/[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 165 by ph10, Wed May 9 10:50:57 2007 UTC revision 619 by ph10, Sun Jul 17 13:23:14 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  /* The chain of eptrblocks for tail recursions uses memory in stack workspace,  /* Values for setting in md->match_function_type to indicate two special types
61  obtained at top level, the size of which is defined by EPTR_WORK_SIZE. */  of call to match(). We do it this way to save on using another stack variable,
62    as stack usage is to be discouraged. */
 #define EPTR_WORK_SIZE (1000)  
   
 /* Flag bits for the match() function */  
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 */
 #define match_tail_recursed  0x04  /* Tail recursive call */  
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 70  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 83  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 116  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 148  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 205  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 };         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, RM64, RM65, RM66 };
281    
282  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
283  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
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,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 235  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,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 249  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;\
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 270  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 289  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      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 313  typedef struct heapframe { Line 380  typedef struct heapframe {
380    BOOL Xcondition;    BOOL Xcondition;
381    BOOL Xprev_is_word;    BOOL Xprev_is_word;
382    
   unsigned long int Xoriginal_ims;  
   
383  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
# Line 326  typedef struct heapframe { Line 391  typedef struct heapframe {
391    uschar Xocchars[8];    uschar Xocchars[8];
392  #endif  #endif
393    
394      int Xcodelink;
395    int Xctype;    int Xctype;
396    unsigned int Xfc;    unsigned int Xfc;
397    int Xfi;    int Xfi;
# Line 361  typedef struct heapframe { Line 427  typedef struct heapframe {
427    
428  /* This function is called recursively in many circumstances. Whenever it  /* This function is called recursively in many circumstances. Whenever it
429  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
430  same response.  same response. */
431    
432    /* These macros pack up tests that are used for partial matching, and which
433    appears several times in the code. We set the "hit end" flag if the pointer is
434    at the end of the subject and also past the start of the subject (i.e.
435    something has been matched). For hard partial matching, we then return
436    immediately. The second one is used when we already know we are past the end of
437    the subject. */
438    
439    #define CHECK_PARTIAL()\
440      if (md->partial != 0 && eptr >= md->end_subject && \
441          eptr > md->start_used_ptr) \
442        { \
443        md->hitend = TRUE; \
444        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
445        }
446    
447    #define SCHECK_PARTIAL()\
448      if (md->partial != 0 && eptr > md->start_used_ptr) \
449        { \
450        md->hitend = TRUE; \
451        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
452        }
453    
454    
455  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
456  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
457  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
458  made performance worse.  made performance worse.
459    
460  Arguments:  Arguments:
461     eptr        pointer to current character in subject     eptr        pointer to current character in subject
462     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
463       mstart      pointer to the current match start position (can be modified
464                     by encountering \K)
465       markptr     pointer to the most recent MARK name, or NULL
466     offset_top  current top pointer     offset_top  current top pointer
467     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
468     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
469                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
                  match_tail_recursed - this is a tail_recursed group  
470     rdepth      the recursion depth     rdepth      the recursion depth
471    
472  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
473                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
474                   a negative MATCH_xxx value for PRUNE, SKIP, etc
475                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
476                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
477  */  */
478    
479  static int  static int
480  match(REGISTER USPTR eptr, REGISTER const uschar *ecode,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
481    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
482    int flags, unsigned int rdepth)    unsigned int rdepth)
483  {  {
484  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
485  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 404  register unsigned int c;   /* Character Line 491  register unsigned int c;   /* Character
491  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
492    
493  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
494    BOOL caseless;
495    int condcode;
496    
497  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
498  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 411  heap storage. Set up the top-level frame Line 500  heap storage. Set up the top-level frame
500  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
501    
502  #ifdef NO_RECURSE  #ifdef NO_RECURSE
503  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
504    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
505  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
506    
507  /* Copy in the original argument variables */  /* Copy in the original argument variables */
508    
509  frame->Xeptr = eptr;  frame->Xeptr = eptr;
510  frame->Xecode = ecode;  frame->Xecode = ecode;
511    frame->Xmstart = mstart;
512    frame->Xmarkptr = markptr;
513  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
514  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
515  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
516    
517  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 432  HEAP_RECURSE: Line 522  HEAP_RECURSE:
522    
523  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
524  #define ecode              frame->Xecode  #define ecode              frame->Xecode
525    #define mstart             frame->Xmstart
526    #define markptr            frame->Xmarkptr
527  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
528  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
529  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
530    
531  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 444  HEAP_RECURSE: Line 534  HEAP_RECURSE:
534  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
535  #endif  #endif
536  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
537    #define codelink           frame->Xcodelink
538  #define data               frame->Xdata  #define data               frame->Xdata
539  #define next               frame->Xnext  #define next               frame->Xnext
540  #define pp                 frame->Xpp  #define pp                 frame->Xpp
# Line 456  HEAP_RECURSE: Line 547  HEAP_RECURSE:
547  #define condition          frame->Xcondition  #define condition          frame->Xcondition
548  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
549    
 #define original_ims       frame->Xoriginal_ims  
   
550  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
551  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
552  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
# Line 494  i, and fc and c, can be the same variabl Line 583  i, and fc and c, can be the same variabl
583  #define fi i  #define fi i
584  #define fc c  #define fc c
585    
586    /* Many of the following variables are used only in small blocks of the code.
587  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  My normal style of coding would have declared them within each of those blocks.
588  const uschar *charptr;             /* in small blocks of the code. My normal */  However, in order to accommodate the version of this code that uses an external
589  #endif                             /* style of coding would have declared    */  "stack" implemented on the heap, it is easier to declare them all here, so the
590  const uschar *callpat;             /* them within each of those blocks.      */  declarations can be cut out in a block. The only declarations within blocks
591  const uschar *data;                /* However, in order to accommodate the   */  below are for variables that do not have to be preserved over a recursive call
592  const uschar *next;                /* version of this code that uses an      */  to RMATCH(). */
593  USPTR         pp;                  /* external "stack" implemented on the    */  
594  const uschar *prev;                /* heap, it is easier to declare them all */  #ifdef SUPPORT_UTF8
595  USPTR         saved_eptr;          /* here, so the declarations can be cut   */  const uschar *charptr;
596                                     /* out in a block. The only declarations  */  #endif
597  recursion_info new_recursive;      /* within blocks below are for variables  */  const uschar *callpat;
598                                     /* that do not have to be preserved over  */  const uschar *data;
599  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  const uschar *next;
600    USPTR         pp;
601    const uschar *prev;
602    USPTR         saved_eptr;
603    
604    recursion_info new_recursive;
605    
606    BOOL cur_is_word;
607  BOOL condition;  BOOL condition;
608  BOOL prev_is_word;  BOOL prev_is_word;
609    
 unsigned long int original_ims;  
   
610  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
611  int prop_type;  int prop_type;
612  int prop_value;  int prop_value;
# Line 524  int oclength; Line 618  int oclength;
618  uschar occhars[8];  uschar occhars[8];
619  #endif  #endif
620    
621    int codelink;
622  int ctype;  int ctype;
623  int length;  int length;
624  int max;  int max;
# Line 538  int stacksave[REC_STACK_SAVE_MAX]; Line 633  int stacksave[REC_STACK_SAVE_MAX];
633  eptrblock newptrb;  eptrblock newptrb;
634  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
635    
636    /* To save space on the stack and in the heap frame, I have doubled up on some
637    of the local variables that are used only in localised parts of the code, but
638    still need to be preserved over recursive calls of match(). These macros define
639    the alternative names that are used. */
640    
641    #define allow_zero    cur_is_word
642    #define cbegroup      condition
643    #define code_offset   codelink
644    #define condassert    condition
645    #define matched_once  prev_is_word
646    
647  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
648  variables. */  variables. */
649    
# Line 557  TAIL_RECURSE: Line 663  TAIL_RECURSE:
663  /* 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
664  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
665  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()
666  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
667  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
668  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,
669  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
# Line 574  haven't exceeded the recursive call limi Line 680  haven't exceeded the recursive call limi
680  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
681  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
682    
 original_ims = ims;    /* Save for resetting on ')' */  
   
683  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
684  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
685  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
686  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
687  When match() is called in other circumstances, don't add to the chain. If this  
688  is a tail recursion, use a block from the workspace, as the one on the stack is  When MATCH_CBEGROUP is set, add the current subject pointer to the chain of
689  already used. */  such remembered pointers, to be checked when we hit the closing ket, in order
690    to break infinite loops that match no characters. When match() is called in
691    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
692    NOT be used with tail recursion, because the memory block that is used is on
693    the stack, so a new one may be required for each match(). */
694    
695  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
696    {    {
697    eptrblock *p;    newptrb.epb_saved_eptr = eptr;
698    if ((flags & match_tail_recursed) != 0)    newptrb.epb_prev = eptrb;
699      {    eptrb = &newptrb;
700      if (md->eptrn >= EPTR_WORK_SIZE) RRETURN(PCRE_ERROR_NULLWSLIMIT);    md->match_function_type = 0;
     p = md->eptrchain + md->eptrn++;  
     }  
   else p = &newptrb;  
   p->epb_saved_eptr = eptr;  
   p->epb_prev = eptrb;  
   eptrb = p;  
701    }    }
702    
703  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 604  for (;;) Line 706  for (;;)
706    {    {
707    minimize = possessive = FALSE;    minimize = possessive = FALSE;
708    op = *ecode;    op = *ecode;
709    
   /* 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 > md->start_match)  
     md->hitend = TRUE;  
   
