/[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 511 by ph10, Mon Mar 29 09:25:38 2010 UTC revision 751 by ph10, Fri Nov 18 11:13:30 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 255  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 276  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
276         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
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 };         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280           RM61,  RM62, RM63, RM64, RM65, RM66 };
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 265  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 277  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 291  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 314  argument of match(), which never changes Line 335  argument of match(), which never changes
335    
336  #define RRETURN(ra)\  #define RRETURN(ra)\
337    {\    {\
338    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
339    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
340    (pcre_stack_free)(newframe);\    (pcre_stack_free)(oldframe);\
341    if (frame != NULL)\    if (frame != NULL)\
342      {\      {\
343      rrc = ra;\      rrc = ra;\
# Line 338  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 361  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 420  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 447  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 465  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 478  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 486  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 496  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 512  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 536  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 574  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
589    const uschar *charptr;
590    #endif
591    const uschar *callpat;
592    const uschar *data;
593    const uschar *next;
594    USPTR         pp;
595    const uschar *prev;
596    USPTR         saved_eptr;
597    
598    recursion_info new_recursive;
599    
600  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  BOOL cur_is_word;
 const uschar *charptr;             /* in small blocks of the code. My normal */  
 #endif                             /* style of coding would have declared    */  
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
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 619  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 655  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 685  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, RM51);        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
710      argument. It is passed back in md->start_match_ptr (an overloading of that      argument. It is passed back in md->start_match_ptr (an overloading of that
711      variable). If it does match, we reset that variable to the current subject      variable). If it does match, we reset that variable to the current subject
712      position and return MATCH_SKIP. Otherwise, pass back the return code      position and return MATCH_SKIP. Otherwise, pass back the return code
713      unaltered. */      unaltered. */
714    
715      if (rrc == MATCH_SKIP_ARG &&      if (rrc == MATCH_SKIP_ARG &&
716          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)          strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
717        {        {
718        md->start_match_ptr = eptr;        md->start_match_ptr = eptr;
719        RRETURN(MATCH_SKIP);        RRETURN(MATCH_SKIP);
720        }        }
721    
722      if (md->mark == NULL) md->mark = markptr;      if (md->mark == NULL) md->mark = markptr;
723      RRETURN(rrc);      RRETURN(rrc);
724    
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, RM51);        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, RM53);        eptrb, RM57);
767      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769      /* Pass back the current skip name by overloading md->start_match_ptr and  
770      returning the special MATCH_SKIP_ARG return code. This will either be      /* Pass back the current skip name by overloading md->start_match_ptr and
771      caught by a matching MARK, or get to the top, where it is treated the same      returning the special MATCH_SKIP_ARG return code. This will either be
772        caught by a matching MARK, or get to the top, where it is treated the same
773      as PRUNE. */      as PRUNE. */
774    
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 opcode, so that
779        the branch in which it occurs can be determined. Overload the start of
780        match pointer to do this. */
781    
782      case OP_THEN:      case OP_THEN:
783      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
784        ims, eptrb, flags, RM54);        eptrb, RM54);
785      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
786        md->start_match_ptr = ecode;
787      MRRETURN(MATCH_THEN);      MRRETURN(MATCH_THEN);
788    
789      case OP_THEN_ARG:      case OP_THEN_ARG:
790      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top,
791        ims, eptrb, flags, RM54);        md, eptrb, RM58);
792      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
793        md->start_match_ptr = ecode;
794      md->mark = ecode + 2;      md->mark = ecode + 2;
795      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
796    
797      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle an atomic group that does not contain any capturing parentheses.
798      the current subject position in the working slot at the top of the vector.      This can be handled like an assertion. Prior to 8.13, all atomic groups
799      We mustn't change the current values of the data slot, because they may be      were handled this way. In 8.13, the code was changed as below for ONCE, so
800      set from a previous iteration of this group, and be referred to by a      that backups pass through the group and thereby reset captured values.
801      reference inside the group.      However, this uses a lot more stack, so in 8.20, atomic groups that do not
802        contain any captures generate OP_ONCE_NC, which can be handled in the old,
803      If the bracket fails to match, we need to restore this value and also the      less stack intensive way.
804      values of the final offsets, in case they were set by a previous iteration  
805      of the same bracket.      Check the alternative branches in turn - the matching won't pass the KET
806        for this kind of subpattern. If any one branch matches, we carry on as at
807        the end of a normal bracket, leaving the subject pointer, but resetting
808        the start-of-match value in case it was changed by \K. */
809    
810        case OP_ONCE_NC:
811        prev = ecode;
812        saved_eptr = eptr;
813        do
814          {
815          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
816          if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */
817            {
818            mstart = md->start_match_ptr;
819            markptr = md->mark;
820            break;
821            }
822          if (rrc == MATCH_THEN)
823            {
824            next = ecode + GET(ecode,1);
825            if (md->start_match_ptr < next &&
826                (*ecode == OP_ALT || *next == OP_ALT))
827              rrc = MATCH_NOMATCH;
828            }
829    
830          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
831          ecode += GET(ecode,1);
832          }
833        while (*ecode == OP_ALT);
834    
835        /* If hit the end of the group (which could be repeated), fail */
836    
837        if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
838    
839        /* Continue as from after the group, updating the offsets high water
840        mark, since extracts may have been taken. */
841    
842        do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
843    
844        offset_top = md->end_offset_top;
845        eptr = md->end_match_ptr;
846    
847        /* For a non-repeating ket, just continue at this level. This also
848        happens for a repeating ket if no characters were matched in the group.
849        This is the forcible breaking of infinite loops as implemented in Perl
850        5.005. */
851    
852        if (*ecode == OP_KET || eptr == saved_eptr)
853          {
854          ecode += 1+LINK_SIZE;
855          break;
856          }
857    
858        /* The repeating kets try the rest of the pattern or restart from the
859        preceding bracket, in the appropriate order. The second "call" of match()
860        uses tail recursion, to avoid using another stack frame. */
861    
862        if (*ecode == OP_KETRMIN)
863          {
864          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
865          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
866          ecode = prev;
867          goto TAIL_RECURSE;
868          }
869        else  /* OP_KETRMAX */
870          {
871          md->match_function_type = MATCH_CBEGROUP;
872          RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
873          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
874          ecode += 1 + LINK_SIZE;
875          goto TAIL_RECURSE;
876          }
877        /* Control never gets here */
878    
879        /* Handle a capturing bracket, other than those that are possessive with an
880        unlimited repeat. If there is space in the offset vector, save the current
881        subject position in the working slot at the top of the vector. We mustn't
882        change the current values of the data slot, because they may be set from a
883        previous iteration of this group, and be referred to by a reference inside
884        the group. A failure to match might occur after the group has succeeded,
885        if something later on doesn't match. For this reason, we need to restore
886        the working value and also the values of the final offsets, in case they
887        were set by a previous iteration of the same bracket.
888    
889      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
890      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 793  for (;;) Line 910  for (;;)
910        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
911    
912        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
913        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
914            (int)(eptr - md->start_subject);
915    
916        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
917          {          {
918            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
919          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
920            ims, eptrb, flags, RM1);            eptrb, RM1);
921          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
922    
923            /* If we backed up to a THEN, check whether it is within the current
924            branch by comparing the address of the THEN that is passed back with
925            the end of the branch. If it is within the current branch, and the
926            branch is one of two or more alternatives (it either starts or ends
927            with OP_ALT), we have reached the limit of THEN's action, so convert
928            the return code to NOMATCH, which will cause normal backtracking to
929            happen from now on. Otherwise, THEN is passed back to an outer
930            alternative. This implements Perl's treatment of parenthesized groups,
931            where a group not containing | does not affect the current alternative,
932            that is, (X) is NOT the same as (X|(*F)). */
933    
934            if (rrc == MATCH_THEN)
935              {
936              next = ecode + GET(ecode,1);
937              if (md->start_match_ptr < next &&
938                  (*ecode == OP_ALT || *next == OP_ALT))
939                rrc = MATCH_NOMATCH;
940              }
941    
942            /* Anything other than NOMATCH is passed back. */
943    
944            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
945          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
946          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
947            if (*ecode != OP_ALT) break;
948          }          }
       while (*ecode == OP_ALT);  
949    
950        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
951        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
952        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
953        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
954    
955        if (rrc != MATCH_THEN) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
956        RRETURN(MATCH_NOMATCH);  
957          if (md->mark == NULL) md->mark = markptr;
958          RRETURN(rrc);
959        }        }
960    
961      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 827  for (;;) Line 969  for (;;)
969      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
970      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
971    
972      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
973      final alternative within the brackets, we would return the result of a      repeat and ONCE group with no captures. Loop for all the alternatives.
     recursive call to match() whatever happened. We can reduce stack usage by  
     turning this into a tail recursion, except in the case when match_cbegroup  
     is set.*/  
974    
975        When we get to the final alternative within the brackets, we used to return
976        the result of a recursive call to match() whatever happened so it was
977        possible to reduce stack usage by turning this into a tail recursion,
978        except in the case of a possibly empty group. However, now that there is
979        the possiblity of (*THEN) occurring in the final alternative, this
980        optimization is no longer always possible.
981    
982        We can optimize if we know there are no (*THEN)s in the pattern; at present
983        this is the best that can be done.
