/[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 597 by ph10, Mon May 2 17:08:52 2011 UTC revision 699 by ph10, Tue Sep 20 10:46:54 2011 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2010 University of Cambridge             Copyright (c) 1997-2011 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 57  possible. There are also some static sup Line 57  possible. There are also some static sup
57  #undef min  #undef min
58  #undef max  #undef max
59    
60  /* Flag bits for the match() function */  /* Values for setting in md->match_function_type to indicate two special types
61    of call to match(). We do it this way to save on using another stack variable,
62    as stack usage is to be discouraged. */
63    
64  #define match_condassert     0x01  /* Called to check a condition assertion */  #define MATCH_CONDASSERT     1  /* Called to check a condition assertion */
65  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */  #define MATCH_CBEGROUP       2  /* Could-be-empty unlimited repeat group */
66    
67  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
68  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 73  negative to avoid the external error cod Line 75  negative to avoid the external error cod
75    
76  #define MATCH_ACCEPT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_COMMIT       (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_PRUNE        (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_SKIP         (-996)  #define MATCH_ONCE         (-996)
80  #define MATCH_SKIP_ARG     (-995)  #define MATCH_PRUNE        (-995)
81  #define MATCH_THEN         (-994)  #define MATCH_SKIP         (-994)
82    #define MATCH_SKIP_ARG     (-993)
83    #define MATCH_THEN         (-992)
84    
85  /* This is a convenience macro for code that occurs many times. */  /* This is a convenience macro for code that occurs many times. */
86    
# Line 134  while (length-- > 0) Line 138  while (length-- > 0)
138    
139  /* Normally, if a back reference hasn't been set, the length that is passed is  /* Normally, if a back reference hasn't been set, the length that is passed is
140  negative, so the match always fails. However, in JavaScript compatibility mode,  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  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.  subject bytes matched may be different to the number of reference bytes.
143    
144  Arguments:  Arguments:
# Line 142  Arguments: Line 146  Arguments:
146    eptr        pointer into the subject    eptr        pointer into the subject
147    length      length of reference to be matched (number of bytes)    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:      < 0 if not matched, otherwise the number of subject bytes matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
152  */  */
153    
154  static int  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 eptr_start = eptr;  USPTR eptr_start = eptr;
159  register USPTR p = md->start_subject + md->offset_vector[offset];  register USPTR p = md->start_subject + md->offset_vector[offset];
# Line 175  if (length < 0) return -1; Line 179  if (length < 0) return -1;
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      /* Match characters up to the end of the reference. NOTE: the number of      /* Match characters up to the end of the reference. NOTE: the number of
189      bytes matched may differ, because there are some characters whose upper and      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      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);      (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      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      the latter. It is important, therefore, to check the length along the
194      reference, not along the subject (earlier code did this wrong). */      reference, not along the subject (earlier code did this wrong). */
195    
196      USPTR endptr = p + length;      USPTR endptr = p + length;
197      while (p < endptr)      while (p < endptr)
198        {        {
# Line 206  if ((ims & PCRE_CASELESS) != 0) Line 210  if ((ims & PCRE_CASELESS) != 0)
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      if (eptr + length > md->end_subject) return -1;      if (eptr + length > md->end_subject) return -1;
214      while (length-- > 0)      while (length-- > 0)
215        { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }        { 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    {    {
224    if (eptr + length > md->end_subject) return -1;    if (eptr + length > md->end_subject) return -1;
225    while (length-- > 0) if (*p++ != *eptr++) return -1;    while (length-- > 0) if (*p++ != *eptr++) return -1;
226    }    }
227    
228  return eptr - eptr_start;  return eptr - eptr_start;
# Line 273  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 277  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
277         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
278         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
279         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280         RM61,  RM62 };         RM61,  RM62, RM63 };
281    
282  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
283  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
# Line 283  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 295  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 309  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 = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
319    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
# Line 319  argument of match(), which never changes Line 323  argument of match(), which never changes
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 357  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 380  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 467  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 485  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 498  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 517  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 533  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 557  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 595  i, and fc and c, can be the same variabl Line 577  i, and fc and c, can be the same variabl
577  #define fi i  #define fi i
578  #define fc c  #define fc c
579    
580    /* Many of the following variables are used only in small blocks of the code.
581    My normal style of coding would have declared them within each of those blocks.
582    However, in order to accommodate the version of this code that uses an external
583    "stack" implemented on the heap, it is easier to declare them all here, so the
584    declarations can be cut out in a block. The only declarations within blocks
585    below are for variables that do not have to be preserved over a recursive call
586    to RMATCH(). */
587    
588  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  #ifdef SUPPORT_UTF8
589  const uschar *charptr;             /* in small blocks of the code. My normal */  const uschar *charptr;
590  #endif                             /* style of coding would have declared    */  #endif
591  const uschar *callpat;             /* them within each of those blocks.      */  const uschar *callpat;
592  const uschar *data;                /* However, in order to accommodate the   */  const uschar *data;
593  const uschar *next;                /* version of this code that uses an      */  const uschar *next;
594  USPTR         pp;                  /* external "stack" implemented on the    */  USPTR         pp;
595  const uschar *prev;                /* heap, it is easier to declare them all */  const uschar *prev;
596  USPTR         saved_eptr;          /* here, so the declarations can be cut   */  USPTR         saved_eptr;
597                                     /* out in a block. The only declarations  */  
598  recursion_info new_recursive;      /* within blocks below are for variables  */  recursion_info new_recursive;
599                                     /* that do not have to be preserved over  */  
600  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  BOOL cur_is_word;
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
 unsigned long int original_ims;  
   
604  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
608  int oclength;  int oclength;
609  uschar occhars[8];  uschar occhars[8];
610  #endif  #endif
# Line 640  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 676  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 706  for (;;) Line 703  for (;;)
703      case OP_MARK:      case OP_MARK:
704      markptr = ecode + 2;      markptr = ecode + 2;
705      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
706        ims, eptrb, flags, RM55);        eptrb, RM55);
707    
708      /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an      /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
709      argument, and we must check whether that argument matches this MARK's      argument, and we must check whether that argument matches this MARK's
# Line 732  for (;;) Line 729  for (;;)
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 && rrc != MATCH_PRUNE &&      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
734          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
735          rrc != MATCH_THEN)          rrc != MATCH_THEN)
# Line 743  for (;;) Line 740  for (;;)
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 && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
745      MRRETURN(MATCH_PRUNE);      MRRETURN(MATCH_PRUNE);
746    
747      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
748      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
749        ims, eptrb, flags, RM56);        eptrb, RM56);
750      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) 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);
# Line 758  for (;;) Line 755  for (;;)
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 && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
760        RRETURN(rrc);        RRETURN(rrc);
761      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
# Line 766  for (;;) Line 763  for (;;)
763    
764      case OP_SKIP_ARG:      case OP_SKIP_ARG:
765      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
766        ims, eptrb, flags, RM57);        eptrb, RM57);
767      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768        RRETURN(rrc);        RRETURN(rrc);
769    
# Line 785  for (;;) Line 782  for (;;)
782    
783      case OP_THEN:      case OP_THEN:
784      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
785        ims, eptrb, flags, RM54);        eptrb, RM54);
786      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
787      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode - GET(ecode, 1);
788      MRRETURN(MATCH_THEN);      MRRETURN(MATCH_THEN);
789    
790      case OP_THEN_ARG:      case OP_THEN_ARG:
791      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
792        offset_top, md, ims, eptrb, flags, RM58);        offset_top, md, eptrb, RM58);
793      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
794      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode - GET(ecode, 1);
795      md->mark = ecode + LINK_SIZE + 2;      md->mark = ecode + LINK_SIZE + 2;
796      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
797    
798      /* Handle a capturing bracket. If there is space in the offset vector, save      /* Handle a capturing bracket, other than those that are possessive with an
799      the current subject position in the working slot at the top of the vector.      unlimited repeat. If there is space in the offset vector, save the current
800      We mustn't change the current values of the data slot, because they may be      subject position in the working slot at the top of the vector. We mustn't
801      set from a previous iteration of this group, and be referred to by a      change the current values of the data slot, because they may be set from a
802      reference inside the group.      previous iteration of this group, and be referred to by a reference inside
803        the group. A failure to match might occur after the group has succeeded,
804      If the bracket fails to match, we need to restore this value and also the      if something later on doesn't match. For this reason, we need to restore
805      values of the final offsets, in case they were set by a previous iteration      the working value and also the values of the final offsets, in case they
806      of the same bracket.      were set by a previous iteration of the same bracket.