710    switch(op)    switch(op)
711      {      {
712      /* Handle a capturing bracket. If there is space in the offset vector, save      case OP_MARK:
713      the current subject position in the working slot at the top of the vector.      markptr = ecode + 2;
714      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,
715      set from a previous iteration of this group, and be referred to by a        eptrb, RM55);
716      reference inside the group.  
717        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
718      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
719      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
720      of the same bracket.      variable). If it does match, we reset that variable to the current subject
721        position and return MATCH_SKIP. Otherwise, pass back the return code
722        unaltered. */
723    
724        if (rrc == MATCH_SKIP_ARG &&
725            strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
726          {
727          md->start_match_ptr = eptr;
728          RRETURN(MATCH_SKIP);
729          }
730    
731        if (md->mark == NULL) md->mark = markptr;
732        RRETURN(rrc);
733    
734        case OP_FAIL:
735        MRRETURN(MATCH_NOMATCH);
736    
737        /* COMMIT overrides PRUNE, SKIP, and THEN */
738    
739        case OP_COMMIT:
740        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
741          eptrb, RM52);
742        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
743            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
744            rrc != MATCH_THEN)
745          RRETURN(rrc);
746        MRRETURN(MATCH_COMMIT);
747    
748        /* PRUNE overrides THEN */
749    
750        case OP_PRUNE:
751        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
752          eptrb, RM51);
753        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
754        MRRETURN(MATCH_PRUNE);
755    
756        case OP_PRUNE_ARG:
757        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
758          eptrb, RM56);
759        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
760        md->mark = ecode + 2;
761        RRETURN(MATCH_PRUNE);
762    
763        /* SKIP overrides PRUNE and THEN */
764    
765        case OP_SKIP:
766        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
767          eptrb, RM53);
768        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
769          RRETURN(rrc);
770        md->start_match_ptr = eptr;   /* Pass back current position */
771        MRRETURN(MATCH_SKIP);
772    
773        case OP_SKIP_ARG:
774        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
775          eptrb, RM57);
776        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
777          RRETURN(rrc);
778    
779        /* Pass back the current skip name by overloading md->start_match_ptr and
780        returning the special MATCH_SKIP_ARG return code. This will either be
781        caught by a matching MARK, or get to the top, where it is treated the same
782        as PRUNE. */
783    
784        md->start_match_ptr = ecode + 2;
785        RRETURN(MATCH_SKIP_ARG);
786    
787        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
788        the alt that is at the start of the current branch. This makes it possible
789        to skip back past alternatives that precede the THEN within the current
790        branch. */
791    
792        case OP_THEN:
793        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
794          eptrb, RM54);
795        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
796        md->start_match_ptr = ecode - GET(ecode, 1);
797        MRRETURN(MATCH_THEN);
798    
799        case OP_THEN_ARG:
800        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
801          offset_top, md, eptrb, RM58);
802        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
803        md->start_match_ptr = ecode - GET(ecode, 1);
804        md->mark = ecode + LINK_SIZE + 2;
805        RRETURN(MATCH_THEN);
806    
807        /* Handle a capturing bracket, other than those that are possessive with an
808        unlimited repeat. If there is space in the offset vector, save the current
809        subject position in the working slot at the top of the vector. We mustn't
810        change the current values of the data slot, because they may be set from a
811        previous iteration of this group, and be referred to by a reference inside
812        the group. A failure to match might occur after the group has succeeded,
813        if something later on doesn't match. For this reason, we need to restore
814        the working value and also the values of the final offsets, in case they
815        were set by a previous iteration of the same bracket.
816    
817      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
818      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 633  for (;;) Line 822  for (;;)
822      case OP_SCBRA:      case OP_SCBRA:
823      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
824      offset = number << 1;      offset = number << 1;
825    
826  #ifdef DEBUG  #ifdef PCRE_DEBUG
827      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
828      printf("subject=");      printf("subject=");
829      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 649  for (;;) Line 838  for (;;)
838        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
839    
840        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
841        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
842            (int)(eptr - md->start_subject);
843    
844        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
845          {          {
846          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
847            ims, eptrb, flags, RM1);          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
848          if (rrc != MATCH_NOMATCH) RRETURN(rrc);            eptrb, RM1);
849            if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
850            if (rrc != MATCH_NOMATCH &&
851                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
852              RRETURN(rrc);
853          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
854          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
855            if (*ecode != OP_ALT) break;
856          }          }
       while (*ecode == OP_ALT);  
857    
858        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
859        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
860        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
861        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
862    
863          /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
864          MATCH_THEN. */
865    
866        RRETURN(MATCH_NOMATCH);        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
867          RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
868        }        }
869    
870      /* Insufficient room for saving captured contents. Treat as a non-capturing      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
871      bracket. */      as a non-capturing bracket. */
872    
873        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
874        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
875    
876      DPRINTF(("insufficient capture room: treat as non-capturing\n"));      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
877    
878      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
879      final alternative within the brackets, we would return the result of a      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
     recursive call to match() whatever happened. We can reduce stack usage by  
     turning this into a tail recursion. */  
880    
881        /* Non-capturing or atomic group, except for possessive with unlimited
882        repeat. Loop for all the alternatives. When we get to the final alternative
883        within the brackets, we used to return the result of a recursive call to
884        match() whatever happened so it was possible to reduce stack usage by
885        turning this into a tail recursion, except in the case of a possibly empty
886        group. However, now that there is the possiblity of (*THEN) occurring in
887        the final alternative, this optimization is no longer possible.
888    
889        MATCH_ONCE is returned when the end of an atomic group is successfully
890        reached, but subsequent matching fails. It passes back up the tree (causing
891        captured values to be reset) until the original atomic group level is
892        reached. This is tested by comparing md->once_target with the start of the
893        group. At this point, the return is converted into MATCH_NOMATCH so that
894        previous backup points can be taken. */
895    
896        case OP_ONCE:
897      case OP_BRA:      case OP_BRA:
898      case OP_SBRA:      case OP_SBRA:
899      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
900      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
901      for (;;)      for (;;)
902        {        {
903        if (ecode[GET(ecode, 1)] != OP_ALT)        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
904          {        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
905          ecode += _pcre_OP_lengths[*ecode];          RM2);
906          flags |= match_tail_recursed;        if (rrc != MATCH_NOMATCH &&
907          DPRINTF(("bracket 0 tail recursion\n"));            (rrc != MATCH_THEN || md->start_match_ptr != ecode))
908          goto TAIL_RECURSE;          {
909            if (rrc == MATCH_ONCE)
910              {
911              const uschar *scode = ecode;
912              if (*scode != OP_ONCE)           /* If not at start, find it */
913                {
914                while (*scode == OP_ALT) scode += GET(scode, 1);
915                scode -= GET(scode, 1);
916                }
917              if (md->once_target == scode) rrc = MATCH_NOMATCH;
918              }
919            RRETURN(rrc);
920            }
921          ecode += GET(ecode, 1);
922          if (*ecode != OP_ALT) break;
923          }
924        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
925        RRETURN(MATCH_NOMATCH);
926    
927        /* Handle possessive capturing brackets with an unlimited repeat. We come
928        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
929        handled similarly to the normal case above. However, the matching is
930        different. The end of these brackets will always be OP_KETRPOS, which
931        returns MATCH_KETRPOS without going further in the pattern. By this means
932        we can handle the group by iteration rather than recursion, thereby
933        reducing the amount of stack needed. */
934    
935        case OP_CBRAPOS:
936        case OP_SCBRAPOS:
937        allow_zero = FALSE;
938    
939        POSSESSIVE_CAPTURE:
940        number = GET2(ecode, 1+LINK_SIZE);
941        offset = number << 1;
942    
943    #ifdef PCRE_DEBUG
944        printf("start possessive bracket %d\n", number);
945        printf("subject=");
946        pchars(eptr, 16, TRUE, md);
947        printf("\n");
948    #endif
949    
950        if (offset < md->offset_max)
951          {
952          matched_once = FALSE;
953          code_offset = ecode - md->start_code;
954    
955          save_offset1 = md->offset_vector[offset];
956          save_offset2 = md->offset_vector[offset+1];
957          save_offset3 = md->offset_vector[md->offset_end - number];
958          save_capture_last = md->capture_last;
959    
960          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
961    
962          /* Each time round the loop, save the current subject position for use
963          when the group matches. For MATCH_MATCH, the group has matched, so we
964          restart it with a new subject starting position, remembering that we had
965          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
966          usual. If we haven't matched any alternatives in any iteration, check to
967          see if a previous iteration matched. If so, the group has matched;
968          continue from afterwards. Otherwise it has failed; restore the previous
969          capture values before returning NOMATCH. */
970    
971          for (;;)
972            {
973            md->offset_vector[md->offset_end - number] =
974              (int)(eptr - md->start_subject);
975            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
976            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
977              eptrb, RM63);
978            if (rrc == MATCH_KETRPOS)
979              {
980              offset_top = md->end_offset_top;
981              eptr = md->end_match_ptr;
982              ecode = md->start_code + code_offset;
983              save_capture_last = md->capture_last;
984              matched_once = TRUE;
985              continue;
986              }
987            if (rrc != MATCH_NOMATCH &&
988                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
989              RRETURN(rrc);
990            md->capture_last = save_capture_last;
991            ecode += GET(ecode, 1);
992            if (*ecode != OP_ALT) break;
993          }          }
994    
995        /* For non-final alternatives, continue the loop for a NOMATCH result;        if (!matched_once)
996        otherwise return. */          {
997            md->offset_vector[offset] = save_offset1;
998            md->offset_vector[offset+1] = save_offset2;
999            md->offset_vector[md->offset_end - number] = save_offset3;
1000            }
1001    
1002          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
1003          if (allow_zero || matched_once)
1004            {
1005            ecode += 1 + LINK_SIZE;
1006            break;
1007            }
1008    
1009          RRETURN(MATCH_NOMATCH);
1010          }
1011    
1012        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1013        as a non-capturing bracket. */
1014    
1015        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1016        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1017    
1018        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1019          eptrb, flags, RM2);  
1020        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1021        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1022    
1023        /* Non-capturing possessive bracket with unlimited repeat. We come here
1024        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1025        without the capturing complication. It is written out separately for speed
1026        and cleanliness. */
1027    
1028        case OP_BRAPOS:
1029        case OP_SBRAPOS:
1030        allow_zero = FALSE;
1031    
1032        POSSESSIVE_NON_CAPTURE:
1033        matched_once = FALSE;
1034        code_offset = ecode - md->start_code;
1035    
1036        for (;;)
1037          {
1038          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1039          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1040            eptrb, RM48);
1041          if (rrc == MATCH_KETRPOS)
1042            {
1043            offset_top = md->end_offset_top;
1044            eptr = md->end_match_ptr;
1045            ecode = md->start_code + code_offset;
1046            matched_once = TRUE;
1047            continue;
1048            }
1049          if (rrc != MATCH_NOMATCH &&
1050              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1051            RRETURN(rrc);
1052        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1053          if (*ecode != OP_ALT) break;
1054        }        }
1055    
1056        if (matched_once || allow_zero)
1057          {
1058          ecode += 1 + LINK_SIZE;
1059          break;
1060          }
1061        RRETURN(MATCH_NOMATCH);
1062    
1063      /* Control never reaches here. */      /* Control never reaches here. */
1064    
1065      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1066      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1067      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1068      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1069    
1070      case OP_COND:      case OP_COND:
1071      case OP_SCOND:      case OP_SCOND:
1072      if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */      codelink = GET(ecode, 1);
1073    
1074        /* Because of the way auto-callout works during compile, a callout item is
1075        inserted between OP_COND and an assertion condition. */
1076    
1077        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1078        {        {
1079        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (pcre_callout != NULL)
1080        condition = md->recursive != NULL &&          {
1081          (offset == RREF_ANY || offset == md->recursive->group_num);          pcre_callout_block cb;
1082        ecode += condition? 3 : GET(ecode, 1);          cb.version          = 1;   /* Version 1 of the callout block */
1083            cb.callout_number   = ecode[LINK_SIZE+2];
1084            cb.offset_vector    = md->offset_vector;
1085            cb.subject          = (PCRE_SPTR)md->start_subject;
1086            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1087            cb.start_match      = (int)(mstart - md->start_subject);
1088            cb.current_position = (int)(eptr - md->start_subject);
1089            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1090            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1091            cb.capture_top      = offset_top/2;
1092            cb.capture_last     = md->capture_last;
1093            cb.callout_data     = md->callout_data;
1094            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1095            if (rrc < 0) RRETURN(rrc);
1096            }
1097          ecode += _pcre_OP_lengths[OP_CALLOUT];
1098          }
1099    
1100        condcode = ecode[LINK_SIZE+1];
1101    
1102        /* Now see what the actual condition is */
1103    
1104        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1105          {
1106          if (md->recursive == NULL)                /* Not recursing => FALSE */
1107            {
1108            condition = FALSE;
1109            ecode += GET(ecode, 1);
1110            }
1111          else
1112            {
1113            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1114            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1115    
