/[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 345 by ph10, Mon Apr 28 15:10:02 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;        int chartype, script;
1657        int othercase;        int category = _pcre_ucp_findprop(c, &chartype, &script);
       int category = _pcre_ucp_findchar(c, &chartype, &othercase);  
   
       rqdtype = *(++ecode);  
       ecode++;  
1658    
1659        if (rqdtype >= 128)        switch(ecode[1])
1660          {          {
1661          if ((rqdtype - 128 != category) == (op == OP_PROP))          case PT_ANY:
1662            if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
1663            break;
1664    
1665            case PT_LAMP:
1666            if ((chartype == ucp_Lu ||
1667                 chartype == ucp_Ll ||
1668                 chartype == ucp_Lt) == (op == OP_NOTPROP))
1669            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1670          }           break;
1671        else  
1672          {          case PT_GC:
1673          if ((rqdtype != chartype) == (op == OP_PROP))          if ((ecode[2] != category) == (op == OP_PROP))
1674              RRETURN(MATCH_NOMATCH);
1675            break;
1676    
1677            case PT_PC:
1678            if ((ecode[2] != chartype) == (op == OP_PROP))
1679            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1680            break;
1681    
1682            case PT_SC:
1683            if ((ecode[2] != script) == (op == OP_PROP))
1684              RRETURN(MATCH_NOMATCH);
1685            break;
1686    
1687            default:
1688            RRETURN(PCRE_ERROR_INTERNAL);
1689          }          }
1690    
1691          ecode += 3;
1692        }        }
1693      break;      break;
1694    
# Line 1379  for (;;) Line 1699  for (;;)
1699      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);      if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1700      GETCHARINCTEST(c, eptr);      GETCHARINCTEST(c, eptr);
1701        {        {
1702        int chartype;        int chartype, script;
1703        int othercase;        int category = _pcre_ucp_findprop(c, &chartype, &script);
       int category = _pcre_ucp_findchar(c, &chartype, &othercase);  
1704        if (category == ucp_M) RRETURN(MATCH_NOMATCH);        if (category == ucp_M) RRETURN(MATCH_NOMATCH);
1705        while (eptr < md->end_subject)        while (eptr < md->end_subject)
1706          {          {
# Line 1390  for (;;) Line 1709  for (;;)
1709            {            {
1710            GETCHARLEN(c, eptr, len);            GETCHARLEN(c, eptr, len);
1711            }            }
1712          category = _pcre_ucp_findchar(c, &chartype, &othercase);          category = _pcre_ucp_findprop(c, &chartype, &script);
1713          if (category != ucp_M) break;          if (category != ucp_M) break;
1714          eptr += len;          eptr += len;
1715          }          }
# Line 1411  for (;;) Line 1730  for (;;)
1730      case OP_REF:      case OP_REF:
1731        {        {
1732        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
1733        ecode += 3;                                 /* Advance past item */        ecode += 3;
1734    
1735        /* If the reference is unset, set the length to be longer than the amount        /* If the reference is unset, there are two possibilities:
1736        of subject left; this ensures that every attempt at a match fails. We  
1737        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
1738        minima. */        than the amount of subject left; this ensures that every attempt at a
1739          match fails. We can't just fail here, because of the possibility of
1740        length = (offset >= offset_top || md->offset_vector[offset] < 0)?        quantifiers with zero minima.
1741          md->end_subject - eptr + 1 :  
1742          md->offset_vector[offset+1] - md->offset_vector[offset];        (b) If the JavaScript compatibility flag is set, set the length to zero
1743          so that the back reference matches an empty string.
1744    
1745          Otherwise, set the length to the length of what was matched by the
1746          referenced subpattern. */
1747    
1748          if (offset >= offset_top || md->offset_vector[offset] < 0)
1749            length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;
1750          else
1751            length = md->offset_vector[offset+1] - md->offset_vector[offset];
1752    
1753        /* Set up for repetition, or handle the non-repeated case */        /* Set up for repetition, or handle the non-repeated case */
1754    
# Line 1480  for (;;) Line 1808  for (;;)
1808          {          {
1809          for (fi = min;; fi++)          for (fi = min;; fi++)
1810            {            {
1811            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);
1812            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1813            if (fi >= max || !match_ref(offset, eptr, length, md, ims))            if (fi >= max || !match_ref(offset, eptr, length, md, ims))
1814              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
# Line 1501  for (;;) Line 1829  for (;;)
1829            }            }
1830          while (eptr >= pp)          while (eptr >= pp)
1831            {            {
1832            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);
1833            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1834            eptr -= length;            eptr -= length;
1835            }            }
# Line 1606  for (;;) Line 1934  for (;;)
1934            {            {
1935            for (fi = min;; fi++)            for (fi = min;; fi++)
1936              {              {
1937              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16);
1938              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1939              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1940              GETCHARINC(c, eptr);              GETCHARINC(c, eptr);
# Line 1626  for (;;) Line 1954  for (;;)
1954            {            {
1955            for (fi = min;; fi++)            for (fi = min;; fi++)
1956              {              {
1957              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17);
1958              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1959              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
1960              c = *eptr++;              c = *eptr++;
# Line 1663  for (;;) Line 1991  for (;;)
1991              }              }
1992            for (;;)            for (;;)
1993              {              {
1994              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM18);
1995              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1996              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
1997              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 1682  for (;;) Line 2010  for (;;)
2010              }              }
2011            while (eptr >= pp)            while (eptr >= pp)
2012              {              {
2013              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM19);
             eptr--;  
2014              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2015                eptr--;
2016              }              }
2017            }            }
2018    
# Line 1753  for (;;) Line 2081  for (;;)
2081          {          {
2082          for (fi = min;; fi++)          for (fi = min;; fi++)
2083            {            {
2084            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20);
2085            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2086            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2087            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
# Line 1777  for (;;) Line 2105  for (;;)
2105            }            }
2106          for(;;)          for(;;)
2107            {            {
2108            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21);
2109            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2110            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
2111            BACKCHAR(eptr)            if (utf8) BACKCHAR(eptr);
2112            }            }
2113          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
2114          }          }
# Line 1836  for (;;) Line 2164  for (;;)
2164    
2165        else        else
2166          {          {
2167          int dc;          unsigned int dc;
2168          GETCHARINC(dc, eptr);          GETCHARINC(dc, eptr);
2169          ecode += length;          ecode += length;
2170    
2171          /* 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
2172          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. */  
2173    
2174          if (fc != dc)          if (fc != dc)
2175            {            {
2176  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2177            int chartype;            if (dc != _pcre_ucp_othercase(fc))
           int othercase;  
           if (_pcre_ucp_findchar(fc, &chartype, &othercase) < 0 || dc != othercase)  
2178  #endif  #endif
2179              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2180            }            }
# Line 1867  for (;;) Line 2191  for (;;)
2191        }        }
2192      break;      break;
2193    
2194      /* Match a single character repeatedly; different opcodes share code. */      /* Match a single character repeatedly. */
2195    
2196      case OP_EXACT:      case OP_EXACT:
2197      min = max = GET2(ecode, 1);      min = max = GET2(ecode, 1);
2198      ecode += 3;      ecode += 3;
2199      goto REPEATCHAR;      goto REPEATCHAR;
2200    
2201        case OP_POSUPTO:
2202        possessive = TRUE;
2203        /* Fall through */
2204    
2205      case OP_UPTO:      case OP_UPTO:
2206      case OP_MINUPTO:      case OP_MINUPTO:
2207      min = 0;      min = 0;
# Line 1882  for (;;) Line 2210  for (;;)
2210      ecode += 3;      ecode += 3;
2211      goto REPEATCHAR;      goto REPEATCHAR;
2212    
2213      case OP_STAR:      case OP_POSSTAR:
2214      case OP_MINSTAR:      possessive = TRUE;
2215      case OP_PLUS:      min = 0;
2216      case OP_MINPLUS:      max = INT_MAX;
2217      case OP_QUERY:      ecode++;
2218      case OP_MINQUERY:      goto REPEATCHAR;
2219      c = *ecode++ - OP_STAR;  
2220      minimize = (c & 1) != 0;      case OP_POSPLUS:
2221        possessive = TRUE;
2222        min = 1;
2223        max = INT_MAX;
2224        ecode++;
2225        goto REPEATCHAR;
2226    
2227        case OP_POSQUERY:
2228        possessive = TRUE;
2229        min = 0;
2230        max = 1;
2231        ecode++;
2232        goto REPEATCHAR;
2233    
2234        case OP_STAR:
2235        case OP_MINSTAR:
2236        case OP_PLUS:
2237        case OP_MINPLUS:
2238        case OP_QUERY:
2239        case OP_MINQUERY:
2240        c = *ecode++ - OP_STAR;
2241        minimize = (c & 1) != 0;
2242      min = rep_min[c];                 /* Pick up values from tables; */      min = rep_min[c];                 /* Pick up values from tables; */
2243      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
2244      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
# Line 1913  for (;;) Line 2262  for (;;)
2262    
2263        if (length > 1)        if (length > 1)
2264          {          {
         int oclength = 0;  
         uschar occhars[8];  
   
2265  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2266          int othercase;          unsigned int othercase;
         int chartype;  
2267          if ((ims & PCRE_CASELESS) != 0 &&          if ((ims & PCRE_CASELESS) != 0 &&
2268               _pcre_ucp_findchar(fc, &chartype, &othercase) >= 0 &&              (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR)
              othercase > 0)  
2269            oclength = _pcre_ord2utf8(othercase, occhars);            oclength = _pcre_ord2utf8(othercase, occhars);
2270            else oclength = 0;
2271  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2272    
2273          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2274            {            {
2275            if (memcmp(eptr, charptr, length) == 0) eptr += length;            if (memcmp(eptr, charptr, length) == 0) eptr += length;
2276    #ifdef SUPPORT_UCP
2277            /* Need braces because of following else */            /* Need braces because of following else */
2278            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }            else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }
2279            else            else
# Line 1935  for (;;) Line 2281  for (;;)
2281              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);              if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);
2282              eptr += oclength;              eptr += oclength;
2283              }              }
2284    #else   /* without SUPPORT_UCP */
2285              else { RRETURN(MATCH_NOMATCH); }
2286    #endif  /* SUPPORT_UCP */
2287            }            }
2288    
2289          if (min == max) continue;          if (min == max) continue;
# Line 1943  for (;;) Line 2292  for (;;)
2292            {            {
2293            for (fi = min;; fi++)            for (fi = min;; fi++)
2294              {              {