984    
985        MATCH_ONCE is returned when the end of an atomic group is successfully
986        reached, but subsequent matching fails. It passes back up the tree (causing
987        captured values to be reset) until the original atomic group level is
988        reached. This is tested by comparing md->once_target with the start of the
989        group. At this point, the return is converted into MATCH_NOMATCH so that
990        previous backup points can be taken. */
991    
992        case OP_ONCE:
993      case OP_BRA:      case OP_BRA:
994      case OP_SBRA:      case OP_SBRA:
995      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
996      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
997      for (;;)      for (;;)
998        {        {
999        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
1000    
1001          /* If this is not a possibly empty group, and there are no (*THEN)s in
1002          the pattern, and this is the final alternative, optimize as described
1003          above. */
1004    
1005          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
1006            {
1007            ecode += _pcre_OP_lengths[*ecode];
1008            goto TAIL_RECURSE;
1009            }
1010    
1011          /* In all other cases, we have to make another call to match(). */
1012    
1013          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
1014            RM2);
1015    
1016          /* See comment in the code for capturing groups above about handling
1017          THEN. */
1018    
1019          if (rrc == MATCH_THEN)
1020          {          {
1021          if (flags == 0)    /* Not a possibly empty group */          next = ecode + GET(ecode,1);
1022            if (md->start_match_ptr < next &&
1023                (*ecode == OP_ALT || *next == OP_ALT))
1024              rrc = MATCH_NOMATCH;
1025            }
1026    
1027          if (rrc != MATCH_NOMATCH)
1028            {
1029            if (rrc == MATCH_ONCE)
1030            {            {
1031            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
1032            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
1033            goto TAIL_RECURSE;              {
1034                while (*scode == OP_ALT) scode += GET(scode, 1);
1035                scode -= GET(scode, 1);
1036                }
1037              if (md->once_target == scode) rrc = MATCH_NOMATCH;
1038            }            }
1039            RRETURN(rrc);
1040            }
1041          ecode += GET(ecode, 1);
1042          if (*ecode != OP_ALT) break;
1043          }
1044    
1045          /* Possibly empty group; can't use tail recursion. */      if (md->mark == NULL) md->mark = markptr;
1046        RRETURN(MATCH_NOMATCH);
1047    
1048          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* Handle possessive capturing brackets with an unlimited repeat. We come
1049            eptrb, flags, RM48);      here from BRAZERO with allow_zero set TRUE. The offset_vector values are
1050          if (rrc == MATCH_NOMATCH) md->mark = markptr;      handled similarly to the normal case above. However, the matching is
1051          RRETURN(rrc);      different. The end of these brackets will always be OP_KETRPOS, which
1052        returns MATCH_KETRPOS without going further in the pattern. By this means
1053        we can handle the group by iteration rather than recursion, thereby
1054        reducing the amount of stack needed. */
1055    
1056        case OP_CBRAPOS:
1057        case OP_SCBRAPOS:
1058        allow_zero = FALSE;
1059    
1060        POSSESSIVE_CAPTURE:
1061        number = GET2(ecode, 1+LINK_SIZE);
1062        offset = number << 1;
1063    
1064    #ifdef PCRE_DEBUG
1065        printf("start possessive bracket %d\n", number);
1066        printf("subject=");
1067        pchars(eptr, 16, TRUE, md);
1068        printf("\n");
1069    #endif
1070    
1071        if (offset < md->offset_max)
1072          {
1073          matched_once = FALSE;
1074          code_offset = ecode - md->start_code;
1075    
1076          save_offset1 = md->offset_vector[offset];
1077          save_offset2 = md->offset_vector[offset+1];
1078          save_offset3 = md->offset_vector[md->offset_end - number];
1079          save_capture_last = md->capture_last;
1080    
1081          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1082    
1083          /* Each time round the loop, save the current subject position for use
1084          when the group matches. For MATCH_MATCH, the group has matched, so we
1085          restart it with a new subject starting position, remembering that we had
1086          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1087          usual. If we haven't matched any alternatives in any iteration, check to
1088          see if a previous iteration matched. If so, the group has matched;
1089          continue from afterwards. Otherwise it has failed; restore the previous
1090          capture values before returning NOMATCH. */
1091    
1092          for (;;)
1093            {
1094            md->offset_vector[md->offset_end - number] =
1095              (int)(eptr - md->start_subject);
1096            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1097            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1098              eptrb, RM63);
1099            if (rrc == MATCH_KETRPOS)
1100              {
1101              offset_top = md->end_offset_top;
1102              eptr = md->end_match_ptr;
1103              ecode = md->start_code + code_offset;
1104              save_capture_last = md->capture_last;
1105              matched_once = TRUE;
1106              continue;
1107              }
1108    
1109            /* See comment in the code for capturing groups above about handling
1110            THEN. */
1111    
1112            if (rrc == MATCH_THEN)
1113              {
1114              next = ecode + GET(ecode,1);
1115              if (md->start_match_ptr < next &&
1116                  (*ecode == OP_ALT || *next == OP_ALT))
1117                rrc = MATCH_NOMATCH;
1118              }
1119    
1120            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1121            md->capture_last = save_capture_last;
1122            ecode += GET(ecode, 1);
1123            if (*ecode != OP_ALT) break;
1124          }          }
1125    
1126        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1127        otherwise return. */          {
1128            md->offset_vector[offset] = save_offset1;
1129            md->offset_vector[offset+1] = save_offset2;
1130            md->offset_vector[md->offset_end - number] = save_offset3;
1131            }
1132    
1133        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (md->mark == NULL) md->mark = markptr;
1134          eptrb, flags, RM2);        if (allow_zero || matched_once)
1135        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          {
1136            ecode += 1 + LINK_SIZE;
1137            break;
1138            }
1139    
1140          RRETURN(MATCH_NOMATCH);
1141          }
1142    
1143        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1144        as a non-capturing bracket. */
1145    
1146        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1147        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1148    
1149        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1150    
1151        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1152        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1153    
1154        /* Non-capturing possessive bracket with unlimited repeat. We come here
1155        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1156        without the capturing complication. It is written out separately for speed
1157        and cleanliness. */
1158    
1159        case OP_BRAPOS:
1160        case OP_SBRAPOS:
1161        allow_zero = FALSE;
1162    
1163        POSSESSIVE_NON_CAPTURE:
1164        matched_once = FALSE;
1165        code_offset = ecode - md->start_code;
1166    
1167        for (;;)
1168          {
1169          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1170          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1171            eptrb, RM48);
1172          if (rrc == MATCH_KETRPOS)
1173            {
1174            offset_top = md->end_offset_top;
1175            eptr = md->end_match_ptr;
1176            ecode = md->start_code + code_offset;
1177            matched_once = TRUE;
1178            continue;
1179            }
1180    
1181          /* See comment in the code for capturing groups above about handling
1182          THEN. */
1183    
1184          if (rrc == MATCH_THEN)
1185            {
1186            next = ecode + GET(ecode,1);
1187            if (md->start_match_ptr < next &&
1188                (*ecode == OP_ALT || *next == OP_ALT))
1189              rrc = MATCH_NOMATCH;
1190            }
1191    
1192          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1193        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1194          if (*ecode != OP_ALT) break;
1195        }        }
1196    
1197        if (matched_once || allow_zero)
1198          {
1199          ecode += 1 + LINK_SIZE;
1200          break;
1201          }
1202        RRETURN(MATCH_NOMATCH);
1203    
1204      /* Control never reaches here. */      /* Control never reaches here. */
1205    
1206      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1207      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1208      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1209      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1210    
1211      case OP_COND:      case OP_COND:
1212      case OP_SCOND:      case OP_SCOND:
1213      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1214    
1215      /* Because of the way auto-callout works during compile, a callout item is      /* Because of the way auto-callout works during compile, a callout item is
1216      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 884  for (;;) Line 1220  for (;;)
1220        if (pcre_callout != NULL)        if (pcre_callout != NULL)
1221          {          {
1222          pcre_callout_block cb;          pcre_callout_block cb;
1223          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1224          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1225          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1226          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
1227          cb.subject_length   = md->end_subject - md->start_subject;          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1228          cb.start_match      = mstart - md->start_subject;          cb.start_match      = (int)(mstart - md->start_subject);
1229          cb.current_position = eptr - md->start_subject;          cb.current_position = (int)(eptr - md->start_subject);
1230          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1231          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1232          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1233          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1234          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1235            cb.mark             = markptr;
1236          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1237          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1238          }          }
# Line 916  for (;;) Line 1253  for (;;)
1253        else        else
1254          {          {
1255          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/          int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1256          condition =  (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1257    
1258          /* If the test is for recursion into a specific subpattern, and it is          /* If the test is for recursion into a specific subpattern, and it is
1259          false, but the test was set up by name, scan the table to see if the          false, but the test was set up by name, scan the table to see if the
1260          name refers to any other numbers, and test them. The condition is true          name refers to any other numbers, and test them. The condition is true
1261          if any one is set. */          if any one is set. */
1262    
1263          if (!condition && condcode == OP_NRREF && recno != RREF_ANY)          if (!condition && condcode == OP_NRREF)
1264            {            {
1265            uschar *slotA = md->name_table;            uschar *slotA = md->name_table;
1266            for (i = 0; i < md->name_count; i++)            for (i = 0; i < md->name_count; i++)
# Line 1049  for (;;) Line 1386  for (;;)
1386        }        }
1387    
1388      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1389      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1390      assertion. */      an assertion. */
1391    
1392      else      else
1393        {        {
1394        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1395            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1396        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1397          {          {
1398            if (md->end_offset_top > offset_top)
1399              offset_top = md->end_offset_top;  /* Captures may have happened */
1400          condition = TRUE;          condition = TRUE;
1401          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1402          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1403          }          }
1404    
1405          /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1406          assertion; it is therefore treated as NOMATCH. */
1407    
1408        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1409          {          {
1410          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
# Line 1073  for (;;) Line 1416  for (;;)
1416          }          }
1417        }        }
1418    
1419      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one, can
1420      we can use tail recursion to avoid using another stack frame, except when      use tail recursion to avoid using another stack frame, except when there is
1421      match_cbegroup is required for an unlimited repeat of a possibly empty      unlimited repeat of a possibly empty group. In the latter case, a recursive
1422      group. If the second alternative doesn't exist, we can just plough on. */      call to match() is always required, unless the second alternative doesn't
1423        exist, in which case we can just plough on. Note that, for compatibility
1424        with Perl, the | in a conditional group is NOT treated as creating two
1425        alternatives. If a THEN is encountered in the branch, it propagates out to
1426        the enclosing alternative (unless nested in a deeper set of alternatives,
1427        of course). */
1428    
1429      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1430        {        {
1431        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
         {  
         RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);  
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
1432          {          {
1433          flags = 0;          ecode += 1 + LINK_SIZE;
1434          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1435          }          }
1436    
1437          md->match_function_type = MATCH_CBEGROUP;
1438          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1439          RRETURN(rrc);
1440        }        }
1441      else                         /* Condition false & no alternative */  
1442         /* Condition false & no alternative; continue after the group. */
1443    
1444        else
1445        {        {
1446        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1447        }        }
# Line 1116  for (;;) Line 1465  for (;;)
1465        {        {
1466        md->offset_vector[offset] =        md->offset_vector[offset] =
1467          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
1468        md->offset_vector[offset+1] = eptr - md->start_subject;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1469        if (offset_top <= offset) offset_top = offset + 2;        if (offset_top <= offset) offset_top = offset + 2;
1470        }        }
1471      ecode += 3;      ecode += 3;
1472      break;      break;
1473    
1474    
1475      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. */
     recursion, we should restore the offsets appropriately and continue from  
     after the call. */  
1476    
     case OP_ACCEPT:  
1477      case OP_END:      case OP_END:
1478      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1479        {      case OP_ASSERT_ACCEPT:
       recursion_info *rec = md->recursive;  
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
       md->recursive = rec->prevrec;  
       memmove(md->offset_vector, rec->offset_save,  
         rec->saved_max * sizeof(int));  
       offset_top = rec->save_offset_top;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1480    
