/[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 595 by ph10, Mon May 2 10:33:29 2011 UTC revision 630 by ph10, Fri Jul 22 10:00:10 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        {        {
199        int c, d;        int c, d;
200          if (eptr >= md->end_subject) return -1;
201        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
202        GETCHARINC(d, p);        GETCHARINC(d, p);
203        if (c != d && c != UCD_OTHERCASE(d)) return -1;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
# Line 204  if ((ims & PCRE_CASELESS) != 0) Line 209  if ((ims & PCRE_CASELESS) != 0)
209    
210    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
211    is no UCP support. */    is no UCP support. */
212        {
213    while (length-- > 0)      if (eptr + length > md->end_subject) return -1;
214      { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }      while (length-- > 0)
215          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
216        }
217    }    }
218    
219  /* In the caseful case, we can just compare the bytes, whether or not we  /* In the caseful case, we can just compare the bytes, whether or not we
220  are in UTF-8 mode. */  are in UTF-8 mode. */
221    
222  else  else
223    { while (length-- > 0) if (*p++ != *eptr++) return -1; }    {
224      if (eptr + length > md->end_subject) return -1;
225      while (length-- > 0) if (*p++ != *eptr++) return -1;
226      }
227    
228  return eptr - eptr_start;  return eptr - eptr_start;
229  }  }
# Line 267  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 277  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 289  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 303  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 313  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 351  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 374  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 461  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 479  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 492  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 511  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 527  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 551  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 589  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 634  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 670  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 700  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 726  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 737  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 752  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 760  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 779  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 829  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 863  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        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        RRETURN(MATCH_NOMATCH);
1001          eptrb, flags, RM2);        }
1002    
1003        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1004        as a non-capturing bracket. */
1005    
1006        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1007        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008    
1009        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1010    
1011        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1012        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1013    
1014        /* Non-capturing possessive bracket with unlimited repeat. We come here
1015        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1016        without the capturing complication. It is written out separately for speed
1017        and cleanliness. */
1018    
1019        case OP_BRAPOS:
1020        case OP_SBRAPOS:
1021        allow_zero = FALSE;
1022    
1023        POSSESSIVE_NON_CAPTURE:
1024        matched_once = FALSE;
1025        code_offset = ecode - md->start_code;
1026    
1027        for (;;)
1028          {
1029          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1030          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1031            eptrb, RM48);
1032          if (rrc == MATCH_KETRPOS)
1033            {
1034            offset_top = md->end_offset_top;
1035            eptr = md->end_match_ptr;
1036            ecode = md->start_code + code_offset;
1037            matched_once = TRUE;
1038            continue;
1039            }
1040        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
1041            (rrc != MATCH_THEN || md->start_match_ptr != ecode))            (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1042          RRETURN(rrc);          RRETURN(rrc);
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045        }        }
1046    
1047        if (matched_once || allow_zero)
1048          {
1049          ecode += 1 + LINK_SIZE;
1050          break;
1051          }
1052        RRETURN(MATCH_NOMATCH);
1053    
1054      /* Control never reaches here. */      /* Control never reaches here. */
1055    
1056      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1057      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1058      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1059      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1060    
1061      case OP_COND:      case OP_COND:
1062      case OP_SCOND:      case OP_SCOND:
1063      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1064    
1065      /* Because of the way auto-callout works during compile, a callout item is      /* Because of the way auto-callout works during compile, a callout item is
1066      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 1087  for (;;) Line 1235  for (;;)
1235        }        }
1236    
1237      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1238      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1239      assertion. */      an assertion. */
1240    
1241      else      else
1242        {        {
1243        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1244            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1245        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1246          {          {
1247            if (md->end_offset_top > offset_top)
1248              offset_top = md->end_offset_top;  /* Captures may have happened */
1249          condition = TRUE;          condition = TRUE;
1250          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1251          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
# Line 1113  for (;;) Line 1263  for (;;)
1263        }        }
1264    
1265      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one,
1266      we can use tail recursion to avoid using another stack frame, except when      we used to use tail recursion to avoid using another stack frame, except
1267      match_cbegroup is required for an unlimited repeat of a possibly empty      when there was unlimited repeat of a possibly empty group. However, that
1268      group. If the second alternative doesn't exist, we can just plough on. */      strategy no longer works because of the possibilty of (*THEN) being
1269        encountered in the branch. A recursive call to match() is always required,
1270        unless the second alternative doesn't exist, in which case we can just
1271        plough on. */
1272    
1273      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1274        {        {
1275        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1276        if (op == OP_SCOND)        /* Possibly empty group */        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1277          {        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1278          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          rrc = MATCH_NOMATCH;
1279          RRETURN(rrc);        RRETURN(rrc);
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
         goto TAIL_RECURSE;  
         }  
1280        }        }
1281      else                         /* Condition false & no alternative */      else                         /* Condition false & no alternative */
1282        {        {
# Line 1162  for (;;) Line 1309  for (;;)
1309      break;      break;
1310    
1311    
1312      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. */
     recursion, we should restore the offsets appropriately and continue from  
     after the call. */  
1313    
     case OP_ACCEPT:  
1314      case OP_END:      case OP_END:
1315      if (md->recursive != NULL && md->recursive->group_num == 0)      case OP_ACCEPT:
1316        {      case OP_ASSERT_ACCEPT:
       recursion_info *rec = md->recursive;  
       DPRINTF(("End of pattern in a (?0) recursion\n"));  
       md->recursive = rec->prevrec;  
       memmove(md->offset_vector, rec->offset_save,  
         rec->saved_max * sizeof(int));  
       offset_top = rec->save_offset_top;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
       }  
1317    
1318      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is      /* If we have matched an empty string, fail if not in an assertion and not
1319      set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1320      the subject. In both cases, backtracking will then try other alternatives,      is set and we have matched at the start of the subject. In both cases,
1321      if any. */      backtracking will then try other alternatives, if any. */
1322    
1323      if (eptr == mstart &&      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1324          (md->notempty ||           md->recursive == NULL &&
1325            (md->notempty_atstart &&           (md->notempty ||
1326              mstart == md->start_subject + md->start_offset)))             (md->notempty_atstart &&
1327                 mstart == md->start_subject + md->start_offset)))
1328        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1329    
1330      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
# Line 1204  for (;;) Line 1339  for (;;)
1339      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1340      MRRETURN(rrc);      MRRETURN(rrc);
1341    
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
   
1342      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1343      matching won't pass the KET for an assertion. If any one branch matches,      matching won't pass the KET for an assertion. If any one branch matches,
1344      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
1345      start of each branch to move the current point backwards, so the code at      start of each branch to move the current point backwards, so the code at
1346      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1347        of a condition, we want to return immediately afterwards. The caller of
1348        this incarnation of the match() function will have set MATCH_CONDASSERT in
1349        md->match_function type, and one of these opcodes will be the first opcode
1350        that is processed. We use a local variable that is preserved over calls to
1351        match() to remember this case. */
1352    
1353      case OP_ASSERT:      case OP_ASSERT:
1354      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1355        if (md->match_function_type == MATCH_CONDASSERT)
1356          {
1357          condassert = TRUE;
1358          md->match_function_type = 0;
1359          }
1360        else condassert = FALSE;
1361    
1362      do      do
1363        {        {
1364        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1365        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1366          {          {
