/[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 172 by ph10, Tue Jun 5 10:40:13 2007 UTC revision 699 by ph10, Tue Sep 20 10:46:54 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 };
281    
282  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
283  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
284  actuall used in this definition. */  actually used in this definition. */
285    
286  #ifndef NO_RECURSE  #ifndef NO_RECURSE
287  #define REGISTER register  #define REGISTER register
288    
289  #ifdef DEBUG  #ifdef PCRE_DEBUG
290  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \  #define RMATCH(ra,rb,rc,rd,re,rw) \
291    { \    { \
292    printf("match() called in line %d\n", __LINE__); \    printf("match() called in line %d\n", __LINE__); \
293    rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \
294    printf("to line %d\n", __LINE__); \    printf("to line %d\n", __LINE__); \
295    }    }
296  #define RRETURN(ra) \  #define RRETURN(ra) \
# Line 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,mstart,rc,rd,re,rf,rg,rdepth+1)    rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1)
304  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
305  #endif  #endif
306    
# Line 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;\    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 271  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 290  typedef struct heapframe { Line 354  typedef struct heapframe {
354    
355    /* Function arguments that may change */    /* Function arguments that may change */
356    
357    const uschar *Xeptr;    USPTR Xeptr;
358    const uschar *Xecode;    const uschar *Xecode;
359    const uschar *Xmstart;    USPTR Xmstart;
360      USPTR Xmarkptr;
361    int Xoffset_top;    int Xoffset_top;
   long int Xims;  
362    eptrblock *Xeptrb;    eptrblock *Xeptrb;
   int Xflags;  
363    unsigned int Xrdepth;    unsigned int Xrdepth;
364    
365    /* Function local variables */    /* Function local variables */
366    
367    const uschar *Xcallpat;    USPTR Xcallpat;
368    const uschar *Xcharptr;  #ifdef SUPPORT_UTF8
369    const uschar *Xdata;    USPTR Xcharptr;
370    const uschar *Xnext;  #endif
371    const uschar *Xpp;    USPTR Xdata;
372    const uschar *Xprev;    USPTR Xnext;
373    const uschar *Xsaved_eptr;    USPTR Xpp;
374      USPTR Xprev;
375      USPTR Xsaved_eptr;
376    
377    recursion_info Xnew_recursive;    recursion_info Xnew_recursive;
378    
# Line 315  typedef struct heapframe { Line 380  typedef struct heapframe {
380    BOOL Xcondition;    BOOL Xcondition;
381    BOOL Xprev_is_word;    BOOL Xprev_is_word;
382    
   unsigned long int Xoriginal_ims;  
   
383  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
384    int Xprop_type;    int Xprop_type;
385    int Xprop_value;    int Xprop_value;
386    int Xprop_fail_result;    int Xprop_fail_result;
   int Xprop_category;  
   int Xprop_chartype;  
   int Xprop_script;  
387    int Xoclength;    int Xoclength;
388    uschar Xocchars[8];    uschar Xocchars[8];
389  #endif  #endif
390    
391      int Xcodelink;
392    int Xctype;    int Xctype;
393    unsigned int Xfc;    unsigned int Xfc;
394    int Xfi;    int Xfi;
# Line 363  typedef struct heapframe { Line 424  typedef struct heapframe {
424    
425  /* This function is called recursively in many circumstances. Whenever it  /* This function is called recursively in many circumstances. Whenever it
426  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
427  same response.  same response. */
428    
429    /* These macros pack up tests that are used for partial matching, and which
430    appears several times in the code. We set the "hit end" flag if the pointer is
431    at the end of the subject and also past the start of the subject (i.e.
432    something has been matched). For hard partial matching, we then return
433    immediately. The second one is used when we already know we are past the end of
434    the subject. */
435    
436    #define CHECK_PARTIAL()\
437      if (md->partial != 0 && eptr >= md->end_subject && \
438          eptr > md->start_used_ptr) \
439        { \
440        md->hitend = TRUE; \
441        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
442        }
443    
444    #define SCHECK_PARTIAL()\
445      if (md->partial != 0 && eptr > md->start_used_ptr) \
446        { \
447        md->hitend = TRUE; \
448        if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \
449        }
450    
451    
452  Performance note: It might be tempting to extract commonly used fields from the  /* Performance note: It might be tempting to extract commonly used fields from
453  md structure (e.g. utf8, end_subject) into individual variables to improve  the md structure (e.g. utf8, end_subject) into individual variables to improve
454  performance. Tests using gcc on a SPARC disproved this; in the first case, it  performance. Tests using gcc on a SPARC disproved this; in the first case, it
455  made performance worse.  made performance worse.
456    
# Line 375  Arguments: Line 459  Arguments:
459     ecode       pointer to current position in compiled code     ecode       pointer to current position in compiled code
460     mstart      pointer to the current match start position (can be modified     mstart      pointer to the current match start position (can be modified
461                   by encountering \K)                   by encountering \K)
462       markptr     pointer to the most recent MARK name, or NULL
463     offset_top  current top pointer     offset_top  current top pointer
464     md          pointer to "static" info for the match     md          pointer to "static" info for the match
    ims         current /i, /m, and /s options  
465     eptrb       pointer to chain of blocks containing eptr at start of     eptrb       pointer to chain of blocks containing eptr at start of
466                   brackets - for testing for empty matches                   brackets - for testing for empty matches
    flags       can contain  
                  match_condassert - this is an assertion condition  
                  match_cbegroup - this is the start of an unlimited repeat  
                    group that can match an empty string  
                  match_tail_recursed - this is a tail_recursed group  
467     rdepth      the recursion depth     rdepth      the recursion depth
468    
469  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
470                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
471                   a negative MATCH_xxx value for PRUNE, SKIP, etc
472                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
473                   (e.g. stopped by repeated call or recursion limit)                   (e.g. stopped by repeated call or recursion limit)
474  */  */
475    
476  static int  static int
477  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
478    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb,
479    int flags, unsigned int rdepth)    unsigned int rdepth)
480  {  {
481  /* These variables do not need to be preserved over recursion in this function,  /* These variables do not need to be preserved over recursion in this function,
482  so they can be ordinary variables in all cases. Mark some of them with  so they can be ordinary variables in all cases. Mark some of them with
# Line 408  register unsigned int c;   /* Character Line 488  register unsigned int c;   /* Character
488  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
489    
490  BOOL minimize, possessive; /* Quantifier options */  BOOL minimize, possessive; /* Quantifier options */
491    BOOL caseless;
492    int condcode;
493    
494  /* When recursion is not being used, all "local" variables that have to be  /* When recursion is not being used, all "local" variables that have to be
495  preserved over calls to RMATCH() are part of a "frame" which is obtained from  preserved over calls to RMATCH() are part of a "frame" which is obtained from
# Line 415  heap storage. Set up the top-level frame Line 497  heap storage. Set up the top-level frame
497  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
498    
499  #ifdef NO_RECURSE  #ifdef NO_RECURSE
500  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
501    if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
502  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
503    
504  /* Copy in the original argument variables */  /* Copy in the original argument variables */
# Line 423  frame->Xprevframe = NULL;            /* Line 506  frame->Xprevframe = NULL;            /*
506  frame->Xeptr = eptr;  frame->Xeptr = eptr;
507  frame->Xecode = ecode;  frame->Xecode = ecode;
508  frame->Xmstart = mstart;  frame->Xmstart = mstart;
509    frame->Xmarkptr = markptr;
510  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
 frame->Xims = ims;  
511  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
 frame->Xflags = flags;  
512  frame->Xrdepth = rdepth;  frame->Xrdepth = rdepth;
513    
514  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
# Line 438  HEAP_RECURSE: Line 520  HEAP_RECURSE:
520  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
521  #define ecode              frame->Xecode  #define ecode              frame->Xecode
522  #define mstart             frame->Xmstart  #define mstart             frame->Xmstart
523    #define markptr            frame->Xmarkptr
524  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
 #define ims                frame->Xims  
525  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
 #define flags              frame->Xflags  
526  #define rdepth             frame->Xrdepth  #define rdepth             frame->Xrdepth
527    
528  /* Ditto for the local variables */  /* Ditto for the local variables */
# Line 450  HEAP_RECURSE: Line 531  HEAP_RECURSE:
531  #define charptr            frame->Xcharptr  #define charptr            frame->Xcharptr
532  #endif  #endif
533  #define callpat            frame->Xcallpat  #define callpat            frame->Xcallpat
534    #define codelink           frame->Xcodelink
535  #define data               frame->Xdata  #define data               frame->Xdata
536  #define next               frame->Xnext  #define next               frame->Xnext
537  #define pp                 frame->Xpp  #define pp                 frame->Xpp
# Line 462  HEAP_RECURSE: Line 544  HEAP_RECURSE:
544  #define condition          frame->Xcondition  #define condition          frame->Xcondition
545  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
546    
 #define original_ims       frame->Xoriginal_ims  
   
547  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
548  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
549  #define prop_value         frame->Xprop_value  #define prop_value         frame->Xprop_value
550  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
 #define prop_category      frame->Xprop_category  
 #define prop_chartype      frame->Xprop_chartype  
 #define prop_script        frame->Xprop_script  
551  #define oclength           frame->Xoclength  #define oclength           frame->Xoclength
552  #define occhars            frame->Xocchars  #define occhars            frame->Xocchars
553  #endif  #endif
# Line 500  i, and fc and c, can be the same variabl Line 577  i, and fc and c, can be the same variabl
577  #define fi i  #define fi i
578  #define fc c  #define fc c
579    
580    /* Many of the following variables are used only in small blocks of the code.
581    My normal style of coding would have declared them within each of those blocks.
582    However, in order to accommodate the version of this code that uses an external
583    "stack" implemented on the heap, it is easier to declare them all here, so the
584    declarations can be cut out in a block. The only declarations within blocks
585    below are for variables that do not have to be preserved over a recursive call
586    to RMATCH(). */
587    
588    #ifdef SUPPORT_UTF8
589    const uschar *charptr;
590    #endif
591    const uschar *callpat;
592    const uschar *data;
593    const uschar *next;
594    USPTR         pp;
595    const uschar *prev;
596    USPTR         saved_eptr;
597    
598    recursion_info new_recursive;
599    
600  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */  BOOL cur_is_word;
 const uschar *charptr;             /* in small blocks of the code. My normal */  
 #endif                             /* style of coding would have declared    */  
 const uschar *callpat;             /* them within each of those blocks.      */  
 const uschar *data;                /* However, in order to accommodate the   */  
 const uschar *next;                /* version of this code that uses an      */  
 USPTR         pp;                  /* external "stack" implemented on the    */  
 const uschar *prev;                /* heap, it is easier to declare them all */  
 USPTR         saved_eptr;          /* here, so the declarations can be cut   */  
                                    /* out in a block. The only declarations  */  
 recursion_info new_recursive;      /* within blocks below are for variables  */  
                                    /* that do not have to be preserved over  */  
 BOOL cur_is_word;                  /* a recursive call to RMATCH().          */  
601  BOOL condition;  BOOL condition;
602  BOOL prev_is_word;  BOOL prev_is_word;
603    
 unsigned long int original_ims;  
   
604  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
605  int prop_type;  int prop_type;
606  int prop_value;  int prop_value;
607  int prop_fail_result;  int prop_fail_result;
 int prop_category;  
 int prop_chartype;  
 int prop_script;  
608  int oclength;  int oclength;
609  uschar occhars[8];  uschar occhars[8];
610  #endif  #endif
611    
612    int codelink;
613  int ctype;  int ctype;
614  int length;  int length;
615  int max;  int max;
# Line 544  int stacksave[REC_STACK_SAVE_MAX]; Line 624  int stacksave[REC_STACK_SAVE_MAX];
624  eptrblock newptrb;  eptrblock newptrb;
625  #endif     /* NO_RECURSE */  #endif     /* NO_RECURSE */
626    
627    /* To save space on the stack and in the heap frame, I have doubled up on some
628    of the local variables that are used only in localised parts of the code, but
629    still need to be preserved over recursive calls of match(). These macros define
630    the alternative names that are used. */
631    
632    #define allow_zero    cur_is_word
633    #define cbegroup      condition
634    #define code_offset   codelink
635    #define condassert    condition
636    #define matched_once  prev_is_word
637    
638  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
639  variables. */  variables. */
640    
# Line 563  TAIL_RECURSE: Line 654  TAIL_RECURSE:
654  /* OK, now we can get on with the real code of the function. Recursive calls  /* OK, now we can get on with the real code of the function. Recursive calls
655  are specified by the macro RMATCH and RRETURN is used to return. When  are specified by the macro RMATCH and RRETURN is used to return. When
656  NO_RECURSE is *not* defined, these just turn into a recursive call to match()  NO_RECURSE is *not* defined, these just turn into a recursive call to match()
657  and a "return", respectively (possibly with some debugging if DEBUG is  and a "return", respectively (possibly with some debugging if PCRE_DEBUG is
658  defined). However, RMATCH isn't like a function call because it's quite a  defined). However, RMATCH isn't like a function call because it's quite a
659  complicated macro. It has to be used in one particular way. This shouldn't,  complicated macro. It has to be used in one particular way. This shouldn't,
660  however, impact performance when true recursion is being used. */  however, impact performance when true recursion is being used. */
# Line 580  haven't exceeded the recursive call limi Line 671  haven't exceeded the recursive call limi
671  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
672  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);  if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
673    
 original_ims = ims;    /* Save for resetting on ')' */  
   
674  /* At the start of a group with an unlimited repeat that may match an empty  /* At the start of a group with an unlimited repeat that may match an empty
675  string, the match_cbegroup flag is set. When this is the case, add the current  string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is
676  subject pointer to the chain of such remembered pointers, to be checked when we  done this way to save having to use another function argument, which would take
677  hit the closing ket, in order to break infinite loops that match no characters.  up space on the stack. See also MATCH_CONDASSERT below.
678  When match() is called in other circumstances, don't add to the chain. If this  
679  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
680  already used. */  such remembered pointers, to be checked when we hit the closing ket, in order
681    to break infinite loops that match no characters. When match() is called in
682    other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must
683    NOT be used with tail recursion, because the memory block that is used is on
684    the stack, so a new one may be required for each match(). */
685    
686  if ((flags & match_cbegroup) != 0)  if (md->match_function_type == MATCH_CBEGROUP)
687    {    {
688    eptrblock *p;    newptrb.epb_saved_eptr = eptr;
689    if ((flags & match_tail_recursed) != 0)    newptrb.epb_prev = eptrb;
690      {    eptrb = &newptrb;
691      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;  
692    }    }
693    
694  /* Now start processing the opcodes. */  /* Now start processing the opcodes. */
# Line 611  for (;;) Line 698  for (;;)
698    minimize = possessive = FALSE;    minimize = possessive = FALSE;
699    op = *ecode;    op = *ecode;
700    
   /* For partial matching, remember if we ever hit the end of the subject after  
   matching at least one subject character. */  
   
   if (md->partial &&  
       eptr >= md->end_subject &&  
       eptr > mstart)  
     md->hitend = TRUE;  
   
701    switch(op)    switch(op)
702      {      {
703      /* Handle a capturing bracket. If there is space in the offset vector, save      case OP_MARK:
704      the current subject position in the working slot at the top of the vector.      markptr = ecode + 2;
705      We mustn't change the current values of the data slot, because they may be      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
706      set from a previous iteration of this group, and be referred to by a        eptrb, RM55);
707      reference inside the group.  
