/[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 654 by ph10, Tue Aug 2 11:00:40 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
589    const uschar *charptr;
590    #endif
591    const uschar *callpat;
592    const uschar *data;
593    const uschar *next;
594    USPTR         pp;
595    const uschar *prev;
596    USPTR         saved_eptr;
597    
598    recursion_info new_recursive;
599    
600  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  BOOL cur_is_word;
 const uschar *charptr;             /* in small blocks of the code. My normal */  
 #endif                             /* style of coding would have declared    */  
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
 unsigned long int original_ims;  
   
604  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
608  int oclength;  int oclength;
609  uschar occhars[8];  uschar occhars[8];
610  #endif  #endif
# Line 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        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
987        otherwise return. */          {
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          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
994          if (allow_zero || matched_once)
995            {
996            ecode += 1 + LINK_SIZE;
997            break;
998            }
999    
1000          RRETURN(MATCH_NOMATCH);
1001          }
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        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      for (;;)
1028          eptrb, flags, RM2);        {
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          {        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1279          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          rrc = MATCH_NOMATCH;
1280          RRETURN(rrc);        RRETURN(rrc);
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
         goto TAIL_RECURSE;  
         }  
1281        }        }
1282      else                         /* Condition false & no alternative */      else                         /* Condition false & no alternative */
1283        {        {
# Line 1168  for (;;) Line 1310  for (;;)
1310      break;      break;
1311    
1312    
1313      /* 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. */  
1314    
     case OP_ACCEPT:  
1315      case OP_END:      case OP_END:
1316      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1317        {      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;  
       }  
1318    
1319      /* 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
1320      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
1321      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,
1322      if any. */      backtracking will then try other alternatives, if any. */
1323    
1324      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1325          (md->notempty ||           md->recursive == NULL &&
1326            (md->notempty_atstart &&           (md->notempty ||
1327              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1328                 mstart == md->start_subject + md->start_offset)))
1329        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1330    
1331      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1210  for (;;) Line 1340  for (;;)
1340      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1341      MRRETURN(rrc);      MRRETURN(rrc);
1342    
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
   
1343      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1344      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,
1345      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
1346      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
1347      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1348        of a condition, we want to return immediately afterwards. The caller of
1349        this incarnation of the match() function will have set MATCH_CONDASSERT in
1350        md->match_function type, and one of these opcodes will be the first opcode
1351        that is processed. We use a local variable that is preserved over calls to
1352        match() to remember this case. */
1353    
1354      case OP_ASSERT:      case OP_ASSERT:
1355      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1356        if (md->match_function_type == MATCH_CONDASSERT)
1357          {
1358          condassert = TRUE;
1359          md->match_function_type = 0;
1360          }
1361        else condassert = FALSE;
1362    
1363      do      do
1364        {        {
1365        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1366        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1367          {          {
1368          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1369            markptr = md->mark;
1370          break;          break;
1371          }          }
1372        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
# Line 1241  for (;;) Line 1375  for (;;)
1375        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1376        }        }
1377      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1378    
1379      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1380    
1381      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1382    
1383      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1384    
1385      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1386      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1261  for (;;) Line 1396  for (;;)
1396    
1397      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1398      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1399        if (md->match_function_type == MATCH_CONDASSERT)
1400          {
1401          condassert = TRUE;
1402          md->match_function_type = 0;
1403          }
1404        else condassert = FALSE;
1405    
1406      do      do
1407        {        {
1408        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
         RM5);  
1409        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1410        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1411          {          {
# Line 1278  for (;;) Line 1419  for (;;)
1419        }        }
1420      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1421    
1422      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1423    
1424      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1425      continue;      continue;
# Line 1324  for (;;) Line 1465  for (;;)
1465      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1466        {        {
1467        pcre_callout_block cb;        pcre_callout_block cb;
1468        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1469        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1470        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1471        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 1336  for (;;) Line 1477  for (;;)
1477        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1478        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1479        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1480          cb.mark             = markptr;
1481        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1482        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1483        }        }
# Line 1346  for (;;) Line 1488  for (;;)
1488      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
1489      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1490    
1491      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1492      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
1493      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
1494      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
1495      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
1496      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
1497      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.  
1498    
1499      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
1500      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
1501      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1502        a lot, so he is not to blame for the current way it works. */
1503    
1504      case OP_RECURSE:      case OP_RECURSE:
1505        {        {
1506          recursion_info *ri;
1507          int recno;
1508    
1509        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1510        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1511          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1512    
1513          /* Check for repeating a recursion without advancing the subject pointer.