1481      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is      /* If we have matched an empty string, fail if not in an assertion and not
1482      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1483      the subject. In both cases, backtracking will then try other alternatives,      is set and we have matched at the start of the subject. In both cases,
1484      if any. */      backtracking will then try other alternatives, if any. */
1485    
1486      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1487          (md->notempty ||           md->recursive == NULL &&
1488            (md->notempty_atstart &&           (md->notempty ||
1489              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1490                 mstart == md->start_subject + md->start_offset)))
1491        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1492    
1493      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1158  for (;;) Line 1495  for (;;)
1495      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1496      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1497      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
     MRRETURN(((op == OP_END)? MATCH_MATCH : MATCH_ACCEPT));  
1498    
1499      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1500        given as the argument to MRRETURN when the heap is in use. */
1501    
1502      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1503      ims = ecode[1];      MRRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1504    
1505      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1506      matching won't pass the KET for an assertion. If any one branch matches,      matching won't pass the KET for an assertion. If any one branch matches,
1507      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
1508      start of each branch to move the current point backwards, so the code at      start of each branch to move the current point backwards, so the code at
1509      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1510        of a condition, we want to return immediately afterwards. The caller of
1511        this incarnation of the match() function will have set MATCH_CONDASSERT in
1512        md->match_function type, and one of these opcodes will be the first opcode
1513        that is processed. We use a local variable that is preserved over calls to
1514        match() to remember this case. */
1515    
1516      case OP_ASSERT:      case OP_ASSERT:
1517      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1518        if (md->match_function_type == MATCH_CONDASSERT)
1519          {
1520          condassert = TRUE;
1521          md->match_function_type = 0;
1522          }
1523        else condassert = FALSE;
1524    
1525      do      do
1526        {        {
1527        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1528        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1529          {          {
1530          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1531            markptr = md->mark;
1532          break;          break;
1533          }          }
1534    
1535          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1536          as NOMATCH. */
1537    
1538        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1539        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1540        }        }
1541      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1542    
1543      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1544    
1545      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1546    
1547      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1548    
1549      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1550      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1209  for (;;) Line 1560  for (;;)
1560    
1561      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1562      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1563        if (md->match_function_type == MATCH_CONDASSERT)
1564          {
1565          condassert = TRUE;
1566          md->match_function_type = 0;
1567          }
1568        else condassert = FALSE;
1569    
1570      do      do
1571        {        {
1572        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
         RM5);  
1573        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1574        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1575          {          {
1576          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1577          break;          break;
1578          }          }
1579    
1580          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1581          as NOMATCH. */
1582    
1583        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1584        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1585        }        }
1586      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1587    
1588      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1589    
1590      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1591      continue;      continue;
# Line 1270  for (;;) Line 1631  for (;;)
1631      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1632        {        {
1633        pcre_callout_block cb;        pcre_callout_block cb;
1634        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1635        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1636        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1637        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1638        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1639        cb.start_match      = mstart - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1640        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1641        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1642        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1643        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1644        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1645        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1646          cb.mark             = markptr;
1647        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1648        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1649        }        }
# Line 1292  for (;;) Line 1654  for (;;)
1654      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
1655      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1656    
1657      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1658      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
1659      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
1660      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
1661      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
1662      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
1663      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.  
1664    
1665      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
1666      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
1667      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1668        a lot, so he is not to blame for the current way it works. */
1669    
1670      case OP_RECURSE:      case OP_RECURSE:
1671        {        {
1672          recursion_info *ri;
1673          int recno;
1674    
1675        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1676        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1677          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1678    
1679          /* Check for repeating a recursion without advancing the subject pointer.
1680          This should catch convoluted mutual recursions. (Some simple cases are
1681          caught at compile time.) */
1682    
1683          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1684            if (recno == ri->group_num && eptr == ri->subject_position)
1685              RRETURN(PCRE_ERROR_RECURSELOOP);
1686    
1687        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1688    
1689          new_recursive.group_num = recno;
1690          new_recursive.subject_position = eptr;
1691        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1692        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1693    
1694        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1695    
1696        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1697    
1698        /* Now save the offset data. */        /* Now save the offset data */
1699    
1700        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1701        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1334  for (;;) Line 1706  for (;;)
1706            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1707          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1708          }          }
   
1709        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1710              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1711    
1712        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1713        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1714          might be changed, so reset it before looping. */
1715    
1716        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1717        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1718        do        do
1719          {          {
1720            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1721          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1722            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1723            memcpy(md->offset_vector, new_recursive.offset_save,
1724                new_recursive.saved_max * sizeof(int));
1725            md->recursive = new_recursive.prevrec;
1726          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1727            {            {
1728            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1729            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1730              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1731            MRRETURN(MATCH_MATCH);  
1732              /* Set where we got to in the subject, and reset the start in case
1733              it was changed by \K. This *is* propagated back out of a recursion,
1734              for Perl compatibility. */
1735    
1736              eptr = md->end_match_ptr;
1737              mstart = md->start_match_ptr;
1738              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1739            }            }
1740    
1741            /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1742            as NOMATCH. */
1743    
1744          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1745            {            {
1746            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
# Line 1365  for (;;) Line 1750  for (;;)
1750            }            }
1751    
1752          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1753          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1754          }          }
1755        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1377  for (;;) Line 1760  for (;;)
1760          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1761        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1762        }        }
     /* 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;  
1763    
1764      do      RECURSION_MATCHED:
1765        {      break;
       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;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
   
     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 */  
1766    
1767      /* 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
1768      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1472  for (;;) Line 1778  for (;;)
1778      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1779    
1780      case OP_BRAZERO:      case OP_BRAZERO:
1781        {      next = ecode + 1;
1782        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1783        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1784        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1785        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
1786        ecode = next + 1 + LINK_SIZE;      break;
1787        }  
1788      break;      case OP_BRAMINZERO:
1789        next = ecode + 1;
1790      case OP_BRAMINZERO:      do next += GET(next, 1); while (*next == OP_ALT);
1791        {      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1792        next = ecode+1;      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1793        do next += GET(next, 1); while (*next == OP_ALT);      ecode++;
       RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode++;  
       }  
1794      break;      break;
1795    
1796      case OP_SKIPZERO:      case OP_SKIPZERO:
1797        {      next = ecode+1;
1798        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1799        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1800      break;      break;
1801    
1802        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1803        here; just jump to the group, with allow_zero set TRUE. */
1804    
1805        case OP_BRAPOSZERO:
1806        op = *(++ecode);
1807        allow_zero = TRUE;
1808        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1809          goto POSSESSIVE_NON_CAPTURE;
1810    
1811      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1812    
1813      case OP_KET:      case OP_KET:
1814      case OP_KETRMIN:      case OP_KETRMIN:
1815      case OP_KETRMAX:      case OP_KETRMAX:
1816        case OP_KETRPOS:
1817      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1818    
1819      /* 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
1820      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1821      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1822    
1823      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1824        {        {
1825        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1826        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1827        }        }
1828      else saved_eptr = NULL;      else saved_eptr = NULL;
1829    
1830      /* If we are at the end of an assertion group or an atomic group, stop      /* If we are at the end of an assertion group or a non-capturing atomic
1831      matching and return MATCH_MATCH, but record the current high water mark for      group, stop matching and return MATCH_MATCH, but record the current high
1832      use by positive assertions. We also need to record the match start in case      water mark for use by positive assertions. We also need to record the match
1833      it was changed by \K. */      start in case it was changed by \K. */
1834    
1835      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) ||
1836          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||           *prev == OP_ONCE_NC)
         *prev == OP_ONCE)  
1837        {        {
1838        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE_NC */
1839        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1840        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1841        MRRETURN(MATCH_MATCH);        MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1842        }        }
1843    
1844      /* 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
1845      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1846      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
1847      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
1848      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
1849        the current subject position and start match pointer and give a MATCH
1850        return. */
1851    
1852      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1853            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1854        {        {
1855        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1856        offset = number << 1;        offset = number << 1;
# Line 1548  for (;;) Line 1860  for (;;)
1860        printf("\n");        printf("\n");
1861  #endif  #endif
1862    
1863          /* Handle a recursively called group. */
1864    
1865          if (md->recursive != NULL && md->recursive->group_num == number)
1866            {
1867            md->end_match_ptr = eptr;
1868            md->start_match_ptr = mstart;
1869            RRETURN(MATCH_MATCH);
1870            }
1871    
1872          /* Deal with capturing */
1873    
1874        md->capture_last = number;        md->capture_last = number;
1875        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1876          {          {
1877            /* If offset is greater than offset_top, it means that we are
1878            "skipping" a capturing group, and that group's offsets must be marked
1879            unset. In earlier versions of PCRE, all the offsets were unset at the
1880            start of matching, but this doesn't work because atomic groups and
1881            assertions can cause a value to be set that should later be unset.