708        /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an
709      If the bracket fails to match, we need to restore this value and also the      argument, and we must check whether that argument matches this MARK's
710      values of the final offsets, in case they were set by a previous iteration      argument. It is passed back in md->start_match_ptr (an overloading of that
711      of the same bracket.      variable). If it does match, we reset that variable to the current subject
712        position and return MATCH_SKIP. Otherwise, pass back the return code
713        unaltered. */
714    
715        if (rrc == MATCH_SKIP_ARG &&
716            strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0)
717          {
718          md->start_match_ptr = eptr;
719          RRETURN(MATCH_SKIP);
720          }
721    
722        if (md->mark == NULL) md->mark = markptr;
723        RRETURN(rrc);
724    
725        case OP_FAIL:
726        MRRETURN(MATCH_NOMATCH);
727    
728        /* COMMIT overrides PRUNE, SKIP, and THEN */
729    
730        case OP_COMMIT:
731        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
732          eptrb, RM52);
733        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&
734            rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&
735            rrc != MATCH_THEN)
736          RRETURN(rrc);
737        MRRETURN(MATCH_COMMIT);
738    
739        /* PRUNE overrides THEN */
740    
741        case OP_PRUNE:
742        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
743          eptrb, RM51);
744        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
745        MRRETURN(MATCH_PRUNE);
746    
747        case OP_PRUNE_ARG:
748        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
749          eptrb, RM56);
750        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
751        md->mark = ecode + 2;
752        RRETURN(MATCH_PRUNE);
753    
754        /* SKIP overrides PRUNE and THEN */
755    
756        case OP_SKIP:
757        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
758          eptrb, RM53);
759        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
760          RRETURN(rrc);
761        md->start_match_ptr = eptr;   /* Pass back current position */
762        MRRETURN(MATCH_SKIP);
763    
764        case OP_SKIP_ARG:
765        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md,
766          eptrb, RM57);
767        if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)
768          RRETURN(rrc);
769    
770        /* Pass back the current skip name by overloading md->start_match_ptr and
771        returning the special MATCH_SKIP_ARG return code. This will either be
772        caught by a matching MARK, or get to the top, where it is treated the same
773        as PRUNE. */
774    
775        md->start_match_ptr = ecode + 2;
776        RRETURN(MATCH_SKIP_ARG);
777    
778        /* For THEN (and THEN_ARG) we pass back the address of the bracket or
779        the alt that is at the start of the current branch. This makes it possible
780        to skip back past alternatives that precede the THEN within the current
781        branch. */
782    
783        case OP_THEN:
784        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
785          eptrb, RM54);
786        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
787        md->start_match_ptr = ecode - GET(ecode, 1);
788        MRRETURN(MATCH_THEN);
789    
790        case OP_THEN_ARG:
791        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE],
792          offset_top, md, eptrb, RM58);
793        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
794        md->start_match_ptr = ecode - GET(ecode, 1);
795        md->mark = ecode + LINK_SIZE + 2;
796        RRETURN(MATCH_THEN);
797    
798        /* Handle a capturing bracket, other than those that are possessive with an
799        unlimited repeat. If there is space in the offset vector, save the current
800        subject position in the working slot at the top of the vector. We mustn't
801        change the current values of the data slot, because they may be set from a
802        previous iteration of this group, and be referred to by a reference inside
803        the group. A failure to match might occur after the group has succeeded,
804        if something later on doesn't match. For this reason, we need to restore
805        the working value and also the values of the final offsets, in case they
806        were set by a previous iteration of the same bracket.
807    
808      If there isn't enough space in the offset vector, treat this as if it were      If there isn't enough space in the offset vector, treat this as if it were
809      a non-capturing bracket. Don't worry about setting the flag for the error      a non-capturing bracket. Don't worry about setting the flag for the error
# Line 640  for (;;) Line 814  for (;;)
814      number = GET2(ecode, 1+LINK_SIZE);      number = GET2(ecode, 1+LINK_SIZE);
815      offset = number << 1;      offset = number << 1;
816    
817  #ifdef DEBUG  #ifdef PCRE_DEBUG
818      printf("start bracket %d\n", number);      printf("start bracket %d\n", number);
819      printf("subject=");      printf("subject=");
820      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
# Line 655  for (;;) Line 829  for (;;)
829        save_capture_last = md->capture_last;        save_capture_last = md->capture_last;
830    
831        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
832        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] =
833            (int)(eptr - md->start_subject);
834    
835        flags = (op == OP_SCBRA)? match_cbegroup : 0;        for (;;)
       do  
836          {          {
837            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
838          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
839            ims, eptrb, flags, RM1);            eptrb, RM1);
840          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc == MATCH_ONCE) break;  /* Backing up through an atomic group */
841            if (rrc != MATCH_NOMATCH &&
842                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
843              RRETURN(rrc);
844          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
845          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
846            if (*ecode != OP_ALT) break;
847          }          }
       while (*ecode == OP_ALT);  
848    
849        DPRINTF(("bracket %d failed\n", number));        DPRINTF(("bracket %d failed\n", number));
   
850        md->offset_vector[offset] = save_offset1;        md->offset_vector[offset] = save_offset1;
851        md->offset_vector[offset+1] = save_offset2;        md->offset_vector[offset+1] = save_offset2;
852        md->offset_vector[md->offset_end - number] = save_offset3;        md->offset_vector[md->offset_end - number] = save_offset3;
853    
854        RRETURN(MATCH_NOMATCH);        /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or
855          MATCH_THEN. */
856    
857          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
858          RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH));
859        }        }
860    
861      /* Insufficient room for saving captured contents. Treat as a non-capturing      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
862      bracket. */      as a non-capturing bracket. */
863    
864        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
865        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
866    
867      DPRINTF(("insufficient capture room: treat as non-capturing\n"));      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
868    
869      /* Non-capturing bracket. Loop for all the alternatives. When we get to the      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
870      final alternative within the brackets, we would return the result of a      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
871      recursive call to match() whatever happened. We can reduce stack usage by  
872      turning this into a tail recursion. */      /* Non-capturing or atomic group, except for possessive with unlimited
873        repeat. Loop for all the alternatives. When we get to the final alternative
874        within the brackets, we used to return the result of a recursive call to
875        match() whatever happened so it was possible to reduce stack usage by
876        turning this into a tail recursion, except in the case of a possibly empty
877        group. However, now that there is the possiblity of (*THEN) occurring in
878        the final alternative, this optimization is no longer possible.
879    
880        MATCH_ONCE is returned when the end of an atomic group is successfully
881        reached, but subsequent matching fails. It passes back up the tree (causing
882        captured values to be reset) until the original atomic group level is
883        reached. This is tested by comparing md->once_target with the start of the
884        group. At this point, the return is converted into MATCH_NOMATCH so that
885        previous backup points can be taken. */
886    
887        case OP_ONCE:
888      case OP_BRA:      case OP_BRA:
889      case OP_SBRA:      case OP_SBRA:
890      DPRINTF(("start non-capturing bracket\n"));      DPRINTF(("start non-capturing bracket\n"));
891      flags = (op >= OP_SBRA)? match_cbegroup : 0;  
892      for (;;)      for (;;)
893        {        {
894        if (ecode[GET(ecode, 1)] != OP_ALT)        if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP;
895          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb,
896            RM2);
897          if (rrc != MATCH_NOMATCH &&
898              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
899            {
900            if (rrc == MATCH_ONCE)
901              {
902              const uschar *scode = ecode;
903              if (*scode != OP_ONCE)           /* If not at start, find it */
904                {
905                while (*scode == OP_ALT) scode += GET(scode, 1);
906                scode -= GET(scode, 1);
907                }
908              if (md->once_target == scode) rrc = MATCH_NOMATCH;
909              }
910            RRETURN(rrc);
911            }
912          ecode += GET(ecode, 1);
913          if (*ecode != OP_ALT) break;
914          }
915        if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
916        RRETURN(MATCH_NOMATCH);
917    
918        /* Handle possessive capturing brackets with an unlimited repeat. We come
919        here from BRAZERO with allow_zero set TRUE. The offset_vector values are
920        handled similarly to the normal case above. However, the matching is
921        different. The end of these brackets will always be OP_KETRPOS, which
922        returns MATCH_KETRPOS without going further in the pattern. By this means
923        we can handle the group by iteration rather than recursion, thereby
924        reducing the amount of stack needed. */
925    
926        case OP_CBRAPOS:
927        case OP_SCBRAPOS:
928        allow_zero = FALSE;
929    
930        POSSESSIVE_CAPTURE:
931        number = GET2(ecode, 1+LINK_SIZE);
932        offset = number << 1;
933    
934    #ifdef PCRE_DEBUG
935        printf("start possessive bracket %d\n", number);
936        printf("subject=");
937        pchars(eptr, 16, TRUE, md);
938        printf("\n");
939    #endif
940    
941        if (offset < md->offset_max)
942          {
943          matched_once = FALSE;
944          code_offset = ecode - md->start_code;
945    
946          save_offset1 = md->offset_vector[offset];
947          save_offset2 = md->offset_vector[offset+1];
948          save_offset3 = md->offset_vector[md->offset_end - number];
949          save_capture_last = md->capture_last;
950    
951          DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
952    
953          /* Each time round the loop, save the current subject position for use
954          when the group matches. For MATCH_MATCH, the group has matched, so we
955          restart it with a new subject starting position, remembering that we had
956          at least one match. For MATCH_NOMATCH, carry on with the alternatives, as
957          usual. If we haven't matched any alternatives in any iteration, check to
958          see if a previous iteration matched. If so, the group has matched;
959          continue from afterwards. Otherwise it has failed; restore the previous
960          capture values before returning NOMATCH. */
961    
962          for (;;)
963            {
964            md->offset_vector[md->offset_end - number] =
965              (int)(eptr - md->start_subject);
966            if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
967            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
968              eptrb, RM63);
969            if (rrc == MATCH_KETRPOS)
970              {
971              offset_top = md->end_offset_top;
972              eptr = md->end_match_ptr;
973              ecode = md->start_code + code_offset;
974              save_capture_last = md->capture_last;
975              matched_once = TRUE;
976              continue;
977              }
978            if (rrc != MATCH_NOMATCH &&
979                (rrc != MATCH_THEN || md->start_match_ptr != ecode))
980              RRETURN(rrc);
981            md->capture_last = save_capture_last;
982            ecode += GET(ecode, 1);
983            if (*ecode != OP_ALT) break;
984            }
985    
986          if (!matched_once)
987            {
988            md->offset_vector[offset] = save_offset1;
989            md->offset_vector[offset+1] = save_offset2;
990            md->offset_vector[md->offset_end - number] = save_offset3;
991            }
992    
993          if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr;
994          if (allow_zero || matched_once)
995          {          {
996          ecode += _pcre_OP_lengths[*ecode];          ecode += 1 + LINK_SIZE;
997          flags |= match_tail_recursed;          break;
         DPRINTF(("bracket 0 tail recursion\n"));  
         goto TAIL_RECURSE;  
998          }          }
999    
1000        /* For non-final alternatives, continue the loop for a NOMATCH result;        RRETURN(MATCH_NOMATCH);
1001        otherwise return. */        }
1002    
1003        /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
1004        as a non-capturing bracket. */
1005    
1006        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1007          eptrb, flags, RM2);      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1008        if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
1009        DPRINTF(("insufficient capture room: treat as non-capturing\n"));
1010    
1011        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1012        /* VVVVVVVVVVVVVVVVVVVVVVVVV */
1013    
1014        /* Non-capturing possessive bracket with unlimited repeat. We come here
1015        from BRAZERO with allow_zero = TRUE. The code is similar to the above,
1016        without the capturing complication. It is written out separately for speed
1017        and cleanliness. */
1018    
1019        case OP_BRAPOS:
1020        case OP_SBRAPOS:
1021        allow_zero = FALSE;
1022    
1023        POSSESSIVE_NON_CAPTURE:
1024        matched_once = FALSE;
1025        code_offset = ecode - md->start_code;
1026    
1027        for (;;)
1028          {
1029          if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1030          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
1031            eptrb, RM48);
1032          if (rrc == MATCH_KETRPOS)
1033            {
1034            offset_top = md->end_offset_top;
1035            eptr = md->end_match_ptr;
1036            ecode = md->start_code + code_offset;
1037            matched_once = TRUE;
1038            continue;
1039            }
1040          if (rrc != MATCH_NOMATCH &&
1041              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1042            RRETURN(rrc);
1043        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1044          if (*ecode != OP_ALT) break;
1045          }
1046    
1047        if (matched_once || allow_zero)
1048          {
1049          ecode += 1 + LINK_SIZE;
1050          break;
1051        }        }
1052        RRETURN(MATCH_NOMATCH);
1053    
1054      /* Control never reaches here. */      /* Control never reaches here. */
1055    
1056      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
1057      two branches. If the condition is false, skipping the first branch takes us      two branches. If the condition is false, skipping the first branch takes us
1058      past the end if there is only one branch, but that's OK because that is      past the end if there is only one branch, but that's OK because that is
1059      exactly what going to the ket would do. As there is only one branch to be      exactly what going to the ket would do. */
     obeyed, we can use tail recursion to avoid using another stack frame. */  
1060    
1061      case OP_COND:      case OP_COND:
1062      case OP_SCOND:      case OP_SCOND:
1063      if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */      codelink = GET(ecode, 1);
1064    
1065        /* Because of the way auto-callout works during compile, a callout item is
1066        inserted between OP_COND and an assertion condition. */
1067    
1068        if (ecode[LINK_SIZE+1] == OP_CALLOUT)
1069        {        {
1070        offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/        if (pcre_callout != NULL)
1071        condition = md->recursive != NULL &&          {
1072          (offset == RREF_ANY || offset == md->recursive->group_num);          pcre_callout_block cb;
1073        ecode += condition? 3 : GET(ecode, 1);          cb.version          = 2;   /* Version 1 of the callout block */
1074            cb.callout_number   = ecode[LINK_SIZE+2];
1075            cb.offset_vector    = md->offset_vector;
1076            cb.subject          = (PCRE_SPTR)md->start_subject;
1077            cb.subject_length   = (int)(md->end_subject - md->start_subject);
1078            cb.start_match      = (int)(mstart - md->start_subject);
1079            cb.current_position = (int)(eptr - md->start_subject);
1080            cb.pattern_position = GET(ecode, LINK_SIZE + 3);
1081            cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
1082            cb.capture_top      = offset_top/2;
1083            cb.capture_last     = md->capture_last;
1084            cb.callout_data     = md->callout_data;
1085            cb.mark             = markptr;
1086            if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1087            if (rrc < 0) RRETURN(rrc);
1088            }
1089          ecode += _pcre_OP_lengths[OP_CALLOUT];
1090          }
1091    
1092        condcode = ecode[LINK_SIZE+1];
1093    
1094        /* Now see what the actual condition is */
1095    
1096        if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
1097          {
1098          if (md->recursive == NULL)                /* Not recursing => FALSE */
1099            {
1100            condition = FALSE;
1101            ecode += GET(ecode, 1);
1102            }
1103          else
1104            {
1105            int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
1106            condition =  (recno == RREF_ANY || recno == md->recursive->group_num);
1107    
1108            /* If the test is for recursion into a specific subpattern, and it is
1109            false, but the test was set up by name, scan the table to see if the
1110            name refers to any other numbers, and test them. The condition is true
1111            if any one is set. */
1112    
1113            if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