1514          This should catch convoluted mutual recursions. (Some simple cases are
1515          caught at compile time.) */
1516    
1517          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1518            if (recno == ri->group_num && eptr == ri->subject_position)
1519              RRETURN(PCRE_ERROR_RECURSELOOP);
1520    
1521        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1522    
1523          new_recursive.group_num = recno;
1524          new_recursive.subject_position = eptr;
1525        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1526        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1527    
1528        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1529    
1530        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1531    
1532        /* Now save the offset data. */        /* Now save the offset data */
1533    
1534        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1535        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1388  for (;;) Line 1540  for (;;)
1540            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1541          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1542          }          }
   
1543        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1544              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1545    
1546        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1547        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1548          might be changed, so reset it before looping. */
1549    
1550        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1551        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1552        do        do
1553          {          {
1554            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1555          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1556            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1557            memcpy(md->offset_vector, new_recursive.offset_save,
1558                new_recursive.saved_max * sizeof(int));
1559          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1560            {            {
1561            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1562            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1563            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1564              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1565            MRRETURN(MATCH_MATCH);  
1566              /* Set where we got to in the subject, and reset the start in case
1567              it was changed by \K. This *is* propagated back out of a recursion,
1568              for Perl compatibility. */
1569    
1570              eptr = md->end_match_ptr;
1571              mstart = md->start_match_ptr;
1572              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1573            }            }
1574          else if (rrc != MATCH_NOMATCH &&          else if (rrc != MATCH_NOMATCH &&
1575                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
# Line 1420  for (;;) Line 1581  for (;;)
1581            }            }
1582    
1583          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1584          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1585          }          }
1586        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1432  for (;;) Line 1591  for (;;)
1591          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1592        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1593        }        }
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer, but resetting  
     the start-of-match value in case it was changed by \K. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */  
         {  
         mstart = md->start_match_ptr;  
         break;  
         }  
       if (rrc != MATCH_NOMATCH &&  
           (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
         RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
1594    
1595      if (ecode[1+LINK_SIZE] == OP_OPT)      RECURSION_MATCHED:
1596        {      break;
       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 */  
1597    
1598      /* 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
1599      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1529  for (;;) Line 1609  for (;;)
1609      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1610    
1611      case OP_BRAZERO:      case OP_BRAZERO:
1612        {      next = ecode + 1;
1613        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1614        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1615        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1616        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1617      break;      break;
1618    
1619      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1620        {      next = ecode + 1;
1621        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1622        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1623        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1624        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1625      break;      break;
1626    
1627      case OP_SKIPZERO:      case OP_SKIPZERO:
1628        {      next = ecode+1;
1629        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1630        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1631      break;      break;
1632    
1633        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1634        here; just jump to the group, with allow_zero set TRUE. */
1635    
1636        case OP_BRAPOSZERO:
1637        op = *(++ecode);
1638        allow_zero = TRUE;
1639        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1640          goto POSSESSIVE_NON_CAPTURE;
1641    
1642      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1643    
1644      case OP_KET:      case OP_KET:
1645      case OP_KETRMIN:      case OP_KETRMIN:
1646      case OP_KETRMAX:      case OP_KETRMAX:
1647        case OP_KETRPOS:
1648      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1649    
1650      /* 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
1651      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1652      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1653    
1654      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1655        {        {
1656        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1657        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1658        }        }
1659      else saved_eptr = NULL;      else saved_eptr = NULL;
1660    
1661      /* 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
1662      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
1663      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
1664      it was changed by \K. */      by \K. */
1665    
1666      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1667          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1668        {        {
1669        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1670        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1671        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1672        MRRETURN(MATCH_MATCH);        MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1673        }        }
1674    
1675      /* 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
1676      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1677      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
1678      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
1679      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
1680        the current subject position and start match pointer and give a MATCH
1681        return. */
1682    
1683      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1684            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1685        {        {
1686        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1687        offset = number << 1;        offset = number << 1;
# Line 1605  for (;;) Line 1691  for (;;)
1691        printf("\n");        printf("\n");
1692  #endif  #endif
1693    
1694          /* Handle a recursively called group. */
1695    
1696          if (md->recursive != NULL && md->recursive->group_num == number)
1697            {
1698            md->end_match_ptr = eptr;
1699            md->start_match_ptr = mstart;
1700            RRETURN(MATCH_MATCH);
1701            }
1702    
1703          /* Deal with capturing */
1704    
1705        md->capture_last = number;        md->capture_last = number;
1706        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1707          {          {
1708            /* If offset is greater than offset_top, it means that we are
1709            "skipping" a capturing group, and that group's offsets must be marked
1710            unset. In earlier versions of PCRE, all the offsets were unset at the
1711            start of matching, but this doesn't work because atomic groups and
1712            assertions can cause a value to be set that should later be unset.
1713            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1714            part of the atomic group, but this is not on the final matching path,
1715            so must be unset when 2 is set. (If there is no group 2, there is no
1716            problem, because offset_top will then be 2, indicating no capture.) */
1717    
1718            if (offset > offset_top)
1719              {
1720              register int *iptr = md->offset_vector + offset_top;
1721              register int *iend = md->offset_vector + offset;
1722              while (iptr < iend) *iptr++ = -1;
1723              }
1724    
1725            /* Now make the extraction */
1726    
1727          md->offset_vector[offset] =          md->offset_vector[offset] =
1728            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1729          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1730          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1731          }          }
1732          }
1733    
1734        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1735        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1736        group. This is the forcible breaking of infinite loops as implemented in
1737        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1738        processing the rest of the pattern at a lower level. If this results in a
1739        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1740        bypassing intermediate backup points, but resetting any captures that
1741        happened along the way. */
1742    
