/[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 85 by nigel, Sat Feb 24 21:41:13 2007 UTC revision 349 by ph10, Wed Jul 2 18:42:11 2008 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) 1997-2005 University of Cambridge             Copyright (c) 1997-2008 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 */
50    #define PSSTART start_subject  /* Field containing processed string start */
51    #define PSEND   end_subject    /* Field containing processed string end */
52    
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    /* Undefine some potentially clashing cpp symbols */
56    
57  /* Structure for building a chain of data that actually lives on the  #undef min
58  stack, for holding the values of the subject pointer at the start of each  #undef max
 subpattern, so as to detect when an empty string has been matched by a  
 subpattern - to break infinite loops. When NO_RECURSE is set, these blocks  
 are on the heap, not on the stack. */  
   
 typedef struct eptrblock {  
   struct eptrblock *epb_prev;  
   const uschar *epb_saved_eptr;  
 } eptrblock;  
59    
60  /* Flag bits for the match() function */  /* Flag bits for the match() function */
61    
62  #define match_condassert   0x01    /* Called to check a condition assertion */  #define match_condassert     0x01  /* Called to check a condition assertion */
63  #define match_isgroup      0x02    /* Set if start of bracketed group */  #define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */
64    
65  /* Non-error returns from the match() function. Error returns are externally  /* Non-error returns from the match() function. Error returns are externally
66  defined PCRE_ERROR_xxx codes, which are all negative. */  defined PCRE_ERROR_xxx codes, which are all negative. */
# Line 68  defined PCRE_ERROR_xxx codes, which are Line 68  defined PCRE_ERROR_xxx codes, which are
68  #define MATCH_MATCH        1  #define MATCH_MATCH        1
69  #define MATCH_NOMATCH      0  #define MATCH_NOMATCH      0
70    
71    /* Special internal returns from the match() function. Make them sufficiently
72    negative to avoid the external error codes. */
73    
74    #define MATCH_COMMIT       (-999)
75    #define MATCH_PRUNE        (-998)
76    #define MATCH_SKIP         (-997)
77    #define MATCH_THEN         (-996)
78    
79  /* 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.
80  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,
81  because the offset vector is always a multiple of 3 long. */  because the offset vector is always a multiple of 3 long. */
# Line 101  Returns:     nothing Line 109  Returns:     nothing
109  static void  static void
110  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)  pchars(const uschar *p, int length, BOOL is_subject, match_data *md)
111  {  {
112  int c;  unsigned int c;
113  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;  if (is_subject && length > md->end_subject - p) length = md->end_subject - p;
114  while (length-- > 0)  while (length-- > 0)
115    if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c);    if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c);
# Line 128  Returns:      TRUE if matched Line 136  Returns:      TRUE if matched
136  */  */
137    
138  static BOOL  static BOOL
139  match_ref(int offset, register const uschar *eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
140    unsigned long int ims)    unsigned long int ims)
141  {  {
142  const uschar *p = md->start_subject + md->offset_vector[offset];  USPTR p = md->start_subject + md->offset_vector[offset];
143    
144  #ifdef DEBUG  #ifdef DEBUG
145  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 169  return TRUE; Line 177  return TRUE;
177  ****************************************************************************  ****************************************************************************
178                     RECURSION IN THE match() FUNCTION                     RECURSION IN THE match() FUNCTION
179    
180  The match() function is highly recursive. Some regular expressions can cause  The match() function is highly recursive, though not every recursive call
181  it to recurse thousands of times. I was writing for Unix, so I just let it  increases the recursive depth. Nevertheless, some regular expressions can cause
182  call itself recursively. This uses the stack for saving everything that has  it to recurse to a great depth. I was writing for Unix, so I just let it call
183  to be saved for a recursive call. On Unix, the stack can be large, and this  itself recursively. This uses the stack for saving everything that has to be
184  works fine.  saved for a recursive call. On Unix, the stack can be large, and this works
185    fine.
186  It turns out that on non-Unix systems there are problems with programs that  
187  use a lot of stack. (This despite the fact that every last chip has oodles  It turns out that on some non-Unix-like systems there are problems with
188  of memory these days, and techniques for extending the stack have been known  programs that use a lot of stack. (This despite the fact that every last chip
189  for decades.) So....  has oodles of memory these days, and techniques for extending the stack have
190    been known for decades.) So....
191    
192  There is a fudge, triggered by defining NO_RECURSE, which avoids recursive  There is a fudge, triggered by defining NO_RECURSE, which avoids recursive
193  calls by keeping local variables that need to be preserved in blocks of memory  calls by keeping local variables that need to be preserved in blocks of memory
194  obtained from malloc instead instead of on the stack. Macros are used to  obtained from malloc() instead instead of on the stack. Macros are used to
195  achieve this so that the actual code doesn't look very different to what it  achieve this so that the actual code doesn't look very different to what it
196  always used to.  always used to.
197    
198    The original heap-recursive code used longjmp(). However, it seems that this
199    can be very slow on some operating systems. Following a suggestion from Stan
200    Switzer, the use of longjmp() has been abolished, at the cost of having to
201    provide a unique number for each call to RMATCH. There is no way of generating
202    a sequence of numbers at compile time in C. I have given them names, to make
203    them stand out more clearly.
204    
205    Crude tests on x86 Linux show a small speedup of around 5-8%. However, on
206    FreeBSD, avoiding longjmp() more than halves the time taken to run the standard
207    tests. Furthermore, not using longjmp() means that local dynamic variables
208    don't have indeterminate values; this has meant that the frame size can be
209    reduced because the result can be "passed back" by straight setting of the
210    variable instead of being passed in the frame.
211  ****************************************************************************  ****************************************************************************
212  ***************************************************************************/  ***************************************************************************/
213    
214    /* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN
215    below must be updated in sync.  */
216    
217  /* These versions of the macros use the stack, as normal */  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,
218           RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,
219           RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
220           RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
221           RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
222           RM51,  RM52, RM53, RM54 };
223    
224    /* These versions of the macros use the stack, as normal. There are debugging
225    versions and production versions. Note that the "rw" argument of RMATCH isn't
226    actuall used in this definition. */
227    
228  #ifndef NO_RECURSE  #ifndef NO_RECURSE
229  #define REGISTER register  #define REGISTER register
230  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) rx = match(ra,rb,rc,rd,re,rf,rg)  
231    #ifdef DEBUG
232    #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
233      { \
234      printf("match() called in line %d\n", __LINE__); \
235      rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \
236      printf("to line %d\n", __LINE__); \
237      }
238    #define RRETURN(ra) \
239      { \
240      printf("match() returned %d from line %d ", ra, __LINE__); \
241      return ra; \
242      }
243    #else
244    #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
245      rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1)
246  #define RRETURN(ra) return ra  #define RRETURN(ra) return ra
247    #endif
248    
249  #else  #else
250    
251    
252  /* These versions of the macros manage a private stack on the heap. Note  /* These versions of the macros manage a private stack on the heap. Note that
253  that the rd argument of RMATCH isn't actually used. It's the md argument of  the "rd" argument of RMATCH isn't actually used in this definition. It's the md
254  match(), which never changes. */  argument of match(), which never changes. */
255    
256  #define REGISTER  #define REGISTER
257    
258  #define RMATCH(rx,ra,rb,rc,rd,re,rf,rg)\  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\
259    {\    {\
260    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\
261    if (setjmp(frame->Xwhere) == 0)\    frame->Xwhere = rw; \
262      {\    newframe->Xeptr = ra;\
263      newframe->Xeptr = ra;\    newframe->Xecode = rb;\
264      newframe->Xecode = rb;\    newframe->Xmstart = mstart;\
265      newframe->Xoffset_top = rc;\    newframe->Xoffset_top = rc;\
266      newframe->Xims = re;\    newframe->Xims = re;\
267      newframe->Xeptrb = rf;\    newframe->Xeptrb = rf;\
268      newframe->Xflags = rg;\    newframe->Xflags = rg;\
269      newframe->Xprevframe = frame;\    newframe->Xrdepth = frame->Xrdepth + 1;\
270      frame = newframe;\    newframe->Xprevframe = frame;\
271      DPRINTF(("restarting from line %d\n", __LINE__));\    frame = newframe;\
272      goto HEAP_RECURSE;\    DPRINTF(("restarting from line %d\n", __LINE__));\
273      }\    goto HEAP_RECURSE;\
274    else\    L_##rw:\
275      {\    DPRINTF(("jumped back to line %d\n", __LINE__));\
     DPRINTF(("longjumped back to line %d\n", __LINE__));\  
     frame = md->thisframe;\  
     rx = frame->Xresult;\  
     }\  
276    }    }
277    
278  #define RRETURN(ra)\  #define RRETURN(ra)\
# Line 235  match(), which never changes. */ Line 282  match(), which never changes. */
282    (pcre_stack_free)(newframe);\    (pcre_stack_free)(newframe);\
283    if (frame != NULL)\    if (frame != NULL)\
284      {\      {\
285      frame->Xresult = ra;\      rrc = ra;\
286      md->thisframe = frame;\      goto HEAP_RETURN;\
     longjmp(frame->Xwhere, 1);\  
287      }\      }\
288    return ra;\    return ra;\
289    }    }
# Line 252  typedef struct heapframe { Line 298  typedef struct heapframe {
298    
299    const uschar *Xeptr;    const uschar *Xeptr;
300    const uschar *Xecode;    const uschar *Xecode;
301      const uschar *Xmstart;
302    int Xoffset_top;    int Xoffset_top;
303    long int Xims;    long int Xims;
304    eptrblock *Xeptrb;    eptrblock *Xeptrb;
305    int Xflags;    int Xflags;
306      unsigned int Xrdepth;
307    
308    /* Function local variables */    /* Function local variables */
309    
# Line 271  typedef struct heapframe { Line 319  typedef struct heapframe {
319    
320    BOOL Xcur_is_word;    BOOL Xcur_is_word;
321    BOOL Xcondition;    BOOL Xcondition;
   BOOL Xminimize;  
322    BOOL Xprev_is_word;    BOOL Xprev_is_word;
323    
324    unsigned long int Xoriginal_ims;    unsigned long int Xoriginal_ims;
325    
326  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
327    int Xprop_type;    int Xprop_type;
328      int Xprop_value;
329    int Xprop_fail_result;    int Xprop_fail_result;
330    int Xprop_category;    int Xprop_category;
331    int Xprop_chartype;    int Xprop_chartype;
332    int Xprop_othercase;    int Xprop_script;
333    int Xprop_test_against;    int Xoclength;
334    int *Xprop_test_variable;    uschar Xocchars[8];
335  #endif  #endif
336    
337    int Xctype;    int Xctype;
338    int Xfc;    unsigned int Xfc;
339    int Xfi;    int Xfi;
340    int Xlength;    int Xlength;
341    int Xmax;    int Xmax;
# Line 301  typedef struct heapframe { Line 349  typedef struct heapframe {
349    
350    eptrblock Xnewptrb;    eptrblock Xnewptrb;
351    
352    /* Place to pass back result, and where to jump back to */    /* Where to jump back to */
353    
354    int  Xresult;    int Xwhere;
   jmp_buf Xwhere;  
355    
356  } heapframe;  } heapframe;
357    
# Line 320  typedef struct heapframe { Line 367  typedef struct heapframe {
367  *         Match from current position            *  *         Match from current position            *
368  *************************************************/  *************************************************/
369    
370  /* On entry ecode points to the first opcode, and eptr to the first character  /* This function is called recursively in many circumstances. Whenever it
 in the subject string, while eptrb holds the value of eptr at the start of the  
 last bracketed group - used for breaking infinite loops matching zero-length  
 strings. This function is called recursively in many circumstances. Whenever it  
371  returns a negative (error) response, the outer incarnation must also return the  returns a negative (error) response, the outer incarnation must also return the
372  same response.  same response.
373    
# Line 333  performance. Tests using gcc on a SPARC Line 377  performance. Tests using gcc on a SPARC
377  made performance worse.  made performance worse.
378    
379  Arguments:  Arguments:
380     eptr        pointer in subject     eptr        pointer to current character in subject
381     ecode       position in code     ecode       pointer to current position in compiled code
382       mstart      pointer to the current match start position (can be modified
383                     by encountering \K)
384     offset_top  current top pointer     offset_top  current top pointer
385     md          pointer to "static" info for the match     md          pointer to "static" info for the match
386     ims         current /i, /m, and /s options     ims         current /i, /m, and /s options
# Line 342  Arguments: Line 388  Arguments:
388                   brackets - for testing for empty matches                   brackets - for testing for empty matches
389     flags       can contain     flags       can contain
390                   match_condassert - this is an assertion condition                   match_condassert - this is an assertion condition
391                   match_isgroup - this is the start of a bracketed group                   match_cbegroup - this is the start of an unlimited repeat
392                       group that can match an empty string
393       rdepth      the recursion depth
394    
395  Returns:       MATCH_MATCH if matched            )  these values are >= 0  Returns:       MATCH_MATCH if matched            )  these values are >= 0
396                 MATCH_NOMATCH if failed to match  )                 MATCH_NOMATCH if failed to match  )
397                 a negative PCRE_ERROR_xxx value if aborted by an error condition                 a negative PCRE_ERROR_xxx value if aborted by an error condition
398                   (e.g. stopped by recursion limit)                   (e.g. stopped by repeated call or recursion limit)
399  */  */
400    
401  static int  static int
402  match(REGISTER const uschar *eptr, REGISTER const uschar *ecode,  match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart,
403    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,    int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,
404    int flags)    int flags, unsigned int rdepth)
405  {  {
406  /* 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,
407  so they can be ordinary variables in all cases. Mark them with "register"  so they can be ordinary variables in all cases. Mark some of them with
408  because they are used a lot in loops. */  "register" because they are used a lot in loops. */
409    
410  register int  rrc;    /* Returns from recursive calls */  register int  rrc;         /* Returns from recursive calls */
411  register int  i;      /* Used for loops not involving calls to RMATCH() */  register int  i;           /* Used for loops not involving calls to RMATCH() */
412  register int  c;      /* Character values not kept over RMATCH() calls */  register unsigned int c;   /* Character values not kept over RMATCH() calls */
413  register BOOL utf8;   /* Local copy of UTF-8 flag for speed */  register BOOL utf8;        /* Local copy of UTF-8 flag for speed */
414    
415    BOOL minimize, possessive; /* Quantifier options */
416    
417  /* 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
418  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 377  frame->Xprevframe = NULL;            /* Line 427  frame->Xprevframe = NULL;            /*
427    
428  frame->Xeptr = eptr;  frame->Xeptr = eptr;
429  frame->Xecode = ecode;  frame->Xecode = ecode;
430    frame->Xmstart = mstart;
431  frame->Xoffset_top = offset_top;  frame->Xoffset_top = offset_top;
432  frame->Xims = ims;  frame->Xims = ims;
433  frame->Xeptrb = eptrb;  frame->Xeptrb = eptrb;
434  frame->Xflags = flags;  frame->Xflags = flags;
435    frame->Xrdepth = rdepth;
436    
437  /* This is where control jumps back to to effect "recursion" */  /* This is where control jumps back to to effect "recursion" */
438    
# Line 390  HEAP_RECURSE: Line 442  HEAP_RECURSE:
442    
443  #define eptr               frame->Xeptr  #define eptr               frame->Xeptr
444  #define ecode              frame->Xecode  #define ecode              frame->Xecode
445    #define mstart             frame->Xmstart
446  #define offset_top         frame->Xoffset_top  #define offset_top         frame->Xoffset_top
447  #define ims                frame->Xims  #define ims                frame->Xims
448  #define eptrb              frame->Xeptrb  #define eptrb              frame->Xeptrb
449  #define flags              frame->Xflags  #define flags              frame->Xflags
450    #define rdepth             frame->Xrdepth
451    
452  /* Ditto for the local variables */  /* Ditto for the local variables */
453    
# Line 411  HEAP_RECURSE: Line 465  HEAP_RECURSE:
465    
466  #define cur_is_word        frame->Xcur_is_word  #define cur_is_word        frame->Xcur_is_word
467  #define condition          frame->Xcondition  #define condition          frame->Xcondition
 #define minimize           frame->Xminimize  
468  #define prev_is_word       frame->Xprev_is_word  #define prev_is_word       frame->Xprev_is_word
469    
470  #define original_ims       frame->Xoriginal_ims  #define original_ims       frame->Xoriginal_ims
471    
472  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
473  #define prop_type          frame->Xprop_type  #define prop_type          frame->Xprop_type
474    #define prop_value         frame->Xprop_value
475  #define prop_fail_result   frame->Xprop_fail_result  #define prop_fail_result   frame->Xprop_fail_result
476  #define prop_category      frame->Xprop_category  #define prop_category      frame->Xprop_category
477  #define prop_chartype      frame->Xprop_chartype  #define prop_chartype      frame->Xprop_chartype
478  #define prop_othercase     frame->Xprop_othercase  #define prop_script        frame->Xprop_script
479  #define prop_test_against  frame->Xprop_test_against  #define oclength           frame->Xoclength
480  #define prop_test_variable frame->Xprop_test_variable  #define occhars            frame->Xocchars
481  #endif  #endif
482    
483  #define ctype              frame->Xctype  #define ctype              frame->Xctype
# Line 447  HEAP_RECURSE: Line 501  HEAP_RECURSE:
501  get preserved during recursion in the normal way. In this environment, fi and  get preserved during recursion in the normal way. In this environment, fi and
502  i, and fc and c, can be the same variables. */  i, and fc and c, can be the same variables. */
503    
504  #else  #else         /* NO_RECURSE not defined */
505  #define fi i  #define fi i
506  #define fc c  #define fc c
507    
508    
509  #ifdef SUPPORT_UTF8                /* Many of these variables are used ony */  #ifdef SUPPORT_UTF8                /* Many of these variables are used only  */
510  const uschar *charptr;             /* small blocks of the code. My normal  */  const uschar *charptr;             /* in small blocks of the code. My normal */
511  #endif                             /* style of coding would have declared  */  #endif                             /* style of coding would have declared    */
512  const uschar *callpat;             /* them within each of those blocks.    */  const uschar *callpat;             /* them within each of those blocks.      */
513  const uschar *data;                /* However, in order to accommodate the */  const uschar *data;                /* However, in order to accommodate the   */
514  const uschar *next;                /* version of this code that uses an    */  const uschar *next;                /* version of this code that uses an      */
515  const uschar *pp;                  /* external "stack" implemented on the  */  USPTR         pp;                  /* external "stack" implemented on the    */
516  const uschar *prev;                /* heap, it is easier to declare them   */  const uschar *prev;                /* heap, it is easier to declare them all */
517  const uschar *saved_eptr;          /* all here, so the declarations can    */  USPTR         saved_eptr;          /* here, so the declarations can be cut   */
518                                     /* be cut out in a block. The only      */                                     /* out in a block. The only declarations  */
519  recursion_info new_recursive;      /* declarations within blocks below are */  recursion_info new_recursive;      /* within blocks below are for variables  */
520                                     /* for variables that do not have to    */                                     /* that do not have to be preserved over  */
521  BOOL cur_is_word;                  /* be preserved over a recursive call   */  BOOL cur_is_word;                  /* a recursive call to RMATCH().          */
522  BOOL condition;                    /* to RMATCH().                         */  BOOL condition;
 BOOL minimize;  
523  BOOL prev_is_word;  BOOL prev_is_word;
524    
525  unsigned long int original_ims;  unsigned long int original_ims;
526    
527  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
528  int prop_type;  int prop_type;
529    int prop_value;
530  int prop_fail_result;  int prop_fail_result;
531  int prop_category;  int prop_category;
532  int prop_chartype;  int prop_chartype;
533  int prop_othercase;  int prop_script;
534  int prop_test_against;  int oclength;
535  int *prop_test_variable;  uschar occhars[8];
536  #endif  #endif
537    
538  int ctype;  int ctype;
# Line 493  int save_offset1, save_offset2, save_off Line 547  int save_offset1, save_offset2, save_off
547  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
548    
549  eptrblock newptrb;  eptrblock newptrb;
550  #endif  #endif     /* NO_RECURSE */
551    
552  /* These statements are here to stop the compiler complaining about unitialized  /* These statements are here to stop the compiler complaining about unitialized
553  variables. */  variables. */
554    
555  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
556    prop_value = 0;
557  prop_fail_result = 0;  prop_fail_result = 0;
 prop_test_against = 0;  
 prop_test_variable = NULL;  
558  #endif  #endif
559    
560  /* OK, now we can get on with the real code of the function. Recursion is  
561  specified by the macros RMATCH and RRETURN. When NO_RECURSE is *not* defined,  /* This label is used for tail recursion, which is used in a few cases even
562  these just turn into a recursive call to match() and a "return", respectively.  when NO_RECURSE is not defined, in order to reduce the amount of stack that is
563  However, RMATCH isn't like a function call because it's quite a complicated  used. Thanks to Ian Taylor for noticing this possibility and sending the
564  macro. It has to be used in one particular way. This shouldn't, however, impact  original patch. */
565  performance when true recursion is being used. */  
566    TAIL_RECURSE:
567    
568    /* OK, now we can get on with the real code of the function. Recursive calls
569    are specified by the macro RMATCH and RRETURN is used to return. When
570    NO_RECURSE is *not* defined, these just turn into a recursive call to match()
571    and a "return", respectively (possibly with some debugging if DEBUG is
572    defined). However, RMATCH isn't like a function call because it's quite a
573    complicated macro. It has to be used in one particular way. This shouldn't,
574    however, impact performance when true recursion is being used. */
575    
576    #ifdef SUPPORT_UTF8
577    utf8 = md->utf8;       /* Local copy of the flag */
578    #else
579    utf8 = FALSE;
580    #endif
581    
582    /* First check that we haven't called match() too many times, or that we
583    haven't exceeded the recursive call limit. */
584    
585  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);  if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
586    if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);
587    
588  original_ims = ims;    /* Save for resetting on ')' */  original_ims = ims;    /* Save for resetting on ')' */
 utf8 = md->utf8;       /* Local copy of the flag */  
589    
590  /* At the start of a bracketed group, add the current subject pointer to the  /* At the start of a group with an unlimited repeat that may match an empty
591  stack of such pointers, to be re-instated at the end of the group when we hit  string, the match_cbegroup flag is set. When this is the case, add the current
592  the closing ket. When match() is called in other circumstances, we don't add to  subject pointer to the chain of such remembered pointers, to be checked when we
593  this stack. */  hit the closing ket, in order to break infinite loops that match no characters.