1882            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1883            part of the atomic group, but this is not on the final matching path,
1884            so must be unset when 2 is set. (If there is no group 2, there is no
1885            problem, because offset_top will then be 2, indicating no capture.) */
1886    
1887            if (offset > offset_top)
1888              {
1889              register int *iptr = md->offset_vector + offset_top;
1890              register int *iend = md->offset_vector + offset;
1891              while (iptr < iend) *iptr++ = -1;
1892              }
1893    
1894            /* Now make the extraction */
1895    
1896          md->offset_vector[offset] =          md->offset_vector[offset] =
1897            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1898          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1899          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1900          }          }
1901          }
1902    
1903        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1904        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1905        group. This is the forcible breaking of infinite loops as implemented in
1906        Perl 5.005. For a non-repeating atomic group that includes captures,
1907        establish a backup point by processing the rest of the pattern at a lower
1908        level. If this results in a NOMATCH return, pass MATCH_ONCE back to the
1909        original OP_ONCE level, thereby bypassing intermediate backup points, but
1910        resetting any captures that happened along the way. */
1911    
1912        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1913          {
1914          if (*prev == OP_ONCE)
1915          {          {
1916          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1917          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1918          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1919          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;  
1920          }          }
1921          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1922          break;
1923        }        }
1924    
1925      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1926      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1927        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. */  
1928    
1929      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KETRPOS)
1930        {        {
1931        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1932        break;        md->end_offset_top = offset_top;
1933          RRETURN(MATCH_KETRPOS);
1934        }        }
1935    
1936      /* 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
1937      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
1938      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
1939      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
1940        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1941    
1942      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1943        {        {
1944        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1945        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1946        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1947          {          {
1948          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1949            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1950            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1951            RRETURN(MATCH_ONCE);
1952            }
1953          if (*prev >= OP_SBRA)    /* Could match an empty string */
1954            {
1955            md->match_function_type = MATCH_CBEGROUP;
1956            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1957          RRETURN(rrc);          RRETURN(rrc);
1958          }          }
1959        ecode = prev;        ecode = prev;
# Line 1613  for (;;) Line 1961  for (;;)
1961        }        }
1962      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1963        {        {
1964        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1965          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1966          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1967        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1968          if (*prev == OP_ONCE)
1969            {
1970            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1971            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1972            md->once_target = prev;
1973            RRETURN(MATCH_ONCE);
1974            }
1975        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1976        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1977        }        }
1978      /* Control never gets here */      /* Control never gets here */
1979    
1980      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1981    
1982      case OP_CIRC:      case OP_CIRC:
1983      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 */  
1984    
1985      /* Start of subject assertion */      /* Start of subject assertion */
1986    
# Line 1642  for (;;) Line 1989  for (;;)
1989      ecode++;      ecode++;
1990      break;      break;
1991    
1992        /* Multiline mode: start of subject unless notbol, or after any newline. */
1993    
1994        case OP_CIRCM:
1995        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1996        if (eptr != md->start_subject &&
1997            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1998          MRRETURN(MATCH_NOMATCH);
1999        ecode++;
2000        break;
2001    
2002      /* Start of match assertion */      /* Start of match assertion */
2003    
2004      case OP_SOM:      case OP_SOM:
# Line 1656  for (;;) Line 2013  for (;;)
2013      ecode++;      ecode++;
2014      break;      break;
2015    
2016      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
2017      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
2018    
2019      case OP_DOLL:      case OP_DOLLM:
2020      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
2021        {        { 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;  
       }  
2022      else      else
2023        {        {
2024        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
2025        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           MRRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
2026        }        }
2027        ecode++;
2028        break;
2029    
2030        /* Not multiline mode: assert before a terminating newline or before end of
2031        subject unless noteol is set. */
2032    
2033        case OP_DOLL:
2034        if (md->noteol) MRRETURN(MATCH_NOMATCH);
2035        if (!md->endonly) goto ASSERT_NL_OR_EOS;
2036    
2037      /* ... else fall through for endonly */      /* ... else fall through for endonly */
2038    
2039      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
2040    
2041      case OP_EOD:      case OP_EOD:
2042      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
2043        SCHECK_PARTIAL();
2044      ecode++;      ecode++;
2045      break;      break;
2046    
2047      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
2048    
2049      case OP_EODN:      case OP_EODN:
2050      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
2051        if (eptr < md->end_subject &&
2052          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
2053        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2054    
2055        /* Either at end of string or \n before end. */
2056    
2057        SCHECK_PARTIAL();
2058      ecode++;      ecode++;
2059      break;      break;
2060    
# Line 1713  for (;;) Line 2072  for (;;)
2072  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2073        if (utf8)        if (utf8)
2074          {          {
2075            /* Get status of previous character */
2076    
2077          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2078            {            {
2079            USPTR lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
2080            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
2081            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;            if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
2082            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2083    #ifdef SUPPORT_UCP
2084              if (md->use_ucp)
2085                {
2086                if (c == '_') prev_is_word = TRUE; else
2087                  {
2088                  int cat = UCD_CATEGORY(c);
2089                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2090                  }
2091                }
2092              else
2093    #endif
2094            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2095            }            }
2096    
2097            /* Get status of next character */
2098    
2099          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2100            {            {
2101            SCHECK_PARTIAL();            SCHECK_PARTIAL();
# Line 1729  for (;;) Line 2104  for (;;)
2104          else          else
2105            {            {
2106            GETCHAR(c, eptr);            GETCHAR(c, eptr);
2107    #ifdef SUPPORT_UCP
2108              if (md->use_ucp)
2109                {
2110                if (c == '_') cur_is_word = TRUE; else
2111                  {
2112                  int cat = UCD_CATEGORY(c);
2113                  cur_is_word = (cat == ucp_L || cat == ucp_N);
2114                  }
2115                }
2116              else
2117    #endif
2118            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2119            }            }
2120          }          }
2121        else        else
2122  #endif  #endif
2123    
2124        /* Not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
2125          consistency with the behaviour of \w we do use it in this case. */
2126    
2127          {          {
2128            /* Get status of previous character */
2129    
2130          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
2131            {            {
2132            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;            if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
2133    #ifdef SUPPORT_UCP
2134              if (md->use_ucp)
2135                {
2136                c = eptr[-1];
2137                if (c == '_') prev_is_word = TRUE; else
2138                  {
2139                  int cat = UCD_CATEGORY(c);
2140                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2141                  }
2142                }
2143              else
2144    #endif
2145            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);            prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2146            }            }
2147    
2148            /* Get status of next character */
2149    
2150          if (eptr >= md->end_subject)          if (eptr >= md->end_subject)
2151            {            {
2152            SCHECK_PARTIAL();            SCHECK_PARTIAL();
2153            cur_is_word = FALSE;            cur_is_word = FALSE;
2154            }            }
2155          else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);          else
2156    #ifdef SUPPORT_UCP
2157            if (md->use_ucp)
2158              {
2159              c = *eptr;
2160              if (c == '_') cur_is_word = TRUE; else
2161                {
2162                int cat = UCD_CATEGORY(c);
2163                cur_is_word = (cat == ucp_L || cat == ucp_N);
2164                }
2165              }
2166            else
2167    #endif
2168            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2169          }          }
2170    
2171        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
# Line 1766  for (;;) Line 2183  for (;;)
2183      /* Fall through */      /* Fall through */
2184    
2185      case OP_ALLANY:      case OP_ALLANY:
2186      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2187        {        {                            /* not be updated before SCHECK_PARTIAL. */
2188        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2189        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2190        }        }
2191        eptr++;
2192      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2193      ecode++;      ecode++;
2194      break;      break;
# Line 1779  for (;;) Line 2197  for (;;)
2197      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2198    
2199      case OP_ANYBYTE:      case OP_ANYBYTE:
2200      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2201        {        {                            /* not be updated before SCHECK_PARTIAL. */
2202        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2203        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2204        }        }
2205        eptr++;
2206      ecode++;      ecode++;
2207      break;      break;
2208    
# Line 1899  for (;;) Line 2318  for (;;)
2318      switch(c)      switch(c)
2319        {        {
2320        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2321    
2322        case 0x000d:        case 0x000d:
2323        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2324        break;        break;
# Line 2055  for (;;) Line 2475  for (;;)
2475               prop->chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2476               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2477            MRRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2478           break;          break;
2479    
2480          case PT_GC:          case PT_GC:
2481          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
# Line 2072  for (;;) Line 2492  for (;;)
2492            MRRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2493          break;          break;
2494    
2495            /* These are specials */
2496    
2497            case PT_ALNUM:
2498            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2499                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2500              MRRETURN(MATCH_NOMATCH);
2501            break;
2502    
2503            case PT_SPACE:    /* Perl space */
2504            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2505                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2506                   == (op == OP_NOTPROP))
2507              MRRETURN(MATCH_NOMATCH);
2508            break;
2509    
2510            case PT_PXSPACE:  /* POSIX space */
2511            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2512                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2513                 c == CHAR_FF || c == CHAR_CR)
2514                   == (op == OP_NOTPROP))
2515              MRRETURN(MATCH_NOMATCH);
2516            break;
2517    
2518            case PT_WORD:
2519            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2520                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2521                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2522              MRRETURN(MATCH_NOMATCH);
2523            break;
2524    
2525            /* This should never occur */
2526    
2527          default:          default:
2528          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2529          }          }
# Line 2090  for (;;) Line 2542  for (;;)
2542        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2543        }        }
2544      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2545        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2546        while (eptr < md->end_subject)
2547        {        {
2548        int category = UCD_CATEGORY(c);        int len = 1;
2549        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2550        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2551          {        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;  
         }  
2552        }        }
2553      ecode++;      ecode++;
2554      break;      break;
# Line 2119  for (;;) Line 2564  for (;;)
2564      loops). */      loops). */
2565    
2566      case OP_REF:      case OP_REF:
2567        {      case OP_REFI:
2568        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2569        ecode += 3;      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2570        ecode += 3;
2571    
2572        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2573    
2574        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2575        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
2576        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.  
