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

Diff of /code/trunk/pcre_exec.c

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

revision 597 by ph10, Mon May 2 17:08:52 2011 UTC revision 618 by ph10, Sat Jul 16 17:24:16 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 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
# Line 273  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 277  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
277         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
278         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
279         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280         RM61,  RM62 };         RM61,  RM62, RM63, RM64, RM65, RM66 };
281    
282  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
283  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
# Line 283  actually used in this definition. */ Line 287  actually used in this definition. */
287  #define REGISTER register  #define REGISTER register
288    
289  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
290  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
291    { \    { \
292    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
293    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
294    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
295    }    }
296  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 295  actually used in this definition. */ Line 299  actually used in this definition. */
299    return ra; \    return ra; \
300    }    }
301  #else  #else
302  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
303    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
304  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
305  #endif  #endif
306    
# Line 309  argument of match(), which never changes Line 313  argument of match(), which never changes
313    
314  #define REGISTER  #define REGISTER
315    
316  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
317    {\    {\
318    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
319    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
# Line 319  argument of match(), which never changes Line 323  argument of match(), which never changes
323    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
324    newframe->Xmarkptr = markptr;\    newframe->Xmarkptr = markptr;\
325    newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
326    newframe->Xims = re;\    newframe->Xeptrb = re;\
   newframe->Xeptrb = rf;\  
   newframe->Xflags = rg;\  
327    newframe->Xrdepth = frame->Xrdepth + 1;\    newframe->Xrdepth = frame->Xrdepth + 1;\
328    newframe->Xprevframe = frame;\    newframe->Xprevframe = frame;\
329    frame = newframe;\    frame = newframe;\
# Line 357  typedef struct heapframe { Line 359  typedef struct heapframe {
359    USPTR Xmstart;    USPTR Xmstart;
360    USPTR Xmarkptr;    USPTR Xmarkptr;
361    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
362    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
363    unsigned int Xrdepth;    unsigned int Xrdepth;
364    
365    /* Function local variables */    /* Function local variables */
# Line 380  typedef struct heapframe { Line 380  typedef struct heapframe {
380    BOOL Xcondition;    BOOL Xcondition;
381    BOOL Xprev_is_word;    BOOL Xprev_is_word;
382    
   unsigned long int Xoriginal_ims;  
   
383  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
# Line 467  Arguments: Line 465  Arguments:
465     markptr     pointer to the most recent MARK name, or NULL     markptr     pointer to the most recent MARK name, or NULL
466     offset_top  current top pointer     offset_top  current top pointer
467     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
468     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
469                   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  
470     rdepth      the recursion depth     rdepth      the recursion depth
471    
472  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
# Line 485  Returns:       MATCH_MATCH if matched Line 478  Returns:       MATCH_MATCH if matched
478    
479  static int  static int
480  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
481    const uschar *markptr, int offset_top, match_data *md, unsigned long int ims,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
482    eptrblock *eptrb, int flags, unsigned int rdepth)    unsigned int rdepth)
483  {  {
484  /* 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,
485  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 498  register unsigned int c;   /* Character Line 491  register unsigned int c;   /* Character
491  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
492    
493  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
494    BOOL caseless;
495  int condcode;  int condcode;
496    
497  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
# Line 517  frame->Xecode = ecode; Line 511  frame->Xecode = ecode;
511  frame->Xmstart = mstart;  frame->Xmstart = mstart;
512  frame->Xmarkptr = markptr;  frame->Xmarkptr = markptr;
513  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
514  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
515  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
516    
517  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 533  HEAP_RECURSE: Line 525  HEAP_RECURSE:
525  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
526  #define markptr            frame->Xmarkptr  #define markptr            frame->Xmarkptr
527  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
528  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
529  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
530    
531  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 557  HEAP_RECURSE: Line 547  HEAP_RECURSE:
547  #define condition          frame->Xcondition  #define condition          frame->Xcondition
548  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
549    
 #define original_ims       frame->Xoriginal_ims  
   
550  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
551  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
552  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
# Line 595  i, and fc and c, can be the same variabl Line 583  i, and fc and c, can be the same variabl
583  #define fi i  #define fi i
584  #define fc c  #define fc c
585    
586    /* Many of the following variables are used only in small blocks of the code.
587  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  My normal style of coding would have declared them within each of those blocks.
588  const uschar *charptr;             /* in small blocks of the code. My normal */  However, in order to accommodate the version of this code that uses an external
589  #endif                             /* style of coding would have declared    */  "stack" implemented on the heap, it is easier to declare them all here, so the
590  const uschar *callpat;             /* them within each of those blocks.      */  declarations can be cut out in a block. The only declarations within blocks
591  const uschar *data;                /* However, in order to accommodate the   */  below are for variables that do not have to be preserved over a recursive call
592  const uschar *next;                /* version of this code that uses an      */  to RMATCH(). */
593  USPTR         pp;                  /* external "stack" implemented on the    */  
594  const uschar *prev;                /* heap, it is easier to declare them all */  #ifdef SUPPORT_UTF8
595  USPTR         saved_eptr;          /* here, so the declarations can be cut   */  const uschar *charptr;
596                                     /* out in a block. The only declarations  */  #endif
597  recursion_info new_recursive;      /* within blocks below are for variables  */  const uschar *callpat;
598                                     /* that do not have to be preserved over  */  const uschar *data;
599  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  const uschar *next;
600    USPTR         pp;
601    const uschar *prev;
602    USPTR         saved_eptr;
603    
604    recursion_info new_recursive;
605    
606    BOOL cur_is_word;
607  BOOL condition;  BOOL condition;
608  BOOL prev_is_word;  BOOL prev_is_word;
609    
 unsigned long int original_ims;  
   
610  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
611  int prop_type;  int prop_type;
612  int prop_value;  int prop_value;
# Line 640  int stacksave[REC_STACK_SAVE_MAX]; Line 633  int stacksave[REC_STACK_SAVE_MAX];
633  eptrblock newptrb;  eptrblock newptrb;
634  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
635    
636    /* To save space on the stack and in the heap frame, I have doubled up on some
637    of the local variables that are used only in localised parts of the code, but
638    still need to be preserved over recursive calls of match(). These macros define
639    the alternative names that are used. */
640    
641    #define allow_zero    cur_is_word
642    #define cbegroup      condition
643    #define code_offset   codelink
644    #define condassert    condition
645    #define matched_once  prev_is_word
646    
647  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
648  variables. */  variables. */
649    
# Line 676  haven't exceeded the recursive call limi Line 680  haven't exceeded the recursive call limi
680  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
681  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
682    
 original_ims = ims;    /* Save for resetting on ')' */  
   