594    When match() is called in other circumstances, don't add to the chain. The
595    match_cbegroup flag must NOT be used with tail recursion, because the memory
596    block that is used is on the stack, so a new one may be required for each
597    match(). */
598    
599  if ((flags & match_isgroup) != 0)  if ((flags & match_cbegroup) != 0)
600    {    {
   newptrb.epb_prev = eptrb;  
601    newptrb.epb_saved_eptr = eptr;    newptrb.epb_saved_eptr = eptr;
602      newptrb.epb_prev = eptrb;
603    eptrb = &newptrb;    eptrb = &newptrb;
604    }    }
605    
606  /* Now start processing the operations. */  /* Now start processing the opcodes. */
607    
608  for (;;)  for (;;)
609    {    {
610      minimize = possessive = FALSE;
611    op = *ecode;    op = *ecode;
   minimize = FALSE;  
612    
613    /* For partial matching, remember if we ever hit the end of the subject after    /* For partial matching, remember if we ever hit the end of the subject after
614    matching at least one subject character. */    matching at least one subject character. */
615    
616    if (md->partial &&    if (md->partial &&
617        eptr >= md->end_subject &&        eptr >= md->end_subject &&
618        eptr > md->start_match)        eptr > mstart)
619      md->hitend = TRUE;      md->hitend = TRUE;
620    
621    /* Opening capturing bracket. If there is space in the offset vector, save    switch(op)
   the current subject position in the working slot at the top of the vector. We  
   mustn't change the current values of the data slot, because they may be set  
   from a previous iteration of this group, and be referred to by a reference  
   inside the group.  
   
   If the bracket fails to match, we need to restore this value and also the  
   values of the final offsets, in case they were set by a previous iteration of  
   the same bracket.  
   
   If there isn't enough space in the offset vector, treat this as if it were a  
   non-capturing bracket. Don't worry about setting the flag for the error case  
   here; that is handled in the code for KET. */  
   
   if (op > OP_BRA)  
622      {      {
623      number = op - OP_BRA;      case OP_FAIL:
624        RRETURN(MATCH_NOMATCH);
     /* For extended extraction brackets (large number), we have to fish out the  
     number from a dummy opcode at the start. */  
625    
626      if (number > EXTRACT_BASIC_MAX)      case OP_PRUNE:
627        number = GET2(ecode, 2+LINK_SIZE);      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
628          ims, eptrb, flags, RM51);
629        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
630        RRETURN(MATCH_PRUNE);
631    
632        case OP_COMMIT:
633        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
634          ims, eptrb, flags, RM52);
635        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
636        RRETURN(MATCH_COMMIT);
637    
638        case OP_SKIP:
639        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
640          ims, eptrb, flags, RM53);
641        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
642        md->start_match_ptr = eptr;   /* Pass back current position */
643        RRETURN(MATCH_SKIP);
644    
645        case OP_THEN:
646        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
647          ims, eptrb, flags, RM54);
648        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
649        RRETURN(MATCH_THEN);
650    
651        /* Handle a capturing bracket. If there is space in the offset vector, save
652        the current subject position in the working slot at the top of the vector.
653        We mustn't change the current values of the data slot, because they may be
654        set from a previous iteration of this group, and be referred to by a
655        reference inside the group.
656    
657        If the bracket fails to match, we need to restore this value and also the
658        values of the final offsets, in case they were set by a previous iteration
659        of the same bracket.
660    
661        If there isn't enough space in the offset vector, treat this as if it were
662        a non-capturing bracket. Don't worry about setting the flag for the error
663        case here; that is handled in the code for KET. */
664    
665        case OP_CBRA:
666        case OP_SCBRA:
667        number = GET2(ecode, 1+LINK_SIZE);
668      offset = number << 1;      offset = number << 1;
669    
670  #ifdef DEBUG  #ifdef DEBUG
671      printf("start bracket %d subject=", number);      printf("start bracket %d\n", number);
672        printf("subject=");
673      pchars(eptr, 16, TRUE, md);      pchars(eptr, 16, TRUE, md);
674      printf("\n");      printf("\n");
675  #endif  #endif
# Line 584  for (;;) Line 684  for (;;)
684        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));        DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
685        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;        md->offset_vector[md->offset_end - number] = eptr - md->start_subject;
686    
687          flags = (op == OP_SCBRA)? match_cbegroup : 0;
688        do        do
689          {          {
690          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
691            match_isgroup);            ims, eptrb, flags, RM1);
692          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
693          md->capture_last = save_capture_last;          md->capture_last = save_capture_last;
694          ecode += GET(ecode, 1);          ecode += GET(ecode, 1);
695          }          }
# Line 603  for (;;) Line 704  for (;;)
704        RRETURN(MATCH_NOMATCH);        RRETURN(MATCH_NOMATCH);
705        }        }
706    
707      /* Insufficient room for saving captured contents */      /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
708        as a non-capturing bracket. */
709    
710      else op = OP_BRA;      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
711      }      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
712    
713    /* Other types of node can be handled by a switch */      DPRINTF(("insufficient capture room: treat as non-capturing\n"));
714    
715    switch(op)      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
716      {      /* VVVVVVVVVVVVVVVVVVVVVVVVV */
717      case OP_BRA:     /* Non-capturing bracket: optimized */  
718      DPRINTF(("start bracket 0\n"));      /* Non-capturing bracket. Loop for all the alternatives. When we get to the
719      do      final alternative within the brackets, we would return the result of a
720        recursive call to match() whatever happened. We can reduce stack usage by
721        turning this into a tail recursion, except in the case when match_cbegroup
722        is set.*/
723    
724        case OP_BRA:
725        case OP_SBRA:
726        DPRINTF(("start non-capturing bracket\n"));
727        flags = (op >= OP_SBRA)? match_cbegroup : 0;
728        for (;;)
729        {        {
730        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,        if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */
731          match_isgroup);          {
732        if (rrc != MATCH_NOMATCH) RRETURN(rrc);          if (flags == 0)    /* Not a possibly empty group */
733              {
734              ecode += _pcre_OP_lengths[*ecode];
735              DPRINTF(("bracket 0 tail recursion\n"));
736              goto TAIL_RECURSE;
737              }
738    
739            /* Possibly empty group; can't use tail recursion. */
740    
741            RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,
742              eptrb, flags, RM48);
743            RRETURN(rrc);
744            }
745    
746          /* For non-final alternatives, continue the loop for a NOMATCH result;
747          otherwise return. */
748    
749          RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,
750            eptrb, flags, RM2);
751          if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
752        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
753        }        }
754      while (*ecode == OP_ALT);      /* Control never reaches here. */
     DPRINTF(("bracket 0 failed\n"));  
     RRETURN(MATCH_NOMATCH);  
755    
756      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than
757      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
758      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
759      exactly what going to the ket would do. */      exactly what going to the ket would do. As there is only one branch to be
760        obeyed, we can use tail recursion to avoid using another stack frame. */
761    
762      case OP_COND:      case OP_COND:
763      if (ecode[LINK_SIZE+1] == OP_CREF) /* Condition extract or recurse test */      case OP_SCOND:
764        if (ecode[LINK_SIZE+1] == OP_RREF)         /* Recursion test */
765          {
766          offset = GET2(ecode, LINK_SIZE + 2);     /* Recursion group number*/
767          condition = md->recursive != NULL &&
768            (offset == RREF_ANY || offset == md->recursive->group_num);
769          ecode += condition? 3 : GET(ecode, 1);
770          }
771    
772        else if (ecode[LINK_SIZE+1] == OP_CREF)    /* Group used test */
773        {        {
774        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */        offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
775        condition = (offset == CREF_RECURSE * 2)?        condition = offset < offset_top && md->offset_vector[offset] >= 0;
776          (md->recursive != NULL) :        ecode += condition? 3 : GET(ecode, 1);
777          (offset < offset_top && md->offset_vector[offset] >= 0);        }
778        RMATCH(rrc, eptr, ecode + (condition?  
779          (LINK_SIZE + 4) : (LINK_SIZE + 1 + GET(ecode, 1))),      else if (ecode[LINK_SIZE+1] == OP_DEF)     /* DEFINE - always false */
780          offset_top, md, ims, eptrb, match_isgroup);        {
781        RRETURN(rrc);        condition = FALSE;
782          ecode += GET(ecode, 1);
783        }        }
784    
785      /* The condition is an assertion. Call match() to evaluate it - setting      /* The condition is an assertion. Call match() to evaluate it - setting
786      the final argument TRUE causes it to stop at the end of an assertion. */      the final argument match_condassert causes it to stop at the end of an
787        assertion. */
788    
789      else      else
790        {        {
791        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,
792            match_condassert | match_isgroup);            match_condassert, RM3);
793        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
794          {          {
795          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE+2);          condition = TRUE;
796            ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
797          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
798          }          }
799        else if (rrc != MATCH_NOMATCH)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
800          {          {
801          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
802          }          }
803        else ecode += GET(ecode, 1);        else
804        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,          {
805          match_isgroup);          condition = FALSE;
806        RRETURN(rrc);          ecode += GET(ecode, 1);
807            }
808        }        }
     /* Control never reaches here */  
809    
810      /* Skip over conditional reference or large extraction number data if      /* We are now at the branch that is to be obeyed. As there is only one,
811      encountered. */      we can use tail recursion to avoid using another stack frame, except when
812        match_cbegroup is required for an unlimited repeat of a possibly empty
813        group. If the second alternative doesn't exist, we can just plough on. */
814    
815      case OP_CREF:      if (condition || *ecode == OP_ALT)
816      case OP_BRANUMBER:        {
817      ecode += 3;        ecode += 1 + LINK_SIZE;
818          if (op == OP_SCOND)        /* Possibly empty group */
819            {
820            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);
821            RRETURN(rrc);
822            }
823          else                       /* Group must match something */
824            {
825            flags = 0;
826            goto TAIL_RECURSE;
827            }
828          }
829        else                         /* Condition false & no 2nd alternative */
830          {
831          ecode += 1 + LINK_SIZE;
832          }
833      break;      break;
834    
     /* End of the pattern. If we are in a recursion, we should restore the  
     offsets appropriately and continue from after the call. */  
835    
836        /* End of the pattern, either real or forced. If we are in a top-level
837        recursion, we should restore the offsets appropriately and continue from
838        after the call. */
839    
840        case OP_ACCEPT:
841      case OP_END:      case OP_END:
842      if (md->recursive != NULL && md->recursive->group_num == 0)      if (md->recursive != NULL && md->recursive->group_num == 0)
843        {        {
844        recursion_info *rec = md->recursive;        recursion_info *rec = md->recursive;
845        DPRINTF(("Hit the end in a (?0) recursion\n"));        DPRINTF(("End of pattern in a (?0) recursion\n"));
846        md->recursive = rec->prevrec;        md->recursive = rec->prevrec;
847        memmove(md->offset_vector, rec->offset_save,        memmove(md->offset_vector, rec->offset_save,
848          rec->saved_max * sizeof(int));          rec->saved_max * sizeof(int));
849        md->start_match = rec->save_start;        mstart = rec->save_start;
850        ims = original_ims;        ims = original_ims;
851        ecode = rec->after_call;        ecode = rec->after_call;
852        break;        break;
# Line 694  for (;;) Line 855  for (;;)
855      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty      /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty
856      string - backtracking will then try other alternatives, if any. */      string - backtracking will then try other alternatives, if any. */
857    
858      if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH);      if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH);
859      md->end_match_ptr = eptr;          /* Record where we ended */      md->end_match_ptr = eptr;           /* Record where we ended */
860      md->end_offset_top = offset_top;   /* and how many extracts were taken */      md->end_offset_top = offset_top;    /* and how many extracts were taken */
861        md->start_match_ptr = mstart;       /* and the start (\K can modify) */
862      RRETURN(MATCH_MATCH);      RRETURN(MATCH_MATCH);
863    
864      /* Change option settings */      /* Change option settings */
# Line 717  for (;;) Line 879  for (;;)
879      case OP_ASSERTBACK:      case OP_ASSERTBACK:
880      do      do
881        {        {
882        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
883          match_isgroup);          RM4);
884        if (rrc == MATCH_MATCH) break;        if (rrc == MATCH_MATCH) break;
885        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
886        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
887        }        }
888      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
# Line 744  for (;;) Line 906  for (;;)
906      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
907      do      do
908        {        {
909        RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
910          match_isgroup);          RM5);
911        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);
912        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
913        ecode += GET(ecode,1);        ecode += GET(ecode,1);
914        }        }
915      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
# Line 766  for (;;) Line 928  for (;;)
928  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
929      if (utf8)      if (utf8)
930        {        {
931        c = GET(ecode,1);        i = GET(ecode, 1);
932        for (i = 0; i < c; i++)        while (i-- > 0)
933          {          {
934          eptr--;          eptr--;
935          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);          if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
936          BACKCHAR(eptr)          BACKCHAR(eptr);
937          }          }
938        }        }
939      else      else
# Line 780  for (;;) Line 942  for (;;)
942      /* No UTF-8 support, or not in UTF-8 mode: count is byte count */      /* No UTF-8 support, or not in UTF-8 mode: count is byte count */
943    
944        {        {
945        eptr -= GET(ecode,1);        eptr -= GET(ecode, 1);
946        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
947        }        }
948    
# Line 800  for (;;) Line 962  for (;;)
962        cb.version          = 1;   /* Version 1 of the callout block */        cb.version          = 1;   /* Version 1 of the callout block */
963        cb.callout_number   = ecode[1];        cb.callout_number   = ecode[1];
964        cb.offset_vector    = md->offset_vector;        cb.offset_vector    = md->offset_vector;
965        cb.subject          = (const char *)md->start_subject;        cb.subject          = (PCRE_SPTR)md->start_subject;
966        cb.subject_length   = md->end_subject - md->start_subject;        cb.subject_length   = md->end_subject - md->start_subject;
967        cb.start_match      = md->start_match - md->start_subject;        cb.start_match      = mstart - md->start_subject;
968        cb.current_position = eptr - md->start_subject;        cb.current_position = eptr - md->start_subject;
969        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
970        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
# Line 837  for (;;) Line 999  for (;;)
999      case OP_RECURSE:      case OP_RECURSE:
1000        {        {
1001        callpat = md->start_code + GET(ecode, 1);        callpat = md->start_code + GET(ecode, 1);
1002        new_recursive.group_num = *callpat - OP_BRA;        new_recursive.group_num = (callpat == md->start_code)? 0 :
1003            GET2(callpat, 1 + LINK_SIZE);
       /* For extended extraction brackets (large number), we have to fish out  
       the number from a dummy opcode at the start. */  
   
       if (new_recursive.group_num > EXTRACT_BASIC_MAX)  
         new_recursive.group_num = GET2(callpat, 2+LINK_SIZE);  
1004    
1005        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1006    
# Line 869  for (;;) Line 1026  for (;;)
1026    
1027        memcpy(new_recursive.offset_save, md->offset_vector,        memcpy(new_recursive.offset_save, md->offset_vector,
1028              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1029        new_recursive.save_start = md->start_match;        new_recursive.save_start = mstart;
1030        md->start_match = eptr;        mstart = eptr;
1031    
1032        /* OK, now we can do the recursion. For each top-level alternative we        /* OK, now we can do the recursion. For each top-level alternative we
1033        restore the offset and recursion data. */        restore the offset and recursion data. */
1034    
1035        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1036          flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;
1037        do        do
1038          {          {
1039          RMATCH(rrc, eptr, callpat + 1 + LINK_SIZE, offset_top, md, ims,          RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
1040              eptrb, match_isgroup);            md, ims, eptrb, flags, RM6);
1041          if (rrc == MATCH_MATCH)          if (rrc == MATCH_MATCH)
1042            {            {
1043              DPRINTF(("Recursion matched\n"));
1044            md->recursive = new_recursive.prevrec;            md->recursive = new_recursive.prevrec;
1045            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
1046              (pcre_free)(new_recursive.offset_save);              (pcre_free)(new_recursive.offset_save);
1047            RRETURN(MATCH_MATCH);            RRETURN(MATCH_MATCH);
1048            }            }
1049          else if (rrc != MATCH_NOMATCH) RRETURN(rrc);          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1050              {
1051              DPRINTF(("Recursion gave error %d\n", rrc));
1052              RRETURN(rrc);
1053              }
1054    
1055          md->recursive = &new_recursive;          md->recursive = &new_recursive;
1056          memcpy(md->offset_vector, new_recursive.offset_save,          memcpy(md->offset_vector, new_recursive.offset_save,
# Line 912  for (;;) Line 1075  for (;;)
1075      the end of a normal bracket, leaving the subject pointer. */      the end of a normal bracket, leaving the subject pointer. */
1076    
1077      case OP_ONCE:      case OP_ONCE:
1078        {      prev = ecode;
1079        prev = ecode;      saved_eptr = eptr;
       saved_eptr = eptr;  
1080    
1081        do      do
1082          {        {
1083          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);
1084            eptrb, match_isgroup);        if (rrc == MATCH_MATCH) break;
1085          if (rrc == MATCH_MATCH) break;        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
1086          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        ecode += GET(ecode,1);
1087          ecode += GET(ecode,1);        }
1088          }      while (*ecode == OP_ALT);
       while (*ecode == OP_ALT);  
1089    
1090        /* If hit the end of the group (which could be repeated), fail */      /* If hit the end of the group (which could be repeated), fail */
1091    
1092        if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);      if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);
1093    
1094        /* Continue as from after the assertion, updating the offsets high water      /* Continue as from after the assertion, updating the offsets high water
1095        mark, since extracts may have been taken. */      mark, since extracts may have been taken. */
1096    
1097        do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
1098    
1099        offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1100        eptr = md->end_match_ptr;      eptr = md->end_match_ptr;
1101    
1102        /* For a non-repeating ket, just continue at this level. This also      /* For a non-repeating ket, just continue at this level. This also
1103        happens for a repeating ket if no characters were matched in the group.      happens for a repeating ket if no characters were matched in the group.
