/[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 197 by ph10, Tue Jul 31 10:50:18 2007 UTC revision 642 by ph10, Thu Jul 28 18:59:40 2011 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2007 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 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  pattern matching using an NFA algorithm, trying to mimic Perl as closely as  pattern matching using an NFA algorithm, trying to mimic Perl as closely as
43  possible. There are also some static supporting functions. */  possible. There are also some static supporting functions. */
44    
45    #ifdef HAVE_CONFIG_H
46    #include "config.h"
47    #endif
48    
49  #define NLBLOCK md             /* Block containing newline information */  #define NLBLOCK md             /* Block containing newline information */
50  #define PSSTART start_subject  /* Field containing processed string start */  #define PSSTART start_subject  /* Field containing processed string start */
51  #define PSEND   end_subject    /* Field containing processed string end */  #define PSEND   end_subject    /* Field containing processed string end */
# Line 53  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 64  defined PCRE_ERROR_xxx codes, which are Line 70  defined PCRE_ERROR_xxx codes, which are
70  #define MATCH_MATCH        1  #define MATCH_MATCH        1
71  #define MATCH_NOMATCH      0  #define MATCH_NOMATCH      0
72    
73    /* Special internal returns from the match() function. Make them sufficiently
74    negative to avoid the external error codes. */
75    
76    #define MATCH_ACCEPT       (-999)
77    #define MATCH_COMMIT       (-998)
78    #define MATCH_KETRPOS      (-997)
79    #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,
95  because the offset vector is always a multiple of 3 long. */  because the offset vector is always a multiple of 3 long. */
# Line 77  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 110  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 142  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 199  variable instead of being passed in the Line 268  variable instead of being passed in the
268  ****************************************************************************  ****************************************************************************
269  ***************************************************************************/  ***************************************************************************/
270    
271    /* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN
272  /* Numbers for RMATCH calls */  below must be updated in sync.  */
273    
274  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,
275         RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,         RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,
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, 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 229  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 243  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 265  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 284  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 309  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 357  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 369  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 401  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 408  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 416  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 431  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 443  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 455  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 493  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  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  recursion_info new_recursive;
599  const uschar *charptr;             /* in small blocks of the code. My normal */  
600  #endif                             /* style of coding would have declared    */  BOOL cur_is_word;
 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 537  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 556  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 573  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 598  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      /* Handle a capturing bracket. If there is space in the offset vector, save      case OP_MARK:
704      the current subject position in the working slot at the top of the vector.      markptr = ecode + 2;
705      We mustn't change the current values of the data slot, because they may be      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
706      set from a previous iteration of this group, and be referred to by a        eptrb, RM55);
707      reference inside the group.  
708        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
709      If the bracket fails to match, we need to restore this value and also the      argument, and we must check whether that argument matches this MARK's
710      values of the final offsets, in case they were set by a previous iteration      argument. It is passed back in md->start_match_ptr (an overloading of that
711      of the same bracket.      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:
726        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:
742        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
743          eptrb, RM51);
744        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);
753    
754        /* SKIP overrides PRUNE and THEN */
755    
756        case OP_SKIP:
757        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
758          eptrb, RM53);
759        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
760          RRETURN(rrc);
761        md->start_match_ptr = eptr;   /* Pass back current position */
762        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 bracket or
779        the alt that is at the start of the current branch. This makes it possible
780        to skip back past alternatives that precede the THEN within the current
781        branch. */
782    
783        case OP_THEN:
784        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
785          eptrb, RM54);
786        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
787        md->start_match_ptr = ecode - GET(ecode, 1);
788        MRRETURN(MATCH_THEN);
789    
790        case OP_THEN_ARG:
791        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
792          offset_top, md, eptrb, RM58);
793        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
794        md->start_match_ptr = ecode - GET(ecode, 1);
795        md->mark = ecode + LINK_SIZE + 2;
796        RRETURN(MATCH_THEN);
797    
798        /* Handle a capturing bracket, other than those that are possessive with an
799        unlimited repeat. If there is space in the offset vector, save the current
800        subject position in the working slot at the top of the vector. We mustn't
801        change the current values of the data slot, because they may be set from a
802        previous iteration of this group, and be referred to by a reference inside
803        the group. A failure to match might occur after the group has succeeded,
804        if something later on doesn't match. For this reason, we need to restore
805        the working value and also the values of the final offsets, in case they
806        were set by a previous iteration of the same bracket.
807    
808      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
809      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 627  for (;;) Line 814  for (;;)
814      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
815      offset = number << 1;      offset = number << 1;
816    
817  #ifdef DEBUG  #ifdef PCRE_DEBUG
818      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
819      printf("subject=");      printf("subject=");
820      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 642  for (;;) Line 829  for (;;)
829        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
830    
831        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
832        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
833            (int)(eptr - md->start_subject);
834    
835        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
836          {          {
837            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839            ims, eptrb, flags, RM1);            eptrb, RM1);
840          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841            if (rrc != MATCH_NOMATCH &&
842                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
843              RRETURN(rrc);
844          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
845          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
846            if (*ecode != OP_ALT) break;
847          }          }
       while (*ecode == OP_ALT);  
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855          MATCH_THEN. */
856    
857          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
858          RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
859        }        }
860    
861      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
# Line 675  for (;;) Line 869  for (;;)
869      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870      /* VVVVVVVVVVVVVVVVVVVVVVVVV */      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
871    
872      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* Non-capturing or atomic group, except for possessive with unlimited
873      final alternative within the brackets, we would return the result of a      repeat. Loop for all the alternatives. When we get to the final alternative
874      recursive call to match() whatever happened. We can reduce stack usage by      within the brackets, we used to return the result of a recursive call to
875      turning this into a tail recursion, except in the case when match_cbegroup      match() whatever happened so it was possible to reduce stack usage by
876      is set.*/      turning this into a tail recursion, except in the case of a possibly empty
877        group. However, now that there is the possiblity of (*THEN) occurring in
878        the final alternative, this optimization is no longer possible.
879    
880        MATCH_ONCE is returned when the end of an atomic group is successfully
881        reached, but subsequent matching fails. It passes back up the tree (causing
882        captured values to be reset) until the original atomic group level is
883        reached. This is tested by comparing md->once_target with the start of the
884        group. At this point, the return is converted into MATCH_NOMATCH so that
885        previous backup points can be taken. */
886    
887        case OP_ONCE:
888      case OP_BRA:      case OP_BRA:
889      case OP_SBRA:      case OP_SBRA:
890      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
891      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
892      for (;;)      for (;;)
893        {        {
894        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
895          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
896            RM2);
897          if (rrc != MATCH_NOMATCH &&
898              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
899          {          {
900          if (flags == 0)    /* Not a possibly empty group */          if (rrc == MATCH_ONCE)
901            {            {
902            ecode += _pcre_OP_lengths[*ecode];            const uschar *scode = ecode;
903            DPRINTF(("bracket 0 tail recursion\n"));            if (*scode != OP_ONCE)           /* If not at start, find it */
904            goto TAIL_RECURSE;              {
905                while (*scode == OP_ALT) scode += GET(scode, 1);
906                scode -= GET(scode, 1);
907                }
908              if (md->once_target == scode) rrc = MATCH_NOMATCH;
909            }            }
910            RRETURN(rrc);
911            }
912          ecode += GET(ecode, 1);
913          if (*ecode != OP_ALT) break;
914          }
915        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
916        RRETURN(MATCH_NOMATCH);
917    
918          /* Possibly empty group; can't use tail recursion. */      /* Handle possessive capturing brackets with an unlimited repeat. We come
919        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
920        handled similarly to the normal case above. However, the matching is
921        different. The end of these brackets will always be OP_KETRPOS, which
922        returns MATCH_KETRPOS without going further in the pattern. By this means
923        we can handle the group by iteration rather than recursion, thereby
924        reducing the amount of stack needed. */
925    
926        case OP_CBRAPOS:
927        case OP_SCBRAPOS:
928        allow_zero = FALSE;
929    
930          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      POSSESSIVE_CAPTURE:
931            eptrb, flags, RM48);      number = GET2(ecode, 1+LINK_SIZE);
932          RRETURN(rrc);      offset = number << 1;
933    
934    #ifdef PCRE_DEBUG
935        printf("start possessive bracket %d\n", number);
936        printf("subject=");
937        pchars(eptr, 16, TRUE, md);
938        printf("\n");
939    #endif
940    
941        if (offset < md->offset_max)
942          {
943          matched_once = FALSE;
944          code_offset = ecode - md->start_code;
945    
946          save_offset1 = md->offset_vector[offset];
947          save_offset2 = md->offset_vector[offset+1];
948          save_offset3 = md->offset_vector[md->offset_end - number];
949          save_capture_last = md->capture_last;
950    
951          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
952    
953          /* Each time round the loop, save the current subject position for use
954          when the group matches. For MATCH_MATCH, the group has matched, so we
955          restart it with a new subject starting position, remembering that we had
956          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
957          usual. If we haven't matched any alternatives in any iteration, check to
958          see if a previous iteration matched. If so, the group has matched;
959          continue from afterwards. Otherwise it has failed; restore the previous
960          capture values before returning NOMATCH. */
961    
962          for (;;)
963            {
964            md->offset_vector[md->offset_end - number] =
965              (int)(eptr - md->start_subject);
966            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
967            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
968              eptrb, RM63);
969            if (rrc == MATCH_KETRPOS)
970              {
971              offset_top = md->end_offset_top;
972              eptr = md->end_match_ptr;
973              ecode = md->start_code + code_offset;
974              save_capture_last = md->capture_last;
975              matched_once = TRUE;
976              continue;
977              }
978            if (rrc != MATCH_NOMATCH &&
979                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
980              RRETURN(rrc);
981            md->capture_last = save_capture_last;
982            ecode += GET(ecode, 1);
983            if (*ecode != OP_ALT) break;
984          }          }
985    
986        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
987        otherwise return. */          {
988            md->offset_vector[offset] = save_offset1;
989            md->offset_vector[offset+1] = save_offset2;
990            md->offset_vector[md->offset_end - number] = save_offset3;
991            }
992    
993        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
994          eptrb, flags, RM2);        if (allow_zero || matched_once)
995        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          {
996            ecode += 1 + LINK_SIZE;
997            break;
998            }
999    
1000          RRETURN(MATCH_NOMATCH);
1001          }
1002    
1003        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1004        as a non-capturing bracket. */
1005    
1006        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1007        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008    
1009        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1010    
1011        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1012        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1013    
1014        /* Non-capturing possessive bracket with unlimited repeat. We come here
1015        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1016        without the capturing complication. It is written out separately for speed
1017        and cleanliness. */
1018    
1019        case OP_BRAPOS:
1020        case OP_SBRAPOS:
1021        allow_zero = FALSE;
1022    
1023        POSSESSIVE_NON_CAPTURE:
1024        matched_once = FALSE;
1025        code_offset = ecode - md->start_code;
1026    
1027        for (;;)
1028          {
1029          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1030          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1031            eptrb, RM48);
1032          if (rrc == MATCH_KETRPOS)
1033            {
1034            offset_top = md->end_offset_top;
1035            eptr = md->end_match_ptr;
1036            ecode = md->start_code + code_offset;
1037            matched_once = TRUE;
1038            continue;
1039            }
1040          if (rrc != MATCH_NOMATCH &&
1041              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1042            RRETURN(rrc);
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045          }
1046    
1047        if (matched_once || allow_zero)
1048          {
1049          ecode += 1 + LINK_SIZE;
1050          break;
1051        }        }
1052        RRETURN(MATCH_NOMATCH);
1053    
1054      /* Control never reaches here. */      /* Control never reaches here. */
1055    
1056      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1057      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1058      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1059      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1060    
1061      case OP_COND:      case OP_COND:
1062      case OP_SCOND:      case OP_SCOND:
1063      if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */      codelink = GET(ecode, 1);
1064    
1065        /* Because of the way auto-callout works during compile, a callout item is
1066        inserted between OP_COND and an assertion condition. */
1067    
1068        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1069        {        {
1070        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (pcre_callout != NULL)
1071        condition = md->recursive != NULL &&          {
1072          (offset == RREF_ANY || offset == md->recursive->group_num);          pcre_callout_block cb;
1073        ecode += condition? 3 : GET(ecode, 1);          cb.version          = 1;   /* Version 1 of the callout block */
1074            cb.callout_number   = ecode[LINK_SIZE+2];
1075            cb.offset_vector    = md->offset_vector;
1076            cb.subject          = (PCRE_SPTR)md->start_subject;