807    
808      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
809      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 835  for (;;) Line 832  for (;;)
832        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
833          (int)(eptr - md->start_subject);          (int)(eptr - md->start_subject);
834    
835        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
836          {          {
837            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839            ims, eptrb, flags, RM1);            eptrb, RM1);
840            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841          if (rrc != MATCH_NOMATCH &&          if (rrc != MATCH_NOMATCH &&
842              (rrc != MATCH_THEN || md->start_match_ptr != ecode))              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
843            RRETURN(rrc);            RRETURN(rrc);
844          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
845          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
846            if (*ecode != OP_ALT) break;
847          }          }
       while (*ecode == OP_ALT);  
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854        if (rrc != MATCH_THEN) md->mark = markptr;        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855        RRETURN(MATCH_NOMATCH);        MATCH_THEN. */
856    
857          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
858          RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
859        }        }
860    
861      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 869  for (;;) Line 869  for (;;)
869      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
871    
872      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
873      final alternative within the brackets, we would return the result of a      repeat. Loop for all the alternatives. When we get to the final alternative
874      recursive call to match() whatever happened. We can reduce stack usage by      within the brackets, we used to return the result of a recursive call to
875      turning this into a tail recursion, except in the case when match_cbegroup      match() whatever happened so it was possible to reduce stack usage by
876      is set.*/      turning this into a tail recursion, except in the case of a possibly empty
877        group. However, now that there is the possiblity of (*THEN) occurring in
878        the final alternative, this optimization is no longer possible.
879    
880        MATCH_ONCE is returned when the end of an atomic group is successfully
881        reached, but subsequent matching fails. It passes back up the tree (causing
882        captured values to be reset) until the original atomic group level is
883        reached. This is tested by comparing md->once_target with the start of the
884        group. At this point, the return is converted into MATCH_NOMATCH so that
885        previous backup points can be taken. */
886    
887        case OP_ONCE:
888      case OP_BRA:      case OP_BRA:
889      case OP_SBRA:      case OP_SBRA:
890      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
891      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
892      for (;;)      for (;;)
893        {        {
894        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
895          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
896            RM2);
897          if (rrc != MATCH_NOMATCH &&
898              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
899          {          {
900          if (flags == 0)    /* Not a possibly empty group */          if (rrc == MATCH_ONCE)
901            {            {
902            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
903            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
904            goto TAIL_RECURSE;              {
905                while (*scode == OP_ALT) scode += GET(scode, 1);
906                scode -= GET(scode, 1);
907                }
908              if (md->once_target == scode) rrc = MATCH_NOMATCH;
909            }            }
910            RRETURN(rrc);
911            }
912          ecode += GET(ecode, 1);
913          if (*ecode != OP_ALT) break;
914          }
915        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
916        RRETURN(MATCH_NOMATCH);
917    
918          /* Possibly empty group; can't use tail recursion. */      /* Handle possessive capturing brackets with an unlimited repeat. We come
919        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
920        handled similarly to the normal case above. However, the matching is
921        different. The end of these brackets will always be OP_KETRPOS, which
922        returns MATCH_KETRPOS without going further in the pattern. By this means
923        we can handle the group by iteration rather than recursion, thereby
924        reducing the amount of stack needed. */
925    
926        case OP_CBRAPOS:
927        case OP_SCBRAPOS:
928        allow_zero = FALSE;
929    
930          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      POSSESSIVE_CAPTURE:
931            eptrb, flags, RM48);      number = GET2(ecode, 1+LINK_SIZE);
932          if (rrc == MATCH_NOMATCH) md->mark = markptr;      offset = number << 1;
933          RRETURN(rrc);  
934    #ifdef PCRE_DEBUG
935        printf("start possessive bracket %d\n", number);
936        printf("subject=");
937        pchars(eptr, 16, TRUE, md);
938        printf("\n");
939    #endif
940    
941        if (offset < md->offset_max)
942          {
943          matched_once = FALSE;
944          code_offset = ecode - md->start_code;
945    
946          save_offset1 = md->offset_vector[offset];
947          save_offset2 = md->offset_vector[offset+1];
948          save_offset3 = md->offset_vector[md->offset_end - number];
949          save_capture_last = md->capture_last;
950    
951          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
952    
953          /* Each time round the loop, save the current subject position for use
954          when the group matches. For MATCH_MATCH, the group has matched, so we
955          restart it with a new subject starting position, remembering that we had
956          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
957          usual. If we haven't matched any alternatives in any iteration, check to
958          see if a previous iteration matched. If so, the group has matched;
959          continue from afterwards. Otherwise it has failed; restore the previous
960          capture values before returning NOMATCH. */
961    
962          for (;;)
963            {
964            md->offset_vector[md->offset_end - number] =
965              (int)(eptr - md->start_subject);
966            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
967            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
968              eptrb, RM63);
969            if (rrc == MATCH_KETRPOS)
970              {
971              offset_top = md->end_offset_top;
972              eptr = md->end_match_ptr;
973              ecode = md->start_code + code_offset;
974              save_capture_last = md->capture_last;
975              matched_once = TRUE;
976              continue;
977              }
978            if (rrc != MATCH_NOMATCH &&
979                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
980              RRETURN(rrc);
981            md->capture_last = save_capture_last;
982            ecode += GET(ecode, 1);
983            if (*ecode != OP_ALT) break;
984            }
985    
986          if (!matched_once)
987            {
988            md->offset_vector[offset] = save_offset1;
989            md->offset_vector[offset+1] = save_offset2;
990            md->offset_vector[md->offset_end - number] = save_offset3;
991          }          }
992    
993        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
994        otherwise return. */        if (allow_zero || matched_once)
995            {
996            ecode += 1 + LINK_SIZE;
997            break;
998            }
999    
1000        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        RRETURN(MATCH_NOMATCH);
1001          eptrb, flags, RM2);        }
1002    
1003        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1004        as a non-capturing bracket. */
1005    
1006        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1007        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008    
1009        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1010    
1011        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1012        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1013    
1014        /* Non-capturing possessive bracket with unlimited repeat. We come here
1015        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1016        without the capturing complication. It is written out separately for speed
1017        and cleanliness. */
1018    
1019        case OP_BRAPOS:
1020        case OP_SBRAPOS:
1021        allow_zero = FALSE;
1022    
1023        POSSESSIVE_NON_CAPTURE:
1024        matched_once = FALSE;
1025        code_offset = ecode - md->start_code;
1026    
1027        for (;;)
1028          {
1029          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1030          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1031            eptrb, RM48);
1032          if (rrc == MATCH_KETRPOS)
1033            {
1034            offset_top = md->end_offset_top;
1035            eptr = md->end_match_ptr;
1036            ecode = md->start_code + code_offset;
1037            matched_once = TRUE;
1038            continue;
1039            }
1040        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
1041            (rrc != MATCH_THEN || md->start_match_ptr != ecode))            (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1042          RRETURN(rrc);          RRETURN(rrc);
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045        }        }
1046    
1047        if (matched_once || allow_zero)
1048          {
1049          ecode += 1 + LINK_SIZE;
1050          break;
1051          }
1052        RRETURN(MATCH_NOMATCH);
1053    
1054      /* Control never reaches here. */      /* Control never reaches here. */
1055    
1056      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1057      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1058      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1059      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1060    
1061      case OP_COND:      case OP_COND:
1062      case OP_SCOND:      case OP_SCOND:
1063      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1064    
1065      /* Because of the way auto-callout works during compile, a callout item is      /* Because of the way auto-callout works during compile, a callout item is
1066      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 928  for (;;) Line 1070  for (;;)
1070        if (pcre_callout != NULL)        if (pcre_callout != NULL)
1071          {          {
1072          pcre_callout_block cb;          pcre_callout_block cb;
1073          cb.version          = 1;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1074          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[LINK_SIZE+2];
1075          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1076          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 940  for (;;) Line 1082  for (;;)
1082          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1083          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last;
1084          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1085            cb.mark             = markptr;
1086          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1087          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1088          }          }
# Line 1093  for (;;) Line 1236  for (;;)
1236        }        }
1237    
1238      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1239      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
1240      assertion. */      an assertion. */
1241    
1242      else      else
1243        {        {
1244        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1245            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1246        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1247          {          {
1248            if (md->end_offset_top > offset_top)
1249              offset_top = md->end_offset_top;  /* Captures may have happened */
1250          condition = TRUE;          condition = TRUE;
1251          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1252          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
# Line 1119  for (;;) Line 1264  for (;;)
1264        }        }
1265    
1266      /* 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,
1267      we can use tail recursion to avoid using another stack frame, except when      we used to use tail recursion to avoid using another stack frame, except
1268      match_cbegroup is required for an unlimited repeat of a possibly empty      when there was unlimited repeat of a possibly empty group. However, that
1269      group. If the second alternative doesn't exist, we can just plough on. */      strategy no longer works because of the possibilty of (*THEN) being
1270        encountered in the branch. A recursive call to match() is always required,
1271        unless the second alternative doesn't exist, in which case we can just
1272        plough on. */
1273    
1274      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1275        {        {
1276        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1277        if (op == OP_SCOND)        /* Possibly empty group */        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1278          {  
1279          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);        /* If the result is THEN from within the "true" branch of the condition,
1280          RRETURN(rrc);        md->start_match_ptr will point to the original OP_COND, not to the start
1281          }        of the branch, so we have do work to see if it matches. If THEN comes
1282        else                       /* Group must match something */        from the "false" branch, md->start_match_ptr does point to OP_ALT. */
1283          {  
1284          flags = 0;        if (rrc == MATCH_THEN)
1285          goto TAIL_RECURSE;          {
1286          }          if (*ecode != OP_ALT)
1287              {
1288              do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
1289              ecode -= GET(ecode, 1);
1290              }
1291            if (md->start_match_ptr == ecode) rrc = MATCH_NOMATCH;
1292            }
1293          RRETURN(rrc);
1294        }        }
1295      else                         /* Condition false & no alternative */      else                         /* Condition false & no alternative */
1296        {        {
# Line 1168  for (;;) Line 1323  for (;;)
1323      break;      break;
1324    
1325    
1326      /* 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. */  
1327    
     case OP_ACCEPT:  
1328      case OP_END:      case OP_END:
1329      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1330        {      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;  
       }  
1331    
1332      /* 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
1333      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