2295              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22);
2296              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2297              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2298              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (memcmp(eptr, charptr, length) == 0) eptr += length;
2299    #ifdef SUPPORT_UCP
2300              /* Need braces because of following else */              /* Need braces because of following else */
2301              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }              else if (oclength == 0) { RRETURN(MATCH_NOMATCH); }
2302              else              else
# Line 1954  for (;;) Line 2304  for (;;)
2304                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);                if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH);
2305                eptr += oclength;                eptr += oclength;
2306                }                }
2307    #else   /* without SUPPORT_UCP */
2308                else { RRETURN (MATCH_NOMATCH); }
2309    #endif  /* SUPPORT_UCP */
2310              }              }
2311            /* Control never gets here */            /* Control never gets here */
2312            }            }
2313          else  
2314            else  /* Maximize */
2315            {            {
2316            pp = eptr;            pp = eptr;
2317            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2318              {              {
2319              if (eptr > md->end_subject - length) break;              if (eptr > md->end_subject - length) break;
2320              if (memcmp(eptr, charptr, length) == 0) eptr += length;              if (memcmp(eptr, charptr, length) == 0) eptr += length;
2321    #ifdef SUPPORT_UCP
2322              else if (oclength == 0) break;              else if (oclength == 0) break;
2323              else              else
2324                {                {
2325                if (memcmp(eptr, occhars, oclength) != 0) break;                if (memcmp(eptr, occhars, oclength) != 0) break;
2326                eptr += oclength;                eptr += oclength;
2327                }                }
2328    #else   /* without SUPPORT_UCP */
2329                else break;
2330    #endif  /* SUPPORT_UCP */
2331              }              }
2332            while (eptr >= pp)  
2333              if (possessive) continue;
2334              for(;;)
2335             {             {
2336             RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);             RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23);
2337             if (rrc != MATCH_NOMATCH) RRETURN(rrc);             if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2338               if (eptr == pp) RRETURN(MATCH_NOMATCH);
2339    #ifdef SUPPORT_UCP
2340               eptr--;
2341               BACKCHAR(eptr);
2342    #else   /* without SUPPORT_UCP */
2343             eptr -= length;             eptr -= length;
2344    #endif  /* SUPPORT_UCP */
2345             }             }
           RRETURN(MATCH_NOMATCH);  
2346            }            }
2347          /* Control never gets here */          /* Control never gets here */
2348          }          }
# Line 2017  for (;;) Line 2382  for (;;)
2382          {          {
2383          for (fi = min;; fi++)          for (fi = min;; fi++)
2384            {            {
2385            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24);
2386            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2387            if (fi >= max || eptr >= md->end_subject ||            if (fi >= max || eptr >= md->end_subject ||
2388                fc != md->lcc[*eptr++])                fc != md->lcc[*eptr++])
# Line 2025  for (;;) Line 2390  for (;;)
2390            }            }
2391          /* Control never gets here */          /* Control never gets here */
2392          }          }
2393        else        else  /* Maximize */
2394          {          {
2395          pp = eptr;          pp = eptr;
2396          for (i = min; i < max; i++)          for (i = min; i < max; i++)
# Line 2033  for (;;) Line 2398  for (;;)
2398            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;            if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break;
2399            eptr++;            eptr++;
2400            }            }
2401            if (possessive) continue;
2402          while (eptr >= pp)          while (eptr >= pp)
2403            {            {
2404            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25);
2405            eptr--;            eptr--;
2406            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2407            }            }
# Line 2054  for (;;) Line 2420  for (;;)
2420          {          {
2421          for (fi = min;; fi++)          for (fi = min;; fi++)
2422            {            {
2423            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26);
2424            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2425            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)            if (fi >= max || eptr >= md->end_subject || fc != *eptr++)
2426              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2427            }            }
2428          /* Control never gets here */          /* Control never gets here */
2429          }          }
2430        else        else  /* Maximize */
2431          {          {
2432          pp = eptr;          pp = eptr;
2433          for (i = min; i < max; i++)          for (i = min; i < max; i++)
# Line 2069  for (;;) Line 2435  for (;;)
2435            if (eptr >= md->end_subject || fc != *eptr) break;            if (eptr >= md->end_subject || fc != *eptr) break;
2436            eptr++;            eptr++;
2437            }            }
2438            if (possessive) continue;
2439          while (eptr >= pp)          while (eptr >= pp)
2440            {            {
2441            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27);
2442            eptr--;            eptr--;
2443            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2444            }            }
# Line 2121  for (;;) Line 2488  for (;;)
2488      ecode += 3;      ecode += 3;
2489      goto REPEATNOTCHAR;      goto REPEATNOTCHAR;
2490    
2491        case OP_NOTPOSSTAR:
2492        possessive = TRUE;
2493        min = 0;
2494        max = INT_MAX;
2495        ecode++;
2496        goto REPEATNOTCHAR;
2497    
2498        case OP_NOTPOSPLUS:
2499        possessive = TRUE;
2500        min = 1;
2501        max = INT_MAX;
2502        ecode++;
2503        goto REPEATNOTCHAR;
2504    
2505        case OP_NOTPOSQUERY:
2506        possessive = TRUE;
2507        min = 0;
2508        max = 1;
2509        ecode++;
2510        goto REPEATNOTCHAR;
2511    
2512        case OP_NOTPOSUPTO:
2513        possessive = TRUE;
2514        min = 0;
2515        max = GET2(ecode, 1);
2516        ecode += 3;
2517        goto REPEATNOTCHAR;
2518    
2519      case OP_NOTSTAR:      case OP_NOTSTAR:
2520      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
2521      case OP_NOTPLUS:      case OP_NOTPLUS:
# Line 2160  for (;;) Line 2555  for (;;)
2555        /* UTF-8 mode */        /* UTF-8 mode */
2556        if (utf8)        if (utf8)
2557          {          {
2558          register int d;          register unsigned int d;
2559          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2560            {            {
2561            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
# Line 2185  for (;;) Line 2580  for (;;)
2580          /* UTF-8 mode */          /* UTF-8 mode */
2581          if (utf8)          if (utf8)
2582            {            {
2583            register int d;            register unsigned int d;
2584            for (fi = min;; fi++)            for (fi = min;; fi++)
2585              {              {
2586              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28);
2587              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2588              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
2589              if (d < 256) d = md->lcc[d];              if (d < 256) d = md->lcc[d];
# Line 2202  for (;;) Line 2597  for (;;)
2597            {            {
2598            for (fi = min;; fi++)            for (fi = min;; fi++)
2599              {              {
2600              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29);
2601              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2602              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])              if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++])
2603                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
# Line 2221  for (;;) Line 2616  for (;;)
2616          /* UTF-8 mode */          /* UTF-8 mode */
2617          if (utf8)          if (utf8)
2618            {            {
2619            register int d;            register unsigned int d;
2620            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2621              {              {
2622              int len = 1;              int len = 1;
# Line 2231  for (;;) Line 2626  for (;;)
2626              if (fc == d) break;              if (fc == d) break;
2627              eptr += len;              eptr += len;
2628              }              }
2629            for(;;)          if (possessive) continue;
2630            for(;;)
2631              {              {
2632              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM30);
2633              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2634              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2635              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2248  for (;;) Line 2644  for (;;)
2644              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;              if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break;
2645              eptr++;              eptr++;
2646              }              }
2647              if (possessive) continue;
2648            while (eptr >= pp)            while (eptr >= pp)
2649              {              {
2650              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM31);
2651              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2652              eptr--;              eptr--;
2653              }              }
# Line 2269  for (;;) Line 2666  for (;;)
2666        /* UTF-8 mode */        /* UTF-8 mode */
2667        if (utf8)        if (utf8)
2668          {          {
2669          register int d;          register unsigned int d;
2670          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2671            {            {
2672            GETCHARINC(d, eptr);            GETCHARINC(d, eptr);
# Line 2292  for (;;) Line 2689  for (;;)
2689          /* UTF-8 mode */          /* UTF-8 mode */
2690          if (utf8)          if (utf8)
2691            {            {
2692            register int d;            register unsigned int d;
2693            for (fi = min;; fi++)            for (fi = min;; fi++)
2694              {              {
2695              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32);
2696              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2697              GETCHARINC(d, eptr);              GETCHARINC(d, eptr);
2698              if (fi >= max || eptr >= md->end_subject || fc == d)              if (fi >= max || eptr >= md->end_subject || fc == d)
# Line 2308  for (;;) Line 2705  for (;;)
2705            {            {
2706            for (fi = min;; fi++)            for (fi = min;; fi++)
2707              {              {
2708              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33);
2709              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2710              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)              if (fi >= max || eptr >= md->end_subject || fc == *eptr++)
2711                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
# Line 2327  for (;;) Line 2724  for (;;)
2724          /* UTF-8 mode */          /* UTF-8 mode */
2725          if (utf8)          if (utf8)
2726            {            {
2727            register int d;            register unsigned int d;
2728            for (i = min; i < max; i++)            for (i = min; i < max; i++)
2729              {              {
2730              int len = 1;              int len = 1;
# Line 2336  for (;;) Line 2733  for (;;)
2733              if (fc == d) break;              if (fc == d) break;
2734              eptr += len;              eptr += len;
2735              }              }
2736              if (possessive) continue;
2737            for(;;)            for(;;)
2738              {              {
2739              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM34);
2740              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2741              if (eptr-- == pp) break;        /* Stop if tried at original pos */              if (eptr-- == pp) break;        /* Stop if tried at original pos */
2742              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 2353  for (;;) Line 2751  for (;;)
2751              if (eptr >= md->end_subject || fc == *eptr) break;              if (eptr >= md->end_subject || fc == *eptr) break;