1116            /* If the test is for recursion into a specific subpattern, and it is
1117            false, but the test was set up by name, scan the table to see if the
1118            name refers to any other numbers, and test them. The condition is true
1119            if any one is set. */
1120    
1121            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1122              {
1123              uschar *slotA = md->name_table;
1124              for (i = 0; i < md->name_count; i++)
1125                {
1126                if (GET2(slotA, 0) == recno) break;
1127                slotA += md->name_entry_size;
1128                }
1129    
1130              /* Found a name for the number - there can be only one; duplicate
1131              names for different numbers are allowed, but not vice versa. First
1132              scan down for duplicates. */
1133    
1134              if (i < md->name_count)
1135                {
1136                uschar *slotB = slotA;
1137                while (slotB > md->name_table)
1138                  {
1139                  slotB -= md->name_entry_size;
1140                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1141                    {
1142                    condition = GET2(slotB, 0) == md->recursive->group_num;
1143                    if (condition) break;
1144                    }
1145                  else break;
1146                  }
1147    
1148                /* Scan up for duplicates */
1149    
1150                if (!condition)
1151                  {
1152                  slotB = slotA;
1153                  for (i++; i < md->name_count; i++)
1154                    {
1155                    slotB += md->name_entry_size;
1156                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1157                      {
1158                      condition = GET2(slotB, 0) == md->recursive->group_num;
1159                      if (condition) break;
1160                      }
1161                    else break;
1162                    }
1163                  }
1164                }
1165              }
1166    
1167            /* Chose branch according to the condition */
1168    
1169            ecode += condition? 3 : GET(ecode, 1);
1170            }
1171        }        }
1172    
1173      else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1174        {        {
1175        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1176        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1177    
1178          /* If the numbered capture is unset, but the reference was by name,
1179          scan the table to see if the name refers to any other numbers, and test
1180          them. The condition is true if any one is set. This is tediously similar
1181          to the code above, but not close enough to try to amalgamate. */
1182    
1183          if (!condition && condcode == OP_NCREF)
1184            {
1185            int refno = offset >> 1;
1186            uschar *slotA = md->name_table;
1187    
1188            for (i = 0; i < md->name_count; i++)
1189              {
1190              if (GET2(slotA, 0) == refno) break;
1191              slotA += md->name_entry_size;
1192              }
1193    
1194            /* Found a name for the number - there can be only one; duplicate names
1195            for different numbers are allowed, but not vice versa. First scan down
1196            for duplicates. */
1197    
1198            if (i < md->name_count)
1199              {
1200              uschar *slotB = slotA;
1201              while (slotB > md->name_table)
1202                {
1203                slotB -= md->name_entry_size;
1204                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1205                  {
1206                  offset = GET2(slotB, 0) << 1;
1207                  condition = offset < offset_top &&
1208                    md->offset_vector[offset] >= 0;
1209                  if (condition) break;
1210                  }
1211                else break;
1212                }
1213    
1214              /* Scan up for duplicates */
1215    
1216              if (!condition)
1217                {
1218                slotB = slotA;
1219                for (i++; i < md->name_count; i++)
1220                  {
1221                  slotB += md->name_entry_size;
1222                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1223                    {
1224                    offset = GET2(slotB, 0) << 1;
1225                    condition = offset < offset_top &&
1226                      md->offset_vector[offset] >= 0;
1227                    if (condition) break;
1228                    }
1229                  else break;
1230                  }
1231                }
1232              }
1233            }
1234    
1235          /* Chose branch according to the condition */
1236    
1237        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 3 : GET(ecode, 1);
1238        }        }
1239    
1240      else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
1241        {        {
1242        condition = FALSE;        condition = FALSE;
1243        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1244        }        }
1245    
1246      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1247      the final argument match_condassert causes it to stop at the end of an      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of
1248      assertion. */      an assertion. */
1249    
1250      else      else
1251        {        {
1252        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1253            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1254        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1255          {          {
1256            if (md->end_offset_top > offset_top)
1257              offset_top = md->end_offset_top;  /* Captures may have happened */
1258          condition = TRUE;          condition = TRUE;
1259          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1260          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1261          }          }
1262        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH &&
1263                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1264          {          {
1265          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1266          }          }
1267        else        else
1268          {          {
1269          condition = FALSE;          condition = FALSE;
1270          ecode += GET(ecode, 1);          ecode += codelink;
1271          }          }
1272        }        }
1273    
1274      /* 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,
1275      we can use tail recursion to avoid using another stack frame. If the second      we used to use tail recursion to avoid using another stack frame, except
1276      alternative doesn't exist, we can just plough on. */      when there was unlimited repeat of a possibly empty group. However, that
1277        strategy no longer works because of the possibilty of (*THEN) being
1278        encountered in the branch. A recursive call to match() is always required,
1279        unless the second alternative doesn't exist, in which case we can just
1280        plough on. */
1281    
1282      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1283        {        {
1284        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1285        flags = match_tail_recursed | ((op == OP_SCOND)? match_cbegroup : 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1286        goto TAIL_RECURSE;        if (rrc == MATCH_THEN && md->start_match_ptr == ecode)
1287            rrc = MATCH_NOMATCH;
1288          RRETURN(rrc);
1289        }        }
1290      else      else                         /* Condition false & no alternative */
1291        {        {
1292        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1293        }        }
1294      break;      break;
1295    
1296    
1297      /* 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,
1298      restore the offsets appropriately and continue from after the call. */      to close any currently open capturing brackets. */
1299    
1300      case OP_END:      case OP_CLOSE:
1301      if (md->recursive != NULL && md->recursive->group_num == 0)      number = GET2(ecode, 1);
1302        offset = number << 1;
1303    
1304    #ifdef PCRE_DEBUG
1305          printf("end bracket %d at *ACCEPT", number);
1306          printf("\n");
1307    #endif
1308    
1309        md->capture_last = number;
1310        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1311        {        {
1312        recursion_info *rec = md->recursive;        md->offset_vector[offset] =
1313        DPRINTF(("End of pattern in a (?0) recursion\n"));          md->offset_vector[md->offset_end - number];
1314        md->recursive = rec->prevrec;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1315        memmove(md->offset_vector, rec->offset_save,        if (offset_top <= offset) offset_top = offset + 2;
         rec->saved_max * sizeof(int));  
       md->start_match = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
1316        }        }
1317        ecode += 3;
1318        break;
1319    
     /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty  
     string - backtracking will then try other alternatives, if any. */  
1320    
1321      if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH);      /* End of the pattern, either real or forced. */
1322      md->end_match_ptr = eptr;          /* Record where we ended */  
1323      md->end_offset_top = offset_top;   /* and how many extracts were taken */      case OP_END:
1324      RRETURN(MATCH_MATCH);      case OP_ACCEPT:
1325        case OP_ASSERT_ACCEPT:
1326      /* Change option settings */  
1327        /* If we have matched an empty string, fail if not in an assertion and not
1328      case OP_OPT:      in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1329      ims = ecode[1];      is set and we have matched at the start of the subject. In both cases,
1330      ecode += 2;      backtracking will then try other alternatives, if any. */
1331      DPRINTF(("ims set to %02lx\n", ims));  
1332      break;      if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1333             md->recursive == NULL &&
1334             (md->notempty ||
1335               (md->notempty_atstart &&
1336                 mstart == md->start_subject + md->start_offset)))
1337          MRRETURN(MATCH_NOMATCH);
1338    
1339        /* Otherwise, we have a match. */
1340    
1341        md->end_match_ptr = eptr;           /* Record where we ended */
1342        md->end_offset_top = offset_top;    /* and how many extracts were taken */
1343        md->start_match_ptr = mstart;       /* and the start (\K can modify) */
1344    
1345        /* For some reason, the macros don't work properly if an expression is
1346        given as the argument to MRRETURN when the heap is in use. */
1347    
1348        rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1349        MRRETURN(rrc);
1350    
1351      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1352      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,
1353      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
1354      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
1355      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1356        of a condition, we want to return immediately afterwards. The caller of
1357        this incarnation of the match() function will have set MATCH_CONDASSERT in
1358        md->match_function type, and one of these opcodes will be the first opcode
1359        that is processed. We use a local variable that is preserved over calls to
1360        match() to remember this case. */
1361    
1362      case OP_ASSERT:      case OP_ASSERT:
1363      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1364        if (md->match_function_type == MATCH_CONDASSERT)
1365          {
1366          condassert = TRUE;
1367          md->match_function_type = 0;
1368          }
1369        else condassert = FALSE;
1370    
1371      do      do
1372        {        {
1373        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1374          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1375        if (rrc == MATCH_MATCH) break;          {
1376        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          mstart = md->start_match_ptr;   /* In case \K reset it */
1377            break;
1378            }
1379          if (rrc != MATCH_NOMATCH &&
1380              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1381            RRETURN(rrc);
1382        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1383        }        }
1384      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1385      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1386        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1387    
1388      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1389    
1390      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1391    
1392      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1393      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 840  for (;;) Line 1397  for (;;)
1397      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1398      continue;      continue;
1399    
1400      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1401        PRUNE, or COMMIT means we must assume failure without checking subsequent
1402        branches. */
1403    
1404      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1405      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1406        if (md->match_function_type == MATCH_CONDASSERT)
1407          {
1408          condassert = TRUE;
1409          md->match_function_type = 0;
1410          }
1411        else condassert = FALSE;
1412    
1413      do      do
1414        {        {
1415        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1416          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1417        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1418        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          {
1419            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1420            break;
1421            }
1422          if (rrc != MATCH_NOMATCH &&
1423              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1424            RRETURN(rrc);
1425        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1426        }        }
1427      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1428    
1429      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1430    
1431      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1432      continue;      continue;
1433    
# Line 872  for (;;) Line 1444  for (;;)
1444        while (i-- > 0)        while (i-- > 0)
1445          {          {
1446          eptr--;          eptr--;
1447          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1448          BACKCHAR(eptr)          BACKCHAR(eptr);
1449          }          }
1450        }        }
1451      else      else
# Line 883  for (;;) Line 1455  for (;;)
1455    
1456        {        {
1457        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1458        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1459        }        }
1460    
1461      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1462    
1463        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1464      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1465      break;      break;
1466    
# Line 903  for (;;) Line 1476  for (;;)
1476        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1477        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1478        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1479        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1480        cb.start_match      = md->start_match - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1481        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1482        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1483        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1484        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1485        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1486        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1487        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1488        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1489        }        }
1490      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 920  for (;;) Line 1493  for (;;)
1493      /* Recursion either matches the current regex, or some subexpression. The      /* Recursion either matches the current regex, or some subexpression. The
1494      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
1495      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1496    
1497      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1498      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
1499      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
1500      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
1501      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
1502      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
1503      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.  
1504    
1505      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
1506      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
1507      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1508        a lot, so he is not to blame for the current way it works. */
1509    
1510      case OP_RECURSE:      case OP_RECURSE:
1511        {        {
# Line 947  for (;;) Line 1518  for (;;)
1518        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1519        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1520    
1521        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1522    
1523        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1524    
1525        /* Now save the offset data. */        /* Now save the offset data */
1526    
1527        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1528        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 963  for (;;) Line 1533  for (;;)
1533            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1534          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1535          }          }
   