1743        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1744          {
1745          if (*prev == OP_ONCE)
1746          {          {
1747          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1748          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1749          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1750          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;  
1751          }          }
1752          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1753          break;
1754        }        }
1755    
1756      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1757      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1758        at a time from the outer level, thus saving stack. */
1759    
1760      ims = original_ims;      if (*ecode == OP_KETRPOS)
     DPRINTF(("ims reset to %02lx\n", ims));  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
1761        {        {
1762        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1763        break;        md->end_offset_top = offset_top;
1764          RRETURN(MATCH_KETRPOS);
1765        }        }
1766    
1767      /* 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
1768      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
1769      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
1770      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
1771        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1772    
1773      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1774        {        {
1775        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1776        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1777        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1778            {
1779            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1780            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1781            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1782            RRETURN(MATCH_ONCE);
1783            }
1784          if (*prev >= OP_SBRA)    /* Could match an empty string */
1785          {          {
1786          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1787            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1788          RRETURN(rrc);          RRETURN(rrc);
1789          }          }
1790        ecode = prev;        ecode = prev;
# Line 1670  for (;;) Line 1792  for (;;)
1792        }        }
1793      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1794        {        {
1795        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1796          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1797          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1798        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1799          if (*prev == OP_ONCE)
1800            {
1801            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1802            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1803            md->once_target = prev;
1804            RRETURN(MATCH_ONCE);
1805            }
1806        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1807        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1808        }        }
1809      /* Control never gets here */      /* Control never gets here */
1810    
1811      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1812    
1813      case OP_CIRC:      case OP_CIRC:
1814      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 */  
1815    
1816      /* Start of subject assertion */      /* Start of subject assertion */
1817    
# Line 1699  for (;;) Line 1820  for (;;)
1820      ecode++;      ecode++;
1821      break;      break;
1822    
1823        /* Multiline mode: start of subject unless notbol, or after any newline. */
1824    
1825        case OP_CIRCM:
1826        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1827        if (eptr != md->start_subject &&
1828            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1829          MRRETURN(MATCH_NOMATCH);
1830        ecode++;
1831        break;
1832    
1833      /* Start of match assertion */      /* Start of match assertion */
1834    
1835      case OP_SOM:      case OP_SOM:
# Line 1713  for (;;) Line 1844  for (;;)
1844      ecode++;      ecode++;
1845      break;      break;
1846    
1847      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1848      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1849    
1850      case OP_DOLL:      case OP_DOLLM:
1851      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1852        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1853        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 */  
1854        {        {
1855        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1856        if (!md->endonly) goto ASSERT_NL_OR_EOS;        SCHECK_PARTIAL();
1857        }        }
1858        ecode++;
1859        break;
1860    
1861        /* Not multiline mode: assert before a terminating newline or before end of
1862        subject unless noteol is set. */
1863    
1864        case OP_DOLL:
1865        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1866        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1867    
1868      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1869    
# Line 1884  for (;;) Line 2014  for (;;)
2014      /* Fall through */      /* Fall through */
2015    
2016      case OP_ALLANY:      case OP_ALLANY:
2017      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2018        {        {                            /* not be updated before SCHECK_PARTIAL. */
2019        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2020        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2021        }        }
2022        eptr++;
2023      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2024      ecode++;      ecode++;
2025      break;      break;
# Line 1897  for (;;) Line 2028  for (;;)
2028      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2029    
2030      case OP_ANYBYTE:      case OP_ANYBYTE:
2031      if (eptr++ >= md->end_subject)      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2032        {        {                            /* not be updated before SCHECK_PARTIAL. */
2033        SCHECK_PARTIAL();        SCHECK_PARTIAL();
2034        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2035        }        }
2036        eptr++;
2037      ecode++;      ecode++;
2038      break;      break;
2039    
# Line 2017  for (;;) Line 2149  for (;;)
2149      switch(c)      switch(c)
2150        {        {
2151        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2152    
2153        case 0x000d:        case 0x000d:
2154        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2155        break;        break;
# Line 2240  for (;;) Line 2373  for (;;)
2373        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2374        }        }
2375      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2376        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2377        while (eptr < md->end_subject)
2378        {        {
2379        int category = UCD_CATEGORY(c);        int len = 1;
2380        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2381        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2382          {        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;  
         }  
2383        }        }
2384      ecode++;      ecode++;
2385      break;      break;
# Line 2269  for (;;) Line 2395  for (;;)
2395      loops). */      loops). */
2396    
2397      case OP_REF:      case OP_REF:
2398        case OP_REFI:
2399        caseless = op == OP_REFI;
2400      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2401      ecode += 3;      ecode += 3;
2402    
# Line 2316  for (;;) Line 2444  for (;;)
2444        break;        break;
2445    
2446        default:               /* No repeat follows */        default:               /* No repeat follows */
2447        if ((length = match_ref(offset, eptr, length, md, ims)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2448          {          {
2449          CHECK_PARTIAL();          CHECK_PARTIAL();
2450          MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
# Line 2336  for (;;) Line 2464  for (;;)
2464    
2465      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2466        {        {
2467        int slength;        int slength;
2468        if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2469          {          {
2470          CHECK_PARTIAL();          CHECK_PARTIAL();
2471          MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
# Line 2356  for (;;) Line 2484  for (;;)
2484        {        {
2485        for (fi = min;; fi++)        for (fi = min;; fi++)
2486          {          {
2487          int slength;          int slength;
2488          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2489          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2490          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) MRRETURN(MATCH_NOMATCH);
2491          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2492            {            {
2493            CHECK_PARTIAL();            CHECK_PARTIAL();
2494            MRRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
# Line 2377  for (;;) Line 2505  for (;;)
2505        pp = eptr;        pp = eptr;
2506        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2507          {          {
2508          int slength;          int slength;
2509          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2510            {            {
2511            CHECK_PARTIAL();            CHECK_PARTIAL();
2512            break;            break;
# Line 2387  for (;;) Line 2515  for (;;)
2515          }          }
2516        while (eptr >= pp)        while (eptr >= pp)
2517          {          {
2518          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2519          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2520          eptr -= length;          eptr -= length;
2521          }          }
# Line 2497  for (;;) Line 2625  for (;;)
2625            {            {
2626            for (fi = min;; fi++)            for (fi = min;; fi++)
2627              {              {
2628              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2629              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2630              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2631              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2522  for (;;) Line 2650  for (;;)
2650            {            {
2651            for (fi = min;; fi++)            for (fi = min;; fi++)