1077            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1078            cb.start_match      = (int)(mstart - md->start_subject);
1079            cb.current_position = (int)(eptr - md->start_subject);
1080            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1081            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1082            cb.capture_top      = offset_top/2;
1083            cb.capture_last     = md->capture_last;
1084            cb.callout_data     = md->callout_data;
1085            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1086            if (rrc < 0) RRETURN(rrc);
1087            }
1088          ecode += _pcre_OP_lengths[OP_CALLOUT];
1089          }
1090    
1091        condcode = ecode[LINK_SIZE+1];
1092    
1093        /* Now see what the actual condition is */
1094    
1095        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1096          {
1097          if (md->recursive == NULL)                /* Not recursing => FALSE */
1098            {
1099            condition = FALSE;
1100            ecode += GET(ecode, 1);
1101            }
1102          else
1103            {
1104            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1105            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1106    
1107            /* If the test is for recursion into a specific subpattern, and it is
1108            false, but the test was set up by name, scan the table to see if the
1109            name refers to any other numbers, and test them. The condition is true
1110            if any one is set. */
1111    
1112            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1113              {
1114              uschar *slotA = md->name_table;
1115              for (i = 0; i < md->name_count; i++)
1116                {
1117                if (GET2(slotA, 0) == recno) break;
1118                slotA += md->name_entry_size;
1119                }
1120    
1121              /* Found a name for the number - there can be only one; duplicate
1122              names for different numbers are allowed, but not vice versa. First
1123              scan down for duplicates. */
1124    
1125              if (i < md->name_count)
1126                {
1127                uschar *slotB = slotA;
1128                while (slotB > md->name_table)
1129                  {
1130                  slotB -= md->name_entry_size;
1131                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1132                    {
1133                    condition = GET2(slotB, 0) == md->recursive->group_num;
1134                    if (condition) break;
1135                    }
1136                  else break;
1137                  }
1138    
1139                /* Scan up for duplicates */
1140    
1141                if (!condition)
1142                  {
1143                  slotB = slotA;
1144                  for (i++; i < md->name_count; i++)
1145                    {
1146                    slotB += md->name_entry_size;
1147                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1148                      {
1149                      condition = GET2(slotB, 0) == md->recursive->group_num;
1150                      if (condition) break;
1151                      }
1152                    else break;
1153                    }
1154                  }
1155                }
1156              }
1157    
1158            /* Chose branch according to the condition */
1159    
1160            ecode += condition? 3 : GET(ecode, 1);
1161            }
1162        }        }
1163    
1164      else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1165        {        {
1166        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1167        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1168    
1169          /* If the numbered capture is unset, but the reference was by name,
1170          scan the table to see if the name refers to any other numbers, and test
1171          them. The condition is true if any one is set. This is tediously similar
1172          to the code above, but not close enough to try to amalgamate. */
1173    
1174          if (!condition && condcode == OP_NCREF)
1175            {
1176            int refno = offset >> 1;
1177            uschar *slotA = md->name_table;
1178    
1179            for (i = 0; i < md->name_count; i++)
1180              {
1181              if (GET2(slotA, 0) == refno) break;
1182              slotA += md->name_entry_size;
1183              }
1184    
1185            /* Found a name for the number - there can be only one; duplicate names
1186            for different numbers are allowed, but not vice versa. First scan down
1187            for duplicates. */
1188    
1189            if (i < md->name_count)
1190              {
1191              uschar *slotB = slotA;
1192              while (slotB > md->name_table)
1193                {
1194                slotB -= md->name_entry_size;
1195                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1196                  {
1197                  offset = GET2(slotB, 0) << 1;
1198                  condition = offset < offset_top &&
1199                    md->offset_vector[offset] >= 0;
1200                  if (condition) break;
1201                  }
1202                else break;
1203                }
1204    
1205              /* Scan up for duplicates */
1206    
1207              if (!condition)
1208                {
1209                slotB = slotA;
1210                for (i++; i < md->name_count; i++)
1211                  {
1212                  slotB += md->name_entry_size;
1213                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1214                    {
1215                    offset = GET2(slotB, 0) << 1;
1216                    condition = offset < offset_top &&
1217                      md->offset_vector[offset] >= 0;
1218                    if (condition) break;
1219                    }
1220                  else break;
1221                  }
1222                }
1223              }
1224            }
1225    
1226          /* Chose branch according to the condition */
1227    
1228        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 3 : GET(ecode, 1);
1229        }        }
1230    
1231      else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
1232        {        {
1233        condition = FALSE;        condition = FALSE;
1234        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1235        }        }
1236    
1237      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1238      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1239      assertion. */      an assertion. */
1240    
1241      else      else
1242        {        {
1243        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1244            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1245        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1246          {          {
1247            if (md->end_offset_top > offset_top)
1248              offset_top = md->end_offset_top;  /* Captures may have happened */
1249          condition = TRUE;          condition = TRUE;
1250          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1251          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1252          }          }
1253        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH &&
1254                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1255          {          {
1256          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1257          }          }
1258        else        else
1259          {          {
1260          condition = FALSE;          condition = FALSE;
1261          ecode += GET(ecode, 1);          ecode += codelink;
1262          }          }
1263        }        }
1264    
1265      /* We are now at the branch that is to be obeyed. As there is only one,      /* We are now at the branch that is to be obeyed. As there is only one,
1266      we can use tail recursion to avoid using another stack frame, except when      we used to use tail recursion to avoid using another stack frame, except
1267      match_cbegroup is required for an unlimited repeat of a possibly empty      when there was unlimited repeat of a possibly empty group. However, that
1268      group. If the second alternative doesn't exist, we can just plough on. */      strategy no longer works because of the possibilty of (*THEN) being
1269        encountered in the branch. A recursive call to match() is always required,
1270        unless the second alternative doesn't exist, in which case we can just
1271        plough on. */
1272    
1273      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1274        {        {
1275        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1276        if (op == OP_SCOND)        /* Possibly empty group */        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1277          {        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1278          RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);          rrc = MATCH_NOMATCH;
1279          RRETURN(rrc);        RRETURN(rrc);
         }  
       else                       /* Group must match something */  
         {  
         flags = 0;  
         goto TAIL_RECURSE;  
         }  
1280        }        }
1281      else                         /* Condition false & no 2nd alternative */      else                         /* Condition false & no alternative */
1282        {        {
1283        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1284        }        }
1285      break;      break;
1286    
1287    
1288      /* End of the pattern. If we are in a top-level recursion, we should      /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
1289      restore the offsets appropriately and continue from after the call. */      to close any currently open capturing brackets. */
1290    
1291      case OP_END:      case OP_CLOSE:
1292      if (md->recursive != NULL && md->recursive->group_num == 0)      number = GET2(ecode, 1);
1293        offset = number << 1;
1294    
1295    #ifdef PCRE_DEBUG
1296          printf("end bracket %d at *ACCEPT", number);
1297          printf("\n");
1298    #endif
1299    
1300        md->capture_last = number;
1301        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1302        {        {
1303        recursion_info *rec = md->recursive;        md->offset_vector[offset] =
1304        DPRINTF(("End of pattern in a (?0) recursion\n"));          md->offset_vector[md->offset_end - number];
1305        md->recursive = rec->prevrec;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1306        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;  
1307        }        }
1308        ecode += 3;
1309        break;
1310    
1311    
1312        /* End of the pattern, either real or forced. */
1313    
1314        case OP_END:
1315        case OP_ACCEPT:
1316        case OP_ASSERT_ACCEPT:
1317    
1318      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty      /* If we have matched an empty string, fail if not in an assertion and not
1319      string - backtracking will then try other alternatives, if any. */      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1320        is set and we have matched at the start of the subject. In both cases,
1321        backtracking will then try other alternatives, if any. */
1322    
1323        if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1324             md->recursive == NULL &&
1325             (md->notempty ||
1326               (md->notempty_atstart &&
1327                 mstart == md->start_subject + md->start_offset)))
1328          MRRETURN(MATCH_NOMATCH);
1329    
1330        /* Otherwise, we have a match. */
1331    
     if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH);  
1332      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1333      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1334      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);  
1335    
1336      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1337        given as the argument to MRRETURN when the heap is in use. */
1338    
1339      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1340      ims = ecode[1];      MRRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1341    
1342      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1343      matching won't pass the KET for an assertion. If any one branch matches,      matching won't pass the KET for an assertion. If any one branch matches,
1344      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the      the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
1345      start of each branch to move the current point backwards, so the code at      start of each branch to move the current point backwards, so the code at
1346      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1347        of a condition, we want to return immediately afterwards. The caller of
1348        this incarnation of the match() function will have set MATCH_CONDASSERT in
1349        md->match_function type, and one of these opcodes will be the first opcode
1350        that is processed. We use a local variable that is preserved over calls to
1351        match() to remember this case. */
1352    
1353      case OP_ASSERT:      case OP_ASSERT:
1354      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1355        if (md->match_function_type == MATCH_CONDASSERT)
1356          {
1357          condassert = TRUE;
1358          md->match_function_type = 0;
1359          }
1360        else condassert = FALSE;
1361    
1362      do      do
1363        {        {
1364        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1365          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1366        if (rrc == MATCH_MATCH) break;          {
1367        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          mstart = md->start_match_ptr;   /* In case \K reset it */
1368            markptr = md->mark;
1369            break;
1370            }
1371          if (rrc != MATCH_NOMATCH &&
1372              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1373            RRETURN(rrc);
1374        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1375        }        }
1376      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1377      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1378        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1379    
1380      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1381    
1382      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1383    
1384      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1385      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 858  for (;;) Line 1389  for (;;)
1389      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1390      continue;      continue;
1391    
1392      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1393        PRUNE, or COMMIT means we must assume failure without checking subsequent
1394        branches. */
1395    
1396      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1397      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1398        if (md->match_function_type == MATCH_CONDASSERT)
1399          {
1400          condassert = TRUE;
1401          md->match_function_type = 0;
1402          }
1403        else condassert = FALSE;
1404    
1405      do      do
1406        {        {
1407        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1408          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1409        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1410        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          {
1411            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1412            break;
1413            }
1414          if (rrc != MATCH_NOMATCH &&
1415              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1416            RRETURN(rrc);
1417        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1418        }        }
1419      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1420    
1421      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1422    
1423      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1424      continue;      continue;
# Line 890  for (;;) Line 1436  for (;;)
1436        while (i-- > 0)        while (i-- > 0)
1437          {          {
1438          eptr--;          eptr--;
1439          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1440          BACKCHAR(eptr)          BACKCHAR(eptr);
1441          }          }
1442        }        }
1443      else      else
# Line 901  for (;;) Line 1447  for (;;)
1447    
1448        {        {
1449        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1450        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1451        }        }
1452    
1453      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1454    
1455        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1456      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1457      break;      break;
1458    
# Line 921  for (;;) Line 1468  for (;;)
1468        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1469        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1470        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1471        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1472        cb.start_match      = mstart - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1473        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1474        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1475        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1476        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1477        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1478        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1479        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1480        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1481        }        }
1482      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 939  for (;;) Line 1486  for (;;)
1486      offset data is the offset to the starting bracket from the start of the      offset data is the offset to the starting bracket from the start of the
1487      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1488    
1489      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1490      save their starting points and reinstate them after the recursion. However,      re-instated afterwards. We don't know how many are started and not yet
1491      we don't know how many such there are (offset_top records the completed      finished (offset_top records the completed total) so we just have to save
1492      total) so we just have to save all the potential data. There may be up to      all the potential data. There may be up to 65535 such values, which is too
1493      65535 such values, which is too large to put on the stack, but using malloc      large to put on the stack, but using malloc for small numbers seems
1494      for small numbers seems expensive. As a compromise, the stack is used when      expensive. As a compromise, the stack is used when there are no more than
1495      there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc      REC_STACK_SAVE_MAX values to store; otherwise malloc is used.