1536        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1537              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1538        new_recursive.save_start = md->start_match;  
1539        md->start_match = eptr;        /* OK, now we can do the recursion. After processing each alternative,
1540          restore the offset data. If there were nested recursions, md->recursive
1541        /* OK, now we can do the recursion. For each top-level alternative we        might be changed, so reset it before looping. */
       restore the offset and recursion data. */  
1542    
1543        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1544        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1545        do        do
1546          {          {
1547            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1548          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1549            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1550          if (rrc == MATCH_MATCH)          memcpy(md->offset_vector, new_recursive.offset_save,
1551                new_recursive.saved_max * sizeof(int));
1552            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1553            {            {
1554            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
1555            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1556            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1557              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1558            RRETURN(MATCH_MATCH);  
1559              /* Set where we got to in the subject, and reset the start in case
1560              it was changed by \K. This *is* propagated back out of a recursion,
1561              for Perl compatibility. */
1562    
1563              eptr = md->end_match_ptr;
1564              mstart = md->start_match_ptr;
1565              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1566            }            }
1567          else if (rrc != MATCH_NOMATCH)          else if (rrc != MATCH_NOMATCH &&
1568                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1569            {            {
1570            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1571              if (new_recursive.offset_save != stacksave)
1572                (pcre_free)(new_recursive.offset_save);
1573            RRETURN(rrc);            RRETURN(rrc);
1574            }            }
1575    
1576          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1577          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1578          }          }
1579        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1003  for (;;) Line 1582  for (;;)
1582        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1583        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1584          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1585        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
       }  
     /* Control never reaches here */  
   