2752              eptr++;              eptr++;
2753              }              }
2754              if (possessive) continue;
2755            while (eptr >= pp)            while (eptr >= pp)
2756              {              {
2757              RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM35);
2758              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2759              eptr--;              eptr--;
2760              }              }
# Line 2384  for (;;) Line 2783  for (;;)
2783      ecode += 3;      ecode += 3;
2784      goto REPEATTYPE;      goto REPEATTYPE;
2785    
2786        case OP_TYPEPOSSTAR:
2787        possessive = TRUE;
2788        min = 0;
2789        max = INT_MAX;
2790        ecode++;
2791        goto REPEATTYPE;
2792    
2793        case OP_TYPEPOSPLUS:
2794        possessive = TRUE;
2795        min = 1;
2796        max = INT_MAX;
2797        ecode++;
2798        goto REPEATTYPE;
2799    
2800        case OP_TYPEPOSQUERY:
2801        possessive = TRUE;
2802        min = 0;
2803        max = 1;
2804        ecode++;
2805        goto REPEATTYPE;
2806    
2807        case OP_TYPEPOSUPTO:
2808        possessive = TRUE;
2809        min = 0;
2810        max = GET2(ecode, 1);
2811        ecode += 3;
2812        goto REPEATTYPE;
2813    
2814      case OP_TYPESTAR:      case OP_TYPESTAR:
2815      case OP_TYPEMINSTAR:      case OP_TYPEMINSTAR:
2816      case OP_TYPEPLUS:      case OP_TYPEPLUS:
# Line 2408  for (;;) Line 2835  for (;;)
2835        {        {
2836        prop_fail_result = ctype == OP_NOTPROP;        prop_fail_result = ctype == OP_NOTPROP;
2837        prop_type = *ecode++;        prop_type = *ecode++;
2838        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;  
         }  
2839        }        }
2840      else prop_type = -1;      else prop_type = -1;
2841  #endif  #endif
# Line 2434  for (;;) Line 2852  for (;;)
2852      if (min > 0)      if (min > 0)
2853        {        {
2854  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2855        if (prop_type > 0)        if (prop_type >= 0)
2856          {          {
2857          for (i = 1; i <= min; i++)          switch(prop_type)
2858            {            {
2859            GETCHARINC(c, eptr);            case PT_ANY:
2860            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            if (prop_fail_result) RRETURN(MATCH_NOMATCH);
2861            if ((*prop_test_variable == prop_test_against) == prop_fail_result)            for (i = 1; i <= min; i++)
2862              RRETURN(MATCH_NOMATCH);              {
2863                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2864                GETCHARINCTEST(c, eptr);
2865                }
2866              break;
2867    
2868              case PT_LAMP:
2869              for (i = 1; i <= min; i++)
2870                {
2871                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2872                GETCHARINCTEST(c, eptr);
2873                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
2874                if ((prop_chartype == ucp_Lu ||
2875                     prop_chartype == ucp_Ll ||
2876                     prop_chartype == ucp_Lt) == prop_fail_result)
2877                  RRETURN(MATCH_NOMATCH);
2878                }
2879              break;
2880    
2881              case PT_GC:
2882              for (i = 1; i <= min; i++)
2883                {
2884                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2885                GETCHARINCTEST(c, eptr);
2886                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
2887                if ((prop_category == prop_value) == prop_fail_result)
2888                  RRETURN(MATCH_NOMATCH);
2889                }
2890              break;
2891    
2892              case PT_PC:
2893              for (i = 1; i <= min; i++)
2894                {
2895                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2896                GETCHARINCTEST(c, eptr);
2897                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
2898                if ((prop_chartype == prop_value) == prop_fail_result)
2899                  RRETURN(MATCH_NOMATCH);
2900                }
2901              break;
2902    
2903              case PT_SC:
2904              for (i = 1; i <= min; i++)
2905                {
2906                if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2907                GETCHARINCTEST(c, eptr);
2908                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
2909                if ((prop_script == prop_value) == prop_fail_result)
2910                  RRETURN(MATCH_NOMATCH);
2911                }
2912              break;
2913    
2914              default:
2915              RRETURN(PCRE_ERROR_INTERNAL);
2916            }            }
2917          }          }
2918    
# Line 2453  for (;;) Line 2924  for (;;)
2924          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2925            {            {
2926            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
2927            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
2928            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);
2929            while (eptr < md->end_subject)            while (eptr < md->end_subject)
2930              {              {
# Line 2462  for (;;) Line 2933  for (;;)
2933                {                {
2934                GETCHARLEN(c, eptr, len);                GETCHARLEN(c, eptr, len);
2935                }                }
2936              prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
2937              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
2938              eptr += len;              eptr += len;
2939              }              }
# Line 2480  for (;;) Line 2951  for (;;)
2951          case OP_ANY:          case OP_ANY:
2952          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
2953            {            {
2954            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject || IS_NEWLINE(eptr))
              (*eptr++ == NEWLINE && (ims & PCRE_DOTALL) == 0))  
2955              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
2956              eptr++;
2957              while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2958              }
2959            break;
2960    
2961            case OP_ALLANY:
2962            for (i = 1; i <= min; i++)
2963              {
2964              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2965              eptr++;
2966            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
2967            }            }
2968          break;          break;
# Line 2491  for (;;) Line 2971  for (;;)
2971          eptr += min;          eptr += min;
2972          break;          break;
2973    
2974            case OP_ANYNL:
2975            for (i = 1; i <= min; i++)
2976              {
2977              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
2978              GETCHARINC(c, eptr);
2979              switch(c)
2980                {
2981                default: RRETURN(MATCH_NOMATCH);
2982                case 0x000d:
2983                if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
2984                break;
2985    
2986                case 0x000a:
2987                break;
2988    
2989                case 0x000b:
2990                case 0x000c:
2991                case 0x0085:
2992                case 0x2028:
2993                case 0x2029:
2994                if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
2995                break;
2996                }
2997              }
2998            break;
2999    
3000            case OP_NOT_HSPACE:
3001            for (i = 1; i <= min; i++)
3002              {
3003              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3004              GETCHARINC(c, eptr);
3005              switch(c)
3006                {
3007                default: break;
3008                case 0x09:      /* HT */
3009                case 0x20:      /* SPACE */
3010                case 0xa0:      /* NBSP */
3011                case 0x1680:    /* OGHAM SPACE MARK */
3012                case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3013                case 0x2000:    /* EN QUAD */
3014                case 0x2001:    /* EM QUAD */
3015                case 0x2002:    /* EN SPACE */
3016                case 0x2003:    /* EM SPACE */
3017                case 0x2004:    /* THREE-PER-EM SPACE */
3018                case 0x2005:    /* FOUR-PER-EM SPACE */
3019                case 0x2006:    /* SIX-PER-EM SPACE */
3020                case 0x2007:    /* FIGURE SPACE */
3021                case 0x2008:    /* PUNCTUATION SPACE */
3022                case 0x2009:    /* THIN SPACE */
3023                case 0x200A:    /* HAIR SPACE */
3024                case 0x202f:    /* NARROW NO-BREAK SPACE */
3025                case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3026                case 0x3000:    /* IDEOGRAPHIC SPACE */
3027                RRETURN(MATCH_NOMATCH);
3028                }
3029              }
3030            break;
3031    
3032            case OP_HSPACE:
3033            for (i = 1; i <= min; i++)
3034              {
3035              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3036              GETCHARINC(c, eptr);
3037              switch(c)
3038                {
3039                default: RRETURN(MATCH_NOMATCH);
3040                case 0x09:      /* HT */
3041                case 0x20:      /* SPACE */
3042                case 0xa0:      /* NBSP */
3043                case 0x1680:    /* OGHAM SPACE MARK */
3044                case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3045                case 0x2000:    /* EN QUAD */
3046                case 0x2001:    /* EM QUAD */
3047                case 0x2002:    /* EN SPACE */
3048                case 0x2003:    /* EM SPACE */
3049                case 0x2004:    /* THREE-PER-EM SPACE */
3050                case 0x2005:    /* FOUR-PER-EM SPACE */
3051                case 0x2006:    /* SIX-PER-EM SPACE */
3052                case 0x2007:    /* FIGURE SPACE */
3053                case 0x2008:    /* PUNCTUATION SPACE */
3054                case 0x2009:    /* THIN SPACE */
3055                case 0x200A:    /* HAIR SPACE */
3056                case 0x202f:    /* NARROW NO-BREAK SPACE */
3057                case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3058                case 0x3000:    /* IDEOGRAPHIC SPACE */
3059                break;
3060                }
3061              }
3062            break;
3063    
3064            case OP_NOT_VSPACE:
3065            for (i = 1; i <= min; i++)
3066              {
3067              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3068              GETCHARINC(c, eptr);
3069              switch(c)
3070                {
3071                default: break;
3072                case 0x0a:      /* LF */
3073                case 0x0b:      /* VT */
3074                case 0x0c:      /* FF */
3075                case 0x0d:      /* CR */
3076                case 0x85:      /* NEL */
3077                case 0x2028:    /* LINE SEPARATOR */
3078                case 0x2029:    /* PARAGRAPH SEPARATOR */
3079                RRETURN(MATCH_NOMATCH);
3080                }
3081              }
3082            break;
3083    
3084            case OP_VSPACE:
3085            for (i = 1; i <= min; i++)
3086              {
3087              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3088              GETCHARINC(c, eptr);
3089              switch(c)
3090                {
3091                default: RRETURN(MATCH_NOMATCH);
3092                case 0x0a:      /* LF */
3093                case 0x0b:      /* VT */
3094                case 0x0c:      /* FF */
3095                case 0x0d:      /* CR */
3096                case 0x85:      /* NEL */
3097                case 0x2028:    /* LINE SEPARATOR */
3098                case 0x2029:    /* PARAGRAPH SEPARATOR */
3099                break;
3100                }
3101              }
3102            break;
3103    
3104          case OP_NOT_DIGIT:          case OP_NOT_DIGIT:
3105          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3106            {            {
# Line 2515  for (;;) Line 3125  for (;;)
3125          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3126            {            {
3127            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject ||
3128               (*eptr < 128 && (md->ctypes[*eptr++] & ctype_space) != 0))               (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0))
3129              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3130            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
3131            }            }
3132          break;          break;
3133    
# Line 2535  for (;;) Line 3145  for (;;)
3145          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3146            {            {
3147            if (eptr >= md->end_subject ||            if (eptr >= md->end_subject ||
3148               (*eptr < 128 && (md->ctypes[*eptr++] & ctype_word) != 0))               (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0))