683  /* 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
684  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
685  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
686  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
687  When match() is called in other circumstances, don't add to the chain. The  
688  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
689  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
690  match(). */  to break infinite loops that match no characters. When match() is called in
691    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
692    NOT be used with tail recursion, because the memory block that is used is on
693    the stack, so a new one may be required for each match(). */
694    
695  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
696    {    {
697    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
698    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
699    eptrb = &newptrb;    eptrb = &newptrb;
700      md->match_function_type = 0;
701    }    }
702    
703  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 700  for (;;) Line 706  for (;;)
706    {    {
707    minimize = possessive = FALSE;    minimize = possessive = FALSE;
708    op = *ecode;    op = *ecode;
709    
710    switch(op)    switch(op)
711      {      {
712      case OP_MARK:      case OP_MARK:
713      markptr = ecode + 2;      markptr = ecode + 2;
714      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
715        ims, eptrb, flags, RM55);        eptrb, RM55);
716    
717      /* 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
718      argument, and we must check whether that argument matches this MARK's      argument, and we must check whether that argument matches this MARK's
# Line 732  for (;;) Line 738  for (;;)
738    
739      case OP_COMMIT:      case OP_COMMIT:
740      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
741        ims, eptrb, flags, RM52);        eptrb, RM52);
742      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
743          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&          rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
744          rrc != MATCH_THEN)          rrc != MATCH_THEN)
# Line 743  for (;;) Line 749  for (;;)
749    
750      case OP_PRUNE:      case OP_PRUNE:
751      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
752        ims, eptrb, flags, RM51);        eptrb, RM51);
753      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
754      MRRETURN(MATCH_PRUNE);      MRRETURN(MATCH_PRUNE);
755    
756      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
757      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
758        ims, eptrb, flags, RM56);        eptrb, RM56);
759      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
760      md->mark = ecode + 2;      md->mark = ecode + 2;
761      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
# Line 758  for (;;) Line 764  for (;;)
764    
765      case OP_SKIP:      case OP_SKIP:
766      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
767        ims, eptrb, flags, RM53);        eptrb, RM53);
768      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
769        RRETURN(rrc);        RRETURN(rrc);
770      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
# Line 766  for (;;) Line 772  for (;;)
772    
773      case OP_SKIP_ARG:      case OP_SKIP_ARG:
774      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
775        ims, eptrb, flags, RM57);        eptrb, RM57);
776      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
777        RRETURN(rrc);        RRETURN(rrc);
778    
# Line 785  for (;;) Line 791  for (;;)
791    
792      case OP_THEN:      case OP_THEN:
793      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
794        ims, eptrb, flags, RM54);        eptrb, RM54);
795      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
796      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode - GET(ecode, 1);
797      MRRETURN(MATCH_THEN);      MRRETURN(MATCH_THEN);
798    
799      case OP_THEN_ARG:      case OP_THEN_ARG:
800      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
801        offset_top, md, ims, eptrb, flags, RM58);        offset_top, md, eptrb, RM58);
802      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
803      md->start_match_ptr = ecode - GET(ecode, 1);      md->start_match_ptr = ecode - GET(ecode, 1);
804      md->mark = ecode + LINK_SIZE + 2;      md->mark = ecode + LINK_SIZE + 2;
805      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
806    
807      /* 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
808      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
809      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
810      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
811      reference inside the group.      previous iteration of this group, and be referred to by a reference inside
812        the group. A failure to match might occur after the group has succeeded,
813      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
814      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
815      of the same bracket.      were set by a previous iteration of the same bracket.
816    
817      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
818      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 816  for (;;) Line 822  for (;;)
822      case OP_SCBRA:      case OP_SCBRA:
823      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
824      offset = number << 1;      offset = number << 1;
825    
826  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
827      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
828      printf("subject=");      printf("subject=");
# Line 835  for (;;) Line 841  for (;;)
841        md->offset_vector[md->offset_end - number] =        md->offset_vector[md->offset_end - number] =
842          (int)(eptr - md->start_subject);          (int)(eptr - md->start_subject);
843    
844        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
845          {          {
846          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
847            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
848              eptrb, RM1);
849            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
850          if (rrc != MATCH_NOMATCH &&          if (rrc != MATCH_NOMATCH &&
851              (rrc != MATCH_THEN || md->start_match_ptr != ecode))              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
852            RRETURN(rrc);            RRETURN(rrc);
853          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
854          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
855            if (*ecode != OP_ALT) break;
856          }          }
       while (*ecode == OP_ALT);  
857    
858        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
859        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
860        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
861        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
862    
863          /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
864          MATCH_THEN. */
865    
866        if (rrc != MATCH_THEN) md->mark = markptr;        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
867        RRETURN(MATCH_NOMATCH);        RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
868        }        }
869    
870      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 869  for (;;) Line 878  for (;;)
878      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
879      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
880    
881      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
882      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
883      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
884      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
885      is set.*/      turning this into a tail recursion, except in the case of a possibly empty
886        group. However, now that there is the possiblity of (*THEN) occurring in
887        the final alternative, this optimization is no longer possible.
888    
889        MATCH_ONCE is returned when the end of an atomic group is successfully
890        reached, but subsequent matching fails. It passes back up the tree (causing
891        captured values to be reset) until the original atomic group level is
892        reached. This is tested by comparing md->once_target with the start of the
893        group. At this point, the return is converted into MATCH_NOMATCH so that
894        previous backup points can be taken. */
895    
896        case OP_ONCE:
897      case OP_BRA:      case OP_BRA:
898      case OP_SBRA:      case OP_SBRA:
899      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
900      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
901      for (;;)      for (;;)
902        {        {
903        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
904          {        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
905          if (flags == 0)    /* Not a possibly empty group */          RM2);
906          if (rrc != MATCH_NOMATCH &&
907              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
908            {
909            if (rrc == MATCH_ONCE)
910            {            {
911            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
912            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
913            goto TAIL_RECURSE;              {
914            }              while (*scode == OP_ALT) scode += GET(scode, 1);
915                scode -= GET(scode, 1);
916                }
917              if (md->once_target == scode) rrc = MATCH_NOMATCH;
918              }
919            RRETURN(rrc);
920            }
921          ecode += GET(ecode, 1);
922          if (*ecode != OP_ALT) break;
923          }
924        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
925        RRETURN(MATCH_NOMATCH);
926    
927          /* Possibly empty group; can't use tail recursion. */      /* Handle possessive capturing brackets with an unlimited repeat. We come
928        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
929        handled similarly to the normal case above. However, the matching is
930        different. The end of these brackets will always be OP_KETRPOS, which
931        returns MATCH_KETRPOS without going further in the pattern. By this means
932        we can handle the group by iteration rather than recursion, thereby
933        reducing the amount of stack needed. */
934    
935        case OP_CBRAPOS:
936        case OP_SCBRAPOS:
937        allow_zero = FALSE;
938    
939        POSSESSIVE_CAPTURE:
940        number = GET2(ecode, 1+LINK_SIZE);
941        offset = number << 1;
942    