1367          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1368            markptr = md->mark;
1369          break;          break;
1370          }          }
1371        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
# Line 1235  for (;;) Line 1374  for (;;)
1374        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1375        }        }
1376      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1377    
1378      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1379    
1380      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1381    
1382      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1383    
1384      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1385      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1255  for (;;) Line 1395  for (;;)
1395    
1396      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1397      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1398        if (md->match_function_type == MATCH_CONDASSERT)
1399          {
1400          condassert = TRUE;
1401          md->match_function_type = 0;
1402          }
1403        else condassert = FALSE;
1404    
1405      do      do
1406        {        {
1407        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
         RM5);  
1408        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1409        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1410          {          {
# Line 1272  for (;;) Line 1418  for (;;)
1418        }        }
1419      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1420    
1421      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1422    
1423      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1424      continue;      continue;
# Line 1340  for (;;) Line 1486  for (;;)
1486      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
1487      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1488    
1489      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1490      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
1491      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
1492      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
1493      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
1494      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
1495      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.  
1496    
1497      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
1498      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
1499      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1500        a lot, so he is not to blame for the current way it works. */
1501    
1502      case OP_RECURSE:      case OP_RECURSE:
1503        {        {
# Line 1366  for (;;) Line 1510  for (;;)
1510        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1511        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1512    
1513        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1514    
1515        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1516    
1517        /* Now save the offset data. */        /* Now save the offset data */
1518    
1519        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1520        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1382  for (;;) Line 1525  for (;;)
1525            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1526          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1527          }          }
   
1528        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1529              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_offset_top = offset_top;  
1530    
1531        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1532        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1533          might be changed, so reset it before looping. */
1534    
1535        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1536        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1537        do        do
1538          {          {
1539            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1540          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1541            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1542            memcpy(md->offset_vector, new_recursive.offset_save,
1543                new_recursive.saved_max * sizeof(int));
1544          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1545            {            {
1546            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1547            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1548            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1549              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1550            MRRETURN(MATCH_MATCH);  
1551              /* Set where we got to in the subject, and reset the start in case
1552              it was changed by \K. This *is* propagated back out of a recursion,
1553              for Perl compatibility. */
1554    
1555              eptr = md->end_match_ptr;
1556              mstart = md->start_match_ptr;
1557              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1558            }            }
1559          else if (rrc != MATCH_NOMATCH &&          else if (rrc != MATCH_NOMATCH &&
1560                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
# Line 1414  for (;;) Line 1566  for (;;)
1566            }            }
1567    
1568          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1569          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1570          }          }
1571        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1426  for (;;) Line 1576  for (;;)
1576          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1577        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1578        }        }
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer, but resetting  
     the start-of-match value in case it was changed by \K. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH)  /* Note: _not_ MATCH_ACCEPT */  
         {  
         mstart = md->start_match_ptr;  
         break;  
         }  
       if (rrc != MATCH_NOMATCH &&  
           (rrc != MATCH_THEN || md->start_match_ptr != ecode))  
         RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
   
     /* The repeating kets try the rest of the pattern or restart from the  
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
1579    
1580      if (*ecode == OP_KETRMIN)      RECURSION_MATCHED:
1581        {      break;
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
1582    
1583      /* 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
1584      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1523  for (;;) Line 1594  for (;;)
1594      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1595    
1596      case OP_BRAZERO:      case OP_BRAZERO:
1597        {      next = ecode + 1;
1598        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1599        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1600        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1601        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1602      break;      break;
1603    
1604      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1605        {      next = ecode + 1;
1606        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1607        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1608        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1609        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1610      break;      break;
1611    
1612      case OP_SKIPZERO:      case OP_SKIPZERO:
1613        {      next = ecode+1;
1614        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1615        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1616      break;      break;
1617    
1618        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1619        here; just jump to the group, with allow_zero set TRUE. */
1620    
1621        case OP_BRAPOSZERO:
1622        op = *(++ecode);
1623        allow_zero = TRUE;
1624        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1625          goto POSSESSIVE_NON_CAPTURE;
1626    
1627      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1628    
1629      case OP_KET:      case OP_KET:
1630      case OP_KETRMIN:      case OP_KETRMIN:
1631      case OP_KETRMAX:      case OP_KETRMAX:
1632        case OP_KETRPOS:
1633      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1634    
1635      /* 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
1636      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1637      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1638    
1639      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1640        {        {
1641        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1642        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1643        }        }
1644      else saved_eptr = NULL;      else saved_eptr = NULL;
1645    
1646      /* 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
1647      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
1648      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
1649      it was changed by \K. */      by \K. */
1650    
1651      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1652          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1653        {        {
1654        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1655        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1656        md->start_match_ptr = mstart;        md->start_match_ptr = mstart;
1657        MRRETURN(MATCH_MATCH);        MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1658        }        }
1659    
1660      /* 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
1661      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1662      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
1663      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
1664      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
1665        the current subject position and start match pointer and give a MATCH
1666        return. */
1667    
1668      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1669            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1670        {        {
1671        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1672        offset = number << 1;        offset = number << 1;
# Line 1599  for (;;) Line 1676  for (;;)
1676        printf("\n");        printf("\n");
1677  #endif  #endif
1678    
1679          /* Handle a recursively called group. */
1680    
1681          if (md->recursive != NULL && md->recursive->group_num == number)
1682            {
1683            md->end_match_ptr = eptr;
1684            md->start_match_ptr = mstart;
1685            RRETURN(MATCH_MATCH);
1686            }
1687    
1688          /* Deal with capturing */
1689    
1690        md->capture_last = number;        md->capture_last = number;
1691        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1692          {          {
1693            /* If offset is greater than offset_top, it means that we are
1694            "skipping" a capturing group, and that group's offsets must be marked
1695            unset. In earlier versions of PCRE, all the offsets were unset at the
1696            start of matching, but this doesn't work because atomic groups and
1697            assertions can cause a value to be set that should later be unset.
