/[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 226 by ph10, Tue Aug 21 11:46:08 2007 UTC revision 721 by ph10, Fri Oct 7 15:51:39 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) 1997-2007 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 43  pattern matching using an NFA algorithm, Line 43  pattern matching using an NFA algorithm,
43  possible. There are also some static supporting functions. */  possible. There are also some static supporting functions. */
44    
45  #ifdef HAVE_CONFIG_H  #ifdef HAVE_CONFIG_H
46  #include <config.h>  #include "config.h"
47  #endif  #endif
48    
49  #define NLBLOCK md             /* Block containing newline information */  #define NLBLOCK md             /* Block containing newline information */
# 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 71  defined PCRE_ERROR_xxx codes, which are Line 73  defined PCRE_ERROR_xxx codes, which are
73  /* Special internal returns from the match() function. Make them sufficiently  /* Special internal returns from the match() function. Make them sufficiently
74  negative to avoid the external error codes. */  negative to avoid the external error codes. */
75    
76  #define MATCH_COMMIT       (-999)  #define MATCH_ACCEPT       (-999)
77  #define MATCH_PRUNE        (-998)  #define MATCH_COMMIT       (-998)
78  #define MATCH_SKIP         (-997)  #define MATCH_KETRPOS      (-997)
79  #define MATCH_THEN         (-996)  #define MATCH_ONCE         (-996)
80    #define MATCH_PRUNE        (-995)
81    #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. */
86    
87    #define MRRETURN(ra) \
88      { \
89      md->mark = markptr; \
90      RRETURN(ra); \
91      }
92    
93  /* Maximum number of ints of offset to save on the stack for recursive calls.  /* Maximum number of ints of offset to save on the stack for recursive calls.
94  If the offset vector is bigger, malloc is used. This should be a multiple of 3,  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
# Line 89  static const char rep_max[] = { 0, 0, 0, Line 103  static const char rep_max[] = { 0, 0, 0,
103    
104    
105    
106  #ifdef DEBUG  #ifdef PCRE_DEBUG
107  /*************************************************  /*************************************************
108  *        Debugging function to print chars       *  *        Debugging function to print chars       *
109  *************************************************/  *************************************************/
# Line 122  while (length-- > 0) Line 136  while (length-- > 0)
136  *          Match a back-reference                *  *          Match a back-reference                *
137  *************************************************/  *************************************************/
138    
139  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
140  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
141    the length passed is zero. Note that in caseless UTF-8 mode, the number of
142    subject bytes matched may be different to the number of reference bytes.
143    
144  Arguments:  Arguments:
145    offset      index into the offset vector    offset      index into the offset vector
146    eptr        points into the subject    eptr        pointer into the subject
147    length      length to be matched    length      length of reference to be matched (number of bytes)
148    md          points to match data block    md          points to match data block
149    ims         the ims flags    caseless    TRUE if caseless
150    
151  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
152  */  */
153    
154  static BOOL  static int
155  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
156    unsigned long int ims)    BOOL caseless)
157  {  {
158  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
159    register USPTR p = md->start_subject + md->offset_vector[offset];
160    
161  #ifdef DEBUG  #ifdef PCRE_DEBUG
162  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
163    printf("matching subject <null>");    printf("matching subject <null>");
164  else  else
# Line 154  pchars(p, length, FALSE, md); Line 171  pchars(p, length, FALSE, md);
171  printf("\n");  printf("\n");
172  #endif  #endif
173    
174  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
175    
176  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
177    
178  /* Separate the caselesss case for speed */  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
179    properly if Unicode properties are supported. Otherwise, we can check only
180    ASCII characters. */
181    
182  if ((ims & PCRE_CASELESS) != 0)  if (caseless)
183    {    {
184    while (length-- > 0)  #ifdef SUPPORT_UTF8
185      if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE;  #ifdef SUPPORT_UCP
186      if (md->utf8)
187        {
188        /* Match characters up to the end of the reference. NOTE: the number of
189        bytes matched may differ, because there are some characters whose upper and
190        lower case versions code as different numbers of bytes. For example, U+023A
191        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
192        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
193        the latter. It is important, therefore, to check the length along the
194        reference, not along the subject (earlier code did this wrong). */
195    
196        USPTR endptr = p + length;
197        while (p < endptr)
198          {
199          int c, d;
200          if (eptr >= md->end_subject) return -1;
201          GETCHARINC(c, eptr);
202          GETCHARINC(d, p);
203          if (c != d && c != UCD_OTHERCASE(d)) return -1;
204          }
205        }
206      else
207    #endif
208    #endif
209    
210      /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
211      is no UCP support. */
212        {
213        if (eptr + length > md->end_subject) return -1;
214        while (length-- > 0)
215          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
216        }
217    }    }
218    
219    /* In the caseful case, we can just compare the bytes, whether or not we
220    are in UTF-8 mode. */
221    
222  else  else
223    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
224      if (eptr + length > md->end_subject) return -1;
225      while (length-- > 0) if (*p++ != *eptr++) return -1;
226      }
227    
228  return TRUE;  return eptr - eptr_start;
229  }  }
230    
231    
# Line 219  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 276  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
276         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,         RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
277         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
278         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
279         RM51,  RM52, RM53, RM54 };         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
280           RM61,  RM62, RM63 };
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
284  actuall used in this definition. */  actually used in this definition. */
285    
286  #ifndef NO_RECURSE  #ifndef NO_RECURSE
287  #define REGISTER register  #define REGISTER register
288    
289  #ifdef 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,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 241  actuall used in this definition. */ Line 299  actuall 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,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 255  argument of match(), which never changes Line 313  argument of match(), which never changes
313    
314  #define REGISTER  #define REGISTER
315    
316  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rw)\
317    {\    {\
318    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
319      if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
320    frame->Xwhere = rw; \    frame->Xwhere = rw; \
321    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
322    newframe->Xecode = rb;\    newframe->Xecode = rb;\
323    newframe->Xmstart = mstart;\    newframe->Xmstart = mstart;\
324      newframe->Xmarkptr = markptr;\
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 277  argument of match(), which never changes Line 335  argument of match(), which never changes
335    
336  #define RRETURN(ra)\  #define RRETURN(ra)\
337    {\    {\
338    heapframe *newframe = frame;\    heapframe *oldframe = frame;\
339    frame = newframe->Xprevframe;\    frame = oldframe->Xprevframe;\
340    (pcre_stack_free)(newframe);\    (pcre_stack_free)(oldframe);\
341    if (frame != NULL)\    if (frame != NULL)\
342      {\      {\
343      rrc = ra;\      rrc = ra;\
# Line 296  typedef struct heapframe { Line 354  typedef struct heapframe {
354    
355    /* Function arguments that may change */    /* Function arguments that may change */
356    
357    const uschar *Xeptr;    USPTR Xeptr;
358    const uschar *Xecode;    const uschar *Xecode;
359    const uschar *Xmstart;    USPTR Xmstart;
360      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 */
366    
367    const uschar *Xcallpat;    USPTR Xcallpat;
368    const uschar *Xcharptr;  #ifdef SUPPORT_UTF8
369    const uschar *Xdata;    USPTR Xcharptr;
370    const uschar *Xnext;  #endif
371    const uschar *Xpp;    USPTR Xdata;
372    const uschar *Xprev;    USPTR Xnext;
373    const uschar *Xsaved_eptr;    USPTR Xpp;
374      USPTR Xprev;
375      USPTR Xsaved_eptr;
376    
377    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
378    
# Line 321  typedef struct heapframe { Line 380  typedef struct heapframe {
380    BOOL Xcondition;    BOOL Xcondition;
381    BOOL Xprev_is_word;    BOOL Xprev_is_word;
382    
   unsigned long int Xoriginal_ims;  
   
383  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
386    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
387    int Xoclength;    int Xoclength;
388    uschar Xocchars[8];    uschar Xocchars[8];
389  #endif  #endif
390    
391      int Xcodelink;
392    int Xctype;    int Xctype;
393    unsigned int Xfc;    unsigned int Xfc;
394    int Xfi;    int Xfi;
# Line 369  typedef struct heapframe { Line 424  typedef struct heapframe {
424    
425  /* This function is called recursively in many circumstances. Whenever it  /* This function is called recursively in many circumstances. Whenever it
426  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
427  same response.  same response. */
428    
429    /* These macros pack up tests that are used for partial matching, and which
430    appears several times in the code. We set the "hit end" flag if the pointer is
431    at the end of the subject and also past the start of the subject (i.e.
432    something has been matched). For hard partial matching, we then return
433    immediately. The second one is used when we already know we are past the end of
434    the subject. */
435    
436    #define CHECK_PARTIAL()\
437      if (md->partial != 0 && eptr >= md->end_subject && \
438          eptr > md->start_used_ptr) \
439        { \
440        md->hitend = TRUE; \
441        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
442        }
443    
444    #define SCHECK_PARTIAL()\
445      if (md->partial != 0 && eptr > md->start_used_ptr) \
446        { \
447        md->hitend = TRUE; \
448        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
449        }
450    
451    
452  Performance note: It might be tempting to extract commonly used fields from the  /* Performance note: It might be tempting to extract commonly used fields from
453  md structure (e.g. utf8, end_subject) into individual variables to improve  the md structure (e.g. utf8, end_subject) into individual variables to improve
454  performance. Tests using gcc on a SPARC disproved this; in the first case, it  performance. Tests using gcc on a SPARC disproved this; in the first case, it
455  made performance worse.  made performance worse.
456    
# Line 381  Arguments: Line 459  Arguments:
459     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
460     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
461                   by encountering \K)                   by encountering \K)
462       markptr     pointer to the most recent MARK name, or NULL
463     offset_top  current top pointer     offset_top  current top pointer
464     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
465     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
466                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
467     rdepth      the recursion depth     rdepth      the recursion depth
468    
469  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
470                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
471                   a negative MATCH_xxx value for PRUNE, SKIP, etc
472                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
473                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
474  */  */
475    
476  static int  static int
477  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
478    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
479    int flags, unsigned int rdepth)    unsigned int rdepth)
480  {  {
481  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
482  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 413  register unsigned int c;   /* Character Line 488  register unsigned int c;   /* Character
488  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
489    
490  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
491    BOOL caseless;
492    int condcode;
493    
494  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
495  preserved over calls to RMATCH() are part of a "frame" which is obtained from  preserved over calls to RMATCH() are part of a "frame" which is obtained from
# Line 420  heap storage. Set up the top-level frame Line 497  heap storage. Set up the top-level frame
497  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
498    
499  #ifdef NO_RECURSE  #ifdef NO_RECURSE
500  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
501    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
502  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
503    
504  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 428  frame->Xprevframe = NULL;            /* Line 506  frame->Xprevframe = NULL;            /*
506  frame->Xeptr = eptr;  frame->Xeptr = eptr;
507  frame->Xecode = ecode;  frame->Xecode = ecode;
508  frame->Xmstart = mstart;  frame->Xmstart = mstart;
509    frame->Xmarkptr = markptr;
510  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
511  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
512  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
513    
514  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 443  HEAP_RECURSE: Line 520  HEAP_RECURSE:
520  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
521  #define ecode              frame->Xecode  #define ecode              frame->Xecode
522  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
523    #define markptr            frame->Xmarkptr
524  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
525  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
526  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
527    
528  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 455  HEAP_RECURSE: Line 531  HEAP_RECURSE:
531  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
532  #endif  #endif
533  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
534    #define codelink           frame->Xcodelink
535  #define data               frame->Xdata  #define data               frame->Xdata
536  #define next               frame->Xnext  #define next               frame->Xnext
537  #define pp                 frame->Xpp  #define pp                 frame->Xpp
# Line 467  HEAP_RECURSE: Line 544  HEAP_RECURSE:
544  #define condition          frame->Xcondition  #define condition          frame->Xcondition
545  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
546    
 #define original_ims       frame->Xoriginal_ims  
   
547  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
548  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
549  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
550  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
 #define prop_category      frame->Xprop_category  
 #define prop_chartype      frame->Xprop_chartype  
 #define prop_script        frame->Xprop_script  
551  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
552  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
553  #endif  #endif
# Line 505  i, and fc and c, can be the same variabl Line 577  i, and fc and c, can be the same variabl
577  #define fi i  #define fi i
578  #define fc c  #define fc c
579    
580    /* Many of the following variables are used only in small blocks of the code.
581    My normal style of coding would have declared them within each of those blocks.
582    However, in order to accommodate the version of this code that uses an external
583    "stack" implemented on the heap, it is easier to declare them all here, so the
584    declarations can be cut out in a block. The only declarations within blocks
585    below are for variables that do not have to be preserved over a recursive call
586    to RMATCH(). */
587    
588    #ifdef SUPPORT_UTF8
589    const uschar *charptr;
590    #endif
591    const uschar *callpat;
592    const uschar *data;
593    const uschar *next;
594    USPTR         pp;
595    const uschar *prev;
596    USPTR         saved_eptr;
597    
598    recursion_info new_recursive;
599    
600  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  BOOL cur_is_word;
 const uschar *charptr;             /* in small blocks of the code. My normal */  
 #endif                             /* style of coding would have declared    */  
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
 unsigned long int original_ims;  
   