943          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,  #ifdef PCRE_DEBUG
944            eptrb, flags, RM48);      printf("start possessive bracket %d\n", number);
945          if (rrc == MATCH_NOMATCH) md->mark = markptr;      printf("subject=");
946          RRETURN(rrc);      pchars(eptr, 16, TRUE, md);
947        printf("\n");
948    #endif
949    
950        if (offset < md->offset_max)
951          {
952          matched_once = FALSE;
953          code_offset = ecode - md->start_code;
954    
955          save_offset1 = md->offset_vector[offset];
956          save_offset2 = md->offset_vector[offset+1];
957          save_offset3 = md->offset_vector[md->offset_end - number];
958          save_capture_last = md->capture_last;
959    
960          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
961    
962          /* Each time round the loop, save the current subject position for use
963          when the group matches. For MATCH_MATCH, the group has matched, so we
964          restart it with a new subject starting position, remembering that we had
965          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
966          usual. If we haven't matched any alternatives in any iteration, check to
967          see if a previous iteration matched. If so, the group has matched;
968          continue from afterwards. Otherwise it has failed; restore the previous
969          capture values before returning NOMATCH. */
970    
971          for (;;)
972            {
973            md->offset_vector[md->offset_end - number] =
974              (int)(eptr - md->start_subject);
975            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
976            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
977              eptrb, RM63);
978            if (rrc == MATCH_KETRPOS)
979              {
980              offset_top = md->end_offset_top;
981              eptr = md->end_match_ptr;
982              ecode = md->start_code + code_offset;
983              save_capture_last = md->capture_last;
984              matched_once = TRUE;
985              continue;
986              }
987            if (rrc != MATCH_NOMATCH &&
988                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
989              RRETURN(rrc);
990            md->capture_last = save_capture_last;
991            ecode += GET(ecode, 1);
992            if (*ecode != OP_ALT) break;
993          }          }
994    
995        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
996        otherwise return. */          {
997            md->offset_vector[offset] = save_offset1;
998            md->offset_vector[offset+1] = save_offset2;
999            md->offset_vector[md->offset_end - number] = save_offset3;
1000            }
1001    
1002          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
1003          if (allow_zero || matched_once)
1004            {
1005            ecode += 1 + LINK_SIZE;
1006            break;
1007            }
1008    
1009          RRETURN(MATCH_NOMATCH);
1010          }
1011    
1012        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1013        as a non-capturing bracket. */
1014    
1015        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1016        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1017    
1018        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1019    
1020        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1021          eptrb, flags, RM2);      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1022    
1023        /* Non-capturing possessive bracket with unlimited repeat. We come here
1024        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1025        without the capturing complication. It is written out separately for speed
1026        and cleanliness. */
1027    
1028        case OP_BRAPOS:
1029        case OP_SBRAPOS:
1030        allow_zero = FALSE;
1031    
1032        POSSESSIVE_NON_CAPTURE:
1033        matched_once = FALSE;
1034        code_offset = ecode - md->start_code;
1035    
1036        for (;;)
1037          {
1038          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1039          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1040            eptrb, RM48);
1041          if (rrc == MATCH_KETRPOS)
1042            {
1043            offset_top = md->end_offset_top;
1044            eptr = md->end_match_ptr;
1045            ecode = md->start_code + code_offset;
1046            matched_once = TRUE;
1047            continue;
1048            }
1049        if (rrc != MATCH_NOMATCH &&        if (rrc != MATCH_NOMATCH &&
1050            (rrc != MATCH_THEN || md->start_match_ptr != ecode))            (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1051          RRETURN(rrc);          RRETURN(rrc);
1052        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1053          if (*ecode != OP_ALT) break;
1054        }        }
1055    
1056        if (matched_once || allow_zero)
1057          {
1058          ecode += 1 + LINK_SIZE;
1059          break;
1060          }
1061        RRETURN(MATCH_NOMATCH);
1062    
1063      /* Control never reaches here. */      /* Control never reaches here. */
1064    
1065      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1066      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
1067      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
1068      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. */  
1069    
1070      case OP_COND:      case OP_COND:
1071      case OP_SCOND:      case OP_SCOND:
1072      codelink= GET(ecode, 1);      codelink = GET(ecode, 1);
1073    
1074      /* 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
1075      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
# Line 1093  for (;;) Line 1244  for (;;)
1244        }        }
1245    
1246      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1247      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
1248      assertion. */      an assertion. */
1249    
1250      else      else
1251        {        {
1252        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1253            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1254        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1255          {          {
1256          condition = TRUE;          condition = TRUE;
# Line 1119  for (;;) Line 1270  for (;;)
1270        }        }
1271    
1272      /* 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,
1273      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
1274      match_cbegroup is required for an unlimited repeat of a possibly empty      when there was unlimited repeat of a possibly empty group. However, that
1275      group. If the second alternative doesn't exist, we can just plough on. */      strategy no longer works because of the possibilty of (*THEN) being
1276        encountered in the branch. A recursive call to match() is always required,
1277        unless the second alternative doesn't exist, in which case we can just
1278        plough on. */
1279    
1280      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1281        {        {
1282        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1283        if (op == OP_SCOND)        /* Possibly empty group */        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1284          {        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1285          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          rrc = MATCH_NOMATCH;
1286          RRETURN(rrc);        RRETURN(rrc);
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
         goto TAIL_RECURSE;  
         }  
1287        }        }
1288      else                         /* Condition false & no alternative */      else                         /* Condition false & no alternative */
1289        {        {
# Line 1168  for (;;) Line 1316  for (;;)
1316      break;      break;
1317    
1318    
1319      /* End of the pattern, either real or forced. If we are in a top-level      /* End of the pattern, either real or forced. If we are in a recursion, we
1320      recursion, we should restore the offsets appropriately and continue from      should restore the offsets appropriately, and if it's a top-level
1321      after the call. */      recursion, continue from after the call. */
1322    
1323      case OP_ACCEPT:      case OP_ACCEPT:
1324        case OP_ASSERT_ACCEPT:
1325      case OP_END:      case OP_END:
1326      if (md->recursive != NULL && md->recursive->group_num == 0)  
1327    /*
1328        if (md->recursive != NULL)
1329        {        {
1330        recursion_info *rec = md->recursive;        recursion_info *rec = md->recursive;
1331        DPRINTF(("End of pattern in a (?0) recursion\n"));  
1332        md->recursive = rec->prevrec;        md->recursive = rec->prevrec;
1333        memmove(md->offset_vector, rec->offset_save,  
1334          memmove(md->offset_vector, rec->offset_save,
1335          rec->saved_max * sizeof(int));          rec->saved_max * sizeof(int));
1336        offset_top = rec->save_offset_top;        offset_top = rec->save_offset_top;
1337        ims = original_ims;        if (rec->group_num == 0)
1338        ecode = rec->after_call;          {
1339        break;          ecode = rec->after_call;
1340            break;
1341            }
1342        }        }
1343    */
1344        /* Otherwise, if we have matched an empty string, fail if not in an
1345        assertion and if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1346        is set and we have matched at the start of the subject. In both cases,
1347        backtracking will then try other alternatives, if any. */
1348    
1349    /*    else */ if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1350    
1351      /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is           md->recursive == NULL &&
     set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of  
     the subject. In both cases, backtracking will then try other alternatives,  
     if any. */  
1352    
     if (eptr == mstart &&  
1353          (md->notempty ||          (md->notempty ||
1354            (md->notempty_atstart &&            (md->notempty_atstart &&
1355              mstart == md->start_subject + md->start_offset)))              mstart == md->start_subject + md->start_offset)))
1356        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1357    
1358      /* Otherwise, we have a match. */      /* Otherwise, we have a match. */
1359    
1360      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1361      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1362      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
# Line 1210  for (;;) Line 1367  for (;;)
1367      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1368      MRRETURN(rrc);      MRRETURN(rrc);
1369    
     /* Change option settings */  
   