1698            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1699            part of the atomic group, but this is not on the final matching path,
1700            so must be unset when 2 is set. (If there is no group 2, there is no
1701            problem, because offset_top will then be 2, indicating no capture.) */
1702    
1703            if (offset > offset_top)
1704              {
1705              register int *iptr = md->offset_vector + offset_top;
1706              register int *iend = md->offset_vector + offset;
1707              while (iptr < iend) *iptr++ = -1;
1708              }
1709    
1710            /* Now make the extraction */
1711    
1712          md->offset_vector[offset] =          md->offset_vector[offset] =
1713            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1714          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1715          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1716          }          }
1717          }
1718    
1719        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1720        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1721        group. This is the forcible breaking of infinite loops as implemented in
1722        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1723        processing the rest of the pattern at a lower level. If this results in a
1724        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1725        bypassing intermediate backup points, but resetting any captures that
1726        happened along the way. */
1727    
1728        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1729          {
1730          if (*prev == OP_ONCE)
1731          {          {
1732          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1733          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1734          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1735          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;  
1736          }          }
1737          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1738          break;
1739        }        }
1740    
1741      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1742      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1743        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
1744    
1745      /* For a non-repeating ket, just continue at this level. This also      if (*ecode == OP_KETRPOS)
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
1746        {        {
1747        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1748        break;        md->end_offset_top = offset_top;
1749          RRETURN(MATCH_KETRPOS);
1750        }        }
1751    
1752      /* 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
1753      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
1754      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
1755      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
1756        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1757    
1758      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1759        {        {
1760        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1761        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1762        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1763            {
1764            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1765            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1766            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1767            RRETURN(MATCH_ONCE);
1768            }
1769          if (*prev >= OP_SBRA)    /* Could match an empty string */
1770          {          {
1771          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1772            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1773          RRETURN(rrc);          RRETURN(rrc);
1774          }          }
1775        ecode = prev;        ecode = prev;
# Line 1664  for (;;) Line 1777  for (;;)
1777        }        }
1778      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1779        {        {
1780        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1781          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1782          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1783        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1784          if (*prev == OP_ONCE)
1785            {
1786            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1787            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1788            md->once_target = prev;
1789            RRETURN(MATCH_ONCE);
1790            }
1791        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1792        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1793        }        }
1794      /* Control never gets here */      /* Control never gets here */
1795    
1796      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1797    
1798      case OP_CIRC:      case OP_CIRC:
1799      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 */  
1800    
1801      /* Start of subject assertion */      /* Start of subject assertion */
1802    
# Line 1693  for (;;) Line 1805  for (;;)
1805      ecode++;      ecode++;
1806      break;      break;
1807    
1808        /* Multiline mode: start of subject unless notbol, or after any newline. */
1809    
1810        case OP_CIRCM:
1811        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1812        if (eptr != md->start_subject &&
1813            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1814          MRRETURN(MATCH_NOMATCH);
1815        ecode++;
1816        break;
1817    
1818      /* Start of match assertion */      /* Start of match assertion */
1819    
1820      case OP_SOM:      case OP_SOM:
# Line 1707  for (;;) Line 1829  for (;;)
1829      ecode++;      ecode++;
1830      break;      break;
1831    
1832      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1833      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1834    
1835      case OP_DOLL:      case OP_DOLLM:
1836      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1837        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1838        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 */  
1839        {        {
1840        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1841        if (!md->endonly) goto ASSERT_NL_OR_EOS;        SCHECK_PARTIAL();
1842        }        }
1843        ecode++;
1844        break;
1845    
1846        /* Not multiline mode: assert before a terminating newline or before end of
1847        subject unless noteol is set. */
1848    
1849        case OP_DOLL:
1850        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1851        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1852    
1853      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1854    
# Line 2011  for (;;) Line 2132  for (;;)
2132      switch(c)      switch(c)
2133        {        {
2134        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2135    
2136        case 0x000d:        case 0x000d:
2137        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2138        break;        break;
# Line 2234  for (;;) Line 2356  for (;;)
2356        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2357        }        }
2358      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2359        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2360        while (eptr < md->end_subject)
2361        {        {
2362        int category = UCD_CATEGORY(c);        int len = 1;
2363        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2364        while (eptr < md->end_subject)        if (UCD_CATEGORY(c) != ucp_M) break;
2365          {        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;  
         }  
2366        }        }
2367      ecode++;      ecode++;
2368      break;      break;
# Line 2263  for (;;) Line 2378  for (;;)
2378      loops). */      loops). */
2379    
2380      case OP_REF:      case OP_REF:
2381        case OP_REFI:
2382        caseless = op == OP_REFI;
2383      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2384      ecode += 3;      ecode += 3;
2385    
# Line 2310  for (;;) Line 2427  for (;;)
2427        break;        break;
2428    
2429        default:               /* No repeat follows */        default:               /* No repeat follows */
2430        if ((length = match_ref(offset, eptr, length, md, ims)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2431          {          {
2432          CHECK_PARTIAL();          CHECK_PARTIAL();
2433          MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
# Line 2330  for (;;) Line 2447  for (;;)
2447    
2448      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2449        {        {
2450        int slength;        int slength;
2451        if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2452          {          {
2453          CHECK_PARTIAL();          CHECK_PARTIAL();
2454          MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
# Line 2350  for (;;) Line 2467  for (;;)
2467        {        {
2468        for (fi = min;; fi++)        for (fi = min;; fi++)
2469          {          {
2470          int slength;          int slength;
2471          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2472          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2473          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) MRRETURN(MATCH_NOMATCH);
2474          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2475            {            {
2476            CHECK_PARTIAL();            CHECK_PARTIAL();
2477            MRRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
# Line 2371  for (;;) Line 2488  for (;;)
2488        pp = eptr;        pp = eptr;
2489        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2490          {          {
2491          int slength;          int slength;
2492          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2493            {            {
2494            CHECK_PARTIAL();            CHECK_PARTIAL();
2495            break;            break;
# Line 2381  for (;;) Line 2498  for (;;)
2498          }          }
2499        while (eptr >= pp)        while (eptr >= pp)
2500          {          {
2501          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2502          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2503          eptr -= length;          eptr -= length;
2504          }          }
# Line 2491  for (;;) Line 2608  for (;;)
2608            {            {
2609            for (fi = min;; fi++)            for (fi = min;; fi++)
2610              {              {