2652              {              {
2653              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2654              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2655              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2656              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2568  for (;;) Line 2696  for (;;)
2696              }              }
2697            for (;;)            for (;;)
2698              {              {
2699              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2700              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2701              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2702              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2591  for (;;) Line 2719  for (;;)
2719              }              }
2720            while (eptr >= pp)            while (eptr >= pp)
2721              {              {
2722              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2723              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2724              eptr--;              eptr--;
2725              }              }
# Line 2667  for (;;) Line 2795  for (;;)
2795          {          {
2796          for (fi = min;; fi++)          for (fi = min;; fi++)
2797            {            {
2798            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2799            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2800            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2801            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2700  for (;;) Line 2828  for (;;)
2828            }            }
2829          for(;;)          for(;;)
2830            {            {
2831            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2832            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2833            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2834            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2745  for (;;) Line 2873  for (;;)
2873    
2874      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2875    
2876      case OP_CHARNC:      case OP_CHARI:
2877  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2878      if (utf8)      if (utf8)
2879        {        {
# Line 2805  for (;;) Line 2933  for (;;)
2933      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2934    
2935      case OP_EXACT:      case OP_EXACT:
2936        case OP_EXACTI:
2937      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2938      ecode += 3;      ecode += 3;
2939      goto REPEATCHAR;      goto REPEATCHAR;
2940    
2941      case OP_POSUPTO:      case OP_POSUPTO:
2942        case OP_POSUPTOI:
2943      possessive = TRUE;      possessive = TRUE;
2944      /* Fall through */      /* Fall through */
2945    
2946      case OP_UPTO:      case OP_UPTO:
2947        case OP_UPTOI:
2948      case OP_MINUPTO:      case OP_MINUPTO:
2949        case OP_MINUPTOI:
2950      min = 0;      min = 0;
2951      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2952      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2953      ecode += 3;      ecode += 3;
2954      goto REPEATCHAR;      goto REPEATCHAR;
2955    
2956      case OP_POSSTAR:      case OP_POSSTAR:
2957        case OP_POSSTARI:
2958      possessive = TRUE;      possessive = TRUE;
2959      min = 0;      min = 0;
2960      max = INT_MAX;      max = INT_MAX;
# Line 2829  for (;;) Line 2962  for (;;)
2962      goto REPEATCHAR;      goto REPEATCHAR;
2963    
2964      case OP_POSPLUS:      case OP_POSPLUS:
2965        case OP_POSPLUSI:
2966      possessive = TRUE;      possessive = TRUE;
2967      min = 1;      min = 1;
2968      max = INT_MAX;      max = INT_MAX;
# Line 2836  for (;;) Line 2970  for (;;)
2970      goto REPEATCHAR;      goto REPEATCHAR;
2971    
2972      case OP_POSQUERY:      case OP_POSQUERY:
2973        case OP_POSQUERYI:
2974      possessive = TRUE;      possessive = TRUE;
2975      min = 0;      min = 0;
2976      max = 1;      max = 1;
# Line 2843  for (;;) Line 2978  for (;;)
2978      goto REPEATCHAR;      goto REPEATCHAR;
2979    
2980      case OP_STAR:      case OP_STAR:
2981        case OP_STARI:
2982      case OP_MINSTAR:      case OP_MINSTAR:
2983        case OP_MINSTARI:
2984      case OP_PLUS:      case OP_PLUS:
2985        case OP_PLUSI:
2986      case OP_MINPLUS:      case OP_MINPLUS:
2987        case OP_MINPLUSI:
2988      case OP_QUERY:      case OP_QUERY:
2989        case OP_QUERYI:
2990      case OP_MINQUERY:      case OP_MINQUERY:
2991      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
2992        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
2993      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
2994      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2995      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2996      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2873  for (;;) Line 3013  for (;;)
3013          {          {
3014  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3015          unsigned int othercase;          unsigned int othercase;
3016          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3017              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3018            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3019          else oclength = 0;          else oclength = 0;
# Line 2901  for (;;) Line 3041  for (;;)
3041            {            {
3042            for (fi = min;; fi++)            for (fi = min;; fi++)
3043              {              {
3044              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3045              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3046              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3047              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2943  for (;;) Line 3083  for (;;)
3083    
3084            for(;;)            for(;;)
3085              {              {
3086              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3087              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3088              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3089  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2980  for (;;) Line 3120  for (;;)
3120      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3121        max, eptr));        max, eptr));
3122    
3123      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3124        {        {
3125        fc = md->lcc[fc];        fc = md->lcc[fc];
3126        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2997  for (;;) Line 3137  for (;;)
3137          {          {
3138          for (fi = min;; fi++)          for (fi = min;; fi++)
3139            {            {
3140            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3141            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3142            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3143            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3027  for (;;) Line 3167  for (;;)
3167    
3168          while (eptr >= pp)          while (eptr >= pp)
3169            {            {
3170            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3171            eptr--;            eptr--;
3172            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3173            }            }
# Line 3056  for (;;) Line 3196  for (;;)
3196          {          {
3197          for (fi = min;; fi++)          for (fi = min;; fi++)
3198            {            {
3199            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3200            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3201            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3202            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3085  for (;;) Line 3225  for (;;)
3225    
3226          while (eptr >= pp)          while (eptr >= pp)
3227            {            {
3228            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3229            eptr--;            eptr--;
3230            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3231            }            }
# Line 3098  for (;;) Line 3238  for (;;)
3238      checking can be multibyte. */      checking can be multibyte. */
3239    
3240      case OP_NOT:      case OP_NOT:
3241        case OP_NOTI:
3242      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3243        {        {
3244        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 3105  for (;;) Line 3246  for (;;)
3246        }        }
3247      ecode++;      ecode++;
3248      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3249      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3250        {        {
3251  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3252        if (c < 256)        if (c < 256)
# Line 3113  for (;;) Line 3254  for (;;)
3254        c = md->lcc[c];        c = md->lcc[c];
3255        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3256        }        }
3257      else      else    /* Caseful */
3258        {        {
3259        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3260        }        }
# Line 3127  for (;;) Line 3268  for (;;)
3268      about... */      about... */
3269    
3270      case OP_NOTEXACT:      case OP_NOTEXACT:
3271        case OP_NOTEXACTI:
3272      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3273      ecode += 3;      ecode += 3;
3274      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3275    
3276      case OP_NOTUPTO:      case OP_NOTUPTO:
3277        case OP_NOTUPTOI:
3278      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3279        case OP_NOTMINUPTOI:
3280      min = 0;      min = 0;
3281      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3282      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3283      ecode += 3;      ecode += 3;
3284      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3285    
3286      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3287        case OP_NOTPOSSTARI:
3288      possessive = TRUE;      possessive = TRUE;
3289      min = 0;      min = 0;
3290      max = INT_MAX;      max = INT_MAX;
# Line 3147  for (;;) Line 3292  for (;;)
3292      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3293    
3294      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3295        case OP_NOTPOSPLUSI:
3296      possessive = TRUE;      possessive = TRUE;
3297      min = 1;      min = 1;
3298      max = INT_MAX;      max = INT_MAX;
# Line 3154  for (;;) Line 3300  for (;;)
3300      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3301    