     case OP_OPT:  
     ims = ecode[1];  
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
   
1370      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1371      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,
1372      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
1373      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
1374      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1375        of a condition, we want to return immediately afterwards. The caller of
1376        this incarnation of the match() function will have set MATCH_CONDASSERT in
1377        md->match_function type, and one of these opcodes will be the first opcode
1378        that is processed. We use a local variable that is preserved over calls to
1379        match() to remember this case. */
1380    
1381      case OP_ASSERT:      case OP_ASSERT:
1382      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1383        if (md->match_function_type == MATCH_CONDASSERT)
1384          {
1385          condassert = TRUE;
1386          md->match_function_type = 0;
1387          }
1388        else condassert = FALSE;
1389    
1390      do      do
1391        {        {
1392        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
         RM4);  
1393        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1394          {          {
1395          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
# Line 1241  for (;;) Line 1401  for (;;)
1401        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1402        }        }
1403      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1404    
1405      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1406    
1407      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1408    
1409      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1410    
1411      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1412      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 1261  for (;;) Line 1422  for (;;)
1422    
1423      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1424      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1425        if (md->match_function_type == MATCH_CONDASSERT)
1426          {
1427          condassert = TRUE;
1428          md->match_function_type = 0;
1429          }
1430        else condassert = FALSE;
1431    
1432      do      do
1433        {        {
1434        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
         RM5);  
1435        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1436        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1437          {          {
# Line 1278  for (;;) Line 1445  for (;;)
1445        }        }
1446      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1447    
1448      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1449    
1450      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1451      continue;      continue;
1452    
# Line 1345  for (;;) Line 1512  for (;;)
1512      /* Recursion either matches the current regex, or some subexpression. The      /* Recursion either matches the current regex, or some subexpression. The
1513      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
1514      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1515    
1516      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1517      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
1518      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
1519      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
1520      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
1521      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
1522      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.  
1523    
1524      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
1525      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
1526      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1527        a lot, so he is not to blame for the current way it works. */
1528    
1529      case OP_RECURSE:      case OP_RECURSE:
1530        {        {
# Line 1372  for (;;) Line 1537  for (;;)
1537        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1538        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1539    
1540        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1541    
1542        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1543    
1544        /* Now save the offset data. */        /* Now save the offset data */
1545    
1546        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1547        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1388  for (;;) Line 1552  for (;;)
1552            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1553          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1554          }          }
   
1555        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1556              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1557        new_recursive.save_offset_top = offset_top;  
1558          /* OK, now we can do the recursion. After processing each alternative,
1559        /* OK, now we can do the recursion. For each top-level alternative we        restore the offset data. If there were nested recursions, md->recursive
1560        restore the offset and recursion data. */        might be changed, so reset it before looping. */
1561    
1562        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1563        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1564        do        do
1565          {          {
1566            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1567          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1568            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1569            memcpy(md->offset_vector, new_recursive.offset_save,
1570                new_recursive.saved_max * sizeof(int));
1571          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1572            {            {
1573            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1574            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1575            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1576              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1577            MRRETURN(MATCH_MATCH);  
1578              /* Set where we got to in the subject, and reset the start in case
1579              it was changed by \K. This *is* propagated back out of a recursion,
1580              for Perl compatibility. */
1581    
1582              eptr = md->end_match_ptr;
1583              mstart = md->start_match_ptr;
1584              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1585            }            }
1586          else if (rrc != MATCH_NOMATCH &&          else if (rrc != MATCH_NOMATCH &&
1587                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
# Line 1420  for (;;) Line 1593  for (;;)
1593            }            }
1594    
1595          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1596          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1597          }          }
1598        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1432  for (;;) Line 1603  for (;;)
1603          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1604        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1605        }        }
1606      /* Control never reaches here */  
1607        RECURSION_MATCHED:
1608      /* "Once" brackets are like assertion brackets except that after a match,      break;
     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));  
       }  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
1609    
1610      /* 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
1611      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1527  for (;;) Line 1619  for (;;)
1619      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1620      with fixed upper repeat limits are compiled as a number of copies, with the      with fixed upper repeat limits are compiled as a number of copies, with the
1621      optional ones preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1622    
1623      case OP_BRAZERO:      case OP_BRAZERO:
1624        {      next = ecode + 1;
1625        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1626        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1627        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1628        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1629      break;      break;
1630    
1631      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1632        {      next = ecode + 1;
1633        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1634        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1635        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1636        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
       ecode++;  
       }  
1637      break;      break;
1638    
1639      case OP_SKIPZERO:      case OP_SKIPZERO:
1640        {      next = ecode+1;
1641        next = ecode+1;      do next += GET(next,1); while (*next == OP_ALT);
1642        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1643      break;      break;
1644    
1645        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1646        here; just jump to the group, with allow_zero set TRUE. */
1647    
1648        case OP_BRAPOSZERO:
1649        op = *(++ecode);
1650        allow_zero = TRUE;
1651        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1652          goto POSSESSIVE_NON_CAPTURE;
1653    
1654      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1655    
1656      case OP_KET:      case OP_KET:
1657      case OP_KETRMIN:      case OP_KETRMIN:
1658      case OP_KETRMAX:      case OP_KETRMAX:
1659        case OP_KETRPOS:
1660      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1661    
1662      /* 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
1663      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1664      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1665    
1666      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1667        {        {
1668        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1669        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1670        }        }
1671      else saved_eptr = NULL;      else saved_eptr = NULL;
1672    
1673      /* 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
1674      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
1675      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
1676      it was changed by \K. */      by \K. */
1677    
1678      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1679          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1680        {        {
1681        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1682        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
# Line 1591  for (;;) Line 1686  for (;;)
1686    
1687      /* 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
1688      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1689      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
1690      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
1691      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
1692        the current subject position and start match pointer and give a MATCH
1693        return. */
1694    
1695      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1696            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1697        {        {
1698        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1699        offset = number << 1;        offset = number << 1;
# Line 1605  for (;;) Line 1703  for (;;)
1703        printf("\n");        printf("\n");
1704  #endif  #endif
1705    
1706          /* Handle a recursively called group. */
1707    
1708          if (md->recursive != NULL && md->recursive->group_num == number)
1709            {
1710            md->end_match_ptr = eptr;
1711            md->start_match_ptr = mstart;
1712            RRETURN(MATCH_MATCH);
1713            }
1714    
1715          /* Deal with capturing */
1716    
1717        md->capture_last = number;        md->capture_last = number;
1718        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1719          {          {
1720            /* If offset is greater than offset_top, it means that we are
1721            "skipping" a capturing group, and that group's offsets must be marked
1722            unset. In earlier versions of PCRE, all the offsets were unset at the
1723            start of matching, but this doesn't work because atomic groups and
1724            assertions can cause a value to be set that should later be unset.
1725            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1726            part of the atomic group, but this is not on the final matching path,
1727            so must be unset when 2 is set. (If there is no group 2, there is no
1728            problem, because offset_top will then be 2, indicating no capture.) */
1729    
1730            if (offset > offset_top)
1731              {
1732              register int *iptr = md->offset_vector + offset_top;
1733              register int *iend = md->offset_vector + offset;
1734              while (iptr < iend) *iptr++ = -1;
1735              }
1736    
1737            /* Now make the extraction */
1738    
1739          md->offset_vector[offset] =          md->offset_vector[offset] =
1740            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1741          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1742          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1743          }          }
   