2577    
2578        (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
2579        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2580    
2581        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
2582        referenced subpattern. */      referenced subpattern. */
2583    
2584        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2585          length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;        length = (md->jscript_compat)? 0 : -1;
2586        else      else
2587          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2588    
2589        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2590    
2591        switch (*ecode)      switch (*ecode)
2592          {        {
2593          case OP_CRSTAR:        case OP_CRSTAR:
2594          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2595          case OP_CRPLUS:        case OP_CRPLUS:
2596          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2597          case OP_CRQUERY:        case OP_CRQUERY:
2598          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2599          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2600          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2601          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2602          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2603          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2604          break;        break;
2605    
2606          case OP_CRRANGE:        case OP_CRRANGE:
2607          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2608          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2609          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2610          max = GET2(ecode, 3);        max = GET2(ecode, 3);
2611          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2612          ecode += 5;        ecode += 5;
2613          break;        break;
2614    
2615          default:               /* No repeat follows */        default:               /* No repeat follows */
2616          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2617            {          {
2618            CHECK_PARTIAL();          CHECK_PARTIAL();
2619            MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
           }  
         eptr += length;  
         continue;              /* With the main loop */  
2620          }          }
2621          eptr += length;
2622          continue;              /* With the main loop */
2623          }
2624    
2625        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2626        main loop. */      zero, just continue with the main loop. */
2627    
2628        if (length == 0) continue;      if (length == 0) continue;
2629    
2630        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2631        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2632        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2633    
2634        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2635          {
2636          int slength;
2637          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2638          {          {
2639          if (!match_ref(offset, eptr, length, md, ims))          CHECK_PARTIAL();
2640            {          MRRETURN(MATCH_NOMATCH);
           CHECK_PARTIAL();  
           MRRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2641          }          }
2642          eptr += slength;
2643          }
2644    
2645        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2646        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2647    
2648        if (min == max) continue;      if (min == max) continue;
2649    
2650        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2651    
2652        if (minimize)      if (minimize)
2653          {
2654          for (fi = min;; fi++)
2655          {          {
2656          for (fi = min;; fi++)          int slength;
2657            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2658            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2659            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2660            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2661            {            {
2662            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2663            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;  
2664            }            }
2665          /* Control never gets here */          eptr += slength;
2666          }          }
2667          /* Control never gets here */
2668          }
2669    
2670        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2671    
2672        else      else
2673          {
2674          pp = eptr;
2675          for (i = min; i < max; i++)
2676          {          {
2677          pp = eptr;          int slength;
2678          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2679            {            {
2680            if (!match_ref(offset, eptr, length, md, ims))            CHECK_PARTIAL();
2681              {            break;
             CHECK_PARTIAL();  
             break;  
             }  
           eptr += length;  
           }  
         while (eptr >= pp)  
           {  
           RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);  
           if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
           eptr -= length;  
2682            }            }
2683          MRRETURN(MATCH_NOMATCH);          eptr += slength;
2684            }
2685          while (eptr >= pp)
2686            {
2687            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2688            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2689            eptr -= length;
2690          }          }
2691          MRRETURN(MATCH_NOMATCH);
2692        }        }
2693      /* Control never gets here */      /* Control never gets here */
2694    
# Line 2347  for (;;) Line 2794  for (;;)
2794            {            {
2795            for (fi = min;; fi++)            for (fi = min;; fi++)
2796              {              {
2797              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2798              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2799              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2800              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2372  for (;;) Line 2819  for (;;)
2819            {            {
2820            for (fi = min;; fi++)            for (fi = min;; fi++)
2821              {              {
2822              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2823              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2824              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2825              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2418  for (;;) Line 2865  for (;;)
2865              }              }
2866            for (;;)            for (;;)
2867              {              {
2868              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2869              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2870              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2871              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2441  for (;;) Line 2888  for (;;)
2888              }              }
2889            while (eptr >= pp)            while (eptr >= pp)
2890              {              {
2891              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2892              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2893              eptr--;              eptr--;
2894              }              }
# Line 2517  for (;;) Line 2964  for (;;)
2964          {          {
2965          for (fi = min;; fi++)          for (fi = min;; fi++)
2966            {            {
2967            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2968            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2969            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2970            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2550  for (;;) Line 2997  for (;;)
2997            }            }
2998          for(;;)          for(;;)
2999            {            {
3000            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
3001            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3002            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3003            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2595  for (;;) Line 3042  for (;;)
3042    
3043      /* Match a single character, caselessly */      /* Match a single character, caselessly */
3044    
3045      case OP_CHARNC:      case OP_CHARI:
3046  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3047      if (utf8)      if (utf8)
3048        {        {
# Line 2655  for (;;) Line 3102  for (;;)
3102      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3103    
3104      case OP_EXACT:      case OP_EXACT:
3105        case OP_EXACTI:
3106      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3107      ecode += 3;      ecode += 3;
3108      goto REPEATCHAR;      goto REPEATCHAR;
3109    
3110      case OP_POSUPTO:      case OP_POSUPTO:
3111        case OP_POSUPTOI:
3112      possessive = TRUE;      possessive = TRUE;
3113      /* Fall through */      /* Fall through */
3114    
3115      case OP_UPTO:      case OP_UPTO:
3116        case OP_UPTOI:
3117      case OP_MINUPTO:      case OP_MINUPTO:
3118        case OP_MINUPTOI:
3119      min = 0;      min = 0;
3120      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3121      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3122      ecode += 3;      ecode += 3;
3123      goto REPEATCHAR;      goto REPEATCHAR;
3124    
3125      case OP_POSSTAR:      case OP_POSSTAR:
3126        case OP_POSSTARI:
3127      possessive = TRUE;      possessive = TRUE;
3128      min = 0;      min = 0;
3129      max = INT_MAX;      max = INT_MAX;
# Line 2679  for (;;) Line 3131  for (;;)
3131      goto REPEATCHAR;      goto REPEATCHAR;
3132    
3133      case OP_POSPLUS:      case OP_POSPLUS:
3134        case OP_POSPLUSI:
3135      possessive = TRUE;      possessive = TRUE;
3136      min = 1;      min = 1;
3137      max = INT_MAX;      max = INT_MAX;
# Line 2686  for (;;) Line 3139  for (;;)
3139      goto REPEATCHAR;      goto REPEATCHAR;
3140    
3141      case OP_POSQUERY:      case OP_POSQUERY:
3142        case OP_POSQUERYI:
3143      possessive = TRUE;      possessive = TRUE;
3144      min = 0;      min = 0;
3145      max = 1;      max = 1;
# Line 2693  for (;;) Line 3147  for (;;)
3147      goto REPEATCHAR;      goto REPEATCHAR;
3148    
3149      case OP_STAR:      case OP_STAR:
3150        case OP_STARI:
3151      case OP_MINSTAR:      case OP_MINSTAR:
3152        case OP_MINSTARI:
3153      case OP_PLUS:      case OP_PLUS:
3154        case OP_PLUSI:
3155      case OP_MINPLUS:      case OP_MINPLUS:
3156        case OP_MINPLUSI:
3157      case OP_QUERY:      case OP_QUERY:
3158        case OP_QUERYI:
3159      case OP_MINQUERY:      case OP_MINQUERY:
3160      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3161        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3162      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3163      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3164      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3165      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2723  for (;;) Line 3182  for (;;)
3182          {          {
3183  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3184          unsigned int othercase;          unsigned int othercase;
3185          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3186              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3187            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3188          else oclength = 0;          else oclength = 0;
# Line 2751  for (;;) Line 3210  for (;;)
3210            {            {
3211            for (fi = min;; fi++)            for (fi = min;; fi++)
3212              {              {
3213              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3214              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3215              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3216              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2793  for (;;) Line 3252  for (;;)
3252    
3253            for(;;)            for(;;)
3254              {              {
3255              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3256              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3257              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3258  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2830  for (;;) Line 3289  for (;;)
3289      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3290        max, eptr));        max, eptr));
3291    
3292      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3293        {        {
3294        fc = md->lcc[fc];        fc = md->lcc[fc];
3295        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2847  for (;;) Line 3306  for (;;)
3306          {          {
3307          for (fi = min;; fi++)          for (fi = min;; fi++)
3308            {            {
3309            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3310            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3311            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3312            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2877  for (;;) Line 3336  for (;;)
3336    
3337          while (eptr >= pp)          while (eptr >= pp)
3338            {            {
3339            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3340            eptr--;            eptr--;
3341            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3342            }            }
# Line 2906  for (;;) Line 3365  for (;;)
3365          {          {
3366          for (fi = min;; fi++)          for (fi = min;; fi++)
3367            {            {
3368            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3369            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3370            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3371            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2935  for (;;) Line 3394  for (;;)
3394    
3395          while (eptr >= pp)          while (eptr >= pp)
3396            {            {
3397            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3398            eptr--;            eptr--;
3399            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3400            }            }
# Line 2948  for (;;) Line 3407  for (;;)
3407      checking can be multibyte. */      checking can be multibyte. */
3408    
3409      case OP_NOT:      case OP_NOT:
3410        case OP_NOTI:
3411      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3412        {        {
3413        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 2955  for (;;) Line 3415  for (;;)
3415        }        }
3416      ecode++;      ecode++;
3417      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3418      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3419        {        {
3420  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3421        if (c < 256)        if (c < 256)
# Line 2963  for (;;) Line 3423  for (;;)
3423        c = md->lcc[c];        c = md->lcc[c];
3424        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3425        }        }
3426      else      else    /* Caseful */
3427        {        {
3428        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3429        }        }
# Line 2977  for (;;) Line 3437  for (;;)
3437      about... */      about... */
3438    
3439      case OP_NOTEXACT:      case OP_NOTEXACT:
3440        case OP_NOTEXACTI:
3441      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3442      ecode += 3;      ecode += 3;
3443      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3444    
3445      case OP_NOTUPTO:      case OP_NOTUPTO:
3446        case OP_NOTUPTOI:
3447      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3448        case OP_NOTMINUPTOI:
3449      min = 0;      min = 0;
3450      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3451      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3452      ecode += 3;      ecode += 3;
3453      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3454    
3455      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3456        case OP_NOTPOSSTARI:
3457      possessive = TRUE;      possessive = TRUE;
3458      min = 0;      min = 0;
3459      max = INT_MAX;      max = INT_MAX;
# Line 2997  for (;;) Line 3461  for (;;)
3461      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3462    