3149              RRETURN(MATCH_NOMATCH);              RRETURN(MATCH_NOMATCH);
3150            while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;            while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80);
3151            }            }
3152          break;          break;
3153    
# Line 2559  for (;;) Line 3169  for (;;)
3169  #endif     /* SUPPORT_UTF8 */  #endif     /* SUPPORT_UTF8 */
3170    
3171        /* 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
3172        than OP_PROP and OP_NOTPROP. */        than OP_PROP and OP_NOTPROP. We can assume that there are the minimum
3173          number of bytes present, as this was tested above. */
3174    
3175        switch(ctype)        switch(ctype)
3176          {          {
3177          case OP_ANY:          case OP_ANY:
3178          if ((ims & PCRE_DOTALL) == 0)          for (i = 1; i <= min; i++)
3179            {            {
3180            for (i = 1; i <= min; i++)            if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
3181              if (*eptr++ == NEWLINE) RRETURN(MATCH_NOMATCH);            eptr++;
3182            }            }
3183          else eptr += min;          break;
3184    
3185            case OP_ALLANY:
3186            eptr += min;
3187          break;          break;
3188    
3189          case OP_ANYBYTE:          case OP_ANYBYTE:
3190          eptr += min;          eptr += min;
3191          break;          break;
3192    
3193            /* Because of the CRLF case, we can't assume the minimum number of
3194            bytes are present in this case. */
3195    
3196            case OP_ANYNL:
3197            for (i = 1; i <= min; i++)
3198              {
3199              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3200              switch(*eptr++)
3201                {
3202                default: RRETURN(MATCH_NOMATCH);
3203                case 0x000d:
3204                if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3205                break;
3206                case 0x000a:
3207                break;
3208    
3209                case 0x000b:
3210                case 0x000c:
3211                case 0x0085:
3212                if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
3213                break;
3214                }
3215              }
3216            break;
3217    
3218            case OP_NOT_HSPACE:
3219            for (i = 1; i <= min; i++)
3220              {
3221              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3222              switch(*eptr++)
3223                {
3224                default: break;
3225                case 0x09:      /* HT */
3226                case 0x20:      /* SPACE */
3227                case 0xa0:      /* NBSP */
3228                RRETURN(MATCH_NOMATCH);
3229                }
3230              }
3231            break;
3232    
3233            case OP_HSPACE:
3234            for (i = 1; i <= min; i++)
3235              {
3236              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3237              switch(*eptr++)
3238                {
3239                default: RRETURN(MATCH_NOMATCH);
3240                case 0x09:      /* HT */
3241                case 0x20:      /* SPACE */
3242                case 0xa0:      /* NBSP */
3243                break;
3244                }
3245              }
3246            break;
3247    
3248            case OP_NOT_VSPACE:
3249            for (i = 1; i <= min; i++)
3250              {
3251              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3252              switch(*eptr++)
3253                {
3254                default: break;
3255                case 0x0a:      /* LF */
3256                case 0x0b:      /* VT */
3257                case 0x0c:      /* FF */
3258                case 0x0d:      /* CR */
3259                case 0x85:      /* NEL */
3260                RRETURN(MATCH_NOMATCH);
3261                }
3262              }
3263            break;
3264    
3265            case OP_VSPACE:
3266            for (i = 1; i <= min; i++)
3267              {
3268              if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3269              switch(*eptr++)
3270                {
3271                default: RRETURN(MATCH_NOMATCH);
3272                case 0x0a:      /* LF */
3273                case 0x0b:      /* VT */
3274                case 0x0c:      /* FF */
3275                case 0x0d:      /* CR */
3276                case 0x85:      /* NEL */
3277                break;
3278                }
3279              }
3280            break;
3281    
3282          case OP_NOT_DIGIT:          case OP_NOT_DIGIT:
3283          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
3284            if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH);            if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH);
# Line 2624  for (;;) Line 3327  for (;;)
3327      if (minimize)      if (minimize)
3328        {        {
3329  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3330        if (prop_type > 0)        if (prop_type >= 0)
3331          {          {
3332          for (fi = min;; fi++)          switch(prop_type)
3333            {            {
3334            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            case PT_ANY:
3335            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            for (fi = min;; fi++)
3336            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);              {
3337            GETCHARINC(c, eptr);              RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36);
3338            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3339            if ((*prop_test_variable == prop_test_against) == prop_fail_result)              if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3340              RRETURN(MATCH_NOMATCH);              GETCHARINC(c, eptr);
3341            }              if (prop_fail_result) RRETURN(MATCH_NOMATCH);
3342          }              }
3343              /* Control never gets here */
3344    
3345        /* Match extended Unicode sequences. We will get here only if the            case PT_LAMP:
3346        support is in the binary; otherwise a compile-time error occurs. */            for (fi = min;; fi++)
3347                {
3348                RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37);
3349                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3350                if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3351                GETCHARINC(c, eptr);
3352                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3353                if ((prop_chartype == ucp_Lu ||
3354                     prop_chartype == ucp_Ll ||
3355                     prop_chartype == ucp_Lt) == prop_fail_result)
3356                  RRETURN(MATCH_NOMATCH);
3357                }
3358              /* Control never gets here */
3359    
3360        else if (ctype == OP_EXTUNI)            case PT_GC:
3361          {            for (fi = min;; fi++)
3362                {
3363                RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38);
3364                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3365                if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3366                GETCHARINC(c, eptr);
3367                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3368                if ((prop_category == prop_value) == prop_fail_result)
3369                  RRETURN(MATCH_NOMATCH);
3370                }
3371              /* Control never gets here */
3372    
3373              case PT_PC:
3374              for (fi = min;; fi++)
3375                {
3376                RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39);
3377                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3378                if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3379                GETCHARINC(c, eptr);
3380                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3381                if ((prop_chartype == prop_value) == prop_fail_result)
3382                  RRETURN(MATCH_NOMATCH);
3383                }
3384              /* Control never gets here */
3385    
3386              case PT_SC:
3387              for (fi = min;; fi++)
3388                {
3389                RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40);
3390                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3391                if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3392                GETCHARINC(c, eptr);
3393                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3394                if ((prop_script == prop_value) == prop_fail_result)
3395                  RRETURN(MATCH_NOMATCH);
3396                }
3397              /* Control never gets here */
3398    
3399              default:
3400              RRETURN(PCRE_ERROR_INTERNAL);
3401              }
3402            }
3403    
3404          /* Match extended Unicode sequences. We will get here only if the
3405          support is in the binary; otherwise a compile-time error occurs. */
3406    
3407          else if (ctype == OP_EXTUNI)
3408            {
3409          for (fi = min;; fi++)          for (fi = min;; fi++)
3410            {            {
3411            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41);
3412            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3413            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);
3414            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3415            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3416            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);            if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH);
3417            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3418              {              {
# Line 2658  for (;;) Line 3421  for (;;)
3421                {                {
3422                GETCHARLEN(c, eptr, len);                GETCHARLEN(c, eptr, len);
3423                }                }
3424              prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3425              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
3426              eptr += len;              eptr += len;
3427              }              }
# Line 2674  for (;;) Line 3437  for (;;)
3437          {          {
3438          for (fi = min;; fi++)          for (fi = min;; fi++)
3439            {            {
3440            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42);
3441            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3442            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max || eptr >= md->end_subject ||
3443                   (ctype == OP_ANY && IS_NEWLINE(eptr)))
3444                RRETURN(MATCH_NOMATCH);
3445    
3446            GETCHARINC(c, eptr);            GETCHARINC(c, eptr);
3447            switch(ctype)            switch(ctype)
3448              {              {
3449              case OP_ANY:              case OP_ANY:        /* This is the non-NL case */
3450              if ((ims & PCRE_DOTALL) == 0 && c == NEWLINE) RRETURN(MATCH_NOMATCH);              case OP_ALLANY:
3451                case OP_ANYBYTE:
3452              break;              break;
3453    
3454              case OP_ANYBYTE:              case OP_ANYNL:
3455                switch(c)
3456                  {
3457                  default: RRETURN(MATCH_NOMATCH);
3458                  case 0x000d:
3459                  if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3460                  break;
3461                  case 0x000a:
3462                  break;
3463    
3464                  case 0x000b:
3465                  case 0x000c:
3466                  case 0x0085:
3467                  case 0x2028:
3468                  case 0x2029:
3469                  if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
3470                  break;
3471                  }
3472                break;
3473    
3474                case OP_NOT_HSPACE:
3475                switch(c)
3476                  {
3477                  default: break;
3478                  case 0x09:      /* HT */
3479                  case 0x20:      /* SPACE */
3480                  case 0xa0:      /* NBSP */
3481                  case 0x1680:    /* OGHAM SPACE MARK */
3482                  case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3483                  case 0x2000:    /* EN QUAD */
3484                  case 0x2001:    /* EM QUAD */
3485                  case 0x2002:    /* EN SPACE */
3486                  case 0x2003:    /* EM SPACE */
3487                  case 0x2004:    /* THREE-PER-EM SPACE */
3488                  case 0x2005:    /* FOUR-PER-EM SPACE */
3489                  case 0x2006:    /* SIX-PER-EM SPACE */
3490                  case 0x2007:    /* FIGURE SPACE */
3491                  case 0x2008:    /* PUNCTUATION SPACE */
3492                  case 0x2009:    /* THIN SPACE */
3493                  case 0x200A:    /* HAIR SPACE */
3494                  case 0x202f:    /* NARROW NO-BREAK SPACE */
3495                  case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3496                  case 0x3000:    /* IDEOGRAPHIC SPACE */