1334      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,
1335      if any. */      backtracking will then try other alternatives, if any. */
1336    
1337      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1338          (md->notempty ||           md->recursive == NULL &&
1339            (md->notempty_atstart &&           (md->notempty ||
1340              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1341                 mstart == md->start_subject + md->start_offset)))
1342        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1343    
1344      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1210  for (;;) Line 1353  for (;;)
1353      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1354      MRRETURN(rrc);      MRRETURN(rrc);
1355    
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
   
1356      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1357      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,
1358      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
1359      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
1360      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1361        of a condition, we want to return immediately afterwards. The caller of
1362        this incarnation of the match() function will have set MATCH_CONDASSERT in
1363        md->match_function type, and one of these opcodes will be the first opcode
1364        that is processed. We use a local variable that is preserved over calls to
1365        match() to remember this case. */
1366    
1367      case OP_ASSERT:      case OP_ASSERT:
1368      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1369        if (md->match_function_type == MATCH_CONDASSERT)
1370          {
1371          condassert = TRUE;
1372          md->match_function_type = 0;
1373          }
1374        else condassert = FALSE;
1375    
1376      do      do
1377        {        {
1378        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1379        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1380          {          {
1381          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1382            markptr = md->mark;
1383          break;          break;
1384          }          }
1385        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
# Line 1241  for (;;) Line 1388  for (;;)
1388        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1389        }        }
1390      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1391    
1392      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1393    
1394      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1395    
1396      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1397    
1398      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1399      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1261  for (;;) Line 1409  for (;;)
1409    
1410      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1411      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1412        if (md->match_function_type == MATCH_CONDASSERT)
1413          {
1414          condassert = TRUE;
1415          md->match_function_type = 0;
1416          }
1417        else condassert = FALSE;
1418    
1419      do      do
1420        {        {
1421        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
         RM5);  
1422        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1423        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1424          {          {
# Line 1278  for (;;) Line 1432  for (;;)
1432        }        }
1433      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1434    
1435      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1436    
1437      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1438      continue;      continue;
# Line 1324  for (;;) Line 1478  for (;;)
1478      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1479        {        {
1480        pcre_callout_block cb;        pcre_callout_block cb;
1481        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1482        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1483        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1484        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 1336  for (;;) Line 1490  for (;;)
1490        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1491        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1492        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1493          cb.mark             = markptr;
1494        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1495        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1496        }        }
# Line 1346  for (;;) Line 1501  for (;;)
1501      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
1502      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1503    
1504      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1505      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
1506      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
1507      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
1508      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
1509      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
1510      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.  
1511    
1512      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
1513      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
1514      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1515        a lot, so he is not to blame for the current way it works. */
1516    
1517      case OP_RECURSE:      case OP_RECURSE:
1518        {        {
1519          recursion_info *ri;
1520          int recno;
1521    
1522        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1523        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1524          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1525    
1526          /* Check for repeating a recursion without advancing the subject pointer.
1527          This should catch convoluted mutual recursions. (Some simple cases are
1528          caught at compile time.) */
1529    
1530          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1531            if (recno == ri->group_num && eptr == ri->subject_position)
1532              RRETURN(PCRE_ERROR_RECURSELOOP);
1533    
1534        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1535    
1536          new_recursive.group_num = recno;
1537          new_recursive.subject_position = eptr;
1538        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1539        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1540    
1541        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1542    
1543        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1544    
1545        /* Now save the offset data. */        /* Now save the offset data */
1546    
1547        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1548        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1388  for (;;) Line 1553  for (;;)
1553            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1554          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1555          }          }
   
1556        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1557              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1558    
1559        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1560        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1561          might be changed, so reset it before looping. */
1562    
1563        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1564        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1565        do        do
1566          {          {
1567            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1568          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1569            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1570            memcpy(md->offset_vector, new_recursive.offset_save,
1571                new_recursive.saved_max * sizeof(int));
1572            md->recursive = new_recursive.prevrec;
1573          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1574            {            {
1575            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1576            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1577              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1578            MRRETURN(MATCH_MATCH);  
1579              /* Set where we got to in the subject, and reset the start in case
1580              it was changed by \K. This *is* propagated back out of a recursion,
1581              for Perl compatibility. */
1582    
1583              eptr = md->end_match_ptr;
1584              mstart = md->start_match_ptr;
1585              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1586            }            }
1587          else if (rrc != MATCH_NOMATCH &&          else if (rrc != MATCH_NOMATCH &&
1588                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
# Line 1420  for (;;) Line 1594  for (;;)
1594            }            }
1595    
1596          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1597          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1598          }          }
1599        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1432  for (;;) Line 1604  for (;;)
1604          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1605        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1606        }        }
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer, but resetting  
     the start-of-match value in case it was changed by \K. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */  
         {  
         mstart = md->start_match_ptr;  
         break;  
         }  
       if (rrc != MATCH_NOMATCH &&  
           (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
         RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
1607    
1608      offset_top = md->end_offset_top;      RECURSION_MATCHED:
1609      eptr = md->end_match_ptr;      break;
   
     /* 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 */  
1610    
1611      /* 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
1612      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1529  for (;;) Line 1622  for (;;)
1622      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1623    
1624      case OP_BRAZERO:      case OP_BRAZERO:
1625        {      next = ecode + 1;
1626        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1627        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1628        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1629        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1630      break;      break;
1631    
1632      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1633        {      next = ecode + 1;
1634        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1635        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1636        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1637        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1638      break;      break;
1639    
1640      case OP_SKIPZERO:      case OP_SKIPZERO:
1641        {      next = ecode+1;
1642        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1643        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1644      break;      break;
1645    
1646        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1647        here; just jump to the group, with allow_zero set TRUE. */
1648    
1649        case OP_BRAPOSZERO:
1650        op = *(++ecode);
1651        allow_zero = TRUE;
1652        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1653          goto POSSESSIVE_NON_CAPTURE;
1654    
1655      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1656    
1657      case OP_KET:      case OP_KET:
1658      case OP_KETRMIN:      case OP_KETRMIN:
1659      case OP_KETRMAX:      case OP_KETRMAX:
1660        case OP_KETRPOS:
1661      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1662    
1663      /* 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
1664      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1665      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1666    
1667      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1668        {        {
1669        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1670        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1671        }        }
1672      else saved_eptr = NULL;      else saved_eptr = NULL;
1673    
1674      /* If we are at the end of an assertion group or an atomic group, stop      /* If we are at the end of an assertion group, stop matching and return
1675      matching and return MATCH_MATCH, but record the current high water mark for      MATCH_MATCH, but record the current high water mark for use by positive
1676      use by positive assertions. We also need to record the match start in case      assertions. We also need to record the match start in case it was changed
1677      it was changed by \K. */      by \K. */
1678    
1679      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1680          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1681        {        {
1682        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1683        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1684        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1685        MRRETURN(MATCH_MATCH);        MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1686        }        }
1687    
1688      /* 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
1689      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1690      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
1691      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
1692      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
1693        the current subject position and start match pointer and give a MATCH
1694        return. */
1695    
1696      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1697            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1698        {        {
1699        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1700        offset = number << 1;        offset = number << 1;
# Line 1605  for (;;) Line 1704  for (;;)
1704        printf("\n");        printf("\n");
1705  #endif  #endif
1706    
1707          /* Handle a recursively called group. */
1708    
1709          if (md->recursive != NULL && md->recursive->group_num == number)
1710            {
1711            md->end_match_ptr = eptr;
1712            md->start_match_ptr = mstart;
1713            RRETURN(MATCH_MATCH);
1714            }
1715    
1716          /* Deal with capturing */
1717    
1718        md->capture_last = number;        md->capture_last = number;
1719        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1720          {          {
1721            /* If offset is greater than offset_top, it means that we are
1722            "skipping" a capturing group, and that group's offsets must be marked
1723            unset. In earlier versions of PCRE, all the offsets were unset at the
1724            start of matching, but this doesn't work because atomic groups and
1725            assertions can cause a value to be set that should later be unset.