3463      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3464        case OP_NOTPOSPLUSI:
3465      possessive = TRUE;      possessive = TRUE;
3466      min = 1;      min = 1;
3467      max = INT_MAX;      max = INT_MAX;
# Line 3004  for (;;) Line 3469  for (;;)
3469      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3470    
3471      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3472        case OP_NOTPOSQUERYI:
3473      possessive = TRUE;      possessive = TRUE;
3474      min = 0;      min = 0;
3475      max = 1;      max = 1;
# Line 3011  for (;;) Line 3477  for (;;)
3477      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3478    
3479      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3480        case OP_NOTPOSUPTOI:
3481      possessive = TRUE;      possessive = TRUE;
3482      min = 0;      min = 0;
3483      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 3018  for (;;) Line 3485  for (;;)
3485      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3486    
3487      case OP_NOTSTAR:      case OP_NOTSTAR:
3488        case OP_NOTSTARI:
3489      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3490        case OP_NOTMINSTARI:
3491      case OP_NOTPLUS:      case OP_NOTPLUS:
3492        case OP_NOTPLUSI:
3493      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3494        case OP_NOTMINPLUSI:
3495      case OP_NOTQUERY:      case OP_NOTQUERY:
3496        case OP_NOTQUERYI:
3497      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3498      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3499        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3500      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3501      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3502      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3045  for (;;) Line 3518  for (;;)
3518      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,
3519        max, eptr));        max, eptr));
3520    
3521      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3522        {        {
3523        fc = md->lcc[fc];        fc = md->lcc[fc];
3524    
# Line 3093  for (;;) Line 3566  for (;;)
3566            register unsigned int d;            register unsigned int d;
3567            for (fi = min;; fi++)            for (fi = min;; fi++)
3568              {              {
3569              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3570              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3571              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3572              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3112  for (;;) Line 3585  for (;;)
3585            {            {
3586            for (fi = min;; fi++)            for (fi = min;; fi++)
3587              {              {
3588              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3589              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3590              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3591              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3153  for (;;) Line 3626  for (;;)
3626          if (possessive) continue;          if (possessive) continue;
3627          for(;;)          for(;;)
3628              {              {
3629              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3630              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3631              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3632              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3176  for (;;) Line 3649  for (;;)
3649            if (possessive) continue;            if (possessive) continue;
3650            while (eptr >= pp)            while (eptr >= pp)
3651              {              {
3652              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3653              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3654              eptr--;              eptr--;
3655              }              }
# Line 3233  for (;;) Line 3706  for (;;)
3706            register unsigned int d;            register unsigned int d;
3707            for (fi = min;; fi++)            for (fi = min;; fi++)
3708              {              {
3709              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3710              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3711              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3712              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3251  for (;;) Line 3724  for (;;)
3724            {            {
3725            for (fi = min;; fi++)            for (fi = min;; fi++)
3726              {              {
3727              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3728              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3729              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3730              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3291  for (;;) Line 3764  for (;;)
3764            if (possessive) continue;            if (possessive) continue;
3765            for(;;)            for(;;)
3766              {              {
3767              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3768              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3769              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3770              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3314  for (;;) Line 3787  for (;;)
3787            if (possessive) continue;            if (possessive) continue;
3788            while (eptr >= pp)            while (eptr >= pp)
3789              {              {
3790              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3791              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3792              eptr--;              eptr--;
3793              }              }
# Line 3406  for (;;) Line 3879  for (;;)
3879      is tidier. Also separate the UCP code, which can be the same for both UTF-8      is tidier. Also separate the UCP code, which can be the same for both UTF-8
3880      and single-bytes. */      and single-bytes. */
3881    
3882      if (min > 0)      if (min > 0)
3883        {        {
3884  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3885        if (prop_type >= 0)        if (prop_type >= 0)
3886          {          {
3887          switch(prop_type)          switch(prop_type)
3888            {            {
3889            case PT_ANY:            case PT_ANY:
3890            if (prop_fail_result) MRRETURN(MATCH_NOMATCH);            if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
3891              for (i = 1; i <= min; i++)
3892                {
3893                if (eptr >= md->end_subject)
3894                  {
3895                  SCHECK_PARTIAL();
3896                  MRRETURN(MATCH_NOMATCH);
3897                  }
3898                GETCHARINCTEST(c, eptr);
3899                }
3900              break;
3901    
3902              case PT_LAMP:
3903              for (i = 1; i <= min; i++)
3904                {
3905                int chartype;
3906                if (eptr >= md->end_subject)
3907                  {
3908                  SCHECK_PARTIAL();
3909                  MRRETURN(MATCH_NOMATCH);
3910                  }
3911                GETCHARINCTEST(c, eptr);
3912                chartype = UCD_CHARTYPE(c);
3913                if ((chartype == ucp_Lu ||
3914                     chartype == ucp_Ll ||
3915                     chartype == ucp_Lt) == prop_fail_result)
3916                  MRRETURN(MATCH_NOMATCH);
3917                }
3918              break;
3919    
3920              case PT_GC:
3921              for (i = 1; i <= min; i++)
3922                {
3923                if (eptr >= md->end_subject)
3924                  {
3925                  SCHECK_PARTIAL();
3926                  MRRETURN(MATCH_NOMATCH);
3927                  }
3928                GETCHARINCTEST(c, eptr);
3929                if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
3930                  MRRETURN(MATCH_NOMATCH);
3931                }
3932              break;
3933    
3934              case PT_PC:
3935              for (i = 1; i <= min; i++)
3936                {
3937                if (eptr >= md->end_subject)
3938                  {
3939                  SCHECK_PARTIAL();
3940                  MRRETURN(MATCH_NOMATCH);
3941                  }
3942                GETCHARINCTEST(c, eptr);
3943                if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
3944                  MRRETURN(MATCH_NOMATCH);
3945                }
3946              break;
3947    
3948              case PT_SC:
3949            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3950              {              {
3951              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3423  for (;;) Line 3954  for (;;)
3954                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3955                }                }
3956              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3957                if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
3958                  MRRETURN(MATCH_NOMATCH);
3959              }              }
3960            break;            break;
3961    
3962            case PT_LAMP:            case PT_ALNUM:
3963            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3964              {              {
3965                int category;
3966              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3967                {                {
3968                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3969                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3970                }                }
3971              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3972              prop_chartype = UCD_CHARTYPE(c);              category = UCD_CATEGORY(c);
3973              if ((prop_chartype == ucp_Lu ||              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                  prop_chartype == ucp_Ll ||  
                  prop_chartype == ucp_Lt) == prop_fail_result)  
3974                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3975              }              }
3976            break;            break;
3977    
3978            case PT_GC:            case PT_SPACE:    /* Perl space */
3979            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3980              {              {
3981              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3452  for (;;) Line 3984  for (;;)
3984                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3985                }                }
3986              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3987              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3988              if ((prop_category == prop_value) == prop_fail_result)                   c == CHAR_FF || c == CHAR_CR)
3989                       == prop_fail_result)
3990                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3991              }              }
3992            break;            break;
3993    
3994            case PT_PC:            case PT_PXSPACE:  /* POSIX space */
3995            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3996              {              {
3997              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3467  for (;;) Line 4000  for (;;)
4000                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4001                }                }
4002              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4003              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4004              if ((prop_chartype == prop_value) == prop_fail_result)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4005                       == prop_fail_result)
4006                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4007              }              }
4008            break;            break;
4009    
4010            case PT_SC:            case PT_WORD:
4011            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4012              {              {
4013                int category;
4014              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4015                {                {
4016                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4017                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4018                }                }
4019              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4020              prop_script = UCD_SCRIPT(c);              category = UCD_CATEGORY(c);
4021              if ((prop_script == prop_value) == prop_fail_result)              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
4022                       == prop_fail_result)
4023                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4024              }              }
4025            break;            break;
4026    
4027              /* This should not occur */
4028    
4029            default:            default:
4030            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
4031            }            }
# Line 3506  for (;;) Line 4044  for (;;)
4044              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4045              }              }
4046            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4047            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4048            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4049              {              {
4050              int len = 1;              int len = 1;
4051              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4052                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4053              eptr += len;              eptr += len;
4054              }              }
4055            }            }
# Line 3572  for (;;) Line 4107  for (;;)
4107            switch(c)            switch(c)
4108              {              {
4109              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4110    
4111              case 0x000d:              case 0x000d:
4112              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4113              break;              break;
# Line 3848  for (;;) Line 4384  for (;;)
4384            switch(*eptr++)            switch(*eptr++)
4385              {              {
4386              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4387    
4388              case 0x000d:              case 0x000d:
4389              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4390              break;              break;
4391    
4392              case 0x000a:              case 0x000a:
4393              break;              break;
4394    
# Line 4040  for (;;) Line 4578  for (;;)
4578            case PT_ANY:            case PT_ANY:
4579            for (fi = min;; fi++)            for (fi = min;; fi++)
4580              {              {
4581              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4582              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4583              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4584              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4048  for (;;) Line 4586  for (;;)
4586                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4587                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4588                }                }
4589              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4590              if (prop_fail_result) MRRETURN(MATCH_NOMATCH);              if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
4591              }              }
4592            /* Control never gets here */            /* Control never gets here */
# Line 4056  for (;;) Line 4594  for (;;)
4594            case PT_LAMP:            case PT_LAMP:
4595            for (fi = min;; fi++)            for (fi = min;; fi++)
4596              {              {
4597              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              int chartype;
4598                RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4599              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4600              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4601              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4064  for (;;) Line 4603  for (;;)
4603                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4604                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4605                }                }