3497                  RRETURN(MATCH_NOMATCH);
3498                  }
3499                break;
3500    
3501                case OP_HSPACE:
3502                switch(c)
3503                  {
3504                  default: RRETURN(MATCH_NOMATCH);
3505                  case 0x09:      /* HT */
3506                  case 0x20:      /* SPACE */
3507                  case 0xa0:      /* NBSP */
3508                  case 0x1680:    /* OGHAM SPACE MARK */
3509                  case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3510                  case 0x2000:    /* EN QUAD */
3511                  case 0x2001:    /* EM QUAD */
3512                  case 0x2002:    /* EN SPACE */
3513                  case 0x2003:    /* EM SPACE */
3514                  case 0x2004:    /* THREE-PER-EM SPACE */
3515                  case 0x2005:    /* FOUR-PER-EM SPACE */
3516                  case 0x2006:    /* SIX-PER-EM SPACE */
3517                  case 0x2007:    /* FIGURE SPACE */
3518                  case 0x2008:    /* PUNCTUATION SPACE */
3519                  case 0x2009:    /* THIN SPACE */
3520                  case 0x200A:    /* HAIR SPACE */
3521                  case 0x202f:    /* NARROW NO-BREAK SPACE */
3522                  case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3523                  case 0x3000:    /* IDEOGRAPHIC SPACE */
3524                  break;
3525                  }
3526                break;
3527    
3528                case OP_NOT_VSPACE:
3529                switch(c)
3530                  {
3531                  default: break;
3532                  case 0x0a:      /* LF */
3533                  case 0x0b:      /* VT */
3534                  case 0x0c:      /* FF */
3535                  case 0x0d:      /* CR */
3536                  case 0x85:      /* NEL */
3537                  case 0x2028:    /* LINE SEPARATOR */
3538                  case 0x2029:    /* PARAGRAPH SEPARATOR */
3539                  RRETURN(MATCH_NOMATCH);
3540                  }
3541                break;
3542    
3543                case OP_VSPACE:
3544                switch(c)
3545                  {
3546                  default: RRETURN(MATCH_NOMATCH);
3547                  case 0x0a:      /* LF */
3548                  case 0x0b:      /* VT */
3549                  case 0x0c:      /* FF */
3550                  case 0x0d:      /* CR */
3551                  case 0x85:      /* NEL */
3552                  case 0x2028:    /* LINE SEPARATOR */
3553                  case 0x2029:    /* PARAGRAPH SEPARATOR */
3554                  break;
3555                  }
3556              break;              break;
3557    
3558              case OP_NOT_DIGIT:              case OP_NOT_DIGIT:
# Line 2729  for (;;) Line 3596  for (;;)
3596          {          {
3597          for (fi = min;; fi++)          for (fi = min;; fi++)
3598            {            {
3599            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43);
3600            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3601            if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH);            if (fi >= max || eptr >= md->end_subject ||
3602                   (ctype == OP_ANY && IS_NEWLINE(eptr)))
3603                RRETURN(MATCH_NOMATCH);
3604    
3605            c = *eptr++;            c = *eptr++;
3606            switch(ctype)            switch(ctype)
3607              {              {
3608              case OP_ANY:              case OP_ANY:     /* This is the non-NL case */
3609              if ((ims & PCRE_DOTALL) == 0 && c == NEWLINE) RRETURN(MATCH_NOMATCH);              case OP_ALLANY:
3610                case OP_ANYBYTE:
3611              break;              break;
3612    
3613              case OP_ANYBYTE:              case OP_ANYNL:
3614                switch(c)
3615                  {
3616                  default: RRETURN(MATCH_NOMATCH);
3617                  case 0x000d:
3618                  if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
3619                  break;
3620    
3621                  case 0x000a:
3622                  break;
3623    
3624                  case 0x000b:
3625                  case 0x000c:
3626                  case 0x0085:
3627                  if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
3628                  break;
3629                  }
3630                break;
3631    
3632                case OP_NOT_HSPACE:
3633                switch(c)
3634                  {
3635                  default: break;
3636                  case 0x09:      /* HT */
3637                  case 0x20:      /* SPACE */
3638                  case 0xa0:      /* NBSP */
3639                  RRETURN(MATCH_NOMATCH);
3640                  }
3641                break;
3642    
3643                case OP_HSPACE:
3644                switch(c)
3645                  {
3646                  default: RRETURN(MATCH_NOMATCH);
3647                  case 0x09:      /* HT */
3648                  case 0x20:      /* SPACE */
3649                  case 0xa0:      /* NBSP */
3650                  break;
3651                  }
3652                break;
3653    
3654                case OP_NOT_VSPACE:
3655                switch(c)
3656                  {
3657                  default: break;
3658                  case 0x0a:      /* LF */
3659                  case 0x0b:      /* VT */
3660                  case 0x0c:      /* FF */
3661                  case 0x0d:      /* CR */
3662                  case 0x85:      /* NEL */
3663                  RRETURN(MATCH_NOMATCH);
3664                  }
3665                break;
3666    
3667                case OP_VSPACE:
3668                switch(c)
3669                  {
3670                  default: RRETURN(MATCH_NOMATCH);
3671                  case 0x0a:      /* LF */
3672                  case 0x0b:      /* VT */
3673                  case 0x0c:      /* FF */
3674                  case 0x0d:      /* CR */
3675                  case 0x85:      /* NEL */
3676                  break;
3677                  }
3678              break;              break;
3679    
3680              case OP_NOT_DIGIT:              case OP_NOT_DIGIT:
# Line 2774  for (;;) Line 3709  for (;;)
3709        /* Control never gets here */        /* Control never gets here */
3710        }        }
3711    
3712      /* 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
3713      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
3714      UTF-8 and UCP stuff separate. */      UTF-8 and UCP stuff separate. */
3715    
# Line 2783  for (;;) Line 3718  for (;;)
3718        pp = eptr;  /* Remember where we started */        pp = eptr;  /* Remember where we started */
3719    
3720  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3721        if (prop_type > 0)        if (prop_type >= 0)
3722          {          {
3723          for (i = min; i < max; i++)          switch(prop_type)
3724            {            {
3725            int len = 1;            case PT_ANY:
3726            if (eptr >= md->end_subject) break;            for (i = min; i < max; i++)
3727            GETCHARLEN(c, eptr, len);              {
3728            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              int len = 1;
3729            if ((*prop_test_variable == prop_test_against) == prop_fail_result)              if (eptr >= md->end_subject) break;
3730              break;              GETCHARLEN(c, eptr, len);
3731            eptr+= len;              if (prop_fail_result) break;
3732                eptr+= len;
3733                }
3734              break;
3735    
3736              case PT_LAMP:
3737              for (i = min; i < max; i++)
3738                {
3739                int len = 1;
3740                if (eptr >= md->end_subject) break;
3741                GETCHARLEN(c, eptr, len);
3742                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3743                if ((prop_chartype == ucp_Lu ||
3744                     prop_chartype == ucp_Ll ||
3745                     prop_chartype == ucp_Lt) == prop_fail_result)
3746                  break;
3747                eptr+= len;
3748                }
3749              break;
3750    
3751              case PT_GC:
3752              for (i = min; i < max; i++)
3753                {
3754                int len = 1;
3755                if (eptr >= md->end_subject) break;
3756                GETCHARLEN(c, eptr, len);
3757                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3758                if ((prop_category == prop_value) == prop_fail_result)
3759                  break;
3760                eptr+= len;
3761                }
3762              break;
3763    
3764              case PT_PC:
3765              for (i = min; i < max; i++)
3766                {
3767                int len = 1;
3768                if (eptr >= md->end_subject) break;
3769                GETCHARLEN(c, eptr, len);
3770                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3771                if ((prop_chartype == prop_value) == prop_fail_result)
3772                  break;
3773                eptr+= len;
3774                }
3775              break;
3776    
3777              case PT_SC:
3778              for (i = min; i < max; i++)
3779                {
3780                int len = 1;
3781                if (eptr >= md->end_subject) break;
3782                GETCHARLEN(c, eptr, len);
3783                prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3784                if ((prop_script == prop_value) == prop_fail_result)
3785                  break;
3786                eptr+= len;
3787                }
3788              break;
3789            }            }
3790    
3791          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
3792    
3793            if (possessive) continue;
3794          for(;;)          for(;;)
3795            {            {
3796            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44);
3797            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3798            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3799            BACKCHAR(eptr);            if (utf8) BACKCHAR(eptr);
3800            }            }
3801          }          }
3802    
# Line 2816  for (;;) Line 3809  for (;;)
3809            {            {
3810            if (eptr >= md->end_subject) break;            if (eptr >= md->end_subject) break;
3811            GETCHARINCTEST(c, eptr);            GETCHARINCTEST(c, eptr);
3812            prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);            prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3813            if (prop_category == ucp_M) break;            if (prop_category == ucp_M) break;
3814            while (eptr < md->end_subject)            while (eptr < md->end_subject)
3815              {              {
# Line 2825  for (;;) Line 3818  for (;;)
3818                {                {
3819                GETCHARLEN(c, eptr, len);                GETCHARLEN(c, eptr, len);
3820                }                }
3821              prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3822              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
3823              eptr += len;              eptr += len;
3824              }              }
# Line 2833  for (;;) Line 3826  for (;;)
3826    
3827          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
3828    
3829            if (possessive) continue;
3830          for(;;)          for(;;)
3831            {            {
3832            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45);
3833            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3834            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
3835            for (;;)                        /* Move back over one extended */            for (;;)                        /* Move back over one extended */
3836              {              {
3837              int len = 1;              int len = 1;
             BACKCHAR(eptr);  
3838              if (!utf8) c = *eptr; else              if (!utf8) c = *eptr; else
3839                {                {
3840                  BACKCHAR(eptr);
3841                GETCHARLEN(c, eptr, len);                GETCHARLEN(c, eptr, len);
3842                }                }
3843              prop_category = _pcre_ucp_findchar(c, &prop_chartype, &prop_othercase);              prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script);
3844              if (prop_category != ucp_M) break;              if (prop_category != ucp_M) break;
3845              eptr--;              eptr--;
3846              }              }
# Line 2864  for (;;) Line 3858  for (;;)
3858          switch(ctype)          switch(ctype)
3859            {            {
3860            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. */  
   
3861            if (max < INT_MAX)            if (max < INT_MAX)
3862              {              {
3863              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  
3864                {                {
3865                for (i = min; i < max; i++)                if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
3866                  {                eptr++;