     is used. A problem is what to do if the malloc fails ... there is no way of  
     returning to the top level with an error. Save the top REC_STACK_SAVE_MAX  
     values on the stack, and accept that the rest may be wrong.  
1496    
1497      There are also other values that have to be saved. We use a chained      There are also other values that have to be saved. We use a chained
1498      sequence of blocks that actually live on the stack. Thanks to Robin Houston      sequence of blocks that actually live on the stack. Thanks to Robin Houston
1499      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1500        a lot, so he is not to blame for the current way it works. */
1501    
1502      case OP_RECURSE:      case OP_RECURSE:
1503        {        {
1504          recursion_info *ri;
1505          int recno;
1506    
1507        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1508        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1509          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1510    
1511          /* Check for repeating a recursion without advancing the subject pointer.
1512          This should catch convoluted mutual recursions. (Some simple cases are
1513          caught at compile time.) */
1514    
1515          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1516            if (recno == ri->group_num && eptr == ri->subject_position)
1517              RRETURN(PCRE_ERROR_RECURSELOOP);
1518    
1519        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1520    
1521          new_recursive.group_num = recno;
1522          new_recursive.subject_position = eptr;
1523        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1524        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1525    
1526        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1527    
1528        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1529    
1530        /* Now save the offset data. */        /* Now save the offset data */
1531    
1532        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1533        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 981  for (;;) Line 1538  for (;;)
1538            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1539          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1540          }          }
   
1541        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1542              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = mstart;  
       mstart = eptr;  
1543    
1544        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1545        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1546          might be changed, so reset it before looping. */
1547    
1548        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1549        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1550        do        do
1551          {          {
1552            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1553          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1554            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1555          if (rrc == MATCH_MATCH)          memcpy(md->offset_vector, new_recursive.offset_save,
1556                new_recursive.saved_max * sizeof(int));
1557            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1558            {            {
1559            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1560            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1561            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1562              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1563            RRETURN(MATCH_MATCH);  
1564              /* Set where we got to in the subject, and reset the start in case
1565              it was changed by \K. This *is* propagated back out of a recursion,
1566              for Perl compatibility. */
1567    
1568              eptr = md->end_match_ptr;
1569              mstart = md->start_match_ptr;
1570              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1571            }            }
1572          else if (rrc != MATCH_NOMATCH)          else if (rrc != MATCH_NOMATCH &&
1573                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1574            {            {
1575            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1576              if (new_recursive.offset_save != stacksave)
1577                (pcre_free)(new_recursive.offset_save);
1578            RRETURN(rrc);            RRETURN(rrc);
1579            }            }
1580    
1581          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1582          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1583          }          }
1584        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1021  for (;;) Line 1587  for (;;)
1587        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1588        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1589          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1590        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1591        }        }
     /* Control never reaches here */  
1592    
1593      /* "Once" brackets are like assertion brackets except that after a match,      RECURSION_MATCHED:
1594      the point in the subject string is not moved back. Thus there can never be      break;
     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;  
1595    
1596      do      /* An alternation is the end of a branch; scan along to find the end of the
1597        {      bracketed group and go to there. */
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH) break;  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
1598    
1599      /* If hit the end of the group (which could be repeated), fail */      case OP_ALT:
1600        do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1601        break;
1602    
1603      if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1604        indicating that it may occur zero times. It may repeat infinitely, or not
1605        at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1606        with fixed upper repeat limits are compiled as a number of copies, with the
1607        optional ones preceded by BRAZERO or BRAMINZERO. */
1608    
1609      /* Continue as from after the assertion, updating the offsets high water      case OP_BRAZERO:
1610      mark, since extracts may have been taken. */      next = ecode + 1;
1611        RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1612        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1613        do next += GET(next, 1); while (*next == OP_ALT);
1614        ecode = next + 1 + LINK_SIZE;
1615        break;
1616    
1617      do ecode += GET(ecode, 1); while (*ecode == OP_ALT);      case OP_BRAMINZERO:
1618        next = ecode + 1;
1619        do next += GET(next, 1); while (*next == OP_ALT);
1620        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1621        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1622        ecode++;
1623        break;
1624    
1625      offset_top = md->end_offset_top;      case OP_SKIPZERO:
1626      eptr = md->end_match_ptr;      next = ecode+1;
1627        do next += GET(next,1); while (*next == OP_ALT);
1628        ecode = next + 1 + LINK_SIZE;
1629        break;
1630    
1631      /* For a non-repeating ket, just continue at this level. This also      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1632      happens for a repeating ket if no characters were matched in the group.      here; just jump to the group, with allow_zero set TRUE. */
     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. */  
1633    
1634      if (*ecode == OP_KET || eptr == saved_eptr)      case OP_BRAPOSZERO:
1635        {      op = *(++ecode);
1636        ecode += 1+LINK_SIZE;      allow_zero = TRUE;
1637        break;      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1638        }        goto POSSESSIVE_NON_CAPTURE;
1639    
1640      /* The repeating kets try the rest of the pattern or restart from the      /* End of a group, repeated or non-repeating. */
     preceding bracket, in the appropriate order. The second "call" of match()  
     uses tail recursion, to avoid using another stack frame. We need to reset  
     any options that changed within the bracket before re-running it, so  
     check the next opcode. */  
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = 0;  
       goto TAIL_RECURSE;  
       }  
     /* Control never gets here */  
   
     /* An alternation is the end of a branch; scan along to find the end of the  
     bracketed group and go to there. */  
   
     case OP_ALT:  
     do ecode += GET(ecode,1); while (*ecode == OP_ALT);  
     break;  
   
     /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating  
     that it may occur zero times. It may repeat infinitely, or not at all -  
     i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper  
     repeat limits are compiled as a number of copies, with the optional ones  
     preceded by BRAZERO or BRAMINZERO. */  
   
     case OP_BRAZERO:  
       {  
       next = ecode+1;  
       RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       do next += GET(next,1); while (*next == OP_ALT);  
       ecode = next + 1 + LINK_SIZE;  
       }  
     break;  
   
     case OP_BRAMINZERO:  
       {  
       next = ecode+1;  
       do next += GET(next, 1); while (*next == OP_ALT);  
       RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode++;  
       }  
     break;  
   
     /* End of a group, repeated or non-repeating. */  
1641    
1642      case OP_KET:      case OP_KET:
1643      case OP_KETRMIN:      case OP_KETRMIN:
1644      case OP_KETRMAX:      case OP_KETRMAX:
1645        case OP_KETRPOS:
1646      prev = ecode - GET(ecode, 1);      prev = ecode - GET(ecode, 1);
1647    
1648      /* 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
1649      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1650      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1651    
1652      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1653        {        {
1654        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1655        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
# Line 1152  for (;;) Line 1658  for (;;)
1658    
1659      /* 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
1660      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
1661      assertions. Do this also for the "once" (atomic) groups. */      assertions. We also need to record the match start in case it was changed
1662        by \K. */
1663    
1664      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1665          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1666        {        {
1667        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1668        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1669        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1670          MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1671        }        }
1672    
1673      /* 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
1674      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1675      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
1676      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
1677      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
1678        the current subject position and start match pointer and give a MATCH
1679        return. */
1680    
1681      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1682            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1683        {        {
1684        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1685        offset = number << 1;        offset = number << 1;
1686    
1687  #ifdef DEBUG  #ifdef PCRE_DEBUG
1688        printf("end bracket %d", number);        printf("end bracket %d", number);
1689        printf("\n");        printf("\n");
1690  #endif  #endif
1691    
1692          /* Handle a recursively called group. */
1693    
1694          if (md->recursive != NULL && md->recursive->group_num == number)
1695            {
1696            md->end_match_ptr = eptr;
1697            md->start_match_ptr = mstart;
1698            RRETURN(MATCH_MATCH);
1699            }
1700    
1701          /* Deal with capturing */
1702    
1703        md->capture_last = number;        md->capture_last = number;
1704        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1705          {          {
1706            /* If offset is greater than offset_top, it means that we are
1707            "skipping" a capturing group, and that group's offsets must be marked
1708            unset. In earlier versions of PCRE, all the offsets were unset at the
1709            start of matching, but this doesn't work because atomic groups and
1710            assertions can cause a value to be set that should later be unset.