4606              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4607              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4608              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4609                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4610                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4611                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4612              }              }
4613            /* Control never gets here */            /* Control never gets here */
# Line 4076  for (;;) Line 4615  for (;;)
4615            case PT_GC:            case PT_GC:
4616            for (fi = min;; fi++)            for (fi = min;; fi++)
4617              {              {
4618              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4619              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4620              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4621              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4084  for (;;) Line 4623  for (;;)
4623                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4624                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4625                }                }
4626              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4627              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4628                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4629              }              }
4630            /* Control never gets here */            /* Control never gets here */
# Line 4094  for (;;) Line 4632  for (;;)
4632            case PT_PC:            case PT_PC:
4633            for (fi = min;; fi++)            for (fi = min;; fi++)
4634              {              {
4635              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
4636              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4637              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4638              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4102  for (;;) Line 4640  for (;;)
4640                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4641                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4642                }                }
4643              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4644              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4645                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4646              }              }
4647            /* Control never gets here */            /* Control never gets here */
# Line 4112  for (;;) Line 4649  for (;;)
4649            case PT_SC:            case PT_SC:
4650            for (fi = min;; fi++)            for (fi = min;; fi++)
4651              {              {
4652              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
4653              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4654              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4655              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4120  for (;;) Line 4657  for (;;)
4657                SCHECK_PARTIAL();                SCHECK_PARTIAL();
4658                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4659                }                }
4660              GETCHARINC(c, eptr);              GETCHARINCTEST(c, eptr);
4661              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
4662              if ((prop_script == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
4663                }
4664              /* Control never gets here */
4665    
4666              case PT_ALNUM:
4667              for (fi = min;; fi++)
4668                {
4669                int category;
4670                RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4671                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4672                if (fi >= max) MRRETURN(MATCH_NOMATCH);
4673                if (eptr >= md->end_subject)
4674                  {
4675                  SCHECK_PARTIAL();
4676                  MRRETURN(MATCH_NOMATCH);
4677                  }
4678                GETCHARINCTEST(c, eptr);
4679                category = UCD_CATEGORY(c);
4680                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
4681                  MRRETURN(MATCH_NOMATCH);
4682                }
4683              /* Control never gets here */
4684    
4685              case PT_SPACE:    /* Perl space */
4686              for (fi = min;; fi++)
4687                {
4688                RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4689                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4690                if (fi >= max) MRRETURN(MATCH_NOMATCH);
4691                if (eptr >= md->end_subject)
4692                  {
4693                  SCHECK_PARTIAL();
4694                  MRRETURN(MATCH_NOMATCH);
4695                  }
4696                GETCHARINCTEST(c, eptr);
4697                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4698                     c == CHAR_FF || c == CHAR_CR)
4699                       == prop_fail_result)
4700                  MRRETURN(MATCH_NOMATCH);
4701                }
4702              /* Control never gets here */
4703    
4704              case PT_PXSPACE:  /* POSIX space */
4705              for (fi = min;; fi++)
4706                {
4707                RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4708                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4709                if (fi >= max) MRRETURN(MATCH_NOMATCH);
4710                if (eptr >= md->end_subject)
4711                  {
4712                  SCHECK_PARTIAL();
4713                  MRRETURN(MATCH_NOMATCH);
4714                  }
4715                GETCHARINCTEST(c, eptr);
4716                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
4717                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4718                       == prop_fail_result)
4719                  MRRETURN(MATCH_NOMATCH);
4720                }
4721              /* Control never gets here */
4722    
4723              case PT_WORD:
4724              for (fi = min;; fi++)
4725                {
4726                int category;
4727                RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4728                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4729                if (fi >= max) MRRETURN(MATCH_NOMATCH);
4730                if (eptr >= md->end_subject)
4731                  {
4732                  SCHECK_PARTIAL();
4733                  MRRETURN(MATCH_NOMATCH);
4734                  }
4735                GETCHARINCTEST(c, eptr);
4736                category = UCD_CATEGORY(c);
4737                if ((category == ucp_L ||
4738                     category == ucp_N ||
4739                     c == CHAR_UNDERSCORE)
4740                       == prop_fail_result)
4741                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4742              }              }
4743            /* Control never gets here */            /* Control never gets here */
4744    
4745              /* This should never occur */
4746    
4747            default:            default:
4748            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
4749            }            }
# Line 4139  for (;;) Line 4756  for (;;)
4756          {          {
4757          for (fi = min;; fi++)          for (fi = min;; fi++)
4758            {            {
4759            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4760            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4761            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4762            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4148  for (;;) Line 4765  for (;;)
4765              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4766              }              }
4767            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4768            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4769            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4770              {              {
4771              int len = 1;              int len = 1;
4772              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4773                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4774              eptr += len;              eptr += len;
4775              }              }
4776            }            }
4777          }          }
   
4778        else        else
4779  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4780    
# Line 4171  for (;;) Line 4784  for (;;)
4784          {          {
4785          for (fi = min;; fi++)          for (fi = min;; fi++)
4786            {            {
4787            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4788            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4789            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4790            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4334  for (;;) Line 4947  for (;;)
4947          {          {
4948          for (fi = min;; fi++)          for (fi = min;; fi++)
4949            {            {
4950            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4951            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4952            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4953            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4473  for (;;) Line 5086  for (;;)
5086                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5087                break;                break;
5088                }                }
5089              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5090              if (prop_fail_result) break;              if (prop_fail_result) break;
5091              eptr+= len;              eptr+= len;
5092              }              }
# Line 4482  for (;;) Line 5095  for (;;)
5095            case PT_LAMP:            case PT_LAMP:
5096            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5097              {              {
5098                int chartype;
5099              int len = 1;              int len = 1;
5100              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5101                {                {
5102                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5103                break;                break;
5104                }                }
5105              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5106              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
5107              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
5108                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
5109                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
5110                break;                break;
5111              eptr+= len;              eptr+= len;
5112              }              }
# Line 4507  for (;;) Line 5121  for (;;)
5121                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5122                break;                break;
5123                }                }
5124              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5125              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
5126              eptr+= len;              eptr+= len;
5127              }              }
5128            break;            break;
# Line 4524  for (;;) Line 5136  for (;;)
5136                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5137                break;                break;
5138                }                }
5139              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5140              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
5141              eptr+= len;              eptr+= len;
5142              }              }
5143            break;            break;
# Line 4541  for (;;) Line 5151  for (;;)
5151                SCHECK_PARTIAL();                SCHECK_PARTIAL();
5152                break;                break;
5153                }                }
5154              GETCHARLEN(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5155              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
5156              if ((prop_script == prop_value) == prop_fail_result)              eptr+= len;
5157                }
5158              break;
5159    
5160              case PT_ALNUM:
5161              for (i = min; i < max; i++)
5162                {
5163                int category;
5164                int len = 1;
5165                if (eptr >= md->end_subject)
5166                  {
5167                  SCHECK_PARTIAL();
5168                  break;
5169                  }
5170                GETCHARLENTEST(c, eptr, len);
5171                category = UCD_CATEGORY(c);
5172                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
5173                  break;
5174                eptr+= len;
5175                }
5176              break;
5177    
5178              case PT_SPACE:    /* Perl space */
5179              for (i = min; i < max; i++)
5180                {
5181                int len = 1;
5182                if (eptr >= md->end_subject)
5183                  {
5184                  SCHECK_PARTIAL();
5185                  break;
5186                  }
5187                GETCHARLENTEST(c, eptr, len);
5188                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
5189                     c == CHAR_FF || c == CHAR_CR)
5190                     == prop_fail_result)
5191                  break;
5192                eptr+= len;
5193                }
5194              break;
5195    
5196              case PT_PXSPACE:  /* POSIX space */
5197              for (i = min; i < max; i++)
5198                {
5199                int len = 1;
5200                if (eptr >= md->end_subject)
5201                  {
5202                  SCHECK_PARTIAL();
5203                  break;
5204                  }
5205                GETCHARLENTEST(c, eptr, len);
5206                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
5207                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5208                     == prop_fail_result)
5209                  break;
5210                eptr+= len;
5211                }
5212              break;
5213    
5214              case PT_WORD:
5215              for (i = min; i < max; i++)
5216                {
5217                int category;
5218                int len = 1;
5219                if (eptr >= md->end_subject)
5220                  {
5221                  SCHECK_PARTIAL();
5222                  break;
5223                  }
5224                GETCHARLENTEST(c, eptr, len);
5225                category = UCD_CATEGORY(c);
5226                if ((category == ucp_L || category == ucp_N ||
5227                     c == CHAR_UNDERSCORE) == prop_fail_result)
5228                break;                break;
5229              eptr+= len;              eptr+= len;
5230              }              }
5231            break;            break;
5232    
5233              default:
5234              RRETURN(PCRE_ERROR_INTERNAL);
5235            }            }
5236    
5237          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
# Line 4555  for (;;) Line 5239  for (;;)
5239          if (possessive) continue;          if (possessive) continue;
5240          for(;;)          for(;;)
5241            {            {
5242            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5243            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5244            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5245            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4569  for (;;) Line 5253  for (;;)
5253          {          {
5254          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5255            {            {
5256              int len = 1;
5257            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5258              {              {
5259              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5260              break;              break;
5261              }              }
5262            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5263            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5264            if (prop_category == ucp_M) break;            eptr += len;
5265            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5266              {              {
5267              int len = 1;              len = 1;
5268              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5269                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5270              eptr += len;              eptr += len;
5271              }              }
5272            }            }
# Line 4596  for (;;) Line 5277  for (;;)
5277    
5278          for(;;)          for(;;)