1726            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1727            part of the atomic group, but this is not on the final matching path,
1728            so must be unset when 2 is set. (If there is no group 2, there is no
1729            problem, because offset_top will then be 2, indicating no capture.) */
1730    
1731            if (offset > offset_top)
1732              {
1733              register int *iptr = md->offset_vector + offset_top;
1734              register int *iend = md->offset_vector + offset;
1735              while (iptr < iend) *iptr++ = -1;
1736              }
1737    
1738            /* Now make the extraction */
1739    
1740          md->offset_vector[offset] =          md->offset_vector[offset] =
1741            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1742          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1743          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1744          }          }
1745          }
1746    
1747        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1748        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1749        group. This is the forcible breaking of infinite loops as implemented in
1750        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1751        processing the rest of the pattern at a lower level. If this results in a
1752        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1753        bypassing intermediate backup points, but resetting any captures that
1754        happened along the way. */
1755    
1756        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1757          {
1758          if (*prev == OP_ONCE)
1759          {          {
1760          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1761          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1762          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1763          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;  
1764          }          }
1765          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1766          break;
1767        }        }
1768    
1769      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1770      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1771        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
1772    
1773      /* For a non-repeating ket, just continue at this level. This also      if (*ecode == OP_KETRPOS)
     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)  
1774        {        {
1775        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1776        break;        md->end_offset_top = offset_top;
1777          RRETURN(MATCH_KETRPOS);
1778        }        }
1779    
1780      /* 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
1781      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
1782      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
1783      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
1784        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1785    
1786      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1787        {        {
1788        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1789        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1790        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1791          {          {
1792          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1793            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1794            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1795            RRETURN(MATCH_ONCE);
1796            }
1797          if (*prev >= OP_SBRA)    /* Could match an empty string */
1798            {
1799            md->match_function_type = MATCH_CBEGROUP;
1800            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1801          RRETURN(rrc);          RRETURN(rrc);
1802          }          }
1803        ecode = prev;        ecode = prev;
# Line 1670  for (;;) Line 1805  for (;;)
1805        }        }
1806      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1807        {        {
1808        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1809          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1810          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1811        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1812          if (*prev == OP_ONCE)
1813            {
1814            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1815            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1816            md->once_target = prev;
1817            RRETURN(MATCH_ONCE);
1818            }
1819        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1820        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1821        }        }
1822      /* Control never gets here */      /* Control never gets here */
1823    
1824      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1825    
1826      case OP_CIRC:      case OP_CIRC:
1827      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 */  
1828    
1829      /* Start of subject assertion */      /* Start of subject assertion */
1830    
# Line 1699  for (;;) Line 1833  for (;;)
1833      ecode++;      ecode++;
1834      break;      break;
1835    
1836        /* Multiline mode: start of subject unless notbol, or after any newline. */
1837    
1838        case OP_CIRCM:
1839        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1840        if (eptr != md->start_subject &&
1841            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1842          MRRETURN(MATCH_NOMATCH);
1843        ecode++;
1844        break;
1845    
1846      /* Start of match assertion */      /* Start of match assertion */
1847    
1848      case OP_SOM:      case OP_SOM:
# Line 1713  for (;;) Line 1857  for (;;)
1857      ecode++;      ecode++;
1858      break;      break;
1859    
1860      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1861      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1862    
1863      case OP_DOLL:      case OP_DOLLM:
1864      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1865        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1866        if (eptr < md->end_subject)      else
         { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }  
       else  
         {  
         if (md->noteol) MRRETURN(MATCH_NOMATCH);  
         SCHECK_PARTIAL();  
         }  
       ecode++;  
       break;  
       }  
     else  /* Not multiline */  
1867        {        {
1868        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1869        if (!md->endonly) goto ASSERT_NL_OR_EOS;        SCHECK_PARTIAL();
1870        }        }
1871        ecode++;
1872        break;
1873    
1874        /* Not multiline mode: assert before a terminating newline or before end of
1875        subject unless noteol is set. */
1876    
1877        case OP_DOLL:
1878        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1879        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1880    
1881      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1882    
# Line 1884  for (;;) Line 2027  for (;;)
2027      /* Fall through */      /* Fall through */
2028    
2029      case OP_ALLANY:      case OP_ALLANY:
2030      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2031        {        {                            /* not be updated before SCHECK_PARTIAL. */
2032        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2033        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2034        }        }
2035        eptr++;
2036      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2037      ecode++;      ecode++;
2038      break;      break;
# Line 1897  for (;;) Line 2041  for (;;)
2041      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2042    
2043      case OP_ANYBYTE:      case OP_ANYBYTE:
2044      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2045        {        {                            /* not be updated before SCHECK_PARTIAL. */
2046        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2047        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2048        }        }
2049        eptr++;
2050      ecode++;      ecode++;
2051      break;      break;
2052    
# Line 2017  for (;;) Line 2162  for (;;)
2162      switch(c)      switch(c)
2163        {        {
2164        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2165    
2166        case 0x000d:        case 0x000d:
2167        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2168        break;        break;
# Line 2240  for (;;) Line 2386  for (;;)
2386        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2387        }        }
2388      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2389        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2390        while (eptr < md->end_subject)
2391        {        {
2392        int category = UCD_CATEGORY(c);        int len = 1;
2393        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2394        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2395          {        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;  
         }  
2396        }        }
2397      ecode++;      ecode++;
2398      break;      break;
# Line 2269  for (;;) Line 2408  for (;;)
2408      loops). */      loops). */
2409    
2410      case OP_REF:      case OP_REF:
2411        case OP_REFI:
2412        caseless = op == OP_REFI;
2413      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2414      ecode += 3;      ecode += 3;
2415    
# Line 2316  for (;;) Line 2457  for (;;)
2457        break;        break;
2458    
2459        default:               /* No repeat follows */        default:               /* No repeat follows */
2460        if ((length = match_ref(offset, eptr, length, md, ims)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2461          {          {
2462          CHECK_PARTIAL();          CHECK_PARTIAL();
2463          MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
# Line 2336  for (;;) Line 2477  for (;;)
2477    
2478      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2479        {        {
2480        int slength;        int slength;
2481        if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2482          {          {
2483          CHECK_PARTIAL();          CHECK_PARTIAL();
2484          MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
# Line 2356  for (;;) Line 2497  for (;;)
2497        {        {
2498        for (fi = min;; fi++)        for (fi = min;; fi++)
2499          {          {
2500          int slength;          int slength;
2501          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2502          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2503          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) MRRETURN(MATCH_NOMATCH);
2504          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2505            {            {
2506            CHECK_PARTIAL();            CHECK_PARTIAL();
2507            MRRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
# Line 2377  for (;;) Line 2518  for (;;)
2518        pp = eptr;        pp = eptr;
2519        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2520          {          {
2521          int slength;          int slength;
2522          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2523            {            {
2524            CHECK_PARTIAL();            CHECK_PARTIAL();
2525            break;            break;
# Line 2387  for (;;) Line 2528  for (;;)
2528          }          }
2529        while (eptr >= pp)        while (eptr >= pp)
2530          {          {
2531          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2532          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2533          eptr -= length;          eptr -= length;
2534          }          }
# Line 2497  for (;;) Line 2638  for (;;)
2638            {            {
2639            for (fi = min;; fi++)            for (fi = min;; fi++)
2640              {              {
2641              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2642              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2643              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2644              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2522  for (;;) Line 2663  for (;;)
2663            {            {
2664            for (fi = min;; fi++)            for (fi = min;; fi++)
2665              {              {
2666              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2667              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2668              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2669              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2568  for (;;) Line 2709  for (;;)
2709              }              }
2710            for (;;)            for (;;)
2711              {              {
2712              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2713              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2714              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2715              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2591  for (;;) Line 2732  for (;;)
2732              }              }
2733            while (eptr >= pp)            while (eptr >= pp)
2734              {              {
2735              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2736              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2737              eptr--;              eptr--;
2738              }              }
# Line 2667  for (;;) Line 2808  for (;;)
2808          {          {
2809          for (fi = min;; fi++)          for (fi = min;; fi++)
2810            {            {