604  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
608  int oclength;  int oclength;
609  uschar occhars[8];  uschar occhars[8];
610  #endif  #endif
611    
612    int codelink;
613  int ctype;  int ctype;
614  int length;  int length;
615  int max;  int max;
# Line 549  int stacksave[REC_STACK_SAVE_MAX]; Line 624  int stacksave[REC_STACK_SAVE_MAX];
624  eptrblock newptrb;  eptrblock newptrb;
625  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
626    
627    /* To save space on the stack and in the heap frame, I have doubled up on some
628    of the local variables that are used only in localised parts of the code, but
629    still need to be preserved over recursive calls of match(). These macros define
630    the alternative names that are used. */
631    
632    #define allow_zero    cur_is_word
633    #define cbegroup      condition
634    #define code_offset   codelink
635    #define condassert    condition
636    #define matched_once  prev_is_word
637    
638  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
639  variables. */  variables. */
640    
# Line 568  TAIL_RECURSE: Line 654  TAIL_RECURSE:
654  /* OK, now we can get on with the real code of the function. Recursive calls  /* OK, now we can get on with the real code of the function. Recursive calls
655  are specified by the macro RMATCH and RRETURN is used to return. When  are specified by the macro RMATCH and RRETURN is used to return. When
656  NO_RECURSE is *not* defined, these just turn into a recursive call to match()  NO_RECURSE is *not* defined, these just turn into a recursive call to match()
657  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
658  defined). However, RMATCH isn't like a function call because it's quite a  defined). However, RMATCH isn't like a function call because it's quite a
659  complicated macro. It has to be used in one particular way. This shouldn't,  complicated macro. It has to be used in one particular way. This shouldn't,
660  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
# Line 585  haven't exceeded the recursive call limi Line 671  haven't exceeded the recursive call limi
671  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
672  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
673    
 original_ims = ims;    /* Save for resetting on ')' */  
   
674  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
675  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
676  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
677  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
678  When match() is called in other circumstances, don't add to the chain. The  
679  match_cbegroup flag must NOT be used with tail recursion, because the memory  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
680  block that is used is on the stack, so a new one may be required for each  such remembered pointers, to be checked when we hit the closing ket, in order
681  match(). */  to break infinite loops that match no characters. When match() is called in
682    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
683    NOT be used with tail recursion, because the memory block that is used is on
684    the stack, so a new one may be required for each match(). */
685    
686  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
687    {    {
688    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
689    newptrb.epb_prev = eptrb;    newptrb.epb_prev = eptrb;
690    eptrb = &newptrb;    eptrb = &newptrb;
691      md->match_function_type = 0;
692    }    }
693    
694  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 610  for (;;) Line 698  for (;;)
698    minimize = possessive = FALSE;    minimize = possessive = FALSE;
699    op = *ecode;    op = *ecode;
700    
   /* For partial matching, remember if we ever hit the end of the subject after  
   matching at least one subject character. */  
   
   if (md->partial &&  
       eptr >= md->end_subject &&  
       eptr > mstart)  
     md->hitend = TRUE;  
   