1711            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1712            part of the atomic group, but this is not on the final matching path,
1713            so must be unset when 2 is set. (If there is no group 2, there is no
1714            problem, because offset_top will then be 2, indicating no capture.) */
1715    
1716            if (offset > offset_top)
1717              {
1718              register int *iptr = md->offset_vector + offset_top;
1719              register int *iend = md->offset_vector + offset;
1720              while (iptr < iend) *iptr++ = -1;
1721              }
1722    
1723            /* Now make the extraction */
1724    
1725          md->offset_vector[offset] =          md->offset_vector[offset] =
1726            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1727          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1728          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1729          }          }
1730          }
1731    
1732        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1733        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1734        group. This is the forcible breaking of infinite loops as implemented in
1735        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1736        processing the rest of the pattern at a lower level. If this results in a
1737        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1738        bypassing intermediate backup points, but resetting any captures that
1739        happened along the way. */
1740    
1741        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1742          {
1743          if (*prev == OP_ONCE)
1744          {          {
1745          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1746          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1747          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1748          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;  
1749          }          }
1750          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1751          break;
1752        }        }
1753    
1754      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1755      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1756        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
1757    
1758      /* 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)  
1759        {        {
1760        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1761        break;        md->end_offset_top = offset_top;
1762          RRETURN(MATCH_KETRPOS);
1763        }        }
1764    
1765      /* 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
1766      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
1767      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
1768      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
1769        string. */
     flags = (*prev >= OP_SBRA)? match_cbegroup : 0;  
1770    
1771      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1772        {        {
1773        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
1774        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1775        if (flags != 0)    /* Could match an empty string */        if (*prev == OP_ONCE)
1776            {
1777            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1778            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1779            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1780            RRETURN(MATCH_ONCE);
1781            }
1782          if (*prev >= OP_SBRA)    /* Could match an empty string */
1783          {          {
1784          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);          md->match_function_type = MATCH_CBEGROUP;
1785            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1786          RRETURN(rrc);          RRETURN(rrc);
1787          }          }
1788        ecode = prev;        ecode = prev;
# Line 1244  for (;;) Line 1790  for (;;)
1790        }        }
1791      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1792        {        {
1793        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1794          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1795          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1796        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1797          if (*prev == OP_ONCE)
1798            {
1799            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1800            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1801            md->once_target = prev;
1802            RRETURN(MATCH_ONCE);
1803            }
1804        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = 0;  
1805        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1806        }        }
1807      /* Control never gets here */      /* Control never gets here */
1808    
1809      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1810    
1811      case OP_CIRC:      case OP_CIRC:
1812      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 */  
1813    
1814      /* Start of subject assertion */      /* Start of subject assertion */
1815    
1816      case OP_SOD:      case OP_SOD:
1817      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1818        ecode++;
1819        break;
1820    
1821        /* Multiline mode: start of subject unless notbol, or after any newline. */
1822    
1823        case OP_CIRCM:
1824        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1825        if (eptr != md->start_subject &&
1826            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1827          MRRETURN(MATCH_NOMATCH);
1828      ecode++;      ecode++;
1829      break;      break;
1830    
1831      /* Start of match assertion */      /* Start of match assertion */
1832    
1833      case OP_SOM:      case OP_SOM:
1834      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1835      ecode++;      ecode++;
1836      break;      break;
1837    
# Line 1287  for (;;) Line 1842  for (;;)
1842      ecode++;      ecode++;
1843      break;      break;
1844    
1845      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1846      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1847    
1848      case OP_DOLL:      case OP_DOLLM:
1849      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1850        {        { 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;  
       }  
1851      else      else
1852        {        {
1853        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1854        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1855        }        }
1856        ecode++;
1857        break;
1858    
1859        /* Not multiline mode: assert before a terminating newline or before end of
1860        subject unless noteol is set. */
1861    
1862        case OP_DOLL:
1863        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1864        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1865    
1866      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1867    
1868      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1869    
1870      case OP_EOD:      case OP_EOD:
1871      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1872        SCHECK_PARTIAL();
1873      ecode++;      ecode++;
1874      break;      break;
1875    
1876      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1877    
1878      case OP_EODN:      case OP_EODN:
1879      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1880        if (eptr < md->end_subject &&
1881          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1882        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1883    
1884        /* Either at end of string or \n before end. */
1885    
1886        SCHECK_PARTIAL();
1887      ecode++;      ecode++;
1888      break;      break;
1889    
# Line 1338  for (;;) Line 1895  for (;;)
1895    
1896        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1897        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
1898        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1899          partial matching. */
1900    
1901  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1902        if (utf8)        if (utf8)
1903          {          {
1904            /* Get status of previous character */
1905    
1906          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1907            {            {
1908            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1909            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1910              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1911            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1912    #ifdef SUPPORT_UCP
1913              if (md->use_ucp)
1914                {
1915                if (c == '_') prev_is_word = TRUE; else
1916                  {
1917                  int cat = UCD_CATEGORY(c);
1918                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1919                  }
1920                }
1921              else
1922    #endif
1923            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1924            }            }
1925          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
1926            /* Get status of next character */
1927    
1928            if (eptr >= md->end_subject)
1929              {
1930              SCHECK_PARTIAL();
1931              cur_is_word = FALSE;
1932              }
1933            else
1934            {            {
1935            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1936    #ifdef SUPPORT_UCP
1937              if (md->use_ucp)
1938                {
1939                if (c == '_') cur_is_word = TRUE; else
1940                  {
1941                  int cat = UCD_CATEGORY(c);
1942                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1943                  }
1944                }
1945              else
1946    #endif
1947            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1948            }            }
1949          }          }
1950        else        else
1951  #endif  #endif
1952    
1953        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1954          consistency with the behaviour of \w we do use it in this case. */
1955    
1956          {          {
1957          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
1958            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
1959          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
1960            ((md->ctypes[*eptr] & ctype_word) != 0);            {
1961              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1962    #ifdef SUPPORT_UCP
1963              if (md->use_ucp)
1964                {
1965                c = eptr[-1];
1966                if (c == '_') prev_is_word = TRUE; else
1967                  {
1968                  int cat = UCD_CATEGORY(c);
1969                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1970                  }
1971                }
1972              else
1973    #endif
1974              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1975              }
1976    
1977            /* Get status of next character */
1978    
1979            if (eptr >= md->end_subject)
1980              {
1981              SCHECK_PARTIAL();
1982              cur_is_word = FALSE;
1983              }
1984            else
1985    #ifdef SUPPORT_UCP
1986            if (md->use_ucp)
1987              {
1988              c = *eptr;
1989              if (c == '_') cur_is_word = TRUE; else
1990                {
1991                int cat = UCD_CATEGORY(c);
1992                cur_is_word = (cat == ucp_L || cat == ucp_N);
1993                }
1994              }
1995            else
1996    #endif
1997            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
1998          }          }
1999    
2000        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2001    
2002        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2003             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2004          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2005        }        }
2006      break;      break;
2007    
2008      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2009    
2010      case OP_ANY:      case OP_ANY:
2011      if ((ims & PCRE_DOTALL) == 0)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2012        /* Fall through */
2013    
2014        case OP_ALLANY:
2015        if (eptr++ >= md->end_subject)
2016        {        {
2017        if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);        SCHECK_PARTIAL();
2018          MRRETURN(MATCH_NOMATCH);
2019        }        }
2020      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
     if (utf8)  
       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
2021      ecode++;      ecode++;
2022      break;      break;
2023    
# Line 1393  for (;;) Line 2025  for (;;)
2025      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2026    
2027      case OP_ANYBYTE:      case OP_ANYBYTE:
2028      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr++ >= md->end_subject)
2029          {
2030          SCHECK_PARTIAL();
2031          MRRETURN(MATCH_NOMATCH);
2032          }
2033      ecode++;      ecode++;
2034      break;      break;
2035    
2036      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2037      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2038          {
2039          SCHECK_PARTIAL();
2040          MRRETURN(MATCH_NOMATCH);
2041          }
2042      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2043      if (      if (
2044  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1406  for (;;) Line 2046  for (;;)
2046  #endif  #endif
2047         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2048         )         )
2049        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2050      ecode++;      ecode++;
2051      break;      break;
2052    
2053      case OP_DIGIT:      case OP_DIGIT:
2054      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2055          {
2056          SCHECK_PARTIAL();
2057          MRRETURN(MATCH_NOMATCH);
2058          }
2059      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2060      if (      if (
2061  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1419  for (;;) Line 2063  for (;;)
2063  #endif  #endif
2064         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2065         )         )
2066        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2067      ecode++;      ecode++;
2068      break;      break;
2069    
2070      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2071      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2072          {
2073          SCHECK_PARTIAL();
2074          MRRETURN(MATCH_NOMATCH);
2075          }
2076      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2077      if (      if (
2078  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1432  for (;;) Line 2080  for (;;)
2080  #endif  #endif
2081         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2082         )         )
2083        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2084      ecode++;      ecode++;
2085      break;      break;
2086    
2087      case OP_WHITESPACE:      case OP_WHITESPACE:
2088      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2089          {
2090          SCHECK_PARTIAL();
2091          MRRETURN(MATCH_NOMATCH);
2092          }
2093      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2094      if (      if (
2095  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1445  for (;;) Line 2097  for (;;)
2097  #endif  #endif
2098         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2099         )         )
2100        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2101      ecode++;      ecode++;
2102      break;      break;
2103    
2104      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2105      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2106          {
2107          SCHECK_PARTIAL();
2108          MRRETURN(MATCH_NOMATCH);
2109          }
2110      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2111      if (      if (
2112  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1458  for (;;) Line 2114  for (;;)
2114  #endif  #endif
2115         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2116         )         )
2117        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2118      ecode++;      ecode++;
2119      break;      break;
2120    
2121      case OP_WORDCHAR:      case OP_WORDCHAR:
2122      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2123          {