     /* "Once" brackets are like assertion brackets except that after a match,  
     the point in the subject string is not moved back. Thus there can never be  
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer. */  
   
     case OP_ONCE:  
     prev = ecode;  
     saved_eptr = eptr;  
   
     do  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,  
         eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH) break;  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += GET(ecode,1);  
1586        }        }
1587      while (*ecode == OP_ALT);  
1588        RECURSION_MATCHED:
1589        break;
1590    
1591      /* If hit the end of the group (which could be repeated), fail */      /* An alternation is the end of a branch; scan along to find the end of the
1592        bracketed group and go to there. */
1593    
1594      if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);      case OP_ALT:
1595        do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1596        break;
1597    
1598      /* Continue as from after the assertion, updating the offsets high water      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1599      mark, since extracts may have been taken. */      indicating that it may occur zero times. It may repeat infinitely, or not
1600        at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1601        with fixed upper repeat limits are compiled as a number of copies, with the
1602        optional ones preceded by BRAZERO or BRAMINZERO. */
1603    
1604        case OP_BRAZERO:
1605        next = ecode + 1;
1606        RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1607        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1608        do next += GET(next, 1); while (*next == OP_ALT);
1609        ecode = next + 1 + LINK_SIZE;
1610        break;
1611    
1612        case OP_BRAMINZERO:
1613        next = ecode + 1;
1614        do next += GET(next, 1); while (*next == OP_ALT);
1615        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1616        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1617        ecode++;
1618        break;
1619    
1620      do ecode += GET(ecode, 1); while (*ecode == OP_ALT);      case OP_SKIPZERO:
1621        next = ecode+1;
1622        do next += GET(next,1); while (*next == OP_ALT);
1623        ecode = next + 1 + LINK_SIZE;
1624        break;
1625    
1626        /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1627        here; just jump to the group, with allow_zero set TRUE. */
1628    
1629        case OP_BRAPOSZERO:
1630        op = *(++ecode);
1631        allow_zero = TRUE;
1632        if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1633          goto POSSESSIVE_NON_CAPTURE;
1634    
1635      offset_top = md->end_offset_top;      /* End of a group, repeated or non-repeating. */
     eptr = md->end_match_ptr;  
1636    
1637      /* For a non-repeating ket, just continue at this level. This also      case OP_KET:
1638      happens for a repeating ket if no characters were matched in the group.      case OP_KETRMIN:
1639      This is the forcible breaking of infinite loops as implemented in Perl      case OP_KETRMAX:
1640      5.005. If there is an options reset, it will get obeyed in the normal      case OP_KETRPOS:
1641      course of events. */      prev = ecode - GET(ecode, 1);
1642    
1643        /* If this was a group that remembered the subject start, in order to break
1644        infinite repeats of empty string matches, retrieve the subject start from
1645        the chain. Otherwise, set it NULL. */
1646    
1647      if (*ecode == OP_KET || eptr == saved_eptr)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1648        {        {
1649        ecode += 1+LINK_SIZE;        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1650        break;        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1651        }        }
1652        else saved_eptr = NULL;
1653    
1654      /* The repeating kets try the rest of the pattern or restart from the      /* If we are at the end of an assertion group, stop matching and return
1655      preceding bracket, in the appropriate order. The second "call" of match()      MATCH_MATCH, but record the current high water mark for use by positive
1656      uses tail recursion, to avoid using another stack frame. We need to reset      assertions. We also need to record the match start in case it was changed
1657      any options that changed within the bracket before re-running it, so      by \K. */
     check the next opcode. */  
1658    
1659      if (ecode[1+LINK_SIZE] == OP_OPT)      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1660            *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
1661        {        {
1662        ims = (ims & ~PCRE_IMS) | ecode[4];        md->end_match_ptr = eptr;      /* For ONCE */
1663        DPRINTF(("ims set to %02lx at group repeat\n", ims));        md->end_offset_top = offset_top;
1664          md->start_match_ptr = mstart;
1665          MRRETURN(MATCH_MATCH);
1666        }        }
1667    
1668      if (*ecode == OP_KETRMIN)      /* For capturing groups we have to check the group number back at the start
1669        {      and if necessary complete handling an extraction by setting the offsets and
1670        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0,      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1671          RM8);      into group 0, so it won't be picked up here. Instead, we catch it when the
1672        if (rrc != MATCH_NOMATCH) RRETURN(rrc);      OP_END is reached. Other recursion is handled here. We just have to record
1673        ecode = prev;      the current subject position and start match pointer and give a MATCH
1674        flags = match_tail_recursed;      return. */
1675        goto TAIL_RECURSE;  
1676        }      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1677      else  /* OP_KETRMAX */          *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = match_tail_recursed;  
       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. */  
   
     case OP_KET:  
     case OP_KETRMIN:  
     case OP_KETRMAX:  
     prev = ecode - GET(ecode, 1);  
   
     /* If this was a group that remembered the subject start, in order to break  
     infinite repeats of empty string matches, retrieve the subject start from  
     the chain. Otherwise, set it NULL. */  
   
     if (*prev >= OP_SBRA)  
       {  
       saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */  
       eptrb = eptrb->epb_prev;              /* Backup to previous group */  
       }  
     else saved_eptr = NULL;  
   
     /* If we are at the end of an assertion group, stop matching and return  
     MATCH_MATCH, but record the current high water mark for use by positive  
     assertions. Do this also for the "once" (atomic) groups. */  
   
     if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||  
         *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||  
         *prev == OP_ONCE)  
       {  
       md->end_match_ptr = eptr;      /* For ONCE */  
       md->end_offset_top = offset_top;  
       RRETURN(MATCH_MATCH);  
       }  
   
     /* For capturing groups we have to check the group number back at the start  
     and if necessary complete handling an extraction by setting the offsets and  
     bumping the high water mark. Note that whole-pattern recursion is coded as  
     a recurse into group 0, so it won't be picked up here. Instead, we catch it  
     when the OP_END is reached. Other recursion is handled here. */  
   
     if (*prev == OP_CBRA || *prev == OP_SCBRA)  
1678        {        {
1679        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1680        offset = number << 1;        offset = number << 1;
1681    
1682  #ifdef DEBUG  #ifdef PCRE_DEBUG
1683        printf("end bracket %d", number);        printf("end bracket %d", number);
1684        printf("\n");        printf("\n");
1685  #endif  #endif
1686    
1687          /* Handle a recursively called group. */
1688    
1689          if (md->recursive != NULL && md->recursive->group_num == number)
1690            {
1691            md->end_match_ptr = eptr;
1692            md->start_match_ptr = mstart;
1693            RRETURN(MATCH_MATCH);
1694            }
1695    
1696          /* Deal with capturing */
1697    
1698        md->capture_last = number;        md->capture_last = number;
1699        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1700          {          {
1701            /* If offset is greater than offset_top, it means that we are
1702            "skipping" a capturing group, and that group's offsets must be marked
1703            unset. In earlier versions of PCRE, all the offsets were unset at the
1704            start of matching, but this doesn't work because atomic groups and
1705            assertions can cause a value to be set that should later be unset.
1706            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1707            part of the atomic group, but this is not on the final matching path,
1708            so must be unset when 2 is set. (If there is no group 2, there is no
1709            problem, because offset_top will then be 2, indicating no capture.) */
1710    
1711            if (offset > offset_top)
1712              {
1713              register int *iptr = md->offset_vector + offset_top;
1714              register int *iend = md->offset_vector + offset;
1715              while (iptr < iend) *iptr++ = -1;
1716              }
1717    
1718            /* Now make the extraction */
1719    
1720          md->offset_vector[offset] =          md->offset_vector[offset] =
1721            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1722          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1723          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1724          }          }
   
       /* Handle a recursively called group. Restore the offsets  
       appropriately and continue from after the call. */  
   
       if (md->recursive != NULL && md->recursive->group_num == number)  
         {  
         recursion_info *rec = md->recursive;  
         DPRINTF(("Recursion (%d) succeeded - continuing\n", number));  
         md->recursive = rec->prevrec;  
         md->start_match = rec->save_start;  
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
         }  
1725        }        }
1726    
1727      /* For both capturing and non-capturing groups, reset the value of the ims      /* For an ordinary non-repeating ket, just continue at this level. This
1728      flags, in case they got changed during the group. */      also happens for a repeating ket if no characters were matched in the
1729        group. This is the forcible breaking of infinite loops as implemented in
1730      ims = original_ims;      Perl 5.005. For a non-repeating atomic group, establish a backup point by
1731      DPRINTF(("ims reset to %02lx\n", ims));      processing the rest of the pattern at a lower level. If this results in a
1732        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1733      /* For a non-repeating ket, just continue at this level. This also      bypassing intermediate backup points, but resetting any captures that
1734      happens for a repeating ket if no characters were matched in the group.      happened along the way. */
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
1735    
1736      if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1737        {        {
1738        ecode += 1 + LINK_SIZE;        if (*prev == OP_ONCE)
1739            {
1740            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1741            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1742            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1743            RRETURN(MATCH_ONCE);
1744            }
1745          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1746        break;        break;
1747        }        }
1748    
1749      /* The repeating kets try the rest of the pattern or restart from the      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1750      preceding bracket, in the appropriate order. In the second case, we can use      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1751      tail recursion to avoid using another stack frame. */      at a time from the outer level, thus saving stack. */
1752    
1753      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;      if (*ecode == OP_KETRPOS)
1754          {
1755          md->end_match_ptr = eptr;
1756          md->end_offset_top = offset_top;
1757          RRETURN(MATCH_KETRPOS);
1758          }
1759    
1760        /* The normal repeating kets try the rest of the pattern or restart from
1761        the preceding bracket, in the appropriate order. In the second case, we can
1762        use tail recursion to avoid using another stack frame, unless we have an
1763        an atomic group or an unlimited repeat of a group that can match an empty
1764        string. */
1765    
1766      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1767        {        {
1768        RMATCH(eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64);
         RM12);  
1769        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1770          if (*prev == OP_ONCE)
1771            {
1772            RMATCH(eptr, prev, offset_top, md, eptrb, RM66);
1773            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1774            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1775            RRETURN(MATCH_ONCE);
1776            }
1777          if (*prev >= OP_SBRA)    /* Could match an empty string */
1778            {
1779            md->match_function_type = MATCH_CBEGROUP;
1780            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1781            RRETURN(rrc);
1782            }
1783        ecode = prev;        ecode = prev;
       flags |= match_tail_recursed;  
1784        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1785        }        }
1786      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1787        {        {
1788        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1789          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1790          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1791        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1792          if (*prev == OP_ONCE)
1793            {
1794            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65);
1795            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1796            md->once_target = prev;
1797            RRETURN(MATCH_ONCE);
1798            }
1799        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = match_tail_recursed;  
1800        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1801        }        }
1802      /* Control never gets here */      /* Control never gets here */
1803    
1804      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1805    
1806      case OP_CIRC:      case OP_CIRC:
1807      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1808      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 */  
   