701    switch(op)    switch(op)
702      {      {
703        case OP_MARK:
704        markptr = ecode + 2;
705        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
706          eptrb, RM55);
707    
708        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
709        argument, and we must check whether that argument matches this MARK's
710        argument. It is passed back in md->start_match_ptr (an overloading of that
711        variable). If it does match, we reset that variable to the current subject
712        position and return MATCH_SKIP. Otherwise, pass back the return code
713        unaltered. */
714    
715        if (rrc == MATCH_SKIP_ARG &&
716            strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
717          {
718          md->start_match_ptr = eptr;
719          RRETURN(MATCH_SKIP);
720          }
721    
722        if (md->mark == NULL) md->mark = markptr;
723        RRETURN(rrc);
724    
725      case OP_FAIL:      case OP_FAIL:
726      RRETURN(MATCH_NOMATCH);      MRRETURN(MATCH_NOMATCH);
727    
728        /* COMMIT overrides PRUNE, SKIP, and THEN */
729    
730        case OP_COMMIT:
731        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
732          eptrb, RM52);
733        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
734            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
735            rrc != MATCH_THEN)
736          RRETURN(rrc);
737        MRRETURN(MATCH_COMMIT);
738    
739        /* PRUNE overrides THEN */
740    
741      case OP_PRUNE:      case OP_PRUNE:
742      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
743        ims, eptrb, flags, RM51);        eptrb, RM51);
744      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
745        MRRETURN(MATCH_PRUNE);
746    
747        case OP_PRUNE_ARG:
748        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
749          eptrb, RM56);
750        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
751        md->mark = ecode + 2;
752      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
753    
754      case OP_COMMIT:      /* SKIP overrides PRUNE and THEN */
     RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,  
       ims, eptrb, flags, RM52);  
     if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
     RRETURN(MATCH_COMMIT);  
755    
756      case OP_SKIP:      case OP_SKIP:
757      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
758        ims, eptrb, flags, RM53);        eptrb, RM53);
759      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
760          RRETURN(rrc);
761      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
762      RRETURN(MATCH_SKIP);      MRRETURN(MATCH_SKIP);
763    
764        case OP_SKIP_ARG:
765        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
766          eptrb, RM57);
767        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769    
770        /* Pass back the current skip name by overloading md->start_match_ptr and
771        returning the special MATCH_SKIP_ARG return code. This will either be
772        caught by a matching MARK, or get to the top, where it is treated the same
773        as PRUNE. */
774    
775        md->start_match_ptr = ecode + 2;
776        RRETURN(MATCH_SKIP_ARG);
777    
778        /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that
779        the branch in which it occurs can be determined. Overload the start of
780        match pointer to do this. */
781    
782      case OP_THEN:      case OP_THEN:
783      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
784        ims, eptrb, flags, RM54);        eptrb, RM54);
785        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
786        md->start_match_ptr = ecode;
787        MRRETURN(MATCH_THEN);
788    
789        case OP_THEN_ARG:
790        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top,
791          md, eptrb, RM58);
792      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
793        md->start_match_ptr = ecode;
794        md->mark = ecode + 2;
795      RRETURN(MATCH_THEN);      RRETURN(MATCH_THEN);
796    
797      /* 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
798      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
799      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
800      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
801      reference inside the group.      previous iteration of this group, and be referred to by a reference inside
802        the group. A failure to match might occur after the group has succeeded,
803      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
804      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
805      of the same bracket.      were set by a previous iteration of the same bracket.
806    
807      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
808      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 667  for (;;) Line 813  for (;;)
813      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
814      offset = number << 1;      offset = number << 1;
815    
816  #ifdef DEBUG  #ifdef PCRE_DEBUG
817      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
818      printf("subject=");      printf("subject=");
819      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 682  for (;;) Line 828  for (;;)
828        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
829    
830        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
831        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
832            (int)(eptr - md->start_subject);
833    
834        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
835          {          {
836            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
837          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
838            ims, eptrb, flags, RM1);            eptrb, RM1);
839          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
840    
841            /* If we backed up to a THEN, check whether it is within the current
842            branch by comparing the address of the THEN that is passed back with
843            the end of the branch. If it is within the current branch, and the
844            branch is one of two or more alternatives (it either starts or ends
845            with OP_ALT), we have reached the limit of THEN's action, so convert
846            the return code to NOMATCH, which will cause normal backtracking to
847            happen from now on. Otherwise, THEN is passed back to an outer
848            alternative. This implements Perl's treatment of parenthesized groups,
849            where a group not containing | does not affect the current alternative,
850            that is, (X) is NOT the same as (X|(*F)). */
851    
852            if (rrc == MATCH_THEN)
853              {
854              next = ecode + GET(ecode,1);
855              if (md->start_match_ptr < next &&
856                  (*ecode == OP_ALT || *next == OP_ALT))
857                rrc = MATCH_NOMATCH;
858              }
859    
860            /* Anything other than NOMATCH is passed back. */
861    
862            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
863          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
864          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
865            if (*ecode != OP_ALT) break;
866          }          }
       while (*ecode == OP_ALT);  
867    
868        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
869        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
870        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
871        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
872    
873        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */
874    
875          if (md->mark == NULL) md->mark = markptr;
876          RRETURN(rrc);
877        }        }
878    
879      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 715  for (;;) Line 887  for (;;)
887      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
888      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
889    
890      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
891      final alternative within the brackets, we would return the result of a      repeat. Loop for all the alternatives.
892      recursive call to match() whatever happened. We can reduce stack usage by  
893      turning this into a tail recursion, except in the case when match_cbegroup      When we get to the final alternative within the brackets, we used to return
894      is set.*/      the result of a recursive call to match() whatever happened so it was
895        possible to reduce stack usage by turning this into a tail recursion,
896        except in the case of a possibly empty group. However, now that there is
897        the possiblity of (*THEN) occurring in the final alternative, this
898        optimization is no longer always possible.
899    
900        We can optimize if we know there are no (*THEN)s in the pattern; at present
901        this is the best that can be done.
902    
903        MATCH_ONCE is returned when the end of an atomic group is successfully
904        reached, but subsequent matching fails. It passes back up the tree (causing
905        captured values to be reset) until the original atomic group level is
906        reached. This is tested by comparing md->once_target with the start of the
907        group. At this point, the return is converted into MATCH_NOMATCH so that
908        previous backup points can be taken. */
909    
910        case OP_ONCE:
911      case OP_BRA:      case OP_BRA:
912      case OP_SBRA:      case OP_SBRA:
913      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
914      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
915      for (;;)      for (;;)
916        {        {
917        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
918    
919          /* If this is not a possibly empty group, and there are no (*THEN)s in
920          the pattern, and this is the final alternative, optimize as described
921          above. */
922    
923          else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT)
924            {
925            ecode += _pcre_OP_lengths[*ecode];
926            goto TAIL_RECURSE;
927            }
928    
929          /* In all other cases, we have to make another call to match(). */
930    
931          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
932            RM2);
933    
934          /* See comment in the code for capturing groups above about handling
935          THEN. */
936    
937          if (rrc == MATCH_THEN)
938          {          {
939          if (flags == 0)    /* Not a possibly empty group */          next = ecode + GET(ecode,1);
940            if (md->start_match_ptr < next &&
941                (*ecode == OP_ALT || *next == OP_ALT))
942              rrc = MATCH_NOMATCH;
943            }
944    
945          if (rrc != MATCH_NOMATCH)
946            {
947            if (rrc == MATCH_ONCE)
948            {            {
949            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
950            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
951            goto TAIL_RECURSE;              {
952                while (*scode == OP_ALT) scode += GET(scode, 1);
953                scode -= GET(scode, 1);
954                }
955              if (md->once_target == scode) rrc = MATCH_NOMATCH;
956            }            }
957            RRETURN(rrc);
958            }
959          ecode += GET(ecode, 1);
960          if (*ecode != OP_ALT) break;
961          }
962    
963        if (md->mark == NULL) md->mark = markptr;
964        RRETURN(MATCH_NOMATCH);
965    
966          /* Possibly empty group; can't use tail recursion. */      /* Handle possessive capturing brackets with an unlimited repeat. We come
967        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
968        handled similarly to the normal case above. However, the matching is
969        different. The end of these brackets will always be OP_KETRPOS, which
970        returns MATCH_KETRPOS without going further in the pattern. By this means
971        we can handle the group by iteration rather than recursion, thereby
972        reducing the amount of stack needed. */
973    
974        case OP_CBRAPOS:
975        case OP_SCBRAPOS:
976        allow_zero = FALSE;
977    
978          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      POSSESSIVE_CAPTURE:
979            eptrb, flags, RM48);      number = GET2(ecode, 1+LINK_SIZE);
980          RRETURN(rrc);      offset = number << 1;
981    
982    #ifdef PCRE_DEBUG
983        printf("start possessive bracket %d\n", number);
984        printf("subject=");
985        pchars(eptr, 16, TRUE, md);
986        printf("\n");
987    #endif
988    
989        if (offset < md->offset_max)
990          {
991          matched_once = FALSE;
992          code_offset = ecode - md->start_code;
993    
994          save_offset1 = md->offset_vector[offset];
995          save_offset2 = md->offset_vector[offset+1];
996          save_offset3 = md->offset_vector[md->offset_end - number];
997          save_capture_last = md->capture_last;
998    
999          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
1000    
1001          /* Each time round the loop, save the current subject position for use
1002          when the group matches. For MATCH_MATCH, the group has matched, so we
1003          restart it with a new subject starting position, remembering that we had
1004          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
1005          usual. If we haven't matched any alternatives in any iteration, check to
1006          see if a previous iteration matched. If so, the group has matched;
1007          continue from afterwards. Otherwise it has failed; restore the previous
1008          capture values before returning NOMATCH. */
1009    
1010          for (;;)
1011            {
1012            md->offset_vector[md->offset_end - number] =
1013              (int)(eptr - md->start_subject);
1014            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1015            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1016              eptrb, RM63);
1017            if (rrc == MATCH_KETRPOS)
1018              {
1019              offset_top = md->end_offset_top;
1020              eptr = md->end_match_ptr;
1021              ecode = md->start_code + code_offset;
1022              save_capture_last = md->capture_last;
1023              matched_once = TRUE;
1024              continue;
1025              }
1026    
1027            /* See comment in the code for capturing groups above about handling
1028            THEN. */
1029    
1030            if (rrc == MATCH_THEN)
1031              {
1032              next = ecode + GET(ecode,1);
1033              if (md->start_match_ptr < next &&
1034                  (*ecode == OP_ALT || *next == OP_ALT))
1035                rrc = MATCH_NOMATCH;
1036              }
1037    
1038            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1039            md->capture_last = save_capture_last;
1040            ecode += GET(ecode, 1);
1041            if (*ecode != OP_ALT) break;
1042          }          }
1043    
1044        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
1045        otherwise return. */          {
1046            md->offset_vector[offset] = save_offset1;
1047            md->offset_vector[offset+1] = save_offset2;
1048            md->offset_vector[md->offset_end - number] = save_offset3;
1049            }
1050    
1051        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (md->mark == NULL) md->mark = markptr;
1052          eptrb, flags, RM2);        if (allow_zero || matched_once)
1053        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);          {
1054            ecode += 1 + LINK_SIZE;
1055            break;
1056            }
1057    
1058          RRETURN(MATCH_NOMATCH);
1059          }
1060    
1061        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1062        as a non-capturing bracket. */
1063    
1064        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1065        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1066    
1067        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1068    
1069        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1070        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1071    
1072        /* Non-capturing possessive bracket with unlimited repeat. We come here
1073        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1074        without the capturing complication. It is written out separately for speed
1075        and cleanliness. */
1076    
1077        case OP_BRAPOS:
1078        case OP_SBRAPOS:
1079        allow_zero = FALSE;
1080    
1081        POSSESSIVE_NON_CAPTURE:
1082        matched_once = FALSE;
1083        code_offset = ecode - md->start_code;
1084    
1085        for (;;)
1086          {
1087          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1088          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1089            eptrb, RM48);
1090          if (rrc == MATCH_KETRPOS)
1091            {
1092            offset_top = md->end_offset_top;
1093            eptr = md->end_match_ptr;
1094            ecode = md->start_code + code_offset;
1095            matched_once = TRUE;
1096            continue;
1097            }
1098    
1099          /* See comment in the code for capturing groups above about handling
1100          THEN. */
1101    
1102          if (rrc == MATCH_THEN)
1103            {
1104            next = ecode + GET(ecode,1);
1105            if (md->start_match_ptr < next &&
1106                (*ecode == OP_ALT || *next == OP_ALT))
1107              rrc = MATCH_NOMATCH;
1108            }
1109    
1110          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1111        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1112          if (*ecode != OP_ALT) break;
1113        }        }
1114    
1115        if (matched_once || allow_zero)
1116          {
1117          ecode += 1 + LINK_SIZE;
1118          break;
1119          }
1120        RRETURN(MATCH_NOMATCH);
1121    
1122      /* Control never reaches here. */      /* Control never reaches here. */
1123    
1124      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1125      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
1126      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
1127      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. */  
1128    
1129      case OP_COND:      case OP_COND:
1130      case OP_SCOND:      case OP_SCOND:
1131      if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */      codelink = GET(ecode, 1);
1132    
1133        /* Because of the way auto-callout works during compile, a callout item is
1134        inserted between OP_COND and an assertion condition. */
1135    
1136        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1137        {        {
1138        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (pcre_callout != NULL)
1139        condition = md->recursive != NULL &&          {
1140          (offset == RREF_ANY || offset == md->recursive->group_num);          pcre_callout_block cb;
1141        ecode += condition? 3 : GET(ecode, 1);          cb.version          = 2;   /* Version 1 of the callout block */
1142            cb.callout_number   = ecode[LINK_SIZE+2];
1143            cb.offset_vector    = md->offset_vector;
1144            cb.subject          = (PCRE_SPTR)md->start_subject;
1145            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1146            cb.start_match      = (int)(mstart - md->start_subject);
1147            cb.current_position = (int)(eptr - md->start_subject);
1148            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1149            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1150            cb.capture_top      = offset_top/2;
1151            cb.capture_last     = md->capture_last;
1152            cb.callout_data     = md->callout_data;
1153            cb.mark             = markptr;
1154            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1155            if (rrc < 0) RRETURN(rrc);
1156            }
1157          ecode += _pcre_OP_lengths[OP_CALLOUT];
1158          }
1159    
1160        condcode = ecode[LINK_SIZE+1];
1161    
1162        /* Now see what the actual condition is */
1163    
1164        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1165          {
1166          if (md->recursive == NULL)                /* Not recursing => FALSE */
1167            {
1168            condition = FALSE;
1169            ecode += GET(ecode, 1);
1170            }
1171          else
1172            {
1173            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1174            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1175    
1176            /* If the test is for recursion into a specific subpattern, and it is
1177            false, but the test was set up by name, scan the table to see if the
1178            name refers to any other numbers, and test them. The condition is true
1179            if any one is set. */
1180    
1181            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1182              {
1183              uschar *slotA = md->name_table;
1184              for (i = 0; i < md->name_count; i++)
1185                {
1186                if (GET2(slotA, 0) == recno) break;
1187                slotA += md->name_entry_size;
1188                }
1189    
1190              /* Found a name for the number - there can be only one; duplicate
1191              names for different numbers are allowed, but not vice versa. First
1192              scan down for duplicates. */
1193    
1194              if (i < md->name_count)
1195                {
1196                uschar *slotB = slotA;
1197                while (slotB > md->name_table)
1198                  {
1199                  slotB -= md->name_entry_size;
1200                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1201                    {
1202                    condition = GET2(slotB, 0) == md->recursive->group_num;
1203                    if (condition) break;
1204                    }
1205                  else break;
1206                  }
1207    
1208                /* Scan up for duplicates */
1209    
1210                if (!condition)
1211                  {
1212                  slotB = slotA;
1213                  for (i++; i < md->name_count; i++)
1214                    {
1215                    slotB += md->name_entry_size;
1216                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1217                      {
1218                      condition = GET2(slotB, 0) == md->recursive->group_num;
1219                      if (condition) break;
1220                      }
1221                    else break;
1222                    }
1223                  }
1224                }
1225              }
1226    
1227            /* Chose branch according to the condition */
1228    
1229            ecode += condition? 3 : GET(ecode, 1);
1230            }
1231        }        }
1232    
1233      else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1234        {        {
1235        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1236        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1237    
1238          /* If the numbered capture is unset, but the reference was by name,
1239          scan the table to see if the name refers to any other numbers, and test
1240          them. The condition is true if any one is set. This is tediously similar
1241          to the code above, but not close enough to try to amalgamate. */
1242    
1243          if (!condition && condcode == OP_NCREF)
1244            {
1245            int refno = offset >> 1;
1246            uschar *slotA = md->name_table;
1247    
1248            for (i = 0; i < md->name_count; i++)
1249              {
1250              if (GET2(slotA, 0) == refno) break;
1251              slotA += md->name_entry_size;
1252              }
1253    
1254            /* Found a name for the number - there can be only one; duplicate names
1255            for different numbers are allowed, but not vice versa. First scan down
1256            for duplicates. */
1257    
1258            if (i < md->name_count)
1259              {
1260              uschar *slotB = slotA;
1261              while (slotB > md->name_table)
1262                {
1263                slotB -= md->name_entry_size;
1264                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1265                  {
1266                  offset = GET2(slotB, 0) << 1;
1267                  condition = offset < offset_top &&
1268                    md->offset_vector[offset] >= 0;
1269                  if (condition) break;
1270                  }
1271                else break;
1272                }
1273    
1274              /* Scan up for duplicates */
1275    
1276              if (!condition)
1277                {
1278                slotB = slotA;
1279                for (i++; i < md->name_count; i++)
1280                  {
1281                  slotB += md->name_entry_size;
1282                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1283                    {
1284                    offset = GET2(slotB, 0) << 1;
1285                    condition = offset < offset_top &&
1286                      md->offset_vector[offset] >= 0;
1287                    if (condition) break;
1288                    }
1289                  else break;
1290                  }
1291                }
1292              }
1293            }
1294    
1295          /* Chose branch according to the condition */
1296    
1297        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 3 : GET(ecode, 1);
1298        }        }
1299    
1300      else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
1301        {        {
1302        condition = FALSE;        condition = FALSE;
1303        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1304        }        }
1305    
1306      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1307      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
1308      assertion. */      an assertion. */
1309    
1310      else      else
1311        {        {
1312        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1313            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1314        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1315          {          {
1316            if (md->end_offset_top > offset_top)
1317              offset_top = md->end_offset_top;  /* Captures may have happened */
1318          condition = TRUE;          condition = TRUE;
1319          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1320          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1321          }          }
1322        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)  
1323          /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1324          assertion; it is therefore treated as NOMATCH. */
1325    
1326          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1327          {          {
1328          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1329          }          }
1330        else        else
1331          {          {
1332          condition = FALSE;          condition = FALSE;
1333          ecode += GET(ecode, 1);          ecode += codelink;
1334          }          }
1335        }        }
1336    
1337      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one, can
1338      we can use tail recursion to avoid using another stack frame, except when      use tail recursion to avoid using another stack frame, except when there is
1339      match_cbegroup is required for an unlimited repeat of a possibly empty      unlimited repeat of a possibly empty group. In the latter case, a recursive
1340      group. If the second alternative doesn't exist, we can just plough on. */      call to match() is always required, unless the second alternative doesn't
1341        exist, in which case we can just plough on. Note that, for compatibility
1342        with Perl, the | in a conditional group is NOT treated as creating two
1343        alternatives. If a THEN is encountered in the branch, it propagates out to
1344        the enclosing alternative (unless nested in a deeper set of alternatives,
1345        of course). */
1346    
1347      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1348        {        {
1349        ecode += 1 + LINK_SIZE;        if (op != OP_SCOND)
       if (op == OP_SCOND)        /* Possibly empty group */  
         {  
         RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);  
         RRETURN(rrc);  
         }  
       else                       /* Group must match something */  
1350          {          {
1351          flags = 0;          ecode += 1 + LINK_SIZE;
1352          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1353          }          }
1354    
1355          md->match_function_type = MATCH_CBEGROUP;
1356          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1357          RRETURN(rrc);
1358        }        }
1359      else                         /* Condition false & no 2nd alternative */  
1360         /* Condition false & no alternative; continue after the group. */
1361    
1362        else
1363        {        {
1364        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1365        }        }
1366      break;      break;
1367    
1368    
1369      /* End of the pattern, either real or forced. If we are in a top-level      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1370      recursion, we should restore the offsets appropriately and continue from      to close any currently open capturing brackets. */
     after the call. */  
1371    
1372      case OP_ACCEPT:      case OP_CLOSE:
1373      case OP_END:      number = GET2(ecode, 1);
1374      if (md->recursive != NULL && md->recursive->group_num == 0)      offset = number << 1;
1375    
1376    #ifdef PCRE_DEBUG
1377          printf("end bracket %d at *ACCEPT", number);
1378          printf("\n");
1379    #endif
1380    
1381        md->capture_last = number;
1382        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1383        {        {
1384        recursion_info *rec = md->recursive;        md->offset_vector[offset] =
1385        DPRINTF(("End of pattern in a (?0) recursion\n"));          md->offset_vector[md->offset_end - number];
1386        md->recursive = rec->prevrec;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1387        memmove(md->offset_vector, rec->offset_save,        if (offset_top <= offset) offset_top = offset + 2;
         rec->saved_max * sizeof(int));  
       mstart = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
1388        }        }
1389        ecode += 3;
1390        break;
1391    
     /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty  
     string - backtracking will then try other alternatives, if any. */  
1392    
1393      if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH);      /* End of the pattern, either real or forced. */
1394    
1395        case OP_END:
1396        case OP_ACCEPT:
1397        case OP_ASSERT_ACCEPT:
1398    
1399        /* If we have matched an empty string, fail if not in an assertion and not
1400        in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1401        is set and we have matched at the start of the subject. In both cases,
1402        backtracking will then try other alternatives, if any. */
1403    
1404        if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1405             md->recursive == NULL &&
1406             (md->notempty ||
1407               (md->notempty_atstart &&
1408                 mstart == md->start_subject + md->start_offset)))
1409          MRRETURN(MATCH_NOMATCH);
1410    
1411        /* Otherwise, we have a match. */
1412    
1413      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1414      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1415      md->start_match_ptr = mstart;       /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
     RRETURN(MATCH_MATCH);  
1416    
1417      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1418        given as the argument to MRRETURN when the heap is in use. */
1419    
1420      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1421      ims = ecode[1];      MRRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1422    
1423      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1424      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,
1425      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
1426      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
1427      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1428        of a condition, we want to return immediately afterwards. The caller of
1429        this incarnation of the match() function will have set MATCH_CONDASSERT in
1430        md->match_function type, and one of these opcodes will be the first opcode
1431        that is processed. We use a local variable that is preserved over calls to
1432        match() to remember this case. */
1433    
1434      case OP_ASSERT:      case OP_ASSERT:
1435      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1436        if (md->match_function_type == MATCH_CONDASSERT)
1437          {
1438          condassert = TRUE;
1439          md->match_function_type = 0;
1440          }
1441        else condassert = FALSE;
1442    
1443      do      do
1444        {        {
1445        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1446          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1447        if (rrc == MATCH_MATCH) break;          {
1448            mstart = md->start_match_ptr;   /* In case \K reset it */
1449            markptr = md->mark;
1450            break;
1451            }
1452    
1453          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1454          as NOMATCH. */
1455    
1456        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1457        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1458        }        }
1459      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1460      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1461        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1462    
1463      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1464    
1465      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1466    
1467      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1468      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 900  for (;;) Line 1472  for (;;)
1472      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1473      continue;      continue;
1474    
1475      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1476        PRUNE, or COMMIT means we must assume failure without checking subsequent
1477        branches. */
1478    
1479      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1480      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1481        if (md->match_function_type == MATCH_CONDASSERT)
1482          {
1483          condassert = TRUE;
1484          md->match_function_type = 0;
1485          }
1486        else condassert = FALSE;
1487    
1488      do      do
1489        {        {
1490        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1491          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1492        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1493            {
1494            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1495            break;
1496            }
1497    
1498          /* PCRE does not allow THEN to escape beyond an assertion; it is treated
1499          as NOMATCH. */
1500    
1501        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1502        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1503        }        }
1504      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1505    
1506      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1507    
1508      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1509      continue;      continue;
# Line 932  for (;;) Line 1521  for (;;)
1521        while (i-- > 0)        while (i-- > 0)
1522          {          {
1523          eptr--;          eptr--;
1524          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1525          BACKCHAR(eptr);          BACKCHAR(eptr);
1526          }          }
1527        }        }
# Line 943  for (;;) Line 1532  for (;;)
1532    
1533        {        {
1534        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1535        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1536        }        }
1537    
1538      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1539    
1540        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1541      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1542      break;      break;
1543    
# Line 959  for (;;) Line 1549  for (;;)
1549      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1550        {        {
1551        pcre_callout_block cb;        pcre_callout_block cb;
1552        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1553        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1554        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1555        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1556        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1557        cb.start_match      = mstart - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1558        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1559        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1560        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1561        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1562        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1563        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1564        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        cb.mark             = markptr;
1565          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1566        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1567        }        }
1568      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 981  for (;;) Line 1572  for (;;)
1572      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
1573      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1574    
1575      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1576      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
1577      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
1578      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
1579      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
1580      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
1581      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.  
1582    
1583      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
1584      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
1585      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1586        a lot, so he is not to blame for the current way it works. */
1587    
1588      case OP_RECURSE:      case OP_RECURSE:
1589        {        {
1590          recursion_info *ri;
1591          int recno;
1592    
1593        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1594        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1595          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1596    
1597          /* Check for repeating a recursion without advancing the subject pointer.
1598          This should catch convoluted mutual recursions. (Some simple cases are
1599          caught at compile time.) */
1600    
1601          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1602            if (recno == ri->group_num && eptr == ri->subject_position)
1603              RRETURN(PCRE_ERROR_RECURSELOOP);
1604    
1605        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1606    
1607          new_recursive.group_num = recno;
1608          new_recursive.subject_position = eptr;
1609        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1610        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1611    
1612        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1613    
1614        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1615    
1616        /* Now save the offset data. */        /* Now save the offset data */
1617    
1618        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1619        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 1023  for (;;) Line 1624  for (;;)
1624            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1625          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1626          }          }
   