       /* Handle a recursively called group. Restore the offsets  
       appropriately and continue from after the call. */  
   
       if (md->recursive != NULL && md->recursive->group_num == number)  
         {  
         recursion_info *rec = md->recursive;  
         DPRINTF(("Recursion (%d) succeeded - continuing\n", number));  
         md->recursive = rec->prevrec;  
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         offset_top = rec->save_offset_top;  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
         }  
1744        }        }
1745    
1746      /* For both capturing and non-capturing groups, reset the value of the ims      /* For an ordinary non-repeating ket, just continue at this level. This
1747      flags, in case they got changed during the group. */      also happens for a repeating ket if no characters were matched in the
1748        group. This is the forcible breaking of infinite loops as implemented in
1749      ims = original_ims;      Perl 5.005. For a non-repeating atomic group, establish a backup point by
1750      DPRINTF(("ims reset to %02lx\n", ims));      processing the rest of the pattern at a lower level. If this results in a
1751        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1752      /* For a non-repeating ket, just continue at this level. This also      bypassing intermediate backup points, but resetting any captures that
1753      happens for a repeating ket if no characters were matched in the group.      happened along the way. */
     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. */  
1754    
1755      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1756        {        {
1757        ecode += 1 + LINK_SIZE;        if (*prev == OP_ONCE)
1758            {
1759            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1760            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1761            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1762            RRETURN(MATCH_ONCE);
1763            }
1764          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1765        break;        break;
1766        }        }
1767    
1768      /* The repeating kets try the rest of the pattern or restart from the      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1769      preceding bracket, in the appropriate order. In the second case, we can use      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1770      tail recursion to avoid using another stack frame, unless we have an      at a time from the outer level, thus saving stack. */
1771      unlimited repeat of a group that can match an empty string. */  
1772        if (*ecode == OP_KETRPOS)
1773      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;        {
1774          md->end_match_ptr = eptr;
1775          md->end_offset_top = offset_top;
1776          RRETURN(MATCH_KETRPOS);
1777          }
1778    
1779        /* The normal repeating kets try the rest of the pattern or restart from
1780        the preceding bracket, in the appropriate order. In the second case, we can
1781        use tail recursion to avoid using another stack frame, unless we have an
1782        an atomic group or an unlimited repeat of a group that can match an empty
1783        string. */
1784    
1785      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1786        {        {
1787        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
1788        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1789        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1790            {
1791            RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
1792            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1793            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1794            RRETURN(MATCH_ONCE);
1795            }
1796          if (*prev >= OP_SBRA)    /* Could match an empty string */
1797          {          {
1798          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1799            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1800          RRETURN(rrc);          RRETURN(rrc);
1801          }          }
1802        ecode = prev;        ecode = prev;
# Line 1670  for (;;) Line 1804  for (;;)
1804        }        }
1805      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1806        {        {
1807        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1808          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1809          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1810        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1811          if (*prev == OP_ONCE)
1812            {
1813            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
1814            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1815            md->once_target = prev;
1816            RRETURN(MATCH_ONCE);
1817            }
1818        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1819        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1820        }        }
1821      /* Control never gets here */      /* Control never gets here */
1822    
1823      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1824    
1825      case OP_CIRC:      case OP_CIRC:
1826      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1827      if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         MRRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
   
1828      /* Start of subject assertion */      /* Start of subject assertion */
1829    
1830      case OP_SOD:      case OP_SOD:
1831      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1832      ecode++;      ecode++;
1833      break;      break;
1834    
1835        /* Multiline mode: start of subject unless notbol, or after any newline. */
1836    
1837        case OP_CIRCM:
1838        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1839        if (eptr != md->start_subject &&
1840            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1841          MRRETURN(MATCH_NOMATCH);
1842        ecode++;
1843        break;
1844    
1845      /* Start of match assertion */      /* Start of match assertion */
1846    
# Line 1713  for (;;) Line 1856  for (;;)
1856      ecode++;      ecode++;
1857      break;      break;
1858    
1859      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1860      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1861    
1862      case OP_DOLL:      case OP_DOLLM:
1863      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1864        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1865        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 */  
1866        {        {
1867        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1868        if (!md->endonly) goto ASSERT_NL_OR_EOS;        SCHECK_PARTIAL();
1869        }        }
1870        ecode++;
1871        break;
1872    
1873        /* Not multiline mode: assert before a terminating newline or before end of
1874        subject unless noteol is set. */
1875    
1876        case OP_DOLL:
1877        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1878        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1879    
1880      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1881    
# Line 2017  for (;;) Line 2159  for (;;)
2159      switch(c)      switch(c)
2160        {        {
2161        default: MRRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2162    
2163        case 0x000d:        case 0x000d:
2164        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2165        break;        break;
# Line 2269  for (;;) Line 2412  for (;;)
2412      loops). */      loops). */
2413    
2414      case OP_REF:      case OP_REF:
2415        case OP_REFI:
2416        caseless = op == OP_REFI;
2417      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2418      ecode += 3;      ecode += 3;
2419    
# Line 2316  for (;;) Line 2461  for (;;)
2461        break;        break;
2462    
2463        default:               /* No repeat follows */        default:               /* No repeat follows */
2464        if ((length = match_ref(offset, eptr, length, md, ims)) < 0)        if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2465          {          {
2466          CHECK_PARTIAL();          CHECK_PARTIAL();
2467          MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
# Line 2337  for (;;) Line 2482  for (;;)
2482      for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2483        {        {
2484        int slength;        int slength;
2485        if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)        if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2486          {          {
2487          CHECK_PARTIAL();          CHECK_PARTIAL();
2488          MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
# Line 2357  for (;;) Line 2502  for (;;)
2502        for (fi = min;; fi++)        for (fi = min;; fi++)
2503          {          {
2504          int slength;          int slength;
2505          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2506          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2507          if (fi >= max) MRRETURN(MATCH_NOMATCH);          if (fi >= max) MRRETURN(MATCH_NOMATCH);
2508          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2509            {            {
2510            CHECK_PARTIAL();            CHECK_PARTIAL();
2511            MRRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
# Line 2378  for (;;) Line 2523  for (;;)
2523        for (i = min; i < max; i++)        for (i = min; i < max; i++)
2524          {          {
2525          int slength;          int slength;
2526          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2527            {            {
2528            CHECK_PARTIAL();            CHECK_PARTIAL();
2529            break;            break;
# Line 2387  for (;;) Line 2532  for (;;)
2532          }          }
2533        while (eptr >= pp)        while (eptr >= pp)
2534          {          {
2535          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);          RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2536          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2537          eptr -= length;          eptr -= length;
2538          }          }
# Line 2497  for (;;) Line 2642  for (;;)
2642            {            {
2643            for (fi = min;; fi++)            for (fi = min;; fi++)
2644              {              {
2645              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2646              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2647              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2648              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2522  for (;;) Line 2667  for (;;)
2667            {            {
2668            for (fi = min;; fi++)            for (fi = min;; fi++)
2669              {              {
2670              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2671              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2672              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2673              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 2568  for (;;) Line 2713  for (;;)
2713              }              }
2714            for (;;)            for (;;)
2715              {              {
2716              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2717              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2718              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2719              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2591  for (;;) Line 2736  for (;;)
2736              }              }
2737            while (eptr >= pp)            while (eptr >= pp)
2738              {              {
2739              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2740              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2741              eptr--;              eptr--;
2742              }              }
# Line 2667  for (;;) Line 2812  for (;;)
2812          {          {
2813          for (fi = min;; fi++)          for (fi = min;; fi++)
2814            {            {
2815            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2816            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2817            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2818            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 2700  for (;;) Line 2845  for (;;)
2845            }            }
2846          for(;;)          for(;;)
2847            {            {
2848            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2849            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2850            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2851            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 2745  for (;;) Line 2890  for (;;)
2890    
2891      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2892    
2893      case OP_CHARNC:      case OP_CHARI:
2894  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2895      if (utf8)      if (utf8)
2896        {        {
# Line 2805  for (;;) Line 2950  for (;;)
2950      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2951    
2952      case OP_EXACT:      case OP_EXACT:
2953        case OP_EXACTI:
2954      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2955      ecode += 3;      ecode += 3;
2956      goto REPEATCHAR;      goto REPEATCHAR;
2957    
2958      case OP_POSUPTO:      case OP_POSUPTO:
2959        case OP_POSUPTOI:
2960      possessive = TRUE;      possessive = TRUE;
2961      /* Fall through */      /* Fall through */
2962    
2963      case OP_UPTO:      case OP_UPTO:
2964        case OP_UPTOI:
2965      case OP_MINUPTO:      case OP_MINUPTO:
2966        case OP_MINUPTOI:
2967      min = 0;      min = 0;
2968      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2969      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2970      ecode += 3;      ecode += 3;
2971      goto REPEATCHAR;      goto REPEATCHAR;
2972    
2973      case OP_POSSTAR:      case OP_POSSTAR:
2974        case OP_POSSTARI:
2975      possessive = TRUE;      possessive = TRUE;
2976      min = 0;      min = 0;
2977      max = INT_MAX;      max = INT_MAX;
# Line 2829  for (;;) Line 2979  for (;;)
2979      goto REPEATCHAR;      goto REPEATCHAR;
2980    
2981      case OP_POSPLUS:      case OP_POSPLUS:
2982        case OP_POSPLUSI:
2983      possessive = TRUE;      possessive = TRUE;
2984      min = 1;      min = 1;
2985      max = INT_MAX;      max = INT_MAX;
# Line 2836  for (;;) Line 2987  for (;;)
2987      goto REPEATCHAR;      goto REPEATCHAR;
2988    
2989      case OP_POSQUERY:      case OP_POSQUERY:
2990        case OP_POSQUERYI:
2991      possessive = TRUE;      possessive = TRUE;
2992      min = 0;      min = 0;
2993      max = 1;      max = 1;
# Line 2843  for (;;) Line 2995  for (;;)
2995      goto REPEATCHAR;      goto REPEATCHAR;
2996    
2997      case OP_STAR:      case OP_STAR:
2998        case OP_STARI:
2999      case OP_MINSTAR:      case OP_MINSTAR:
3000        case OP_MINSTARI:
3001      case OP_PLUS:      case OP_PLUS:
3002        case OP_PLUSI:
3003      case OP_MINPLUS:      case OP_MINPLUS:
3004        case OP_MINPLUSI:
3005      case OP_QUERY:      case OP_QUERY:
3006        case OP_QUERYI:
3007      case OP_MINQUERY:      case OP_MINQUERY:
3008      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3009        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3010      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
   