1104        This is the forcible breaking of infinite loops as implemented in Perl      This is the forcible breaking of infinite loops as implemented in Perl
1105        5.005. If there is an options reset, it will get obeyed in the normal      5.005. If there is an options reset, it will get obeyed in the normal
1106        course of events. */      course of events. */
1107    
1108        if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1109          {        {
1110          ecode += 1+LINK_SIZE;        ecode += 1+LINK_SIZE;
1111          break;        break;
1112          }        }
1113    
1114        /* The repeating kets try the rest of the pattern or restart from the      /* The repeating kets try the rest of the pattern or restart from the
1115        preceding bracket, in the appropriate order. We need to reset any options      preceding bracket, in the appropriate order. The second "call" of match()
1116        that changed within the bracket before re-running it, so check the next      uses tail recursion, to avoid using another stack frame. We need to reset
1117        opcode. */      any options that changed within the bracket before re-running it, so
1118        check the next opcode. */
1119    
1120        if (ecode[1+LINK_SIZE] == OP_OPT)      if (ecode[1+LINK_SIZE] == OP_OPT)
1121          {        {
1122          ims = (ims & ~PCRE_IMS) | ecode[4];        ims = (ims & ~PCRE_IMS) | ecode[4];
1123          DPRINTF(("ims set to %02lx at group repeat\n", ims));        DPRINTF(("ims set to %02lx at group repeat\n", ims));
1124          }        }
1125    
1126        if (*ecode == OP_KETRMIN)      if (*ecode == OP_KETRMIN)
1127          {        {
1128          RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);
1129          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1130          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);        ecode = prev;
1131          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        flags = 0;
1132          }        goto TAIL_RECURSE;
       else  /* OP_KETRMAX */  
         {  
         RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
         }  
1133        }        }
1134      RRETURN(MATCH_NOMATCH);      else  /* OP_KETRMAX */
1135          {
1136          RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);
1137          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1138          ecode += 1 + LINK_SIZE;
1139          flags = 0;
1140          goto TAIL_RECURSE;
1141          }
1142        /* Control never gets here */
1143    
1144      /* An alternation is the end of a branch; scan along to find the end of the      /* An alternation is the end of a branch; scan along to find the end of the
1145      bracketed group and go to there. */      bracketed group and go to there. */
# Line 985  for (;;) Line 1148  for (;;)
1148      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1149      break;      break;
1150    
1151      /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating      /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
1152      that it may occur zero times. It may repeat infinitely, or not at all -      indicating that it may occur zero times. It may repeat infinitely, or not
1153      i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper      at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
1154      repeat limits are compiled as a number of copies, with the optional ones      with fixed upper repeat limits are compiled as a number of copies, with the
1155      preceded by BRAZERO or BRAMINZERO. */      optional ones preceded by BRAZERO or BRAMINZERO. */
1156    
1157      case OP_BRAZERO:      case OP_BRAZERO:
1158        {        {
1159        next = ecode+1;        next = ecode+1;
1160        RMATCH(rrc, eptr, next, offset_top, md, ims, eptrb, match_isgroup);        RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);
1161        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1162        do next += GET(next,1); while (*next == OP_ALT);        do next += GET(next,1); while (*next == OP_ALT);
1163        ecode = next + 1+LINK_SIZE;        ecode = next + 1 + LINK_SIZE;
1164        }        }
1165      break;      break;
1166    
1167      case OP_BRAMINZERO:      case OP_BRAMINZERO:
1168        {        {
1169        next = ecode+1;        next = ecode+1;
1170        do next += GET(next,1); while (*next == OP_ALT);        do next += GET(next, 1); while (*next == OP_ALT);
1171        RMATCH(rrc, eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb,        RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);
         match_isgroup);  
1172        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1173        ecode++;        ecode++;
1174        }        }
1175      break;      break;
1176    
1177      /* End of a group, repeated or non-repeating. If we are at the end of      case OP_SKIPZERO:
1178      an assertion "group", stop matching and return MATCH_MATCH, but record the        {
1179      current high water mark for use by positive assertions. Do this also        next = ecode+1;
1180      for the "once" (not-backup up) groups. */        do next += GET(next,1); while (*next == OP_ALT);
1181          ecode = next + 1 + LINK_SIZE;
1182          }
1183        break;
1184    
1185        /* End of a group, repeated or non-repeating. */
1186    
1187      case OP_KET:      case OP_KET:
1188      case OP_KETRMIN:      case OP_KETRMIN:
1189      case OP_KETRMAX:      case OP_KETRMAX:
1190        {      prev = ecode - GET(ecode, 1);
       prev = ecode - GET(ecode, 1);  
       saved_eptr = eptrb->epb_saved_eptr;  
1191    
1192        /* Back up the stack of bracket start pointers. */      /* If this was a group that remembered the subject start, in order to break
1193        infinite repeats of empty string matches, retrieve the subject start from
1194        the chain. Otherwise, set it NULL. */
1195    
1196        eptrb = eptrb->epb_prev;      if (*prev >= OP_SBRA)
1197          {
1198        if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||        saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
1199            *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||        eptrb = eptrb->epb_prev;              /* Backup to previous group */
1200            *prev == OP_ONCE)        }
1201          {      else saved_eptr = NULL;
         md->end_match_ptr = eptr;      /* For ONCE */  
         md->end_offset_top = offset_top;  
         RRETURN(MATCH_MATCH);  
         }  
1202    
1203        /* In all other cases except a conditional group we have to check the      /* If we are at the end of an assertion group, stop matching and return
1204        group number back at the start and if necessary complete handling an      MATCH_MATCH, but record the current high water mark for use by positive
1205        extraction by setting the offsets and bumping the high water mark. */      assertions. Do this also for the "once" (atomic) groups. */
1206    
1207        if (*prev != OP_COND)      if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
1208          {          *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||
1209          number = *prev - OP_BRA;          *prev == OP_ONCE)
1210          {
1211          md->end_match_ptr = eptr;      /* For ONCE */
1212          md->end_offset_top = offset_top;
1213          RRETURN(MATCH_MATCH);
1214          }
1215    
1216          /* For extended extraction brackets (large number), we have to fish out      /* For capturing groups we have to check the group number back at the start
1217          the number from a dummy opcode at the start. */      and if necessary complete handling an extraction by setting the offsets and
1218        bumping the high water mark. Note that whole-pattern recursion is coded as
1219        a recurse into group 0, so it won't be picked up here. Instead, we catch it
1220        when the OP_END is reached. Other recursion is handled here. */
1221    
1222          if (number > EXTRACT_BASIC_MAX) number = GET2(prev, 2+LINK_SIZE);      if (*prev == OP_CBRA || *prev == OP_SCBRA)
1223          offset = number << 1;        {
1224          number = GET2(prev, 1+LINK_SIZE);
1225          offset = number << 1;
1226    
1227  #ifdef DEBUG  #ifdef DEBUG
1228          printf("end bracket %d", number);        printf("end bracket %d", number);
1229          printf("\n");        printf("\n");
1230  #endif  #endif
1231    
1232          /* Test for a numbered group. This includes groups called as a result        md->capture_last = number;
1233          of recursion. Note that whole-pattern recursion is coded as a recurse        if (offset >= md->offset_max) md->offset_overflow = TRUE; else
1234          into group 0, so it won't be picked up here. Instead, we catch it when          {
1235          the OP_END is reached. */          md->offset_vector[offset] =
1236              md->offset_vector[md->offset_end - number];
1237          if (number > 0)          md->offset_vector[offset+1] = eptr - md->start_subject;
1238            {          if (offset_top <= offset) offset_top = offset + 2;
1239            md->capture_last = number;          }
1240            if (offset >= md->offset_max) md->offset_overflow = TRUE; else  
1241              {        /* Handle a recursively called group. Restore the offsets
1242              md->offset_vector[offset] =        appropriately and continue from after the call. */
1243                md->offset_vector[md->offset_end - number];  
1244              md->offset_vector[offset+1] = eptr - md->start_subject;        if (md->recursive != NULL && md->recursive->group_num == number)
1245              if (offset_top <= offset) offset_top = offset + 2;          {
1246              }          recursion_info *rec = md->recursive;
1247            DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
1248            /* Handle a recursively called group. Restore the offsets          md->recursive = rec->prevrec;
1249            appropriately and continue from after the call. */          mstart = rec->save_start;
1250            memcpy(md->offset_vector, rec->offset_save,
1251            if (md->recursive != NULL && md->recursive->group_num == number)            rec->saved_max * sizeof(int));
1252              {          ecode = rec->after_call;
1253              recursion_info *rec = md->recursive;          ims = original_ims;
1254              DPRINTF(("Recursion (%d) succeeded - continuing\n", number));          break;
             md->recursive = rec->prevrec;  
             md->start_match = rec->save_start;  
             memcpy(md->offset_vector, rec->offset_save,  
               rec->saved_max * sizeof(int));  
             ecode = rec->after_call;  
             ims = original_ims;  
             break;  
             }  
           }  
1255          }          }
1256          }
1257    
1258        /* Reset the value of the ims flags, in case they got changed during      /* For both capturing and non-capturing groups, reset the value of the ims
1259        the group. */      flags, in case they got changed during the group. */
1260    
1261        ims = original_ims;      ims = original_ims;
1262        DPRINTF(("ims reset to %02lx\n", ims));      DPRINTF(("ims reset to %02lx\n", ims));
1263    
1264        /* For a non-repeating ket, just continue at this level. This also      /* For a non-repeating ket, just continue at this level. This also
1265        happens for a repeating ket if no characters were matched in the group.      happens for a repeating ket if no characters were matched in the group.