1627        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1628              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = mstart;  
       mstart = eptr;  
1629    
1630        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1631        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1632          might be changed, so reset it before looping. */
1633    
1634        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1635        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1636        do        do
1637          {          {
1638            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1639          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1640            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1641          if (rrc == MATCH_MATCH)          memcpy(md->offset_vector, new_recursive.offset_save,
1642                new_recursive.saved_max * sizeof(int));
1643            md->recursive = new_recursive.prevrec;
1644            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1645            {            {
1646            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1647            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1648              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1649            RRETURN(MATCH_MATCH);  
1650              /* Set where we got to in the subject, and reset the start in case
1651              it was changed by \K. This *is* propagated back out of a recursion,
1652              for Perl compatibility. */
1653    
1654              eptr = md->end_match_ptr;
1655              mstart = md->start_match_ptr;
1656              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1657            }            }
1658          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)  
1659            /* PCRE does not allow THEN to escape beyond a recursion; it is treated
1660            as NOMATCH. */
1661    
1662            else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1663            {            {
1664            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1665              if (new_recursive.offset_save != stacksave)
1666                (pcre_free)(new_recursive.offset_save);
1667            RRETURN(rrc);            RRETURN(rrc);
1668            }            }
1669    
1670          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1671          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1672          }          }
1673        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1063  for (;;) Line 1676  for (;;)
1676        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1677        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1678          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1679        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
       }  
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer. */  
   
     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) break;  
       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
   
     /* If hit the end of the group (which could be repeated), fail */  
   
     if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);  
   
     /* Continue as from after the assertion, updating the offsets high water  
     mark, since extracts may have been taken. */  
   
     do ecode += GET(ecode, 1); while (*ecode == OP_ALT);  
   
     offset_top = md->end_offset_top;  
     eptr = md->end_match_ptr;  
   