1114              {
1115              uschar *slotA = md->name_table;
1116              for (i = 0; i < md->name_count; i++)
1117                {
1118                if (GET2(slotA, 0) == recno) break;
1119                slotA += md->name_entry_size;
1120                }
1121    
1122              /* Found a name for the number - there can be only one; duplicate
1123              names for different numbers are allowed, but not vice versa. First
1124              scan down for duplicates. */
1125    
1126              if (i < md->name_count)
1127                {
1128                uschar *slotB = slotA;
1129                while (slotB > md->name_table)
1130                  {
1131                  slotB -= md->name_entry_size;
1132                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1133                    {
1134                    condition = GET2(slotB, 0) == md->recursive->group_num;
1135                    if (condition) break;
1136                    }
1137                  else break;
1138                  }
1139    
1140                /* Scan up for duplicates */
1141    
1142                if (!condition)
1143                  {
1144                  slotB = slotA;
1145                  for (i++; i < md->name_count; i++)
1146                    {
1147                    slotB += md->name_entry_size;
1148                    if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1149                      {
1150                      condition = GET2(slotB, 0) == md->recursive->group_num;
1151                      if (condition) break;
1152                      }
1153                    else break;
1154                    }
1155                  }
1156                }
1157              }
1158    
1159            /* Chose branch according to the condition */
1160    
1161            ecode += condition? 3 : GET(ecode, 1);
1162            }
1163        }        }
1164    
1165      else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
1166        {        {
1167        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
1168        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1169    
1170          /* If the numbered capture is unset, but the reference was by name,
1171          scan the table to see if the name refers to any other numbers, and test
1172          them. The condition is true if any one is set. This is tediously similar
1173          to the code above, but not close enough to try to amalgamate. */
1174    
1175          if (!condition && condcode == OP_NCREF)
1176            {
1177            int refno = offset >> 1;
1178            uschar *slotA = md->name_table;
1179    
1180            for (i = 0; i < md->name_count; i++)
1181              {
1182              if (GET2(slotA, 0) == refno) break;
1183              slotA += md->name_entry_size;
1184              }
1185    
1186            /* Found a name for the number - there can be only one; duplicate names
1187            for different numbers are allowed, but not vice versa. First scan down
1188            for duplicates. */
1189    
1190            if (i < md->name_count)
1191              {
1192              uschar *slotB = slotA;
1193              while (slotB > md->name_table)
1194                {
1195                slotB -= md->name_entry_size;
1196                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1197                  {
1198                  offset = GET2(slotB, 0) << 1;
1199                  condition = offset < offset_top &&
1200                    md->offset_vector[offset] >= 0;
1201                  if (condition) break;
1202                  }
1203                else break;
1204                }
1205    
1206              /* Scan up for duplicates */
1207    
1208              if (!condition)
1209                {
1210                slotB = slotA;
1211                for (i++; i < md->name_count; i++)
1212                  {
1213                  slotB += md->name_entry_size;
1214                  if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
1215                    {
1216                    offset = GET2(slotB, 0) << 1;
1217                    condition = offset < offset_top &&
1218                      md->offset_vector[offset] >= 0;
1219                    if (condition) break;
1220                    }
1221                  else break;
1222                  }
1223                }
1224              }
1225            }
1226    
1227          /* Chose branch according to the condition */
1228    
1229        ecode += condition? 3 : GET(ecode, 1);        ecode += condition? 3 : GET(ecode, 1);
1230        }        }
1231    
1232      else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */      else if (condcode == OP_DEF)     /* DEFINE - always false */
1233        {        {
1234        condition = FALSE;        condition = FALSE;
1235        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1236        }        }
1237    
1238      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
1239      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
1240      assertion. */      an assertion. */
1241    
1242      else      else
1243        {        {
1244        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        md->match_function_type = MATCH_CONDASSERT;
1245            match_condassert, RM3);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);
1246        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1247          {          {
1248            if (md->end_offset_top > offset_top)
1249              offset_top = md->end_offset_top;  /* Captures may have happened */
1250          condition = TRUE;          condition = TRUE;
1251          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
1252          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1253          }          }
1254        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH &&
1255                  (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1256          {          {
1257          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1258          }          }
1259        else        else
1260          {          {
1261          condition = FALSE;          condition = FALSE;
1262          ecode += GET(ecode, 1);          ecode += codelink;
1263          }          }
1264        }        }
1265    
1266      /* 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,
1267      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
1268      alternative doesn't exist, we can just plough on. */      when there was unlimited repeat of a possibly empty group. However, that
1269        strategy no longer works because of the possibilty of (*THEN) being
1270        encountered in the branch. A recursive call to match() is always required,
1271        unless the second alternative doesn't exist, in which case we can just
1272        plough on. */
1273    
1274      if (condition || *ecode == OP_ALT)      if (condition || *ecode == OP_ALT)
1275        {        {
1276        ecode += 1 + LINK_SIZE;        if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP;
1277        flags = match_tail_recursed | ((op == OP_SCOND)? match_cbegroup : 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);
1278        goto TAIL_RECURSE;  
1279          /* If the result is THEN from within the "true" branch of the condition,
1280          md->start_match_ptr will point to the original OP_COND, not to the start
1281          of the branch, so we have do work to see if it matches. If THEN comes
1282          from the "false" branch, md->start_match_ptr does point to OP_ALT. */
1283    
1284          if (rrc == MATCH_THEN)
1285            {
1286            if (*ecode != OP_ALT)
1287              {
1288              do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
1289              ecode -= GET(ecode, 1);
1290              }
1291            if (md->start_match_ptr == ecode) rrc = MATCH_NOMATCH;
1292            }
1293          RRETURN(rrc);
1294        }        }
1295      else      else                         /* Condition false & no alternative */
1296        {        {
1297        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1298        }        }
1299      break;      break;
1300    
1301    
1302      /* 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,
1303      restore the offsets appropriately and continue from after the call. */      to close any currently open capturing brackets. */
1304    
1305      case OP_END:      case OP_CLOSE:
1306      if (md->recursive != NULL && md->recursive->group_num == 0)      number = GET2(ecode, 1);
1307        offset = number << 1;
1308    
1309    #ifdef PCRE_DEBUG
1310          printf("end bracket %d at *ACCEPT", number);
1311          printf("\n");
1312    #endif
1313    
1314        md->capture_last = number;
1315        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1316        {        {
1317        recursion_info *rec = md->recursive;        md->offset_vector[offset] =
1318        DPRINTF(("End of pattern in a (?0) recursion\n"));          md->offset_vector[md->offset_end - number];
1319        md->recursive = rec->prevrec;        md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1320        memmove(md->offset_vector, rec->offset_save,        if (offset_top <= offset) offset_top = offset + 2;
         rec->saved_max * sizeof(int));  
       mstart = rec->save_start;  
       ims = original_ims;  
       ecode = rec->after_call;  
       break;  
1321        }        }
1322        ecode += 3;
1323        break;
1324    
1325    
1326        /* End of the pattern, either real or forced. */
1327    
1328        case OP_END:
1329        case OP_ACCEPT:
1330        case OP_ASSERT_ACCEPT:
1331    
1332        /* If we have matched an empty string, fail if not in an assertion and not
1333        in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART
1334        is set and we have matched at the start of the subject. In both cases,
1335        backtracking will then try other alternatives, if any. */
1336    
1337        if (eptr == mstart && op != OP_ASSERT_ACCEPT &&
1338             md->recursive == NULL &&
1339             (md->notempty ||
1340               (md->notempty_atstart &&
1341                 mstart == md->start_subject + md->start_offset)))
1342          MRRETURN(MATCH_NOMATCH);
1343    
1344      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty      /* Otherwise, we have a match. */
     string - backtracking will then try other alternatives, if any. */  
1345    
     if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH);  
1346      md->end_match_ptr = eptr;           /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
1347      md->end_offset_top = offset_top;    /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
1348      md->start_match_ptr = mstart;  /* and the start (\K can modify) */      md->start_match_ptr = mstart;       /* and the start (\K can modify) */
     RRETURN(MATCH_MATCH);  
1349    
1350      /* Change option settings */      /* For some reason, the macros don't work properly if an expression is
1351        given as the argument to MRRETURN when the heap is in use. */
1352    
1353      case OP_OPT:      rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT;
1354      ims = ecode[1];      MRRETURN(rrc);
     ecode += 2;  
     DPRINTF(("ims set to %02lx\n", ims));  
     break;  
1355    
1356      /* Assertion brackets. Check the alternative branches in turn - the      /* Assertion brackets. Check the alternative branches in turn - the
1357      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,
1358      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
1359      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
1360      this level is identical to the lookahead case. */      this level is identical to the lookahead case. When the assertion is part
1361        of a condition, we want to return immediately afterwards. The caller of
1362        this incarnation of the match() function will have set MATCH_CONDASSERT in
1363        md->match_function type, and one of these opcodes will be the first opcode
1364        that is processed. We use a local variable that is preserved over calls to
1365        match() to remember this case. */
1366    
1367      case OP_ASSERT:      case OP_ASSERT:
1368      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1369        if (md->match_function_type == MATCH_CONDASSERT)
1370          {
1371          condassert = TRUE;
1372          md->match_function_type = 0;
1373          }
1374        else condassert = FALSE;
1375    
1376      do      do
1377        {        {
1378        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1379          RM4);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1380        if (rrc == MATCH_MATCH) break;          {
1381        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          mstart = md->start_match_ptr;   /* In case \K reset it */
1382            markptr = md->mark;
1383            break;
1384            }
1385          if (rrc != MATCH_NOMATCH &&
1386              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1387            RRETURN(rrc);
1388        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1389        }        }
1390      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1391      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);  
1392        if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH);
1393    
1394      /* If checking an assertion for a condition, return MATCH_MATCH. */      /* If checking an assertion for a condition, return MATCH_MATCH. */
1395    
1396      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1397    
1398      /* Continue from after the assertion, updating the offsets high water      /* Continue from after the assertion, updating the offsets high water
1399      mark, since extracts may have been taken during the assertion. */      mark, since extracts may have been taken during the assertion. */
# Line 847  for (;;) Line 1403  for (;;)
1403      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1404      continue;      continue;
1405    
1406      /* Negative assertion: all branches must fail to match */      /* Negative assertion: all branches must fail to match. Encountering SKIP,
1407        PRUNE, or COMMIT means we must assume failure without checking subsequent
1408        branches. */
1409    
1410      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1411      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1412        if (md->match_function_type == MATCH_CONDASSERT)
1413          {
1414          condassert = TRUE;
1415          md->match_function_type = 0;
1416          }
1417        else condassert = FALSE;
1418    
1419      do      do
1420        {        {
1421        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1422          RM5);        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH);
1423        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)
1424        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          {
1425            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1426            break;
1427            }
1428          if (rrc != MATCH_NOMATCH &&
1429              (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1430            RRETURN(rrc);
1431        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1432        }        }
1433      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1434    
1435      if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1436    
1437      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1438      continue;      continue;
# Line 879  for (;;) Line 1450  for (;;)
1450        while (i-- > 0)        while (i-- > 0)
1451          {          {
1452          eptr--;          eptr--;
1453          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1454          BACKCHAR(eptr)          BACKCHAR(eptr);
1455          }          }
1456        }        }
1457      else      else
# Line 890  for (;;) Line 1461  for (;;)
1461    
1462        {        {
1463        eptr -= GET(ecode, 1);        eptr -= GET(ecode, 1);
1464        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH);
1465        }        }
1466    
1467      /* Skip to next op code */      /* Save the earliest consulted character, then skip to next op code */
1468    
1469        if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
1470      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1471      break;      break;
1472    
# Line 906  for (;;) Line 1478  for (;;)
1478      if (pcre_callout != NULL)      if (pcre_callout != NULL)
1479        {        {
1480        pcre_callout_block cb;        pcre_callout_block cb;
1481        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 2;   /* Version 1 of the callout block */
1482        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
1483        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
1484        cb.subject          = (PCRE_SPTR)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
1485        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = (int)(md->end_subject - md->start_subject);
1486        cb.start_match      = mstart - md->start_subject;        cb.start_match      = (int)(mstart - md->start_subject);
1487        cb.current_position = eptr - md->start_subject;        cb.current_position = (int)(eptr - md->start_subject);
1488        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1489        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1490        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1491        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last;
1492        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1493        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);        cb.mark             = markptr;
1494          if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH);
1495        if (rrc < 0) RRETURN(rrc);        if (rrc < 0) RRETURN(rrc);
1496        }        }
1497      ecode += 2 + 2*LINK_SIZE;      ecode += 2 + 2*LINK_SIZE;
# Line 928  for (;;) Line 1501  for (;;)
1501      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
1502      whole pattern. (This is so that it works from duplicated subpatterns.)      whole pattern. (This is so that it works from duplicated subpatterns.)
1503    
1504      If there are any capturing brackets started but not finished, we have to      The state of the capturing groups is preserved over recursion, and
1505      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
1506      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
1507      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
1508      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
1509      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
1510      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.  
1511    
1512      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
1513      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
1514      for the original version of this logic. */      for the original version of this logic. It has, however, been hacked around
1515        a lot, so he is not to blame for the current way it works. */
1516    
1517      case OP_RECURSE:      case OP_RECURSE:
1518        {        {
1519          recursion_info *ri;
1520          int recno;
1521    
1522        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1523        new_recursive.group_num = (callpat == md->start_code)? 0 :        recno = (callpat == md->start_code)? 0 :
1524          GET2(callpat, 1 + LINK_SIZE);          GET2(callpat, 1 + LINK_SIZE);
1525    
1526          /* Check for repeating a recursion without advancing the subject pointer.