2124          SCHECK_PARTIAL();
2125          MRRETURN(MATCH_NOMATCH);
2126          }
2127      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2128      if (      if (
2129  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1471  for (;;) Line 2131  for (;;)
2131  #endif  #endif
2132         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2133         )         )
2134        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2135      ecode++;      ecode++;
2136      break;      break;
2137    
2138      case OP_ANYNL:      case OP_ANYNL:
2139      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2140          {
2141          SCHECK_PARTIAL();
2142          MRRETURN(MATCH_NOMATCH);
2143          }
2144      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2145      switch(c)      switch(c)
2146        {        {
2147        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2148    
2149        case 0x000d:        case 0x000d:
2150        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2151        break;        break;
2152    
2153        case 0x000a:        case 0x000a:
2154          break;
2155    
2156        case 0x000b:        case 0x000b:
2157        case 0x000c:        case 0x000c:
2158        case 0x0085:        case 0x0085:
2159        case 0x2028:        case 0x2028:
2160        case 0x2029:        case 0x2029:
2161          if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2162        break;        break;
2163        }        }
2164      ecode++;      ecode++;
2165      break;      break;
2166    
2167      case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
2168      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2169          {
2170          SCHECK_PARTIAL();
2171          MRRETURN(MATCH_NOMATCH);
2172          }
2173      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2174      switch(c)      switch(c)
2175        {        {
# Line 1520  for (;;) Line 2193  for (;;)
2193        case 0x202f:    /* NARROW NO-BREAK SPACE */        case 0x202f:    /* NARROW NO-BREAK SPACE */
2194        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */        case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2195        case 0x3000:    /* IDEOGRAPHIC SPACE */        case 0x3000:    /* IDEOGRAPHIC SPACE */
2196        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2197        }        }
2198      ecode++;      ecode++;
2199      break;      break;
2200    
2201      case OP_HSPACE:      case OP_HSPACE:
2202      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2203          {
2204          SCHECK_PARTIAL();
2205          MRRETURN(MATCH_NOMATCH);
2206          }
2207      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2208      switch(c)      switch(c)
2209        {        {
2210        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2211        case 0x09:      /* HT */        case 0x09:      /* HT */
2212        case 0x20:      /* SPACE */        case 0x20:      /* SPACE */
2213        case 0xa0:      /* NBSP */        case 0xa0:      /* NBSP */
# Line 1556  for (;;) Line 2233  for (;;)
2233      break;      break;
2234    
2235      case OP_NOT_VSPACE:      case OP_NOT_VSPACE:
2236      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2237          {
2238          SCHECK_PARTIAL();
2239          MRRETURN(MATCH_NOMATCH);
2240          }
2241      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2242      switch(c)      switch(c)
2243        {        {
# Line 1568  for (;;) Line 2249  for (;;)
2249        case 0x85:      /* NEL */        case 0x85:      /* NEL */
2250        case 0x2028:    /* LINE SEPARATOR */        case 0x2028:    /* LINE SEPARATOR */
2251        case 0x2029:    /* PARAGRAPH SEPARATOR */        case 0x2029:    /* PARAGRAPH SEPARATOR */
2252        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2253        }        }
2254      ecode++;      ecode++;
2255      break;      break;
2256    
2257      case OP_VSPACE:      case OP_VSPACE:
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        {        {
2266        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2267        case 0x0a:      /* LF */        case 0x0a:      /* LF */
2268        case 0x0b:      /* VT */        case 0x0b:      /* VT */
2269        case 0x0c:      /* FF */        case 0x0c:      /* FF */
# Line 1597  for (;;) Line 2282  for (;;)
2282    
2283      case OP_PROP:      case OP_PROP:
2284      case OP_NOTPROP:      case OP_NOTPROP:
2285      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2286          {
2287          SCHECK_PARTIAL();
2288          MRRETURN(MATCH_NOMATCH);
2289          }
2290      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2291        {        {
2292        int chartype, script;        const ucd_record *prop = GET_UCD(c);
       int category = _pcre_ucp_findprop(c, &chartype, &script);  
2293    
2294        switch(ecode[1])        switch(ecode[1])
2295          {          {
2296          case PT_ANY:          case PT_ANY:
2297          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2298          break;          break;
2299    
2300          case PT_LAMP:          case PT_LAMP:
2301          if ((chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2302               chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2303               chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2304            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2305           break;          break;
2306    
2307          case PT_GC:          case PT_GC:
2308          if ((ecode[2] != category) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2309            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2310          break;          break;
2311    
2312          case PT_PC:          case PT_PC:
2313          if ((ecode[2] != chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2314            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2315          break;          break;
2316    
2317          case PT_SC:          case PT_SC:
2318          if ((ecode[2] != script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2319            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2320          break;          break;
2321    
2322            /* These are specials */
2323    
2324            case PT_ALNUM:
2325            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2326                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2327              MRRETURN(MATCH_NOMATCH);
2328            break;
2329    
2330            case PT_SPACE:    /* Perl space */
2331            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2332                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2333                   == (op == OP_NOTPROP))
2334              MRRETURN(MATCH_NOMATCH);
2335            break;
2336    
2337            case PT_PXSPACE:  /* POSIX space */
2338            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2339                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2340                 c == CHAR_FF || c == CHAR_CR)
2341                   == (op == OP_NOTPROP))
2342              MRRETURN(MATCH_NOMATCH);
2343            break;
2344    
2345            case PT_WORD:
2346            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2347                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2348                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2349              MRRETURN(MATCH_NOMATCH);
2350            break;
2351    
2352            /* This should never occur */
2353    
2354          default:          default:
2355          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2356          }          }
# Line 1643  for (;;) Line 2363  for (;;)
2363      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2364    
2365      case OP_EXTUNI:      case OP_EXTUNI:
2366      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2367          {
2368          SCHECK_PARTIAL();
2369          MRRETURN(MATCH_NOMATCH);
2370          }
2371      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2372        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2373        while (eptr < md->end_subject)
2374        {        {
2375        int chartype, script;        int len = 1;
2376        int category = _pcre_ucp_findprop(c, &chartype, &script);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2377        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (UCD_CATEGORY(c) != ucp_M) break;
2378        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;  
         }  
2379        }        }
2380      ecode++;      ecode++;
2381      break;      break;
# Line 1675  for (;;) Line 2391  for (;;)
2391      loops). */      loops). */
2392    
2393      case OP_REF:      case OP_REF:
2394        {      case OP_REFI:
2395        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2396        ecode += 3;                                 /* Advance past item */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2397        ecode += 3;
2398    
2399        /* 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];  
2400    
2401        /* Set up for repetition, or handle the non-repeated case */      (a) In the default, Perl-compatible state, set the length negative;
2402        this ensures that every attempt at a match fails. We can't just fail
2403        here, because of the possibility of quantifiers with zero minima.
2404    
2405        switch (*ecode)      (b) If the JavaScript compatibility flag is set, set the length to zero
2406          {      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;  
2407    
2408          case OP_CRRANGE:      Otherwise, set the length to the length of what was matched by the
2409          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;  
2410    
2411          default:               /* No repeat follows */      if (offset >= offset_top || md->offset_vector[offset] < 0)
2412          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);        length = (md->jscript_compat)? 0 : -1;
2413          eptr += length;      else
2414          continue;              /* With the main loop */        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2415    
2416        /* Set up for repetition, or handle the non-repeated case */
2417    
2418        switch (*ecode)
2419          {
2420          case OP_CRSTAR:
2421          case OP_CRMINSTAR:
2422          case OP_CRPLUS:
2423          case OP_CRMINPLUS:
2424          case OP_CRQUERY:
2425          case OP_CRMINQUERY:
2426          c = *ecode++ - OP_CRSTAR;
2427          minimize = (c & 1) != 0;
2428          min = rep_min[c];                 /* Pick up values from tables; */
2429          max = rep_max[c];                 /* zero for max => infinity */
2430          if (max == 0) max = INT_MAX;
2431          break;
2432    
2433          case OP_CRRANGE:
2434          case OP_CRMINRANGE:
2435          minimize = (*ecode == OP_CRMINRANGE);
2436          min = GET2(ecode, 1);
2437          max = GET2(ecode, 3);
2438          if (max == 0) max = INT_MAX;
2439          ecode += 5;
2440          break;
2441    
2442          default:               /* No repeat follows */
2443          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2444            {
2445            CHECK_PARTIAL();
2446            MRRETURN(MATCH_NOMATCH);
2447          }          }
2448          eptr += length;
2449          continue;              /* With the main loop */
2450          }
2451    
2452        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2453        main loop. */      zero, just continue with the main loop. */
2454    
2455        if (length == 0) continue;      if (length == 0) continue;
2456    
2457        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2458        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2459        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2460    
2461        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2462          {
2463          int slength;
2464          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2465          {          {
2466          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2467          eptr += length;          MRRETURN(MATCH_NOMATCH);
2468          }          }
2469          eptr += slength;
2470          }
2471    
2472        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2473        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2474    
2475        if (min == max) continue;      if (min == max) continue;
2476    
2477        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2478    
2479        if (minimize)      if (minimize)
2480          {
2481          for (fi = min;; fi++)
2482          {          {
2483          for (fi = min;; fi++)          int slength;
2484            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2485            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2486            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2487            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2488            {            {
2489            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2490            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2491            }            }
2492          /* Control never gets here */          eptr += slength;
2493          }          }
2494          /* Control never gets here */
2495          }
2496    
2497        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2498    
2499        else      else
2500          {
2501          pp = eptr;
2502          for (i = min; i < max; i++)
2503          {          {
2504          pp = eptr;          int slength;
2505          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)  
2506            {            {
2507            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2508            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2509            }            }
2510          RRETURN(MATCH_NOMATCH);          eptr += slength;
2511            }
2512          while (eptr >= pp)
2513            {
2514            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2515            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2516            eptr -= length;
2517          }          }
2518          MRRETURN(MATCH_NOMATCH);
2519        }        }
2520      /* Control never gets here */      /* Control never gets here */
2521    
   
   
2522      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2523      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,
2524      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 1832  for (;;) Line 2573  for (;;)
2573          {          {
2574          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2575            {            {
2576            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2577                {
2578                SCHECK_PARTIAL();
2579                MRRETURN(MATCH_NOMATCH);
2580                }
2581            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2582            if (c > 255)            if (c > 255)
2583              {              {
2584              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2585              }              }
2586            else            else
2587              {              {
2588              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2589              }              }
2590            }            }
2591          }          }
# Line 1850  for (;;) Line 2595  for (;;)
2595          {          {
2596          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2597            {            {
2598            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2599                {
2600                SCHECK_PARTIAL();
2601                MRRETURN(MATCH_NOMATCH);
2602                }
2603            c = *eptr++;            c = *eptr++;
2604            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2605            }            }
2606          }          }
2607    
# Line 1872  for (;;) Line 2621  for (;;)
2621            {            {
2622            for (fi = min;; fi++)            for (fi = min;; fi++)
2623              {              {
2624              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2625              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2626              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2627                if (eptr >= md->end_subject)
2628                  {
2629                  SCHECK_PARTIAL();
2630                  MRRETURN(MATCH_NOMATCH);
2631                  }
2632              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2633              if (c > 255)              if (c > 255)