     /* For a non-repeating ket, just continue at this level. This also  
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
       {  
       ecode += 1+LINK_SIZE;  
       break;  
1680        }        }
1681    
1682      /* The repeating kets try the rest of the pattern or restart from the      RECURSION_MATCHED:
1683      preceding bracket, in the appropriate order. The second "call" of match()      break;
     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 */  
1684    
1685      /* 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
1686      bracketed group and go to there. */      bracketed group and go to there. */
# Line 1148  for (;;) Line 1689  for (;;)
1689      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1690      break;      break;
1691    
1692      /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1693      that it may occur zero times. It may repeat infinitely, or not at all -      indicating that it may occur zero times. It may repeat infinitely, or not
1694      i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1695      repeat limits are compiled as a number of copies, with the optional ones      with fixed upper repeat limits are compiled as a number of copies, with the
1696      preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1697    
1698      case OP_BRAZERO:      case OP_BRAZERO:
1699        {      next = ecode + 1;
1700        next = ecode+1;      RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1701        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1702        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      do next += GET(next, 1); while (*next == OP_ALT);
1703        do next += GET(next,1); while (*next == OP_ALT);      ecode = next + 1 + LINK_SIZE;
       ecode = next + 1 + LINK_SIZE;  
       }  
1704      break;      break;
1705    
1706      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1707        {      next = ecode + 1;
1708        next = ecode+1;      do next += GET(next, 1); while (*next == OP_ALT);
1709        do next += GET(next, 1); while (*next == OP_ALT);      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1710        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1711        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      ecode++;
1712        ecode++;      break;
1713        }  
1714        case OP_SKIPZERO:
1715        next = ecode+1;
1716        do next += GET(next,1); while (*next == OP_ALT);
1717        ecode = next + 1 + LINK_SIZE;
1718      break;      break;
1719    
1720        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1721        here; just jump to the group, with allow_zero set TRUE. */
1722    
1723        case OP_BRAPOSZERO:
1724        op = *(++ecode);
1725        allow_zero = TRUE;
1726        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1727          goto POSSESSIVE_NON_CAPTURE;
1728    
1729      /* End of a group, repeated or non-repeating. */      /* End of a group, repeated or non-repeating. */
1730    
1731      case OP_KET:      case OP_KET:
1732      case OP_KETRMIN:      case OP_KETRMIN:
1733      case OP_KETRMAX:      case OP_KETRMAX:
1734        case OP_KETRPOS:
1735      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1736    
1737      /* 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
1738      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1739      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1740    
1741      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1742        {        {
1743        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1744        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
# Line 1194  for (;;) Line 1747  for (;;)
1747    
1748      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group, stop matching and return
1749      MATCH_MATCH, but record the current high water mark for use by positive      MATCH_MATCH, but record the current high water mark for use by positive
1750      assertions. Do this also for the "once" (atomic) groups. */      assertions. We also need to record the match start in case it was changed
1751        by \K. */
1752    
1753      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT)
         *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||  
         *prev == OP_ONCE)  
1754        {        {
1755        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1756        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1757        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1758          MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1759        }        }
1760    
1761      /* 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
1762      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1763      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
1764      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
1765      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
1766        the current subject position and start match pointer and give a MATCH
1767        return. */
1768    
1769      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1770            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1771        {        {
1772        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1773        offset = number << 1;        offset = number << 1;
1774    
1775  #ifdef DEBUG  #ifdef PCRE_DEBUG
1776        printf("end bracket %d", number);        printf("end bracket %d", number);
1777        printf("\n");        printf("\n");
1778  #endif  #endif
1779    
1780          /* Handle a recursively called group. */
1781    
1782          if (md->recursive != NULL && md->recursive->group_num == number)
1783            {
1784            md->end_match_ptr = eptr;
1785            md->start_match_ptr = mstart;
1786            RRETURN(MATCH_MATCH);
1787            }
1788    
1789          /* Deal with capturing */
1790    
1791        md->capture_last = number;        md->capture_last = number;
1792        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1793          {          {
1794            /* If offset is greater than offset_top, it means that we are
1795            "skipping" a capturing group, and that group's offsets must be marked
1796            unset. In earlier versions of PCRE, all the offsets were unset at the
1797            start of matching, but this doesn't work because atomic groups and
1798            assertions can cause a value to be set that should later be unset.
1799            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1800            part of the atomic group, but this is not on the final matching path,
1801            so must be unset when 2 is set. (If there is no group 2, there is no
1802            problem, because offset_top will then be 2, indicating no capture.) */
1803    
1804            if (offset > offset_top)
1805              {
1806              register int *iptr = md->offset_vector + offset_top;
1807              register int *iend = md->offset_vector + offset;
1808              while (iptr < iend) *iptr++ = -1;
1809              }
1810    
1811            /* Now make the extraction */
1812    
1813          md->offset_vector[offset] =          md->offset_vector[offset] =
1814            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1815          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1816          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1817          }          }
1818          }
1819    
1820        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1821        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1822        group. This is the forcible breaking of infinite loops as implemented in
1823        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1824        processing the rest of the pattern at a lower level. If this results in a
1825        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1826        bypassing intermediate backup points, but resetting any captures that
1827        happened along the way. */
1828    
1829        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1830          {
1831          if (*prev == OP_ONCE)
1832          {          {
1833          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1834          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1835          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1836          mstart = rec->save_start;          RRETURN(MATCH_ONCE);
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1837          }          }
1838          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1839          break;
1840        }        }
1841    
1842      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1843      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1844        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
1845    
1846      /* For a non-repeating ket, just continue at this level. This also      if (*ecode == OP_KETRPOS)
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
1847        {        {
1848        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1849        break;        md->end_offset_top = offset_top;
1850          RRETURN(MATCH_KETRPOS);
1851        }        }
1852    
1853      /* The repeating kets try the rest of the pattern or restart from the      /* The normal repeating kets try the rest of the pattern or restart from
1854      preceding bracket, in the appropriate order. In the second case, we can use      the preceding bracket, in the appropriate order. In the second case, we can
1855      tail recursion to avoid using another stack frame, unless we have an      use tail recursion to avoid using another stack frame, unless we have an
1856      unlimited repeat of a group that can match an empty string. */      an atomic group or an unlimited repeat of a group that can match an empty
1857        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1858    
1859      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1860        {        {
1861        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1862        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1863        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1864            {
1865            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1866            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1867            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1868            RRETURN(MATCH_ONCE);
1869            }
1870          if (*prev >= OP_SBRA)    /* Could match an empty string */
1871          {          {
1872          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1873            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1874          RRETURN(rrc);          RRETURN(rrc);
1875          }          }
1876        ecode = prev;        ecode = prev;
# Line 1286  for (;;) Line 1878  for (;;)
1878        }        }
1879      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1880        {        {
1881        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1882          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1883          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1884        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1885          if (*prev == OP_ONCE)
1886            {
1887            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1888            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1889            md->once_target = prev;
1890            RRETURN(MATCH_ONCE);
1891            }
1892        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1893        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1894        }        }
1895      /* Control never gets here */      /* Control never gets here */
1896    
1897      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1898    
1899      case OP_CIRC:      case OP_CIRC:
1900      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
     if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         RRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
1901    
1902      /* Start of subject assertion */      /* Start of subject assertion */
1903    
1904      case OP_SOD:      case OP_SOD:
1905      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1906        ecode++;
1907        break;
1908    
1909        /* Multiline mode: start of subject unless notbol, or after any newline. */
1910    
1911        case OP_CIRCM:
1912        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1913        if (eptr != md->start_subject &&
1914            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1915          MRRETURN(MATCH_NOMATCH);
1916      ecode++;      ecode++;
1917      break;      break;
1918    
1919      /* Start of match assertion */      /* Start of match assertion */
1920    
1921      case OP_SOM:      case OP_SOM:
1922      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1923      ecode++;      ecode++;
1924      break;      break;
1925    
# Line 1329  for (;;) Line 1930  for (;;)
1930      ecode++;      ecode++;
1931      break;      break;
1932    
1933      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1934      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1935    
1936      case OP_DOLL:      case OP_DOLLM:
1937      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1938        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
       if (eptr < md->end_subject)  
         { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }  
       else  
         { if (md->noteol) RRETURN(MATCH_NOMATCH); }  
       ecode++;  
       break;  
       }  
1939      else      else
1940        {        {
1941        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1942        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1943        }        }
1944        ecode++;
1945        break;
1946    
1947        /* Not multiline mode: assert before a terminating newline or before end of
1948        subject unless noteol is set. */
1949    
1950        case OP_DOLL:
1951        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1952        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1953    
1954      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1955    
1956      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1957    
1958      case OP_EOD:      case OP_EOD:
1959      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1960        SCHECK_PARTIAL();
1961      ecode++;      ecode++;
1962      break;      break;
1963    
1964      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1965    
1966      case OP_EODN:      case OP_EODN:
1967      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1968        if (eptr < md->end_subject &&
1969          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1970        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1971    
1972        /* Either at end of string or \n before end. */
1973    
1974        SCHECK_PARTIAL();
1975      ecode++;      ecode++;
1976      break;      break;
1977    
# Line 1380  for (;;) Line 1983  for (;;)
1983    
1984        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1985        It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to        It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to
1986        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1987          partial matching. */
1988    
1989  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1990        if (utf8)        if (utf8)
1991          {          {
1992            /* Get status of previous character */
1993    
1994          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1995            {            {
1996            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1997            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1998              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1999            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
2000    #ifdef SUPPORT_UCP
2001              if (md->use_ucp)
2002                {
2003                if (c == '_') prev_is_word = TRUE; else
2004                  {
2005                  int cat = UCD_CATEGORY(c);
2006                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2007                  }
2008                }
2009              else
2010    #endif
2011            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2012            }            }
2013          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
2014            /* Get status of next character */
2015    
2016            if (eptr >= md->end_subject)
2017              {
2018              SCHECK_PARTIAL();
2019              cur_is_word = FALSE;
2020              }
2021            else
2022            {            {
2023            GETCHAR(c, eptr);            GETCHAR(c, eptr);
2024    #ifdef SUPPORT_UCP
2025              if (md->use_ucp)
2026                {
2027                if (c == '_') cur_is_word = TRUE; else
2028                  {
2029                  int cat = UCD_CATEGORY(c);
2030                  cur_is_word = (cat == ucp_L || cat == ucp_N);
2031                  }
2032                }
2033              else
2034    #endif
2035            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
2036            }            }
2037          }          }
2038        else        else
2039  #endif  #endif
2040    
2041        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
2042          consistency with the behaviour of \w we do use it in this case. */
2043    
2044          {          {
2045          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
2046            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
2047          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
2048            ((md->ctypes[*eptr] & ctype_word) != 0);            {
2049              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
2050    #ifdef SUPPORT_UCP
2051              if (md->use_ucp)
2052                {
2053                c = eptr[-1];
2054                if (c == '_') prev_is_word = TRUE; else
2055                  {
2056                  int cat = UCD_CATEGORY(c);
2057                  prev_is_word = (cat == ucp_L || cat == ucp_N);
2058                  }
2059                }
2060              else
2061    #endif
2062              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
2063              }
2064    
2065            /* Get status of next character */
2066    
2067            if (eptr >= md->end_subject)
2068              {
2069              SCHECK_PARTIAL();
2070              cur_is_word = FALSE;
2071              }
2072            else
2073    #ifdef SUPPORT_UCP
2074            if (md->use_ucp)
2075              {
2076              c = *eptr;
2077              if (c == '_') cur_is_word = TRUE; else
2078                {
2079                int cat = UCD_CATEGORY(c);
2080                cur_is_word = (cat == ucp_L || cat == ucp_N);
2081                }
2082              }
2083            else
2084    #endif
2085            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2086          }          }
2087    
2088        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2089    
2090        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2091             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2092          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2093        }        }
2094      break;      break;
2095    
2096      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2097    
2098      case OP_ANY:      case OP_ANY:
2099      if ((ims & PCRE_DOTALL) == 0)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2100        {      /* Fall through */
2101        if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);  
2102        case OP_ALLANY:
2103        if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2104          {                            /* not be updated before SCHECK_PARTIAL. */
2105          SCHECK_PARTIAL();
2106          MRRETURN(MATCH_NOMATCH);
2107        }        }
2108      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      eptr++;
2109      if (utf8)      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
2110      ecode++;      ecode++;
2111      break;      break;
2112    
# Line 1435  for (;;) Line 2114  for (;;)
2114      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2115    
2116      case OP_ANYBYTE:      case OP_ANYBYTE:
2117      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2118          {                            /* not be updated before SCHECK_PARTIAL. */
2119          SCHECK_PARTIAL();
2120          MRRETURN(MATCH_NOMATCH);
2121          }
2122        eptr++;
2123      ecode++;      ecode++;
2124      break;      break;
2125    
2126      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2127      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2128          {
2129          SCHECK_PARTIAL();
2130          MRRETURN(MATCH_NOMATCH);
2131          }
2132      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2133      if (      if (
2134  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1448  for (;;) Line 2136  for (;;)
2136  #endif  #endif
2137         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2138         )         )
2139        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2140      ecode++;      ecode++;
2141      break;      break;
2142    
2143      case OP_DIGIT:      case OP_DIGIT:
2144      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2145          {
2146          SCHECK_PARTIAL();
2147          MRRETURN(MATCH_NOMATCH);
2148          }
2149      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2150      if (      if (
2151  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1461  for (;;) Line 2153  for (;;)
2153  #endif  #endif
2154         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2155         )         )
2156        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2157      ecode++;      ecode++;
2158      break;      break;
2159    
2160      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2161      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2162          {
2163          SCHECK_PARTIAL();
2164          MRRETURN(MATCH_NOMATCH);
2165          }
2166      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2167      if (      if (
2168  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1474  for (;;) Line 2170  for (;;)
2170  #endif  #endif
2171         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2172         )         )
2173        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2174      ecode++;      ecode++;
2175      break;      break;
2176    
2177      case OP_WHITESPACE:      case OP_WHITESPACE:
2178      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2179          {
2180          SCHECK_PARTIAL();
2181          MRRETURN(MATCH_NOMATCH);
2182          }
2183      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2184      if (      if (
2185  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1487  for (;;) Line 2187  for (;;)
2187  #endif  #endif
2188         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2189         )         )
2190        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2191      ecode++;      ecode++;
2192      break;      break;
2193    
2194      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2195      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2196          {
2197          SCHECK_PARTIAL();
2198          MRRETURN(MATCH_NOMATCH);
2199          }
2200      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2201      if (      if (
2202  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1500  for (;;) Line 2204  for (;;)
2204  #endif  #endif
2205         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2206         )         )
2207        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2208      ecode++;      ecode++;
2209      break;      break;
2210    
2211      case OP_WORDCHAR:      case OP_WORDCHAR:
2212      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2213          {
2214          SCHECK_PARTIAL();
2215          MRRETURN(MATCH_NOMATCH);
2216          }
2217      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2218      if (      if (
2219  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1513  for (;;) Line 2221  for (;;)
2221  #endif  #endif
2222         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2223         )         )
2224        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2225      ecode++;      ecode++;
2226      break;      break;
2227    
2228      case OP_ANYNL:      case OP_ANYNL:
2229      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2230          {
2231          SCHECK_PARTIAL();
2232          MRRETURN(MATCH_NOMATCH);
2233          }
2234      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2235      switch(c)      switch(c)
2236        {        {
2237        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2238    
2239        case 0x000d:        case 0x000d:
2240        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2241        break;        break;
2242    
2243        case 0x000a:        case 0x000a:
2244          break;
2245    
2246        case 0x000b:        case 0x000b:
2247        case 0x000c:        case 0x000c:
2248        case 0x0085:        case 0x0085:
2249        case 0x2028:        case 0x2028:
2250        case 0x2029:        case 0x2029:
2251          if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2252        break;        break;
2253        }        }
2254      ecode++;      ecode++;
2255      break;      break;
2256    
2257      case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
2258      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2259          {
2260          SCHECK_PARTIAL();
2261          MRRETURN(MATCH_NOMATCH);
2262          }
2263      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2264      switch(c)      switch(c)
2265        {        {
# Line 1562  for (;;) Line 2283  for (;;)
2283        case 0x202f:    /* NARROW NO-BREAK SPACE */        case 0x202f:    /* NARROW NO-BREAK SPACE */
2284        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2285        case 0x3000:    /* IDEOGRAPHIC SPACE */        case 0x3000:    /* IDEOGRAPHIC SPACE */
2286        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2287        }        }
2288      ecode++;      ecode++;
2289      break;      break;
2290    
2291      case OP_HSPACE:      case OP_HSPACE:
2292      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2293          {
2294          SCHECK_PARTIAL();
2295          MRRETURN(MATCH_NOMATCH);
2296          }
2297      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2298      switch(c)      switch(c)
2299        {        {
2300        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2301        case 0x09:      /* HT */        case 0x09:      /* HT */
2302        case 0x20:      /* SPACE */        case 0x20:      /* SPACE */
2303        case 0xa0:      /* NBSP */        case 0xa0:      /* NBSP */
# Line 1598  for (;;) Line 2323  for (;;)
2323      break;      break;
2324    
2325      case OP_NOT_VSPACE:      case OP_NOT_VSPACE:
2326      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2327          {
2328          SCHECK_PARTIAL();
2329          MRRETURN(MATCH_NOMATCH);
2330          }
2331      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2332      switch(c)      switch(c)
2333        {        {
# Line 1610  for (;;) Line 2339  for (;;)
2339        case 0x85:      /* NEL */        case 0x85:      /* NEL */
2340        case 0x2028:    /* LINE SEPARATOR */        case 0x2028:    /* LINE SEPARATOR */
2341        case 0x2029:    /* PARAGRAPH SEPARATOR */        case 0x2029:    /* PARAGRAPH SEPARATOR */
2342        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2343        }        }
2344      ecode++;      ecode++;
2345      break;      break;
2346    
2347      case OP_VSPACE:      case OP_VSPACE:
2348      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2349          {
2350          SCHECK_PARTIAL();
2351          MRRETURN(MATCH_NOMATCH);
2352          }
2353      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2354      switch(c)      switch(c)
2355        {        {
2356        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2357        case 0x0a:      /* LF */        case 0x0a:      /* LF */
2358        case 0x0b:      /* VT */        case 0x0b:      /* VT */
2359        case 0x0c:      /* FF */        case 0x0c:      /* FF */
# Line 1639  for (;;) Line 2372  for (;;)
2372    
2373      case OP_PROP:      case OP_PROP:
2374      case OP_NOTPROP:      case OP_NOTPROP:
2375      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2376          {
2377          SCHECK_PARTIAL();
2378          MRRETURN(MATCH_NOMATCH);
2379          }
2380      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2381        {        {
2382        int chartype, script;        const ucd_record *prop = GET_UCD(c);
       int category = _pcre_ucp_findprop(c, &chartype, &script);  
2383    
2384        switch(ecode[1])        switch(ecode[1])
2385          {          {
2386          case PT_ANY:          case PT_ANY:
2387          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2388          break;          break;
2389    
2390          case PT_LAMP:          case PT_LAMP:
2391          if ((chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2392               chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2393               chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2394            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2395           break;          break;
2396    
2397          case PT_GC:          case PT_GC:
2398          if ((ecode[2] != category) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2399            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2400          break;          break;
2401    
2402          case PT_PC:          case PT_PC:
2403          if ((ecode[2] != chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2404            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2405          break;          break;
2406    
2407          case PT_SC:          case PT_SC:
2408          if ((ecode[2] != script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2409            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2410          break;          break;
2411    
2412            /* These are specials */
2413    
2414            case PT_ALNUM:
2415            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2416                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2417              MRRETURN(MATCH_NOMATCH);
2418            break;
2419    
2420            case PT_SPACE:    /* Perl space */
2421            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2422                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2423                   == (op == OP_NOTPROP))
2424              MRRETURN(MATCH_NOMATCH);
2425            break;
2426    
2427            case PT_PXSPACE:  /* POSIX space */
2428            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2429                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2430                 c == CHAR_FF || c == CHAR_CR)
2431                   == (op == OP_NOTPROP))
2432              MRRETURN(MATCH_NOMATCH);
2433            break;
2434    
2435            case PT_WORD:
2436            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2437                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2438                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2439              MRRETURN(MATCH_NOMATCH);
2440            break;
2441    
2442            /* This should never occur */
2443    
2444          default:          default:
2445          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2446          }          }
# Line 1685  for (;;) Line 2453  for (;;)
2453      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2454    
2455      case OP_EXTUNI:      case OP_EXTUNI:
2456      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2457          {
2458          SCHECK_PARTIAL();
2459          MRRETURN(MATCH_NOMATCH);
2460          }
2461      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2462        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2463        while (eptr < md->end_subject)
2464        {        {
2465        int chartype, script;        int len = 1;
2466        int category = _pcre_ucp_findprop(c, &chartype, &script);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2467        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (UCD_CATEGORY(c) != ucp_M) break;
2468        while (eptr < md->end_subject)        eptr += len;
         {  
         int len = 1;  
         if (!utf8) c = *eptr; else  
           {  
           GETCHARLEN(c, eptr, len);  
           }  
         category = _pcre_ucp_findprop(c, &chartype, &script);  
         if (category != ucp_M) break;  
         eptr += len;  
         }  
2469        }        }
2470      ecode++;      ecode++;
2471      break;      break;
# Line 1717  for (;;) Line 2481  for (;;)
2481      loops). */      loops). */
2482    
2483      case OP_REF:      case OP_REF:
2484        {      case OP_REFI:
2485        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2486        ecode += 3;                                 /* Advance past item */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2487        ecode += 3;
2488    
2489        /* If the reference is unset, set the length to be longer than the amount      /* If the reference is unset, there are two possibilities:
       of subject left; this ensures that every attempt at a match fails. We  
       can't just fail here, because of the possibility of quantifiers with zero  
       minima. */  
   