2811            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2812            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2813            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2814            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2700  for (;;) Line 2841  for (;;)
2841            }            }
2842          for(;;)          for(;;)
2843            {            {
2844            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2845            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2846            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2847            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2745  for (;;) Line 2886  for (;;)
2886    
2887      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2888    
2889      case OP_CHARNC:      case OP_CHARI:
2890  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2891      if (utf8)      if (utf8)
2892        {        {
# Line 2805  for (;;) Line 2946  for (;;)
2946      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2947    
2948      case OP_EXACT:      case OP_EXACT:
2949        case OP_EXACTI:
2950      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2951      ecode += 3;      ecode += 3;
2952      goto REPEATCHAR;      goto REPEATCHAR;
2953    
2954      case OP_POSUPTO:      case OP_POSUPTO:
2955        case OP_POSUPTOI:
2956      possessive = TRUE;      possessive = TRUE;
2957      /* Fall through */      /* Fall through */
2958    
2959      case OP_UPTO:      case OP_UPTO:
2960        case OP_UPTOI:
2961      case OP_MINUPTO:      case OP_MINUPTO:
2962        case OP_MINUPTOI:
2963      min = 0;      min = 0;
2964      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2965      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2966      ecode += 3;      ecode += 3;
2967      goto REPEATCHAR;      goto REPEATCHAR;
2968    
2969      case OP_POSSTAR:      case OP_POSSTAR:
2970        case OP_POSSTARI:
2971      possessive = TRUE;      possessive = TRUE;
2972      min = 0;      min = 0;
2973      max = INT_MAX;      max = INT_MAX;
# Line 2829  for (;;) Line 2975  for (;;)
2975      goto REPEATCHAR;      goto REPEATCHAR;
2976    
2977      case OP_POSPLUS:      case OP_POSPLUS:
2978        case OP_POSPLUSI:
2979      possessive = TRUE;      possessive = TRUE;
2980      min = 1;      min = 1;
2981      max = INT_MAX;      max = INT_MAX;
# Line 2836  for (;;) Line 2983  for (;;)
2983      goto REPEATCHAR;      goto REPEATCHAR;
2984    
2985      case OP_POSQUERY:      case OP_POSQUERY:
2986        case OP_POSQUERYI:
2987      possessive = TRUE;      possessive = TRUE;
2988      min = 0;      min = 0;
2989      max = 1;      max = 1;
# Line 2843  for (;;) Line 2991  for (;;)
2991      goto REPEATCHAR;      goto REPEATCHAR;
2992    
2993      case OP_STAR:      case OP_STAR:
2994        case OP_STARI:
2995      case OP_MINSTAR:      case OP_MINSTAR:
2996        case OP_MINSTARI:
2997      case OP_PLUS:      case OP_PLUS:
2998        case OP_PLUSI:
2999      case OP_MINPLUS:      case OP_MINPLUS:
3000        case OP_MINPLUSI:
3001      case OP_QUERY:      case OP_QUERY:
3002        case OP_QUERYI:
3003      case OP_MINQUERY:      case OP_MINQUERY:
3004      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3005        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3006      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3007      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3008      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3009      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2873  for (;;) Line 3026  for (;;)
3026          {          {
3027  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3028          unsigned int othercase;          unsigned int othercase;
3029          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3030              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3031            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3032          else oclength = 0;          else oclength = 0;
# Line 2901  for (;;) Line 3054  for (;;)
3054            {            {
3055            for (fi = min;; fi++)            for (fi = min;; fi++)
3056              {              {
3057              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3058              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3059              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3060              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2943  for (;;) Line 3096  for (;;)
3096    
3097            for(;;)            for(;;)
3098              {              {
3099              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3100              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3101              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3102  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2980  for (;;) Line 3133  for (;;)
3133      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3134        max, eptr));        max, eptr));
3135    
3136      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3137        {        {
3138        fc = md->lcc[fc];        fc = md->lcc[fc];
3139        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2997  for (;;) Line 3150  for (;;)
3150          {          {
3151          for (fi = min;; fi++)          for (fi = min;; fi++)
3152            {            {
3153            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3154            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3155            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3156            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3027  for (;;) Line 3180  for (;;)
3180    
3181          while (eptr >= pp)          while (eptr >= pp)
3182            {            {
3183            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3184            eptr--;            eptr--;
3185            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3186            }            }
# Line 3056  for (;;) Line 3209  for (;;)
3209          {          {
3210          for (fi = min;; fi++)          for (fi = min;; fi++)
3211            {            {
3212            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3213            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3214            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3215            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3085  for (;;) Line 3238  for (;;)
3238    
3239          while (eptr >= pp)          while (eptr >= pp)
3240            {            {
3241            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3242            eptr--;            eptr--;
3243            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3244            }            }
# Line 3098  for (;;) Line 3251  for (;;)
3251      checking can be multibyte. */      checking can be multibyte. */
3252    
3253      case OP_NOT:      case OP_NOT:
3254        case OP_NOTI:
3255      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3256        {        {
3257        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 3105  for (;;) Line 3259  for (;;)
3259        }        }
3260      ecode++;      ecode++;
3261      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3262      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3263        {        {
3264  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3265        if (c < 256)        if (c < 256)
# Line 3113  for (;;) Line 3267  for (;;)
3267        c = md->lcc[c];        c = md->lcc[c];
3268        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3269        }        }
3270      else      else    /* Caseful */
3271        {        {
3272        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3273        }        }
# Line 3127  for (;;) Line 3281  for (;;)
3281      about... */      about... */
3282    
3283      case OP_NOTEXACT:      case OP_NOTEXACT:
3284        case OP_NOTEXACTI:
3285      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3286      ecode += 3;      ecode += 3;
3287      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3288    
3289      case OP_NOTUPTO:      case OP_NOTUPTO:
3290        case OP_NOTUPTOI:
3291      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3292        case OP_NOTMINUPTOI:
3293      min = 0;      min = 0;
3294      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3295      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3296      ecode += 3;      ecode += 3;
3297      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3298    
3299      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3300        case OP_NOTPOSSTARI:
3301      possessive = TRUE;      possessive = TRUE;
3302      min = 0;      min = 0;
3303      max = INT_MAX;      max = INT_MAX;
# Line 3147  for (;;) Line 3305  for (;;)
3305      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3306    
3307      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3308        case OP_NOTPOSPLUSI:
3309      possessive = TRUE;      possessive = TRUE;
3310      min = 1;      min = 1;
3311      max = INT_MAX;      max = INT_MAX;
# Line 3154  for (;;) Line 3313  for (;;)
3313      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3314    
3315      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3316        case OP_NOTPOSQUERYI:
3317      possessive = TRUE;      possessive = TRUE;
3318      min = 0;      min = 0;
3319      max = 1;      max = 1;
# Line 3161  for (;;) Line 3321  for (;;)
3321      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3322    
3323      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3324        case OP_NOTPOSUPTOI:
3325      possessive = TRUE;      possessive = TRUE;
3326      min = 0;      min = 0;
3327      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 3168  for (;;) Line 3329  for (;;)
3329      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3330    
3331      case OP_NOTSTAR:      case OP_NOTSTAR:
3332        case OP_NOTSTARI:
3333      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3334        case OP_NOTMINSTARI:
3335      case OP_NOTPLUS:      case OP_NOTPLUS:
3336        case OP_NOTPLUSI:
3337      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3338        case OP_NOTMINPLUSI:
3339      case OP_NOTQUERY:      case OP_NOTQUERY:
3340        case OP_NOTQUERYI:
3341      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3342      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3343        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3344      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3345      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3346      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3195  for (;;) Line 3362  for (;;)
3362      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,
3363        max, eptr));        max, eptr));
3364    
3365      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3366        {        {
3367        fc = md->lcc[fc];        fc = md->lcc[fc];
3368    
# Line 3243  for (;;) Line 3410  for (;;)
3410            register unsigned int d;            register unsigned int d;
3411            for (fi = min;; fi++)            for (fi = min;; fi++)
3412              {              {
3413              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3414              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3415              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3416              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3262  for (;;) Line 3429  for (;;)
3429            {            {
3430            for (fi = min;; fi++)            for (fi = min;; fi++)
3431              {              {
3432              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3433              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3434              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3435              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3303  for (;;) Line 3470  for (;;)
3470          if (possessive) continue;          if (possessive) continue;
3471          for(;;)          for(;;)
3472              {              {
3473              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3474              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3475              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3476              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3326  for (;;) Line 3493  for (;;)
3493            if (possessive) continue;            if (possessive) continue;
3494            while (eptr >= pp)            while (eptr >= pp)
3495              {              {
3496              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3497              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3498              eptr--;              eptr--;
3499              }              }
# Line 3383  for (;;) Line 3550  for (;;)
3550            register unsigned int d;            register unsigned int d;
3551            for (fi = min;; fi++)            for (fi = min;; fi++)
3552              {              {
3553              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3554              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3555              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3556              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3401  for (;;) Line 3568  for (;;)
3568            {            {
3569            for (fi = min;; fi++)            for (fi = min;; fi++)
3570              {              {
3571              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3572              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3573              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3574              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3441  for (;;) Line 3608  for (;;)
3608            if (possessive) continue;            if (possessive) continue;