3302      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3303        case OP_NOTPOSQUERYI:
3304      possessive = TRUE;      possessive = TRUE;
3305      min = 0;      min = 0;
3306      max = 1;      max = 1;
# Line 3161  for (;;) Line 3308  for (;;)
3308      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3309    
3310      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3311        case OP_NOTPOSUPTOI:
3312      possessive = TRUE;      possessive = TRUE;
3313      min = 0;      min = 0;
3314      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 3168  for (;;) Line 3316  for (;;)
3316      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3317    
3318      case OP_NOTSTAR:      case OP_NOTSTAR:
3319        case OP_NOTSTARI:
3320      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3321        case OP_NOTMINSTARI:
3322      case OP_NOTPLUS:      case OP_NOTPLUS:
3323        case OP_NOTPLUSI:
3324      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3325        case OP_NOTMINPLUSI:
3326      case OP_NOTQUERY:      case OP_NOTQUERY:
3327        case OP_NOTQUERYI:
3328      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3329      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3330        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3331      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3332      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3333      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3195  for (;;) Line 3349  for (;;)
3349      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,
3350        max, eptr));        max, eptr));
3351    
3352      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3353        {        {
3354        fc = md->lcc[fc];        fc = md->lcc[fc];
3355    
# Line 3243  for (;;) Line 3397  for (;;)
3397            register unsigned int d;            register unsigned int d;
3398            for (fi = min;; fi++)            for (fi = min;; fi++)
3399              {              {
3400              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3401              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3402              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3403              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3262  for (;;) Line 3416  for (;;)
3416            {            {
3417            for (fi = min;; fi++)            for (fi = min;; fi++)
3418              {              {
3419              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3420              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3421              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3422              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3303  for (;;) Line 3457  for (;;)
3457          if (possessive) continue;          if (possessive) continue;
3458          for(;;)          for(;;)
3459              {              {
3460              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3461              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3462              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3463              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3326  for (;;) Line 3480  for (;;)
3480            if (possessive) continue;            if (possessive) continue;
3481            while (eptr >= pp)            while (eptr >= pp)
3482              {              {
3483              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3484              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3485              eptr--;              eptr--;
3486              }              }
# Line 3383  for (;;) Line 3537  for (;;)
3537            register unsigned int d;            register unsigned int d;
3538            for (fi = min;; fi++)            for (fi = min;; fi++)
3539              {              {
3540              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3541              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3542              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3543              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3401  for (;;) Line 3555  for (;;)
3555            {            {
3556            for (fi = min;; fi++)            for (fi = min;; fi++)
3557              {              {
3558              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3559              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3560              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3561              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3441  for (;;) Line 3595  for (;;)
3595            if (possessive) continue;            if (possessive) continue;
3596            for(;;)            for(;;)
3597              {              {
3598              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3599              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3600              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3601              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3464  for (;;) Line 3618  for (;;)
3618            if (possessive) continue;            if (possessive) continue;
3619            while (eptr >= pp)            while (eptr >= pp)
3620              {              {
3621              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3622              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3623              eptr--;              eptr--;
3624              }              }
# Line 3579  for (;;) Line 3733  for (;;)
3733            case PT_LAMP:            case PT_LAMP:
3734            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3735              {              {
3736                int chartype;
3737              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3738                {                {
3739                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3740                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3741                }                }
3742              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3743              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
3744              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
3745                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
3746                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
3747                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3748              }              }
3749            break;            break;
# Line 3602  for (;;) Line 3757  for (;;)
3757                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3758                }                }
3759              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3760              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
3761                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3762              }              }
3763            break;            break;
# Line 3617  for (;;) Line 3771  for (;;)
3771                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3772                }                }
3773              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3774              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
3775                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3776              }              }
3777            break;            break;
# Line 3632  for (;;) Line 3785  for (;;)
3785                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3786                }                }
3787              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3788              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
3789                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3790              }              }
3791            break;            break;
# Line 3641  for (;;) Line 3793  for (;;)
3793            case PT_ALNUM:            case PT_ALNUM:
3794            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3795              {              {
3796                int category;
3797              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3798                {                {
3799                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3800                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3801                }                }
3802              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3803              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3804              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
3805                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3806              }              }
3807            break;            break;
# Line 3663  for (;;) Line 3815  for (;;)
3815                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3816                }                }
3817              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3818              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 ||  
3819                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
3820                     == prop_fail_result)                     == prop_fail_result)
3821                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3680  for (;;) Line 3831  for (;;)
3831                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3832                }                }
3833              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3834              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 ||  
3835                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3836                     == prop_fail_result)                     == prop_fail_result)
3837                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3691  for (;;) Line 3841  for (;;)
3841            case PT_WORD:            case PT_WORD:
3842            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3843              {              {
3844                int category;
3845              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3846                {                {
3847                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3848                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3849                }                }
3850              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3851              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3852              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
                  c == CHAR_UNDERSCORE)  
3853                     == prop_fail_result)                     == prop_fail_result)