2634                {                {
2635                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2636                }                }
2637              else              else
2638                {                {
2639                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2640                }                }
2641              }              }
2642            }            }
# Line 1892  for (;;) Line 2646  for (;;)
2646            {            {
2647            for (fi = min;; fi++)            for (fi = min;; fi++)
2648              {              {
2649              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2650              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2651              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2652                if (eptr >= md->end_subject)
2653                  {
2654                  SCHECK_PARTIAL();
2655                  MRRETURN(MATCH_NOMATCH);
2656                  }
2657              c = *eptr++;              c = *eptr++;
2658              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2659              }              }
2660            }            }
2661          /* Control never gets here */          /* Control never gets here */
# Line 1915  for (;;) Line 2674  for (;;)
2674            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2675              {              {
2676              int len = 1;              int len = 1;
2677              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2678                  {
2679                  SCHECK_PARTIAL();
2680                  break;
2681                  }
2682              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2683              if (c > 255)              if (c > 255)
2684                {                {
# Line 1929  for (;;) Line 2692  for (;;)
2692              }              }
2693            for (;;)            for (;;)
2694              {              {
2695              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2696              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2697              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2698              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1941  for (;;) Line 2704  for (;;)
2704            {            {
2705            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2706              {              {
2707              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2708                  {
2709                  SCHECK_PARTIAL();
2710                  break;
2711                  }
2712              c = *eptr;              c = *eptr;
2713              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2714              eptr++;              eptr++;
2715              }              }
2716            while (eptr >= pp)            while (eptr >= pp)
2717              {              {
2718              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2719              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2720              eptr--;              eptr--;
2721              }              }
2722            }            }
2723    
2724          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2725          }          }
2726        }        }
2727      /* Control never gets here */      /* Control never gets here */
2728    
2729    
2730      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2731      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
2732        mode, because Unicode properties are supported in non-UTF-8 mode. */
2733    
2734  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2735      case OP_XCLASS:      case OP_XCLASS:
# Line 2002  for (;;) Line 2770  for (;;)
2770    
2771        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2772          {          {
2773          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2774          GETCHARINC(c, eptr);            {
2775          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2776              MRRETURN(MATCH_NOMATCH);
2777              }
2778            GETCHARINCTEST(c, eptr);
2779            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2780          }          }
2781    
2782        /* 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 2019  for (;;) Line 2791  for (;;)
2791          {          {
2792          for (fi = min;; fi++)          for (fi = min;; fi++)
2793            {            {
2794            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2795            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2796            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2797            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2798            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2799                SCHECK_PARTIAL();
2800                MRRETURN(MATCH_NOMATCH);
2801                }
2802              GETCHARINCTEST(c, eptr);
2803              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2804            }            }
2805          /* Control never gets here */          /* Control never gets here */
2806          }          }
# Line 2036  for (;;) Line 2813  for (;;)
2813          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2814            {            {
2815            int len = 1;            int len = 1;
2816            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2817            GETCHARLEN(c, eptr, len);              {
2818                SCHECK_PARTIAL();
2819                break;
2820                }
2821              GETCHARLENTEST(c, eptr, len);
2822            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2823            eptr += len;            eptr += len;
2824            }            }
2825          for(;;)          for(;;)
2826            {            {
2827            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2828            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2829            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2830            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2831            }            }
2832          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2833          }          }
2834    
2835        /* Control never gets here */        /* Control never gets here */
# Line 2064  for (;;) Line 2845  for (;;)
2845        length = 1;        length = 1;
2846        ecode++;        ecode++;
2847        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2848        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2849        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2850            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2851            MRRETURN(MATCH_NOMATCH);
2852            }
2853          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2854        }        }
2855      else      else
2856  #endif  #endif
2857    
2858      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2859        {        {
2860        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2861        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2862            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2863            MRRETURN(MATCH_NOMATCH);
2864            }
2865          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2866        ecode += 2;        ecode += 2;
2867        }        }
2868      break;      break;
2869    
2870      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2871    
2872      case OP_CHARNC:      case OP_CHARI:
2873  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2874      if (utf8)      if (utf8)
2875        {        {
# Line 2088  for (;;) Line 2877  for (;;)
2877        ecode++;        ecode++;
2878        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2879    
2880        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2881            {
2882            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2883            MRRETURN(MATCH_NOMATCH);
2884            }
2885    
2886        /* 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
2887        can use the fast lookup table. */        can use the fast lookup table. */
2888    
2889        if (fc < 128)        if (fc < 128)
2890          {          {
2891          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2892          }          }
2893    
2894        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
# Line 2112  for (;;) Line 2905  for (;;)
2905          if (fc != dc)          if (fc != dc)
2906            {            {
2907  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2908            if (dc != _pcre_ucp_othercase(fc))            if (dc != UCD_OTHERCASE(fc))
2909  #endif  #endif
2910              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
2911            }            }
2912          }          }
2913        }        }
# Line 2123  for (;;) Line 2916  for (;;)
2916    
2917      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2918        {        {
2919        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2920        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
2921            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2922            MRRETURN(MATCH_NOMATCH);
2923            }
2924          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2925        ecode += 2;        ecode += 2;
2926        }        }
2927      break;      break;
# Line 2132  for (;;) Line 2929  for (;;)
2929      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2930    
2931      case OP_EXACT:      case OP_EXACT:
2932        case OP_EXACTI:
2933      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2934      ecode += 3;      ecode += 3;
2935      goto REPEATCHAR;      goto REPEATCHAR;
2936    
2937      case OP_POSUPTO:      case OP_POSUPTO:
2938        case OP_POSUPTOI:
2939      possessive = TRUE;      possessive = TRUE;
2940      /* Fall through */      /* Fall through */
2941    
2942      case OP_UPTO:      case OP_UPTO:
2943        case OP_UPTOI:
2944      case OP_MINUPTO:      case OP_MINUPTO:
2945        case OP_MINUPTOI:
2946      min = 0;      min = 0;
2947      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2948      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2949      ecode += 3;      ecode += 3;
2950      goto REPEATCHAR;      goto REPEATCHAR;
2951    
2952      case OP_POSSTAR:      case OP_POSSTAR:
2953        case OP_POSSTARI:
2954      possessive = TRUE;      possessive = TRUE;
2955      min = 0;      min = 0;
2956      max = INT_MAX;      max = INT_MAX;
# Line 2156  for (;;) Line 2958  for (;;)
2958      goto REPEATCHAR;      goto REPEATCHAR;
2959    
2960      case OP_POSPLUS:      case OP_POSPLUS:
2961        case OP_POSPLUSI:
2962      possessive = TRUE;      possessive = TRUE;
2963      min = 1;      min = 1;
2964      max = INT_MAX;      max = INT_MAX;
# Line 2163  for (;;) Line 2966  for (;;)
2966      goto REPEATCHAR;      goto REPEATCHAR;
2967    
2968      case OP_POSQUERY:      case OP_POSQUERY:
2969        case OP_POSQUERYI:
2970      possessive = TRUE;      possessive = TRUE;
2971      min = 0;      min = 0;
2972      max = 1;      max = 1;
# Line 2170  for (;;) Line 2974  for (;;)
2974      goto REPEATCHAR;      goto REPEATCHAR;
2975    
2976      case OP_STAR:      case OP_STAR:
2977        case OP_STARI:
2978      case OP_MINSTAR:      case OP_MINSTAR:
2979        case OP_MINSTARI:
2980      case OP_PLUS:      case OP_PLUS:
2981        case OP_PLUSI:
2982      case OP_MINPLUS:      case OP_MINPLUS:
2983        case OP_MINPLUSI:
2984      case OP_QUERY:      case OP_QUERY:
2985        case OP_QUERYI:
2986      case OP_MINQUERY:      case OP_MINQUERY:
2987      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
2988        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
2989      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
2990      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2991      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2992      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
2993    
2994      /* 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. */  
2995    
2996      REPEATCHAR:      REPEATCHAR:
2997  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2192  for (;;) Line 3000  for (;;)
3000        length = 1;        length = 1;
3001        charptr = ecode;        charptr = ecode;
3002        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3003        ecode += length;        ecode += length;
3004    
3005        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 2202  for (;;) Line 3009  for (;;)
3009          {          {
3010  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3011          unsigned int othercase;          unsigned int othercase;
3012          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3013              (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)              (othercase = UCD_OTHERCASE(fc)) != fc)
3014            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3015          else oclength = 0;          else oclength = 0;
3016  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3017    
3018          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3019            {            {
3020            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3021                memcmp(eptr, charptr, length) == 0) eptr += length;
3022  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3023            /* Need braces because of following else */            else if (oclength > 0 &&
3024            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                     eptr <= md->end_subject - oclength &&
3025                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3026    #endif  /* SUPPORT_UCP */
3027            else            else
3028              {              {
3029              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3030              eptr += oclength;              MRRETURN(MATCH_NOMATCH);
3031              }              }
 #else   /* without SUPPORT_UCP */  
           else { RRETURN(MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3032            }            }
3033    
3034          if (min == max) continue;          if (min == max) continue;
# Line 2230  for (;;) Line 3037  for (;;)
3037            {            {
3038            for (fi = min;; fi++)            for (fi = min;; fi++)
3039              {              {
3040              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3041              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3042              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3043              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3044                  memcmp(eptr, charptr, length) == 0) eptr += length;
3045  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3046              /* Need braces because of following else */              else if (oclength > 0 &&
3047              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                       eptr <= md->end_subject - oclength &&
3048                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3049    #endif  /* SUPPORT_UCP */
3050              else              else
3051                {                {
3052                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3053                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3054                }                }
 #else   /* without SUPPORT_UCP */  
             else { RRETURN (MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3055              }              }
3056            /* Control never gets here */            /* Control never gets here */
3057            }            }
# Line 2254  for (;;) Line 3061  for (;;)
3061            pp = eptr;            pp = eptr;
3062            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3063              {              {
3064              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3065              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3066  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3067              else if (oclength == 0) break;              else if (oclength > 0 &&
3068                         eptr <= md->end_subject - oclength &&
3069                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3070    #endif  /* SUPPORT_UCP */
3071              else              else
3072                {                {
3073                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3074                eptr += oclength;                break;
3075                }                }
 #else   /* without SUPPORT_UCP */  
             else break;  