3609            for(;;)            for(;;)
3610              {              {
3611              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3612              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3613              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3614              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3464  for (;;) Line 3631  for (;;)
3631            if (possessive) continue;            if (possessive) continue;
3632            while (eptr >= pp)            while (eptr >= pp)
3633              {              {
3634              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3635              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3636              eptr--;              eptr--;
3637              }              }
# Line 3579  for (;;) Line 3746  for (;;)
3746            case PT_LAMP:            case PT_LAMP:
3747            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3748              {              {
3749                int chartype;
3750              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3751                {                {
3752                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3753                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3754                }                }
3755              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3756              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
3757              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
3758                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
3759                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
3760                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3761              }              }
3762            break;            break;
# Line 3602  for (;;) Line 3770  for (;;)
3770                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3771                }                }
3772              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3773              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
3774                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3775              }              }
3776            break;            break;
# Line 3617  for (;;) Line 3784  for (;;)
3784                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3785                }                }
3786              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3787              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
3788                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3789              }              }
3790            break;            break;
# Line 3632  for (;;) Line 3798  for (;;)
3798                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3799                }                }
3800              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3801              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
3802                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3803              }              }
3804            break;            break;
# Line 3641  for (;;) Line 3806  for (;;)
3806            case PT_ALNUM:            case PT_ALNUM:
3807            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3808              {              {
3809                int category;
3810              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3811                {                {
3812                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3813                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3814                }                }
3815              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3816              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3817              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
3818                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3819              }              }
3820            break;            break;
# Line 3663  for (;;) Line 3828  for (;;)
3828                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3829                }                }
3830              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3831              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
3832                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
3833                     == prop_fail_result)                     == prop_fail_result)
3834                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3680  for (;;) Line 3844  for (;;)
3844                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3845                }                }
3846              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3847              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
3848                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3849                     == prop_fail_result)                     == prop_fail_result)
3850                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3691  for (;;) Line 3854  for (;;)
3854            case PT_WORD:            case PT_WORD:
3855            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3856              {              {
3857                int category;
3858              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3859                {                {
3860                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3861                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3862                }                }
3863              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3864              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3865              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
                  c == CHAR_UNDERSCORE)  
3866                     == prop_fail_result)                     == prop_fail_result)
3867                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3868              }              }
# Line 3725  for (;;) Line 3888  for (;;)
3888              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
3889              }              }
3890            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3891            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
3892            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3893              {              {
3894              int len = 1;              int len = 1;
3895              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
3896                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
3897              eptr += len;              eptr += len;
3898              }              }
3899            }            }
# Line 3791  for (;;) Line 3951  for (;;)
3951            switch(c)            switch(c)
3952              {              {
3953              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3954    
3955              case 0x000d:              case 0x000d:
3956              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3957              break;              break;
# Line 4067  for (;;) Line 4228  for (;;)
4228            switch(*eptr++)            switch(*eptr++)
4229              {              {
4230              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4231    
4232              case 0x000d:              case 0x000d:
4233              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4234              break;              break;
4235    
4236              case 0x000a:              case 0x000a:
4237              break;              break;
4238    
# Line 4259  for (;;) Line 4422  for (;;)
4422            case PT_ANY:            case PT_ANY:
4423            for (fi = min;; fi++)            for (fi = min;; fi++)
4424              {              {
4425              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4426              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4427              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4428              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4275  for (;;) Line 4438  for (;;)
4438            case PT_LAMP:            case PT_LAMP:
4439            for (fi = min;; fi++)            for (fi = min;; fi++)
4440              {              {
4441              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              int chartype;
4442                RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4443              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4444              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4445              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4284  for (;;) Line 4448  for (;;)
4448                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4449                }                }
4450              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4451              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4452              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4453                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4454                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4455                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4456              }              }
4457            /* Control never gets here */            /* Control never gets here */
# Line 4295  for (;;) Line 4459  for (;;)
4459            case PT_GC:            case PT_GC:
4460            for (fi = min;; fi++)            for (fi = min;; fi++)
4461              {              {
4462              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4463              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4464              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4465              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4304  for (;;) Line 4468  for (;;)
4468                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4469                }                }
4470              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4471              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4472                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4473              }              }
4474            /* Control never gets here */            /* Control never gets here */
# Line 4313  for (;;) Line 4476  for (;;)
4476            case PT_PC:            case PT_PC:
4477            for (fi = min;; fi++)            for (fi = min;; fi++)
4478              {              {
4479              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
4480              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4481              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4482              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4322  for (;;) Line 4485  for (;;)
4485                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4486                }                }
4487              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4488              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4489                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4490              }              }
4491            /* Control never gets here */            /* Control never gets here */
# Line 4331  for (;;) Line 4493  for (;;)
4493            case PT_SC:            case PT_SC:
4494            for (fi = min;; fi++)            for (fi = min;; fi++)
4495              {              {
4496              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
4497              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4498              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4499              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4340  for (;;) Line 4502  for (;;)
4502                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4503                }                }
4504              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4505              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
4506                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4507              }              }
4508            /* Control never gets here */            /* Control never gets here */
# Line 4349  for (;;) Line 4510  for (;;)
4510            case PT_ALNUM:            case PT_ALNUM:
4511            for (fi = min;; fi++)            for (fi = min;; fi++)
4512              {              {
4513              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59);              int category;
4514                RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4515              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4516              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4517              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4358  for (;;) Line 4520  for (;;)
4520                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4521                }                }
4522              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4523              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4524              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
4525                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4526              }              }
4527            /* Control never gets here */            /* Control never gets here */
# Line 4368  for (;;) Line 4529  for (;;)
4529            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4530            for (fi = min;; fi++)            for (fi = min;; fi++)
4531              {              {
4532              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4533              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4534              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4535              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4377  for (;;) Line 4538  for (;;)