2611              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2612              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2613              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2614              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2516  for (;;) Line 2633  for (;;)
2633            {            {
2634            for (fi = min;; fi++)            for (fi = min;; fi++)
2635              {              {
2636              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2637              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2638              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2639              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2562  for (;;) Line 2679  for (;;)
2679              }              }
2680            for (;;)            for (;;)
2681              {              {
2682              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2683              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2684              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2685              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2585  for (;;) Line 2702  for (;;)
2702              }              }
2703            while (eptr >= pp)            while (eptr >= pp)
2704              {              {
2705              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2706              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2707              eptr--;              eptr--;
2708              }              }
# Line 2661  for (;;) Line 2778  for (;;)
2778          {          {
2779          for (fi = min;; fi++)          for (fi = min;; fi++)
2780            {            {
2781            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2782            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2783            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2784            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2694  for (;;) Line 2811  for (;;)
2811            }            }
2812          for(;;)          for(;;)
2813            {            {
2814            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2815            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2816            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2817            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2739  for (;;) Line 2856  for (;;)
2856    
2857      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2858    
2859      case OP_CHARNC:      case OP_CHARI:
2860  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2861      if (utf8)      if (utf8)
2862        {        {
# Line 2799  for (;;) Line 2916  for (;;)
2916      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2917    
2918      case OP_EXACT:      case OP_EXACT:
2919        case OP_EXACTI:
2920      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2921      ecode += 3;      ecode += 3;
2922      goto REPEATCHAR;      goto REPEATCHAR;
2923    
2924      case OP_POSUPTO:      case OP_POSUPTO:
2925        case OP_POSUPTOI:
2926      possessive = TRUE;      possessive = TRUE;
2927      /* Fall through */      /* Fall through */
2928    
2929      case OP_UPTO:      case OP_UPTO:
2930        case OP_UPTOI:
2931      case OP_MINUPTO:      case OP_MINUPTO:
2932        case OP_MINUPTOI:
2933      min = 0;      min = 0;
2934      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2935      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2936      ecode += 3;      ecode += 3;
2937      goto REPEATCHAR;      goto REPEATCHAR;
2938    
2939      case OP_POSSTAR:      case OP_POSSTAR:
2940        case OP_POSSTARI:
2941      possessive = TRUE;      possessive = TRUE;
2942      min = 0;      min = 0;
2943      max = INT_MAX;      max = INT_MAX;
# Line 2823  for (;;) Line 2945  for (;;)
2945      goto REPEATCHAR;      goto REPEATCHAR;
2946    
2947      case OP_POSPLUS:      case OP_POSPLUS:
2948        case OP_POSPLUSI:
2949      possessive = TRUE;      possessive = TRUE;
2950      min = 1;      min = 1;
2951      max = INT_MAX;      max = INT_MAX;
# Line 2830  for (;;) Line 2953  for (;;)
2953      goto REPEATCHAR;      goto REPEATCHAR;
2954    
2955      case OP_POSQUERY:      case OP_POSQUERY:
2956        case OP_POSQUERYI:
2957      possessive = TRUE;      possessive = TRUE;
2958      min = 0;      min = 0;
2959      max = 1;      max = 1;
# Line 2837  for (;;) Line 2961  for (;;)
2961      goto REPEATCHAR;      goto REPEATCHAR;
2962    
2963      case OP_STAR:      case OP_STAR:
2964        case OP_STARI:
2965      case OP_MINSTAR:      case OP_MINSTAR:
2966        case OP_MINSTARI:
2967      case OP_PLUS:      case OP_PLUS:
2968        case OP_PLUSI:
2969      case OP_MINPLUS:      case OP_MINPLUS:
2970        case OP_MINPLUSI:
2971      case OP_QUERY:      case OP_QUERY:
2972        case OP_QUERYI:
2973      case OP_MINQUERY:      case OP_MINQUERY:
2974      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
2975        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
2976      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
2977      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2978      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2979      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2867  for (;;) Line 2996  for (;;)
2996          {          {
2997  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2998          unsigned int othercase;          unsigned int othercase;
2999          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3000              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3001            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3002          else oclength = 0;          else oclength = 0;
# Line 2895  for (;;) Line 3024  for (;;)
3024            {            {
3025            for (fi = min;; fi++)            for (fi = min;; fi++)
3026              {              {
3027              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3028              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3029              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3030              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2937  for (;;) Line 3066  for (;;)
3066    
3067            for(;;)            for(;;)
3068              {              {
3069              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3070              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3071              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3072  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2974  for (;;) Line 3103  for (;;)
3103      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3104        max, eptr));        max, eptr));
3105    
3106      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3107        {        {
3108        fc = md->lcc[fc];        fc = md->lcc[fc];
3109        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2991  for (;;) Line 3120  for (;;)
3120          {          {
3121          for (fi = min;; fi++)          for (fi = min;; fi++)
3122            {            {
3123            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3124            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3125            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3126            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3021  for (;;) Line 3150  for (;;)
3150    
3151          while (eptr >= pp)          while (eptr >= pp)
3152            {            {
3153            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3154            eptr--;            eptr--;
3155            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3156            }            }
# Line 3050  for (;;) Line 3179  for (;;)
3179          {          {
3180          for (fi = min;; fi++)          for (fi = min;; fi++)
3181            {            {
3182            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3183            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3184            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3185            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3079  for (;;) Line 3208  for (;;)
3208    
3209          while (eptr >= pp)          while (eptr >= pp)
3210            {            {
3211            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3212            eptr--;            eptr--;
3213            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3214            }            }
# Line 3092  for (;;) Line 3221  for (;;)
3221      checking can be multibyte. */      checking can be multibyte. */
3222    
3223      case OP_NOT:      case OP_NOT:
3224        case OP_NOTI:
3225      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3226        {        {
3227        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 3099  for (;;) Line 3229  for (;;)
3229        }        }
3230      ecode++;      ecode++;
3231      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3232      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3233        {        {
3234  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3235        if (c < 256)        if (c < 256)
# Line 3107  for (;;) Line 3237  for (;;)
3237        c = md->lcc[c];        c = md->lcc[c];
3238        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3239        }        }
3240      else      else    /* Caseful */
3241        {        {
3242        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3243        }        }
# Line 3121  for (;;) Line 3251  for (;;)
3251      about... */      about... */
3252    
3253      case OP_NOTEXACT:      case OP_NOTEXACT:
3254        case OP_NOTEXACTI:
3255      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3256      ecode += 3;      ecode += 3;
3257      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3258    
3259      case OP_NOTUPTO:      case OP_NOTUPTO:
3260        case OP_NOTUPTOI:
3261      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3262        case OP_NOTMINUPTOI:
3263      min = 0;      min = 0;
3264      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3265      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3266      ecode += 3;      ecode += 3;
3267      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3268    
3269      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3270        case OP_NOTPOSSTARI:
3271      possessive = TRUE;      possessive = TRUE;
3272      min = 0;      min = 0;
3273      max = INT_MAX;      max = INT_MAX;
# Line 3141  for (;;) Line 3275  for (;;)
3275      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3276    
3277      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3278        case OP_NOTPOSPLUSI:
3279      possessive = TRUE;      possessive = TRUE;
3280      min = 1;      min = 1;
3281      max = INT_MAX;      max = INT_MAX;
# Line 3148  for (;;) Line 3283  for (;;)
3283      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3284    
3285      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3286        case OP_NOTPOSQUERYI:
3287      possessive = TRUE;      possessive = TRUE;
3288      min = 0;      min = 0;
3289      max = 1;      max = 1;
# Line 3155  for (;;) Line 3291  for (;;)
3291      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3292    
3293      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3294        case OP_NOTPOSUPTOI:
3295      possessive = TRUE;      possessive = TRUE;
3296      min = 0;      min = 0;
3297      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 3162  for (;;) Line 3299  for (;;)
3299      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3300    
3301      case OP_NOTSTAR:      case OP_NOTSTAR:
3302        case OP_NOTSTARI:
3303      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3304        case OP_NOTMINSTARI:
3305      case OP_NOTPLUS:      case OP_NOTPLUS:
3306        case OP_NOTPLUSI:
3307      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3308        case OP_NOTMINPLUSI:
3309      case OP_NOTQUERY:      case OP_NOTQUERY:
3310        case OP_NOTQUERYI:
3311      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3312      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3313        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3314      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3315      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3316      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3189  for (;;) Line 3332  for (;;)
3332      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,
3333        max, eptr));        max, eptr));
3334    
3335      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3336        {        {
3337        fc = md->lcc[fc];        fc = md->lcc[fc];
3338    
# Line 3237  for (;;) Line 3380  for (;;)
3380            register unsigned int d;            register unsigned int d;
3381            for (fi = min;; fi++)            for (fi = min;; fi++)
3382              {              {
3383              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3384              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3385              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3386              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3256  for (;;) Line 3399  for (;;)
3399            {            {
3400            for (fi = min;; fi++)            for (fi = min;; fi++)
3401              {              {
3402              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3403              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3404              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3405              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3297  for (;;) Line 3440  for (;;)
3440          if (possessive) continue;          if (possessive) continue;
3441          for(;;)          for(;;)
3442              {              {
3443              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3444              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3445              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3446              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3320  for (;;) Line 3463  for (;;)
3463            if (possessive) continue;            if (possessive) continue;
3464            while (eptr >= pp)            while (eptr >= pp)
3465              {              {
3466              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3467              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3468              eptr--;              eptr--;
3469              }              }
# Line 3377  for (;;) Line 3520  for (;;)
3520            register unsigned int d;            register unsigned int d;
3521            for (fi = min;; fi++)            for (fi = min;; fi++)
3522              {              {
3523              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3524              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3525              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3526              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3395  for (;;) Line 3538  for (;;)
3538            {            {
3539            for (fi = min;; fi++)            for (fi = min;; fi++)
3540              {              {
3541              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3542              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3543              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3544              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3435  for (;;) Line 3578  for (;;)
3578            if (possessive) continue;            if (possessive) continue;
3579            for(;;)            for(;;)
3580              {              {
3581              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3582              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3583              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3584              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3458  for (;;) Line 3601  for (;;)
3601            if (possessive) continue;            if (possessive) continue;
3602            while (eptr >= pp)            while (eptr >= pp)
3603              {              {
3604              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3605              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3606              eptr--;              eptr--;
3607              }              }
# Line 3573  for (;;) Line 3716  for (;;)
3716            case PT_LAMP:            case PT_LAMP:
3717            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3718              {              {
3719                int chartype;
3720              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3721                {                {
3722                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3723                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3724                }                }
3725              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3726              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
3727              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
3728                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
3729                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
3730                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3731              }              }
3732            break;            break;
# Line 3596  for (;;) Line 3740  for (;;)
3740                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3741                }                }
3742              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3743              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
3744                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3745              }              }
3746            break;            break;
# Line 3611  for (;;) Line 3754  for (;;)
3754                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3755                }                }
3756              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3757              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
3758                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3759              }              }
3760            break;            break;
# Line 3626  for (;;) Line 3768  for (;;)
3768                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3769                }                }
3770              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3771              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
3772                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3773              }              }
3774            break;            break;
# Line 3635  for (;;) Line 3776  for (;;)
3776            case PT_ALNUM:            case PT_ALNUM:
3777            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3778              {              {
3779                int category;
3780              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3781                {                {
3782                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3783                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3784                }                }
3785              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3786              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3787              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
3788                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3789              }              }
3790            break;            break;
# Line 3657  for (;;) Line 3798  for (;;)
3798                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3799                }                }
3800              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3801              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 ||  
3802                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
3803                     == prop_fail_result)                     == prop_fail_result)
3804                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3674  for (;;) Line 3814  for (;;)
3814                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3815                }                }
3816              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3817              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 ||  
3818                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3819                     == prop_fail_result)                     == prop_fail_result)
3820                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 3685  for (;;) Line 3824  for (;;)
3824            case PT_WORD:            case PT_WORD:
3825            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3826              {              {
3827                int category;
3828              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
3829                {                {
3830                SCHECK_PARTIAL();                SCHECK_PARTIAL();
3831                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3832                }                }
3833              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3834              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