1527          This should catch convoluted mutual recursions. (Some simple cases are
1528          caught at compile time.) */
1529    
1530          for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
1531            if (recno == ri->group_num && eptr == ri->subject_position)
1532              RRETURN(PCRE_ERROR_RECURSELOOP);
1533    
1534        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1535    
1536          new_recursive.group_num = recno;
1537          new_recursive.subject_position = eptr;
1538        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1539        md->recursive = &new_recursive;        md->recursive = &new_recursive;
1540    
1541        /* Find where to continue from afterwards */        /* Where to continue from afterwards */
1542    
1543        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       new_recursive.after_call = ecode;  
1544    
1545        /* Now save the offset data. */        /* Now save the offset data */
1546    
1547        new_recursive.saved_max = md->offset_end;        new_recursive.saved_max = md->offset_end;
1548        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)        if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
# Line 970  for (;;) Line 1553  for (;;)
1553            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));            (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
1554          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);          if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
1555          }          }
   
1556        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1557              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
       new_recursive.save_start = mstart;  
       mstart = eptr;  
1558    
1559        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. After processing each alternative,
1560        restore the offset and recursion data. */        restore the offset data. If there were nested recursions, md->recursive
1561          might be changed, so reset it before looping. */
1562    
1563        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1564        flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;        cbegroup = (*callpat >= OP_SBRA);
1565        do        do
1566          {          {
1567            if (cbegroup) md->match_function_type = MATCH_CBEGROUP;
1568          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1569            md, ims, eptrb, flags, RM6);            md, eptrb, RM6);
1570          if (rrc == MATCH_MATCH)          memcpy(md->offset_vector, new_recursive.offset_save,
1571                new_recursive.saved_max * sizeof(int));
1572            md->recursive = new_recursive.prevrec;
1573            if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1574            {            {
1575            DPRINTF(("Recursion matched\n"));            DPRINTF(("Recursion matched\n"));
           md->recursive = new_recursive.prevrec;  
1576            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1577              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1578            RRETURN(MATCH_MATCH);  
1579              /* Set where we got to in the subject, and reset the start in case
1580              it was changed by \K. This *is* propagated back out of a recursion,
1581              for Perl compatibility. */
1582    
1583              eptr = md->end_match_ptr;
1584              mstart = md->start_match_ptr;
1585              goto RECURSION_MATCHED;        /* Exit loop; end processing */
1586            }            }
1587          else if (rrc != MATCH_NOMATCH)          else if (rrc != MATCH_NOMATCH &&
1588                    (rrc != MATCH_THEN || md->start_match_ptr != ecode))
1589            {            {
1590            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1591              if (new_recursive.offset_save != stacksave)
1592                (pcre_free)(new_recursive.offset_save);
1593            RRETURN(rrc);            RRETURN(rrc);
1594            }            }
1595    
1596          md->recursive = &new_recursive;          md->recursive = &new_recursive;
         memcpy(md->offset_vector, new_recursive.offset_save,  
             new_recursive.saved_max * sizeof(int));  
1597          callpat += GET(callpat, 1);          callpat += GET(callpat, 1);
1598          }          }
1599        while (*callpat == OP_ALT);        while (*callpat == OP_ALT);
# Line 1010  for (;;) Line 1602  for (;;)
1602        md->recursive = new_recursive.prevrec;        md->recursive = new_recursive.prevrec;
1603        if (new_recursive.offset_save != stacksave)        if (new_recursive.offset_save != stacksave)
1604          (pcre_free)(new_recursive.offset_save);          (pcre_free)(new_recursive.offset_save);
1605        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1606        }        }
     /* Control never reaches here */  
1607    
1608      /* "Once" brackets are like assertion brackets except that after a match,      RECURSION_MATCHED:
1609      the point in the subject string is not moved back. Thus there can never be      break;
     a move back into the brackets. Friedl calls these "atomic" subpatterns.  
     Check the alternative branches in turn - the matching won't pass the KET  
     for this kind of subpattern. If any one branch matches, we carry on as at  
     the end of a normal bracket, leaving the subject pointer. */  
1610    
1611      case OP_ONCE:      /* An alternation is the end of a branch; scan along to find the end of the
1612      prev = ecode;      bracketed group and go to there. */
     saved_eptr = eptr;  
1613    
1614      do      case OP_ALT:
1615        {      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1616        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,      break;
         eptrb, 0, RM7);  
       if (rrc == MATCH_MATCH) break;  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += GET(ecode,1);  
       }  
     while (*ecode == OP_ALT);  
1617    
1618      /* If hit the end of the group (which could be repeated), fail */      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1619        indicating that it may occur zero times. It may repeat infinitely, or not
1620        at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1621        with fixed upper repeat limits are compiled as a number of copies, with the
1622        optional ones preceded by BRAZERO or BRAMINZERO. */
1623    
1624      if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);      case OP_BRAZERO:
1625        next = ecode + 1;
1626        RMATCH(eptr, next, offset_top, md, eptrb, RM10);
1627        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1628        do next += GET(next, 1); while (*next == OP_ALT);
1629        ecode = next + 1 + LINK_SIZE;
1630        break;
1631    
1632      /* Continue as from after the assertion, updating the offsets high water      case OP_BRAMINZERO:
1633      mark, since extracts may have been taken. */      next = ecode + 1;
1634        do next += GET(next, 1); while (*next == OP_ALT);
1635        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11);
1636        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1637        ecode++;
1638        break;
1639    
1640      do ecode += GET(ecode, 1); while (*ecode == OP_ALT);      case OP_SKIPZERO:
1641        next = ecode+1;
1642        do next += GET(next,1); while (*next == OP_ALT);
1643        ecode = next + 1 + LINK_SIZE;
1644        break;
1645    
1646      offset_top = md->end_offset_top;      /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything
1647      eptr = md->end_match_ptr;      here; just jump to the group, with allow_zero set TRUE. */
1648    
1649      /* For a non-repeating ket, just continue at this level. This also      case OP_BRAPOSZERO:
1650      happens for a repeating ket if no characters were matched in the group.      op = *(++ecode);
1651      This is the forcible breaking of infinite loops as implemented in Perl      allow_zero = TRUE;
1652      5.005. If there is an options reset, it will get obeyed in the normal      if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE;
1653      course of events. */        goto POSSESSIVE_NON_CAPTURE;
1654    
1655      if (*ecode == OP_KET || eptr == saved_eptr)      /* End of a group, repeated or non-repeating. */
       {  
       ecode += 1+LINK_SIZE;  
       break;  
       }  
1656    
1657      /* The repeating kets try the rest of the pattern or restart from the      case OP_KET:
1658      preceding bracket, in the appropriate order. The second "call" of match()      case OP_KETRMIN:
1659      uses tail recursion, to avoid using another stack frame. We need to reset      case OP_KETRMAX:
1660      any options that changed within the bracket before re-running it, so      case OP_KETRPOS:
1661      check the next opcode. */      prev = ecode - GET(ecode, 1);
   
     if (ecode[1+LINK_SIZE] == OP_OPT)  
       {  
       ims = (ims & ~PCRE_IMS) | ecode[4];  
       DPRINTF(("ims set to %02lx at group repeat\n", ims));  
       }  
   
     if (*ecode == OP_KETRMIN)  
       {  
       RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0,  
         RM8);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode = prev;  
       flags = match_tail_recursed;  
       goto TAIL_RECURSE;  
       }  
     else  /* OP_KETRMAX */  
       {  
       RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);  
       if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
       ecode += 1 + LINK_SIZE;  
       flags = 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);  
1662    
1663      /* If this was a group that remembered the subject start, in order to break      /* If this was a group that remembered the subject start, in order to break
1664      infinite repeats of empty string matches, retrieve the subject start from      infinite repeats of empty string matches, retrieve the subject start from
1665      the chain. Otherwise, set it NULL. */      the chain. Otherwise, set it NULL. */
1666    
1667      if (*prev >= OP_SBRA)      if (*prev >= OP_SBRA || *prev == OP_ONCE)
1668        {        {
1669        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1670        eptrb = eptrb->epb_prev;              /* Backup to previous group */        eptrb = eptrb->epb_prev;              /* Backup to previous group */
# Line 1143  for (;;) Line 1673  for (;;)
1673    
1674      /* If we are at the end of an assertion group, stop matching and return      /* If we are at the end of an assertion group, stop matching and return
1675      MATCH_MATCH, but record the current high water mark for use by positive      MATCH_MATCH, but record the current high water mark for use by positive
1676      assertions. Do this also for the "once" (atomic) groups. */      assertions. We also need to record the match start in case it was changed
1677        by \K. */
1678    
1679      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1680          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT)
         *prev == OP_ONCE)  
1681        {        {
1682        md->end_match_ptr = eptr;      /* For ONCE */        md->end_match_ptr = eptr;      /* For ONCE */
1683        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
1684        RRETURN(MATCH_MATCH);        md->start_match_ptr = mstart;
1685          MRRETURN(MATCH_MATCH);         /* Sets md->mark */
1686        }        }
1687    
1688      /* For capturing groups we have to check the group number back at the start      /* For capturing groups we have to check the group number back at the start
1689      and if necessary complete handling an extraction by setting the offsets and      and if necessary complete handling an extraction by setting the offsets and
1690      bumping the high water mark. Note that whole-pattern recursion is coded as      bumping the high water mark. Whole-pattern recursion is coded as a recurse
1691      a recurse into group 0, so it won't be picked up here. Instead, we catch it      into group 0, so it won't be picked up here. Instead, we catch it when the
1692      when the OP_END is reached. Other recursion is handled here. */      OP_END is reached. Other recursion is handled here. We just have to record
1693        the current subject position and start match pointer and give a MATCH
1694        return. */
1695    
1696      if (*prev == OP_CBRA || *prev == OP_SCBRA)      if (*prev == OP_CBRA || *prev == OP_SCBRA ||
1697            *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS)
1698        {        {
1699        number = GET2(prev, 1+LINK_SIZE);        number = GET2(prev, 1+LINK_SIZE);
1700        offset = number << 1;        offset = number << 1;
1701    
1702  #ifdef DEBUG  #ifdef PCRE_DEBUG
1703        printf("end bracket %d", number);        printf("end bracket %d", number);
1704        printf("\n");        printf("\n");
1705  #endif  #endif
1706    
1707          /* Handle a recursively called group. */
1708    
1709          if (md->recursive != NULL && md->recursive->group_num == number)
1710            {
1711            md->end_match_ptr = eptr;
1712            md->start_match_ptr = mstart;
1713            RRETURN(MATCH_MATCH);
1714            }
1715    
1716          /* Deal with capturing */
1717    
1718        md->capture_last = number;        md->capture_last = number;
1719        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1720          {          {
1721            /* If offset is greater than offset_top, it means that we are
1722            "skipping" a capturing group, and that group's offsets must be marked
1723            unset. In earlier versions of PCRE, all the offsets were unset at the
1724            start of matching, but this doesn't work because atomic groups and
1725            assertions can cause a value to be set that should later be unset.
1726            Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as
1727            part of the atomic group, but this is not on the final matching path,
1728            so must be unset when 2 is set. (If there is no group 2, there is no
1729            problem, because offset_top will then be 2, indicating no capture.) */
1730    
1731            if (offset > offset_top)
1732              {
1733              register int *iptr = md->offset_vector + offset_top;
1734              register int *iend = md->offset_vector + offset;
1735              while (iptr < iend) *iptr++ = -1;
1736              }
1737    
1738            /* Now make the extraction */
1739    
1740          md->offset_vector[offset] =          md->offset_vector[offset] =
1741            md->offset_vector[md->offset_end - number];            md->offset_vector[md->offset_end - number];
1742          md->offset_vector[offset+1] = eptr - md->start_subject;          md->offset_vector[offset+1] = (int)(eptr - md->start_subject);
1743          if (offset_top <= offset) offset_top = offset + 2;          if (offset_top <= offset) offset_top = offset + 2;
1744          }          }
1745          }
1746    
1747        /* Handle a recursively called group. Restore the offsets      /* For an ordinary non-repeating ket, just continue at this level. This
1748        appropriately and continue from after the call. */      also happens for a repeating ket if no characters were matched in the
1749        group. This is the forcible breaking of infinite loops as implemented in
1750        Perl 5.005. For a non-repeating atomic group, establish a backup point by
1751        processing the rest of the pattern at a lower level. If this results in a
1752        NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby
1753        bypassing intermediate backup points, but resetting any captures that
1754        happened along the way. */
1755    
1756        if (md->recursive != NULL && md->recursive->group_num == number)      if (*ecode == OP_KET || eptr == saved_eptr)
1757          {
1758          if (*prev == OP_ONCE)
1759          {          {
1760          recursion_info *rec = md->recursive;          RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12);
1761          DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1762          md->recursive = rec->prevrec;          md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1763          mstart = rec->save_start;          RRETURN(MATCH_ONCE);
         memcpy(md->offset_vector, rec->offset_save,  
           rec->saved_max * sizeof(int));  
         ecode = rec->after_call;  
         ims = original_ims;  
         break;  
1764          }          }
1765          ecode += 1 + LINK_SIZE;    /* Carry on at this level */
1766          break;
1767        }        }
1768    
1769      /* For both capturing and non-capturing groups, reset the value of the ims      /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
1770      flags, in case they got changed during the group. */      and return the MATCH_KETRPOS. This makes it possible to do the repeats one
1771        at a time from the outer level, thus saving stack. */
     ims = original_ims;  
     DPRINTF(("ims reset to %02lx\n", ims));  
1772    
1773      /* For a non-repeating ket, just continue at this level. This also      if (*ecode == OP_KETRPOS)
     happens for a repeating ket if no characters were matched in the group.  