1266        This is the forcible breaking of infinite loops as implemented in Perl      This is the forcible breaking of infinite loops as implemented in Perl
1267        5.005. If there is an options reset, it will get obeyed in the normal      5.005. If there is an options reset, it will get obeyed in the normal
1268        course of events. */      course of events. */
1269    
1270        if (*ecode == OP_KET || eptr == saved_eptr)      if (*ecode == OP_KET || eptr == saved_eptr)
1271          {        {
1272          ecode += 1 + LINK_SIZE;        ecode += 1 + LINK_SIZE;
1273          break;        break;
1274          }        }
1275    
1276        /* The repeating kets try the rest of the pattern or restart from the      /* The repeating kets try the rest of the pattern or restart from the
1277        preceding bracket, in the appropriate order. */      preceding bracket, in the appropriate order. In the second case, we can use
1278        tail recursion to avoid using another stack frame, unless we have an
1279        unlimited repeat of a group that can match an empty string. */
1280    
1281        if (*ecode == OP_KETRMIN)      flags = (*prev >= OP_SBRA)? match_cbegroup : 0;
1282          {  
1283          RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);      if (*ecode == OP_KETRMIN)
1284          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        {
1285          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);
1286          if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1287          }        if (flags != 0)    /* Could match an empty string */
       else  /* OP_KETRMAX */  
1288          {          {
1289          RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);
1290          if (rrc != MATCH_NOMATCH) RRETURN(rrc);          RRETURN(rrc);
         RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0);  
         if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
1291          }          }
1292          ecode = prev;
1293          goto TAIL_RECURSE;
1294        }        }
1295        else  /* OP_KETRMAX */
1296      RRETURN(MATCH_NOMATCH);        {
1297          RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);
1298          if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1299          ecode += 1 + LINK_SIZE;
1300          flags = 0;
1301          goto TAIL_RECURSE;
1302          }
1303        /* Control never gets here */
1304    
1305      /* Start of subject unless notbol, or after internal newline if multiline */      /* Start of subject unless notbol, or after internal newline if multiline */
1306    
# Line 1135  for (;;) Line 1308  for (;;)
1308      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);      if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
1309      if ((ims & PCRE_MULTILINE) != 0)      if ((ims & PCRE_MULTILINE) != 0)
1310        {        {
1311        if (eptr != md->start_subject && eptr[-1] != NEWLINE)        if (eptr != md->start_subject &&
1312              (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
1313          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
1314        ecode++;        ecode++;
1315        break;        break;
# Line 1156  for (;;) Line 1330  for (;;)
1330      ecode++;      ecode++;
1331      break;      break;
1332    
1333        /* Reset the start of match point */
1334    
1335        case OP_SET_SOM:
1336        mstart = eptr;
1337        ecode++;
1338        break;
1339    
1340      /* Assert before internal newline if multiline, or before a terminating      /* Assert before internal newline if multiline, or before a terminating
1341      newline unless endonly is set, else end of subject unless noteol is set. */      newline unless endonly is set, else end of subject unless noteol is set. */
1342    
# Line 1163  for (;;) Line 1344  for (;;)
1344      if ((ims & PCRE_MULTILINE) != 0)      if ((ims & PCRE_MULTILINE) != 0)
1345        {        {
1346        if (eptr < md->end_subject)        if (eptr < md->end_subject)
1347          { if (*eptr != NEWLINE) RRETURN(MATCH_NOMATCH); }          { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }
1348        else        else
1349          { if (md->noteol) RRETURN(MATCH_NOMATCH); }          { if (md->noteol) RRETURN(MATCH_NOMATCH); }
1350        ecode++;        ecode++;
# Line 1174  for (;;) Line 1355  for (;;)
1355        if (md->noteol) RRETURN(MATCH_NOMATCH);        if (md->noteol) RRETURN(MATCH_NOMATCH);
1356        if (!md->endonly)        if (!md->endonly)
1357          {          {
1358          if (eptr < md->end_subject - 1 ||          if (eptr != md->end_subject &&
1359             (eptr == md->end_subject - 1 && *eptr != NEWLINE))              (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1360            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1361          ecode++;          ecode++;
1362          break;          break;
1363          }          }
1364        }        }
1365      /* ... else fall through */      /* ... else fall through for endonly */
1366    
1367      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
1368    
# Line 1193  for (;;) Line 1374  for (;;)
1374      /* End of subject or ending \n assertion (\Z) */      /* End of subject or ending \n assertion (\Z) */
1375    
1376      case OP_EODN:      case OP_EODN:
1377      if (eptr < md->end_subject - 1 ||      if (eptr != md->end_subject &&
1378         (eptr == md->end_subject - 1 && *eptr != NEWLINE)) RRETURN(MATCH_NOMATCH);          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1379          RRETURN(MATCH_NOMATCH);
1380      ecode++;      ecode++;
1381      break;      break;
1382    
# Line 1247  for (;;) Line 1429  for (;;)
1429      /* Match a single character type; inline for speed */      /* Match a single character type; inline for speed */
1430    
1431      case OP_ANY:      case OP_ANY:
1432      if ((ims & PCRE_DOTALL) == 0 && eptr < md->end_subject && *eptr == NEWLINE)      if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
1433        RRETURN(MATCH_NOMATCH);      /* Fall through */
1434    
1435        case OP_ALLANY:
1436      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH);
1437  #ifdef SUPPORT_UTF8      if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
     if (utf8)  
       while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
 #endif  
1438      ecode++;      ecode++;
1439      break;      break;
1440    
# Line 1343  for (;;) Line 1524  for (;;)
1524      ecode++;      ecode++;
1525      break;      break;
1526    
1527        case OP_ANYNL:
1528        if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1529        GETCHARINCTEST(c, eptr);
1530        switch(c)
1531          {
1532          default: RRETURN(MATCH_NOMATCH);
1533          case 0x000d:
1534          if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
1535          break;
1536    
1537          case 0x000a:
1538          break;
1539    
1540          case 0x000b:
1541          case 0x000c:
1542          case 0x0085:
1543          case 0x2028:
1544          case 0x2029:
1545          if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
1546          break;
1547          }
1548        ecode++;
1549        break;
1550    
1551        case OP_NOT_HSPACE:
1552        if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1553        GETCHARINCTEST(c, eptr);
1554        switch(c)
1555          {
1556          default: break;
1557          case 0x09:      /* HT */
1558          case 0x20:      /* SPACE */
1559          case 0xa0:      /* NBSP */
1560          case 0x1680:    /* OGHAM SPACE MARK */
1561          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
1562          case 0x2000:    /* EN QUAD */
1563          case 0x2001:    /* EM QUAD */
1564          case 0x2002:    /* EN SPACE */
1565          case 0x2003:    /* EM SPACE */
1566          case 0x2004:    /* THREE-PER-EM SPACE */
1567          case 0x2005:    /* FOUR-PER-EM SPACE */
1568          case 0x2006:    /* SIX-PER-EM SPACE */
1569          case 0x2007:    /* FIGURE SPACE */
1570          case 0x2008:    /* PUNCTUATION SPACE */
1571          case 0x2009:    /* THIN SPACE */
1572          case 0x200A:    /* HAIR SPACE */
1573          case 0x202f:    /* NARROW NO-BREAK SPACE */
1574          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
1575          case 0x3000:    /* IDEOGRAPHIC SPACE */
1576          RRETURN(MATCH_NOMATCH);
1577          }
1578        ecode++;
1579        break;
1580    
1581        case OP_HSPACE:
1582        if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1583        GETCHARINCTEST(c, eptr);
1584        switch(c)
1585          {
1586          default: RRETURN(MATCH_NOMATCH);
1587          case 0x09:      /* HT */
1588          case 0x20:      /* SPACE */
1589          case 0xa0:      /* NBSP */
1590          case 0x1680:    /* OGHAM SPACE MARK */
1591          case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
1592          case 0x2000:    /* EN QUAD */
1593          case 0x2001:    /* EM QUAD */
1594          case 0x2002:    /* EN SPACE */
1595          case 0x2003:    /* EM SPACE */
1596          case 0x2004:    /* THREE-PER-EM SPACE */
1597          case 0x2005:    /* FOUR-PER-EM SPACE */
1598          case 0x2006:    /* SIX-PER-EM SPACE */
1599          case 0x2007:    /* FIGURE SPACE */
1600          case 0x2008:    /* PUNCTUATION SPACE */
1601          case 0x2009:    /* THIN SPACE */
1602          case 0x200A:    /* HAIR SPACE */
1603          case 0x202f:    /* NARROW NO-BREAK SPACE */
1604          case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
1605          case 0x3000:    /* IDEOGRAPHIC SPACE */
1606          break;
1607          }
1608        ecode++;
1609        break;
1610    
1611        case OP_NOT_VSPACE:
1612        if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1613        GETCHARINCTEST(c, eptr);
1614        switch(c)
1615          {
1616          default: break;
1617          case 0x0a:      /* LF */
1618          case 0x0b:      /* VT */
1619          case 0x0c:      /* FF */
1620          case 0x0d:      /* CR */
1621          case 0x85:      /* NEL */
1622          case 0x2028:    /* LINE SEPARATOR */
1623          case 0x2029:    /* PARAGRAPH SEPARATOR */
1624          RRETURN(MATCH_NOMATCH);
1625          }
1626        ecode++;
1627        break;
1628    
1629        case OP_VSPACE:
1630        if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1631        GETCHARINCTEST(c, eptr);
1632        switch(c)
1633          {
1634          default: RRETURN(MATCH_NOMATCH);
1635          case 0x0a:      /* LF */
1636          case 0x0b:      /* VT */
1637          case 0x0c:      /* FF */
1638          case 0x0d:      /* CR */
1639          case 0x85:      /* NEL */
1640          case 0x2028:    /* LINE SEPARATOR */
1641          case 0x2029:    /* PARAGRAPH SEPARATOR */
1642          break;
1643          }
1644        ecode++;
1645        break;
1646    
1647  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1648      /* Check the next character by Unicode property. We will get here only      /* Check the next character by Unicode property. We will get here only
1649      if the support is in the binary; otherwise a compile-time error occurs. */      if the support is in the binary; otherwise a compile-time error occurs. */
# Line 1352  for (;;) Line 1653  for (;;)
1653      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1654      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1655        {        {
1656        int chartype, rqdtype;        const ucd_record * prop = GET_UCD(c);
       int othercase;  
       int category = _pcre_ucp_findchar(c, &chartype, &othercase);  
   
       rqdtype = *(++ecode);  
       ecode++;  
1657    
1658        if (rqdtype >= 128)        switch(ecode[1])
1659          {          {
1660          if ((rqdtype - 128 != category) == (op == OP_PROP))          case PT_ANY:
1661            if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
1662            break;
1663    
1664            case PT_LAMP:
1665            if ((prop->chartype == ucp_Lu ||
1666                 prop->chartype == ucp_Ll ||
1667                 prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
1668            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1669          }           break;
1670        else  
1671          {          case PT_GC:
1672          if ((rqdtype != chartype) == (op == OP_PROP))          if ((ecode[2] != ucp_gentype[prop->chartype]) == (op == OP_PROP))
1673              RRETURN(MATCH_NOMATCH);
1674            break;
1675    
1676            case PT_PC:
1677            if ((ecode[2] != prop->chartype) == (op == OP_PROP))
1678            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1679            break;
1680    
1681            case PT_SC:
1682            if ((ecode[2] != prop->script) == (op == OP_PROP))
1683              RRETURN(MATCH_NOMATCH);
1684            break;
1685    
1686            default:
1687            RRETURN(PCRE_ERROR_INTERNAL);
1688          }          }
1689    
1690          ecode += 3;
1691        }        }
1692      break;      break;
1693    
# Line 1379  for (;;) Line 1698  for (;;)
1698      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1699      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1700        {        {
1701        int chartype;        int category = UCD_CATEGORY(c);
       int othercase;  
       int category = _pcre_ucp_findchar(c, &chartype, &othercase);  
1702        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (category == ucp_M) RRETURN(MATCH_NOMATCH);
1703        while (eptr < md->end_subject)        while (eptr < md->end_subject)
1704          {          {
# Line 1390  for (;;) Line 1707  for (;;)
1707            {            {
1708            GETCHARLEN(c, eptr, len);            GETCHARLEN(c, eptr, len);
1709            }            }
1710          category = _pcre_ucp_findchar(c, &chartype, &othercase);          category = UCD_CATEGORY(c);
1711          if (category != ucp_M) break;          if (category != ucp_M) break;
1712          eptr += len;          eptr += len;
1713          }          }
# Line 1411  for (;;) Line 1728  for (;;)
1728      case OP_REF:      case OP_REF:
1729        {        {
1730        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
1731        ecode += 3;                                 /* Advance past item */        ecode += 3;
1732    
1733        /* If the reference is unset, set the length to be longer than the amount        /* If the reference is unset, there are two possibilities:
1734        of subject left; this ensures that every attempt at a match fails. We  
1735        can't just fail here, because of the possibility of quantifiers with zero        (a) In the default, Perl-compatible state, set the length to be longer
1736        minima. */        than the amount of subject left; this ensures that every attempt at a
1737          match fails. We can't just fail here, because of the possibility of
1738        length = (offset >= offset_top || md->offset_vector[offset] < 0)?        quantifiers with zero minima.
1739          md->end_subject - eptr + 1 :  
1740          md->offset_vector[offset+1] - md->offset_vector[offset];        (b) If the JavaScript compatibility flag is set, set the length to zero
1741          so that the back reference matches an empty string.
1742    
1743          Otherwise, set the length to the length of what was matched by the
1744          referenced subpattern. */
1745    
1746          if (offset >= offset_top || md->offset_vector[offset] < 0)
1747            length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;
1748          else
1749            length = md->offset_vector[offset+1] - md->offset_vector[offset];
1750    
1751        /* Set up for repetition, or handle the non-repeated case */        /* Set up for repetition, or handle the non-repeated case */
1752    
# Line 1480  for (;;) Line 1806  for (;;)
1806          {          {
1807          for (fi = min;; fi++)          for (fi = min;; fi++)
1808            {            {
1809            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);
1810            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1811            if (fi >= max || !match_ref(offset, eptr, length, md, ims))            if (fi >= max || !match_ref(offset, eptr, length, md, ims))
1812              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
# Line 1501  for (;;) Line 1827  for (;;)
1827            }            }
1828          while (eptr >= pp)          while (eptr >= pp)
1829            {            {
1830            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);
1831            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1832            eptr -= length;            eptr -= length;
1833            }            }
# Line 1606  for (;;) Line 1932  for (;;)
1932            {            {
1933            for (fi = min;; fi++)            for (fi = min;; fi++)
1934              {              {
1935              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);
1936              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1937              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1938              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
# Line 1626  for (;;) Line 1952  for (;;)
1952            {            {
1953            for (fi = min;; fi++)            for (fi = min;; fi++)
1954              {              {
1955              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);
1956              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1957              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1958              c = *eptr++;              c = *eptr++;
# Line 1663  for (;;) Line 1989  for (;;)
1989              }              }
1990            for (;;)            for (;;)
1991              {              {
1992              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);
1993              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1994              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
1995              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1682  for (;;) Line 2008  for (;;)
2008              }              }
2009            while (eptr >= pp)            while (eptr >= pp)
2010              {              {
2011              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);
             eptr--;  
2012              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2013                eptr--;
2014              }              }
2015            }            }
2016    
# Line 1753  for (;;) Line 2079  for (;;)
2079          {          {
2080          for (fi = min;; fi++)          for (fi = min;; fi++)
2081            {            {
2082            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);
2083            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2084            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2085            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
# Line 1777  for (;;) Line 2103  for (;;)
2103            }            }
2104          for(;;)          for(;;)
2105            {            {
2106            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);
2107            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2108            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2109            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2110            }            }
2111          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2112          }          }
# Line 1836  for (;;) Line 2162  for (;;)
2162    
2163        else        else
2164          {          {
2165          int dc;          unsigned int dc;
2166          GETCHARINC(dc, eptr);          GETCHARINC(dc, eptr);
2167          ecode += length;          ecode += length;
2168    
2169          /* If we have Unicode property support, we can use it to test the other          /* If we have Unicode property support, we can use it to test the other
2170          case of the character, if there is one. The result of _pcre_ucp_findchar() is          case of the character, if there is one. */
         < 0 if the char isn't found, and othercase is returned as zero if there  
         isn't one. */  
2171    
2172          if (fc != dc)          if (fc != dc)
2173            {            {
2174  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2175            int chartype;            if (dc != UCD_OTHERCASE(fc))
           int othercase;  
           if (_pcre_ucp_findchar(fc, &chartype, &othercase) < 0 || dc != othercase)  
2176  #endif  #endif
2177              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2178            }            }
# Line 1867  for (;;) Line 2189  for (;;)
2189        }        }
2190      break;      break;
2191    
2192      /* Match a single character repeatedly; different opcodes share code. */      /* Match a single character repeatedly. */
2193    
2194      case OP_EXACT:      case OP_EXACT:
2195      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2196      ecode += 3;      ecode += 3;
2197      goto REPEATCHAR;      goto REPEATCHAR;
2198    
2199        case OP_POSUPTO:
2200        possessive = TRUE;
2201        /* Fall through */
2202    
2203      case OP_UPTO:      case OP_UPTO:
2204      case OP_MINUPTO:      case OP_MINUPTO:
2205      min = 0;      min = 0;
# Line 1882  for (;;) Line 2208  for (;;)
2208      ecode += 3;      ecode += 3;
2209      goto REPEATCHAR;      goto REPEATCHAR;
2210    
2211      case OP_STAR:      case OP_POSSTAR:
2212      case OP_MINSTAR:      possessive = TRUE;
2213      case OP_PLUS:      min = 0;
2214      case OP_MINPLUS:      max = INT_MAX;
2215      case OP_QUERY:      ecode++;
2216      case OP_MINQUERY:      goto REPEATCHAR;
2217      c = *ecode++ - OP_STAR;  
2218      minimize = (c & 1) != 0;      case OP_POSPLUS:
2219        possessive = TRUE;
2220        min = 1;
2221        max = INT_MAX;
2222        ecode++;
2223        goto REPEATCHAR;
2224    
2225        case OP_POSQUERY:
2226        possessive = TRUE;
2227        min = 0;
2228        max = 1;
2229        ecode++;
2230        goto REPEATCHAR;
2231    
2232        case OP_STAR:
2233        case OP_MINSTAR:
2234        case OP_PLUS:
2235        case OP_MINPLUS:
2236        case OP_QUERY:
2237        case OP_MINQUERY:
2238        c = *ecode++ - OP_STAR;
2239        minimize = (c & 1) != 0;
2240      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2241      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2242      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 1913  for (;;) Line 2260  for (;;)
2260    
2261        if (length > 1)        if (length > 1)
2262          {          {
         int oclength = 0;  
         uschar occhars[8];  
   