       length = (offset >= offset_top || md->offset_vector[offset] < 0)?  
         md->end_subject - eptr + 1 :  
         md->offset_vector[offset+1] - md->offset_vector[offset];  
2490    
2491        /* Set up for repetition, or handle the non-repeated case */      (a) In the default, Perl-compatible state, set the length negative;
2492        this ensures that every attempt at a match fails. We can't just fail
2493        here, because of the possibility of quantifiers with zero minima.
2494    
2495        switch (*ecode)      (b) If the JavaScript compatibility flag is set, set the length to zero
2496          {      so that the back reference matches an empty string.
         case OP_CRSTAR:  
         case OP_CRMINSTAR:  
         case OP_CRPLUS:  
         case OP_CRMINPLUS:  
         case OP_CRQUERY:  
         case OP_CRMINQUERY:  
         c = *ecode++ - OP_CRSTAR;  
         minimize = (c & 1) != 0;  
         min = rep_min[c];                 /* Pick up values from tables; */  
         max = rep_max[c];                 /* zero for max => infinity */  
         if (max == 0) max = INT_MAX;  
         break;  
2497    
2498          case OP_CRRANGE:      Otherwise, set the length to the length of what was matched by the
2499          case OP_CRMINRANGE:      referenced subpattern. */
         minimize = (*ecode == OP_CRMINRANGE);  
         min = GET2(ecode, 1);  
         max = GET2(ecode, 3);  
         if (max == 0) max = INT_MAX;  
         ecode += 5;  
         break;  
2500    
2501          default:               /* No repeat follows */      if (offset >= offset_top || md->offset_vector[offset] < 0)
2502          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);        length = (md->jscript_compat)? 0 : -1;
2503          eptr += length;      else
2504          continue;              /* With the main loop */        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2505    
2506        /* Set up for repetition, or handle the non-repeated case */
2507    
2508        switch (*ecode)
2509          {
2510          case OP_CRSTAR:
2511          case OP_CRMINSTAR:
2512          case OP_CRPLUS:
2513          case OP_CRMINPLUS:
2514          case OP_CRQUERY:
2515          case OP_CRMINQUERY:
2516          c = *ecode++ - OP_CRSTAR;
2517          minimize = (c & 1) != 0;
2518          min = rep_min[c];                 /* Pick up values from tables; */
2519          max = rep_max[c];                 /* zero for max => infinity */
2520          if (max == 0) max = INT_MAX;
2521          break;
2522    
2523          case OP_CRRANGE:
2524          case OP_CRMINRANGE:
2525          minimize = (*ecode == OP_CRMINRANGE);
2526          min = GET2(ecode, 1);
2527          max = GET2(ecode, 3);
2528          if (max == 0) max = INT_MAX;
2529          ecode += 5;
2530          break;
2531    
2532          default:               /* No repeat follows */
2533          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2534            {
2535            CHECK_PARTIAL();
2536            MRRETURN(MATCH_NOMATCH);
2537          }          }
2538          eptr += length;
2539          continue;              /* With the main loop */
2540          }
2541    
2542        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2543        main loop. */      zero, just continue with the main loop. */
2544    
2545        if (length == 0) continue;      if (length == 0) continue;
2546    
2547        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2548        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2549        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2550    
2551        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2552          {
2553          int slength;
2554          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2555          {          {
2556          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2557          eptr += length;          MRRETURN(MATCH_NOMATCH);
2558          }          }
2559          eptr += slength;
2560          }
2561    
2562        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2563        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2564    
2565        if (min == max) continue;      if (min == max) continue;
2566    
2567        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2568    
2569        if (minimize)      if (minimize)
2570          {
2571          for (fi = min;; fi++)
2572          {          {
2573          for (fi = min;; fi++)          int slength;
2574            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2575            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2576            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2577            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2578            {            {
2579            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2580            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2581            }            }
2582          /* Control never gets here */          eptr += slength;
2583          }          }
2584          /* Control never gets here */
2585          }
2586    
2587        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2588    
2589        else      else
2590          {
2591          pp = eptr;
2592          for (i = min; i < max; i++)
2593          {          {
2594          pp = eptr;          int slength;
2595          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
           {  
           if (!match_ref(offset, eptr, length, md, ims)) break;  
           eptr += length;  
           }  
         while (eptr >= pp)  
2596            {            {
2597            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2598            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2599            }            }
2600          RRETURN(MATCH_NOMATCH);          eptr += slength;
2601            }
2602          while (eptr >= pp)
2603            {
2604            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2605            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2606            eptr -= length;
2607          }          }
2608          MRRETURN(MATCH_NOMATCH);
2609        }        }
2610      /* Control never gets here */      /* Control never gets here */
2611    
   