 #endif  /* SUPPORT_UCP */  
3076              }              }
3077    
3078            if (possessive) continue;            if (possessive) continue;
3079    
3080            for(;;)            for(;;)
3081             {              {
3082             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3083             if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3084             if (eptr == pp) RRETURN(MATCH_NOMATCH);              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3085  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3086             eptr--;              eptr--;
3087             BACKCHAR(eptr);              BACKCHAR(eptr);
3088  #else   /* without SUPPORT_UCP */  #else   /* without SUPPORT_UCP */
3089             eptr -= length;              eptr -= length;
3090  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3091             }              }
3092            }            }
3093          /* Control never gets here */          /* Control never gets here */
3094          }          }
# Line 2293  for (;;) Line 3101  for (;;)
3101  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3102    
3103      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3104        {  
3105        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3106    
3107      /* 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
3108      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 2310  for (;;) Line 3116  for (;;)
3116      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3117        max, eptr));        max, eptr));
3118    
3119      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3120        {        {
3121        fc = md->lcc[fc];        fc = md->lcc[fc];
3122        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3123          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3124            if (eptr >= md->end_subject)
3125              {
3126              SCHECK_PARTIAL();
3127              MRRETURN(MATCH_NOMATCH);
3128              }
3129            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3130            }
3131        if (min == max) continue;        if (min == max) continue;
3132        if (minimize)        if (minimize)
3133          {          {
3134          for (fi = min;; fi++)          for (fi = min;; fi++)
3135            {            {
3136            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3137            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3138            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3139                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3140              RRETURN(MATCH_NOMATCH);              {
3141                SCHECK_PARTIAL();
3142                MRRETURN(MATCH_NOMATCH);
3143                }
3144              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3145            }            }
3146          /* Control never gets here */          /* Control never gets here */
3147          }          }
# Line 2333  for (;;) Line 3150  for (;;)
3150          pp = eptr;          pp = eptr;
3151          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3152            {            {
3153            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3154                {
3155                SCHECK_PARTIAL();
3156                break;
3157                }
3158              if (fc != md->lcc[*eptr]) break;
3159            eptr++;            eptr++;
3160            }            }
3161    
3162          if (possessive) continue;          if (possessive) continue;
3163    
3164          while (eptr >= pp)          while (eptr >= pp)
3165            {            {
3166            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3167            eptr--;            eptr--;
3168            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3169            }            }
3170          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3171          }          }
3172        /* Control never gets here */        /* Control never gets here */
3173        }        }
# Line 2352  for (;;) Line 3176  for (;;)
3176    
3177      else      else
3178        {        {
3179        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3180            {
3181            if (eptr >= md->end_subject)
3182              {
3183              SCHECK_PARTIAL();
3184              MRRETURN(MATCH_NOMATCH);
3185              }
3186            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3187            }
3188    
3189        if (min == max) continue;        if (min == max) continue;
3190    
3191        if (minimize)        if (minimize)
3192          {          {
3193          for (fi = min;; fi++)          for (fi = min;; fi++)
3194            {            {
3195            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3196            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3197            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3198              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3199                {
3200                SCHECK_PARTIAL();
3201                MRRETURN(MATCH_NOMATCH);
3202                }
3203              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3204            }            }
3205          /* Control never gets here */          /* Control never gets here */
3206          }          }
# Line 2370  for (;;) Line 3209  for (;;)
3209          pp = eptr;          pp = eptr;
3210          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3211            {            {
3212            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3213                {
3214                SCHECK_PARTIAL();
3215                break;
3216                }
3217              if (fc != *eptr) break;
3218            eptr++;            eptr++;
3219            }            }
3220          if (possessive) continue;          if (possessive) continue;
3221    
3222          while (eptr >= pp)          while (eptr >= pp)
3223            {            {
3224            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3225            eptr--;            eptr--;
3226            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3227            }            }
3228          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3229          }          }
3230        }        }
3231      /* Control never gets here */      /* Control never gets here */
# Line 2389  for (;;) Line 3234  for (;;)
3234      checking can be multibyte. */      checking can be multibyte. */
3235    
3236      case OP_NOT:      case OP_NOT:
3237      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3238        if (eptr >= md->end_subject)
3239          {
3240          SCHECK_PARTIAL();
3241          MRRETURN(MATCH_NOMATCH);
3242          }
3243      ecode++;      ecode++;
3244      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3245      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3246        {        {
3247  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3248        if (c < 256)        if (c < 256)
3249  #endif  #endif
3250        c = md->lcc[c];        c = md->lcc[c];
3251        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3252        }        }
3253      else      else    /* Caseful */
3254        {        {
3255        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3256        }        }
3257      break;      break;
3258    
# Line 2414  for (;;) Line 3264  for (;;)
3264      about... */      about... */
3265    
3266      case OP_NOTEXACT:      case OP_NOTEXACT:
3267        case OP_NOTEXACTI:
3268      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3269      ecode += 3;      ecode += 3;
3270      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3271    
3272      case OP_NOTUPTO:      case OP_NOTUPTO:
3273        case OP_NOTUPTOI:
3274      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3275        case OP_NOTMINUPTOI:
3276      min = 0;      min = 0;
3277      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3278      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3279      ecode += 3;      ecode += 3;
3280      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3281    
3282      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3283        case OP_NOTPOSSTARI:
3284      possessive = TRUE;      possessive = TRUE;
3285      min = 0;      min = 0;
3286      max = INT_MAX;      max = INT_MAX;
# Line 2434  for (;;) Line 3288  for (;;)
3288      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3289    
3290      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3291        case OP_NOTPOSPLUSI:
3292      possessive = TRUE;      possessive = TRUE;
3293      min = 1;      min = 1;
3294      max = INT_MAX;      max = INT_MAX;
# Line 2441  for (;;) Line 3296  for (;;)
3296      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3297    
3298      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3299        case OP_NOTPOSQUERYI:
3300      possessive = TRUE;      possessive = TRUE;
3301      min = 0;      min = 0;
3302      max = 1;      max = 1;
# Line 2448  for (;;) Line 3304  for (;;)
3304      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3305    
3306      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3307        case OP_NOTPOSUPTOI:
3308      possessive = TRUE;      possessive = TRUE;
3309      min = 0;      min = 0;
3310      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 2455  for (;;) Line 3312  for (;;)
3312      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3313    
3314      case OP_NOTSTAR:      case OP_NOTSTAR:
3315        case OP_NOTSTARI:
3316      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3317        case OP_NOTMINSTARI:
3318      case OP_NOTPLUS:      case OP_NOTPLUS:
3319        case OP_NOTPLUSI:
3320      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3321        case OP_NOTMINPLUSI:
3322      case OP_NOTQUERY:      case OP_NOTQUERY:
3323        case OP_NOTQUERYI:
3324      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3325      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3326        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3327      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3328      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3329      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3330      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3331    
3332      /* 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. */  
3333    
3334      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3335      fc = *ecode++;      fc = *ecode++;
3336    
3337      /* 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 2485  for (;;) Line 3345  for (;;)
3345      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,
3346        max, eptr));        max, eptr));
3347    
3348      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3349        {        {
3350        fc = md->lcc[fc];        fc = md->lcc[fc];
3351    
# Line 2496  for (;;) Line 3356  for (;;)
3356          register unsigned int d;          register unsigned int d;
3357          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3358            {            {
3359              if (eptr >= md->end_subject)
3360                {
3361                SCHECK_PARTIAL();
3362                MRRETURN(MATCH_NOMATCH);
3363                }
3364            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3365            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3366            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3367            }            }
3368          }          }
3369        else        else
# Line 2507  for (;;) Line 3372  for (;;)
3372        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3373          {          {
3374          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3375            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3376              if (eptr >= md->end_subject)
3377                {
3378                SCHECK_PARTIAL();
3379                MRRETURN(MATCH_NOMATCH);
3380                }
3381              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3382              }
3383          }          }
3384    
3385        if (min == max) continue;        if (min == max) continue;
# Line 2521  for (;;) Line 3393  for (;;)
3393            register unsigned int d;            register unsigned int d;
3394            for (fi = min;; fi++)            for (fi = min;; fi++)
3395              {              {
3396              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3397              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3398                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3399                if (eptr >= md->end_subject)
3400                  {
3401                  SCHECK_PARTIAL();
3402                  MRRETURN(MATCH_NOMATCH);
3403                  }
3404              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3405              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3406              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3407              }              }
3408            }            }
3409          else          else
# Line 2535  for (;;) Line 3412  for (;;)
3412            {            {
3413            for (fi = min;; fi++)            for (fi = min;; fi++)
3414              {              {
3415              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3416              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3417              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3418                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3419                  {
3420                  SCHECK_PARTIAL();
3421                  MRRETURN(MATCH_NOMATCH);
3422                  }
3423                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3424              }              }
3425            }            }
3426          /* Control never gets here */          /* Control never gets here */
# Line 2558  for (;;) Line 3440  for (;;)
3440            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3441              {              {
3442              int len = 1;              int len = 1;
3443              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3444                  {
3445                  SCHECK_PARTIAL();
3446                  break;
3447                  }
3448              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3449              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3450              if (fc == d) break;              if (fc == d) break;
# Line 2567  for (;;) Line 3453  for (;;)
3453          if (possessive) continue;          if (possessive) continue;
3454          for(;;)          for(;;)
3455              {              {
3456              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3457              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3458              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3459              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2579  for (;;) Line 3465  for (;;)
3465            {            {
3466            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3467              {              {
3468              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3469                  {
3470                  SCHECK_PARTIAL();
3471                  break;
3472                  }
3473                if (fc == md->lcc[*eptr]) break;
3474              eptr++;              eptr++;
3475              }              }
3476            if (possessive) continue;            if (possessive) continue;
3477            while (eptr >= pp)            while (eptr >= pp)
3478              {              {
3479              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3480              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3481              eptr--;              eptr--;
3482              }              }
3483            }            }
3484    
3485          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3486          }          }
3487        /* Control never gets here */        /* Control never gets here */
3488        }        }
# Line 2607  for (;;) Line 3498  for (;;)
3498          register unsigned int d;          register unsigned int d;
3499          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3500            {            {
3501              if (eptr >= md->end_subject)
3502                {
3503                SCHECK_PARTIAL();
3504                MRRETURN(MATCH_NOMATCH);
3505                }
3506            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3507            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3508            }            }