4538                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4539                }                }
4540              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4541              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
4542                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
4543                     == prop_fail_result)                     == prop_fail_result)
4544                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4388  for (;;) Line 4548  for (;;)
4548            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4549            for (fi = min;; fi++)            for (fi = min;; fi++)
4550              {              {
4551              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4552              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4553              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4554              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4397  for (;;) Line 4557  for (;;)
4557                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4558                }                }
4559              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4560              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
4561                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4562                     == prop_fail_result)                     == prop_fail_result)
4563                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4408  for (;;) Line 4567  for (;;)
4567            case PT_WORD:            case PT_WORD:
4568            for (fi = min;; fi++)            for (fi = min;; fi++)
4569              {              {
4570              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62);              int category;
4571                RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4572              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4573              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4574              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4417  for (;;) Line 4577  for (;;)
4577                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4578                }                }
4579              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4580              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4581              if ((prop_category == ucp_L ||              if ((category == ucp_L ||
4582                   prop_category == ucp_N ||                   category == ucp_N ||
4583                   c == CHAR_UNDERSCORE)                   c == CHAR_UNDERSCORE)
4584                     == prop_fail_result)                     == prop_fail_result)
4585                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4440  for (;;) Line 4600  for (;;)
4600          {          {
4601          for (fi = min;; fi++)          for (fi = min;; fi++)
4602            {            {
4603            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4604            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4605            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4606            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4449  for (;;) Line 4609  for (;;)
4609              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4610              }              }
4611            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4612            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4613            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4614              {              {
4615              int len = 1;              int len = 1;
4616              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4617                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4618              eptr += len;              eptr += len;
4619              }              }
4620            }            }
4621          }          }
   
4622        else        else
4623  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4624    
# Line 4472  for (;;) Line 4628  for (;;)
4628          {          {
4629          for (fi = min;; fi++)          for (fi = min;; fi++)
4630            {            {
4631            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4632            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4633            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4634            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4635  for (;;) Line 4791  for (;;)
4791          {          {
4792          for (fi = min;; fi++)          for (fi = min;; fi++)
4793            {            {
4794            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4795            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4796            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4797            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4783  for (;;) Line 4939  for (;;)
4939            case PT_LAMP:            case PT_LAMP:
4940            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4941              {              {
4942                int chartype;
4943              int len = 1;              int len = 1;
4944              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4945                {                {
# Line 4790  for (;;) Line 4947  for (;;)
4947                break;                break;
4948                }                }
4949              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4950              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4951              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4952                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4953                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4954                break;                break;
4955              eptr+= len;              eptr+= len;
4956              }              }
# Line 4809  for (;;) Line 4966  for (;;)
4966                break;                break;
4967                }                }
4968              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4969              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
4970              eptr+= len;              eptr+= len;
4971              }              }
4972            break;            break;
# Line 4826  for (;;) Line 4981  for (;;)
4981                break;                break;
4982                }                }
4983              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4984              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
4985              eptr+= len;              eptr+= len;
4986              }              }
4987            break;            break;
# Line 4843  for (;;) Line 4996  for (;;)
4996                break;                break;
4997                }                }
4998              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4999              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
             if ((prop_script == prop_value) == prop_fail_result)  
               break;  
5000              eptr+= len;              eptr+= len;
5001              }              }
5002            break;            break;
# Line 4853  for (;;) Line 5004  for (;;)
5004            case PT_ALNUM:            case PT_ALNUM:
5005            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5006              {              {
5007                int category;
5008              int len = 1;              int len = 1;
5009              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5010                {                {
# Line 4860  for (;;) Line 5012  for (;;)
5012                break;                break;
5013                }                }
5014              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5015              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5016              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                  == prop_fail_result)  
5017                break;                break;
5018              eptr+= len;              eptr+= len;
5019              }              }
# Line 4878  for (;;) Line 5029  for (;;)
5029                break;                break;
5030                }                }
5031              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5032              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
5033                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
5034                   == prop_fail_result)                   == prop_fail_result)
5035                break;                break;
# Line 4897  for (;;) Line 5047  for (;;)
5047                break;                break;
5048                }                }
5049              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5050              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
5051                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5052                   == prop_fail_result)                   == prop_fail_result)
5053                break;                break;
# Line 4909  for (;;) Line 5058  for (;;)
5058            case PT_WORD:            case PT_WORD:
5059            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5060              {              {
5061                int category;
5062              int len = 1;              int len = 1;
5063              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5064                {                {
# Line 4916  for (;;) Line 5066  for (;;)
5066                break;                break;
5067                }                }
5068              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5069              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5070              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N ||
5071                   c == CHAR_UNDERSCORE) == prop_fail_result)                   c == CHAR_UNDERSCORE) == prop_fail_result)
5072                break;                break;
5073              eptr+= len;              eptr+= len;
# Line 4933  for (;;) Line 5083  for (;;)
5083          if (possessive) continue;          if (possessive) continue;
5084          for(;;)          for(;;)
5085            {            {
5086            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5087            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5088            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5089            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4947  for (;;) Line 5097  for (;;)
5097          {          {
5098          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5099            {            {
5100              int len = 1;
5101            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5102              {              {
5103              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5104              break;              break;
5105              }              }
5106            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5107            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5108            if (prop_category == ucp_M) break;            eptr += len;
5109            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5110              {              {
5111              int len = 1;              len = 1;
5112              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5113                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5114              eptr += len;              eptr += len;
5115              }              }
5116            }            }
# Line 4974  for (;;) Line 5121  for (;;)
5121    
5122          for(;;)          for(;;)
5123            {            {
5124            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5125            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5126            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5127            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5128              {              {
             int len = 1;  
5129              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5130                {                {
5131                BACKCHAR(eptr);                BACKCHAR(eptr);
5132                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5133                }                }
5134              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5135              eptr--;              eptr--;
5136              }              }
5137            }            }
# Line 5050  for (;;) Line 5195  for (;;)
5195                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
5196                }                }
5197              }              }
5198            else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */            else
5199                {
5200                eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
5201                SCHECK_PARTIAL();
5202                }
5203            break;            break;
5204    
5205            /* The byte case is the same as non-UTF8 */            /* The byte case is the same as non-UTF8 */
# Line 5258  for (;;) Line 5407  for (;;)
5407            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5408            }            }
5409    
5410          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5411            done (no backing up). Otherwise, match at this position; anything other
5412            than no match is immediately returned. For nomatch, back up one
5413            character, unless we are matching \R and the last thing matched was
5414            \r\n, in which case, back up two bytes. */
5415    
5416          if (possessive) continue;          if (possessive) continue;
5417          for(;;)          for(;;)
5418            {            {