3854                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3855              }              }
# Line 3725  for (;;) Line 3875  for (;;)
3875              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
3876              }              }
3877            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3878            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
3879            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3880              {              {
3881              int len = 1;              int len = 1;
3882              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
3883                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
3884              eptr += len;              eptr += len;
3885              }              }
3886            }            }
# Line 3791  for (;;) Line 3938  for (;;)
3938            switch(c)            switch(c)
3939              {              {
3940              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3941    
3942              case 0x000d:              case 0x000d:
3943              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3944              break;              break;
# Line 4067  for (;;) Line 4215  for (;;)
4215            switch(*eptr++)            switch(*eptr++)
4216              {              {
4217              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4218    
4219              case 0x000d:              case 0x000d:
4220              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4221              break;              break;
4222    
4223              case 0x000a:              case 0x000a:
4224              break;              break;
4225    
# Line 4259  for (;;) Line 4409  for (;;)
4409            case PT_ANY:            case PT_ANY:
4410            for (fi = min;; fi++)            for (fi = min;; fi++)
4411              {              {
4412              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4413              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4414              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4415              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4275  for (;;) Line 4425  for (;;)
4425            case PT_LAMP:            case PT_LAMP:
4426            for (fi = min;; fi++)            for (fi = min;; fi++)
4427              {              {
4428              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              int chartype;
4429                RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4430              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4431              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4432              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4284  for (;;) Line 4435  for (;;)
4435                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4436                }                }
4437              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4438              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4439              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4440                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4441                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4442                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4443              }              }
4444            /* Control never gets here */            /* Control never gets here */
# Line 4295  for (;;) Line 4446  for (;;)
4446            case PT_GC:            case PT_GC:
4447            for (fi = min;; fi++)            for (fi = min;; fi++)
4448              {              {
4449              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4450              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4451              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4452              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4304  for (;;) Line 4455  for (;;)
4455                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4456                }                }
4457              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4458              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4459                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4460              }              }
4461            /* Control never gets here */            /* Control never gets here */
# Line 4313  for (;;) Line 4463  for (;;)
4463            case PT_PC:            case PT_PC:
4464            for (fi = min;; fi++)            for (fi = min;; fi++)
4465              {              {
4466              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
4467              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4468              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4469              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4322  for (;;) Line 4472  for (;;)
4472                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4473                }                }
4474              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4475              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4476                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4477              }              }
4478            /* Control never gets here */            /* Control never gets here */
# Line 4331  for (;;) Line 4480  for (;;)
4480            case PT_SC:            case PT_SC:
4481            for (fi = min;; fi++)            for (fi = min;; fi++)
4482              {              {
4483              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
4484              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4485              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4486              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4340  for (;;) Line 4489  for (;;)
4489                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4490                }                }
4491              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4492              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
4493                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4494              }              }
4495            /* Control never gets here */            /* Control never gets here */
# Line 4349  for (;;) Line 4497  for (;;)
4497            case PT_ALNUM:            case PT_ALNUM:
4498            for (fi = min;; fi++)            for (fi = min;; fi++)
4499              {              {
4500              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59);              int category;
4501                RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4502              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4503              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4504              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4358  for (;;) Line 4507  for (;;)
4507                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4508                }                }
4509              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4510              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4511              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
4512                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4513              }              }
4514            /* Control never gets here */            /* Control never gets here */
# Line 4368  for (;;) Line 4516  for (;;)
4516            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4517            for (fi = min;; fi++)            for (fi = min;; fi++)
4518              {              {
4519              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4520              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4521              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4522              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4377  for (;;) Line 4525  for (;;)
4525                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4526                }                }
4527              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4528              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 ||  
4529                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
4530                     == prop_fail_result)                     == prop_fail_result)
4531                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4388  for (;;) Line 4535  for (;;)
4535            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4536            for (fi = min;; fi++)            for (fi = min;; fi++)
4537              {              {
4538              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4539              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4540              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4541              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4397  for (;;) Line 4544  for (;;)
4544                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4545                }                }
4546              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4547              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 ||  
4548                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4549                     == prop_fail_result)                     == prop_fail_result)
4550                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4408  for (;;) Line 4554  for (;;)
4554            case PT_WORD:            case PT_WORD:
4555            for (fi = min;; fi++)            for (fi = min;; fi++)
4556              {              {
4557              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62);              int category;
4558                RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4559              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4560              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4561              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4417  for (;;) Line 4564  for (;;)
4564                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4565                }                }
4566              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4567              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4568              if ((prop_category == ucp_L ||              if ((category == ucp_L ||
4569                   prop_category == ucp_N ||                   category == ucp_N ||
4570                   c == CHAR_UNDERSCORE)                   c == CHAR_UNDERSCORE)
4571                     == prop_fail_result)                     == prop_fail_result)
