall through */
6254    
6255        /* If MATCH_SKIP_ARG reaches this level it means that a MARK that matched
6256        the SKIP's arg was not found. We also treat this as NOMATCH. */
6257    
6258        case MATCH_SKIP_ARG:
6259        /* Fall through */
6260    
6261        /* NOMATCH and PRUNE advance by one character. THEN at this level acts
6262        exactly like PRUNE. */
6263    
6264      case MATCH_NOMATCH:      case MATCH_NOMATCH:
6265      case MATCH_PRUNE:      case MATCH_PRUNE:
     case MATCH_SKIP_ARG:  
6266      case MATCH_THEN:      case MATCH_THEN:
6267      new_start_match = start_match + 1;      new_start_match = start_match + 1;
6268  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 6075  for(;;) Line 6272  for(;;)
6272  #endif  #endif
6273      break;      break;
6274    
     /* SKIP passes back the next starting point explicitly. */  
   
     case MATCH_SKIP:  
     new_start_match = md->start_match_ptr;  
     break;  
   
6275      /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */      /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */
6276    
6277      case MATCH_COMMIT:      case MATCH_COMMIT:
# Line 6171  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6362  if (rc == MATCH_MATCH || rc == MATCH_ACC
6362    
6363    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = md->offset_overflow? 0 : md->end_offset_top/2;
6364    
6365      /* If there is space in the offset vector, set any unused pairs at the end of
6366      the pattern to -1 for backwards compatibility. It is documented that this
6367      happens. In earlier versions, the whole set of potential capturing offsets
6368      was set to -1 each time round the loop, but this is handled differently now.
6369      "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only
6370      those at the end that need unsetting here. We can't just unset them all at
6371      the start of the whole thing because they may get set in one branch that is
6372      not the final matching branch. */
6373    
6374      if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL)
6375        {
6376        register int *iptr, *iend;
6377        int resetcount = 2 + re->top_bracket * 2;
6378        if (resetcount > offsetcount) resetcount = ocount;
6379        iptr = offsets + md->end_offset_top;
6380        iend = offsets + resetcount;
6381        while (iptr < iend) *iptr++ = -1;
6382        }
6383    
6384    /* If there is space, set up the whole thing as substring 0. The value of    /* If there is space, set up the whole thing as substring 0. The value of
6385    md->start_match_ptr might be modified if \K was encountered on the success    md->start_match_ptr might be modified if \K was encountered on the success
6386    matching path. */    matching path. */

Legend:
Removed from v.530  
changed lines
  Added in v.626

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12