3835              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
                  c == CHAR_UNDERSCORE)  
3836                     == prop_fail_result)                     == prop_fail_result)
3837                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3838              }              }
# Line 3719  for (;;) Line 3858  for (;;)
3858              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
3859              }              }
3860            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3861            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
3862            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3863              {              {
3864              int len = 1;              int len = 1;
3865              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
3866                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
3867              eptr += len;              eptr += len;
3868              }              }
3869            }            }
# Line 3785  for (;;) Line 3921  for (;;)
3921            switch(c)            switch(c)
3922              {              {
3923              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3924    
3925              case 0x000d:              case 0x000d:
3926              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3927              break;              break;
# Line 4061  for (;;) Line 4198  for (;;)
4198            switch(*eptr++)            switch(*eptr++)
4199              {              {
4200              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4201    
4202              case 0x000d:              case 0x000d:
4203              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4204              break;              break;
4205    
4206              case 0x000a:              case 0x000a:
4207              break;              break;
4208    
# Line 4253  for (;;) Line 4392  for (;;)
4392            case PT_ANY:            case PT_ANY:
4393            for (fi = min;; fi++)            for (fi = min;; fi++)
4394              {              {
4395              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4396              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4397              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4398              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4269  for (;;) Line 4408  for (;;)
4408            case PT_LAMP:            case PT_LAMP:
4409            for (fi = min;; fi++)            for (fi = min;; fi++)
4410              {              {
4411              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              int chartype;
4412                RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
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 4278  for (;;) Line 4418  for (;;)
4418                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4419                }                }
4420              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4421              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4422              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4423                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4424                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4425                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4426              }              }
4427            /* Control never gets here */            /* Control never gets here */
# Line 4289  for (;;) Line 4429  for (;;)
4429            case PT_GC:            case PT_GC:
4430            for (fi = min;; fi++)            for (fi = min;; fi++)
4431              {              {
4432              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4433              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4434              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4435              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4298  for (;;) Line 4438  for (;;)
4438                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4439                }                }
4440              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4441              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
             if ((prop_category == prop_value) == prop_fail_result)  
4442                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4443              }              }
4444            /* Control never gets here */            /* Control never gets here */
# Line 4307  for (;;) Line 4446  for (;;)
4446            case PT_PC:            case PT_PC:
4447            for (fi = min;; fi++)            for (fi = min;; fi++)
4448              {              {
4449              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
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 4316  for (;;) Line 4455  for (;;)
4455                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4456                }                }
4457              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4458              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
             if ((prop_chartype == prop_value) == prop_fail_result)  
4459                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4460              }              }
4461            /* Control never gets here */            /* Control never gets here */
# Line 4325  for (;;) Line 4463  for (;;)
4463            case PT_SC:            case PT_SC:
4464            for (fi = min;; fi++)            for (fi = min;; fi++)
4465              {              {
4466              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
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 4334  for (;;) Line 4472  for (;;)
4472                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4473                }                }
4474              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4475              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
             if ((prop_script == prop_value) == prop_fail_result)  
4476                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4477              }              }
4478            /* Control never gets here */            /* Control never gets here */
# Line 4343  for (;;) Line 4480  for (;;)
4480            case PT_ALNUM:            case PT_ALNUM:
4481            for (fi = min;; fi++)            for (fi = min;; fi++)
4482              {              {
4483              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59);              int category;
4484                RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4485              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4486              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4487              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4352  for (;;) Line 4490  for (;;)
4490                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4491                }                }
4492              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4493              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4494              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                    == prop_fail_result)  
4495                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4496              }              }
4497            /* Control never gets here */            /* Control never gets here */
# Line 4362  for (;;) Line 4499  for (;;)
4499            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4500            for (fi = min;; fi++)            for (fi = min;; fi++)
4501              {              {
4502              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4503              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4504              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4505              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4371  for (;;) Line 4508  for (;;)
4508                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4509                }                }
4510              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4511              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 ||  
4512                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
4513                     == prop_fail_result)                     == prop_fail_result)
4514                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4382  for (;;) Line 4518  for (;;)
4518            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4519            for (fi = min;; fi++)            for (fi = min;; fi++)
4520              {              {
4521              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4522              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4523              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4524              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4391  for (;;) Line 4527  for (;;)
4527                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4528                }                }
4529              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4530              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 ||  
4531                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
4532                     == prop_fail_result)                     == prop_fail_result)
4533                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4402  for (;;) Line 4537  for (;;)
4537            case PT_WORD:            case PT_WORD:
4538            for (fi = min;; fi++)            for (fi = min;; fi++)
4539              {              {
4540              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62);              int category;
4541                RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4542              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4543              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4544              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4411  for (;;) Line 4547  for (;;)
4547                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
4548                }                }
4549              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4550              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4551              if ((prop_category == ucp_L ||              if ((category == ucp_L ||
4552                   prop_category == ucp_N ||                   category == ucp_N ||
4553                   c == CHAR_UNDERSCORE)                   c == CHAR_UNDERSCORE)
4554                     == prop_fail_result)                     == prop_fail_result)