     This is the forcible breaking of infinite loops as implemented in Perl  
     5.005. If there is an options reset, it will get obeyed in the normal  
     course of events. */  
   
     if (*ecode == OP_KET || eptr == saved_eptr)  
1774        {        {
1775        ecode += 1 + LINK_SIZE;        md->end_match_ptr = eptr;
1776        break;        md->end_offset_top = offset_top;
1777          RRETURN(MATCH_KETRPOS);
1778        }        }
1779    
1780      /* The repeating kets try the rest of the pattern or restart from the      /* The normal repeating kets try the rest of the pattern or restart from
1781      preceding bracket, in the appropriate order. In the second case, we can use      the preceding bracket, in the appropriate order. In the second case, we can
1782      tail recursion to avoid using another stack frame. */      use tail recursion to avoid using another stack frame, unless we have an
1783        an atomic group or an unlimited repeat of a group that can match an empty
1784      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;      string. */
1785    
1786      if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1787        {        {
1788        RMATCH(eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7);
         RM12);  
1789        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1790          if (*prev == OP_ONCE)
1791            {
1792            RMATCH(eptr, prev, offset_top, md, eptrb, RM8);
1793            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1794            md->once_target = prev;  /* Level at which to change to MATCH_NOMATCH */
1795            RRETURN(MATCH_ONCE);
1796            }
1797          if (*prev >= OP_SBRA)    /* Could match an empty string */
1798            {
1799            md->match_function_type = MATCH_CBEGROUP;
1800            RMATCH(eptr, prev, offset_top, md, eptrb, RM50);
1801            RRETURN(rrc);
1802            }
1803        ecode = prev;        ecode = prev;
       flags |= match_tail_recursed;  
1804        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1805        }        }
1806      else  /* OP_KETRMAX */      else  /* OP_KETRMAX */
1807        {        {
1808        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);        if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP;
1809          RMATCH(eptr, prev, offset_top, md, eptrb, RM13);
1810          if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH;
1811        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1812          if (*prev == OP_ONCE)
1813            {
1814            RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9);
1815            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1816            md->once_target = prev;
1817            RRETURN(MATCH_ONCE);
1818            }
1819        ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
       flags = match_tail_recursed;  
1820        goto TAIL_RECURSE;        goto TAIL_RECURSE;
1821        }        }
1822      /* Control never gets here */      /* Control never gets here */
1823    
1824      /* Start of subject unless notbol, or after internal newline if multiline */      /* Not multiline mode: start of subject assertion, unless notbol. */
1825    
1826      case OP_CIRC:      case OP_CIRC:
1827      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
     if ((ims & PCRE_MULTILINE) != 0)  
       {  
       if (eptr != md->start_subject &&  
           (eptr == md->end_subject || !WAS_NEWLINE(eptr)))  
         RRETURN(MATCH_NOMATCH);  
       ecode++;  
       break;  
       }  
     /* ... else fall through */  
1828    
1829      /* Start of subject assertion */      /* Start of subject assertion */
1830    
1831      case OP_SOD:      case OP_SOD:
1832      if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH);
1833        ecode++;
1834        break;
1835    
1836        /* Multiline mode: start of subject unless notbol, or after any newline. */
1837    
1838        case OP_CIRCM:
1839        if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH);
1840        if (eptr != md->start_subject &&
1841            (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1842          MRRETURN(MATCH_NOMATCH);
1843      ecode++;      ecode++;
1844      break;      break;
1845    
1846      /* Start of match assertion */      /* Start of match assertion */
1847    
1848      case OP_SOM:      case OP_SOM:
1849      if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);      if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH);
1850      ecode++;      ecode++;
1851      break;      break;
1852    
# Line 1274  for (;;) Line 1857  for (;;)
1857      ecode++;      ecode++;
1858      break;      break;
1859    
1860      /* Assert before internal newline if multiline, or before a terminating      /* Multiline mode: assert before any newline, or before end of subject
1861      newline unless endonly is set, else end of subject unless noteol is set. */      unless noteol is set. */
1862    
1863      case OP_DOLL:      case OP_DOLLM:
1864      if ((ims & PCRE_MULTILINE) != 0)      if (eptr < md->end_subject)
1865        {        { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
       if (eptr < md->end_subject)  
         { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }  
       else  
         { if (md->noteol) RRETURN(MATCH_NOMATCH); }  
       ecode++;  
       break;  
       }  
1866      else      else
1867        {        {
1868        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1869        if (!md->endonly)        SCHECK_PARTIAL();
         {  
         if (eptr != md->end_subject &&  
             (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))  
           RRETURN(MATCH_NOMATCH);  
         ecode++;  
         break;  
         }  
1870        }        }
1871        ecode++;
1872        break;
1873    
1874        /* Not multiline mode: assert before a terminating newline or before end of
1875        subject unless noteol is set. */
1876    
1877        case OP_DOLL:
1878        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1879        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1880    
1881      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1882    
1883      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1884    
1885      case OP_EOD:      case OP_EOD:
1886      if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH);
1887        SCHECK_PARTIAL();
1888      ecode++;      ecode++;
1889      break;      break;
1890    
1891      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1892    
1893      case OP_EODN:      case OP_EODN:
1894      if (eptr != md->end_subject &&      ASSERT_NL_OR_EOS:
1895        if (eptr < md->end_subject &&
1896          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1897        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1898    
1899        /* Either at end of string or \n before end. */
1900    
1901        SCHECK_PARTIAL();
1902      ecode++;      ecode++;
1903      break;      break;
1904    
# Line 1325  for (;;) Line 1910  for (;;)
1910    
1911        /* Find out if the previous and current characters are "word" characters.        /* Find out if the previous and current characters are "word" characters.
1912        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
1913        be "non-word" characters. */        be "non-word" characters. Remember the earliest consulted character for
1914          partial matching. */
1915    
1916  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1917        if (utf8)        if (utf8)
1918          {          {
1919            /* Get status of previous character */
1920    
1921          if (eptr == md->start_subject) prev_is_word = FALSE; else          if (eptr == md->start_subject) prev_is_word = FALSE; else
1922            {            {
1923            const uschar *lastptr = eptr - 1;            USPTR lastptr = eptr - 1;
1924            while((*lastptr & 0xc0) == 0x80) lastptr--;            while((*lastptr & 0xc0) == 0x80) lastptr--;
1925              if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
1926            GETCHAR(c, lastptr);            GETCHAR(c, lastptr);
1927    #ifdef SUPPORT_UCP
1928              if (md->use_ucp)
1929                {
1930                if (c == '_') prev_is_word = TRUE; else
1931                  {
1932                  int cat = UCD_CATEGORY(c);
1933                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1934                  }
1935                }
1936              else
1937    #endif
1938            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1939            }            }
1940          if (eptr >= md->end_subject) cur_is_word = FALSE; else  
1941            /* Get status of next character */
1942    
1943            if (eptr >= md->end_subject)
1944              {
1945              SCHECK_PARTIAL();
1946              cur_is_word = FALSE;
1947              }
1948            else
1949            {            {
1950            GETCHAR(c, eptr);            GETCHAR(c, eptr);
1951    #ifdef SUPPORT_UCP
1952              if (md->use_ucp)
1953                {
1954                if (c == '_') cur_is_word = TRUE; else
1955                  {
1956                  int cat = UCD_CATEGORY(c);
1957                  cur_is_word = (cat == ucp_L || cat == ucp_N);
1958                  }
1959                }
1960              else
1961    #endif
1962            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;            cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
1963            }            }
1964          }          }
1965        else        else
1966  #endif  #endif
1967    
1968        /* More streamlined when not in UTF-8 mode */        /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for
1969          consistency with the behaviour of \w we do use it in this case. */
1970    
1971          {          {
1972          prev_is_word = (eptr != md->start_subject) &&          /* Get status of previous character */
1973            ((md->ctypes[eptr[-1]] & ctype_word) != 0);  
1974          cur_is_word = (eptr < md->end_subject) &&          if (eptr == md->start_subject) prev_is_word = FALSE; else
1975            ((md->ctypes[*eptr] & ctype_word) != 0);            {
1976              if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
1977    #ifdef SUPPORT_UCP
1978              if (md->use_ucp)
1979                {
1980                c = eptr[-1];
1981                if (c == '_') prev_is_word = TRUE; else
1982                  {
1983                  int cat = UCD_CATEGORY(c);
1984                  prev_is_word = (cat == ucp_L || cat == ucp_N);
1985                  }
1986                }
1987              else
1988    #endif
1989              prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
1990              }
1991    
1992            /* Get status of next character */
1993    
1994            if (eptr >= md->end_subject)
1995              {
1996              SCHECK_PARTIAL();
1997              cur_is_word = FALSE;
1998              }
1999            else
2000    #ifdef SUPPORT_UCP
2001            if (md->use_ucp)
2002              {
2003              c = *eptr;
2004              if (c == '_') cur_is_word = TRUE; else
2005                {
2006                int cat = UCD_CATEGORY(c);
2007                cur_is_word = (cat == ucp_L || cat == ucp_N);
2008                }
2009              }
2010            else
2011    #endif
2012            cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
2013          }          }
2014    
2015        /* Now see if the situation is what we want */        /* Now see if the situation is what we want */
2016    
2017        if ((*ecode++ == OP_WORD_BOUNDARY)?        if ((*ecode++ == OP_WORD_BOUNDARY)?
2018             cur_is_word == prev_is_word : cur_is_word != prev_is_word)             cur_is_word == prev_is_word : cur_is_word != prev_is_word)
2019          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2020        }        }
2021      break;      break;
2022    
2023      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
2024    
2025      case OP_ANY:      case OP_ANY:
2026      if ((ims & PCRE_DOTALL) == 0)      if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH);
2027        {      /* Fall through */
2028        if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);  
2029        case OP_ALLANY:
2030        if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2031          {                            /* not be updated before SCHECK_PARTIAL. */
2032          SCHECK_PARTIAL();
2033          MRRETURN(MATCH_NOMATCH);
2034        }        }
2035      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      eptr++;
2036      if (utf8)      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
2037      ecode++;      ecode++;
2038      break;      break;
2039    
# Line 1380  for (;;) Line 2041  for (;;)
2041      any byte, even newline, independent of the setting of PCRE_DOTALL. */      any byte, even newline, independent of the setting of PCRE_DOTALL. */
2042    
2043      case OP_ANYBYTE:      case OP_ANYBYTE:
2044      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)   /* DO NOT merge the eptr++ here; it must */
2045          {                            /* not be updated before SCHECK_PARTIAL. */
2046          SCHECK_PARTIAL();
2047          MRRETURN(MATCH_NOMATCH);
2048          }
2049        eptr++;
2050      ecode++;      ecode++;
2051      break;      break;
2052    
2053      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2054      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2055          {
2056          SCHECK_PARTIAL();
2057          MRRETURN(MATCH_NOMATCH);
2058          }
2059      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2060      if (      if (
2061  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1393  for (;;) Line 2063  for (;;)
2063  #endif  #endif
2064         (md->ctypes[c] & ctype_digit) != 0         (md->ctypes[c] & ctype_digit) != 0
2065         )         )
2066        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2067      ecode++;      ecode++;
2068      break;      break;
2069    
2070      case OP_DIGIT:      case OP_DIGIT:
2071      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2072          {
2073          SCHECK_PARTIAL();
2074          MRRETURN(MATCH_NOMATCH);
2075          }
2076      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2077      if (      if (
2078  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1406  for (;;) Line 2080  for (;;)
2080  #endif  #endif
2081         (md->ctypes[c] & ctype_digit) == 0         (md->ctypes[c] & ctype_digit) == 0
2082         )         )
2083        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2084      ecode++;      ecode++;
2085      break;      break;
2086    
2087      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2088      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2089          {
2090          SCHECK_PARTIAL();
2091          MRRETURN(MATCH_NOMATCH);
2092          }
2093      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2094      if (      if (
2095  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1419  for (;;) Line 2097  for (;;)
2097  #endif  #endif
2098         (md->ctypes[c] & ctype_space) != 0         (md->ctypes[c] & ctype_space) != 0
2099         )         )
2100        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2101      ecode++;      ecode++;
2102      break;      break;
2103    
2104      case OP_WHITESPACE:      case OP_WHITESPACE:
2105      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2106          {
2107          SCHECK_PARTIAL();
2108          MRRETURN(MATCH_NOMATCH);
2109          }
2110      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2111      if (      if (
2112  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1432  for (;;) Line 2114  for (;;)
2114  #endif  #endif
2115         (md->ctypes[c] & ctype_space) == 0         (md->ctypes[c] & ctype_space) == 0
2116         )         )
2117        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2118      ecode++;      ecode++;
2119      break;      break;
2120    
2121      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2122      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2123          {
2124          SCHECK_PARTIAL();
2125          MRRETURN(MATCH_NOMATCH);
2126          }
2127      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2128      if (      if (
2129  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1445  for (;;) Line 2131  for (;;)
2131  #endif  #endif
2132         (md->ctypes[c] & ctype_word) != 0         (md->ctypes[c] & ctype_word) != 0
2133         )         )
2134        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2135      ecode++;      ecode++;
2136      break;      break;
2137    
2138      case OP_WORDCHAR:      case OP_WORDCHAR:
2139      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2140          {
2141          SCHECK_PARTIAL();
2142          MRRETURN(MATCH_NOMATCH);
2143          }
2144      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2145      if (      if (
2146  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1458  for (;;) Line 2148  for (;;)
2148  #endif  #endif
2149         (md->ctypes[c] & ctype_word) == 0         (md->ctypes[c] & ctype_word) == 0
2150         )         )
2151        RRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
2152      ecode++;      ecode++;
2153      break;      break;
2154    
2155      case OP_ANYNL:      case OP_ANYNL:
2156      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2157          {
2158          SCHECK_PARTIAL();
2159          MRRETURN(MATCH_NOMATCH);
2160          }
2161      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2162      switch(c)      switch(c)
2163        {        {
2164        default: RRETURN(MATCH_NOMATCH);        default: MRRETURN(MATCH_NOMATCH);
2165    
2166        case 0x000d:        case 0x000d:
2167        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;        if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2168        break;        break;
2169    
2170        case 0x000a:        case 0x000a:
2171          break;
2172    
2173        case 0x000b:        case 0x000b:
2174        case 0x000c:        case 0x000c:
2175        case 0x0085:        case 0x0085:
2176        case 0x2028:        case 0x2028:
2177        case 0x2029:        case 0x2029:
2178          if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH);
2179          break;
2180          }
2181        ecode++;
2182        break;
2183    
2184        case OP_NOT_HSPACE:
2185        if (eptr >= md->end_subject)
2186          {
2187          SCHECK_PARTIAL();
2188          MRRETURN(MATCH_NOMATCH);
2189          }
2190        GETCHARINCTEST(c, eptr);
2191        switch(c)
2192          {
2193          default: break;
2194          case 0x09:      /* HT */
2195          case 0x20:      /* SPACE */
2196          case 0xa0:      /* NBSP */
2197          case 0x1680:    /* OGHAM SPACE MARK */
2198          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2199          case 0x2000:    /* EN QUAD */
2200          case 0x2001:    /* EM QUAD */
2201          case 0x2002:    /* EN SPACE */
2202          case 0x2003:    /* EM SPACE */
2203          case 0x2004:    /* THREE-PER-EM SPACE */
2204          case 0x2005:    /* FOUR-PER-EM SPACE */
2205          case 0x2006:    /* SIX-PER-EM SPACE */
2206          case 0x2007:    /* FIGURE SPACE */
2207          case 0x2008:    /* PUNCTUATION SPACE */
2208          case 0x2009:    /* THIN SPACE */
2209          case 0x200A:    /* HAIR SPACE */
2210          case 0x202f:    /* NARROW NO-BREAK SPACE */
2211          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2212          case 0x3000:    /* IDEOGRAPHIC SPACE */
2213          MRRETURN(MATCH_NOMATCH);
2214          }
2215        ecode++;
2216        break;
2217    
2218        case OP_HSPACE:
2219        if (eptr >= md->end_subject)
2220          {
2221          SCHECK_PARTIAL();
2222          MRRETURN(MATCH_NOMATCH);
2223          }
2224        GETCHARINCTEST(c, eptr);
2225        switch(c)
2226          {
2227          default: MRRETURN(MATCH_NOMATCH);
2228          case 0x09:      /* HT */
2229          case 0x20:      /* SPACE */
2230          case 0xa0:      /* NBSP */
2231          case 0x1680:    /* OGHAM SPACE MARK */
2232          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
2233          case 0x2000:    /* EN QUAD */