2263  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2264          int othercase;          unsigned int othercase;
         int chartype;  
2265          if ((ims & PCRE_CASELESS) != 0 &&          if ((ims & PCRE_CASELESS) != 0 &&
2266               _pcre_ucp_findchar(fc, &chartype, &othercase) >= 0 &&              (othercase = UCD_OTHERCASE(fc)) != fc)
              othercase > 0)  
2267            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
2268            else oclength = 0;
2269  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2270    
2271          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2272            {            {
2273            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (memcmp(eptr, charptr, length) == 0) eptr += length;
2274    #ifdef SUPPORT_UCP
2275            /* Need braces because of following else */            /* Need braces because of following else */
2276            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }
2277            else            else
# Line 1935  for (;;) Line 2279  for (;;)
2279              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);
2280              eptr += oclength;              eptr += oclength;
2281              }              }
2282    #else   /* without SUPPORT_UCP */
2283              else { RRETURN(MATCH_NOMATCH); }
2284    #endif  /* SUPPORT_UCP */
2285            }            }
2286    
2287          if (min == max) continue;          if (min == max) continue;
# Line 1943  for (;;) Line 2290  for (;;)
2290            {            {
2291            for (fi = min;; fi++)            for (fi = min;; fi++)
2292              {              {
2293              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);
2294              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2295              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2296              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (memcmp(eptr, charptr, length) == 0) eptr += length;
2297    #ifdef SUPPORT_UCP
2298              /* Need braces because of following else */              /* Need braces because of following else */
2299              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }
2300              else              else
# Line 1954  for (;;) Line 2302  for (;;)
2302                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);
2303                eptr += oclength;                eptr += oclength;
2304                }                }
2305    #else   /* without SUPPORT_UCP */
2306                else { RRETURN (MATCH_NOMATCH); }
2307    #endif  /* SUPPORT_UCP */
2308              }              }
2309            /* Control never gets here */            /* Control never gets here */
2310            }            }
2311          else  
2312            else  /* Maximize */
2313            {            {
2314            pp = eptr;            pp = eptr;
2315            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2316              {              {
2317              if (eptr > md->end_subject - length) break;              if (eptr > md->end_subject - length) break;
2318              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (memcmp(eptr, charptr, length) == 0) eptr += length;
2319    #ifdef SUPPORT_UCP
2320              else if (oclength == 0) break;              else if (oclength == 0) break;
2321              else              else
2322                {                {
2323                if (memcmp(eptr, occhars, oclength) != 0) break;                if (memcmp(eptr, occhars, oclength) != 0) break;
2324                eptr += oclength;                eptr += oclength;
2325                }                }
2326    #else   /* without SUPPORT_UCP */
2327                else break;
2328    #endif  /* SUPPORT_UCP */
2329              }              }
2330            while (eptr >= pp)  
2331              if (possessive) continue;
2332              for(;;)
2333             {             {
2334             RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);
2335             if (rrc != MATCH_NOMATCH) RRETURN(rrc);             if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2336               if (eptr == pp) RRETURN(MATCH_NOMATCH);
2337    #ifdef SUPPORT_UCP
2338               eptr--;
2339               BACKCHAR(eptr);
2340    #else   /* without SUPPORT_UCP */
2341             eptr -= length;             eptr -= length;
2342    #endif  /* SUPPORT_UCP */
2343             }             }
           RRETURN(MATCH_NOMATCH);  
2344            }            }
2345          /* Control never gets here */          /* Control never gets here */
2346          }          }
# Line 2017  for (;;) Line 2380  for (;;)
2380          {          {
2381          for (fi = min;; fi++)          for (fi = min;; fi++)
2382            {            {
2383            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);
2384            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2385            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max || eptr >= md->end_subject ||
2386                fc != md->lcc[*eptr++])                fc != md->lcc[*eptr++])
# Line 2025  for (;;) Line 2388  for (;;)
2388            }            }
2389          /* Control never gets here */          /* Control never gets here */
2390          }          }
2391        else        else  /* Maximize */
2392          {          {
2393          pp = eptr;          pp = eptr;
2394          for (i = min; i < max; i++)          for (i = min; i < max; i++)
# Line 2033  for (;;) Line 2396  for (;;)
2396            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;
2397            eptr++;            eptr++;
2398            }            }
2399            if (possessive) continue;
2400          while (eptr >= pp)          while (eptr >= pp)
2401            {            {
2402            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);
2403            eptr--;            eptr--;
2404            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2405            }            }
# Line 2054  for (;;) Line 2418  for (;;)
2418          {          {
2419          for (fi = min;; fi++)          for (fi = min;; fi++)
2420            {            {
2421            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);
2422            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2423            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)
2424              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2425            }            }
2426          /* Control never gets here */          /* Control never gets here */
2427          }          }
2428        else        else  /* Maximize */
2429          {          {
2430          pp = eptr;          pp = eptr;
2431          for (i = min; i < max; i++)          for (i = min; i < max; i++)
# Line 2069  for (;;) Line 2433  for (;;)
2433            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject || fc != *eptr) break;
2434            eptr++;            eptr++;
2435            }            }
2436            if (possessive) continue;
2437          while (eptr >= pp)          while (eptr >= pp)
2438            {            {
2439            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);
2440            eptr--;            eptr--;
2441            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2442            }            }
# Line 2121  for (;;) Line 2486  for (;;)
2486      ecode += 3;      ecode += 3;
2487      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
2488    
2489        case OP_NOTPOSSTAR:
2490        possessive = TRUE;
2491        min = 0;
2492        max = INT_MAX;
2493        ecode++;
2494        goto REPEATNOTCHAR;
2495    
2496        case OP_NOTPOSPLUS:
2497        possessive = TRUE;
2498        min = 1;
2499        max = INT_MAX;
2500        ecode++;
2501        goto REPEATNOTCHAR;
2502    
2503        case OP_NOTPOSQUERY:
2504        possessive = TRUE;
2505        min = 0;
2506        max = 1;
2507        ecode++;
2508        goto REPEATNOTCHAR;
2509    
2510        case OP_NOTPOSUPTO:
2511        possessive = TRUE;
2512        min = 0;
2513        max = GET2(ecode, 1);
2514        ecode += 3;
2515        goto REPEATNOTCHAR;
2516    
2517      case OP_NOTSTAR:      case OP_NOTSTAR:
2518      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
2519      case OP_NOTPLUS:      case OP_NOTPLUS:
# Line 2160  for (;;) Line 2553  for (;;)
2553        /* UTF-8 mode */        /* UTF-8 mode */
2554        if (utf8)        if (utf8)
2555          {          {
2556          register int d;          register unsigned int d;
2557          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2558            {            {
2559            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
# Line 2185  for (;;) Line 2578  for (;;)
2578          /* UTF-8 mode */          /* UTF-8 mode */
2579          if (utf8)          if (utf8)
2580            {            {
2581            register int d;            register unsigned int d;
2582            for (fi = min;; fi++)            for (fi = min;; fi++)
2583              {              {
2584              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);
2585              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2586              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
2587              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
# Line 2202  for (;;) Line 2595  for (;;)
2595            {            {
2596            for (fi = min;; fi++)            for (fi = min;; fi++)
2597              {              {
2598              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);
2599              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2600              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])
2601                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
# Line 2221  for (;;) Line 2614  for (;;)
2614          /* UTF-8 mode */          /* UTF-8 mode */
2615          if (utf8)          if (utf8)
2616            {            {
2617            register int d;            register unsigned int d;
2618            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2619              {              {
2620              int len = 1;              int len = 1;
# Line 2231  for (;;) Line 2624  for (;;)
2624              if (fc == d) break;              if (fc == d) break;
2625              eptr += len;              eptr += len;
2626              }              }
2627            for(;;)          if (possessive) continue;
2628            for(;;)
2629              {              {
2630              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);
2631              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2632              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2633              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2248  for (;;) Line 2642  for (;;)
2642              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;
2643              eptr++;              eptr++;
2644              }              }
2645              if (possessive) continue;
2646            while (eptr >= pp)            while (eptr >= pp)
2647              {              {
2648              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);
2649              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2650              eptr--;              eptr--;
2651              }              }
# Line 2269  for (;;) Line 2664  for (;;)
2664        /* UTF-8 mode */        /* UTF-8 mode */
2665        if (utf8)        if (utf8)
2666          {          {
2667          register int d;          register unsigned int d;
2668          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2669            {            {
2670            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
# Line 2292  for (;;) Line 2687  for (;;)
2687          /* UTF-8 mode */          /* UTF-8 mode */
2688          if (utf8)          if (utf8)
2689            {            {
2690            register int d;            register unsigned int d;
2691            for (fi = min;; fi++)            for (fi = min;; fi++)
2692              {              {
2693              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);
2694              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2695              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
2696              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fi >= max || eptr >= md->end_subject || fc == d)
# Line 2308  for (;;) Line 2703  for (;;)
2703            {            {
2704            for (fi = min;; fi++)            for (fi = min;; fi++)
2705              {              {
2706              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);
2707              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2708              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)
2709                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
# Line 2327  for (;;) Line 2722  for (;;)
2722          /* UTF-8 mode */          /* UTF-8 mode */
2723          if (utf8)          if (utf8)
2724            {            {
2725            register int d;            register unsigned int d;
2726            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2727              {              {
2728              int len = 1;              int len = 1;
# Line 2336  for (;;) Line 2731  for (;;)
2731              if (fc == d) break;              if (fc == d) break;
2732              eptr += len;              eptr += len;
2733              }              }
2734              if (possessive) continue;
2735            for(;;)            for(;;)
2736              {              {
2737              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);
2738              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2739              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2740              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2353  for (;;) Line 2749  for (;;)
2749              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject || fc == *eptr) break;
2750              eptr++;              eptr++;
2751              }              }
2752              if (possessive) continue;
2753            while (eptr >= pp)            while (eptr >= pp)
2754              {              {
2755              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);
2756              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2757              eptr--;              eptr--;
2758              }              }
# Line 2384  for (;;) Line 2781  for (;;)
2781      ecode += 3;      ecode += 3;
2782      goto REPEATTYPE;      goto REPEATTYPE;
2783    
2784        case OP_TYPEPOSSTAR:
2785        possessive = TRUE;
2786        min = 0;
2787        max = INT_MAX;
2788        ecode++;
2789        goto REPEATTYPE;
2790    
2791        case OP_TYPEPOSPLUS:
2792        possessive = TRUE;
2793        min = 1;
2794        max = INT_MAX;
2795        ecode++;
2796        goto REPEATTYPE;
2797    
2798        case OP_TYPEPOSQUERY:
2799        possessive = TRUE;
2800        min = 0;
2801        max = 1;
2802        ecode++;
2803        goto REPEATTYPE;
2804    
2805        case OP_TYPEPOSUPTO:
2806        possessive = TRUE;
2807        min = 0;
2808        max = GET2(ecode, 1);
2809        ecode += 3;
2810        goto REPEATTYPE;
2811    
2812      case OP_TYPESTAR:      case OP_TYPESTAR:
2813      case OP_TYPEMINSTAR:      case OP_TYPEMINSTAR:
2814      case OP_TYPEPLUS:      case OP_TYPEPLUS:
# Line 2408  for (;;) Line 2833  for (;;)
2833        {        {
2834        prop_fail_result = ctype == OP_NOTPROP;        prop_fail_result = ctype == OP_NOTPROP;
2835        prop_type = *ecode++;        prop_type = *ecode++;
2836        if (prop_type >= 128)        prop_value = *ecode++;
         {  
         prop_test_against = prop_type - 128;  
         prop_test_variable = &prop_category;  
         }  
       else  
         {  
         prop_test_against = prop_type;  
         prop_test_variable = &prop_chartype;  
         }  
2837        }        }
2838      else prop_type = -1;      else prop_type = -1;
2839  #endif  #endif
# Line 2434  for (;;) Line 2850  for (;;)
2850      if (min > 0)      if (min > 0)
2851        {        {
2852  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2853        if (prop_type > 0)        if (prop_type >= 0)
2854          {          {
2855          for (i = 1; i <= min; i++)          switch(prop_type)
2856            {            {
2857            GETCHARINC(c, eptr);            case PT_ANY:
2858            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            if (prop_fail_result) RRETURN(MATCH_NOMATCH);
2859            if ((*prop_test_variable == prop_test_against) == prop_fail_result)            for (i = 1; i <= min; i++)
2860              RRETURN(MATCH_NOMATCH);              {
2861                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2862                GETCHARINCTEST(c, eptr);
2863                }
2864              break;
2865    
2866              case PT_LAMP:
2867              for (i = 1; i <= min; i++)
2868                {
2869                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2870                GETCHARINCTEST(c, eptr);
2871                prop_chartype = UCD_CHARTYPE(c);
2872                if ((prop_chartype == ucp_Lu ||
2873                     prop_chartype == ucp_Ll ||
2874                     prop_chartype == ucp_Lt) == prop_fail_result)
2875                  RRETURN(MATCH_NOMATCH);
2876                }
2877              break;
2878    
2879              case PT_GC:
2880              for (i = 1; i <= min; i++)
2881                {
2882                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2883                GETCHARINCTEST(c, eptr);
2884                prop_category = UCD_CATEGORY(c);
2885                if ((prop_category == prop_value) == prop_fail_result)
2886                  RRETURN(MATCH_NOMATCH);
2887                }
2888              break;
2889    
2890              case PT_PC:
2891              for (i = 1; i <= min; i++)
2892                {
2893                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2894                GETCHARINCTEST(c, eptr);
2895                prop_chartype = UCD_CHARTYPE(c);
2896                if ((prop_chartype == prop_value) == prop_fail_result)
2897                  RRETURN(MATCH_NOMATCH);
2898                }
2899              break;
2900    
2901              case PT_SC:
2902              for (i = 1; i <= min; i++)
2903                {
2904                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2905                GETCHARINCTEST(c, eptr);
2906                prop_script = UCD_SCRIPT(c);
2907                if ((prop_script == prop_value) == prop_fail_result)
2908                  RRETURN(MATCH_NOMATCH);
2909                }
2910              break;
2911    
2912              default:
2913              RRETURN(PCRE_ERROR_INTERNAL);
2914            }            }
2915          }          }
2916    
# Line 2453  for (;;) Line 2922  for (;;)
2922          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2923            {            {
2924            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
2925            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            prop_category = UCD_CATEGORY(c);
2926            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);
2927            while (eptr < md->end_subject)            while (eptr < md->end_subject)
2928              {              {
# Line 2462  for (;;) Line 2931  for (;;)
2931                {                {
2932                GETCHARLEN(c, eptr, len);                GETCHARLEN(c, eptr, len);
2933                }                }
2934              prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              prop_category = UCD_CATEGORY(c);
2935              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
2936              eptr += len;              eptr += len;
2937              }              }
# Line 2480  for (;;) Line 2949  for (;;)
2949          case OP_ANY:          case OP_ANY:
2950          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2951            {            {
2952            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject || IS_NEWLINE(eptr))
              (*eptr++ == NEWLINE && (ims & PCRE_DOTALL) == 0))  
2953              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2954              eptr++;
2955              while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2956              }
2957            break;
2958    
2959            case OP_ALLANY:
2960            for (i = 1; i <= min; i++)
2961              {
2962              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2963              eptr++;
2964            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2965            }            }
2966          break;          break;
# Line 2491  for (;;) Line 2969  for (;;)
2969          eptr += min;          eptr += min;
2970          break;          break;
2971    
2972            case OP_ANYNL:
2973            for (i = 1; i <= min; i++)
2974              {
2975              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2976              GETCHARINC(c, eptr);
2977              switch(c)
2978                {
2979                default: RRETURN(MATCH_NOMATCH);
2980                case 0x000d:
2981                if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2982                break;
2983    
2984                case 0x000a:
2985                break;
2986    
2987                case 0x000b:
2988                case 0x000c:
2989                case 0x0085:
2990                case 0x2028:
2991                case 0x2029:
2992                if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
2993                break;
2994                }
2995              }
2996            break;
2997    
2998            case OP_NOT_HSPACE:
2999            for (i = 1; i <= min; i++)
3000              {
3001              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3002              GETCHARINC(c, eptr);
3003              switch(c)
3004                {
3005                default: break;
3006                case 0x09:      /* HT */
3007                case 0x20:      /* SPACE */
3008                case 0xa0:      /* NBSP */
3009                case 0x1680:    /* OGHAM SPACE MARK */
3010                case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3011                case 0x2000:    /* EN QUAD */
3012                case 0x2001:    /* EM QUAD */
3013                case 0x2002:    /* EN SPACE */
3014                case 0x2003:    /* EM SPACE */
3015                case 0x2004:    /* THREE-PER-EM SPACE */
3016                case 0x2005:    /* FOUR-PER-EM SPACE */
3017                case 0x2006:    /* SIX-PER-EM SPACE */
3018                case 0x2007:    /* FIGURE SPACE */