4572                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4440  for (;;) Line 4587  for (;;)
4587          {          {
4588          for (fi = min;; fi++)          for (fi = min;; fi++)
4589            {            {
4590            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4591            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4592            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4593            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4449  for (;;) Line 4596  for (;;)
4596              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4597              }              }
4598            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4599            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4600            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4601              {              {
4602              int len = 1;              int len = 1;
4603              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4604                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4605              eptr += len;              eptr += len;
4606              }              }
4607            }            }
4608          }          }
   
4609        else        else
4610  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4611    
# Line 4472  for (;;) Line 4615  for (;;)
4615          {          {
4616          for (fi = min;; fi++)          for (fi = min;; fi++)
4617            {            {
4618            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4619            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4620            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4621            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4635  for (;;) Line 4778  for (;;)
4778          {          {
4779          for (fi = min;; fi++)          for (fi = min;; fi++)
4780            {            {
4781            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4782            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4783            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4784            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4783  for (;;) Line 4926  for (;;)
4926            case PT_LAMP:            case PT_LAMP:
4927            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4928              {              {
4929                int chartype;
4930              int len = 1;              int len = 1;
4931              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4932                {                {
# Line 4790  for (;;) Line 4934  for (;;)
4934                break;                break;
4935                }                }
4936              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4937              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4938              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4939                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4940                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4941                break;                break;
4942              eptr+= len;              eptr+= len;
4943              }              }
# Line 4809  for (;;) Line 4953  for (;;)
4953                break;                break;
4954                }                }
4955              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4956              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
4957              eptr+= len;              eptr+= len;
4958              }              }
4959            break;            break;
# Line 4826  for (;;) Line 4968  for (;;)
4968                break;                break;
4969                }                }
4970              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4971              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
4972              eptr+= len;              eptr+= len;
4973              }              }
4974            break;            break;
# Line 4843  for (;;) Line 4983  for (;;)
4983                break;                break;
4984                }                }
4985              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4986              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
             if ((prop_script == prop_value) == prop_fail_result)  
               break;  
4987              eptr+= len;              eptr+= len;
4988              }              }
4989            break;            break;
# Line 4853  for (;;) Line 4991  for (;;)
4991            case PT_ALNUM:            case PT_ALNUM:
4992            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4993              {              {
4994                int category;
4995              int len = 1;              int len = 1;
4996              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4997                {                {
# Line 4860  for (;;) Line 4999  for (;;)
4999                break;                break;
5000                }                }
5001              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5002              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5003              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                  == prop_fail_result)  
5004                break;                break;
5005              eptr+= len;              eptr+= len;
5006              }              }
# Line 4878  for (;;) Line 5016  for (;;)
5016                break;                break;
5017                }                }
5018              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5019              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
             if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
5020                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
5021                   == prop_fail_result)                   == prop_fail_result)
5022                break;                break;
# Line 4897  for (;;) Line 5034  for (;;)
5034                break;                break;
5035                }                }
5036              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5037              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 ||  
5038                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5039                   == prop_fail_result)                   == prop_fail_result)
5040                break;                break;
# Line 4909  for (;;) Line 5045  for (;;)
5045            case PT_WORD:            case PT_WORD:
5046            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5047              {              {
5048                int category;
5049              int len = 1;              int len = 1;
5050              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5051                {                {
# Line 4916  for (;;) Line 5053  for (;;)
5053                break;                break;
5054                }                }
5055              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5056              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5057              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N ||
5058                   c == CHAR_UNDERSCORE) == prop_fail_result)                   c == CHAR_UNDERSCORE) == prop_fail_result)
5059                break;                break;
5060              eptr+= len;              eptr+= len;
# Line 4933  for (;;) Line 5070  for (;;)
5070          if (possessive) continue;          if (possessive) continue;
5071          for(;;)          for(;;)
5072            {            {
5073            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5074            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5075            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5076            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4947  for (;;) Line 5084  for (;;)
5084          {          {
5085          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5086            {            {
5087              int len = 1;
5088            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5089              {              {
5090              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5091              break;              break;
5092              }              }
5093            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5094            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5095            if (prop_category == ucp_M) break;            eptr += len;
5096            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5097              {              {
5098              int len = 1;              len = 1;
5099              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5100                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5101              eptr += len;              eptr += len;
5102              }              }
5103            }            }
# Line 4974  for (;;) Line 5108  for (;;)
5108    
5109          for(;;)          for(;;)
5110            {            {
5111            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5112            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5113            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5114            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5115              {              {
             int len = 1;  
5116              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5117                {                {
5118                BACKCHAR(eptr);                BACKCHAR(eptr);
5119                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5120                }                }
5121              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5122              eptr--;              eptr--;
5123              }              }
5124            }            }
# Line 5050  for (;;) Line 5182  for (;;)
5182                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
5183                }                }
5184              }              }
5185            else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */            else
5186                {
5187                eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
5188                SCHECK_PARTIAL();
5189                }
5190            break;            break;
5191    
5192            /* The byte case is the same as non-UTF8 */            /* The byte case is the same as non-UTF8 */
# Line 5258  for (;;) Line 5394  for (;;)
5394            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5395            }            }
5396    