2234          case 0x2001:    /* EM QUAD */
2235          case 0x2002:    /* EN SPACE */
2236          case 0x2003:    /* EM SPACE */
2237          case 0x2004:    /* THREE-PER-EM SPACE */
2238          case 0x2005:    /* FOUR-PER-EM SPACE */
2239          case 0x2006:    /* SIX-PER-EM SPACE */
2240          case 0x2007:    /* FIGURE SPACE */
2241          case 0x2008:    /* PUNCTUATION SPACE */
2242          case 0x2009:    /* THIN SPACE */
2243          case 0x200A:    /* HAIR SPACE */
2244          case 0x202f:    /* NARROW NO-BREAK SPACE */
2245          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
2246          case 0x3000:    /* IDEOGRAPHIC SPACE */
2247          break;
2248          }
2249        ecode++;
2250        break;
2251    
2252        case OP_NOT_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: break;
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          MRRETURN(MATCH_NOMATCH);
2270          }
2271        ecode++;
2272        break;
2273    
2274        case OP_VSPACE:
2275        if (eptr >= md->end_subject)
2276          {
2277          SCHECK_PARTIAL();
2278          MRRETURN(MATCH_NOMATCH);
2279          }
2280        GETCHARINCTEST(c, eptr);
2281        switch(c)
2282          {
2283          default: MRRETURN(MATCH_NOMATCH);
2284          case 0x0a:      /* LF */
2285          case 0x0b:      /* VT */
2286          case 0x0c:      /* FF */
2287          case 0x0d:      /* CR */
2288          case 0x85:      /* NEL */
2289          case 0x2028:    /* LINE SEPARATOR */
2290          case 0x2029:    /* PARAGRAPH SEPARATOR */
2291        break;        break;
2292        }        }
2293      ecode++;      ecode++;
# Line 1488  for (;;) Line 2299  for (;;)
2299    
2300      case OP_PROP:      case OP_PROP:
2301      case OP_NOTPROP:      case OP_NOTPROP:
2302      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2303          {
2304          SCHECK_PARTIAL();
2305          MRRETURN(MATCH_NOMATCH);
2306          }
2307      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2308        {        {
2309        int chartype, script;        const ucd_record *prop = GET_UCD(c);
       int category = _pcre_ucp_findprop(c, &chartype, &script);  
2310    
2311        switch(ecode[1])        switch(ecode[1])
2312          {          {
2313          case PT_ANY:          case PT_ANY:
2314          if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);          if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH);
2315          break;          break;
2316    
2317          case PT_LAMP:          case PT_LAMP:
2318          if ((chartype == ucp_Lu ||          if ((prop->chartype == ucp_Lu ||
2319               chartype == ucp_Ll ||               prop->chartype == ucp_Ll ||
2320               chartype == ucp_Lt) == (op == OP_NOTPROP))               prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
2321            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2322           break;          break;
2323    
2324          case PT_GC:          case PT_GC:
2325          if ((ecode[2] != category) == (op == OP_PROP))          if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
2326            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2327          break;          break;
2328    
2329          case PT_PC:          case PT_PC:
2330          if ((ecode[2] != chartype) == (op == OP_PROP))          if ((ecode[2] != prop->chartype) == (op == OP_PROP))
2331            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2332          break;          break;
2333    
2334          case PT_SC:          case PT_SC:
2335          if ((ecode[2] != script) == (op == OP_PROP))          if ((ecode[2] != prop->script) == (op == OP_PROP))
2336            RRETURN(MATCH_NOMATCH);            MRRETURN(MATCH_NOMATCH);
2337            break;
2338    
2339            /* These are specials */
2340    
2341            case PT_ALNUM:
2342            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2343                 _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP))
2344              MRRETURN(MATCH_NOMATCH);
2345            break;
2346    
2347            case PT_SPACE:    /* Perl space */
2348            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2349                 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2350                   == (op == OP_NOTPROP))
2351              MRRETURN(MATCH_NOMATCH);
2352            break;
2353    
2354            case PT_PXSPACE:  /* POSIX space */
2355            if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2356                 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2357                 c == CHAR_FF || c == CHAR_CR)
2358                   == (op == OP_NOTPROP))
2359              MRRETURN(MATCH_NOMATCH);
2360          break;          break;
2361    
2362            case PT_WORD:
2363            if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2364                 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2365                 c == CHAR_UNDERSCORE) == (op == OP_NOTPROP))
2366              MRRETURN(MATCH_NOMATCH);
2367            break;
2368    
2369            /* This should never occur */
2370    
2371          default:          default:
2372          RRETURN(PCRE_ERROR_INTERNAL);          RRETURN(PCRE_ERROR_INTERNAL);
2373          }          }
# Line 1534  for (;;) Line 2380  for (;;)
2380      is in the binary; otherwise a compile-time error occurs. */      is in the binary; otherwise a compile-time error occurs. */
2381    
2382      case OP_EXTUNI:      case OP_EXTUNI:
2383      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject)
2384          {
2385          SCHECK_PARTIAL();
2386          MRRETURN(MATCH_NOMATCH);
2387          }
2388      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
2389        if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH);
2390        while (eptr < md->end_subject)
2391        {        {
2392        int chartype, script;        int len = 1;
2393        int category = _pcre_ucp_findprop(c, &chartype, &script);        if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); }
2394        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (UCD_CATEGORY(c) != ucp_M) break;
2395        while (eptr < md->end_subject)        eptr += len;
         {  
         int len = 1;  
         if (!utf8) c = *eptr; else  
           {  
           GETCHARLEN(c, eptr, len);  
           }  
         category = _pcre_ucp_findprop(c, &chartype, &script);  
         if (category != ucp_M) break;  
         eptr += len;  
         }  
2396        }        }
2397      ecode++;      ecode++;
2398      break;      break;
# Line 1566  for (;;) Line 2408  for (;;)
2408      loops). */      loops). */
2409    
2410      case OP_REF:      case OP_REF:
2411        {      case OP_REFI:
2412        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      caseless = op == OP_REFI;
2413        ecode += 3;                                 /* Advance past item */      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2414        ecode += 3;
2415    
2416        /* 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];  
2417    
2418        /* Set up for repetition, or handle the non-repeated case */      (a) In the default, Perl-compatible state, set the length negative;
2419        this ensures that every attempt at a match fails. We can't just fail
2420        here, because of the possibility of quantifiers with zero minima.
2421    
2422        switch (*ecode)      (b) If the JavaScript compatibility flag is set, set the length to zero
2423          {      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;  
2424    
2425          case OP_CRRANGE:      Otherwise, set the length to the length of what was matched by the
2426          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;  
2427    
2428          default:               /* No repeat follows */      if (offset >= offset_top || md->offset_vector[offset] < 0)
2429          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);        length = (md->jscript_compat)? 0 : -1;
2430          eptr += length;      else
2431          continue;              /* With the main loop */        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2432    
2433        /* Set up for repetition, or handle the non-repeated case */
2434    
2435        switch (*ecode)
2436          {
2437          case OP_CRSTAR:
2438          case OP_CRMINSTAR:
2439          case OP_CRPLUS:
2440          case OP_CRMINPLUS:
2441          case OP_CRQUERY:
2442          case OP_CRMINQUERY:
2443          c = *ecode++ - OP_CRSTAR;
2444          minimize = (c & 1) != 0;
2445          min = rep_min[c];                 /* Pick up values from tables; */
2446          max = rep_max[c];                 /* zero for max => infinity */
2447          if (max == 0) max = INT_MAX;
2448          break;
2449    
2450          case OP_CRRANGE:
2451          case OP_CRMINRANGE:
2452          minimize = (*ecode == OP_CRMINRANGE);
2453          min = GET2(ecode, 1);
2454          max = GET2(ecode, 3);
2455          if (max == 0) max = INT_MAX;
2456          ecode += 5;
2457          break;
2458    
2459          default:               /* No repeat follows */
2460          if ((length = match_ref(offset, eptr, length, md, caseless)) < 0)
2461            {
2462            CHECK_PARTIAL();
2463            MRRETURN(MATCH_NOMATCH);
2464          }          }
2465          eptr += length;
2466          continue;              /* With the main loop */
2467          }
2468    
2469        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2470        main loop. */      zero, just continue with the main loop. */
2471    
2472        if (length == 0) continue;      if (length == 0) continue;
2473    
2474        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2475        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2476        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2477    
2478        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2479          {
2480          int slength;
2481          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2482          {          {
2483          if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH);          CHECK_PARTIAL();
2484          eptr += length;          MRRETURN(MATCH_NOMATCH);
2485          }          }
2486          eptr += slength;
2487          }
2488    
2489        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2490        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2491    
2492        if (min == max) continue;      if (min == max) continue;
2493    
2494        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2495    
2496        if (minimize)      if (minimize)
2497          {
2498          for (fi = min;; fi++)
2499          {          {
2500          for (fi = min;; fi++)          int slength;
2501            RMATCH(eptr, ecode, offset_top, md, eptrb, RM14);
2502            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2503            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2504            if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2505            {            {
2506            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2507            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max || !match_ref(offset, eptr, length, md, ims))  
             RRETURN(MATCH_NOMATCH);  
           eptr += length;  
2508            }            }
2509          /* Control never gets here */          eptr += slength;
2510          }          }
2511          /* Control never gets here */
2512          }
2513    
2514        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2515    
2516        else      else
2517          {
2518          pp = eptr;
2519          for (i = min; i < max; i++)
2520          {          {
2521          pp = eptr;          int slength;
2522          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0)
2523            {            {
2524            if (!match_ref(offset, eptr, length, md, ims)) break;            CHECK_PARTIAL();
2525            eptr += length;            break;
           }  
         while (eptr >= pp)  
           {  
           RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);  
           if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
           eptr -= length;  
2526            }            }
2527          RRETURN(MATCH_NOMATCH);          eptr += slength;
2528            }
2529          while (eptr >= pp)
2530            {
2531            RMATCH(eptr, ecode, offset_top, md, eptrb, RM15);
2532            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2533            eptr -= length;
2534          }          }
2535          MRRETURN(MATCH_NOMATCH);
2536        }        }
2537      /* Control never gets here */      /* Control never gets here */
2538    
   
   
2539      /* Match a bit-mapped character class, possibly repeatedly. This op code is      /* Match a bit-mapped character class, possibly repeatedly. This op code is
2540      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,
2541      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 1723  for (;;) Line 2590  for (;;)
2590          {          {
2591          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2592            {            {
2593            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2594                {
2595                SCHECK_PARTIAL();
2596                MRRETURN(MATCH_NOMATCH);
2597                }
2598            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
2599            if (c > 255)            if (c > 255)
2600              {              {
2601              if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);              if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2602              }              }
2603            else            else
2604              {              {
2605              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2606              }              }
2607            }            }
2608          }          }
# Line 1741  for (;;) Line 2612  for (;;)
2612          {          {
2613          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2614            {            {
2615            if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
2616                {
2617                SCHECK_PARTIAL();
2618                MRRETURN(MATCH_NOMATCH);
2619                }
2620            c = *eptr++;            c = *eptr++;
2621            if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);            if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2622            }            }
2623          }          }
2624    
# Line 1763  for (;;) Line 2638  for (;;)
2638            {            {
2639            for (fi = min;; fi++)            for (fi = min;; fi++)
2640              {              {
2641              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM16);
2642              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2643              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2644                if (eptr >= md->end_subject)
2645                  {
2646                  SCHECK_PARTIAL();
2647                  MRRETURN(MATCH_NOMATCH);
2648                  }
2649              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
2650              if (c > 255)              if (c > 255)
2651                {                {
2652                if (op == OP_CLASS) RRETURN(MATCH_NOMATCH);                if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH);
2653                }                }
2654              else              else
2655                {                {
2656                if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);                if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2657                }                }
2658              }              }
2659            }            }
# Line 1783  for (;;) Line 2663  for (;;)
2663            {            {
2664            for (fi = min;; fi++)            for (fi = min;; fi++)
2665              {              {
2666              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM17);
2667              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2668              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
2669                if (eptr >= md->end_subject)
2670                  {
2671                  SCHECK_PARTIAL();
2672                  MRRETURN(MATCH_NOMATCH);
2673                  }
2674              c = *eptr++;              c = *eptr++;
2675              if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH);              if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH);
2676              }              }
2677            }            }
2678          /* Control never gets here */          /* Control never gets here */
# Line 1806  for (;;) Line 2691  for (;;)
2691            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2692              {              {
2693              int len = 1;              int len = 1;
2694              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2695                  {
2696                  SCHECK_PARTIAL();
2697                  break;
2698                  }
2699              GETCHARLEN(c, eptr, len);              GETCHARLEN(c, eptr, len);
2700              if (c > 255)              if (c > 255)
2701                {                {
# Line 1820  for (;;) Line 2709  for (;;)
2709              }              }
2710            for (;;)            for (;;)
2711              {              {
2712              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
2713              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2714              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2715              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1832  for (;;) Line 2721  for (;;)
2721            {            {
2722            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2723              {              {
2724              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
2725                  {
2726                  SCHECK_PARTIAL();
2727                  break;
2728                  }
2729              c = *eptr;              c = *eptr;
2730              if ((data[c/8] & (1 << (c&7))) == 0) break;              if ((data[c/8] & (1 << (c&7))) == 0) break;
2731              eptr++;              eptr++;
2732              }              }
2733            while (eptr >= pp)            while (eptr >= pp)
2734              {              {
2735              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
2736              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2737              eptr--;              eptr--;
2738              }              }
2739            }            }
2740    
2741          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2742          }          }
2743        }        }
2744      /* Control never gets here */      /* Control never gets here */
2745    
2746    
2747      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. This opcode is encountered only
2748      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
2749        mode, because Unicode properties are supported in non-UTF-8 mode. */
2750    
2751  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2752      case OP_XCLASS:      case OP_XCLASS:
# Line 1893  for (;;) Line 2787  for (;;)
2787    
2788        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
2789          {          {
2790          if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);          if (eptr >= md->end_subject)
2791          GETCHARINC(c, eptr);            {
2792          if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);            SCHECK_PARTIAL();
2793              MRRETURN(MATCH_NOMATCH);
2794              }
2795            GETCHARINCTEST(c, eptr);
2796            if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2797          }          }
2798    
2799        /* 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 1910  for (;;) Line 2808  for (;;)
2808          {          {
2809          for (fi = min;; fi++)          for (fi = min;; fi++)
2810            {            {
2811            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM20);
2812            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2813            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2814            GETCHARINC(c, eptr);            if (eptr >= md->end_subject)