3509          }          }
3510        else        else
# Line 2616  for (;;) Line 3512  for (;;)
3512        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3513          {          {
3514          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3515            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3516              if (eptr >= md->end_subject)
3517                {
3518                SCHECK_PARTIAL();
3519                MRRETURN(MATCH_NOMATCH);
3520                }
3521              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3522              }
3523          }          }
3524    
3525        if (min == max) continue;        if (min == max) continue;
# Line 2630  for (;;) Line 3533  for (;;)
3533            register unsigned int d;            register unsigned int d;
3534            for (fi = min;; fi++)            for (fi = min;; fi++)
3535              {              {
3536              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3537              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3538                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3539                if (eptr >= md->end_subject)
3540                  {
3541                  SCHECK_PARTIAL();
3542                  MRRETURN(MATCH_NOMATCH);
3543                  }
3544              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3545              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3546              }              }
3547            }            }
3548          else          else
# Line 2643  for (;;) Line 3551  for (;;)
3551            {            {
3552            for (fi = min;; fi++)            for (fi = min;; fi++)
3553              {              {
3554              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3555              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3556              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3557                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3558                  {
3559                  SCHECK_PARTIAL();
3560                  MRRETURN(MATCH_NOMATCH);
3561                  }
3562                if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3563              }              }
3564            }            }
3565          /* Control never gets here */          /* Control never gets here */
# Line 2666  for (;;) Line 3579  for (;;)
3579            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3580              {              {
3581              int len = 1;              int len = 1;
3582              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3583                  {
3584                  SCHECK_PARTIAL();
3585                  break;
3586                  }
3587              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3588              if (fc == d) break;              if (fc == d) break;
3589              eptr += len;              eptr += len;
# Line 2674  for (;;) Line 3591  for (;;)
3591            if (possessive) continue;            if (possessive) continue;
3592            for(;;)            for(;;)
3593              {              {
3594              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3595              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3596              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3597              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2686  for (;;) Line 3603  for (;;)
3603            {            {
3604            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3605              {              {
3606              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3607                  {
3608                  SCHECK_PARTIAL();
3609                  break;
3610                  }
3611                if (fc == *eptr) break;
3612              eptr++;              eptr++;
3613              }              }
3614            if (possessive) continue;            if (possessive) continue;
3615            while (eptr >= pp)            while (eptr >= pp)
3616              {              {
3617              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3618              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3619              eptr--;              eptr--;
3620              }              }
3621            }            }
3622    
3623          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3624          }          }
3625        }        }
3626      /* Control never gets here */      /* Control never gets here */
# Line 2780  for (;;) Line 3702  for (;;)
3702    
3703      /* First, ensure the minimum number of matches are present. Use inline      /* First, ensure the minimum number of matches are present. Use inline
3704      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
3705      (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  
3706      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
3707      and single-bytes. */      and single-bytes. */
3708    
3709      if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      if (min > 0)
3710      if (min > 0)        {
3711        {  #ifdef SUPPORT_UCP
3712  #ifdef SUPPORT_UCP        if (prop_type >= 0)
3713        if (prop_type >= 0)          {
3714          {          switch(prop_type)
3715          switch(prop_type)            {
3716            {            case PT_ANY:
3717            case PT_ANY:            if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
3718            if (prop_fail_result) RRETURN(MATCH_NOMATCH);            for (i = 1; i <= min; i++)
3719                {
3720                if (eptr >= md->end_subject)
3721                  {
3722                  SCHECK_PARTIAL();
3723                  MRRETURN(MATCH_NOMATCH);
3724                  }
3725                GETCHARINCTEST(c, eptr);
3726                }
3727              break;
3728    
3729              case PT_LAMP:
3730              for (i = 1; i <= min; i++)
3731                {
3732                int chartype;
3733                if (eptr >= md->end_subject)
3734                  {
3735                  SCHECK_PARTIAL();
3736                  MRRETURN(MATCH_NOMATCH);
3737                  }
3738                GETCHARINCTEST(c, eptr);
3739                chartype = UCD_CHARTYPE(c);
3740                if ((chartype == ucp_Lu ||
3741                     chartype == ucp_Ll ||
3742                     chartype == ucp_Lt) == prop_fail_result)
3743                  MRRETURN(MATCH_NOMATCH);
3744                }
3745              break;
3746    
3747              case PT_GC:
3748              for (i = 1; i <= min; i++)
3749                {
3750                if (eptr >= md->end_subject)
3751                  {
3752                  SCHECK_PARTIAL();
3753                  MRRETURN(MATCH_NOMATCH);
3754                  }
3755                GETCHARINCTEST(c, eptr);
3756                if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
3757                  MRRETURN(MATCH_NOMATCH);
3758                }
3759              break;
3760    
3761              case PT_PC:
3762              for (i = 1; i <= min; i++)
3763                {
3764                if (eptr >= md->end_subject)
3765                  {
3766                  SCHECK_PARTIAL();
3767                  MRRETURN(MATCH_NOMATCH);
3768                  }
3769                GETCHARINCTEST(c, eptr);
3770                if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
3771                  MRRETURN(MATCH_NOMATCH);
3772                }
3773              break;
3774    
3775              case PT_SC:
3776            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3777              {              {
3778              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3779                  {
3780                  SCHECK_PARTIAL();
3781                  MRRETURN(MATCH_NOMATCH);
3782                  }
3783              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3784                if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
3785                  MRRETURN(MATCH_NOMATCH);
3786              }              }
3787            break;            break;
3788    
3789            case PT_LAMP:            case PT_ALNUM:
3790            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3791              {              {
3792              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              int category;
3793                if (eptr >= md->end_subject)
3794                  {
3795                  SCHECK_PARTIAL();
3796                  MRRETURN(MATCH_NOMATCH);
3797                  }
3798              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3799              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);              category = UCD_CATEGORY(c);
3800              if ((prop_chartype == ucp_Lu ||              if ((category == ucp_L || category == ucp_N) == prop_fail_result)
3801                   prop_chartype == ucp_Ll ||                MRRETURN(MATCH_NOMATCH);
                  prop_chartype == ucp_Lt) == prop_fail_result)  
               RRETURN(MATCH_NOMATCH);  
3802              }              }
3803            break;            break;
3804    
3805            case PT_GC:            case PT_SPACE:    /* Perl space */
3806            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3807              {              {
3808              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3809                  {
3810                  SCHECK_PARTIAL();
3811                  MRRETURN(MATCH_NOMATCH);
3812                  }
3813              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3814              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3815              if ((prop_category == prop_value) == prop_fail_result)                   c == CHAR_FF || c == CHAR_CR)
3816                RRETURN(MATCH_NOMATCH);                     == prop_fail_result)
3817                  MRRETURN(MATCH_NOMATCH);
3818              }              }
3819            break;            break;
3820    
3821            case PT_PC:            case PT_PXSPACE:  /* POSIX space */
3822            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3823              {              {
3824              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3825                  {
3826                  SCHECK_PARTIAL();
3827                  MRRETURN(MATCH_NOMATCH);
3828                  }
3829              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3830              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3831              if ((prop_chartype == prop_value) == prop_fail_result)                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3832                RRETURN(MATCH_NOMATCH);                     == prop_fail_result)
3833                  MRRETURN(MATCH_NOMATCH);
3834              }              }
3835            break;            break;
3836    
3837            case PT_SC:            case PT_WORD:
3838            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3839              {              {
3840              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              int category;
3841                if (eptr >= md->end_subject)
3842                  {
3843                  SCHECK_PARTIAL();
3844                  MRRETURN(MATCH_NOMATCH);
3845                  }
3846              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
3847              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);              category = UCD_CATEGORY(c);
3848              if ((prop_script == prop_value) == prop_fail_result)              if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE)
3849                RRETURN(MATCH_NOMATCH);                     == prop_fail_result)
3850                  MRRETURN(MATCH_NOMATCH);
3851              }              }
3852            break;            break;
3853    
3854              /* This should not occur */
3855    
3856            default:            default:
3857            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
3858            }            }
# Line 2861  for (;;) Line 3865  for (;;)
3865          {          {
3866          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3867            {            {
3868              if (eptr >= md->end_subject)
3869                {
3870                SCHECK_PARTIAL();
3871                MRRETURN(MATCH_NOMATCH);
3872                }
3873            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3874            prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);            if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
           if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);  
3875            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3876              {              {
3877              int len = 1;              int len = 1;
3878              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
3879                {              if (UCD_CATEGORY(c) != ucp_M) break;
               GETCHARLEN(c, eptr, len);  
               }  
             prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);  
             if (prop_category != ucp_M) break;  
3880              eptr += len;              eptr += len;
3881              }              }
3882            }            }
# Line 2889  for (;;) Line 3893  for (;;)
3893          case OP_ANY:          case OP_ANY:
3894          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3895            {            {
3896            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject)
3897                 ((ims & PCRE_DOTALL) == 0 && IS_NEWLINE(eptr)))              {
3898              RRETURN(MATCH_NOMATCH);              SCHECK_PARTIAL();
3899                MRRETURN(MATCH_NOMATCH);
3900                }
3901              if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
3902              eptr++;
3903              while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
3904              }
3905            break;
3906    
3907            case OP_ALLANY:
3908            for (i = 1; i <= min; i++)
3909              {
3910              if (eptr >= md->end_subject)
3911                {
3912                SCHECK_PARTIAL();
3913                MRRETURN(MATCH_NOMATCH);
3914                }
3915            eptr++;            eptr++;
3916            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
3917            }            }
3918          break;          break;
3919    
3920          case OP_ANYBYTE:          case OP_ANYBYTE:
3921            if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH);
3922          eptr += min;          eptr += min;
3923          break;          break;
3924    
3925          case OP_ANYNL:          case OP_ANYNL:
3926          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3927            {            {
3928            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3929                {
3930                SCHECK_PARTIAL();
3931                MRRETURN(MATCH_NOMATCH);
3932                }
3933            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
3934            switch(c)            switch(c)
3935              {              {
3936              default: RRETURN(MATCH_NOMATCH);              default: MRRETURN(MATCH_NOMATCH);
3937    
3938              case 0x000d:              case 0x000d:
3939              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;              if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3940              break;              break;
3941    
3942              case 0x000a:              case 0x000a:
3943                break;
3944    
3945              case 0x000b:              case 0x000b:
3946              case 0x000c:              case 0x000c:
3947              case 0x0085:              case 0x0085:
3948              case 0x2028:              case 0x2028:
3949              case 0x2029:              case 0x2029:
3950                if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
3951              break;              break;
3952              }              }
3953            }            }
# Line 2926  for (;;) Line 3956  for (;;)
3956          case OP_NOT_HSPACE:          case OP_NOT_HSPACE:
3957          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3958            {            {
3959            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3960                {
3961                SCHECK_PARTIAL();
3962                MRRETURN(MATCH_NOMATCH);
3963                }
3964            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
3965            switch(c)            switc