5397          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5398            done (no backing up). Otherwise, match at this position; anything other
5399            than no match is immediately returned. For nomatch, back up one
5400            character, unless we are matching \R and the last thing matched was
5401            \r\n, in which case, back up two bytes. */
5402    
5403          if (possessive) continue;          if (possessive) continue;
5404          for(;;)          for(;;)
5405            {            {
5406            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5407            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5408            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5409            BACKCHAR(eptr);            BACKCHAR(eptr);
5410              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5411                  eptr[-1] == '\r') eptr--;
5412            }            }
5413          }          }
5414        else        else
# Line 5465  for (;;) Line 5607  for (;;)
5607            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5608            }            }
5609    
5610          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5611            done (no backing up). Otherwise, match at this position; anything other
5612            than no match is immediately returned. For nomatch, back up one
5613            character (byte), unless we are matching \R and the last thing matched
5614            was \r\n, in which case, back up two bytes. */
5615    
5616          if (possessive) continue;          if (possessive) continue;
5617          while (eptr >= pp)          while (eptr >= pp)
5618            {            {
5619            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5620            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5621              eptr--;
5622              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5623                  eptr[-1] == '\r') eptr--;
5624            }            }
5625          }          }
5626    
# Line 5511  switch (frame->Xwhere) Line 5659  switch (frame->Xwhere)
5659    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)
5660    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)
5661    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)
5662    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)
5663  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5664    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)
5665    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 5688  Undefine all the macros that were define
5688  #undef ecode  #undef ecode
5689  #undef mstart  #undef mstart
5690  #undef offset_top  #undef offset_top
 #undef ims  
5691  #undef eptrb  #undef eptrb
5692  #undef flags  #undef flags
5693    
# Line 5558  Undefine all the macros that were define Line 5705  Undefine all the macros that were define
5705  #undef condition  #undef condition
5706  #undef prev_is_word  #undef prev_is_word
5707    
 #undef original_ims  
   
5708  #undef ctype  #undef ctype
5709  #undef length  #undef length
5710  #undef max  #undef max
# Line 5616  pcre_exec(const pcre *argument_re, const Line 5761  pcre_exec(const pcre *argument_re, const
5761    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5762    int offsetcount)    int offsetcount)
5763  {  {
5764  int rc, resetcount, ocount;  int rc, ocount;
5765  int first_byte = -1;  int first_byte = -1;
5766  int req_byte = -1;  int req_byte = -1;
5767  int req_byte2 = -1;  int req_byte2 = -1;
5768  int newline;  int newline;
 unsigned long int ims;  
5769  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5770  BOOL anchored;  BOOL anchored;
5771  BOOL startline;  BOOL startline;
# Line 5725  utf8 = md->utf8 = (re->options & PCRE_UT Line 5869  utf8 = md->utf8 = (re->options & PCRE_UT
5869  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
5870  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
5871    
5872    /* Some options are unpacked into BOOL variables in the hope that testing
5873    them will be faster than individual option bits. */
5874    
5875  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
5876  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
5877  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
5878  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
5879  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5880                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5881    
5882    
5883  md->hitend = FALSE;  md->hitend = FALSE;
5884  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
5885    
# Line 5812  defined (though never set). So there's n Line 5961  defined (though never set). So there's n
5961  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
5962    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
5963    
5964  /* Check a UTF-8 string if required. Unfortunately there's no way of passing  /* Check a UTF-8 string if required. Pass back the character offset and error
5965  back the character offset. */  code for an invalid string if a results vector is available. */
5966    
5967  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5968  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5969    {    {
5970    int tb;    int erroroffset;
5971    if ((tb = _pcre_valid_utf8((USPTR)subject, length)) >= 0)    int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5972      return (tb == length && md->partial > 1)?    if (errorcode != 0)
       PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;  
   if (start_offset > 0 && start_offset < length)  
5973      {      {
5974      tb = ((USPTR)subject)[start_offset] & 0xc0;      if (offsetcount >= 2)
5975      if (tb == 0x80) return PCRE_ERROR_BADUTF8_OFFSET;        {
5976          offsets[0] = erroroffset;
5977          offsets[1] = errorcode;
5978          }
5979        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5980          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5981      }      }
   }  
 #endif  
5982    
5983  /* The ims options can vary during the matching as a result of the presence    /* Check that a start_offset points to the start of a UTF-8 character. */
 of (?ims) items in the pattern. They are kept in a local variable so that  
 restoring at the exit of a group is easy. */  
5984    
5985  ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL);    if (start_offset > 0 && start_offset < length &&
5986          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
5987        return PCRE_ERROR_BADUTF8_OFFSET;
5988      }
5989    #endif
5990    
5991  /* 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
5992  hold, we get a temporary chunk of working store to use during the matching.  hold, we get a temporary chunk of working store to use during the matching.
# Line 5858  md->offset_max = (2*ocount)/3; Line 6010  md->offset_max = (2*ocount)/3;
6010  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
6011  md->capture_last = -1;  md->capture_last = -1;
6012    
 /* 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;  
   
6013  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
6014  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
6015  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
6016    offsets for the matched string. This is really just for tidiness with callouts,
6017    in case they inspect these fields. */
6018    
6019  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
6020    {    {
6021    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6022    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6023      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6024    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6025      md->offset_vector[0] = md->offset_vector[1] = -1;
6026    }    }
6027    
6028  /* 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 6056  if ((re->flags & PCRE_REQCHSET) != 0)
6056    }    }
6057    
6058    
6059    
6060    
6061  /* ==========================================================================*/  /* ==========================================================================*/
6062    
6063  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 5917  for(;;) Line 6068  for(;;)
6068    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6069    USPTR new_start_match;    USPTR new_start_match;
6070    
   /* 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;  
     }  
   
6071    /* 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
6072    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
6073    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 6256  for(;;)
6256    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6257    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6258    md->match_call_count = 0;    md->match_call_count = 0;
6259    rc = match(start_match, md->start_code, start_match, NULL, 2, md, ims, NULL,    md->match_function_type = 0;
6260      0, 0);    md->end_offset_top = 0;
6261      rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6262    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
6263    
6264    switch(rc)    switch(rc)
# Line 6241  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6384  if (rc == MATCH_MATCH || rc == MATCH_ACC
6384    
6385    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = md->offset_overflow? 0 : md->end_offset_top/2;
6386    
6387      /* If there is space in the offset vector, set any unused pairs at the end of
6388      the pattern to -1 for backwards compatibility. It is documented that this
6389      happens. In earlier versions, the whole set of potential capturing offsets
6390      was set to -1 each time round the loop, but this is handled differently now.
6391      "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only
6392      those at the end that need unsetting here. We can't just unset them all at
6393      the start of the whole thing because they may get set in one branch that is
6394      not the final matching branch. */
6395    
6396      if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL)
6397        {
6398        register int *iptr, *iend;
6399        int resetcount = 2 + re->top_bracket * 2;
6400        if (resetcount > offsetcount) resetcount = ocount;
6401        iptr = offsets + md->end_offset_top;
6402        iend = offsets + resetcount;
6403        while (iptr < iend) *iptr++ = -1;
6404        }
6405    
6406    /* 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
6407    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
6408    matching path. */    matching path. */

Legend:
Removed from v.597  
changed lines
  Added in v.654

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12