4555                MRRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
# Line 4434  for (;;) Line 4570  for (;;)
4570          {          {
4571          for (fi = min;; fi++)          for (fi = min;; fi++)
4572            {            {
4573            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4574            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4575            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4576            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4443  for (;;) Line 4579  for (;;)
4579              MRRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
4580              }              }
4581            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
4582            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH);  
4583            while (eptr < md->end_subject)            while (eptr < md->end_subject)
4584              {              {
4585              int len = 1;              int len = 1;
4586              if (!utf8) c = *eptr;              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
4587                else { GETCHARLEN(c, eptr, len); }              if (UCD_CATEGORY(c) != ucp_M) break;
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
4588              eptr += len;              eptr += len;
4589              }              }
4590            }            }
4591          }          }
   
4592        else        else
4593  #endif     /* SUPPORT_UCP */  #endif     /* SUPPORT_UCP */
4594    
# Line 4466  for (;;) Line 4598  for (;;)
4598          {          {
4599          for (fi = min;; fi++)          for (fi = min;; fi++)
4600            {            {
4601            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4602            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4603            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4604            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4629  for (;;) Line 4761  for (;;)
4761          {          {
4762          for (fi = min;; fi++)          for (fi = min;; fi++)
4763            {            {
4764            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4765            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4766            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4767            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4777  for (;;) Line 4909  for (;;)
4909            case PT_LAMP:            case PT_LAMP:
4910            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4911              {              {
4912                int chartype;
4913              int len = 1;              int len = 1;
4914              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4915                {                {
# Line 4784  for (;;) Line 4917  for (;;)
4917                break;                break;
4918                }                }
4919              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4920              prop_chartype = UCD_CHARTYPE(c);              chartype = UCD_CHARTYPE(c);
4921              if ((prop_chartype == ucp_Lu ||              if ((chartype == ucp_Lu ||
4922                   prop_chartype == ucp_Ll ||                   chartype == ucp_Ll ||
4923                   prop_chartype == ucp_Lt) == prop_fail_result)                   chartype == ucp_Lt) == prop_fail_result)
4924                break;                break;
4925              eptr+= len;              eptr+= len;
4926              }              }
# Line 4803  for (;;) Line 4936  for (;;)
4936                break;                break;
4937                }                }
4938              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4939              prop_category = UCD_CATEGORY(c);              if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break;
             if ((prop_category == prop_value) == prop_fail_result)  
               break;  
4940              eptr+= len;              eptr+= len;
4941              }              }
4942            break;            break;
# Line 4820  for (;;) Line 4951  for (;;)
4951                break;                break;
4952                }                }
4953              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4954              prop_chartype = UCD_CHARTYPE(c);              if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break;
             if ((prop_chartype == prop_value) == prop_fail_result)  
               break;  
4955              eptr+= len;              eptr+= len;
4956              }              }
4957            break;            break;
# Line 4837  for (;;) Line 4966  for (;;)
4966                break;                break;
4967                }                }
4968              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4969              prop_script = UCD_SCRIPT(c);              if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break;
             if ((prop_script == prop_value) == prop_fail_result)  
               break;  
4970              eptr+= len;              eptr+= len;
4971              }              }
4972            break;            break;
# Line 4847  for (;;) Line 4974  for (;;)
4974            case PT_ALNUM:            case PT_ALNUM:
4975            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4976              {              {
4977                int category;
4978              int len = 1;              int len = 1;
4979              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
4980                {                {
# Line 4854  for (;;) Line 4982  for (;;)
4982                break;                break;
4983                }                }
4984              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
4985              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
4986              if ((prop_category == ucp_L || prop_category == ucp_N)              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
                  == prop_fail_result)  
4987                break;                break;
4988              eptr+= len;              eptr+= len;
4989              }              }
# Line 4872  for (;;) Line 4999  for (;;)
4999                break;                break;
5000                }                }
5001              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5002              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 ||  
5003                   c == CHAR_FF || c == CHAR_CR)                   c == CHAR_FF || c == CHAR_CR)
5004                   == prop_fail_result)                   == prop_fail_result)
5005                break;                break;
# Line 4891  for (;;) Line 5017  for (;;)
5017                break;                break;
5018                }                }
5019              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5020              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 ||  
5021                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
5022                   == prop_fail_result)                   == prop_fail_result)
5023                break;                break;
# Line 4903  for (;;) Line 5028  for (;;)
5028            case PT_WORD:            case PT_WORD:
5029            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5030              {              {
5031                int category;
5032              int len = 1;              int len = 1;
5033              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
5034                {                {
# Line 4910  for (;;) Line 5036  for (;;)
5036                break;                break;
5037                }                }
5038              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5039              prop_category = UCD_CATEGORY(c);              category = UCD_CATEGORY(c);
5040              if ((prop_category == ucp_L || prop_category == ucp_N ||              if ((category == ucp_L || category == ucp_N ||
5041                   c == CHAR_UNDERSCORE) == prop_fail_result)                   c == CHAR_UNDERSCORE) == prop_fail_result)
5042                break;                break;
5043              eptr+= len;              eptr+= len;
# Line 4927  for (;;) Line 5053  for (;;)
5053          if (possessive) continue;          if (possessive) continue;
5054          for(;;)          for(;;)
5055            {            {
5056            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5057            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5058            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5059            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4941  for (;;) Line 5067  for (;;)
5067          {          {
5068          for (i = min; i < max; i++)          for (i = min; i < max; i++)
5069            {            {
5070              int len = 1;
5071            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
5072              {              {
5073              SCHECK_PARTIAL();              SCHECK_PARTIAL();
5074              break;              break;
5075              }              }
5076            GETCHARINCTEST(c, eptr);            if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5077            prop_category = UCD_CATEGORY(c);            if (UCD_CATEGORY(c) == ucp_M) break;
5078            if (prop_category == ucp_M) break;            eptr += len;
5079            while (eptr < md->end_subject)            while (eptr < md->end_subject)
5080              {              {
5081              int len = 1;              len = 1;
5082              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
5083                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = UCD_CATEGORY(c);  
             if (prop_category != ucp_M) break;  
5084              eptr += len;              eptr += len;
5085              }              }
5086            }            }
# Line 4968  for (;;) Line 5091  for (;;)
5091    
5092          for(;;)          for(;;)
5093            {            {
5094            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5095            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5096            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5097            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
5098              {              {
             int len = 1;  
5099              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
5100                {                {
5101                BACKCHAR(eptr);                BACKCHAR(eptr);
5102                GETCHARLEN(c, eptr, len);                GETCHAR(c, eptr);
5103                }                }
5104              prop_category = UCD_CATEGORY(c);              if (UCD_CATEGORY(c) != ucp_M) break;