   
2612      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2613      used when all the characters in the class have values in the range 0-255,      used when all the characters in the class have values in the range 0-255,
2614      and either the matching is caseful, or the characters are in the range      and either the matching is caseful, or the characters are in the range
# Line 1874  for (;;) Line 2663  for (;;)
2663          {          {
2664          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2665            {            {
2666            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2667                {
2668                SCHECK_PARTIAL();
2669                MRRETURN(MATCH_NOMATCH);
2670                }
2671            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2672            if (c > 255)            if (c > 255)
2673              {              {
2674              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2675              }              }
2676            else            else
2677              {              {
2678              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2679              }              }
2680            }            }
2681          }          }
# Line 1892  for (;;) Line 2685  for (;;)
2685          {          {
2686          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2687            {            {
2688            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2689                {
2690                SCHECK_PARTIAL();
2691                MRRETURN(MATCH_NOMATCH);
2692                }
2693            c = *eptr++;            c = *eptr++;
2694            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2695            }            }
2696          }          }
2697    
# Line 1914  for (;;) Line 2711  for (;;)
2711            {            {
2712            for (fi = min;; fi++)            for (fi = min;; fi++)
2713              {              {
2714              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2715              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2716              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2717                if (eptr >= md->end_subject)
2718                  {
2719                  SCHECK_PARTIAL();
2720                  MRRETURN(MATCH_NOMATCH);
2721                  }
2722              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2723              if (c > 255)              if (c > 255)
2724                {                {
2725                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2726                }                }
2727              else              else
2728                {                {
2729                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2730                }                }
2731              }              }
2732            }            }
# Line 1934  for (;;) Line 2736  for (;;)
2736            {            {
2737            for (fi = min;; fi++)            for (fi = min;; fi++)
2738              {              {
2739              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2740              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2741              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2742                if (eptr >= md->end_subject)
2743                  {
2744                  SCHECK_PARTIAL();
2745                  MRRETURN(MATCH_NOMATCH);
2746                  }
2747              c = *eptr++;              c = *eptr++;
2748              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2749              }              }
2750            }            }
2751          /* Control never gets here */          /* Control never gets here */
# Line 1957  for (;;) Line 2764  for (;;)
2764            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2765              {              {
2766              int len = 1;              int len = 1;
2767              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2768                  {
2769                  SCHECK_PARTIAL();
2770                  break;
2771                  }
2772              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2773              if (c > 255)              if (c > 255)
2774                {                {
# Line 1971  for (;;) Line 2782  for (;;)
2782              }              }
2783            for (;;)            for (;;)
2784              {              {
2785              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2786              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2787              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2788              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1983  for (;;) Line 2794  for (;;)
2794            {            {
2795            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2796              {              {
2797              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2798                  {
2799                  SCHECK_PARTIAL();
2800                  break;
2801                  }
2802              c = *eptr;              c = *eptr;
2803              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2804              eptr++;              eptr++;
2805              }              }
2806            while (eptr >= pp)            while (eptr >= pp)
2807              {              {
2808              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2809              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2810              eptr--;              eptr--;
2811              }              }
2812            }            }
2813    
2814          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2815          }          }
2816        }        }
2817      /* Control never gets here */      /* Control never gets here */
2818    
2819    
2820      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2821      in UTF-8 mode, because that's the only time it is compiled. */      when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8
2822        mode, because Unicode properties are supported in non-UTF-8 mode. */
2823    
2824  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2825      case OP_XCLASS:      case OP_XCLASS:
# Line 2044  for (;;) Line 2860  for (;;)
2860    
2861        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2862          {          {
2863          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2864          GETCHARINC(c, eptr);            {
2865          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2866              MRRETURN(MATCH_NOMATCH);
2867              }
2868            GETCHARINCTEST(c, eptr);
2869            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2870          }          }
2871    
2872        /* If max == min we can continue with the main loop without the        /* If max == min we can continue with the main loop without the
# Line 2061  for (;;) Line 2881  for (;;)
2881          {          {
2882          for (fi = min;; fi++)          for (fi = min;; fi++)
2883            {            {
2884            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2885            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2886            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2887            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2888            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2889                SCHECK_PARTIAL();
2890                MRRETURN(MATCH_NOMATCH);
2891                }
2892              GETCHARINCTEST(c, eptr);
2893              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2894            }            }
2895          /* Control never gets here */          /* Control never gets here */
2896          }          }
# Line 2078  for (;;) Line 2903  for (;;)
2903          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2904            {            {
2905            int len = 1;            int len = 1;
2906            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2907            GETCHARLEN(c, eptr, len);              {
2908                SCHECK_PARTIAL();
2909                break;
2910                }
2911              GETCHARLENTEST(c, eptr, len);
2912            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2913            eptr += len;            eptr += len;
2914            }            }
2915          for(;;)          for(;;)
2916            {            {
2917            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2918            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2919            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2920            if (utf8) BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
2921            }            }
2922          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2923          }          }
2924    
2925        /* Control never gets here */        /* Control never gets here */
# Line 2106  for (;;) Line 2935  for (;;)
2935        length = 1;        length = 1;
2936        ecode++;        ecode++;
2937        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2938        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2939        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2940            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2941            MRRETURN(MATCH_NOMATCH);
2942            }
2943          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2944        }        }
2945      else      else
2946  #endif  #endif
2947    
2948      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2949        {        {
2950        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2951        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2952            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2953            MRRETURN(MATCH_NOMATCH);
2954            }
2955          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2956        ecode += 2;        ecode += 2;
2957        }        }
2958      break;      break;
2959    
2960      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2961    
2962      case OP_CHARNC:      case OP_CHARI:
2963  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2964      if (utf8)      if (utf8)
2965        {        {
# Line 2130  for (;;) Line 2967  for (;;)
2967        ecode++;        ecode++;
2968        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2969    
2970        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2971            {
2972            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2973            MRRETURN(MATCH_NOMATCH);
2974            }
2975    
2976        /* If the pattern character's value is < 128, we have only one byte, and        /* If the pattern character's value is < 128, we have only one byte, and
2977        can use the fast lookup table. */        can use the fast lookup table. */
2978    
2979        if (fc < 128)        if (fc < 128)
2980          {          {
2981          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2982          }          }
2983    
2984        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
# Line 2154  for (;;) Line 2995  for (;;)
2995          if (fc != dc)          if (fc != dc)
2996            {            {
2997  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2998            if (dc != _pcre_ucp_othercase(fc))            if (dc != UCD_OTHERCASE(fc))
2999  #endif  #endif
3000              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
3001            }            }
3002          }          }
3003        }        }
# Line 2165  for (;;) Line 3006  for (;;)
3006    
3007      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
3008        {        {
3009        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
3010        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3011            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
3012            MRRETURN(MATCH_NOMATCH);
3013            }
3014          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3015        ecode += 2;        ecode += 2;
3016        }        }
3017      break;      break;
# Line 2174  for (;;) Line 3019  for (;;)
3019      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
3020    
3021      case OP_EXACT:      case OP_EXACT:
3022        case OP_EXACTI:
3023      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3024      ecode += 3;      ecode += 3;
3025      goto REPEATCHAR;      goto REPEATCHAR;
3026    
3027      case OP_POSUPTO:      case OP_POSUPTO:
3028        case OP_POSUPTOI:
3029      possessive = TRUE;      possessive = TRUE;
3030      /* Fall through */      /* Fall through */
3031    
3032      case OP_UPTO:      case OP_UPTO:
3033        case OP_UPTOI:
3034      case OP_MINUPTO:      case OP_MINUPTO:
3035        case OP_MINUPTOI:
3036      min = 0;      min = 0;
3037      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3038      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
3039      ecode += 3;      ecode += 3;
3040      goto REPEATCHAR;      goto REPEATCHAR;
3041    
3042      case OP_POSSTAR:      case OP_POSSTAR:
3043        case OP_POSSTARI:
3044      possessive = TRUE;      possessive = TRUE;
3045      min = 0;      min = 0;
3046      max = INT_MAX;      max = INT_MAX;
# Line 2198  for (;;) Line 3048  for (;;)
3048      goto REPEATCHAR;      goto REPEATCHAR;
3049    
3050      case OP_POSPLUS:      case OP_POSPLUS:
3051        case OP_POSPLUSI:
3052      possessive = TRUE;      possessive = TRUE;
3053      min = 1;      min = 1;
3054      max = INT_MAX;      max = INT_MAX;
# Line 2205  for (;;) Line 3056  for (;;)
3056      goto REPEATCHAR;      goto REPEATCHAR;
3057    
3058      case OP_POSQUERY:      case OP_POSQUERY:
3059        case OP_POSQUERYI:
3060      possessive = TRUE;      possessive = TRUE;
3061      min = 0;      min = 0;
3062      max = 1;      max = 1;
# Line 2212  for (;;) Line 3064  for (;;)
3064      goto REPEATCHAR;      goto REPEATCHAR;
3065    
3066      case OP_STAR:      case OP_STAR:
3067        case OP_STARI:
3068      case OP_MINSTAR:      case OP_MINSTAR:
3069        case OP_MINSTARI:
3070      case OP_PLUS:      case OP_PLUS:
3071        case OP_PLUSI:
3072      case OP_MINPLUS:      case OP_MINPLUS:
3073        case OP_MINPLUSI:
3074      case OP_QUERY:      case OP_QUERY:
3075        case OP_QUERYI:
3076      case OP_MINQUERY:      case OP_MINQUERY:
3077      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3078        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3079      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3080      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3081      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3082      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3083    
3084      /* Common code for all repeated single-character matches. We can give      /* Common code for all repeated single-character matches. */
     up quickly if there are fewer than the minimum number of characters left in  
     the subject. */  
3085    
3086      REPEATCHAR:      REPEATCHAR:
3087  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2234  for (;;) Line 3090  for (;;)
3090        length = 1;        length = 1;
3091        charptr = ecode;        charptr = ecode;
3092        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3093        ecode += length;        ecode += length;
3094    
3095        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 2244  for (;;) Line 3099  for (;;)
3099          {          {
3100  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3101          unsigned int othercase;          unsigned int othercase;
3102          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3103              (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)              (othercase = UCD_OTHERCASE(fc)) != fc)
3104            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3105          else oclength = 0;          else oclength = 0;
3106  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3107    
3108          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3109            {            {
3110            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3111                memcmp(eptr, charptr, length) == 0) eptr += length;
3112  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3113            /* Need braces because of following else */            else if (oclength > 0 &&
3114            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                     eptr <= md->end_subject - oclength &&
3115                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3116    #endif  /* SUPPORT_UCP */
3117            else            else
3118              {              {
3119              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3120              eptr += oclength;              MRRETURN(MATCH_NOMATCH);
3121              }              }
 #else   /* without SUPPORT_UCP */  
           else { RRETURN(MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3122            }            }
3123    
3124          if (min == max) continue;          if (min == max) continue;
# Line 2272  for (;;) Line 3127  for (;;)
3127            {            {
3128            for (fi = min;; fi++)            for (fi = min;; fi++)
3129              {              {
3130              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3131              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3132              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3133              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3134                  memcmp(eptr, charptr, length) == 0) eptr += length;
3135  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3136              /* Need braces because of following else */              else if (oclength > 0 &&
3137              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                       eptr <= md->end_subject - oclength &&
3138                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3139    #endif  /* SUPPORT_UCP */
3140              else              else
3141                {                {
3142                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3143                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3144                }                }
 #else   /* without SUPPORT_UCP */  
             else { RRETURN (MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3145              }              }
3146            /* Control never gets here */            /* Control never gets here */
3147            }            }
# Line 2296  for (;;) Line 3151  for (;;)
3151            pp = eptr;            pp = eptr;
3152            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3153              {              {
3154              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3155              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3156  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3157              else if (oclength == 0) break;              else if (oclength > 0 &&
3158                         eptr <= md->end_subject - oclength &&
3159                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3160    #endif  /* SUPPORT_UCP */
3161              else              else
3162                {                {
3163                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3164                eptr += oclength;                break;
3165                }                }
 #else   /* without SUPPORT_UCP */  
             else break;  
 #endif  /* SUPPORT_UCP */  
3166              }              }
3167    
3168            if (possessive) continue;            if (possessive) continue;
3169    
3170            for(;;)            for(;;)
3171             {              {
3172             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3173             if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3174             if (eptr == pp) RRETURN(MATCH_NOMATCH);              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3175  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3176             eptr--;              eptr--;
3177             BACKCHAR(eptr);              BACKCHAR(eptr);
3178  #else   /* without SUPPORT_UCP */  #else   /* without SUPPORT_UCP */
3179             eptr -= length;              eptr -= length;
3180  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3181             }              }
3182            }            }
3183          /* Control never gets here */          /* Control never gets here */
3184          }          }
# Line 2335  for (;;) Line 3191  for (;;)
3191  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3192    
3193      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3194        {  
3195        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3196    
3197      /* The value of fc at this point is always less than 256, though we may or      /* The value of fc at this point is always less than 256, though we may or
3198      may not be in UTF-8 mode. The code is duplicated for the caseless and      may not be in UTF-8 mode. The code is duplicated for the caseless and
# Line 2352  for (;;) Line 3206  for (;;)
3206      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3207        max, eptr));        max, eptr));
3208    
3209      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3210        {        {
3211        fc = md->lcc[fc];        fc = md->lcc[fc];
3212        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3213          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3214            if (eptr >= md->end_subject)
3215              {
3216              SCHECK_PARTIAL();
3217              MRRETURN(MATCH_NOMATCH);
3218              }
3219            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3220            }
3221        if (min == max) continue;        if (min == max) continue;
3222        if (minimize)        if (minimize)
3223          {          {
3224          for (fi = min;; fi++)          for (fi = min;; fi++)
3225            {            {
3226            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3227            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3228            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3229                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3230              RRETURN(MATCH_NOMATCH);              {
3231                SCHECK_PARTIAL();
3232                MRRETURN(MATCH_NOMATCH);
3233                }
3234              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3235            }            }
3236          /* Control never gets here */          /* Control never gets here */
3237          }          }
# Line 2375  for (;;) Line 3240  for (;;)
3240          pp = eptr;          pp = eptr;
3241          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3242            {            {
3243            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3244                {
3245                SCHECK_PARTIAL();
3246                break;
3247                }
3248              if (fc != md->lcc[*eptr]) break;