3011      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3012      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3013      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 2873  for (;;) Line 3030  for (;;)
3030          {          {
3031  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3032          unsigned int othercase;          unsigned int othercase;
3033          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3034              (othercase = UCD_OTHERCASE(fc)) != fc)              (othercase = UCD_OTHERCASE(fc)) != fc)
3035            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3036          else oclength = 0;          else oclength = 0;
# Line 2901  for (;;) Line 3058  for (;;)
3058            {            {
3059            for (fi = min;; fi++)            for (fi = min;; fi++)
3060              {              {
3061              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3062              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3063              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3064              if (eptr <= md->end_subject - length &&              if (eptr <= md->end_subject - length &&
# Line 2943  for (;;) Line 3100  for (;;)
3100    
3101            for(;;)            for(;;)
3102              {              {
3103              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3104              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3105              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3106  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2980  for (;;) Line 3137  for (;;)
3137      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3138        max, eptr));        max, eptr));
3139    
3140      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3141        {        {
3142        fc = md->lcc[fc];        fc = md->lcc[fc];
3143        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
# Line 2997  for (;;) Line 3154  for (;;)
3154          {          {
3155          for (fi = min;; fi++)          for (fi = min;; fi++)
3156            {            {
3157            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3158            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3159            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3160            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3027  for (;;) Line 3184  for (;;)
3184    
3185          while (eptr >= pp)          while (eptr >= pp)
3186            {            {
3187            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3188            eptr--;            eptr--;
3189            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3190            }            }
# Line 3056  for (;;) Line 3213  for (;;)
3213          {          {
3214          for (fi = min;; fi++)          for (fi = min;; fi++)
3215            {            {
3216            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3217            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3218            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3219            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 3085  for (;;) Line 3242  for (;;)
3242    
3243          while (eptr >= pp)          while (eptr >= pp)
3244            {            {
3245            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3246            eptr--;            eptr--;
3247            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3248            }            }
# Line 3098  for (;;) Line 3255  for (;;)
3255      checking can be multibyte. */      checking can be multibyte. */
3256    
3257      case OP_NOT:      case OP_NOT:
3258        case OP_NOTI:
3259      if (eptr >= md->end_subject)      if (eptr >= md->end_subject)
3260        {        {
3261        SCHECK_PARTIAL();        SCHECK_PARTIAL();
# Line 3105  for (;;) Line 3263  for (;;)
3263        }        }
3264      ecode++;      ecode++;
3265      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3266      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3267        {        {
3268  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3269        if (c < 256)        if (c < 256)
# Line 3113  for (;;) Line 3271  for (;;)
3271        c = md->lcc[c];        c = md->lcc[c];
3272        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3273        }        }
3274      else      else    /* Caseful */
3275        {        {
3276        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3277        }        }
# Line 3127  for (;;) Line 3285  for (;;)
3285      about... */      about... */
3286    
3287      case OP_NOTEXACT:      case OP_NOTEXACT:
3288        case OP_NOTEXACTI:
3289      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3290      ecode += 3;      ecode += 3;
3291      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3292    
3293      case OP_NOTUPTO:      case OP_NOTUPTO:
3294        case OP_NOTUPTOI:
3295      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3296        case OP_NOTMINUPTOI:
3297      min = 0;      min = 0;
3298      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3299      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3300      ecode += 3;      ecode += 3;
3301      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3302    
3303      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3304        case OP_NOTPOSSTARI:
3305      possessive = TRUE;      possessive = TRUE;
3306      min = 0;      min = 0;
3307      max = INT_MAX;      max = INT_MAX;
# Line 3147  for (;;) Line 3309  for (;;)
3309      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3310    
3311      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3312        case OP_NOTPOSPLUSI:
3313      possessive = TRUE;      possessive = TRUE;
3314      min = 1;      min = 1;
3315      max = INT_MAX;      max = INT_MAX;
# Line 3154  for (;;) Line 3317  for (;;)
3317      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3318    
3319      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3320        case OP_NOTPOSQUERYI:
3321      possessive = TRUE;      possessive = TRUE;
3322      min = 0;      min = 0;
3323      max = 1;      max = 1;
# Line 3161  for (;;) Line 3325  for (;;)
3325      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3326    
3327      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3328        case OP_NOTPOSUPTOI:
3329      possessive = TRUE;      possessive = TRUE;
3330      min = 0;      min = 0;
3331      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 3168  for (;;) Line 3333  for (;;)
3333      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3334    
3335      case OP_NOTSTAR:      case OP_NOTSTAR:
3336        case OP_NOTSTARI:
3337      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3338        case OP_NOTMINSTARI:
3339      case OP_NOTPLUS:      case OP_NOTPLUS:
3340        case OP_NOTPLUSI:
3341      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3342        case OP_NOTMINPLUSI:
3343      case OP_NOTQUERY:      case OP_NOTQUERY:
3344        case OP_NOTQUERYI:
3345      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3346      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3347        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3348      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3349      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3350      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
# Line 3195  for (;;) Line 3366  for (;;)
3366      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,
3367        max, eptr));        max, eptr));
3368    
3369      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3370        {        {
3371        fc = md->lcc[fc];        fc = md->lcc[fc];
3372    
# Line 3243  for (;;) Line 3414  for (;;)
3414            register unsigned int d;            register unsigned int d;
3415            for (fi = min;; fi++)            for (fi = min;; fi++)
3416              {              {
3417              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3418              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3419              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3420              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3262  for (;;) Line 3433  for (;;)
3433            {            {
3434            for (fi = min;; fi++)            for (fi = min;; fi++)
3435              {              {
3436              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3437              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3438              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3439              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3303  for (;;) Line 3474  for (;;)
3474          if (possessive) continue;          if (possessive) continue;
3475          for(;;)          for(;;)
3476              {              {
3477              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3478              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3479              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3480              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3326  for (;;) Line 3497  for (;;)
3497            if (possessive) continue;            if (possessive) continue;
3498            while (eptr >= pp)            while (eptr >= pp)
3499              {              {
3500              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3501              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3502              eptr--;              eptr--;
3503              }              }
# Line 3383  for (;;) Line 3554  for (;;)
3554            register unsigned int d;            register unsigned int d;
3555            for (fi = min;; fi++)            for (fi = min;; fi++)
3556              {              {
3557              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3558              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3559              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3560              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3401  for (;;) Line 3572  for (;;)
3572            {            {
3573            for (fi = min;; fi++)            for (fi = min;; fi++)
3574              {              {
3575              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3576              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3577              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3578              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 3441  for (;;) Line 3612  for (;;)
3612            if (possessive) continue;            if (possessive) continue;
3613            for(;;)            for(;;)
3614              {              {
3615              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3616              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3617              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3618              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3464  for (;;) Line 3635  for (;;)
3635            if (possessive) continue;            if (possessive) continue;
3636            while (eptr >= pp)            while (eptr >= pp)
3637              {              {
3638              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3639              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3640              eptr--;              eptr--;
3641              }              }
# Line 3791  for (;;) Line 3962  for (;;)
3962            switch(c)            switch(c)
3963              {              {
3964              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3965    
3966              case 0x000d:              case 0x000d:
3967              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3968              break;              break;
# Line 4067  for (;;) Line 4239  for (;;)
4239            switch(*eptr++)            switch(*eptr++)
4240              {              {
4241              default: MRRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
4242    
4243              case 0x000d:              case 0x000d:
4244              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
4245              break;              break;
4246    
4247              case 0x000a:              case 0x000a:
4248              break;              break;
4249    
# Line 4259  for (;;) Line 4433  for (;;)
4433            case PT_ANY:            case PT_ANY:
4434            for (fi = min;; fi++)            for (fi = min;; fi++)
4435              {              {
4436              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM36);
4437              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4438              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4439              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4275  for (;;) Line 4449  for (;;)
4449            case PT_LAMP:            case PT_LAMP:
4450            for (fi = min;; fi++)            for (fi = min;; fi++)
4451              {              {
4452              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM37);
4453              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4454              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4455              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4295  for (;;) Line 4469  for (;;)
4469            case PT_GC:            case PT_GC:
4470            for (fi = min;; fi++)            for (fi = min;; fi++)
4471              {              {
4472              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM38);
4473              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4474              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4475              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4313  for (;;) Line 4487  for (;;)
4487            case PT_PC:            case PT_PC:
4488            for (fi = min;; fi++)            for (fi = min;; fi++)
4489              {              {
4490              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM39);
4491              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4492              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4493              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4331  for (;;) Line 4505  for (;;)
4505            case PT_SC:            case PT_SC:
4506            for (fi = min;; fi++)            for (fi = min;; fi++)
4507              {              {
4508              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM40);
4509              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4510              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4511              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4349  for (;;) Line 4523  for (;;)
4523            case PT_ALNUM:            case PT_ALNUM:
4524            for (fi = min;; fi++)            for (fi = min;; fi++)
4525              {              {
4526              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM59);
4527              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4528              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4529              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4368  for (;;) Line 4542  for (;;)
4542            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4543            for (fi = min;; fi++)            for (fi = min;; fi++)
4544              {              {
4545              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
4546              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4547              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4548              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4388  for (;;) Line 4562  for (;;)
4562            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4563            for (fi = min;; fi++)            for (fi = min;; fi++)
4564              {              {
4565              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM61);
4566              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4567              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4568              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4408  for (;;) Line 4582  for (;;)
4582            case PT_WORD:            case PT_WORD:
4583            for (fi = min;; fi++)            for (fi = min;; fi++)
4584              {              {
4585              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM62);
4586              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4587              if (fi >= max) MRRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
4588              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4440  for (;;) Line 4614  for (;;)
4614          {          {
4615          for (fi = min;; fi++)          for (fi = min;; fi++)
4616            {            {
4617            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM41);
4618            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4619            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4620            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4472  for (;;) Line 4646  for (;;)
4646          {          {
4647          for (fi = min;; fi++)          for (fi = min;; fi++)
4648            {            {
4649            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM42);
4650            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4651            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4652            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4635  for (;;) Line 4809  for (;;)
4809          {          {
4810          for (fi = min;; fi++)          for (fi = min;; fi++)
4811            {            {
4812            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM43);
4813            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4814            if (fi >= max) MRRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
4815            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
# Line 4933  for (;;) Line 5107  for (;;)
5107          if (possessive) continue;          if (possessive) continue;
5108          for(;;)          for(;;)
5109            {            {
5110            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5111            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5112            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5113            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
# Line 4974  for (;;) Line 5148  for (;;)
5148    
5149          for(;;)          for(;;)
5150            {            {
5151            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5152            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5153            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5154            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
# Line 5258  for (;;) Line 5432  for (;;)
5432            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5433            }            }
5434    
5435          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5436            done (no backing up). Otherwise, match at this position; anything other
5437            than no match is immediately returned. For nomatch, back up one
5438            character, unless we are matching \R and the last thing matched was
5439            \r\n, in which case, back up two bytes. */
5440    
5441          if (possessive) continue;          if (possessive) continue;
5442          for(;;)          for(;;)
5443            {            {
5444            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5445            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5446            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
5447            BACKCHAR(eptr);            BACKCHAR(eptr);
5448              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5449                  eptr[-1] == '\r') eptr--;
5450            }            }
5451          }          }
5452        else        else
# Line 5465  for (;;) Line 5645  for (;;)
5645            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5646            }            }
5647    
5648          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run. If possessive, we are
5649            done (no backing up). Otherwise, match at this position; anything other
5650            than no match is immediately returned. For nomatch, back up one
5651            character (byte), unless we are matching \R and the last thing matched
5652            was \r\n, in which case, back up two bytes. */
5653    
5654          if (possessive) continue;          if (possessive) continue;
5655          while (eptr >= pp)          while (eptr >= pp)
5656            {            {
5657            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
           eptr--;  
5658            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5659              eptr--;
5660              if (ctype == OP_ANYNL && eptr > pp  && *eptr == '\n' &&
5661                  eptr[-1] == '\r') eptr--;
5662            }            }
5663          }          }
5664    
# Line 5511  switch (frame->Xwhere) Line 5697  switch (frame->Xwhere)
5697    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)
5698    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)
5699    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)
5700    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) LBL(64)
5701      LBL(65) LBL(66)
5702  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5703    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)
5704    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
# Line 5540  Undefine all the macros that were define Line 5727  Undefine all the macros that were define
5727  #undef ecode  #undef ecode
5728  #undef mstart  #undef mstart
5729  #undef offset_top  #undef offset_top
 #undef ims  
5730  #undef eptrb  #undef eptrb
5731  #undef flags  #undef flags
5732    
# Line 5558  Undefine all the macros that were define Line 5744  Undefine all the macros that were define
5744  #undef condition  #undef condition
5745  #undef prev_is_word  #undef prev_is_word
5746    
 #undef original_ims  
   