             if (prop_category != ucp_M) break;  
5105              eptr--;              eptr--;
5106              }              }
5107            }            }
# Line 5252  for (;;) Line 5373  for (;;)
5373            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5374            }            }
5375    
5376          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5377            done (no backing up). Otherwise, match at this position; anything other
5378            than no match is immediately returned. For nomatch, back up one
5379            character, unless we are matching \R and the last thing matched was
5380            \r\n, in which case, back up two bytes. */
5381    
5382          if (possessive) continue;          if (possessive) continue;
5383          for(;;)          for(;;)
5384            {            {
5385            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5386            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5387            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5388            BACKCHAR(eptr);            BACKCHAR(eptr);
5389              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5390                  eptr[-1] == '\r') eptr--;
5391            }            }
5392          }          }
5393        else        else
# Line 5459  for (;;) Line 5586  for (;;)
5586            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5587            }            }
5588    
5589          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5590            done (no backing up). Otherwise, match at this position; anything other
5591            than no match is immediately returned. For nomatch, back up one
5592            character (byte), unless we are matching \R and the last thing matched
5593            was \r\n, in which case, back up two bytes. */
5594    
5595          if (possessive) continue;          if (possessive) continue;
5596          while (eptr >= pp)          while (eptr >= pp)
5597            {            {
5598            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5599            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5600              eptr--;
5601              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5602                  eptr[-1] == '\r') eptr--;
5603            }            }
5604          }          }
5605    
# Line 5505  switch (frame->Xwhere) Line 5638  switch (frame->Xwhere)
5638    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)
5639    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)
5640    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)
5641    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)
5642  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5643    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)
5644    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
# Line 5534  Undefine all the macros that were define Line 5667  Undefine all the macros that were define
5667  #undef ecode  #undef ecode
5668  #undef mstart  #undef mstart
5669  #undef offset_top  #undef offset_top
 #undef ims  
5670  #undef eptrb  #undef eptrb
5671  #undef flags  #undef flags
5672    
# Line 5552  Undefine all the macros that were define Line 5684  Undefine all the macros that were define
5684  #undef condition  #undef condition
5685  #undef prev_is_word  #undef prev_is_word
5686    
 #undef original_ims  
   
5687  #undef ctype  #undef ctype
5688  #undef length  #undef length
5689  #undef max  #undef max
# Line 5610  pcre_exec(const pcre *argument_re, const Line 5740  pcre_exec(const pcre *argument_re, const
5740    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5741    int offsetcount)    int offsetcount)
5742  {  {
5743  int rc, resetcount, ocount;  int rc, ocount;
5744  int first_byte = -1;  int first_byte = -1;
5745  int req_byte = -1;  int req_byte = -1;
5746  int req_byte2 = -1;  int req_byte2 = -1;
5747  int newline;  int newline;
 unsigned long int ims;  
5748  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5749  BOOL anchored;  BOOL anchored;
5750  BOOL startline;  BOOL startline;
# Line 5719  utf8 = md->utf8 = (re->options & PCRE_UT Line 5848  utf8 = md->utf8 = (re->options & PCRE_UT
5848  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
5849  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
5850    
5851    /* Some options are unpacked into BOOL variables in the hope that testing
5852    them will be faster than individual option bits. */
5853    
5854  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
5855  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
5856  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
5857  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
5858  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5859                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5860    
5861    
5862  md->hitend = FALSE;  md->hitend = FALSE;
5863  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
5864    
# Line 5806  defined (though never set). So there's n Line 5940  defined (though never set). So there's n
5940  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
5941    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
5942    
5943  /* 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
5944  back the character offset. */  code for an invalid string if a results vector is available. */
5945    
5946  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5947  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5948    {    {
5949    int tb;    int erroroffset;
5950    if ((tb = _pcre_valid_utf8((USPTR)subject, length)) >= 0)    int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
5951      return (tb == length && md->partial > 1)?    if (errorcode != 0)
       PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;  
   if (start_offset > 0 && start_offset < length)  
5952      {      {
5953      tb = ((USPTR)subject)[start_offset] & 0xc0;      if (offsetcount >= 2)
5954      if (tb == 0x80) return PCRE_ERROR_BADUTF8_OFFSET;        {
5955          offsets[0] = erroroffset;
5956          offsets[1] = errorcode;
5957          }
5958        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5959          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5960      }      }
   }  
 #endif  
5961    
5962  /* 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. */  
5963    
5964  ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL);    if (start_offset > 0 && start_offset < length &&
5965          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
5966        return PCRE_ERROR_BADUTF8_OFFSET;
5967      }
5968    #endif
5969    
5970  /* 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
5971  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 5852  md->offset_max = (2*ocount)/3; Line 5989  md->offset_max = (2*ocount)/3;
5989  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
5990  md->capture_last = -1;  md->capture_last = -1;
5991    
 /* 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;  
   
5992  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
5993  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
5994  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
5995    offsets for the matched string. This is really just for tidiness with callouts,
5996    in case they inspect these fields. */
5997    
5998  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
5999    {    {
6000    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6001    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6002      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6003    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6004      md->offset_vector[0] = md->offset_vector[1] = -1;
6005    }    }
6006    
6007  /* 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 5901  if ((re->flags & PCRE_REQCHSET) != 0) Line 6035  if ((re->flags & PCRE_REQCHSET) != 0)
6035    }    }
6036    
6037    
6038    
6039    
6040  /* ==========================================================================*/  /* ==========================================================================*/
6041    
6042  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 5911  for(;;) Line 6047  for(;;)
6047    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6048    USPTR new_start_match;    USPTR new_start_match;
6049    
   /* 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;  
     }  
   
6050    /* 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
6051    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
6052    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 6108  for(;;) Line 6235  for(;;)
6235    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6236    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6237    md->match_call_count = 0;    md->match_call_count = 0;
6238    rc = match(start_match, md->start_code, start_match, NULL, 2, md, ims, NULL,    md->match_function_type = 0;
6239      0, 0);    md->end_offset_top = 0;
6240      rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6241    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
6242    
6243    switch(rc)    switch(rc)
# Line 6235  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6363  if (rc == MATCH_MATCH || rc == MATCH_ACC
6363    
6364    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = md->offset_overflow? 0 : md->end_offset_top/2;
6365    
6366      /* If there is space in the offset vector, set any unused pairs at the end of
6367      the pattern to -1 for backwards compatibility. It is documented that this
6368      happens. In earlier versions, the whole set of potential capturing offsets
6369      was set to -1 each time round the loop, but this is handled differently now.
6370      "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only
6371      those at the end that need unsetting here. We can't just unset them all at
6372      the start of the whole thing because they may get set in one branch that is
6373      not the final matching branch. */
6374    
6375      if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL)
6376        {
6377        register int *iptr, *iend;
6378        int resetcount = 2 + re->top_bracket * 2;
6379        if (resetcount > offsetcount) resetcount = ocount;
6380        iptr = offsets + md->end_offset_top;
6381        iend = offsets + resetcount;
6382        while (iptr < iend) *iptr++ = -1;
6383        }
6384    
6385    /* 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
6386    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
6387    matching path. */    matching path. */

Legend:
Removed from v.595  
changed lines
  Added in v.630

  ViewVC Help
Powered by ViewVC 1.1.5