2815            if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH);              {
2816                SCHECK_PARTIAL();
2817                MRRETURN(MATCH_NOMATCH);
2818                }
2819              GETCHARINCTEST(c, eptr);
2820              if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH);
2821            }            }
2822          /* Control never gets here */          /* Control never gets here */
2823          }          }
# Line 1927  for (;;) Line 2830  for (;;)
2830          for (i = min; i < max; i++)          for (i = min; i < max; i++)
2831            {            {
2832            int len = 1;            int len = 1;
2833            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject)
2834            GETCHARLEN(c, eptr, len);              {
2835                SCHECK_PARTIAL();
2836                break;
2837                }
2838              GETCHARLENTEST(c, eptr, len);
2839            if (!_pcre_xclass(c, data)) break;            if (!_pcre_xclass(c, data)) break;
2840            eptr += len;            eptr += len;
2841            }            }
2842          for(;;)          for(;;)
2843            {            {
2844            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
2845            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2846            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2847            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2848            }            }
2849          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
2850          }          }
2851    
2852        /* Control never gets here */        /* Control never gets here */
# Line 1955  for (;;) Line 2862  for (;;)
2862        length = 1;        length = 1;
2863        ecode++;        ecode++;
2864        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2865        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2866        while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH);          {
2867            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2868            MRRETURN(MATCH_NOMATCH);
2869            }
2870          while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH);
2871        }        }
2872      else      else
2873  #endif  #endif
2874    
2875      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2876        {        {
2877        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2878        if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH);          {
2879            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2880            MRRETURN(MATCH_NOMATCH);
2881            }
2882          if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH);
2883        ecode += 2;        ecode += 2;
2884        }        }
2885      break;      break;
2886    
2887      /* Match a single character, caselessly */      /* Match a single character, caselessly */
2888    
2889      case OP_CHARNC:      case OP_CHARI:
2890  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2891      if (utf8)      if (utf8)
2892        {        {
# Line 1979  for (;;) Line 2894  for (;;)
2894        ecode++;        ecode++;
2895        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
2896    
2897        if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);        if (length > md->end_subject - eptr)
2898            {
2899            CHECK_PARTIAL();             /* Not SCHECK_PARTIAL() */
2900            MRRETURN(MATCH_NOMATCH);
2901            }
2902    
2903        /* 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
2904        can use the fast lookup table. */        can use the fast lookup table. */
2905    
2906        if (fc < 128)        if (fc < 128)
2907          {          {
2908          if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2909          }          }
2910    
2911        /* Otherwise we must pick up the subject character */        /* Otherwise we must pick up the subject character */
# Line 2003  for (;;) Line 2922  for (;;)
2922          if (fc != dc)          if (fc != dc)
2923            {            {
2924  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2925            if (dc != _pcre_ucp_othercase(fc))            if (dc != UCD_OTHERCASE(fc))
2926  #endif  #endif
2927              RRETURN(MATCH_NOMATCH);              MRRETURN(MATCH_NOMATCH);
2928            }            }
2929          }          }
2930        }        }
# Line 2014  for (;;) Line 2933  for (;;)
2933    
2934      /* Non-UTF-8 mode */      /* Non-UTF-8 mode */
2935        {        {
2936        if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH);        if (md->end_subject - eptr < 1)
2937        if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
2938            SCHECK_PARTIAL();            /* This one can use SCHECK_PARTIAL() */
2939            MRRETURN(MATCH_NOMATCH);
2940            }
2941          if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
2942        ecode += 2;        ecode += 2;
2943        }        }
2944      break;      break;
# Line 2023  for (;;) Line 2946  for (;;)
2946      /* Match a single character repeatedly. */      /* Match a single character repeatedly. */
2947    
2948      case OP_EXACT:      case OP_EXACT:
2949        case OP_EXACTI:
2950      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2951      ecode += 3;      ecode += 3;
2952      goto REPEATCHAR;      goto REPEATCHAR;
2953    
2954      case OP_POSUPTO:      case OP_POSUPTO:
2955        case OP_POSUPTOI:
2956      possessive = TRUE;      possessive = TRUE;
2957      /* Fall through */      /* Fall through */
2958    
2959      case OP_UPTO:      case OP_UPTO:
2960        case OP_UPTOI:
2961      case OP_MINUPTO:      case OP_MINUPTO:
2962        case OP_MINUPTOI:
2963      min = 0;      min = 0;
2964      max = GET2(ecode, 1);      max = GET2(ecode, 1);
2965      minimize = *ecode == OP_MINUPTO;      minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI;
2966      ecode += 3;      ecode += 3;
2967      goto REPEATCHAR;      goto REPEATCHAR;
2968    
2969      case OP_POSSTAR:      case OP_POSSTAR:
2970        case OP_POSSTARI:
2971      possessive = TRUE;      possessive = TRUE;
2972      min = 0;      min = 0;
2973      max = INT_MAX;      max = INT_MAX;
# Line 2047  for (;;) Line 2975  for (;;)
2975      goto REPEATCHAR;      goto REPEATCHAR;
2976    
2977      case OP_POSPLUS:      case OP_POSPLUS:
2978        case OP_POSPLUSI:
2979      possessive = TRUE;      possessive = TRUE;
2980      min = 1;      min = 1;
2981      max = INT_MAX;      max = INT_MAX;
# Line 2054  for (;;) Line 2983  for (;;)
2983      goto REPEATCHAR;      goto REPEATCHAR;
2984    
2985      case OP_POSQUERY:      case OP_POSQUERY:
2986        case OP_POSQUERYI:
2987      possessive = TRUE;      possessive = TRUE;
2988      min = 0;      min = 0;
2989      max = 1;      max = 1;
# Line 2061  for (;;) Line 2991  for (;;)
2991      goto REPEATCHAR;      goto REPEATCHAR;
2992    
2993      case OP_STAR:      case OP_STAR:
2994        case OP_STARI:
2995      case OP_MINSTAR:      case OP_MINSTAR:
2996        case OP_MINSTARI:
2997      case OP_PLUS:      case OP_PLUS:
2998        case OP_PLUSI:
2999      case OP_MINPLUS:      case OP_MINPLUS:
3000        case OP_MINPLUSI:
3001      case OP_QUERY:      case OP_QUERY:
3002        case OP_QUERYI:
3003      case OP_MINQUERY:      case OP_MINQUERY:
3004      c = *ecode++ - OP_STAR;      case OP_MINQUERYI:
3005        c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI);
3006      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3007      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3008      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3009      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3010    
3011      /* 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. */  
3012    
3013      REPEATCHAR:      REPEATCHAR:
3014  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2083  for (;;) Line 3017  for (;;)
3017        length = 1;        length = 1;
3018        charptr = ecode;        charptr = ecode;
3019        GETCHARLEN(fc, ecode, length);        GETCHARLEN(fc, ecode, length);
       if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3020        ecode += length;        ecode += length;
3021    
3022        /* Handle multibyte character matching specially here. There is        /* Handle multibyte character matching specially here. There is
# Line 2093  for (;;) Line 3026  for (;;)
3026          {          {
3027  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3028          unsigned int othercase;          unsigned int othercase;
3029          if ((ims & PCRE_CASELESS) != 0 &&          if (op >= OP_STARI &&     /* Caseless */
3030              (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)              (othercase = UCD_OTHERCASE(fc)) != fc)
3031            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
3032          else oclength = 0;          else oclength = 0;
3033  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3034    
3035          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3036            {            {
3037            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (eptr <= md->end_subject - length &&
3038                memcmp(eptr, charptr, length) == 0) eptr += length;
3039  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3040            /* Need braces because of following else */            else if (oclength > 0 &&
3041            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                     eptr <= md->end_subject - oclength &&
3042                       memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3043    #endif  /* SUPPORT_UCP */
3044            else            else
3045              {              {
3046              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              CHECK_PARTIAL();
3047              eptr += oclength;              MRRETURN(MATCH_NOMATCH);
3048              }              }
 #else   /* without SUPPORT_UCP */  
           else { RRETURN(MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3049            }            }
3050    
3051          if (min == max) continue;          if (min == max) continue;
# Line 2121  for (;;) Line 3054  for (;;)
3054            {            {
3055            for (fi = min;; fi++)            for (fi = min;; fi++)
3056              {              {
3057              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM22);
3058              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3059              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3060              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (eptr <= md->end_subject - length &&
3061                  memcmp(eptr, charptr, length) == 0) eptr += length;
3062  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3063              /* Need braces because of following else */              else if (oclength > 0 &&
3064              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }                       eptr <= md->end_subject - oclength &&
3065                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3066    #endif  /* SUPPORT_UCP */
3067              else              else
3068                {                {
3069                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                CHECK_PARTIAL();
3070                eptr += oclength;                MRRETURN(MATCH_NOMATCH);
3071                }                }
 #else   /* without SUPPORT_UCP */  
             else { RRETURN (MATCH_NOMATCH); }  
 #endif  /* SUPPORT_UCP */  
3072              }              }
3073            /* Control never gets here */            /* Control never gets here */
3074            }            }
# Line 2145  for (;;) Line 3078  for (;;)
3078            pp = eptr;            pp = eptr;
3079            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3080              {              {
3081              if (eptr > md->end_subject - length) break;              if (eptr <= md->end_subject - length &&
3082              if (memcmp(eptr, charptr, length) == 0) eptr += length;                  memcmp(eptr, charptr, length) == 0) eptr += length;
3083  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3084              else if (oclength == 0) break;              else if (oclength > 0 &&
3085                         eptr <= md->end_subject - oclength &&
3086                         memcmp(eptr, occhars, oclength) == 0) eptr += oclength;
3087    #endif  /* SUPPORT_UCP */
3088              else              else
3089                {                {
3090                if (memcmp(eptr, occhars, oclength) != 0) break;                CHECK_PARTIAL();
3091                eptr += oclength;                break;
3092                }                }
 #else   /* without SUPPORT_UCP */  
             else break;  
 #endif  /* SUPPORT_UCP */  
3093              }              }
3094    
3095            if (possessive) continue;            if (possessive) continue;
3096    
3097            for(;;)            for(;;)
3098             {              {
3099             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3100             if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3101             if (eptr == pp) RRETURN(MATCH_NOMATCH);              if (eptr == pp) { MRRETURN(MATCH_NOMATCH); }
3102  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3103             eptr--;              eptr--;
3104             BACKCHAR(eptr);              BACKCHAR(eptr);
3105  #else   /* without SUPPORT_UCP */  #else   /* without SUPPORT_UCP */
3106             eptr -= length;              eptr -= length;
3107  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3108             }              }
3109            }            }
3110          /* Control never gets here */          /* Control never gets here */
3111          }          }
# Line 2184  for (;;) Line 3118  for (;;)
3118  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
3119    
3120      /* When not in UTF-8 mode, load a single-byte character. */      /* When not in UTF-8 mode, load a single-byte character. */
3121        {  
3122        if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);      fc = *ecode++;
       fc = *ecode++;  
       }  
3123    
3124      /* 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
3125      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 2201  for (;;) Line 3133  for (;;)
3133      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,      DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
3134        max, eptr));        max, eptr));
3135    
3136      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_STARI)  /* Caseless */
3137        {        {
3138        fc = md->lcc[fc];        fc = md->lcc[fc];
3139        for (i = 1; i <= min; i++)        for (i = 1; i <= min; i++)
3140          if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);          {
3141            if (eptr >= md->end_subject)
3142              {
3143              SCHECK_PARTIAL();
3144              MRRETURN(MATCH_NOMATCH);
3145              }
3146            if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3147            }
3148        if (min == max) continue;        if (min == max) continue;
3149        if (minimize)        if (minimize)
3150          {          {
3151          for (fi = min;; fi++)          for (fi = min;; fi++)
3152            {            {
3153            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM24);
3154            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3155            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3156                fc != md->lcc[*eptr++])            if (eptr >= md->end_subject)
3157              RRETURN(MATCH_NOMATCH);              {
3158                SCHECK_PARTIAL();
3159                MRRETURN(MATCH_NOMATCH);
3160                }
3161              if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3162            }            }
3163          /* Control never gets here */          /* Control never gets here */
3164          }          }
# Line 2224  for (;;) Line 3167  for (;;)
3167          pp = eptr;          pp = eptr;
3168          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3169            {            {
3170            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject)
3171                {
3172                SCHECK_PARTIAL();
3173                break;
3174                }
3175              if (fc != md->lcc[*eptr]) break;
3176            eptr++;            eptr++;
3177            }            }
3178    
3179          if (possessive) continue;          if (possessive) continue;
3180    
3181          while (eptr >= pp)          while (eptr >= pp)
3182            {            {
3183            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3184            eptr--;            eptr--;
3185            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3186            }            }
3187          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3188          }          }
3189        /* Control never gets here */        /* Control never gets here */
3190        }        }
3191    
3192      /* Caseful comparisons (includes all multi-byte characters) */      /* Caseful comparisons (includes all multi-byte characters) */
3193    
3194      else      else
3195        {        {
3196        for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH);        for (i = 1; i <= min; i++)
3197            {
3198            if (eptr >= md->end_subject)
3199              {
3200              SCHECK_PARTIAL();
3201              MRRETURN(MATCH_NOMATCH);
3202              }
3203            if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3204            }
3205    
3206        if (min == max) continue;        if (min == max) continue;
3207    
3208        if (minimize)        if (minimize)
3209          {          {
3210          for (fi = min;; fi++)          for (fi = min;; fi++)
3211            {            {
3212            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM26);
3213            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3214            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max) MRRETURN(MATCH_NOMATCH);
3215              RRETURN(MATCH_NOMATCH);            if (eptr >= md->end_subject)
3216                {
3217                SCHECK_PARTIAL();
3218                MRRETURN(MATCH_NOMATCH);
3219                }
3220              if (fc != *eptr++) MRRETURN(MATCH_NOMATCH);
3221            }            }
3222          /* Control never gets here */          /* Control never gets here */
3223          }          }
# Line 2261  for (;;) Line 3226  for (;;)
3226          pp = eptr;          pp = eptr;
3227          for (i = min; i < max; i++)          for (i = min; i < max; i++)
3228            {            {
3229            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject)
3230                {
3231                SCHECK_PARTIAL();
3232                break;
3233                }
3234              if (fc != *eptr) break;
3235            eptr++;            eptr++;
3236            }            }
3237          if (possessive) continue;          if (possessive) continue;
3238    
3239          while (eptr >= pp)          while (eptr >= pp)
3240            {            {
3241            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3242            eptr--;            eptr--;
3243            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3244            }            }
3245          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3246          }          }
3247        }        }
3248      /* Control never gets here */      /* Control never gets here */
# Line 2280  for (;;) Line 3251  for (;;)
3251      checking can be multibyte. */      checking can be multibyte. */
3252    
3253      case OP_NOT:      case OP_NOT:
3254      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      case OP_NOTI:
3255        if (eptr >= md->end_subject)
3256          {
3257          SCHECK_PARTIAL();
3258          MRRETURN(MATCH_NOMATCH);
3259          }
3260      ecode++;      ecode++;
3261      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
3262      if ((ims & PCRE_CASELESS) != 0)      if (op == OP_NOTI)         /* The caseless case */
3263        {        {
3264  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3265        if (c < 256)        if (c < 256)
3266  #endif  #endif
3267        c = md->lcc[c];        c = md->lcc[c];