3019                case 0x2008:    /* PUNCTUATION SPACE */
3020                case 0x2009:    /* THIN SPACE */
3021                case 0x200A:    /* HAIR SPACE */
3022                case 0x202f:    /* NARROW NO-BREAK SPACE */
3023                case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3024                case 0x3000:    /* IDEOGRAPHIC SPACE */
3025                RRETURN(MATCH_NOMATCH);
3026                }
3027              }
3028            break;
3029    
3030            case OP_HSPACE:
3031            for (i = 1; i <= min; i++)
3032              {
3033              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3034              GETCHARINC(c, eptr);
3035              switch(c)
3036                {
3037                default: RRETURN(MATCH_NOMATCH);
3038                case 0x09:      /* HT */
3039                case 0x20:      /* SPACE */
3040                case 0xa0:      /* NBSP */
3041                case 0x1680:    /* OGHAM SPACE MARK */
3042                case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3043                case 0x2000:    /* EN QUAD */
3044                case 0x2001:    /* EM QUAD */
3045                case 0x2002:    /* EN SPACE */
3046                case 0x2003:    /* EM SPACE */
3047                case 0x2004:    /* THREE-PER-EM SPACE */
3048                case 0x2005:    /* FOUR-PER-EM SPACE */
3049                case 0x2006:    /* SIX-PER-EM SPACE */
3050                case 0x2007:    /* FIGURE SPACE */
3051                case 0x2008:    /* PUNCTUATION SPACE */
3052                case 0x2009:    /* THIN SPACE */
3053                case 0x200A:    /* HAIR SPACE */
3054                case 0x202f:    /* NARROW NO-BREAK SPACE */
3055                case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3056                case 0x3000:    /* IDEOGRAPHIC SPACE */
3057                break;
3058                }
3059              }
3060            break;
3061    
3062            case OP_NOT_VSPACE:
3063            for (i = 1; i <= min; i++)
3064              {
3065              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3066              GETCHARINC(c, eptr);
3067              switch(c)
3068                {
3069                default: break;
3070                case 0x0a:      /* LF */
3071                case 0x0b:      /* VT */
3072                case 0x0c:      /* FF */
3073                case 0x0d:      /* CR */
3074                case 0x85:      /* NEL */
3075                case 0x2028:    /* LINE SEPARATOR */
3076                case 0x2029:    /* PARAGRAPH SEPARATOR */
3077                RRETURN(MATCH_NOMATCH);
3078                }
3079              }
3080            break;
3081    
3082            case OP_VSPACE:
3083            for (i = 1; i <= min; i++)
3084              {
3085              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3086              GETCHARINC(c, eptr);
3087              switch(c)
3088                {
3089                default: RRETURN(MATCH_NOMATCH);
3090                case 0x0a:      /* LF */
3091                case 0x0b:      /* VT */
3092                case 0x0c:      /* FF */
3093                case 0x0d:      /* CR */
3094                case 0x85:      /* NEL */
3095                case 0x2028:    /* LINE SEPARATOR */
3096                case 0x2029:    /* PARAGRAPH SEPARATOR */
3097                break;
3098                }
3099              }
3100            break;
3101    
3102          case OP_NOT_DIGIT:          case OP_NOT_DIGIT:
3103          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3104            {            {
# Line 2515  for (;;) Line 3123  for (;;)
3123          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3124            {            {
3125            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject ||
3126               (*eptr < 128 && (md->ctypes[*eptr++] & ctype_space) != 0))               (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0))
3127              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3128            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
3129            }            }
3130          break;          break;
3131    
# Line 2535  for (;;) Line 3143  for (;;)
3143          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3144            {            {
3145            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject ||
3146               (*eptr < 128 && (md->ctypes[*eptr++] & ctype_word) != 0))               (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0))
3147              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3148            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
3149            }            }
3150          break;          break;
3151    
# Line 2559  for (;;) Line 3167  for (;;)
3167  #endif     /* SUPPORT_UTF8 */  #endif     /* SUPPORT_UTF8 */
3168    
3169        /* Code for the non-UTF-8 case for minimum matching of operators other        /* Code for the non-UTF-8 case for minimum matching of operators other
3170        than OP_PROP and OP_NOTPROP. */        than OP_PROP and OP_NOTPROP. We can assume that there are the minimum
3171          number of bytes present, as this was tested above. */
3172    
3173        switch(ctype)        switch(ctype)
3174          {          {
3175          case OP_ANY:          case OP_ANY:
3176          if ((ims & PCRE_DOTALL) == 0)          for (i = 1; i <= min; i++)
3177            {            {
3178            for (i = 1; i <= min; i++)            if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
3179              if (*eptr++ == NEWLINE) RRETURN(MATCH_NOMATCH);            eptr++;
3180            }            }
3181          else eptr += min;          break;
3182    
3183            case OP_ALLANY:
3184            eptr += min;
3185          break;          break;
3186    
3187          case OP_ANYBYTE:          case OP_ANYBYTE:
3188          eptr += min;          eptr += min;
3189          break;          break;
3190    
3191            /* Because of the CRLF case, we can't assume the minimum number of
3192            bytes are present in this case. */
3193    
3194            case OP_ANYNL:
3195            for (i = 1; i <= min; i++)
3196              {
3197              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3198              switch(*eptr++)
3199                {
3200                default: RRETURN(MATCH_NOMATCH);
3201                case 0x000d:
3202                if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3203                break;
3204                case 0x000a:
3205                break;
3206    
3207                case 0x000b:
3208                case 0x000c:
3209                case 0x0085:
3210                if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
3211                break;
3212                }
3213              }
3214            break;
3215    
3216            case OP_NOT_HSPACE:
3217            for (i = 1; i <= min; i++)
3218              {
3219              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3220              switch(*eptr++)
3221                {
3222                default: break;
3223                case 0x09:      /* HT */
3224                case 0x20:      /* SPACE */
3225                case 0xa0:      /* NBSP */
3226                RRETURN(MATCH_NOMATCH);
3227                }
3228              }
3229            break;
3230    
3231            case OP_HSPACE:
3232            for (i = 1; i <= min; i++)
3233              {
3234              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3235              switch(*eptr++)
3236                {
3237                default: RRETURN(MATCH_NOMATCH);
3238                case 0x09:      /* HT */
3239                case 0x20:      /* SPACE */
3240                case 0xa0:      /* NBSP */
3241                break;
3242                }
3243              }
3244            break;
3245    
3246            case OP_NOT_VSPACE:
3247            for (i = 1; i <= min; i++)
3248              {
3249              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3250              switch(*eptr++)
3251                {
3252                default: break;
3253                case 0x0a:      /* LF */
3254                case 0x0b:      /* VT */
3255                case 0x0c:      /* FF */
3256                case 0x0d:      /* CR */
3257                case 0x85:      /* NEL */
3258                RRETURN(MATCH_NOMATCH);
3259                }
3260              }
3261            break;
3262    
3263            case OP_VSPACE:
3264            for (i = 1; i <= min; i++)
3265              {
3266              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3267              switch(*eptr++)
3268                {
3269                default: RRETURN(MATCH_NOMATCH);
3270                case 0x0a:      /* LF */
3271                case 0x0b:      /* VT */
3272                case 0x0c:      /* FF */
3273                case 0x0d:      /* CR */
3274                case 0x85:      /* NEL */
3275                break;
3276                }
3277              }
3278            break;
3279    
3280          case OP_NOT_DIGIT:          case OP_NOT_DIGIT:
3281          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3282            if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH);            if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH);
# Line 2624  for (;;) Line 3325  for (;;)
3325      if (minimize)      if (minimize)
3326        {        {
3327  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3328        if (prop_type > 0)        if (prop_type >= 0)
3329          {          {
3330          for (fi = min;; fi++)          switch(prop_type)
3331            {            {
3332            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            case PT_ANY:
3333            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            for (fi = min;; fi++)
3334            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              {
3335            GETCHARINC(c, eptr);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);
3336            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3337            if ((*prop_test_variable == prop_test_against) == prop_fail_result)              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3338              RRETURN(MATCH_NOMATCH);              GETCHARINC(c, eptr);
3339            }              if (prop_fail_result) RRETURN(MATCH_NOMATCH);
3340          }              }
3341              /* Control never gets here */
3342    
3343        /* Match extended Unicode sequences. We will get here only if the            case PT_LAMP:
3344        support is in the binary; otherwise a compile-time error occurs. */            for (fi = min;; fi++)
3345                {
3346                RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);
3347                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3348                if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3349                GETCHARINC(c, eptr);
3350                prop_chartype = UCD_CHARTYPE(c);
3351                if ((prop_chartype == ucp_Lu ||
3352                     prop_chartype == ucp_Ll ||
3353                     prop_chartype == ucp_Lt) == prop_fail_result)
3354                  RRETURN(MATCH_NOMATCH);
3355                }
3356              /* Control never gets here */
3357    
3358        else if (ctype == OP_EXTUNI)            case PT_GC:
3359          {            for (fi = min;; fi++)
3360                {
3361                RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);
3362                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3363                if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3364                GETCHARINC(c, eptr);
3365                prop_category = UCD_CATEGORY(c);
3366                if ((prop_category == prop_value) == prop_fail_result)
3367                  RRETURN(MATCH_NOMATCH);
3368                }
3369              /* Control never gets here */
3370    
3371              case PT_PC:
3372              for (fi = min;; fi++)
3373                {
3374                RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);
3375                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3376                if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3377                GETCHARINC(c, eptr);
3378                prop_chartype = UCD_CHARTYPE(c);
3379                if ((prop_chartype == prop_value) == prop_fail_result)
3380                  RRETURN(MATCH_NOMATCH);
3381                }
3382              /* Control never gets here */
3383    
3384              case PT_SC:
3385              for (fi = min;; fi++)
3386                {
3387                RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);
3388                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3389                if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3390                GETCHARINC(c, eptr);
3391                prop_script = UCD_SCRIPT(c);
3392                if ((prop_script == prop_value) == prop_fail_result)
3393                  RRETURN(MATCH_NOMATCH);
3394                }
3395              /* Control never gets here */
3396    
3397              default:
3398              RRETURN(PCRE_ERROR_INTERNAL);
3399              }
3400            }
3401    
3402          /* Match extended Unicode sequences. We will get here only if the
3403          support is in the binary; otherwise a compile-time error occurs. */
3404    
3405          else if (ctype == OP_EXTUNI)
3406            {
3407          for (fi = min;; fi++)          for (fi = min;; fi++)
3408            {            {
3409            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);
3410            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3411            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3412            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3413            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            prop_category = UCD_CATEGORY(c);
3414            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);
3415            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3416              {              {
# Line 2658  for (;;) Line 3419  for (;;)
3419                {                {
3420                GETCHARLEN(c, eptr, len);                GETCHARLEN(c, eptr, len);
3421                }                }
3422              prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              prop_category = UCD_CATEGORY(c);
3423              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
3424              eptr += len;              eptr += len;
3425              }              }
# Line 2674  for (;;) Line 3435  for (;;)
3435          {          {
3436          for (fi = min;; fi++)          for (fi = min;; fi++)
3437            {            {
3438            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);
3439            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3440            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max || eptr >= md->end_subject ||
3441                   (ctype == OP_ANY && IS_NEWLINE(eptr)))
3442                RRETURN(MATCH_NOMATCH);
3443    
3444            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
3445            switch(ctype)            switch(ctype)
3446              {              {
3447              case OP_ANY:              case OP_ANY:        /* This is the non-NL case */
3448              if ((ims & PCRE_DOTALL) == 0 && c == NEWLINE) RRETURN(MATCH_NOMATCH);              case OP_ALLANY:
3449                case OP_ANYBYTE:
3450              break;              break;
3451    
3452              case OP_ANYBYTE:              case OP_ANYNL:
3453                switch(c)
3454                  {
3455                  default: RRETURN(MATCH_NOMATCH);
3456                  case 0x000d:
3457                  if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3458                  break;
3459                  case 0x000a:
3460                  break;
3461    
3462                  case 0x000b:
3463                  case 0x000c:
3464                  case 0x0085:
3465                  case 0x2028:
3466                  case 0x2029:
3467                  if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
3468                  break;
3469                  }
3470                break;
3471    
3472                case OP_NOT_HSPACE:
3473                switch(c)
3474                  {
3475                  default: break;
3476                  case 0x09:      /* HT */
3477                  case 0x20:      /* SPACE */
3478                  case 0xa0:      /* NBSP */
3479                  case 0x1680:    /* OGHAM SPACE MARK */
3480                  case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3481                  case 0x2000:    /* EN QUAD */
3482                  case 0x2001:    /* EM QUAD */
3483                  case 0x2002:    /* EN SPACE */
3484                  case 0x2003:    /* EM SPACE */
3485                  case 0x2004:    /* THREE-PER-EM SPACE */
3486                  case 0x2005:    /* FOUR-PER-EM SPACE */
3487                  case 0x2006:    /* SIX-PER-EM SPACE */
3488                  case 0x2007:    /* FIGURE SPACE */
3489                  case 0x2008:    /* PUNCTUATION SPACE */
3490                  case 0x2009:    /* THIN SPACE */
3491                  case 0x200A:    /* HAIR SPACE */
3492                  case 0x202f:    /* NARROW NO-BREAK SPACE */
3493                  case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3494                  case 0x3000:    /* IDEOGRAPHIC SPACE */
3495                  RRETURN(MATCH_NOMATCH);
3496                  }
3497                break;
3498    
3499                case OP_HSPACE:
3500                switch(c)
3501                  {
3502                  default: RRETURN(MATCH_NOMATCH);
3503                  case 0x09:      /* HT */
3504                  case 0x20:      /* SPACE */
3505                  case 0xa0:      /* NBSP */
3506                  case 0x1680:    /* OGHAM SPACE MARK */
3507                  case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3508                  case 0x2000:    /* EN QUAD */
3509                  case 0x2001:    /* EM QUAD */
3510                  case 0x2002:    /* EN SPACE */
3511                  case 0x2003:    /* EM SPACE */
3512                  case 0x2004:    /* THREE-PER-EM SPACE */
3513                  case 0x2005:    /* FOUR-PER-EM SPACE */
3514                  case 0x2006:    /* SIX-PER-EM SPACE */
3515                  case 0x2007:    /* FIGURE SPACE */
3516                  case 0x2008:    /* PUNCTUATION SPACE */
3517                  case 0x2009:    /* THIN SPACE */
3518                  case 0x200A:    /* HAIR SPACE */
3519                  case 0x202f:    /* NARROW NO-BREAK SPACE */
3520                  case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3521                  case 0x3000:    /* IDEOGRAPHIC SPACE */
3522                  break;
3523                  }
3524                break;
3525    
3526                case OP_NOT_VSPACE:
3527                switch(c)
3528                  {
3529                  default: break;
3530                  case 0x0a:      /* LF */
3531                  case 0x0b:      /* VT */
3532                  case 0x0c:      /* FF */
3533                  case 0x0d:      /* CR */
3534                  case 0x85:      /* NEL */
3535                  case 0x2028:    /* LINE SEPARATOR */
3536                  case 0x2029:    /* PARAGRAPH SEPARATOR */
3537                  RRETURN(MATCH_NOMATCH);
3538                  }
3539                break;
3540    
3541                case OP_VSPACE:
3542                switch(c)
3543                  {
3544                  default: RRETURN(MATCH_NOMATCH);
3545                  case 0x0a:      /* LF */
3546                  case 0x0b:      /* VT */
3547                  case 0x0c:      /* FF */
3548                  case 0x0d:      /* CR */
3549                  case 0x85:      /* NEL */
3550                  case 0x2028:    /* LINE SEPARATOR */
3551                  case 0x2029:    /* PARAGRAPH SEPARATOR */
3552                  break;
3553                  }
3554              break;              break;
3555    
3556              case OP_NOT_DIGIT:              case OP_NOT_DIGIT:
# Line 2729  for (;;) Line 3594  for (;;)
3594          {          {
3595          for (fi = min;; fi++)          for (fi = min;; fi++)
3596            {            {
3597            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);
3598            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3599            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max || eptr >= md->end_subject ||
3600                   (ctype == OP_ANY && IS_NEWLINE(eptr)))
3601                RRETURN(MATCH_NOMATCH);
3602    
3603            c = *eptr++;            c = *eptr++;
3604            switch(ctype)            switch(ctype)
3605              {              {
3606              case OP_ANY:              case OP_ANY:     /* This is the non-NL case */
3607              if ((ims & PCRE_DOTALL) == 0 && c == NEWLINE) RRETURN(MATCH_NOMATCH);              case OP_ALLANY:
3608                case OP_ANYBYTE:
3609              break;              break;
3610    
3611              case OP_ANYBYTE:              case OP_ANYNL:
3612                switch(c)
3613                  {
3614                  default: RRETURN(MATCH_NOMATCH);
3615                  case 0x000d:
3616                  if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3617                  break;
3618    
3619                  case 0x000a:
3620                  break;
3621    
3622                  case 0x000b:
3623                  case 0x000c:
3624                  case 0x0085:
3625                  if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
3626                  break;
3627                  }
3628                break;
3629    
3630                case OP_NOT_HSPACE:
3631                switch(c)
3632                  {
3633                  default: break;
3634                  case 0x09:      /* HT */
3635                  case 0x20:      /* SPACE */
3636                  case 0xa0:      /* NBSP */
3637                  RRETURN(MATCH_NOMATCH);
3638                  }
3639                break;
3640    
3641                case OP_HSPACE:
3642                switch(c)
3643                  {
3644                  default: RRETURN(MATCH_NOMATCH);
3645                  case 0x09:      /* HT */
3646                  case 0x20:      /* SPACE */
3647                  case 0xa0:      /* NBSP */
3648                  break;
3649                  }
3650                break;
3651    
3652                case OP_NOT_VSPACE:
3653                switch(c)
3654                  {