3867                  eptr++;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                 while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;  
                 }  
3868                }                }
3869              }              }
3870    
# Line 2894  for (;;) Line 3872  for (;;)
3872    
3873            else            else
3874              {              {
3875              if ((ims & PCRE_DOTALL) == 0)              for (i = min; i < max; i++)
3876                {                {
3877                for (i = min; i < max; i++)                if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
3878                  {                eptr++;
3879                  if (eptr >= md->end_subject || *eptr == NEWLINE) break;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
                 eptr++;  
                 }  
               break;  
3880                }                }
3881              else              }
3882              break;
3883    
3884              case OP_ALLANY:
3885              if (max < INT_MAX)
3886                {
3887                for (i = min; i < max; i++)
3888                {                {
3889                c = max - min;                if (eptr >= md->end_subject) break;
3890                if (c > md->end_subject - eptr) c = md->end_subject - eptr;                eptr++;
3891                eptr += c;                while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
3892                }                }
3893              }              }
3894              else eptr = md->end_subject;   /* Unlimited UTF-8 repeat */
3895            break;            break;
3896    
3897            /* The byte case is the same as non-UTF8 */            /* The byte case is the same as non-UTF8 */
3898    
3899            case OP_ANYBYTE:            case OP_ANYBYTE:
3900            c = max - min;            c = max - min;
3901            if (c > md->end_subject - eptr) c = md->end_subject - eptr;            if (c > (unsigned int)(md->end_subject - eptr))
3902                c = md->end_subject - eptr;
3903            eptr += c;            eptr += c;
3904            break;            break;
3905    
3906              case OP_ANYNL:
3907              for (i = min; i < max; i++)
3908                {
3909                int len = 1;
3910                if (eptr >= md->end_subject) break;
3911                GETCHARLEN(c, eptr, len);
3912                if (c == 0x000d)
3913                  {
3914                  if (++eptr >= md->end_subject) break;
3915                  if (*eptr == 0x000a) eptr++;
3916                  }
3917                else
3918                  {
3919                  if (c != 0x000a &&
3920                      (md->bsr_anycrlf ||
3921                       (c != 0x000b && c != 0x000c &&
3922                        c != 0x0085 && c != 0x2028 && c != 0x2029)))
3923                    break;
3924                  eptr += len;
3925                  }
3926                }
3927              break;
3928    
3929              case OP_NOT_HSPACE:
3930              case OP_HSPACE:
3931              for (i = min; i < max; i++)
3932                {
3933                BOOL gotspace;
3934                int len = 1;
3935                if (eptr >= md->end_subject) break;
3936                GETCHARLEN(c, eptr, len);
3937                switch(c)
3938                  {
3939                  default: gotspace = FALSE; break;
3940                  case 0x09:      /* HT */
3941                  case 0x20:      /* SPACE */
3942                  case 0xa0:      /* NBSP */
3943                  case 0x1680:    /* OGHAM SPACE MARK */
3944                  case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
3945                  case 0x2000:    /* EN QUAD */
3946                  case 0x2001:    /* EM QUAD */
3947                  case 0x2002:    /* EN SPACE */
3948                  case 0x2003:    /* EM SPACE */
3949                  case 0x2004:    /* THREE-PER-EM SPACE */
3950                  case 0x2005:    /* FOUR-PER-EM SPACE */
3951                  case 0x2006:    /* SIX-PER-EM SPACE */
3952                  case 0x2007:    /* FIGURE SPACE */
3953                  case 0x2008:    /* PUNCTUATION SPACE */
3954                  case 0x2009:    /* THIN SPACE */
3955                  case 0x200A:    /* HAIR SPACE */
3956                  case 0x202f:    /* NARROW NO-BREAK SPACE */
3957                  case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
3958                  case 0x3000:    /* IDEOGRAPHIC SPACE */
3959                  gotspace = TRUE;
3960                  break;
3961                  }
3962                if (gotspace == (ctype == OP_NOT_HSPACE)) break;
3963                eptr += len;
3964                }
3965              break;
3966    
3967              case OP_NOT_VSPACE:
3968              case OP_VSPACE:
3969              for (i = min; i < max; i++)
3970                {
3971                BOOL gotspace;
3972                int len = 1;
3973                if (eptr >= md->end_subject) break;
3974                GETCHARLEN(c, eptr, len);
3975                switch(c)
3976                  {
3977                  default: gotspace = FALSE; break;
3978                  case 0x0a:      /* LF */
3979                  case 0x0b:      /* VT */
3980                  case 0x0c:      /* FF */
3981                  case 0x0d:      /* CR */
3982                  case 0x85:      /* NEL */
3983                  case 0x2028:    /* LINE SEPARATOR */
3984                  case 0x2029:    /* PARAGRAPH SEPARATOR */
3985                  gotspace = TRUE;
3986                  break;
3987                  }
3988                if (gotspace == (ctype == OP_NOT_VSPACE)) break;
3989                eptr += len;
3990                }
3991              break;
3992    
3993            case OP_NOT_DIGIT:            case OP_NOT_DIGIT:
3994            for (i = min; i < max; i++)            for (i = min; i < max; i++)
3995              {              {
# Line 2992  for (;;) Line 4062  for (;;)
4062    
4063          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
4064    
4065            if (possessive) continue;
4066          for(;;)          for(;;)
4067            {            {
4068            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM46);
4069            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4070            if (eptr-- == pp) break;        /* Stop if tried at original pos */            if (eptr-- == pp) break;        /* Stop if tried at original pos */
4071            BACKCHAR(eptr);            BACKCHAR(eptr);
4072            }            }
4073          }          }
4074        else        else
4075  #endif  #endif  /* SUPPORT_UTF8 */
4076    
4077        /* Not UTF-8 mode */        /* Not UTF-8 mode */
4078          {          {
4079          switch(ctype)          switch(ctype)
4080            {            {
4081            case OP_ANY:            case OP_ANY:
4082            if ((ims & PCRE_DOTALL) == 0)            for (i = min; i < max; i++)
4083              {              {
4084              for (i = min; i < max; i++)              if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break;
4085                {              eptr++;
               if (eptr >= md->end_subject || *eptr == NEWLINE) break;  
               eptr++;  
               }  
             break;  
4086              }              }
4087            /* For DOTALL case, fall through and treat as \C */            break;
4088    
4089              case OP_ALLANY:
4090            case OP_ANYBYTE:            case OP_ANYBYTE:
4091            c = max - min;            c = max - min;
4092            if (c > md->end_subject - eptr) c = md->end_subject - eptr;            if (c > (unsigned int)(md->end_subject - eptr))
4093                c = md->end_subject - eptr;
4094            eptr += c;            eptr += c;
4095            break;            break;
4096    
4097              case OP_ANYNL:
4098              for (i = min; i < max; i++)
4099                {
4100                if (eptr >= md->end_subject) break;
4101                c = *eptr;
4102                if (c == 0x000d)
4103                  {
4104                  if (++eptr >= md->end_subject) break;
4105                  if (*eptr == 0x000a) eptr++;
4106                  }
4107                else
4108                  {
4109                  if (c != 0x000a &&
4110                      (md->bsr_anycrlf ||
4111                        (c != 0x000b && c != 0x000c && c != 0x0085)))
4112                    break;
4113                  eptr++;
4114                  }
4115                }
4116              break;
4117    
4118              case OP_NOT_HSPACE:
4119              for (i = min; i < max; i++)
4120                {
4121                if (eptr >= md->end_subject) break;
4122                c = *eptr;
4123                if (c == 0x09 || c == 0x20 || c == 0xa0) break;
4124                eptr++;
4125                }
4126              break;
4127    
4128              case OP_HSPACE:
4129              for (i = min; i < max; i++)
4130                {
4131                if (eptr >= md->end_subject) break;
4132                c = *eptr;
4133                if (c != 0x09 && c != 0x20 && c != 0xa0) break;
4134                eptr++;
4135                }
4136              break;
4137    
4138              case OP_NOT_VSPACE:
4139              for (i = min; i < max; i++)
4140                {
4141                if (eptr >= md->end_subject) break;
4142                c = *eptr;
4143                if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85)
4144                  break;
4145                eptr++;
4146                }
4147              break;
4148    
4149              case OP_VSPACE:
4150              for (i = min; i < max; i++)
4151                {
4152                if (eptr >= md->end_subject) break;
4153                c = *eptr;
4154                if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85)
4155                  break;
4156                eptr++;
4157                }
4158              break;
4159    
4160            case OP_NOT_DIGIT:            case OP_NOT_DIGIT:
4161            for (i = min; i < max; i++)            for (i = min; i < max; i++)
4162              {              {
# Line 3085  for (;;) Line 4217  for (;;)
4217    
4218          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
4219    
4220            if (possessive) continue;
4221          while (eptr >= pp)          while (eptr >= pp)
4222            {            {
4223            RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM47);
4224            eptr--;            eptr--;
4225            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4226            }            }
# Line 3099  for (;;) Line 4232  for (;;)
4232        }        }
4233      /* Control never gets here */      /* Control never gets here */
4234    
4235      /* 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
4236      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. */  
4237    
4238      default:      default:
4239      DPRINTF(("Unknown opcode %d\n", *ecode));      DPRINTF(("Unknown opcode %d\n", *ecode));
4240      RRETURN(PCRE_ERROR_UNKNOWN_NODE);      RRETURN(PCRE_ERROR_UNKNOWN_OPCODE);
4241      }      }
4242    
4243    /* 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 4246  for (;;)
4246    
4247    }             /* End of main loop */    }             /* End of main loop */
4248  /* Control never reaches here */  /* Control never reaches here */
4249    
4250    
4251    /* When compiling to use the heap rather than the stack for recursive calls to
4252    match(), the RRETURN() macro jumps here. The number that is saved in
4253    frame->Xwhere indicates which label we actually want to return to. */
4254    
4255    #ifdef NO_RECURSE
4256    #define LBL(val) case val: goto L_RM##val;
4257    HEAP_RETURN:
4258    switch (frame->Xwhere)
4259      {
4260      LBL( 1) LBL( 2) LBL( 3) LBL( 4) LBL( 5) LBL( 6) LBL( 7) LBL( 8)
4261      LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17)
4262      LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33)
4263      LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52)
4264      LBL(53) LBL(54)
4265    #ifdef SUPPORT_UTF8
4266      LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30)
4267      LBL(32) LBL(34) LBL(42) LBL(46)
4268    #ifdef SUPPORT_UCP
4269      LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)
4270    #endif  /* SUPPORT_UCP */
4271    #endif  /* SUPPORT_UTF8 */
4272      default:
4273      DPRINTF(("jump error in pcre match: label %d non-existent\n", frame->Xwhere));
4274      return PCRE_ERROR_INTERNAL;
4275      }
4276    #undef LBL
4277    #endif  /* NO_RECURSE */
4278  }  }
4279    
4280    
# Line 3127  Undefine all the macros that were define Line 4287  Undefine all the macros that were define
4287  #ifdef NO_RECURSE  #ifdef NO_RECURSE
4288  #undef eptr  #undef eptr
4289  #undef ecode  #undef ecode
4290    #undef mstart
4291  #undef offset_top  #undef offset_top
4292  #undef ims  #undef ims
4293  #undef eptrb  #undef eptrb
# Line 3144  Undefine all the macros that were define Line 4305  Undefine all the macros that were define
4305    