5419            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5420            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5421            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5422            BACKCHAR(eptr);            BACKCHAR(eptr);
5423              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5424                  eptr[-1] == '\r') eptr--;
5425            }            }
5426          }          }
5427        else        else
# Line 5465  for (;;) Line 5620  for (;;)
5620            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5621            }            }
5622    
5623          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5624            done (no backing up). Otherwise, match at this position; anything other
5625            than no match is immediately returned. For nomatch, back up one
5626            character (byte), unless we are matching \R and the last thing matched
5627            was \r\n, in which case, back up two bytes. */
5628    
5629          if (possessive) continue;          if (possessive) continue;
5630          while (eptr >= pp)          while (eptr >= pp)
5631            {            {
5632            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5633            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5634              eptr--;
5635              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5636                  eptr[-1] == '\r') eptr--;
5637            }            }
5638          }          }
5639    
# Line 5511  switch (frame->Xwhere) Line 5672  switch (frame->Xwhere)
5672    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)
5673    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)
5674    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)
5675    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63)
5676  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5677    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)
5678    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
# Line 5540  Undefine all the macros that were define Line 5701  Undefine all the macros that were define
5701  #undef ecode  #undef ecode
5702  #undef mstart  #undef mstart
5703  #undef offset_top  #undef offset_top
 #undef ims  
5704  #undef eptrb  #undef eptrb
5705  #undef flags  #undef flags
5706    
# Line 5558  Undefine all the macros that were define Line 5718  Undefine all the macros that were define
5718  #undef condition  #undef condition
5719  #undef prev_is_word  #undef prev_is_word
5720    
 #undef original_ims  
   
5721  #undef ctype  #undef ctype
5722  #undef length  #undef length
5723  #undef max  #undef max
# Line 5616  pcre_exec(const pcre *argument_re, const Line 5774  pcre_exec(const pcre *argument_re, const
5774    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5775    int offsetcount)    int offsetcount)
5776  {  {
5777  int rc, resetcount, ocount;  int rc, ocount, arg_offset_max;
5778  int first_byte = -1;  int first_byte = -1;
5779  int req_byte = -1;  int req_byte = -1;
5780  int req_byte2 = -1;  int req_byte2 = -1;
5781  int newline;  int newline;
 unsigned long int ims;  
5782  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5783  BOOL anchored;  BOOL anchored;
5784  BOOL startline;  BOOL startline;
# Line 5653  if (re == NULL || subject == NULL || Line 5810  if (re == NULL || subject == NULL ||
5810  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
5811  if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;  if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
5812    
5813  /* This information is for finding all the numbers associated with a given  /* These two settings are used in the code for checking a UTF-8 string that
5814  name, for condition testing. */  follows immediately afterwards. Other values in the md block are used only
5815    during "normal" pcre_exec() processing, not when the JIT support is in use,
5816    so they are set up later. */
5817    
5818    utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;
5819    md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5820                  ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5821    
5822    /* Check a UTF-8 string if required. Pass back the character offset and error
5823    code for an invalid string if a results vector is available. */
5824    
5825    #ifdef SUPPORT_UTF8
5826    if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5827      {
5828      int erroroffset;
5829      int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5830      if (errorcode != 0)
5831        {
5832        if (offsetcount >= 2)
5833          {
5834          offsets[0] = erroroffset;
5835          offsets[1] = errorcode;
5836          }
5837        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5838          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5839        }
5840    
5841      /* Check that a start_offset points to the start of a UTF-8 character. */
5842      if (start_offset > 0 && start_offset < length &&
5843          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
5844        return PCRE_ERROR_BADUTF8_OFFSET;
5845      }
5846    #endif
5847    
5848    /* If the pattern was successfully studied with JIT support, run the JIT
5849    executable instead of the rest of this function. Most options must be set at
5850    compile time for the JIT code to be usable. Fallback to the normal code path if
5851    an unsupported flag is set. In particular, JIT does not support partial
5852    matching. */
5853    
5854    #ifdef SUPPORT_JIT
5855    if (extra_data != NULL
5856        && (extra_data->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0
5857        && extra_data->executable_jit != NULL
5858        && (options & ~(PCRE_NO_UTF8_CHECK | PCRE_NOTBOL | PCRE_NOTEOL |
5859                        PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART)) == 0)
5860      return _pcre_jit_exec(re, extra_data->executable_jit, subject, length,
5861        start_offset, options, ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0)
5862        ? MATCH_LIMIT : extra_data->match_limit, offsets, offsetcount);
5863    #endif
5864    
5865    /* Carry on with non-JIT matching. This information is for finding all the
5866    numbers associated with a given name, for condition testing. */
5867    
5868  md->name_table = (uschar *)re + re->name_table_offset;  md->name_table = (uschar *)re + re->name_table_offset;
5869  md->name_count = re->name_count;  md->name_count = re->name_count;
# Line 5721  md->end_subject = md->start_subject + le Line 5930  md->end_subject = md->start_subject + le
5930  end_subject = md->end_subject;  end_subject = md->end_subject;
5931    
5932  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
 utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;  
5933  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
5934  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
5935    
5936    /* Some options are unpacked into BOOL variables in the hope that testing
5937    them will be faster than individual option bits. */
5938    
5939  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
5940  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
5941  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
5942  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
5943  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  
               ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;  
5944  md->hitend = FALSE;  md->hitend = FALSE;
5945  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
5946    
# Line 5812  defined (though never set). So there's n Line 6022  defined (though never set). So there's n
6022  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
6023    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
6024    
 /* 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)  
   {  
   int tb;  
   if ((tb = _pcre_valid_utf8((USPTR)subject, length)) >= 0)  
     return (tb == length && md->partial > 1)?  
       PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;  
   if (start_offset > 0 && start_offset < length)  
     {  
     tb = ((USPTR)subject)[start_offset] & 0xc0;  
     if (tb == 0x80) 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);  
   
6025  /* 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
6026  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.
6027  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
6028  of 3. */  of 3. */
6029    
6030  ocount = offsetcount - (offsetcount % 3);  ocount = offsetcount - (offsetcount % 3);
6031    arg_offset_max = (2*ocount)/3;
6032    
6033  if (re->top_backref > 0 && re->top_backref >= ocount/3)  if (re->top_backref > 0 && re->top_backref >= ocount/3)
6034    {    {
# Line 5858  md->offset_max = (2*ocount)/3; Line 6045  md->offset_max = (2*ocount)/3;
6045  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
6046  md->capture_last = -1;  md->capture_last = -1;
6047    
 /* 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;  
   
6048  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
6049  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
6050  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
6051    offsets for the matched string. This is really just for tidiness with callouts,
6052    in case they inspect these fields. */
6053    
6054  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
6055    {    {
6056    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6057    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6058      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6059    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6060      md->offset_vector[0] = md->offset_vector[1] = -1;
6061    }    }
6062    
6063  /* 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 5907  if ((re->flags & PCRE_REQCHSET) != 0) Line 6091  if ((re->flags & PCRE_REQCHSET) != 0)
6091    }    }
6092    
6093    
6094    
6095    
6096  /* ==========================================================================*/  /* ==========================================================================*/
6097    
6098  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 5917  for(;;) Line 6103  for(;;)
6103    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6104    USPTR new_start_match;    USPTR new_start_match;
6105    
   /* 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;  
     }  
   
6106    /* 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
6107    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
6108    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 6114  for(;;) Line 6291  for(;;)
6291    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6292    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6293    md->match_call_count = 0;    md->match_call_count = 0;
6294    rc = match(start_match, md->start_code, start_match, NULL, 2, md, ims, NULL,    md->match_function_type = 0;
6295      0, 0);    md->end_offset_top = 0;
6296      rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6297    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
6298    
6299    switch(rc)    switch(rc)
# Line 6225  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6403  if (rc == MATCH_MATCH || rc == MATCH_ACC
6403    {    {
6404    if (using_temporary_offsets)    if (using_temporary_offsets)
6405      {      {
6406      if (offsetcount >= 4)      if (arg_offset_max >= 4)
6407        {        {
6408        memcpy(offsets + 2, md->offset_vector + 2,        memcpy(offsets + 2, md->offset_vector + 2,
6409          (offsetcount - 2) * sizeof(int));          (arg_offset_max - 2) * sizeof(int));
6410        DPRINTF(("Copied offsets from temporary memory\n"));        DPRINTF(("Copied offsets from temporary memory\n"));
6411        }        }
6412      if (md->end_offset_top > offsetcount) md->offset_overflow = TRUE;      if (md->end_offset_top > arg_offset_max) md->offset_overflow = TRUE;
6413      DPRINTF(("Freeing temporary memory\n"));      DPRINTF(("Freeing temporary memory\n"));
6414      (pcre_free)(md->offset_vector);      (pcre_free)(md->offset_vector);
6415      }      }
6416    
6417    /* Set the return code to the number of captured strings, or 0 if there are    /* Set the return code to the number of captured strings, or 0 if there were
6418    too many to fit into the vector. */    too many to fit into the vector. */
6419    
6420    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = (md->offset_overflow && md->end_offset_top >= arg_offset_max)?
6421        0 : md->end_offset_top/2;
6422    
6423      /* If there is space in the offset vector, set any unused pairs at the end of
6424      the pattern to -1 for backwards compatibility. It is documented that this
6425      happens. In earlier versions, the whole set of potential capturing offsets
6426      was set to -1 each time round the loop, but this is handled differently now.
6427      "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only
6428      those at the end that need unsetting here. We can't just unset them all at
6429      the start of the whole thing because they may get set in one branch that is
6430      not the final matching branch. */
6431    
6432      if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL)
6433        {
6434        register int *iptr, *iend;
6435        int resetcount = 2 + re->top_bracket * 2;
6436        if (resetcount > offsetcount) resetcount = ocount;
6437        iptr = offsets + md->end_offset_top;
6438        iend = offsets + resetcount;
6439        while (iptr < iend) *iptr++ = -1;
6440        }
6441    
6442    /* If there is space, set up the whole thing as substring 0. The value of    /* If there is space, set up the whole thing as substring 0. The value of
6443    md->start_match_ptr might be modified if \K was encountered on the success    md->start_match_ptr might be modified if \K was encountered on the success

Legend:
Removed from v.597  
changed lines
  Added in v.699

  ViewVC Help
Powered by ViewVC 1.1.5