1809      /* Start of subject assertion */      /* Start of subject assertion */
1810    
1811      case OP_SOD:      case OP_SOD:
1812      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1813        ecode++;
1814        break;
1815    
1816        /* Multiline mode: start of subject unless notbol, or after any newline. */
1817    
1818        case OP_CIRCM:
1819        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1820        if (eptr != md->start_subject &&
1821            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1822          MRRETURN(MATCH_NOMATCH);
1823      ecode++;      ecode++;
1824      break;      break;
1825    
1826      /* Start of match assertion */      /* Start of match assertion */
1827    
1828      case OP_SOM:      case OP_SOM:
1829      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1830      ecode++;      ecode++;
1831      break;      break;
1832    
1833      /* Assert before internal newline if multiline, or before a terminating      /* Reset the start of match point */
     newline unless endonly is set, else end of subject unless noteol is set. */  
1834    
1835      case OP_DOLL:      case OP_SET_SOM:
1836      if ((ims & PCRE_MULTILINE) != 0)      mstart = eptr;
1837        {      ecode++;
1838        if (eptr < md->end_subject)      break;
1839          { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }  
1840        else      /* Multiline mode: assert before any newline, or before end of subject
1841          { if (md->noteol) RRETURN(MATCH_NOMATCH); }      unless noteol is set. */
1842        ecode++;  
1843        break;      case OP_DOLLM:
1844        }      if (eptr < md->end_subject)
1845          { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1846      else      else
1847        {        {
1848        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1849        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1850        }        }
1851        ecode++;
1852        break;
1853    
1854        /* Not multiline mode: assert before a terminating newline or before end of
1855        subject unless noteol is set. */
1856    
1857        case OP_DOLL:
1858        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1859        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1860    
1861      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1862    
1863      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1864    
1865      case OP_EOD:      case OP_EOD:
1866      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1867        SCHECK_PARTIAL();
1868      ecode++;      ecode++;
1869      break;      break;
1870    
1871      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1872    
1873      case OP_EODN:      case OP_EODN:
1874      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1875        if (eptr < md->end_subject &&
1876          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1877        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1878    
1879        /* Either at end of string or \n before end. */
1880    
1881        SCHECK_PARTIAL();
1882      ecode++;      ecode++;
1883      break;      break;
1884    
# Line 1311  for (;;) Line 1890  for (;;)
1890    
1891        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1892        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
1893        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1894          partial matching. */
1895    
1896  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1897        if (utf8)        if (utf8)
1898          {          {
1899            /* Get status of previous character */
1900    
1901          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1902            {            {
1903            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1904            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1905              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1906            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1907    #ifdef SUPPORT_UCP
1908              if (md->use_ucp)
1909                {
1910                if (c == '_') prev_is_word = TRUE; else
1911                  {
1912                  int cat = UCD_CATEGORY(c);
1913                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1914                  }
1915                }
1916              else
1917    #endif
1918            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1919            }            }
1920          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
1921            /* Get status of next character */
1922    
1923            if (eptr >= md->end_subject)
1924              {
1925              SCHECK_PARTIAL();
1926              cur_is_word = FALSE;
1927              }
1928            else
1929            {            {
1930            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1931    #ifdef SUPPORT_UCP
1932              if (md->use_ucp)
1933                {
1934                if (c == '_') cur_is_word = TRUE; else
1935                  {
1936                  int cat = UCD_CATEGORY(c);
1937                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1938                  }
1939                }
1940              else
1941    #endif
1942            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1943            }            }
1944          }          }
1945        else        else
1946  #endif  #endif
1947    
1948        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1949          consistency with the behaviour of \w we do use it in this case. */
1950    
1951          {          {
1952          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
1953            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
1954          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
1955            ((md->ctypes[*eptr] & ctype_word) != 0);            {
1956              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1957    #ifdef SUPPORT_UCP
1958              if (md->use_ucp)
1959                {
1960                c = eptr[-1];
1961                if (c == '_') prev_is_word = TRUE; else
1962                  {
1963                  int cat = UCD_CATEGORY(c);
1964                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1965                  }
1966                }
1967              else
1968    #endif
1969              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1970              }
1971    
1972            /* Get status of next character */
1973    
1974            if (eptr >= md->end_subject)
1975              {
1976              SCHECK_PARTIAL();
1977              cur_is_word = FALSE;
1978              }
1979            else
1980    #ifdef SUPPORT_UCP
1981            if (md->use_ucp)
1982              {
1983              c = *eptr;
1984              if (c == '_') cur_is_word = TRUE; else
1985                {
1986                int cat = UCD_CATEGORY(c);
1987                cur_is_word = (cat == ucp_L || cat == ucp_N);
1988                }
1989              }
1990            else
1991    #endif
1992            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
1993          }          }
1994    
1995        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
1996    
1997        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
1998             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
1999          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2000        }        }
2001      break;      break;
2002    
2003      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2004    
2005      case OP_ANY:      case OP_ANY:
2006      if ((ims & PCRE_DOTALL) == 0)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2007        /* Fall through */
2008    
2009        case OP_ALLANY:
2010        if (eptr++ >= md->end_subject)
2011        {        {
2012        if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);        SCHECK_PARTIAL();
2013          MRRETURN(MATCH_NOMATCH);
2014        }        }
2015      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++;  
2016      ecode++;      ecode++;
2017      break;      break;
2018    
# Line 1366  for (;;) Line 2020  for (;;)
2020      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2021    
2022      case OP_ANYBYTE:      case OP_ANYBYTE:
2023      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr++ >= md->end_subject)
2024          {
2025          SCHECK_PARTIAL();
2026          MRRETURN(MATCH_NOMATCH);
2027          }
2028      ecode++;      ecode++;
2029      break;      break;
2030    
2031      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2032      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2033          {
2034          SCHECK_PARTIAL();
2035          MRRETURN(MATCH_NOMATCH);
2036          }
2037      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2038      if (      if (
2039  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1379  for (;;) Line 2041  for (;;)
2041  #endif  #endif
2042         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2043         )         )
2044        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2045      ecode++;      ecode++;
2046      break;      break;
2047    
2048      case OP_DIGIT:      case OP_DIGIT:
2049      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2050          {
2051          SCHECK_PARTIAL();
2052          MRRETURN(MATCH_NOMATCH);
2053          }
2054      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2055      if (      if (
2056  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1392  for (;;) Line 2058  for (;;)
2058  #endif  #endif
2059         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2060         )         )
2061        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2062      ecode++;      ecode++;
2063      break;      break;
2064    
2065      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2066      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2067          {
2068          SCHECK_PARTIAL();
2069          MRRETURN(MATCH_NOMATCH);
2070          }
2071      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2072      if (      if (
2073  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1405  for (;;) Line 2075  for (;;)
2075  #endif  #endif
2076         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2077         )         )
2078        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2079      ecode++;      ecode++;
2080      break;      break;
2081    
2082      case OP_WHITESPACE:      case OP_WHITESPACE:
2083      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2084          {
2085          SCHECK_PARTIAL();
2086          MRRETURN(MATCH_NOMATCH);
2087          }
2088      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2089      if (      if (
2090  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1418  for (;;) Line 2092  for (;;)
2092  #endif  #endif
2093         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2094         )         )
2095        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2096      ecode++;      ecode++;
2097      break;      break;
2098    
2099      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2100      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2101          {
2102          SCHECK_PARTIAL();
2103          MRRETURN(MATCH_NOMATCH);
2104          }
2105      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2106      if (      if (
2107  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1431  for (;;) Line 2109  for (;;)
2109  #endif  #endif
2110         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2111         )         )
2112        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2113      ecode++;      ecode++;
2114      break;      break;
2115    
2116      case OP_WORDCHAR:      case OP_WORDCHAR:
2117      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2118          {
2119          SCHECK_PARTIAL();
2120          MRRETURN(MATCH_NOMATCH);
2121          }
2122      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2123      if (      if (
2124  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1444  for (;;) Line 2126  for (;;)
2126  #endif  #endif
2127         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2128         )         )
2129        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2130      ecode++;      ecode++;
2131      break;      break;
2132    
2133      case OP_ANYNL:      case OP_ANYNL:
2134      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2135          {
2136          SCHECK_PARTIAL();
2137          MRRETURN(MATCH_NOMATCH);
2138          }
2139      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2140      switch(c)      switch(c)
2141        {        {
2142        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2143    
2144        case 0x000d:        case 0x000d:
2145        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2146        break;        break;
2147    
2148        case 0x000a:        case 0x000a:
2149          break;
2150    
2151        case 0x000b:        case 0x000b:
2152        case 0x000c:        case 0x000c:
2153        case 0x0085:        case 0x0085:
2154        case 0x2028:        case 0x2028:
2155        case 0x2029:        case 0x2029:
2156          if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2157          break;
2158          }
2159        ecode++;
2160        break;
2161    
2162        case OP_NOT_HSPACE:
2163        if (eptr >= md->end_subject)
2164          {
2165          SCHECK_PARTIAL();
2166          MRRETURN(MATCH_NOMATCH);
2167          }
2168        GETCHARINCTEST(c, eptr);
2169        switch(c)
2170          {
2171          default: break;
2172          case 0x09:      /* HT */
2173          case 0x20:      /* SPACE */
2174          case 0xa0:      /* NBSP */
2175          case 0x1680:    /* OGHAM SPACE MARK */
2176          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2177          case 0x2000:    /* EN QUAD */
2178          case 0x2001:    /* EM QUAD */
2179          case 0x2002:    /* EN SPACE */
2180          case 0x2003:    /* EM SPACE */
2181          case 0x2004:    /* THREE-PER-EM SPACE */
2182          case 0x2005:    /* FOUR-PER-EM SPACE */
2183          case 0x2006:    /* SIX-PER-EM SPACE */
2184          case 0x2007:    /* FIGURE SPACE */
2185          case 0x2008:    /* PUNCTUATION SPACE */
2186          case 0x2009:    /* THIN SPACE */
2187          case 0x200A:    /* HAIR SPACE */
2188          case 0x202f:    /* NARROW NO-BREAK SPACE */
2189          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2190          case 0x3000:    /* IDEOGRAPHIC SPACE */
2191          MRRETURN(MATCH_NOMATCH);
2192          }
2193        ecode++;
2194        break;
2195    
2196        case OP_HSPACE:
2197        if (eptr >= md->end_subject)
2198          {
2199          SCHECK_PARTIAL();
2200          MRRETURN(MATCH_NOMATCH);
2201          }
2202        GETCHARINCTEST(c, eptr);
2203        switch(c)
2204          {
2205          default: MRRETURN(MATCH_NOMATCH);
2206          case 0x09:      /* HT */
2207          case 0x20:      /* SPACE */
2208          case 0xa0:      /* NBSP */
2209          case 0x1680:    /* OGHAM SPACE MARK */
2210          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2211          case 0x2000:    /* EN QUAD */
2212          case 0x2001:    /* EM QUAD */
2213          case 0x2002:    /* EN SPACE */
2214          case 0x2003:    /* EM SPACE */
2215          case 0x2004:    /* THREE-PER-EM SPACE */
2216          case 0x2005:    /* FOUR-PER-EM SPACE */
2217          case 0x2006:    /* SIX-PER-EM SPACE */
2218          case 0x2007:    /* FIGURE SPACE */
2219          case 0x2008:    /* PUNCTUATION SPACE */
2220          case 0x2009:    /* THIN SPACE */
2221          case 0x200A:    /* HAIR SPACE */
2222          case 0x202f:    /* NARROW NO-BREAK SPACE */
2223          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2224          case 0x3000:    /* IDEOGRAPHIC SPACE */
2225          break;
2226          }
2227        ecode++;
2228        break;
2229    
2230        case OP_NOT_VSPACE:
2231        if (eptr >= md->end_subject)
2232          {
2233          SCHECK_PARTIAL();
2234          MRRETURN(MATCH_NOMATCH);
2235          }
2236        GETCHARINCTEST(c, eptr);
2237        switch(c)
2238          {
2239          default: break;
2240          case 0x0a:      /* LF */
2241          case 0x0b:      /* VT */
2242          case 0x0c:      /* FF */
2243          case 0x0d:      /* CR */
2244          case 0x85:      /* NEL */
2245          case 0x2028:    /* LINE SEPARATOR */
2246          case 0x2029:    /* PARAGRAPH SEPARATOR */
2247          MRRETURN(MATCH_NOMATCH);
2248          }
2249        ecode++;
2250        break;
2251    
2252        case OP_VSPACE:
2253        if (eptr >= md->end_subject)
2254          {
2255          SCHECK_PARTIAL();
2256          MRRETURN(MATCH_NOMATCH);
2257          }
2258        GETCHARINCTEST(c, eptr);
2259        switch(c)
2260          {
2261          default: MRRETURN(MATCH_NOMATCH);
2262          case 0x0a:      /* LF */
2263          case 0x0b:      /* VT */
2264          case 0x0c:      /* FF */
2265          case 0x0d:      /* CR */
2266          case 0x85:      /* NEL */
2267          case 0x2028:    /* LINE SEPARATOR */
2268          case 0x2029:    /* PARAGRAPH SEPARATOR */
2269        break;        break;
2270        }        }
2271      ecode++;      ecode++;
# Line 1474  for (;;) Line 2277  for (;;)
2277    
2278      case OP_PROP:      case OP_PROP:
2279      case OP_NOTPROP:      case OP_NOTPROP:
2280      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2281          {
2282          SCHECK_PARTIAL();
2283          MRRETURN(MATCH_NOMATCH);
2284          }
2285      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2286        {        {
2287        int chartype, script;        const ucd_record *prop = GET_UCD(c);
       int category = _pcre_ucp_findprop(c, &chartype, &script);  
2288    
2289        switch(ecode[1])        switch(ecode[1])
2290          {          {
2291          case PT_ANY:          case PT_ANY:
2292          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2293          break;          break;
2294    
2295          case PT_LAMP:          case PT_LAMP:
2296          if ((chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2297               chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2298               chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2299            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2300           break;          break;
2301    
2302          case PT_GC:          case PT_GC:
2303          if ((ecode[2] != category) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2304            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2305          break;          break;
2306    
2307          case PT_PC:          case PT_PC:
2308          if ((ecode[2] != chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2309            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2310          break;          break;
2311    
2312          case PT_SC:          case PT_SC:
2313          if ((ecode[2] != script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2314            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2315            break;
2316    
2317            /* These are specials */
2318    
2319            case PT_ALNUM:
2320            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2321                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2322              MRRETURN(MATCH_NOMATCH);
2323            break;
2324    
2325            case PT_SPACE:    /* Perl space */
2326            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2327                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2328                   == (op == OP_NOTPROP))
2329              MRRETURN(MATCH_NOMATCH);
2330            break;
2331    
2332            case PT_PXSPACE:  /* POSIX space */
2333            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2334                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2335                 c == CHAR_FF || c == CHAR_CR)
2336                   == (op == OP_NOTPROP))
2337              MRRETURN(MATCH_NOMATCH);
2338            break;
2339    
2340            case PT_WORD:
2341            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2342                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2343                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2344              MRRETURN(MATCH_NOMATCH);
2345          break;          break;
2346    
2347            /* This should never occur */
2348    
2349          default:          default:
2350          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2351          }          }
# Line 1520  for (;;) Line 2358  for (;;)
2358      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2359    
2360      case OP_EXTUNI:      case OP_EXTUNI:
2361      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2362          {
2363          SCHECK_PARTIAL();
2364          MRRETURN(MATCH_NOMATCH);
2365          }
2366      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2367        {        {
2368        int chartype, script;        int category = UCD_CATEGORY(c);
2369        int category = _pcre_ucp_findprop(c, &chartype, &script);        if (category == ucp_M) MRRETURN(MATCH_NOMATCH);
       if (category == ucp_M) RRETURN(MATCH_NOMATCH);  
2370        while (eptr < md->end_subject)        while (eptr < md->end_subject)
2371          {          {
2372          int len = 1;          int len = 1;
# Line 1533  for (;;) Line 2374  for (;;)
2374            {            {
2375            GETCHARLEN(c, eptr, len);            GETCHARLEN(c, eptr, len);
2376            }            }
2377          category = _pcre_ucp_findprop(c, &chartype, &script);          category = UCD_CATEGORY(c);
2378          if (category != ucp_M) break;          if (category != ucp_M) break;
2379          eptr += len;          eptr += len;
2380          }          }
# Line 1552  for (;;) Line 2393  for (;;)
2393      loops). */      loops). */
2394    
2395      case OP_REF:      case OP_REF:
2396        {      case OP_REFI:
2397        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2398        ecode += 3;                                 /* Advance past item */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2399        ecode += 3;
2400    
2401        /* 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];  
2402    
2403        /* Set up for repetition, or handle the non-repeated case */      (a) In the default, Perl-compatible state, set the length negative;
2404        this ensures that every attempt at a match fails. We can't just fail
2405        here, because of the possibility of quantifiers with zero minima.
2406    
2407        switch (*ecode)      (b) If the JavaScript compatibility flag is set, set the length to zero
2408          {      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;  
2409    
2410          case OP_CRRANGE:      Otherwise, set the length to the length of what was matched by the
2411          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;  
2412    
2413          default:               /* No repeat follows */      if (offset >= offset_top || md->offset_vector[offset] < 0)
2414          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);        length = (md->jscript_compat)? 0 : -1;
2415          eptr += length;      else
2416          continue;              /* With the main loop */        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2417    
2418        /* Set up for repetition, or handle the non-repeated case */
2419    
2420        switch (*ecode)
2421          {
2422          case OP_CRSTAR:
2423          case OP_CRMINSTAR:
2424          case OP_CRPLUS:
2425          case OP_CRMINPLUS:
2426          case OP_CRQUERY:
2427          case OP_CRMINQUERY:
2428          c = *ecode++ - OP_CRSTAR;
2429          minimize = (c & 1) != 0;
2430          min = rep_min[c];                 /* Pick up values from tables; */
2431          max = rep_max[c];                 /* zero for max => infinity */
2432          if (max == 0) max = INT_MAX;
2433          break;
2434    
2435          case OP_CRRANGE:
2436          case OP_CRMINRANGE:
2437          minimize = (*ecode == OP_CRMINRANGE);
2438          min = GET2(ecode, 1);
2439          max = GET2(ecode, 3);
2440          if (max == 0) max = INT_MAX;
2441          ecode += 5;
2442          break;
2443    
2444          default:               /* No repeat follows */
2445          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2446            {
2447            CHECK_PARTIAL();
2448            MRRETURN(MATCH_NOMATCH);
2449          }          }
2450          eptr += length;
2451          continue;              /* With the main loop */
2452          }
2453    
2454        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2455        main loop. */      zero, just continue with the main loop. */
2456    
2457        if (length == 0) continue;      if (length == 0) continue;
2458    
2459        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2460        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2461        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2462    
2463        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2464          {
2465          int slength;
2466          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2467          {          {
2468          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2469          eptr += length;          MRRETURN(MATCH_NOMATCH);
2470          }          }
2471          eptr += slength;
2472          }
2473    
2474        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2475        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2476    
2477        if (min == max) continue;      if (min == max) continue;
2478    
2479        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2480    
2481        if (minimize)      if (minimize)
2482          {
2483          for (fi = min;; fi++)
2484          {          {
2485          for (fi = min;; fi++)          int slength;
2486            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2487            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2488            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2489            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2490            {            {
2491            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2492            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2493            }            }
2494          /* Control never gets here */          eptr += slength;
2495          }          }
2496          /* Control never gets here */
2497          }
2498    
2499        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2500    
2501        else      else
2502          {
2503          pp = eptr;
2504          for (i = min; i < max; i++)
2505          {          {
2506          pp = eptr;          int slength;
2507          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2508            {            {
2509            if (!match_ref(offset, eptr, length, md, ims)) break;            CHECK_PARTIAL();
2510            eptr += length;            break;
           }  
         while (eptr >= pp)  
           {  
           RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);  
           if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
           eptr -= length;  
2511            }            }
2512          RRETURN(MATCH_NOMATCH);          eptr += slength;
2513          }          }
2514          while (eptr >= pp)
2515            {
2516            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2517            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2518            eptr -= length;
2519            }
2520          MRRETURN(MATCH_NOMATCH);
2521        }        }
2522      /* Control never gets here */      /* Control never gets here */
2523    
   