5279            {            {
5280            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5281            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5282            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5283            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5284              {              {
             int len = 1;  
5285              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5286                {                {
5287                BACKCHAR(eptr);                BACKCHAR(eptr);
5288                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5289                }                }
5290              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5291              eptr--;              eptr--;
5292              }              }
5293            }            }
# Line 4672  for (;;) Line 5351  for (;;)
5351                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
5352                }                }
5353              }              }
5354            else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */            else
5355                {
5356                eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
5357                SCHECK_PARTIAL();
5358                }
5359            break;            break;
5360    
5361            /* The byte case is the same as non-UTF8 */            /* The byte case is the same as non-UTF8 */
# Line 4880  for (;;) Line 5563  for (;;)
5563            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5564            }            }
5565    
5566          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5567            done (no backing up). Otherwise, match at this position; anything other
5568            than no match is immediately returned. For nomatch, back up one
5569            character, unless we are matching \R and the last thing matched was
5570            \r\n, in which case, back up two bytes. */
5571    
5572          if (possessive) continue;          if (possessive) continue;
5573          for(;;)          for(;;)
5574            {            {
5575            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5576            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5577            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5578            BACKCHAR(eptr);            BACKCHAR(eptr);
5579              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5580                  eptr[-1] == '\r') eptr--;
5581            }            }
5582          }          }
5583        else        else
# Line 5087  for (;;) Line 5776  for (;;)
5776            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5777            }            }
5778    
5779          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5780            done (no backing up). Otherwise, match at this position; anything other
5781            than no match is immediately returned. For nomatch, back up one
5782            character (byte), unless we are matching \R and the last thing matched
5783            was \r\n, in which case, back up two bytes. */
5784    
5785          if (possessive) continue;          if (possessive) continue;
5786          while (eptr >= pp)          while (eptr >= pp)
5787            {            {
5788            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5789            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5790              eptr--;
5791              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5792                  eptr[-1] == '\r') eptr--;
5793            }            }
5794          }          }
5795    
# Line 5133  switch (frame->Xwhere) Line 5828  switch (frame->Xwhere)
5828    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)
5829    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)
5830    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)
5831    LBL(53) LBL(54)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64)
5832      LBL(65) LBL(66)
5833  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5834    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)
5835    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
5836  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5837    LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)    LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)
5838      LBL(59) LBL(60) LBL(61) LBL(62)
5839  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
5840  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
5841    default:    default:
# Line 5161  Undefine all the macros that were define Line 5858  Undefine all the macros that were define
5858  #undef ecode  #undef ecode
5859  #undef mstart  #undef mstart
5860  #undef offset_top  #undef offset_top
 #undef ims  
5861  #undef eptrb  #undef eptrb
5862  #undef flags  #undef flags
5863    
# Line 5179  Undefine all the macros that were define Line 5875  Undefine all the macros that were define
5875  #undef condition  #undef condition
5876  #undef prev_is_word  #undef prev_is_word
5877    
 #undef original_ims  
   
5878  #undef ctype  #undef ctype
5879  #undef length  #undef length
5880  #undef max  #undef max
# Line 5237  pcre_exec(const pcre *argument_re, const Line 5931  pcre_exec(const pcre *argument_re, const
5931    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5932    int offsetcount)    int offsetcount)
5933  {  {
5934  int rc, resetcount, ocount;  int rc, ocount, arg_offset_max;
5935  int first_byte = -1;  int first_byte = -1;
5936  int req_byte = -1;  int req_byte = -1;
5937  int req_byte2 = -1;  int req_byte2 = -1;
5938  int newline;  int newline;
 unsigned long int ims;  
5939  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5940  BOOL anchored;  BOOL anchored;
5941  BOOL startline;  BOOL startline;
# Line 5272  if ((options & ~PUBLIC_EXEC_OPTIONS) != Line 5965  if ((options & ~PUBLIC_EXEC_OPTIONS) !=
5965  if (re == NULL || subject == NULL ||  if (re == NULL || subject == NULL ||
5966     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;
5967  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
5968    if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
5969    
5970    /* These two settings are used in the code for checking a UTF-8 string that
5971    follows immediately afterwards. Other values in the md block are used only
5972    during "normal" pcre_exec() processing, not when the JIT support is in use,
5973    so they are set up later. */
5974    
5975    utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;
5976    md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5977                  ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5978    
5979    /* Check a UTF-8 string if required. Pass back the character offset and error
5980    code for an invalid string if a results vector is available. */
5981    
5982    #ifdef SUPPORT_UTF8
5983    if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5984      {
5985      int erroroffset;
5986      int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5987      if (errorcode != 0)
5988        {
5989        if (offsetcount >= 2)
5990          {
5991          offsets[0] = erroroffset;
5992          offsets[1] = errorcode;
5993          }
5994        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5995          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5996        }
5997    
5998      /* Check that a start_offset points to the start of a UTF-8 character. */
5999      if (start_offset > 0 && start_offset < length &&
6000          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
6001        return PCRE_ERROR_BADUTF8_OFFSET;
6002      }
6003    #endif
6004    
6005    /* If the pattern was successfully studied with JIT support, run the JIT
6006    executable instead of the rest of this function. Most options must be set at
6007    compile time for the JIT code to be usable. Fallback to the normal code path if
6008    an unsupported flag is set. In particular, JIT does not support partial
6009    matching. */
6010    
6011    #ifdef SUPPORT_JIT
6012    if (extra_data != NULL
6013        && (extra_data->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0
6014        && extra_data->executable_jit != NULL
6015        && (extra_data->flags & PCRE_EXTRA_TABLES) == 0
6016        && (options & ~(PCRE_NO_UTF8_CHECK | PCRE_NOTBOL | PCRE_NOTEOL |
6017                        PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART)) == 0)
6018      return _pcre_jit_exec(re, extra_data->executable_jit, subject, length,
6019        start_offset, options, ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0)
6020        ? MATCH_LIMIT : extra_data->match_limit, offsets, offsetcount);
6021    #endif
6022    
6023  /* This information is for finding all the numbers associated with a given  /* Carry on with non-JIT matching. This information is for finding all the
6024  name, for condition testing. */  numbers associated with a given name, for condition testing. */
6025    
6026  md->name_table = (uschar *)re + re->name_table_offset;  md->name_table = (uschar *)re + re->name_table_offset;
6027  md->name_count = re->name_count;  md->name_count = re->name_count;
# Line 5341  md->end_subject = md->start_subject + le Line 6088  md->end_subject = md->start_subject + le
6088  end_subject = md->end_subject;  end_subject = md->end_subject;
6089    
6090  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
6091  utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
6092  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
6093    
6094    /* Some options are unpacked into BOOL variables in the hope that testing
6095    them will be faster than individual option bits. */
6096    
6097  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
6098  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
6099  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
6100  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
6101  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  
               ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;  
6102  md->hitend = FALSE;  md->hitend = FALSE;
6103  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
6104    
6105  md->recursive = NULL;                   /* No recursion at top level */  md->recursive = NULL;                   /* No recursion at top level */
6106    md->hasthen = (re->flags & PCRE_HASTHEN) != 0;
6107    
6108  md->lcc = tables + lcc_offset;  md->lcc = tables + lcc_offset;
6109  md->ctypes = tables + ctypes_offset;  md->ctypes = tables + ctypes_offset;
# Line 5431  defined (though never set). So there's n Line 6181  defined (though never set). So there's n
6181  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
6182    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
6183    
 /* Check a UTF-8 string if required. Unfortunately there's no way of passing  
 back the character offset. */  
   
 #ifdef SUPPORT_UTF8  
 if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  
   {  
   if (_pcre_valid_utf8((USPTR)subject, length) >= 0)  
     return PCRE_ERROR_BADUTF8;  
   if (start_offset > 0 && start_offset < length)  
     {  
     int tb = ((USPTR)subject)[start_offset];  
     if (tb > 127)  
       {  
       tb &= 0xc0;  
       if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET;  
       }  
     }  
   }  
 #endif  
   
 /* The ims options can vary during the matching as a result of the presence  
 of (?ims) items in the pattern. They are kept in a local variable so that  
 restoring at the exit of a group is easy. */  
   
 ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL);  
   
6184  /* 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
6185  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.
6186  Otherwise, we can use the vector supplied, rounding down its size to a multiple  Otherwise, we can use the vector supplied, rounding down its size to a multiple
6187  of 3. */  of 3. */
6188    
6189  ocount = offsetcount - (offsetcount % 3);  ocount = offsetcount - (offsetcount % 3);
6190    arg_offset_max = (2*ocount)/3;
6191    
6192  if (re->top_backref > 0 && re->top_backref >= ocount/3)  if (re->top_backref > 0 && re->top_backref >= ocount/3)
6193    {    {
# Line 5479  md->offset_max = (2*ocount)/3; Line 6204  md->offset_max = (2*ocount)/3;
6204  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
6205  md->capture_last = -1;  md->capture_last = -1;
6206    
 /* 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;  
   
6207  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
6208  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
6209  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
6210    offsets for the matched string. This is really just for tidiness with callouts,
6211    in case they inspect these fields. */
6212    
6213  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
6214    {    {
6215    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6216    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6217      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6218    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6219      md->offset_vector[0] = md->offset_vector[1] = -1;
6220    }    }
6221    
6222  /* 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 5528  if ((re->flags & PCRE_REQCHSET) != 0) Line 6250  if ((re->flags & PCRE_REQCHSET) != 0)
6250    }    }
6251    
6252    
6253    
6254    
6255  /* ==========================================================================*/  /* ==========================================================================*/
6256    
6257  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 5538  for(;;) Line 6262  for(;;)
6262    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6263    USPTR new_start_match;    USPTR new_start_match;
6264    
   /* 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;  
     }  
   
6265    /* 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
6266    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
6267    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 5574  for(;;) Line 6289  for(;;)
6289    /* There are some optimizations that avoid running the match if a known    /* There are some optimizations that avoid running the match if a known
6290    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.
6291    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
6292    that all callouts do actually occur. */    that all callouts do actually occur. The option can be set in the regex by
6293      (*NO_START_OPT) or passed in match-time options. */
6294    
6295    if ((options & PCRE_NO_START_OPTIMIZE) == 0)    if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0)
6296      {      {
6297      /* Advance to a unique first byte if there is one. */      /* Advance to a unique first byte if there is one. */
6298    
# Line 5630  for(;;) Line 6346  for(;;)
6346        while (start_match < end_subject)        while (start_match < end_subject)
6347          {          {
6348          register unsigned int c = *start_match;          register unsigned int c = *start_match;
6349          if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++;          if ((start_bits[c/8] & (1 << (c&7))) == 0)
6350            else break;            {
6351              start_match++;
6352    #ifdef SUPPORT_UTF8
6353              if (utf8)
6354                while(start_match < end_subject && (*start_match & 0xc0) == 0x80)
6355                  start_match++;
6356    #endif
6357              }
6358            else break;
6359          }          }
6360        }        }
6361      }   /* Starting optimizations */      }   /* Starting optimizations */
# Line 5643  for(;;) Line 6367  for(;;)
6367    /* The following two optimizations are disabled for partial matching or if    /* The following two optimizations are disabled for partial matching or if
6368    disabling is explicitly requested. */    disabling is explicitly requested. */
6369    
6370    if ((options & PCRE_NO_START_OPTIMIZE) == 0 && !md->partial)    if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0 && !md->partial)
6371      {      {
6372      /* If the pattern was studied, a minimum subject length may be set. This is      /* If the pattern was studied, a minimum subject length may be set. This is
6373      a lower bound; no actual string of that length may actually match the      a lower bound; no actual string of that length may actually match the
# Line 5722  for(;;) Line 6446  for(;;)
6446    
6447    /* OK, we can now run the match. If "hitend" is set afterwards, remember the    /* OK, we can now run the match. If "hitend" is set afterwards, remember the
6448    first starting point for which a partial match was found. */    first starting point for which a partial match was found. */
6449