4306  #undef cur_is_word  #undef cur_is_word
4307  #undef condition  #undef condition
 #undef minimize  
4308  #undef prev_is_word  #undef prev_is_word
4309    
4310  #undef original_ims  #undef original_ims
# Line 3200  Returns:          > 0 => success; value Line 4360  Returns:          > 0 => success; value
4360                   < -1 => some kind of unexpected problem                   < -1 => some kind of unexpected problem
4361  */  */
4362    
4363  PCRE_EXPORT int  PCRE_EXP_DEFN int
4364  pcre_exec(const pcre *argument_re, const pcre_extra *extra_data,  pcre_exec(const pcre *argument_re, const pcre_extra *extra_data,
4365    const char *subject, int length, int start_offset, int options, int *offsets,    PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
4366    int offsetcount)    int offsetcount)
4367  {  {
4368  int rc, resetcount, ocount;  int rc, resetcount, ocount;
4369  int first_byte = -1;  int first_byte = -1;
4370  int req_byte = -1;  int req_byte = -1;
4371  int req_byte2 = -1;  int req_byte2 = -1;
4372  unsigned long int ims = 0;  int newline;
4373    unsigned long int ims;
4374  BOOL using_temporary_offsets = FALSE;  BOOL using_temporary_offsets = FALSE;
4375  BOOL anchored;  BOOL anchored;
4376  BOOL startline;  BOOL startline;
4377  BOOL firstline;  BOOL firstline;
4378  BOOL first_byte_caseless = FALSE;  BOOL first_byte_caseless = FALSE;
4379  BOOL req_byte_caseless = FALSE;  BOOL req_byte_caseless = FALSE;
4380    BOOL utf8;
4381  match_data match_block;  match_data match_block;
4382    match_data *md = &match_block;
4383  const uschar *tables;  const uschar *tables;
4384  const uschar *start_bits = NULL;  const uschar *start_bits = NULL;
4385  const uschar *start_match = (const uschar *)subject + start_offset;  USPTR start_match = (USPTR)subject + start_offset;
4386  const uschar *end_subject;  USPTR end_subject;
4387  const uschar *req_byte_ptr = start_match - 1;  USPTR req_byte_ptr = start_match - 1;
4388    
4389  pcre_study_data internal_study;  pcre_study_data internal_study;
4390  const pcre_study_data *study;  const pcre_study_data *study;
# Line 3241  if (offsetcount < 0) return PCRE_ERROR_B Line 4404  if (offsetcount < 0) return PCRE_ERROR_B
4404  the default values. */  the default values. */
4405    
4406  study = NULL;  study = NULL;
4407  match_block.match_limit = MATCH_LIMIT;  md->match_limit = MATCH_LIMIT;
4408  match_block.callout_data = NULL;  md->match_limit_recursion = MATCH_LIMIT_RECURSION;
4409    md->callout_data = NULL;
4410    
4411  /* The table pointer is always in native byte order. */  /* The table pointer is always in native byte order. */
4412    
# Line 3254  if (extra_data != NULL) Line 4418  if (extra_data != NULL)
4418    if ((flags & PCRE_EXTRA_STUDY_DATA) != 0)    if ((flags & PCRE_EXTRA_STUDY_DATA) != 0)
4419      study = (const pcre_study_data *)extra_data->study_data;      study = (const pcre_study_data *)extra_data->study_data;
4420    if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0)    if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0)
4421      match_block.match_limit = extra_data->match_limit;      md->match_limit = extra_data->match_limit;
4422      if ((flags & PCRE_EXTRA_MATCH_LIMIT_RECURSION) != 0)
4423        md->match_limit_recursion = extra_data->match_limit_recursion;
4424    if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0)    if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0)
4425      match_block.callout_data = extra_data->callout_data;      md->callout_data = extra_data->callout_data;
4426    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;
4427    }    }
4428    
# Line 3281  if (re->magic_number != MAGIC_NUMBER) Line 4447  if (re->magic_number != MAGIC_NUMBER)
4447  /* Set up other data */  /* Set up other data */
4448    
4449  anchored = ((re->options | options) & PCRE_ANCHORED) != 0;  anchored = ((re->options | options) & PCRE_ANCHORED) != 0;
4450  startline = (re->options & PCRE_STARTLINE) != 0;  startline = (re->flags & PCRE_STARTLINE) != 0;
4451  firstline = (re->options & PCRE_FIRSTLINE) != 0;  firstline = (re->options & PCRE_FIRSTLINE) != 0;
4452    
4453  /* 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. */
4454    
4455  match_block.start_code = (const uschar *)external_re + re->name_table_offset +  md->start_code = (const uschar *)external_re + re->name_table_offset +
4456    re->name_count * re->name_entry_size;    re->name_count * re->name_entry_size;
4457    
4458  match_block.start_subject = (const uschar *)subject;  md->start_subject = (USPTR)subject;
4459  match_block.start_offset = start_offset;  md->start_offset = start_offset;
4460  match_block.end_subject = match_block.start_subject + length;  md->end_subject = md->start_subject + length;
4461  end_subject = match_block.end_subject;  end_subject = md->end_subject;
4462    
4463  match_block.endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
4464  match_block.utf8 = (re->options & PCRE_UTF8) != 0;  utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0;
4465    md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
4466  match_block.notbol = (options & PCRE_NOTBOL) != 0;  
4467  match_block.noteol = (options & PCRE_NOTEOL) != 0;  md->notbol = (options & PCRE_NOTBOL) != 0;
4468  match_block.notempty = (options & PCRE_NOTEMPTY) != 0;  md->noteol = (options & PCRE_NOTEOL) != 0;
4469  match_block.partial = (options & PCRE_PARTIAL) != 0;  md->notempty = (options & PCRE_NOTEMPTY) != 0;
4470  match_block.hitend = FALSE;  md->partial = (options & PCRE_PARTIAL) != 0;
4471    md->hitend = FALSE;
4472    
4473    md->recursive = NULL;                   /* No recursion at top level */
4474    
4475  match_block.recursive = NULL;                   /* No recursion at top level */  md->lcc = tables + lcc_offset;
4476    md->ctypes = tables + ctypes_offset;
4477    
4478  match_block.lcc = tables + lcc_offset;  /* Handle different \R options. */
4479  match_block.ctypes = tables + ctypes_offset;  
4480    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
4481      {
4482      case 0:
4483      if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0)
4484        md->bsr_anycrlf = (re->options & PCRE_BSR_ANYCRLF) != 0;
4485      else
4486    #ifdef BSR_ANYCRLF
4487      md->bsr_anycrlf = TRUE;
4488    #else
4489      md->bsr_anycrlf = FALSE;
4490    #endif
4491      break;
4492    
4493      case PCRE_BSR_ANYCRLF:
4494      md->bsr_anycrlf = TRUE;
4495      break;
4496    
4497      case PCRE_BSR_UNICODE:
4498      md->bsr_anycrlf = FALSE;
4499      break;
4500    
4501      default: return PCRE_ERROR_BADNEWLINE;
4502      }
4503    
4504    /* Handle different types of newline. The three bits give eight cases. If
4505    nothing is set at run time, whatever was used at compile time applies. */
4506    
4507    switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options :
4508            (pcre_uint32)options) & PCRE_NEWLINE_BITS)
4509      {
4510      case 0: newline = NEWLINE; break;   /* Compile-time default */
4511      case PCRE_NEWLINE_CR: newline = '\r'; break;
4512      case PCRE_NEWLINE_LF: newline = '\n'; break;
4513      case PCRE_NEWLINE_CR+
4514           PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;
4515      case PCRE_NEWLINE_ANY: newline = -1; break;
4516      case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
4517      default: return PCRE_ERROR_BADNEWLINE;
4518      }
4519    
4520    if (newline == -2)
4521      {
4522      md->nltype = NLTYPE_ANYCRLF;
4523      }
4524    else if (newline < 0)
4525      {
4526      md->nltype = NLTYPE_ANY;
4527      }
4528    else
4529      {
4530      md->nltype = NLTYPE_FIXED;
4531      if (newline > 255)
4532        {
4533        md->nllen = 2;
4534        md->nl[0] = (newline >> 8) & 255;
4535        md->nl[1] = newline & 255;
4536        }
4537      else
4538        {
4539        md->nllen = 1;
4540        md->nl[0] = newline;
4541        }
4542      }
4543    
4544  /* 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
4545  moment. */  moment. */
4546    
4547  if (match_block.partial && (re->options & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
4548    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
4549    
4550  /* 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
4551  back the character offset. */  back the character offset. */
4552    
4553  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
4554  if (match_block.utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
4555    {    {
4556    if (_pcre_valid_utf8((uschar *)subject, length) >= 0)    if (_pcre_valid_utf8((uschar *)subject, length) >= 0)
4557      return PCRE_ERROR_BADUTF8;      return PCRE_ERROR_BADUTF8;
# Line 3350  ocount = offsetcount - (offsetcount % 3) Line 4583  ocount = offsetcount - (offsetcount % 3)
4583  if (re->top_backref > 0 && re->top_backref >= ocount/3)  if (re->top_backref > 0 && re->top_backref >= ocount/3)
4584    {    {
4585    ocount = re->top_backref * 3 + 3;    ocount = re->top_backref * 3 + 3;
4586    match_block.offset_vector = (int *)(pcre_malloc)(ocount * sizeof(int));    md->offset_vector = (int *)(pcre_malloc)(ocount * sizeof(int));
4587    if (match_block.offset_vector == NULL) return PCRE_ERROR_NOMEMORY;    if (md->offset_vector == NULL) return PCRE_ERROR_NOMEMORY;
4588    using_temporary_offsets = TRUE;    using_temporary_offsets = TRUE;
4589    DPRINTF(("Got memory to hold back references\n"));    DPRINTF(("Got memory to hold back references\n"));
4590    }    }
4591  else match_block.offset_vector = offsets;  else md->offset_vector = offsets;
4592    
4593  match_block.offset_end = ocount;  md->offset_end = ocount;
4594  match_block.offset_max = (2*ocount)/3;  md->offset_max = (2*ocount)/3;
4595  match_block.offset_overflow = FALSE;  md->offset_overflow = FALSE;
4596  match_block.capture_last = -1;  md->capture_last = -1;
4597    
4598  /* 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
4599  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 4606  if (resetcount > offsetcount) resetcount
4606  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
4607  initialize them to avoid reading uninitialized locations. */  initialize them to avoid reading uninitialized locations. */
4608    
4609  if (match_block.offset_vector != NULL)  if (md->offset_vector != NULL)
4610    {    {
4611    register int *iptr = match_block.offset_vector + ocount;    register int *iptr = md->offset_vector + ocount;
4612    register int *iend = iptr - resetcount/2 + 1;    register int *iend = iptr - resetcount/2 + 1;
4613    while (--iptr >= iend) *iptr = -1;    while (--iptr >= iend) *iptr = -1;
4614    }    }
# Line 3388  studied, there may be a bitmap of possib Line 4621  studied, there may be a bitmap of possib
4621    
4622  if (!anchored)  if (!anchored)
4623    {    {
4624    if ((re->options & PCRE_FIRSTSET) != 0)    if ((re->flags & PCRE_FIRSTSET) != 0)
4625      {      {
4626      first_byte = re->first_byte & 255;      first_byte = re->first_byte & 255;
4627      if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE)      if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE)
4628        first_byte = match_block.lcc[first_byte];        first_byte = md->lcc[first_byte];
4629      }      }
4630    else    else
4631      if (!startline && study != NULL &&      if (!startline && study != NULL &&
# Line 3403  if (!anchored) Line 4636  if (!anchored)
4636  /* For anchored or unanchored matches, there may be a "last known required  /* For anchored or unanchored matches, there may be a "last known required
4637  character" set. */  character" set. */
4638    
4639  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
4640    {    {
4641    req_byte = re->req_byte & 255;    req_byte = re->req_byte & 255;
4642    req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;    req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;
4643    req_byte2 = (tables + fcc_offset)[req_byte];  /* case flipped */    req_byte2 = (tables + fcc_offset)[req_byte];  /* case flipped */
4644    }    }
4645    
4646    
4647    /* ==========================================================================*/
4648    
4649  /* Loop for handling unanchored repeated matching attempts; for anchored regexs  /* Loop for handling unanchored repeated matching attempts; for anchored regexs
4650  the loop runs just once. */  the loop runs just once. */
4651    
4652  do  for(;;)
4653    {    {
4654    const uschar *save_end_subject = end_subject;    USPTR save_end_subject = end_subject;
4655      USPTR new_start_match;
4656    
4657    /* Reset the maximum number of extractions we might see. */    /* Reset the maximum number of extractions we might see. */
4658    
4659    if (match_block.offset_vector != NULL)    if (md->offset_vector != NULL)
4660      {      {
4661      register int *iptr = match_block.offset_vector;      register int *iptr = md->offset_vector;
4662      register int *iend = iptr + resetcount;      register int *iend = iptr + resetcount;
4663      while (iptr < iend) *iptr++ = -1;      while (iptr < iend) *iptr++ = -1;
4664      }      }
4665    
4666    /* 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
4667    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.