   
2524      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2525      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,
2526      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 1709  for (;;) Line 2575  for (;;)
2575          {          {
2576          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2577            {            {
2578            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2579                {
2580                SCHECK_PARTIAL();
2581                MRRETURN(MATCH_NOMATCH);
2582                }
2583            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2584            if (c > 255)            if (c > 255)
2585              {              {
2586              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2587              }              }
2588            else            else
2589              {              {
2590              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2591              }              }
2592            }            }
2593          }          }
# Line 1727  for (;;) Line 2597  for (;;)
2597          {          {
2598          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2599            {            {
2600            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2601                {
2602                SCHECK_PARTIAL();
2603                MRRETURN(MATCH_NOMATCH);
2604                }
2605            c = *eptr++;            c = *eptr++;
2606            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2607            }            }
2608          }          }
2609    
# Line 1749  for (;;) Line 2623  for (;;)
2623            {            {
2624            for (fi = min;; fi++)            for (fi = min;; fi++)
2625              {              {
2626              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2627              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2628              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2629                if (eptr >= md->end_subject)
2630                  {
2631                  SCHECK_PARTIAL();
2632                  MRRETURN(MATCH_NOMATCH);
2633                  }
2634              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2635              if (c > 255)              if (c > 255)
2636                {                {
2637                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2638                }                }
2639              else              else
2640                {                {
2641                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2642                }                }
2643              }              }
2644            }            }
# Line 1769  for (;;) Line 2648  for (;;)
2648            {            {
2649            for (fi = min;; fi++)            for (fi = min;; fi++)
2650              {              {
2651              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2652              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2653              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2654                if (eptr >= md->end_subject)
2655                  {
2656                  SCHECK_PARTIAL();
2657                  MRRETURN(MATCH_NOMATCH);
2658                  }
2659              c = *eptr++;              c = *eptr++;
2660              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2661              }              }
2662            }            }
2663          /* Control never gets here */          /* Control never gets here */
# Line 1792  for (;;) Line 2676  for (;;)
2676            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2677              {              {
2678              int len = 1;              int len = 1;
2679              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2680                  {
2681                  SCHECK_PARTIAL();
2682                  break;
2683                  }
2684              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2685              if (c > 255)              if (c > 255)
2686                {                {
# Line 1806  for (;;) Line 2694  for (;;)
2694              }              }
2695            for (;;)            for (;;)
2696              {              {
2697              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2698              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2699              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2700              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1818  for (;;) Line 2706  for (;;)
2706            {            {
2707            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2708              {              {
2709              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2710                  {
2711                  SCHECK_PARTIAL();
2712                  break;
2713                  }
2714              c = *eptr;              c = *eptr;
2715              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2716              eptr++;              eptr++;
2717              }              }
2718            while (eptr >= pp)            while (eptr >= pp)
2719              {              {
2720              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2721              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2722              eptr--;              eptr--;
2723              }              }
2724            }            }
2725    
2726          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2727          }          }
2728        }        }
2729      /* Control never gets here */      /* Control never gets here */
2730    
2731    
2732      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2733      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
2734        mode, because Unicode properties are supported in non-UTF-8 mode. */
2735    
2736  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2737      case OP_XCLASS:      case OP_XCLASS:
# Line 1879  for (;;) Line 2772  for (;;)
2772    
2773        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2774          {          {
2775          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2776          GETCHARINC(c, eptr);            {
2777          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2778              MRRETURN(MATCH_NOMATCH);
2779              }
2780            GETCHARINCTEST(c, eptr);
2781            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2782          }          }
2783    
2784        /* 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 1896  for (;;) Line 2793  for (;;)
2793          {          {
2794          for (fi = min;; fi++)          for (fi = min;; fi++)
2795            {            {
2796            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2797            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2798            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2799            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2800            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2801                SCHECK_PARTIAL();
2802                MRRETURN(MATCH_NOMATCH);
2803                }
2804              GETCHARINCTEST(c, eptr);
2805              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2806            }            }
2807          /* Control never gets here */          /* Control never gets here */
2808          }          }
# Line 1913  for (;;) Line 2815  for (;;)
2815          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2816            {            {
2817            int len = 1;            int len = 1;
2818            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2819            GETCHARLEN(c, eptr, len);              {
2820                SCHECK_PARTIAL();
2821                break;
2822                }
2823              GETCHARLENTEST(c, eptr, len);
2824            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2825            eptr += len;            eptr += len;
2826            }            }
2827          for(;;)          for(;;)
2828            {            {
2829            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2830            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2831            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2832            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2833            }            }
2834          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2835          }          }
2836    
2837        /* Control never gets here */        /* Control never gets here */
# Line 1941  for (;;) Line 2847  for (;;)
2847        length = 1;        length = 1;
2848        ecode++;        ecode++;
2849        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2850        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2851        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2852            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2853            MRRETURN(MATCH_NOMATCH);
2854            }
2855          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2856        }        }
2857      else      else
2858  #endif  #endif
2859    
2860      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2861        {        {
2862        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2863        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2864            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2865            MRRETURN(MATCH_NOMATCH);
2866            }
2867          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2868        ecode += 2;        ecode += 2;
2869        }        }
2870      break;      break;
2871    
2872      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2873    
2874      case OP_CHARNC:      case OP_CHARI:
2875  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2876      if (utf8)      if (utf8)
2877        {        {
# Line 1965  for (;;) Line 2879  for (;;)
2879        ecode++;        ecode++;
2880        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2881    
2882        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2883            {
2884            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2885            MRRETURN(MATCH_NOMATCH);
2886            }
2887    
2888        /* 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
2889        can use the fast lookup table. */        can use the fast lookup table. */
2890    
2891        if (fc < 128)        if (fc < 128)
2892          {          {
2893          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2894          }          }
2895    
2896        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
# Line 1989  for (;;) Line 2907  for (;;)
2907          if (fc != dc)          if (fc != dc)
2908            {            {
2909  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2910            if (dc != _pcre_ucp_othercase(fc))            if (dc != UCD_OTHERCASE(fc))
2911  #endif  #endif
2912              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
2913            }            }
2914          }          }
2915        }        }
# Line 2000  for (;;) Line 2918  for (;;)
2918    
2919      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2920        {        {
2921        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2922        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
2923            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2924            MRRETURN(MATCH_NOMATCH);
2925            }
2926          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2927        ecode += 2;        ecode += 2;
2928        }        }
2929      break;      break;
# Line 2009  for (;;) Line 2931  for (;;)
2931      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2932    
2933      case OP_EXACT:      case OP_EXACT:
2934        case OP_EXACTI:
2935      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2936      ecode += 3;      ecode += 3;
2937      goto REPEATCHAR;      goto REPEATCHAR;
2938    
2939      case OP_POSUPTO:      case OP_POSUPTO:
2940        case OP_POSUPTOI:
2941      possessive = TRUE;      possessive = TRUE;
2942      /* Fall through */      /* Fall through */
2943    
2944      case OP_UPTO:      case OP_UPTO:
2945        case OP_UPTOI:
2946      case OP_MINUPTO:      case OP_MINUPTO:
2947        case OP_MINUPTOI:
2948      min = 0;      min = 0;
2949      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2950      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2951      ecode += 3;      ecode += 3;
2952      goto REPEATCHAR;      goto REPEATCHAR;
2953    
2954      case OP_POSSTAR:      case OP_POSSTAR:
2955        case OP_POSSTARI:
2956      possessive = TRUE;      possessive = TRUE;
2957      min = 0;      min = 0;
2958      max = INT_MAX;      max = INT_MAX;
# Line 2033  for (;;) Line 2960  for (;;)
2960      goto REPEATCHAR;      goto REPEATCHAR;
2961    
2962      case OP_POSPLUS:      case OP_POSPLUS:
2963        case OP_POSPLUSI:
2964      possessive = TRUE;      possessive = TRUE;
2965      min = 1;      min = 1;
2966      max = INT_MAX;      max = INT_MAX;
# Line 2040  for (;;) Line 2968  for (;;)
2968      goto REPEATCHAR;      goto REPEATCHAR;
2969    
2970      case OP_POSQUERY:      case OP_POSQUERY:
2971        case OP_POSQUERYI:
2972      possessive = TRUE;      possessive = TRUE;
2973      min = 0;      min = 0;
2974      max = 1;      max = 1;
# Line 2047  for (;;) Line 2976  for (;;)
2976      goto REPEATCHAR;      goto REPEATCHAR;
2977    
2978      case OP_STAR:      case OP_STAR:
2979        case OP_STARI:
2980      case OP_MINSTAR:      case OP_MINSTAR:
2981        case OP_MINSTARI:
2982      case OP_PLUS:      case OP_PLUS:
2983        case OP_PLUSI:
2984      case OP_MINPLUS:      case OP_MINPLUS:
2985        case OP_MINPLUSI:
2986      case OP_QUERY:      case OP_QUERY:
2987        case OP_QUERYI:
2988      case OP_MINQUERY:      case OP_MINQUERY:
2989      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
2990        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
2991      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
2992      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2993      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2994      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
2995    
2996      /* 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. */  
2997    
2998      REPEATCHAR:      REPEATCHAR:
2999  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2069  for (;;) Line 3002  for (;;)
3002        length = 1;        length = 1;
3003        charptr = ecode;        charptr = ecode;
3004        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3005        ecode += length;        ecode += length;
3006    
3007        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 2079  for (;;) Line 3011  for (;;)
3011          {          {
3012  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3013          unsigned int othercase;          unsigned int othercase;
3014          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3015              (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)              (othercase = UCD_OTHERCASE(fc)) != fc)
3016            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3017          else oclength = 0;          else oclength = 0;
3018  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3019    
3020          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3021            {            {
3022            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3023                memcmp(eptr, charptr, length) == 0) eptr += length;
3024  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3025            /* Need braces because of following else */            else if (oclength > 0 &&
3026            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                     eptr <= md->end_subject - oclength &&
3027                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3028    #endif  /* SUPPORT_UCP */
3029            else            else
3030              {              {
3031              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3032              eptr += oclength;              MRRETURN(MATCH_NOMATCH);
3033              }              }
 #else   /* without SUPPORT_UCP */  
           else { RRETURN(MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3034            }            }
3035    
3036          if (min == max) continue;          if (min == max) continue;
# Line 2107  for (;;) Line 3039  for (;;)
3039            {            {
3040            for (fi = min;; fi++)            for (fi = min;; fi++)
3041              {              {
3042              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3043              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3044              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3045              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3046                  memcmp(eptr, charptr, length) == 0) eptr += length;
3047  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3048              /* Need braces because of following else */              else if (oclength > 0 &&
3049              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                       eptr <= md->end_subject - oclength &&
3050                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3051    #endif  /* SUPPORT_UCP */
3052              else              else
3053                {                {
3054                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3055                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3056                }                }
 #else   /* without SUPPORT_UCP */  
             else { RRETURN (MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3057              }              }
3058            /* Control never gets here */            /* Control never gets here */
3059            }            }
# Line 2131  for (;;) Line 3063  for (;;)
3063            pp = eptr;            pp = eptr;
3064            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3065              {              {
3066              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3067              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3068  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3069              else if (oclength == 0) break;              else if (oclength > 0 &&
3070                         eptr <= md->end_subject - oclength &&
3071                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3072    #endif  /* SUPPORT_UCP */
3073              else              else
3074                {                {
3075                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3076                eptr += oclength;                break;
3077                }                }
 #else   /* without SUPPORT_UCP */  
             else break;  
 #endif  /* SUPPORT_UCP */  
3078              }              }
3079    
3080            if (possessive) continue;            if (possessive) continue;
3081    
3082            for(;;)            for(;;)
3083             {              {
3084             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3085             if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3086             if (eptr == pp) RRETURN(MATCH_NOMATCH);              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3087  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3088             eptr--;              eptr--;
3089             BACKCHAR(eptr);              BACKCHAR(eptr);
3090  #else   /* without SUPPORT_UCP */  #else   /* without SUPPORT_UCP */
3091             eptr -= length;              eptr -= length;
3092  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3093             }              }
3094            }            }
3095          /* Control never gets here */          /* Control never gets here */
3096          }          }
# Line 2170  for (;;) Line 3103  for (;;)
3103  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3104    
3105      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3106        {  
3107        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3108    
3109      /* 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
3110      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 2187  for (;;) Line 3118  for (;;)
3118      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3119        max, eptr));        max, eptr));
3120    
3121      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3122        {        {
3123        fc = md->lcc[fc];        fc = md->lcc[fc];
3124        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3125          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3126            if (eptr >= md->end_subject)
3127              {
3128              SCHECK_PARTIAL();
3129              MRRETURN(MATCH_NOMATCH);
3130              }
3131            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3132            }
3133        if (min == max) continue;        if (min == max) continue;
3134        if (minimize)        if (minimize)
3135          {          {
3136          for (fi = min;; fi++)          for (fi = min;; fi++)
3137            {            {
3138            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3139            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3140            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3141                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3142              RRETURN(MATCH_NOMATCH);              {
3143                SCHECK_PARTIAL();
3144                MRRETURN(MATCH_NOMATCH);
3145                }
3146              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3147            }            }
3148          /* Control never gets here */          /* Control never gets here */
3149          }          }
# Line 2210  for (;;) Line 3152  for (;;)
3152          pp = eptr;          pp = eptr;
3153          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3154            {            {
3155            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3156                {
3157                SCHECK_PARTIAL();
3158                break;
3159                }
3160              if (fc != md->lcc[*eptr]) break;
3161            eptr++;            eptr++;
3162            }            }
3163    
3164          if (possessive) continue;          if (possessive) continue;
3165    
3166          while (eptr >= pp)          while (eptr >= pp)
3167            {            {
3168            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3169            eptr--;            eptr--;
3170            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3171            }            }
3172          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3173          }          }
3174        /* Control never gets here */        /* Control never gets here */
3175        }        }
# Line 2229  for (;;) Line 3178  for (;;)
3178    
3179      else      else
3180        {        {
3181        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3182            {
3183            if (eptr >= md->end_subject)
3184              {
3185              SCHECK_PARTIAL();
3186              MRRETURN(MATCH_NOMATCH);
3187              }
3188            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3189            }
3190    
3191        if (min == max) continue;        if (min == max) continue;
3192    
3193        if (minimize)        if (minimize)
3194          {          {
3195          for (fi = min;; fi++)          for (fi = min;; fi++)
3196            {            {
3197            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3198            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3199            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3200              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3201                {