3268        if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH);        if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH);
3269        }        }
3270      else      else    /* Caseful */
3271        {        {
3272        if (*ecode++ == c) RRETURN(MATCH_NOMATCH);        if (*ecode++ == c) MRRETURN(MATCH_NOMATCH);
3273        }        }
3274      break;      break;
3275    
# Line 2305  for (;;) Line 3281  for (;;)
3281      about... */      about... */
3282    
3283      case OP_NOTEXACT:      case OP_NOTEXACT:
3284        case OP_NOTEXACTI:
3285      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
3286      ecode += 3;      ecode += 3;
3287      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3288    
3289      case OP_NOTUPTO:      case OP_NOTUPTO:
3290        case OP_NOTUPTOI:
3291      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
3292        case OP_NOTMINUPTOI:
3293      min = 0;      min = 0;
3294      max = GET2(ecode, 1);      max = GET2(ecode, 1);
3295      minimize = *ecode == OP_NOTMINUPTO;      minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI;
3296      ecode += 3;      ecode += 3;
3297      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3298    
3299      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
3300        case OP_NOTPOSSTARI:
3301      possessive = TRUE;      possessive = TRUE;
3302      min = 0;      min = 0;
3303      max = INT_MAX;      max = INT_MAX;
# Line 2325  for (;;) Line 3305  for (;;)
3305      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3306    
3307      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
3308        case OP_NOTPOSPLUSI:
3309      possessive = TRUE;      possessive = TRUE;
3310      min = 1;      min = 1;
3311      max = INT_MAX;      max = INT_MAX;
# Line 2332  for (;;) Line 3313  for (;;)
3313      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3314    
3315      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
3316        case OP_NOTPOSQUERYI:
3317      possessive = TRUE;      possessive = TRUE;
3318      min = 0;      min = 0;
3319      max = 1;      max = 1;
# Line 2339  for (;;) Line 3321  for (;;)
3321      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3322    
3323      case OP_NOTPOSUPTO:      case OP_NOTPOSUPTO:
3324        case OP_NOTPOSUPTOI:
3325      possessive = TRUE;      possessive = TRUE;
3326      min = 0;      min = 0;
3327      max = GET2(ecode, 1);      max = GET2(ecode, 1);
# Line 2346  for (;;) Line 3329  for (;;)
3329      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
3330    
3331      case OP_NOTSTAR:      case OP_NOTSTAR:
3332        case OP_NOTSTARI:
3333      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
3334        case OP_NOTMINSTARI:
3335      case OP_NOTPLUS:      case OP_NOTPLUS:
3336        case OP_NOTPLUSI:
3337      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
3338        case OP_NOTMINPLUSI:
3339      case OP_NOTQUERY:      case OP_NOTQUERY:
3340        case OP_NOTQUERYI:
3341      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
3342      c = *ecode++ - OP_NOTSTAR;      case OP_NOTMINQUERYI:
3343        c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR);
3344      minimize = (c & 1) != 0;      minimize = (c & 1) != 0;
3345      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
3346      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3347      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3348    
3349      /* 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. */  
3350    
3351      REPEATNOTCHAR:      REPEATNOTCHAR:
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3352      fc = *ecode++;      fc = *ecode++;
3353    
3354      /* 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 2376  for (;;) Line 3362  for (;;)
3362      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,
3363        max, eptr));        max, eptr));
3364    
3365      if ((ims & PCRE_CASELESS) != 0)      if (op >= OP_NOTSTARI)     /* Caseless */
3366        {        {
3367        fc = md->lcc[fc];        fc = md->lcc[fc];
3368    
# Line 2387  for (;;) Line 3373  for (;;)
3373          register unsigned int d;          register unsigned int d;
3374          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3375            {            {
3376              if (eptr >= md->end_subject)
3377                {
3378                SCHECK_PARTIAL();
3379                MRRETURN(MATCH_NOMATCH);
3380                }
3381            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3382            if (d < 256) d = md->lcc[d];            if (d < 256) d = md->lcc[d];
3383            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3384            }            }
3385          }          }
3386        else        else
# Line 2398  for (;;) Line 3389  for (;;)
3389        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3390          {          {
3391          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3392            if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH);            {
3393              if (eptr >= md->end_subject)
3394                {
3395                SCHECK_PARTIAL();
3396                MRRETURN(MATCH_NOMATCH);
3397                }
3398              if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3399              }
3400          }          }
3401    
3402        if (min == max) continue;        if (min == max) continue;
# Line 2412  for (;;) Line 3410  for (;;)
3410            register unsigned int d;            register unsigned int d;
3411            for (fi = min;; fi++)            for (fi = min;; fi++)
3412              {              {
3413              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM28);
3414              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3415                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3416                if (eptr >= md->end_subject)
3417                  {
3418                  SCHECK_PARTIAL();
3419                  MRRETURN(MATCH_NOMATCH);
3420                  }
3421              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3422              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3423              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3424              }              }
3425            }            }
3426          else          else
# Line 2426  for (;;) Line 3429  for (;;)
3429            {            {
3430            for (fi = min;; fi++)            for (fi = min;; fi++)
3431              {              {
3432              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM29);
3433              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3434              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3435                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3436                  {
3437                  SCHECK_PARTIAL();
3438                  MRRETURN(MATCH_NOMATCH);
3439                  }
3440                if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH);
3441              }              }
3442            }            }
3443          /* Control never gets here */          /* Control never gets here */
# Line 2449  for (;;) Line 3457  for (;;)
3457            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3458              {              {
3459              int len = 1;              int len = 1;
3460              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3461                  {
3462                  SCHECK_PARTIAL();
3463                  break;
3464                  }
3465              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3466              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
3467              if (fc == d) break;              if (fc == d) break;
# Line 2458  for (;;) Line 3470  for (;;)
3470          if (possessive) continue;          if (possessive) continue;
3471          for(;;)          for(;;)
3472              {              {
3473              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3474              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3475              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3476              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2470  for (;;) Line 3482  for (;;)
3482            {            {
3483            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3484              {              {
3485              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject)
3486                  {
3487                  SCHECK_PARTIAL();
3488                  break;
3489                  }
3490                if (fc == md->lcc[*eptr]) break;
3491              eptr++;              eptr++;
3492              }              }
3493            if (possessive) continue;            if (possessive) continue;
3494            while (eptr >= pp)            while (eptr >= pp)
3495              {              {
3496              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3497              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3498              eptr--;              eptr--;
3499              }              }
3500            }            }
3501    
3502          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3503          }          }
3504        /* Control never gets here */        /* Control never gets here */
3505        }        }
# Line 2498  for (;;) Line 3515  for (;;)
3515          register unsigned int d;          register unsigned int d;
3516          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3517            {            {
3518              if (eptr >= md->end_subject)
3519                {
3520                SCHECK_PARTIAL();
3521                MRRETURN(MATCH_NOMATCH);
3522                }
3523            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
3524            if (fc == d) RRETURN(MATCH_NOMATCH);            if (fc == d) MRRETURN(MATCH_NOMATCH);
3525            }            }
3526          }          }
3527        else        else
# Line 2507  for (;;) Line 3529  for (;;)
3529        /* Not UTF-8 mode */        /* Not UTF-8 mode */
3530          {          {
3531          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3532            if (fc == *eptr++) RRETURN(MATCH_NOMATCH);            {
3533              if (eptr >= md->end_subject)
3534                {
3535                SCHECK_PARTIAL();
3536                MRRETURN(MATCH_NOMATCH);
3537                }
3538              if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3539              }
3540          }          }
3541    
3542        if (min == max) continue;        if (min == max) continue;
# Line 2521  for (;;) Line 3550  for (;;)
3550            register unsigned int d;            register unsigned int d;
3551            for (fi = min;; fi++)            for (fi = min;; fi++)
3552              {              {
3553              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM32);
3554              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3555                if (fi >= max) MRRETURN(MATCH_NOMATCH);
3556                if (eptr >= md->end_subject)
3557                  {
3558                  SCHECK_PARTIAL();
3559                  MRRETURN(MATCH_NOMATCH);
3560                  }
3561              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
3562              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fc == d) MRRETURN(MATCH_NOMATCH);
               RRETURN(MATCH_NOMATCH);  
3563              }              }
3564            }            }
3565          else          else
# Line 2534  for (;;) Line 3568  for (;;)
3568            {            {
3569            for (fi = min;; fi++)            for (fi = min;; fi++)
3570              {              {
3571              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM33);
3572              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3573              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max) MRRETURN(MATCH_NOMATCH);
3574                RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3575                  {
3576                  SCHECK_PARTIAL();
3577                  MRRETURN(MATCH_NOMATCH);
3578                  }
3579                if (fc == *eptr++) MRRETURN(MATCH_NOMATCH);
3580              }              }
3581            }            }
3582          /* Control never gets here */          /* Control never gets here */
# Line 2557  for (;;) Line 3596  for (;;)
3596            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3597              {              {
3598              int len = 1;              int len = 1;
3599              if (eptr >= md->end_subject) break;              if (eptr >= md->end_subject)
3600                  {
3601                  SCHECK_PARTIAL();
3602                  break;
3603                  }
3604              GETCHARLEN(d, eptr, len);              GETCHARLEN(d, eptr, len);
3605              if (fc == d) break;              if (fc == d) break;
3606              eptr += len;              eptr += len;
# Line 2565  for (;;) Line 3608  for (;;)
3608            if (possessive) continue;            if (possessive) continue;
3609            for(;;)            for(;;)
3610              {              {
3611              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3612              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3613              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
3614              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2577  for (;;) Line 3620  for (;;)
3620            {            {
3621            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3622              {              {
3623              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject)
3624                  {
3625                  SCHECK_PARTIAL();
3626                  break;
3627                  }
3628                if (fc == *eptr) break;
3629              eptr++;              eptr++;
3630              }              }
3631            if (possessive) continue;            if (possessive) continue;
3632            while (eptr >= pp)            while (eptr >= pp)
3633              {              {
3634              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
3635              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3636              eptr--;              eptr--;
3637              }              }
3638            }            }
3639    
3640          RRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
3641          }          }
3642        }        }
3643      /* Control never gets here */      /* Control never gets here */
# Line 2671  for (;;) Line 3719  for (;;)
3719    
3720      /* First, ensure the minimum number of matches are present. Use inline      /* First, ensure the minimum number of matches are present. Use inline
3721      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
3722      (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  
3723      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
3724      and single-bytes. */      and single-bytes. */
3725    
     if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);  
3726      if (min > 0)      if (min > 0)
3727        {        {
3728  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2686  for (;;) Line 3731  for (;;)
3731          switch(prop_type)          switch(prop_type)
3732            {            {
3733            case PT_ANY:            case PT_ANY:
3734            if (prop_fail_result) RRETURN(MATCH_NOMATCH);            if (prop_fail_result) MRRETURN(MATCH_NOMATCH);
3735            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3736              {              {
3737              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3738              GETCHARINC(c, eptr);                {
3739                  SCHECK_PARTIAL();
3740                  MRRETURN(MATCH_NOMATCH);
3741                  }
3742                GETCHARINCTEST(c, eptr);
3743              }              }
3744            break;            break;
3745    
3746            case PT_LAMP:            case PT_LAMP:
3747            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3748              {              {
3749              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              int chartype;
3750              GETCHARINC(c, eptr);              if (eptr >= md->end_subject)
3751              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                {
3752              if ((prop_chartype == ucp_Lu ||                SCHECK_PARTIAL();
3753                   prop_chartype == ucp_Ll ||                MRRETURN(MATCH_NOMATCH);
3754                   prop_chartype == ucp_Lt) == prop_fail_result)                }
3755                RRETURN(MATCH_NOMATCH);              GETCHARINCTEST(c, eptr);
3756                chartype = UCD_CHARTYPE(c);
3757                if ((chartype == ucp_Lu ||
3758                     chartype == ucp_Ll ||
3759                     chartype == ucp_Lt) == prop_fail_result)
3760                  MRRETURN(MATCH_NOMATCH);
3761              }              }
3762            break;            break;
3763    
3764            case PT_GC:            case PT_GC:
3765            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3766              {              {
3767              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3768              GETCHARINC(c, eptr);                {
3769              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3770              if ((prop_category == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
3771                RRETURN(MATCH_NOMATCH);                }
3772                GETCHARINCTEST(c, eptr);
3773                if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result)
3774                  MRRETURN(MATCH_NOMATCH);
3775              }              }
3776            break;            break;
3777    
3778            case PT_PC:            case PT_PC:
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              if ((prop_chartype == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
3785                RRETURN(MATCH_NOMATCH);                }
3786                GETCHARINCTEST(c, eptr);
3787                if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result)
3788                  MRRETURN(MATCH_NOMATCH);
3789              }              }
3790            break;            break;
3791    
3792            case PT_SC:            case PT_SC:
3793            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
3794              {              {
3795              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (eptr >= md->end_subject)
3796              GETCHARINC(c, eptr);                {
3797              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);                SCHECK_PARTIAL();
3798              if ((prop_script == prop_value) == prop_fail_result)                MRRETURN(MATCH_NOMATCH);
3799                RRETURN(MATCH_NOMATCH);                }
3800                GETCHARINCTEST(c, eptr);
3801                if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result)
3802                  MRRETURN(MATCH_NOMATCH);
3803                }
3804              break;
3805    
3806              case PT_ALNUM:
3807              for (i = 1; i <= min; i++)
3808                {
3809                int category;
3810                if (eptr >= md->end_subject)
3811                  {
3812                  SCHECK_PARTIAL();
3813                  MRRETURN(MATCH_NOMATCH);
3814                  }
3815                GETCHARINCTEST(c, eptr);
3816                category = UCD_CATEGORY(c);
3817                if ((category == ucp_L || category == ucp_N) == prop_fail_result)
3818                  MRRETURN(MATCH_NOMATCH);
3819                }
3820              break;
3821    
3822              case PT_SPACE:    /* Perl space */
3823              for (i = 1; i <= min; i++)
3824                {
3825                if (eptr >= md->end_subject)
3826                  {
3827                  SCHECK_PARTIAL();
3828                  MRRETURN(MATCH_NOMATCH);
3829                  }
3830                GETCHARINCTEST(c, eptr);
3831                if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||
3832                     c == CHAR_FF || c == CHAR_CR)
3833                       == prop_fail_result)
3834                  MRRETURN(MATCH_NOMATCH);
3835                }
3836              break;
3837    
3838              case PT_PXSPACE:  /* POSIX space */