3655                  default: break;
3656                  case 0x0a:      /* LF */
3657                  case 0x0b:      /* VT */
3658                  case 0x0c:      /* FF */
3659                  case 0x0d:      /* CR */
3660                  case 0x85:      /* NEL */
3661                  RRETURN(MATCH_NOMATCH);
3662                  }
3663                break;
3664    
3665                case OP_VSPACE:
3666                switch(c)
3667                  {
3668                  default: RRETURN(MATCH_NOMATCH);
3669                  case 0x0a:      /* LF */
3670                  case 0x0b:      /* VT */
3671                  case 0x0c:      /* FF */
3672                  case 0x0d:      /* CR */
3673                  case 0x85:      /* NEL */
3674                  break;
3675                  }
3676              break;              break;
3677    
3678              case OP_NOT_DIGIT:              case OP_NOT_DIGIT:
# Line 2774  for (;;) Line 3707  for (;;)
3707        /* Control never gets here */        /* Control never gets here */
3708        }        }
3709    
3710      /* If maximizing it is worth using inline code for speed, doing the type      /* If maximizing, it is worth using inline code for speed, doing the type
3711      test once at the start (i.e. keep it out of the loop). Again, keep the      test once at the start (i.e. keep it out of the loop). Again, keep the
3712      UTF-8 and UCP stuff separate. */      UTF-8 and UCP stuff separate. */
3713    
# Line 2783  for (;;) Line 3716  for (;;)
3716        pp = eptr;  /* Remember where we started */        pp = eptr;  /* Remember where we started */
3717    
3718  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3719        if (prop_type > 0)        if (prop_type >= 0)
3720          {          {
3721          for (i = min; i < max; i++)          switch(prop_type)
3722            {            {
3723            int len = 1;            case PT_ANY:
3724            if (eptr >= md->end_subject) break;            for (i = min; i < max; i++)
3725            GETCHARLEN(c, eptr, len);              {
3726            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              int len = 1;
3727            if ((*prop_test_variable == prop_test_against) == prop_fail_result)              if (eptr >= md->end_subject) break;
3728              break;              GETCHARLEN(c, eptr, len);
3729            eptr+= len;              if (prop_fail_result) break;
3730                eptr+= len;
3731                }
3732              break;
3733    
3734              case PT_LAMP:
3735              for (i = min; i < max; i++)
3736                {
3737                int len = 1;
3738                if (eptr >= md->end_subject) break;
3739                GETCHARLEN(c, eptr, len);
3740                prop_chartype = UCD_CHARTYPE(c);
3741                if ((prop_chartype == ucp_Lu ||
3742                     prop_chartype == ucp_Ll ||
3743                     prop_chartype == ucp_Lt) == prop_fail_result)
3744                  break;
3745                eptr+= len;
3746                }
3747              break;
3748    
3749              case PT_GC:
3750              for (i = min; i < max; i++)
3751                {
3752                int len = 1;
3753                if (eptr >= md->end_subject) break;
3754                GETCHARLEN(c, eptr, len);
3755                prop_category = UCD_CATEGORY(c);
3756                if ((prop_category == prop_value) == prop_fail_result)
3757                  break;
3758                eptr+= len;
3759                }
3760              break;
3761    
3762              case PT_PC:
3763              for (i = min; i < max; i++)
3764                {
3765                int len = 1;
3766                if (eptr >= md->end_subject) break;
3767                GETCHARLEN(c, eptr, len);
3768                prop_chartype = UCD_CHARTYPE(c);
3769                if ((prop_chartype == prop_value) == prop_fail_result)
3770                  break;
3771                eptr+= len;
3772                }
3773              break;
3774    
3775              case PT_SC:
3776              for (i = min; i < max; i++)
3777                {
3778                int len = 1;
3779                if (eptr >= md->end_subject) break;
3780                GETCHARLEN(c, eptr, len);
3781                prop_script = UCD_SCRIPT(c);
3782                if ((prop_script == prop_value) == prop_fail_result)
3783                  break;
3784                eptr+= len;
3785                }
3786              break;
3787            }            }
3788    
3789          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
3790    
3791            if (possessive) continue;
3792          for(;;)          for(;;)
3793            {            {
3794            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);
3795            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3796            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3797            BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
3798            }            }
3799          }          }
3800    
# Line 2816  for (;;) Line 3807  for (;;)
3807            {            {
3808            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject) break;
3809            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3810            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            prop_category = UCD_CATEGORY(c);
3811            if (prop_category == ucp_M) break;            if (prop_category == ucp_M) break;
3812            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3813              {              {
# Line 2825  for (;;) Line 3816  for (;;)
3816                {                {
3817                GETCHARLEN(c, eptr, len);                GETCHARLEN(c, eptr, len);
3818                }                }
3819              prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              prop_category = UCD_CATEGORY(c);
3820              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
3821              eptr += len;              eptr += len;
3822              }              }
# Line 2833  for (;;) Line 3824  for (;;)
3824    
3825          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
3826    
3827            if (possessive) continue;
3828          for(;;)          for(;;)
3829            {            {
3830            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);
3831            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3832            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3833            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
3834              {              {
3835              int len = 1;              int len = 1;
             BACKCHAR(eptr);  
3836              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
3837                {                {
3838                  BACKCHAR(eptr);
3839                GETCHARLEN(c, eptr, len);                GETCHARLEN(c, eptr, len);
3840                }                }
3841              prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              prop_category = UCD_CATEGORY(c);
3842              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
3843              eptr--;              eptr--;
3844              }              }
# Line 2864  for (;;) Line 3856  for (;;)
3856          switch(ctype)          switch(ctype)
3857            {            {
3858            case OP_ANY:            case OP_ANY:
   
           /* Special code is required for UTF8, but when the maximum is unlimited  
           we don't need it, so we repeat the non-UTF8 code. This is probably  
           worth it, because .* is quite a common idiom. */  
   
3859            if (max < INT_MAX)            if (max < INT_MAX)
3860              {              {
3861              if ((ims & PCRE_DOTALL) == 0)              for (i = min; i < max; i++)
               {  
               for (i = min; i < max; i++)  
                 {  
                 if (eptr >= md->end_subject || *eptr == NEWLINE) break;  
                 eptr++;  
                 while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
                 }  
               }  
             else  
3862                {                {
3863                for (i = min; i < max; i++)                if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
3864                  {                eptr++;
3865                  eptr++;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                 while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
                 }  
3866                }                }
3867              }              }
3868    
# Line 2894  for (;;) Line 3870  for (;;)
3870    
3871            else            else
3872              {              {
3873              if ((ims & PCRE_DOTALL) == 0)              for (i = min; i < max; i++)
3874                {                {
3875                for (i = min; i < max; i++)                if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
3876                  {                eptr++;
3877                  if (eptr >= md->end_subject || *eptr == NEWLINE) break;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                 eptr++;  
                 }  
               break;  
3878                }                }
3879              else              }
3880              break;
3881    
3882              case OP_ALLANY:
3883              if (max < INT_MAX)
3884                {
3885                for (i = min; i < max; i++)
3886                {                {
3887                c = max - min;                if (eptr >= md->end_subject) break;
3888                if (c > md->end_subject - eptr) c = md->end_subject - eptr;                eptr++;
3889                eptr += c;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
3890                }                }
3891              }              }
3892              else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
3893            break;            break;
3894    
3895            /* The byte case is the same as non-UTF8 */            /* The byte case is the same as non-UTF8 */
3896    
3897            case OP_ANYBYTE:            case OP_ANYBYTE:
3898            c = max - min;            c = max - min;
3899            if (c > md->end_subject - eptr) c = md->end_subject - eptr;            if (c > (unsigned int)(md->end_subject - eptr))
3900                c = md->end_subject - eptr;
3901            eptr += c;            eptr += c;
3902            break;            break;
3903    
3904              case OP_ANYNL:
3905              for (i = min; i < max; i++)
3906                {
3907                int len = 1;
3908                if (eptr >= md->end_subject) break;
3909                GETCHARLEN(c, eptr, len);
3910                if (c == 0x000d)
3911                  {
3912                  if (++eptr >= md->end_subject) break;
3913                  if (*eptr == 0x000a) eptr++;
3914                  }
3915                else
3916                  {
3917                  if (c != 0x000a &&
3918                      (md->bsr_anycrlf ||
3919                       (c != 0x000b && c != 0x000c &&
3920                        c != 0x0085 && c != 0x2028 && c != 0x2029)))
3921                    break;
3922                  eptr += len;
3923                  }
3924                }
3925              break;
3926    
3927              case OP_NOT_HSPACE:
3928              case OP_HSPACE:
3929              for (i = min; i < max; i++)
3930                {
3931                BOOL gotspace;
3932                int len = 1;
3933                if (eptr >= md->end_subject) break;
3934                GETCHARLEN(c, eptr, len);
3935                switch(c)
3936                  {
3937                  default: gotspace = FALSE; break;
3938                  case 0x09:      /* HT */
3939                  case 0x20:      /* SPACE */
3940                  case 0xa0:      /* NBSP */
3941                  case 0x1680:    /* OGHAM SPACE MARK */
3942                  case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3943                  case 0x2000:    /* EN QUAD */
3944                  case 0x2001:    /* EM QUAD */
3945                  case 0x2002:    /* EN SPACE */
3946                  case 0x2003:    /* EM SPACE */
3947                  case 0x2004:    /* THREE-PER-EM SPACE */
3948                  case 0x2005:    /* FOUR-PER-EM SPACE */
3949                  case 0x2006:    /* SIX-PER-EM SPACE */
3950                  case 0x2007:    /* FIGURE SPACE */
3951                  case 0x2008:    /* PUNCTUATION SPACE */
3952                  case 0x2009:    /* THIN SPACE */
3953                  case 0x200A:    /* HAIR SPACE */
3954                  case 0x202f:    /* NARROW NO-BREAK SPACE */
3955                  case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3956                  case 0x3000:    /* IDEOGRAPHIC SPACE */
3957                  gotspace = TRUE;
3958                  break;
3959                  }
3960                if (gotspace == (ctype == OP_NOT_HSPACE)) break;
3961                eptr += len;
3962                }
3963              break;
3964    
3965              case OP_NOT_VSPACE:
3966              case OP_VSPACE:
3967              for (i = min; i < max; i++)
3968                {
3969                BOOL gotspace;
3970                int len = 1;
3971                if (eptr >= md->end_subject) break;
3972                GETCHARLEN(c, eptr, len);
3973                switch(c)
3974                  {
3975                  default: gotspace = FALSE; break;
3976                  case 0x0a:      /* LF */
3977                  case 0x0b:      /* VT */
3978                  case 0x0c:      /* FF */
3979                  case 0x0d:      /* CR */
3980                  case 0x85:      /* NEL */
3981                  case 0x2028:    /* LINE SEPARATOR */
3982                  case 0x2029:    /* PARAGRAPH SEPARATOR */
3983                  gotspace = TRUE;
3984                  break;
3985                  }
3986                if (gotspace == (ctype == OP_NOT_VSPACE)) break;
3987                eptr += len;
3988                }
3989              break;
3990    
3991            case OP_NOT_DIGIT:            case OP_NOT_DIGIT:
3992            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3993              {              {
# Line 2992  for (;;) Line 4060  for (;;)
4060    
4061          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
4062    
4063            if (possessive) continue;
4064          for(;;)          for(;;)
4065            {            {
4066            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);
4067            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4068            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
4069            BACKCHAR(eptr);            BACKCHAR(eptr);
4070            }            }
4071          }          }
4072        else        else
4073  #endif  #endif  /* SUPPORT_UTF8 */
4074    
4075        /* Not UTF-8 mode */        /* Not UTF-8 mode */
4076          {          {
4077          switch(ctype)          switch(ctype)
4078            {            {
4079            case OP_ANY:            case OP_ANY:
4080            if ((ims & PCRE_DOTALL) == 0)            for (i = min; i < max; i++)
4081              {              {
4082              for (i = min; i < max; i++)              if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
4083                {              eptr++;
               if (eptr >= md->end_subject || *eptr == NEWLINE) break;  
               eptr++;  
               }  
             break;  
4084              }              }
4085            /* For DOTALL case, fall through and treat as \C */            break;
4086    
4087              case OP_ALLANY:
4088            case OP_ANYBYTE:            case OP_ANYBYTE:
4089            c = max - min;            c = max - min;
4090            if (c > md->end_subject - eptr) c = md->end_subject - eptr;            if (c > (unsigned int)(md->end_subject - eptr))
4091                c = md->end_subject - eptr;
4092            eptr += c;            eptr += c;
4093            break;            break;
4094    
4095              case OP_ANYNL:
4096              for (i = min; i < max; i++)
4097                {
4098                if (eptr >= md->end_subject) break;
4099                c = *eptr;
4100                if (c == 0x000d)
4101                  {
4102                  if (++eptr >= md->end_subject) break;
4103                  if (*eptr == 0x000a) eptr++;
4104                  }
4105                else
4106                  {
4107                  if (c != 0x000a &&
4108                      (md->bsr_anycrlf ||
4109                        (c != 0x000b && c != 0x000c && c != 0x0085)))
4110                    break;
4111                  eptr++;
4112                  }
4113                }
4114              break;
4115    
4116              case OP_NOT_HSPACE:
4117              for (i = min; i < max; i++)
4118                {
4119                if (eptr >= md->end_subject) break;
4120                c = *eptr;
4121                if (c == 0x09 || c == 0x20 || c == 0xa0) break;
4122                eptr++;
4123                }
4124              break;
4125    
4126              case OP_HSPACE:
4127              for (i = min; i < max; i++)
4128                {
4129                if (eptr >= md->end_subject) break;
4130                c = *eptr;
4131                if (c != 0x09 && c != 0x20 && c != 0xa0) break;
4132                eptr++;
4133                }
4134              break;
4135    
4136              case OP_NOT_VSPACE:
4137              for (i = min; i < max; i++)
4138                {
4139                if (eptr >= md->end_subject) break;
4140                c = *eptr;
4141                if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85)
4142                  break;
4143                eptr++;
4144                }
4145              break;
4146    
4147              case OP_VSPACE:
4148              for (i = min; i < max; i++)
4149                {
4150                if (eptr >= md->end_subject) break;
4151                c = *eptr;
4152                if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85)
4153                  break;
4154                eptr++;
4155                }
4156              break;
4157    
4158            case OP_NOT_DIGIT:            case OP_NOT_DIGIT:
4159            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4160              {              {
# Line 3085  for (;;) Line 4215  for (;;)
4215    
4216          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
4217    
4218            if (possessive) continue;
4219          while (eptr >= pp)          while (eptr >= pp)
4220            {            {
4221            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);
4222            eptr--;            eptr--;
4223            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4224            }            }
# Line 3099  for (;;) Line 4230  for (;;)
4230        }        }
4231      /* Control never gets here */      /* Control never gets here */
4232    
4233      /* There's been some horrible disaster. Since all codes > OP_BRA are      /* There's been some horrible disaster. Arrival here can only mean there is
4234      for capturing brackets, and there shouldn't be any gaps between 0 and      something seriously wrong in the code above or the OP_xxx definitions. */
     OP_BRA, arrival here can only mean there is something seriously wrong  
     in the code above or the OP_xxx definitions. */  
4235    
4236      default:      default:
4237      DPRINTF(("Unknown opcode %d\n", *ecode));      DPRINTF(("Unknown opcode %d\n", *ecode));
4238      RRETURN(PCRE_ERROR_UNKNOWN_NODE);      RRETURN(PCRE_ERROR_UNKNOWN_OPCODE);
4239      }      }
4240    
4241    /* Do not stick any code in here without much thought; it is assumed    /* Do not stick any code in here without much thought; it is assumed
# Line 3115  for (;;) Line 4244  for (;;)
4244    
4245    }             /* End of main loop */    }             /* End of main loop */
4246  /* Control never reaches here */  /* Control never reaches here */
4247    
4248    
4249    /* When compiling to use the heap rather than the stack for recursive calls to
4250    match(), the RRETURN() macro jumps here. The number that is saved in
4251    frame->Xwhere indicates which label we actually want to return to. */
4252    
4253    #ifdef NO_RECURSE
4254    #define LBL(val) case val: goto L_RM##val;
4255    HEAP_RETURN:
4256    switch (frame->Xwhere)
4257      {
4258      LBL( 1) LBL( 2) LBL( 3) LBL( 4) LBL( 5) LBL( 6) LBL( 7) LBL( 8)
4259      LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)
4260      LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)
4261      LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)
4262      LBL(53) LBL(54)
4263    #ifdef SUPPORT_UTF8
4264      LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)
4265      LBL(32) LBL(34) LBL(42) LBL(46)
4266    #ifdef SUPPORT_UCP
4267      LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)
4268    #endif  /* SUPPORT_UCP */
4269    #endif  /* SUPPORT_UTF8 */
4270      default:
4271      DPRINTF(("jump error in pcre match: label %d non-existent\n", frame->Xwhere));
4272      return PCRE_ERROR_INTERNAL;
4273      }
4274    #undef LBL
4275    #endif  /* NO_RECURSE */
4276  }  }
4277    
4278    
# Line 3127  Undefine all the macros that were define Line 4285  Undefine all the macros that were define
4285  #ifdef NO_RECURSE  #ifdef NO_RECURSE
4286  #undef eptr  #undef eptr
4287  #undef ecode  #undef ecode
4288    #undef mstart
4289  #undef offset_top  #undef offset_top
4290  #undef ims  #undef ims
4291  #undef eptrb  #undef eptrb
# Line 3144  Undefine all the macros that were define Line 4303  Undefine all the macros that were define
4303    
4304  #undef cur_is_word  #undef cur_is_word
4305  #undef condition  #undef condition
 #undef minimize  
4306  #undef prev_is_word  #undef prev_is_word
4307    
4308  #undef original_ims  #undef original_ims
# Line 3200  Returns:          > 0 => success; value Line 4358  Returns:          > 0 => success; value
4358                   < -1 => some kind of unexpected problem                   < -1 => some kind of unexpected problem
4359  */  */
4360    
4361  PCRE_EXPORT int  PCRE_EXP_DEFN int
4362  pcre_exec(const pcre *argument_re, const pcre_extra *extra_data,  pcre_exec(const pcre *argument_re, const pcre_extra *extra_data,
4363    const char *subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
4364    int offsetcount)    int offsetcount)
4365  {  {
4366  int rc, resetcount, ocount;  int rc, resetcount, ocount;
4367  int first_byte = -1;  int first_byte = -1;
4368  int req_byte = -1;  int req_byte = -1;
4369  int req_byte2 = -1;  int req_byte2 = -1;
4370  unsigned long int ims = 0;  int newline;