5747  #undef ctype  #undef ctype
5748  #undef length  #undef length
5749  #undef max  #undef max
# Line 5616  pcre_exec(const pcre *argument_re, const Line 5800  pcre_exec(const pcre *argument_re, const
5800    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
5801    int offsetcount)    int offsetcount)
5802  {  {
5803  int rc, resetcount, ocount;  int rc, ocount;
5804  int first_byte = -1;  int first_byte = -1;
5805  int req_byte = -1;  int req_byte = -1;
5806  int req_byte2 = -1;  int req_byte2 = -1;
5807  int newline;  int newline;
 unsigned long int ims;  
5808  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
5809  BOOL anchored;  BOOL anchored;
5810  BOOL startline;  BOOL startline;
# Line 5725  utf8 = md->utf8 = (re->options & PCRE_UT Line 5908  utf8 = md->utf8 = (re->options & PCRE_UT
5908  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
5909  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
5910    
5911    /* Some options are unpacked into BOOL variables in the hope that testing
5912    them will be faster than individual option bits. */
5913    
5914  md->notbol = (options & PCRE_NOTBOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
5915  md->noteol = (options & PCRE_NOTEOL) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
5916  md->notempty = (options & PCRE_NOTEMPTY) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
5917  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;  md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0;
5918  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :  md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 :
5919                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;                ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0;
5920    
5921    
5922  md->hitend = FALSE;  md->hitend = FALSE;
5923  md->mark = NULL;                        /* In case never set */  md->mark = NULL;                        /* In case never set */
5924    
# Line 5812  defined (though never set). So there's n Line 6000  defined (though never set). So there's n
6000  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
6001    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
6002    
6003  /* 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
6004  back the character offset. */  code for an invalid string if a results vector is available. */
6005    
6006  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
6007  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
6008    {    {
6009    int tb;    int erroroffset;
6010    if ((tb = _pcre_valid_utf8((USPTR)subject, length)) >= 0)    int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset);
6011      return (tb == length && md->partial > 1)?    if (errorcode != 0)
       PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;  
   if (start_offset > 0 && start_offset < length)  
6012      {      {
6013      tb = ((USPTR)subject)[start_offset] & 0xc0;      if (offsetcount >= 2)
6014      if (tb == 0x80) return PCRE_ERROR_BADUTF8_OFFSET;        {
6015      }        offsets[0] = erroroffset;
6016          offsets[1] = errorcode;
6017          }
6018        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
6019          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
6020        }
6021    
6022      /* Check that a start_offset points to the start of a UTF-8 character. */
6023    
6024      if (start_offset > 0 && start_offset < length &&
6025          (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
6026        return PCRE_ERROR_BADUTF8_OFFSET;
6027    }    }
6028  #endif  #endif
6029    
 /* The ims options can vary during the matching as a result of the presence  
 of (?ims) items in the pattern. They are kept in a local variable so that  
 restoring at the exit of a group is easy. */  
   
 ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL);  
   
6030  /* 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
6031  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.
6032  Otherwise, we can use the vector supplied, rounding down its size to a multiple  Otherwise, we can use the vector supplied, rounding down its size to a multiple
# Line 5858  md->offset_max = (2*ocount)/3; Line 6049  md->offset_max = (2*ocount)/3;
6049  md->offset_overflow = FALSE;  md->offset_overflow = FALSE;
6050  md->capture_last = -1;  md->capture_last = -1;
6051    
 /* 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;  
   
6052  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
6053  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
6054  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. Also, unset the
6055    offsets for the matched string. This is really just for tidiness with callouts,
6056    in case they inspect these fields. */
6057    
6058  if (md->offset_vector != NULL)  if (md->offset_vector != NULL)
6059    {    {
6060    register int *iptr = md->offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
6061    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - re->top_bracket;
6062      if (iend < md->offset_vector + 2) iend = md->offset_vector + 2;
6063    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
6064      md->offset_vector[0] = md->offset_vector[1] = -1;
6065    }    }
6066    
6067  /* Set up the first character to match, if available. The first_byte value is  /* Set up the first character to match, if available. The first_byte value is
# Line 5907  if ((re->flags & PCRE_REQCHSET) != 0) Line 6095  if ((re->flags & PCRE_REQCHSET) != 0)
6095    }    }
6096    
6097    
6098    
6099    
6100  /* ==========================================================================*/  /* ==========================================================================*/
6101    
6102  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
# Line 5917  for(;;) Line 6107  for(;;)
6107    USPTR save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
6108    USPTR new_start_match;    USPTR new_start_match;
6109    
   /* 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;  
     }  
   
6110    /* 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
6111    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
6112    newline. Implement this by temporarily adjusting end_subject so that we stop    newline. Implement this by temporarily adjusting end_subject so that we stop
# Line 6114  for(;;) Line 6295  for(;;)
6295    md->start_match_ptr = start_match;    md->start_match_ptr = start_match;
6296    md->start_used_ptr = start_match;    md->start_used_ptr = start_match;
6297    md->match_call_count = 0;    md->match_call_count = 0;
6298    rc = match(start_match, md->start_code, start_match, NULL, 2, md, ims, NULL,    md->match_function_type = 0;
6299      0, 0);    md->end_offset_top = 0;
6300      rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0);
6301    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;
6302    
6303    switch(rc)    switch(rc)
# Line 6238  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 6420  if (rc == MATCH_MATCH || rc == MATCH_ACC
6420    
6421    /* Set the return code to the number of captured strings, or 0 if there are    /* Set the return code to the number of captured strings, or 0 if there are
6422    too many to fit into the vector. */    too many to fit into the vector. */
6423    
6424    rc = md->offset_overflow? 0 : md->end_offset_top/2;    rc = md->offset_overflow? 0 : md->end_offset_top/2;
6425    
6426    /* 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

Legend:
Removed from v.597  
changed lines
  Added in v.618

  ViewVC Help
Powered by ViewVC 1.1.5