3249            eptr++;            eptr++;
3250            }            }
3251    
3252          if (possessive) continue;          if (possessive) continue;
3253    
3254          while (eptr >= pp)          while (eptr >= pp)
3255            {            {
3256            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3257            eptr--;            eptr--;
3258            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3259            }            }
3260          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3261          }          }
3262        /* Control never gets here */        /* Control never gets here */
3263        }        }
# Line 2394  for (;;) Line 3266  for (;;)
3266    
3267      else      else
3268        {        {
3269        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3270            {
3271            if (eptr >= md->end_subject)
3272              {
3273              SCHECK_PARTIAL();
3274              MRRETURN(MATCH_NOMATCH);
3275              }
3276            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3277            }
3278    
3279        if (min == max) continue;        if (min == max) continue;
3280    
3281        if (minimize)        if (minimize)
3282          {          {
3283          for (fi = min;; fi++)          for (fi = min;; fi++)
3284            {            {
3285            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3286            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3287            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3288              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3289                {
3290                SCHECK_PARTIAL();
3291                MRRETURN(MATCH_NOMATCH);
3292                }
3293              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3294            }            }
3295          /* Control never gets here */          /* Control never gets here */
3296          }          }
# Line 2412  for (;;) Line 3299  for (;;)
3299          pp = eptr;          pp = eptr;
3300          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3301            {            {
3302            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3303                {
3304                SCHECK_PARTIAL();
3305                break;
3306                }
3307              if (fc != *eptr) break;
3308            eptr++;            eptr++;
3309            }            }
3310          if (possessive) continue;          if (possessive) continue;
3311    
3312          while (eptr >= pp)          while (eptr >= pp)
3313            {            {
3314            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3315            eptr--;            eptr--;
3316            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3317            }            }
3318          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3319          }          }
3320        }        }
3321      /* Control never gets here */      /* Control never gets here */
# Line 2431  for (;;) Line 3324  for (;;)
3324      checking can be multibyte. */      checking can be multibyte. */
3325    
3326      case OP_NOT:      case OP_NOT:
3327      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3328        if (eptr >= md->end_subject)
3329          {
3330          SCHECK_PARTIAL();
3331          MRRETURN(MATCH_NOMATCH);
3332          }
3333      ecode++;      ecode++;
3334      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3335      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3336        {        {
3337  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3338        if (c < 256)        if (c < 256)
3339  #endif  #endif
3340        c = md->lcc[c];        c = md->lcc[c];
3341        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3342        }        }
3343      else      else    /* Caseful */
3344        {        {
3345        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3346        }        }
3347      break;      break;
3348    
# Line 2456  for (;;) Line 3354  for (;;)
3354      about... */      about... */
3355    
3356      case OP_NOTEXACT:      case OP_NOTEXACT:
3357        case OP_NOTEXACTI:
3358      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3359      ecode += 3;      ecode += 3;
3360      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3361    
3362      case OP_NOTUPTO:      case OP_NOTUPTO:
3363        case OP_NOTUPTOI:
3364      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3365        case OP_NOTMINUPTOI:
3366      min = 0;      min = 0;
3367      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3368      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3369      ecode += 3;      ecode += 3;
3370      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3371    
3372      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3373        case OP_NOTPOSSTARI:
3374      possessive = TRUE;      possessive = TRUE;
3375      min = 0;      min = 0;
3376      max = INT_MAX;      max = INT_MAX;
# Line 2476  for (;;) Line 3378  for (;;)
3378      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3379    
3380      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3381        case OP_NOTPOSPLUSI:
3382      possessive = TRUE;      possessive = TRUE;
3383      min = 1;      min = 1;
3384      max = INT_MAX;      max = INT_MAX;
# Line 2483  for (;;) Line 3386  for (;;)
3386      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3387    
3388      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3389        case OP_NOTPOSQUERYI:
3390      possessive = TRUE;      possessive = TRUE;
3391      min = 0;      min = 0;
3392      max = 1;      max = 1;
# Line 2490  for (;;) Line 3394  for (;;)
3394      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3395    
3396      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3397        case OP_NOTPOSUPTOI:
3398      possessive = TRUE;      possessive = TRUE;
3399      min = 0;      min = 0;
3400      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 2497  for (;;) Line 3402  for (;;)
3402      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3403    
3404      case OP_NOTSTAR:      case OP_NOTSTAR:
3405        case OP_NOTSTARI:
3406      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3407        case OP_NOTMINSTARI:
3408      case OP_NOTPLUS:      case OP_NOTPLUS:
3409        case OP_NOTPLUSI:
3410      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3411        case OP_NOTMINPLUSI:
3412      case OP_NOTQUERY:      case OP_NOTQUERY:
3413        case OP_NOTQUERYI:
3414      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3415      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3416        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3417      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3418      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3419      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3420      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3421    
3422      /* Common code for all repeated single-byte matches. We can give up quickly      /* Common code for all repeated single-byte matches. */
     if there are fewer than the minimum number of bytes left in the  
     subject. */  
3423    
3424      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3425      fc = *ecode++;      fc = *ecode++;
3426    
3427      /* The code is duplicated for the caseless and caseful cases, for speed,      /* The code is duplicated for the caseless and caseful cases, for speed,
# Line 2527  for (;;) Line 3435  for (;;)
3435      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,
3436        max, eptr));        max, eptr));
3437    
3438      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3439        {        {
3440        fc = md->lcc[fc];        fc = md->lcc[fc];
3441    
# Line 2538  for (;;) Line 3446  for (;;)
3446          register unsigned int d;          register unsigned int d;
3447          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3448            {            {
3449              if (eptr >= md->end_subject)
3450                {
3451                SCHECK_PARTIAL();
3452                MRRETURN(MATCH_NOMATCH);
3453                }
3454            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3455            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3456            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3457            }            }
3458          }          }
3459        else        else
# Line 2549  for (;;) Line 3462  for (;;)
3462        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3463          {          {
3464          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3465            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3466              if (eptr >= md->end_subject)
3467                {
3468                SCHECK_PARTIAL();
3469                MRRETURN(MATCH_NOMATCH);
3470                }
3471              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3472              }
3473          }          }
3474    
3475        if (min == max) continue;        if (min == max) continue;
# Line 2563  for (;;) Line 3483  for (;;)
3483            register unsigned int d;            register unsigned int d;
3484            for (fi = min;; fi++)            for (fi = min;; fi++)
3485              {              {
3486              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3487              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3488                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3489                if (eptr >= md->end_subject)
3490                  {
3491                  SCHECK_PARTIAL();
3492                  MRRETURN(MATCH_NOMATCH);
3493                  }
3494              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3495              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3496              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3497              }              }
3498            }            }
3499          else          else
# Line 2577  for (;;) Line 3502  for (;;)
3502            {            {
3503            for (fi = min;; fi++)            for (fi = min;; fi++)
3504              {              {
3505              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3506              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3507              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3508                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3509                  {
3510                  SCHECK_PARTIAL();
3511                  MRRETURN(MATCH_NOMATCH);
3512                  }
3513                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3514              }              }
3515            }            }
3516          /* Control never gets here */          /* Control never gets here */
# Line 2600  for (;;) Line 3530  for (;;)
3530            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3531              {              {
3532              int len = 1;              int len = 1;
3533              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3534                  {
3535                  SCHECK_PARTIAL();
3536                  break;
3537                  }
3538              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3539              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3540              if (fc == d) break;              if (fc == d) break;
# Line 2609  for (;;) Line 3543  for (;;)
3543          if (possessive) continue;          if (possessive) continue;
3544          for(;;)          for(;;)
3545              {              {
3546              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3547              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3548              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3549              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2621  for (;;) Line 3555  for (;;)
3555            {            {
3556            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3557              {              {
3558              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3559                  {
3560                  SCHECK_PARTIAL();
3561                  break;
3562                  }
3563                if (fc == md->lcc[*eptr]) break;
3564              eptr++;              eptr++;
3565              }              }
3566            if (possessive) continue;            if (possessive) continue;
3567            while (eptr >= pp)            while (eptr >= pp)
3568              {              {
3569              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3570              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3571              eptr--;              eptr--;
3572              }              }
3573            }            }
3574    
3575          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3576          }          }
3577        /* Control never gets here */        /* Control never gets here */
3578        }        }
# Line 2649  for (;;) Line 3588  for (;;)
3588          register unsigned int d;          register unsigned int d;
3589          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3590            {            {
3591              if (eptr >= md->end_subject)
3592                {
3593                SCHECK_PARTIAL();
3594                MRRETURN(MATCH_NOMATCH);
3595                }
3596            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3597            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3598            }            }
3599          }          }
3600        else        else
# Line 2658  for (;;) Line 3602  for (;;)
3602        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3603          {          {
3604          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3605            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3606              if (eptr >= md->end_subject)
3607                {
3608                SCHECK_PARTIAL();
3609                MRRETURN(MATCH_NOMATCH);
3610                }
3611              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3612              }
3613          }          }
3614    
3615        if (min == max) continue;        if (min == max) continue;
# Line 2672  for (;;) Line 3623  for (;;)
3623            register unsigned int d;            register unsigned int d;
3624            for (fi = min;; fi++)            for (fi = min;; fi++)
3625              {              {
3626              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3627              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3628                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3629                if (eptr >= md->end_subject)
3630                  {
3631                  SCHECK_PARTIAL();
3632                  MRRETURN(MATCH_NOMATCH);
3633                  }
3634              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3635              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3636              }              }
3637            }            }
3638          else          else
# Line 2685  for (;;) Line 3641  for (;;)
3641            {            {
3642            for (fi = min;; fi++)            for (fi = min;; fi++)
3643              {              {
3644              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3645              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3646              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3647                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3648                  {
3649                  SCHECK_PARTIAL();
3650                  MRRETURN(MATCH_NOMATCH);
3651                  }
3652                if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3653              }              }
3654            }            }
3655          /* Control never gets here */          /* Control never gets here */
# Line 2708  for (;;) Line 3669  for (;;)
3669            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3670              {              {
3671              int len = 1;              int len = 1;
3672              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3673                  {
3674                  SCHECK_PARTIAL();
3675                  break;
3676                  }
3677              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3678              if (fc == d) break;              if (fc == d) break;
3679              eptr += len;              eptr += len;
# Line 2716  for (;;) Line 3681  for (;;)
3681            if (possessive) continue;            if (possessive) continue;
3682            for(;;)            for(;;)
3683              {              {
3684              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3685              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3686              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3687              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2728  for (;;) Line 3693  for (;;)
3693            {            {
3694            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3695              {              {
3696              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3697                  {
3698                  SCHECK_PARTIAL();
3699                  break;
3700                  }
3701                if (fc == *eptr) break;
3702              eptr++;              eptr++;
3703              }              }
3704            if (possessive) continue;            if (possessive) continue;
3705            while (eptr >= pp)            while (eptr >= pp)
3706              {              {
3707              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3708              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3709              eptr--;              eptr--;
3710              }              }
3711            }            }
3712    
3713          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3714          }          }
3715        }        }
3716      /* Control never gets here */      /* Control never gets here */
# Line 2822  for (;;) Line 3792  for (;;)
3792    
3793      /* First, ensure the minimum number of matches are present. Use inline      /* First, ensure the minimum number of matches are present. Use inline
3794      code for maximizing the speed, and do the type test once at the start      code for maximizing the speed, and do the type test once at the start
3795      (i.e. keep it out of the loop). Also we can test that there are at least      (i.e. keep it out of the loop). Separate the UTF-8 code completely as that
     the minimum number of bytes before we start. This isn't as effective in  
     UTF-8 mode, but it does no harm. Separate the UTF-8 code completely as that  
3796      is tidier. Also separate the UCP code, which can be the same for both UTF-8      is tidier. Also separate the UCP code, which can be the same for both UTF-8
3797      and single-bytes. */      and single-bytes. */
3798    
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3799      if (min > 0)      if (min > 0)
3800        {        {
3801  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2837  for (;;) Line 3804  for (;;)
3804          switch(prop_type)          switch(prop_type)
3805            {            {
3806            case PT_ANY:            case PT_ANY:
3807            if (prop_fail_result) RRETURN(MATCH_NOMATCH);            if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
3808              for (i = 1; i <= min; i++)
3809                {
3810                if (eptr >= md->end_subject)
3811                  {
3812                  SCHECK_PARTIAL();
3813                  MRRETURN(MATCH_NOMATCH);
3814                  }
3815                GETCHARINCTEST(c, eptr);
3816                }
3817              break;
3818    
3819              case PT_LAMP:
3820              for (i = 1; i <= min; i++)
3821                {
3822                int chartype;
3823                if (eptr >= md->end_subject)
3824                  {
3825                  SCHECK_PARTIAL();
3826                  MRRETURN(MATCH_NOMATCH);
3827                  }
3828                GETCHARINCTEST(c, eptr);
3829                chartype = UCD_CHARTYPE(c);
3830                if ((chartype == ucp_Lu ||
3831                     chartype == ucp_Ll ||
3832                     chartype == ucp_Lt) == prop_fail_result)
3833                  MRRETURN(MATCH_NOMATCH);
3834                }
3835              break;
3836    
3837              case PT_GC:
3838              for (i = 1; i <= min; i++)
3839                {
3840                if (eptr >= md->end_subject)
3841                  {
3842                  SCHECK_PARTIAL();
3843                  MRRETURN(MATCH_NOMATCH);
3844                  }
3845                GETCHARINCTEST(c, eptr);
3846                if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
3847                  MRRETURN(MATCH_NOMATCH);
3848                }
3849              break;
3850    
3851              case PT_PC:
3852              for (i = 1; i <= min; i++)
3853                {
3854                if (eptr >= md->end_subject)
3855                  {
3856                  SCHECK_PARTIAL();
3857                  MRRETURN(MATCH_NOMATCH);
3858                  }
3859                GETCHARINCTEST(c, eptr);
3860                if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
3861                  MRRETURN(MATCH_NOMATCH);
3862                }
3863              break;
3864    
3865              case PT_SC:
3866            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3867              {              {
3868              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3869                  {
3870                  SCHECK_PARTIAL();
3871                  MRRETURN(MATCH_NOMATCH);
3872                  }
3873              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3874                if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
3875                  MRRETURN(MATCH_NOMATCH);
3876              }              }
3877            break;            break;
3878    
3879            case PT_LAMP:            case PT_ALNUM:
3880            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3881              {              {
3882              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              int category;
3883                if (eptr >= md->end_subject)
3884                  {
3885                  SCHECK_PARTIAL();
3886                  MRRETURN(MATCH_NOMATCH);
3887                  }
3888              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3889              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);              category = UCD_CATEGORY(c);
3890              if ((prop_chartype == ucp_Lu ||              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
3891                   prop_chartype == ucp_Ll ||                MRRETURN(MATCH_NOMATCH);
                  prop_chartype == ucp_Lt) == prop_fail_result)  
               RRETURN(MATCH_NOMATCH);  
3892              }