4668    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
4669    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
4670    */    the match fails at the newline, later code breaks this loop. */
4671    
4672    if (firstline)    if (firstline)
4673      {      {
4674      const uschar *t = start_match;      USPTR t = start_match;
4675      while (t < save_end_subject && *t != '\n') t++;      while (t < md->end_subject && !IS_NEWLINE(t)) t++;
4676      end_subject = t;      end_subject = t;
4677      }      }
4678    
# Line 3445  do Line 4682  do
4682      {      {
4683      if (first_byte_caseless)      if (first_byte_caseless)
4684        while (start_match < end_subject &&        while (start_match < end_subject &&
4685               match_block.lcc[*start_match] != first_byte)               md->lcc[*start_match] != first_byte)
4686          start_match++;          { NEXTCHAR(start_match); }
4687      else      else
4688        while (start_match < end_subject && *start_match != first_byte)        while (start_match < end_subject && *start_match != first_byte)
4689          start_match++;          { NEXTCHAR(start_match); }
4690      }      }
4691    
4692    /* Or to just after \n for a multiline match if possible */    /* Or to just after a linebreak for a multiline match if possible */
4693    
4694    else if (startline)    else if (startline)
4695      {      {
4696      if (start_match > match_block.start_subject + start_offset)      if (start_match > md->start_subject + start_offset)
4697        {        {
4698        while (start_match < end_subject && start_match[-1] != NEWLINE)        while (start_match <= end_subject && !WAS_NEWLINE(start_match))
4699            { NEXTCHAR(start_match); }
4700    
4701          /* If we have just passed a CR and the newline option is ANY or ANYCRLF,
4702          and we are now at a LF, advance the match position by one more character.
4703          */
4704    
4705          if (start_match[-1] == '\r' &&
4706               (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
4707               start_match < end_subject &&
4708               *start_match == '\n')
4709          start_match++;          start_match++;
4710        }        }
4711      }      }
# Line 3470  do Line 4717  do
4717      while (start_match < end_subject)      while (start_match < end_subject)
4718        {        {
4719        register unsigned int c = *start_match;        register unsigned int c = *start_match;
4720        if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++; else break;        if ((start_bits[c/8] & (1 << (c&7))) == 0)
4721            { NEXTCHAR(start_match); }
4722          else break;
4723        }        }
4724      }      }
4725    
# Line 3480  do Line 4729  do
4729    
4730  #ifdef DEBUG  /* Sigh. Some compilers never learn. */  #ifdef DEBUG  /* Sigh. Some compilers never learn. */
4731    printf(">>>> Match against: ");    printf(">>>> Match against: ");
4732    pchars(start_match, end_subject - start_match, TRUE, &match_block);    pchars(start_match, end_subject - start_match, TRUE, md);
4733    printf("\n");    printf("\n");
4734  #endif  #endif
4735    
# Line 3494  do Line 4743  do
4743    
4744    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
4745    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
4746    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
4747    don't do this when the string is sufficiently long.    string... so we don't do this when the string is sufficiently long.
4748    
4749    ALSO: this processing is disabled when partial matching is requested.    ALSO: this processing is disabled when partial matching is requested.
4750    */    */
4751    
4752    if (req_byte >= 0 &&    if (req_byte >= 0 &&
4753        end_subject - start_match < REQ_BYTE_MAX &&        end_subject - start_match < REQ_BYTE_MAX &&
4754        !match_block.partial)        !md->partial)
4755      {      {
4756      register const uschar *p = start_match + ((first_byte >= 0)? 1 : 0);      register USPTR p = start_match + ((first_byte >= 0)? 1 : 0);
4757    
4758      /* 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
4759      place we found it at last time. */      place we found it at last time. */
# Line 3527  do Line 4776  do
4776            }            }
4777          }          }
4778    
4779        /* If we can't find the required character, break the matching loop */        /* If we can't find the required character, break the matching loop,
4780          forcing a match failure. */
4781    
4782        if (p >= end_subject) break;        if (p >= end_subject)
4783            {
4784            rc = MATCH_NOMATCH;
4785            break;
4786            }
4787    
4788        /* If we have found the required character, save the point where we        /* If we have found the required character, save the point where we
4789        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 4793  do
4793        }        }
4794      }      }
4795    
4796    /* When a match occurs, substrings will be set for all internal extractions;    /* OK, we can now run the match. */
4797    we just need to set up the whole thing as substring 0 before returning. If  
4798    there were too many extractions, set the return code to zero. In the case    md->start_match_ptr = start_match;
4799    where we had to get some local store to hold offsets for backreferences, copy    md->match_call_count = 0;
4800    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. */  
4801    
4802    if (rc == MATCH_NOMATCH)    switch(rc)
4803      {      {
4804      if (firstline && *start_match == NEWLINE) break;      /* NOMATCH and PRUNE advance by one character. THEN at this level acts
4805      start_match++;      exactly like PRUNE. */
4806    
4807        case MATCH_NOMATCH:
4808        case MATCH_PRUNE:
4809        case MATCH_THEN:
4810        new_start_match = start_match + 1;
4811  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
4812      if (match_block.utf8)      if (utf8)
4813        while(start_match < end_subject && (*start_match & 0xc0) == 0x80)        while(new_start_match < end_subject && (*new_start_match & 0xc0) == 0x80)
4814          start_match++;          new_start_match++;
4815  #endif  #endif
4816      continue;      break;
     }  
4817    
4818    if (rc != MATCH_MATCH)      /* SKIP passes back the next starting point explicitly. */
4819      {  
4820      DPRINTF((">>>> error: returning %d\n", rc));      case MATCH_SKIP:
4821      return rc;      new_start_match = md->start_match_ptr;
4822        break;
4823    
4824        /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */
4825    
4826        case MATCH_COMMIT:
4827        rc = MATCH_NOMATCH;
4828        goto ENDLOOP;
4829    
4830        /* Any other return is some kind of error. */
4831    
4832        default:
4833        goto ENDLOOP;
4834      }      }
4835    
4836    /* 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"
4837    necessary */    result. Reset the code to MATCH_NOMATCH for subsequent checking. */
4838    
4839      rc = MATCH_NOMATCH;
4840    
4841      /* If PCRE_FIRSTLINE is set, the match must happen before or at the first
4842      newline in the subject (though it may continue over the newline). Therefore,
4843      if we have just failed to match, starting at a newline, do not continue. */
4844    
4845      if (firstline && IS_NEWLINE(start_match)) break;
4846    
4847      /* Advance to new matching position */
4848    
4849      start_match = new_start_match;
4850    
4851      /* Break the loop if the pattern is anchored or if we have passed the end of
4852      the subject. */
4853    
4854      if (anchored || start_match > end_subject) break;
4855    
4856      /* If we have just passed a CR and we are now at a LF, and the pattern does
4857      not contain any explicit matches for \r or \n, and the newline option is CRLF
4858      or ANY or ANYCRLF, advance the match position by one more character. */
4859    
4860      if (start_match[-1] == '\r' &&
4861          start_match < end_subject &&
4862          *start_match == '\n' &&
4863          (re->flags & PCRE_HASCRORLF) == 0 &&
4864            (md->nltype == NLTYPE_ANY ||
4865             md->nltype == NLTYPE_ANYCRLF ||
4866             md->nllen == 2))
4867        start_match++;
4868    
4869      }   /* End of for(;;) "bumpalong" loop */
4870    
4871    /* ==========================================================================*/
4872