3202                SCHECK_PARTIAL();
3203                MRRETURN(MATCH_NOMATCH);
3204                }
3205              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3206            }            }
3207          /* Control never gets here */          /* Control never gets here */
3208          }          }
# Line 2247  for (;;) Line 3211  for (;;)
3211          pp = eptr;          pp = eptr;
3212          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3213            {            {
3214            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3215                {
3216                SCHECK_PARTIAL();
3217                break;
3218                }
3219              if (fc != *eptr) break;
3220            eptr++;            eptr++;
3221            }            }
3222          if (possessive) continue;          if (possessive) continue;
3223    
3224          while (eptr >= pp)          while (eptr >= pp)
3225            {            {
3226            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3227            eptr--;            eptr--;
3228            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3229            }            }
3230          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3231          }          }
3232        }        }
3233      /* Control never gets here */      /* Control never gets here */
# Line 2266  for (;;) Line 3236  for (;;)
3236      checking can be multibyte. */      checking can be multibyte. */
3237    
3238      case OP_NOT:      case OP_NOT:
3239      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3240        if (eptr >= md->end_subject)
3241          {
3242          SCHECK_PARTIAL();
3243          MRRETURN(MATCH_NOMATCH);
3244          }
3245      ecode++;      ecode++;
3246      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3247      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3248        {        {
3249  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3250        if (c < 256)        if (c < 256)
3251  #endif  #endif
3252        c = md->lcc[c];        c = md->lcc[c];
3253        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3254        }        }
3255      else      else    /* Caseful */
3256        {        {
3257        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3258        }        }
3259      break;      break;
3260    
# Line 2291  for (;;) Line 3266  for (;;)
3266      about... */      about... */
3267    
3268      case OP_NOTEXACT:      case OP_NOTEXACT:
3269        case OP_NOTEXACTI:
3270      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3271      ecode += 3;      ecode += 3;
3272      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3273    
3274      case OP_NOTUPTO:      case OP_NOTUPTO:
3275        case OP_NOTUPTOI:
3276      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3277        case OP_NOTMINUPTOI:
3278      min = 0;      min = 0;
3279      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3280      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3281      ecode += 3;      ecode += 3;
3282      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3283    
3284      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3285        case OP_NOTPOSSTARI:
3286      possessive = TRUE;      possessive = TRUE;
3287      min = 0;      min = 0;
3288      max = INT_MAX;      max = INT_MAX;
# Line 2311  for (;;) Line 3290  for (;;)
3290      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3291    
3292      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3293        case OP_NOTPOSPLUSI:
3294      possessive = TRUE;      possessive = TRUE;
3295      min = 1;      min = 1;
3296      max = INT_MAX;      max = INT_MAX;
# Line 2318  for (;;) Line 3298  for (;;)
3298      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3299    
3300      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3301        case OP_NOTPOSQUERYI:
3302      possessive = TRUE;      possessive = TRUE;
3303      min = 0;      min = 0;
3304      max = 1;      max = 1;
# Line 2325  for (;;) Line 3306  for (;;)
3306      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3307    
3308      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3309        case OP_NOTPOSUPTOI:
3310      possessive = TRUE;      possessive = TRUE;
3311      min = 0;      min = 0;
3312      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 2332  for (;;) Line 3314  for (;;)
3314      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3315    
3316      case OP_NOTSTAR:      case OP_NOTSTAR:
3317        case OP_NOTSTARI:
3318      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3319        case OP_NOTMINSTARI:
3320      case OP_NOTPLUS:      case OP_NOTPLUS:
3321        case OP_NOTPLUSI:
3322      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3323        case OP_NOTMINPLUSI:
3324      case OP_NOTQUERY:      case OP_NOTQUERY:
3325        case OP_NOTQUERYI:
3326      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3327      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3328        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3329      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3330      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3331      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3332      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3333    
3334      /* 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. */  
3335    
3336      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3337      fc = *ecode++;      fc = *ecode++;
3338    
3339      /* 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 2362  for (;;) Line 3347  for (;;)
3347      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,
3348        max, eptr));        max, eptr));
3349    
3350      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3351        {        {
3352        fc = md->lcc[fc];        fc = md->lcc[fc];
3353    
# Line 2373  for (;;) Line 3358  for (;;)
3358          register unsigned int d;          register unsigned int d;
3359          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3360            {            {
3361              if (eptr >= md->end_subject)
3362                {
3363                SCHECK_PARTIAL();
3364                MRRETURN(MATCH_NOMATCH);
3365                }
3366            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3367            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3368            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3369            }            }
3370          }          }
3371        else        else
# Line 2384  for (;;) Line 3374  for (;;)
3374        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3375          {          {
3376          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3377            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3378              if (eptr >= md->end_subject)
3379                {
3380                SCHECK_PARTIAL();
3381                MRRETURN(MATCH_NOMATCH);
3382                }
3383              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3384              }
3385          }          }
3386    
3387        if (min == max) continue;        if (min == max) continue;
# Line 2398  for (;;) Line 3395  for (;;)
3395            register unsigned int d;            register unsigned int d;
3396            for (fi = min;; fi++)            for (fi = min;; fi++)
3397              {              {
3398              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3399              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3400                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3401                if (eptr >= md->end_subject)
3402                  {
3403                  SCHECK_PARTIAL();
3404                  MRRETURN(MATCH_NOMATCH);
3405                  }
3406              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3407              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3408              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3409              }              }
3410            }            }
3411          else          else
# Line 2412  for (;;) Line 3414  for (;;)
3414            {            {
3415            for (fi = min;; fi++)            for (fi = min;; fi++)
3416              {              {
3417              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3418              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3419              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3420                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3421                  {
3422                  SCHECK_PARTIAL();
3423                  MRRETURN(MATCH_NOMATCH);
3424                  }
3425                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3426              }              }
3427            }            }
3428          /* Control never gets here */          /* Control never gets here */
# Line 2435  for (;;) Line 3442  for (;;)
3442            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3443              {              {
3444              int len = 1;              int len = 1;
3445              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3446                  {
3447                  SCHECK_PARTIAL();
3448                  break;
3449                  }
3450              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3451              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3452              if (fc == d) break;              if (fc == d) break;
# Line 2444  for (;;) Line 3455  for (;;)
3455          if (possessive) continue;          if (possessive) continue;
3456          for(;;)          for(;;)
3457              {              {
3458              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3459              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3460              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3461              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2456  for (;;) Line 3467  for (;;)
3467            {            {
3468            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3469              {              {
3470              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3471                  {
3472                  SCHECK_PARTIAL();
3473                  break;
3474                  }
3475                if (fc == md->lcc[*eptr]) break;
3476              eptr++;              eptr++;
3477              }              }
3478            if (possessive) continue;            if (possessive) continue;
3479            while (eptr >= pp)            while (eptr >= pp)
3480              {              {
3481              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3482              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3483              eptr--;              eptr--;
3484              }              }
3485            }            }
3486    
3487          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3488          }          }
3489        /* Control never gets here */        /* Control never gets here */
3490        }        }
# Line 2484  for (;;) Line 3500  for (;;)
3500          register unsigned int d;          register unsigned int d;
3501          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3502            {            {
3503              if (eptr >= md->end_subject)
3504                {
3505                SCHECK_PARTIAL();
3506                MRRETURN(MATCH_NOMATCH);
3507                }
3508            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3509            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3510            }            }
3511          }          }
3512        else        else
# Line 2493  for (;;) Line 3514  for (;;)
3514        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3515          {          {
3516          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3517            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3518              if (eptr >= md->end_subject)
3519                {
3520                SCHECK_PARTIAL();
3521                MRRETURN(MATCH_NOMATCH);
3522                }
3523              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3524              }
3525          }          }
3526    
3527        if (min == max) continue;        if (min == max) continue;
# Line 2507  for (;;) Line 3535  for (;;)
3535            register unsigned int d;            register unsigned int d;
3536            for (fi = min;; fi++)            for (fi = min;; fi++)
3537              {              {
3538              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3539              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3540                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3541                if (eptr >= md->end_subject)
3542                  {
3543                  SCHECK_PARTIAL();
3544                  MRRETURN(MATCH_NOMATCH);
3545                  }
3546              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3547              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3548              }              }
3549            }            }
3550          else          else
# Line 2520  for (;;) Line 3553  for (;;)
3553            {            {
3554            for (fi = min;; fi++)            for (fi = min;; fi++)
3555              {              {
3556              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3557              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3558              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3559                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3560                  {
3561                  SCHECK_PARTIAL();
3562                  MRRETURN(MATCH_NOMATCH);
3563                  }
3564                if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3565              }              }
3566            }            }
3567          /* Control never gets here */          /* Control never gets here */
# Line 2543  for (;;) Line 3581  for (;;)
3581            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3582              {              {
3583              int len = 1;              int len = 1;
3584              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3585                  {
3586                  SCHECK_PARTIAL();
3587                  break;
3588                  }
3589              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3590              if (fc == d) break;              if (fc == d) break;
3591              eptr += len;              eptr += len;
# Line 2551  for (;;) Line 3593  for (;;)
3593            if (possessive) continue;            if (possessive) continue;
3594            for(;;)            for(;;)
3595              {              {
3596              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3597              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3598              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3599              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2563  for (;;) Line 3605  for (;;)
3605            {            {
3606            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3607              {              {
3608              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3609                  {
3610                  SCHECK_PARTIAL();
3611                  break;
3612                  }
3613                if (fc == *eptr) break;
3614              eptr++;              eptr++;
3615              }              }
3616            if (possessive) continue;            if (possessive) continue;
3617            while (eptr >= pp)            while (eptr >= pp)
3618              {              {
3619              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3620              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3621              eptr--;              eptr--;
3622              }              }
3623            }            }
3624    
3625          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3626          }          }
3627        }        }
3628      /* Control never gets here */      /* Control never gets here */
# Line 2657  for (;;) Line 3704  for (;;)
3704    
3705      /* First, ensure the minimum number of matches are present. Use inline      /* First, ensure the minimum number of matches are present. Use inline
3706      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
3707      (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  
3708      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
3709      and single-bytes. */      and single-bytes. */
3710    
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3711      if (min > 0)      if (min > 0)
3712        {        {
3713  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2672  for (;;) Line 3716  for (;;)
3716          switch(prop_type)          switch(prop_type)
3717            {            {
3718            case PT_ANY:            case PT_ANY:
3719            if (prop_fail_result) RRETURN(MATCH_NOMATCH);            if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
3720            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3721              {              {
3722              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3723              GETCHARINC(c, eptr);                {
3724                  SCHECK_PARTIAL();
3725                  MRRETURN(MATCH_NOMATCH);
3726                  }
3727                GETCHARINCTEST(c, eptr);
3728              }              }
3729            break;            break;
3730    
3731            case PT_LAMP:            case PT_LAMP:
3732            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3733              {              {
3734              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3735              GETCHARINC(c, eptr);                {
3736              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3737                  MRRETURN(MATCH_NOMATCH);
3738                  }
3739                GETCHARINCTEST(c, eptr);
3740                prop_chartype = UCD_CHARTYPE(c);
3741              if ((prop_chartype == ucp_Lu ||              if ((prop_chartype == ucp_Lu ||
3742                   prop_chartype == ucp_Ll ||                   prop_chartype == ucp_Ll ||
3743                   prop_chartype == ucp_Lt) == prop_fail_result)                   prop_chartype == ucp_Lt) == prop_fail_result)
3744                RRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3745              }              }
3746            break;            break;
3747    
3748            case PT_GC:            case PT_GC:
3749            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3750              {              {
3751              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3752              GETCHARINC(c, eptr);                {
3753              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3754                  MRRETURN(MATCH_NOMATCH);
3755                  }
3756                GETCHARINCTEST(c, eptr);
3757                prop_category = UCD_CATEGORY(c);
3758              if ((prop_category == prop_value) == prop_fail_result)              if ((prop_category == prop_value) == prop_fail_result)
3759                RRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3760              }              }
3761            break;            break;
3762    
3763            case PT_PC:            case PT_PC:
3764            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3765              {              {
3766              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3767              GETCHARINC(c, eptr);                {
3768              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3769                  MRRETURN(MATCH_NOMATCH);
3770                  }
3771                GETCHARINCTEST(c, eptr);
3772                prop_chartype = UCD_CHARTYPE(c);
3773              if ((prop_chartype == prop_value) == prop_fail_result)              if ((prop_chartype == prop_value) == prop_fail_result)
3774                RRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3775              }              }
3776            break;            break;
3777    
3778            case PT_SC:            case PT_SC:
3779            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3780              {              {
3781              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3782              GETCHARINC(c, eptr);                {
3783              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3784                  MRRETURN(MATCH_NOMATCH);
3785                  }
3786                GETCHARINCTEST(c, eptr);
3787                prop_script = UCD_SCRIPT(c);
3788              if ((prop_script == prop_value) == prop_fail_result)              if ((prop_script == prop_value) == prop_fail_result)
3789                RRETURN(MATCH_NOMATCH);                MRRETURN(MATCH_NOMATCH);
3790                }
3791              break;
3792    
3793              case PT_ALNUM:
3794              for (i = 1; i <= min; i++)
3795                {
3796                if (eptr >= md->end_subject)
3797                  {
3798                  SCHECK_PARTIAL();
3799                  MRRETURN(MATCH_NOMATCH);
3800                  }
3801                GETCHARINCTEST(c, eptr);
3802                prop_category = UCD_CATEGORY(c);
3803                if ((prop_category == ucp_L || prop_category == ucp_N)
3804                       == prop_fail_result)
3805                  MRRETURN(MATCH_NOMATCH);
3806                }
3807              break;
3808    
3809              case PT_SPACE:    /* Perl space */
3810              for (i = 1; i <= min; i++)
3811                {
3812                if (eptr >= md->end_subject)
3813                  {
3814                  SCHECK_PARTIAL();
3815                  MRRETURN(MATCH_NOMATCH);
3816                  }
3817                GETCHARINCTEST(c, eptr);
3818                prop_category = UCD_CATEGORY(c);
3819                if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3820                     c == CHAR_FF || c == CHAR_CR)
3821                       == prop_fail_result)
3822                  MRRETURN(MATCH_NOMATCH);
3823                }
3824              break;
3825    
3826              case PT_PXSPACE:  /* POSIX space */
3827              for (i = 1; i <= min; i++)
3828                {
3829                if (eptr >= md->end_subject)
3830                  {
3831                  SCHECK_PARTIAL();
3832                  MRRETURN(MATCH_NOMATCH);
3833                  }
3834                GETCHARINCTEST(c, eptr);
3835                prop_category = UCD_CATEGORY(c);
3836                if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3837                     c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)
3838                       == prop_fail_result)
3839                  MRRETURN(MATCH_NOMATCH);
3840                }
3841              break;
3842    
3843              case PT_WORD:
3844              for (i = 1; i <= min; i++)
3845                {
3846                if (eptr >= md->end_subject)
3847                  {
3848                  SCHECK_PARTIAL();
3849                  MRRETURN(MATCH_NOMATCH);
3850                  }
3851                GETCHARINCTEST(c, eptr);
3852                prop_category = UCD_CATEGORY(c);
3853                if ((prop_category == ucp_L || prop_category == ucp_N ||
3854                     c == CHAR_UNDERSCORE)
3855                       == prop_fail_result)
3856                  MRRETURN(MATCH_NOMATCH);
3857              }              }
3858            break;            break;
3859    
3860              /* This should not occur */
3861    
3862            default:            default:
3863            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
3864            }            }
# Line 2738  for (