4371    unsigned long int ims;
4372  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
4373  BOOL anchored;  BOOL anchored;
4374  BOOL startline;  BOOL startline;
4375  BOOL firstline;  BOOL firstline;
4376  BOOL first_byte_caseless = FALSE;  BOOL first_byte_caseless = FALSE;
4377  BOOL req_byte_caseless = FALSE;  BOOL req_byte_caseless = FALSE;
4378    BOOL utf8;
4379  match_data match_block;  match_data match_block;
4380    match_data *md = &match_block;
4381  const uschar *tables;  const uschar *tables;
4382  const uschar *start_bits = NULL;  const uschar *start_bits = NULL;
4383  const uschar *start_match = (const uschar *)subject + start_offset;  USPTR start_match = (USPTR)subject + start_offset;
4384  const uschar *end_subject;  USPTR end_subject;
4385  const uschar *req_byte_ptr = start_match - 1;  USPTR req_byte_ptr = start_match - 1;
4386    
4387  pcre_study_data internal_study;  pcre_study_data internal_study;
4388  const pcre_study_data *study;  const pcre_study_data *study;
# Line 3241  if (offsetcount < 0) return PCRE_ERROR_B Line 4402  if (offsetcount < 0) return PCRE_ERROR_B
4402  the default values. */  the default values. */
4403    
4404  study = NULL;  study = NULL;
4405  match_block.match_limit = MATCH_LIMIT;  md->match_limit = MATCH_LIMIT;
4406  match_block.callout_data = NULL;  md->match_limit_recursion = MATCH_LIMIT_RECURSION;
4407    md->callout_data = NULL;
4408    
4409  /* The table pointer is always in native byte order. */  /* The table pointer is always in native byte order. */
4410    
# Line 3254  if (extra_data != NULL) Line 4416  if (extra_data != NULL)
4416    if ((flags & PCRE_EXTRA_STUDY_DATA) != 0)    if ((flags & PCRE_EXTRA_STUDY_DATA) != 0)
4417      study = (const pcre_study_data *)extra_data->study_data;      study = (const pcre_study_data *)extra_data->study_data;
4418    if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0)    if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0)
4419      match_block.match_limit = extra_data->match_limit;      md->match_limit = extra_data->match_limit;
4420      if ((flags & PCRE_EXTRA_MATCH_LIMIT_RECURSION) != 0)
4421        md->match_limit_recursion = extra_data->match_limit_recursion;
4422    if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0)    if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0)
4423      match_block.callout_data = extra_data->callout_data;      md->callout_data = extra_data->callout_data;
4424    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;
4425    }    }
4426    
# Line 3281  if (re->magic_number != MAGIC_NUMBER) Line 4445  if (re->magic_number != MAGIC_NUMBER)
4445  /* Set up other data */  /* Set up other data */
4446    
4447  anchored = ((re->options | options) & PCRE_ANCHORED) != 0;  anchored = ((re->options | options) & PCRE_ANCHORED) != 0;
4448  startline = (re->options & PCRE_STARTLINE) != 0;  startline = (re->flags & PCRE_STARTLINE) != 0;
4449  firstline = (re->options & PCRE_FIRSTLINE) != 0;  firstline = (re->options & PCRE_FIRSTLINE) != 0;
4450    
4451  /* The code starts after the real_pcre block and the capture name table. */  /* The code starts after the real_pcre block and the capture name table. */
4452    
4453  match_block.start_code = (const uschar *)external_re + re->name_table_offset +  md->start_code = (const uschar *)external_re + re->name_table_offset +
4454    re->name_count * re->name_entry_size;    re->name_count * re->name_entry_size;
4455    
4456  match_block.start_subject = (const uschar *)subject;  md->start_subject = (USPTR)subject;
4457  match_block.start_offset = start_offset;  md->start_offset = start_offset;
4458  match_block.end_subject = match_block.start_subject + length;  md->end_subject = md->start_subject + length;
4459  end_subject = match_block.end_subject;  end_subject = md->end_subject;
4460    
4461  match_block.endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
4462  match_block.utf8 = (re->options & PCRE_UTF8) != 0;  utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;
4463    md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
4464  match_block.notbol = (options & PCRE_NOTBOL) != 0;  
4465  match_block.noteol = (options & PCRE_NOTEOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
4466  match_block.notempty = (options & PCRE_NOTEMPTY) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
4467  match_block.partial = (options & PCRE_PARTIAL) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
4468  match_block.hitend = FALSE;  md->partial = (options & PCRE_PARTIAL) != 0;
4469    md->hitend = FALSE;
4470    
4471    md->recursive = NULL;                   /* No recursion at top level */
4472    
4473  match_block.recursive = NULL;                   /* No recursion at top level */  md->lcc = tables + lcc_offset;
4474    md->ctypes = tables + ctypes_offset;
4475    
4476  match_block.lcc = tables + lcc_offset;  /* Handle different \R options. */
4477  match_block.ctypes = tables + ctypes_offset;  
4478    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
4479      {
4480      case 0:
4481      if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0)
4482        md->bsr_anycrlf = (re->options & PCRE_BSR_ANYCRLF) != 0;
4483      else
4484    #ifdef BSR_ANYCRLF
4485      md->bsr_anycrlf = TRUE;
4486    #else
4487      md->bsr_anycrlf = FALSE;
4488    #endif
4489      break;
4490    
4491      case PCRE_BSR_ANYCRLF:
4492      md->bsr_anycrlf = TRUE;
4493      break;
4494    
4495      case PCRE_BSR_UNICODE:
4496      md->bsr_anycrlf = FALSE;
4497      break;
4498    
4499      default: return PCRE_ERROR_BADNEWLINE;
4500      }
4501    
4502    /* Handle different types of newline. The three bits give eight cases. If
4503    nothing is set at run time, whatever was used at compile time applies. */
4504    
4505    switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options :
4506            (pcre_uint32)options) & PCRE_NEWLINE_BITS)
4507      {
4508      case 0: newline = NEWLINE; break;   /* Compile-time default */
4509      case PCRE_NEWLINE_CR: newline = '\r'; break;
4510      case PCRE_NEWLINE_LF: newline = '\n'; break;
4511      case PCRE_NEWLINE_CR+
4512           PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;
4513      case PCRE_NEWLINE_ANY: newline = -1; break;
4514      case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
4515      default: return PCRE_ERROR_BADNEWLINE;
4516      }
4517    
4518    if (newline == -2)
4519      {
4520      md->nltype = NLTYPE_ANYCRLF;
4521      }
4522    else if (newline < 0)
4523      {
4524      md->nltype = NLTYPE_ANY;
4525      }
4526    else
4527      {
4528      md->nltype = NLTYPE_FIXED;
4529      if (newline > 255)
4530        {
4531        md->nllen = 2;
4532        md->nl[0] = (newline >> 8) & 255;
4533        md->nl[1] = newline & 255;
4534        }
4535      else
4536        {
4537        md->nllen = 1;
4538        md->nl[0] = newline;
4539        }
4540      }
4541    
4542  /* Partial matching is supported only for a restricted set of regexes at the  /* Partial matching is supported only for a restricted set of regexes at the
4543  moment. */  moment. */
4544    
4545  if (match_block.partial && (re->options & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
4546    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
4547    
4548  /* Check a UTF-8 string if required. Unfortunately there's no way of passing  /* Check a UTF-8 string if required. Unfortunately there's no way of passing
4549  back the character offset. */  back the character offset. */
4550    
4551  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
4552  if (match_block.utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
4553    {    {
4554    if (_pcre_valid_utf8((uschar *)subject, length) >= 0)    if (_pcre_valid_utf8((uschar *)subject, length) >= 0)
4555      return PCRE_ERROR_BADUTF8;      return PCRE_ERROR_BADUTF8;
# Line 3350  ocount = offsetcount - (offsetcount % 3) Line 4581  ocount = offsetcount - (offsetcount % 3)
4581  if (re->top_backref > 0 && re->top_backref >= ocount/3)  if (re->top_backref > 0 && re->top_backref >= ocount/3)
4582    {    {
4583    ocount = re->top_backref * 3 + 3;    ocount = re->top_backref * 3 + 3;
4584    match_block.offset_vector = (int *)(pcre_malloc)(ocount * sizeof(int));    md->offset_vector = (int *)(pcre_malloc)(ocount * sizeof(int));
4585    if (match_block.offset_vector == NULL) return PCRE_ERROR_NOMEMORY;    if (md->offset_vector == NULL) return PCRE_ERROR_NOMEMORY;
4586    using_temporary_offsets = TRUE;    using_temporary_offsets = TRUE;
4587    DPRINTF(("Got memory to hold back references\n"));    DPRINTF(("Got memory to hold back references\n"));
4588    }    }
4589  else match_block.offset_vector = offsets;  else md->offset_vector = offsets;
4590    
4591  match_block.offset_end = ocount;  md->offset_end = ocount;
4592  match_block.offset_max = (2*ocount)/3;  md->offset_max = (2*ocount)/3;
4593  match_block.offset_overflow = FALSE;  md->offset_overflow = FALSE;
4594  match_block.capture_last = -1;  md->capture_last = -1;
4595    
4596  /* Compute the minimum number of offsets that we need to reset each time. Doing  /* Compute the minimum number of offsets that we need to reset each time. Doing
4597  this makes a huge difference to execution time when there aren't many brackets  this makes a huge difference to execution time when there aren't many brackets
# Line 3373  if (resetcount > offsetcount) resetcount Line 4604  if (resetcount > offsetcount) resetcount
4604  never be used unless previously set, but they get saved and restored, and so we  never be used unless previously set, but they get saved and restored, and so we
4605  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. */
4606    
4607  if (match_block.offset_vector != NULL)  if (md->offset_vector != NULL)
4608    {    {
4609    register int *iptr = match_block.offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
4610    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - resetcount/2 + 1;
4611    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
4612    }    }
# Line 3388  studied, there may be a bitmap of possib Line 4619  studied, there may be a bitmap of possib
4619    
4620  if (!anchored)  if (!anchored)
4621    {    {
4622    if ((re->options & PCRE_FIRSTSET) != 0)    if ((re->flags & PCRE_FIRSTSET) != 0)
4623      {      {
4624      first_byte = re->first_byte & 255;      first_byte = re->first_byte & 255;
4625      if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE)      if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE)
4626        first_byte = match_block.lcc[first_byte];        first_byte = md->lcc[first_byte];
4627      }      }
4628    else    else
4629      if (!startline && study != NULL &&      if (!startline && study != NULL &&
# Line 3403  if (!anchored) Line 4634  if (!anchored)
4634  /* For anchored or unanchored matches, there may be a "last known required  /* For anchored or unanchored matches, there may be a "last known required
4635  character" set. */  character" set. */
4636    
4637  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
4638    {    {
4639    req_byte = re->req_byte & 255;    req_byte = re->req_byte & 255;
4640    req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;    req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;
4641    req_byte2 = (tables + fcc_offset)[req_byte];  /* case flipped */    req_byte2 = (tables + fcc_offset)[req_byte];  /* case flipped */
4642    }    }
4643    
4644    
4645    /* ==========================================================================*/
4646    
4647  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
4648  the loop runs just once. */  the loop runs just once. */
4649    
4650  do  for(;;)
4651    {    {
4652    const uschar *save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
4653      USPTR new_start_match;
4654    
4655    /* Reset the maximum number of extractions we might see. */    /* Reset the maximum number of extractions we might see. */
4656    
4657    if (match_block.offset_vector != NULL)    if (md->offset_vector != NULL)
4658      {      {
4659      register int *iptr = match_block.offset_vector;      register int *iptr = md->offset_vector;
4660      register int *iend = iptr + resetcount;      register int *iend = iptr + resetcount;
4661      while (iptr < iend) *iptr++ = -1;      while (iptr < iend) *iptr++ = -1;
4662      }      }
4663    
4664    /* Advance to a unique first char if possible. If firstline is TRUE, the    /* Advance to a unique first char if possible. If firstline is TRUE, the
4665    start of the match is constrained to the first line of a multiline string.    start of the match is constrained to the first line of a multiline string.
4666    Implement this by temporarily adjusting end_subject so that we stop scanning    That is, the match must be before or at the first newline. Implement this by
4667    at a newline. If the match fails at the newline, later code breaks this loop.    temporarily adjusting end_subject so that we stop scanning at a newline. If
4668    */    the match fails at the newline, later code breaks this loop. */
4669    
4670    if (firstline)    if (firstline)
4671      {      {
4672      const uschar *t = start_match;      USPTR t = start_match;
4673      while (t < save_end_subject && *t != '\n') t++;      while (t < md->end_subject && !IS_NEWLINE(t)) t++;
4674      end_subject = t;      end_subject = t;
4675      }      }
4676    
# Line 3445  do Line 4680  do
4680      {      {
4681      if (first_byte_caseless)      if (first_byte_caseless)
4682        while (start_match < end_subject &&        while (start_match < end_subject &&
4683               match_block.lcc[*start_match] != first_byte)               md->lcc[*start_match] != first_byte)
4684          start_match++;          { NEXTCHAR(start_match); }
4685      else      else
4686        while (start_match < end_subject && *start_match != first_byte)        while (start_match < end_subject && *start_match != first_byte)
4687          start_match++;          { NEXTCHAR(start_match); }
4688      }      }
4689    
4690    /* Or to just after \n for a multiline match if possible */    /* Or to just after a linebreak for a multiline match if possible */
4691    
4692    else if (startline)    else if (startline)
4693      {      {
4694      if (start_match > match_block.start_subject + start_offset)      if (start_match > md->start_subject + start_offset)
4695        {        {
4696        while (start_match < end_subject && start_match[-1] != NEWLINE)        while (start_match <= end_subject && !WAS_NEWLINE(start_match))
4697            { NEXTCHAR(start_match); }
4698    
4699          /* If we have just passed a CR and the newline option is ANY or ANYCRLF,
4700          and we are now at a LF, advance the match position by one more character.
4701          */
4702    
4703          if (start_match[-1] == '\r' &&
4704               (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
4705               start_match < end_subject &&
4706               *start_match == '\n')
4707          start_match++;          start_match++;
4708        }        }
4709      }      }
# Line 3470  do Line 4715  do
4715      while (start_match < end_subject)      while (start_match < end_subject)
4716        {        {
4717        register unsigned int c = *start_match;        register unsigned int c = *start_match;
4718        if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++; else break;        if ((start_bits[c/8] & (1 << (c&7))) == 0)
4719            { NEXTCHAR(start_match); }
4720          else break;
4721        }        }
4722      }      }
4723    
# Line 3480  do Line 4727  do
4727    
4728  #ifdef DEBUG  /* Sigh. Some compilers never learn. */  #ifdef DEBUG  /* Sigh. Some compilers never learn. */
4729    printf(">>>> Match against: ");    printf(">>>> Match against: ");
4730    pchars(start_match, end_subject - start_match, TRUE, &match_block);    pchars(start_match, end_subject - start_match, TRUE, md);
4731    printf("\n");    printf("\n");
4732  #endif  #endif
4733    
# Line 3494  do Line 4741  do
4741    
4742    HOWEVER: when the subject string is very, very long, searching to its end can    HOWEVER: when the subject string is very, very long, searching to its end can
4743    take a long time, and give bad performance on quite ordinary patterns. This    take a long time, and give bad performance on quite ordinary patterns. This
4744    showed up when somebody was matching /^C/ on a 32-megabyte string... so we    showed up when somebody was matching something like /^\d+C/ on a 32-megabyte
4745    don't do this when the string is sufficiently long.    string... so we don't do this when the string is sufficiently long.
4746    
4747    ALSO: this processing is disabled when partial matching is requested.    ALSO: this processing is disabled when partial matching is requested.
4748    */    */
4749    
4750    if (req_byte >= 0 &&    if (req_byte >= 0 &&
4751        end_subject - start_match < REQ_BYTE_MAX &&        end_subject - start_match < REQ_BYTE_MAX &&
4752        !match_block.partial)        !md->partial)
4753      {      {
4754      register const uschar *p = start_match + ((first_byte >= 0)? 1 : 0);      register USPTR p = start_match + ((first_byte >= 0)? 1 : 0);
4755    
4756      /* We don't need to repeat the search if we haven't yet reached the      /* We don't need to repeat the search if we haven't yet reached the
4757      place we found it at last time. */      place we found it at last time. */
# Line 3527  do Line 4774  do
4774            }            }
4775          }          }
4776    
4777        /* If we can't find the required character, break the matching loop */        /* If we can't find the required character, break the matching loop,
4778          forcing a match failure. */
4779    
4780        if (p >= end_subject) break;        if (p >= end_subject)
4781            {
4782            rc = MATCH_NOMATCH;
4783            break;
4784            }
4785    
4786        /* If we have found the required character, save the point where we        /* If we have found the required character, save the point where we
4787        found it, so that we don't search again next time round the loop if        found it, so that we don't search again next time round the loop if
# Line 3539  do Line 4791  do
4791        }        }
4792      }      }
4793    
4794    /* When a match occurs, substrings will be set for all internal extractions;    /* OK, we can now run the match. */
4795    we just need to set up the whole thing as substring 0 before returning. If  
4796    there were too many extractions, set the return code to zero. In the case    md->start_match_ptr = start_match;
4797    where we had to get some local store to hold offsets for backreferences, copy    md->match_call_count = 0;
4798    those back references that we can. In this case there need not be overflow    rc = match(start_match, md->start_code, start_match, 2, md, ims, NULL, 0, 0);
   if certain parts of the pattern were not used. */  
   
   match_block.start_match = start_match;  
   match_block.match_call_count = 0;  
   
   rc = match(start_match, match_block.start_code, 2, &match_block, ims, NULL,  
     match_isgroup);  
   
   /* When the result is no match, if the subject's first character was a  
   newline and the PCRE_FIRSTLINE option is set, break (which will return  
   PCRE_ERROR_NOMATCH). The option requests that a match occur before the first  
   newline in the subject. Otherwise, advance the pointer to the next character  
   and continue - but the continuation will actually happen only when the  
   pattern is not anchored. */  
4799    
4800    if (rc == MATCH_NOMATCH)    switch(rc)
4801      {      {
4802      if (firstline && *start_match == NEWLINE) break;      /* NOMATCH and PRUNE advance by one character. THEN at this level acts
4803      start_match++;      exactly like PRUNE. */
4804    
4805        case MATCH_NOMATCH:
4806        case MATCH_PRUNE:
4807        case MATCH_THEN:
4808        new_start_match = start_match + 1;
4809  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
4810      if (match_block.utf8)      if (utf8)
4811        while(start_match < end_subject && (*start_match & 0xc0) == 0x80)        while(new_start_match < end_subject && (*new_start_match & 0xc0) == 0x80)
4812          start_match++;          new_start_match++;
4813  #endif  #endif
4814      continue;      break;
     }  
4815    
4816    if (rc != MATCH_MATCH)      /* SKIP passes back the next starting point explicitly. */
4817      {  
4818      DPRINTF((">>>> error: returning %d\n", rc));      case MATCH_SKIP:
4819      return rc;      new_start_match = md->start_match_ptr;
4820        break;
4821    
4822        /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */
4823    
4824        case MATCH_COMMIT:
4825        rc = MATCH_NOMATCH;
4826        goto ENDLOOP;
4827    
4828        /* Any other return is some kind of error. */
4829    
4830        default:
4831        goto ENDLOOP;
4832      }      }
4833    
4834    /* We have a match! Copy the offset information from temporary store if    /* Control reaches here for the various types of "no match at this point"
4835    necessary */    result. Reset the code to MATCH_NOMATCH for subsequent checking. */
4836    
4837      rc = MATCH_NOMATCH;
4838    
4839      /* If PCRE_FIRSTLINE is set, the match must happen before or at the first
4840      newline in the subject (though it may continue over the newline). Therefore,
4841      if we have just failed to match, starting at a newline, do not continue. */
4842    
4843      if (firstline && IS_NEWLINE(start_match)) break;
4844    
4845      /* Advance to new matching position */
4846    
4847      start_match = new_start_match;
4848    
4849      /* Break the loop if the pattern is anchored or if we have passed the end of
4850      the subject. */
4851    
4852      if (anchored || start_match > end_subject) break;
4853    
4854      /* If we have just passed a CR and we are now at a LF, and the pattern does
4855      not contain any explicit matches for \r or \n, and the newline option is CRLF
4856      or ANY or ANYCRLF, advance the match position by one more character. */
4857    
4858      if (start_match[-1] == '\r' &&
4859          start_match < end_subject &&
4860          *start_match == '\n' &&
4861          (re->flags & PCRE_HASCRORLF) == 0 &&
4862            (md->nltype == NLTYPE_ANY ||
4863             md->nltype == NLTYPE_ANYCRLF ||
4864             md->nllen == 2))
4865        start_match++;
4866    
4867      }   /* End of for(;;) "bumpalong" loop */
4868    
4869    /* ==========================================================================*/
4870    
4871    /* We reach here when rc is not MATCH_NOMATCH, or if one of the stopping
4872    conditions is true:
4873    
4874    (1) The pattern is anchored or the match was failed by (*COMMIT);
4875