/[pcre]/code/trunk/pcre_internal.h
ViewVC logotype

Diff of /code/trunk/pcre_internal.h

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 850 by zherczeg, Wed Jan 4 17:29:11 2012 UTC revision 1408 by ph10, Tue Dec 3 16:27:00 2013 UTC
# Line 7  Line 7 
7  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.
8    
9                         Written by Philip Hazel                         Written by Philip Hazel
10             Copyright (c) 1997-2012 University of Cambridge             Copyright (c) 1997-2013 University of Cambridge
11    
12  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
13  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 40  POSSIBILITY OF SUCH DAMAGE. Line 40  POSSIBILITY OF SUCH DAMAGE.
40    
41  /* This header contains definitions that are shared between the different  /* This header contains definitions that are shared between the different
42  modules, but which are not relevant to the exported API. This includes some  modules, but which are not relevant to the exported API. This includes some
43  functions whose names all begin with "_pcre_" or "_pcre16_" depending on  functions whose names all begin with "_pcre_", "_pcre16_" or "_pcre32_"
44  the PRIV macro. */  depending on the PRIV macro. */
45    
46  #ifndef PCRE_INTERNAL_H  #ifndef PCRE_INTERNAL_H
47  #define PCRE_INTERNAL_H  #define PCRE_INTERNAL_H
# Line 53  the PRIV macro. */ Line 53  the PRIV macro. */
53  #endif  #endif
54    
55  /* PCRE is compiled as an 8 bit library if it is not requested otherwise. */  /* PCRE is compiled as an 8 bit library if it is not requested otherwise. */
56  #ifndef COMPILE_PCRE16  
57    #if !defined COMPILE_PCRE16 && !defined COMPILE_PCRE32
58  #define COMPILE_PCRE8  #define COMPILE_PCRE8
59  #endif  #endif
60    
# Line 78  Until then we define it if SUPPORT_UTF i Line 79  Until then we define it if SUPPORT_UTF i
79  #define SUPPORT_UTF8 1  #define SUPPORT_UTF8 1
80  #endif  #endif
81    
82  /* We do not support both EBCDIC and UTF-8/16 at the same time. The "configure"  /* We do not support both EBCDIC and UTF-8/16/32 at the same time. The "configure"
83  script prevents both being selected, but not everybody uses "configure". */  script prevents both being selected, but not everybody uses "configure". */
84    
85  #if defined EBCDIC && defined SUPPORT_UTF  #if defined EBCDIC && defined SUPPORT_UTF
86  #error The use of both EBCDIC and SUPPORT_UTF8/16 is not supported.  #error The use of both EBCDIC and SUPPORT_UTF is not supported.
87  #endif  #endif
88    
89  /* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef  /* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef
# Line 111  setjmp and stdarg are used is when NO_RE Line 112  setjmp and stdarg are used is when NO_RE
112  #include <stdlib.h>  #include <stdlib.h>
113  #include <string.h>  #include <string.h>
114    
115    /* Valgrind (memcheck) support */
116    
117    #ifdef SUPPORT_VALGRIND
118    #include <valgrind/memcheck.h>
119    #endif
120    
121  /* When compiling a DLL for Windows, the exported symbols have to be declared  /* When compiling a DLL for Windows, the exported symbols have to be declared
122  using some MS magic. I found some useful information on this web page:  using some MS magic. I found some useful information on this web page:
123  http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the  http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the
# Line 187  preprocessor time in standard C environm Line 194  preprocessor time in standard C environm
194  typedef unsigned char pcre_uint8;  typedef unsigned char pcre_uint8;
195    
196  #if USHRT_MAX == 65535  #if USHRT_MAX == 65535
197    typedef unsigned short pcre_uint16;  typedef unsigned short pcre_uint16;
198    typedef short pcre_int16;  typedef short pcre_int16;
199    #define PCRE_UINT16_MAX USHRT_MAX
200    #define PCRE_INT16_MAX SHRT_MAX
201  #elif UINT_MAX == 65535  #elif UINT_MAX == 65535
202    typedef unsigned int pcre_uint16;  typedef unsigned int pcre_uint16;
203    typedef int pcre_int16;  typedef int pcre_int16;
204    #define PCRE_UINT16_MAX UINT_MAX
205    #define PCRE_INT16_MAX INT_MAX
206  #else  #else
207    #error Cannot determine a type for 16-bit unsigned integers  #error Cannot determine a type for 16-bit integers
208  #endif  #endif
209    
210  #if UINT_MAX == 4294967295  #if UINT_MAX == 4294967295U
211    typedef unsigned int pcre_uint32;  typedef unsigned int pcre_uint32;
212    typedef int pcre_int32;  typedef int pcre_int32;
213  #elif ULONG_MAX == 4294967295  #define PCRE_UINT32_MAX UINT_MAX
214    typedef unsigned long int pcre_uint32;  #define PCRE_INT32_MAX INT_MAX
215    typedef long int pcre_int32;  #elif ULONG_MAX == 4294967295UL
216    typedef unsigned long int pcre_uint32;
217    typedef long int pcre_int32;
218    #define PCRE_UINT32_MAX ULONG_MAX
219    #define PCRE_INT32_MAX LONG_MAX
220  #else  #else
221    #error Cannot determine a type for 32-bit unsigned integers  #error Cannot determine a type for 32-bit integers
222  #endif  #endif
223    
224  /* When checking for integer overflow in pcre_compile(), we need to handle  /* When checking for integer overflow in pcre_compile(), we need to handle
# Line 214  stdint.h is available, include it; it ma Line 229  stdint.h is available, include it; it ma
229  have stdint.h (e.g. Solaris) may have inttypes.h. The macro int64_t may be set  have stdint.h (e.g. Solaris) may have inttypes.h. The macro int64_t may be set
230  by "configure". */  by "configure". */
231    
232  #if HAVE_STDINT_H  #if defined HAVE_STDINT_H
233  #include <stdint.h>  #include <stdint.h>
234  #elif HAVE_INTTYPES_H  #elif defined HAVE_INTTYPES_H
235  #include <inttypes.h>  #include <inttypes.h>
236  #endif  #endif
237    
# Line 243  exactly 256 items. When the character is Line 258  exactly 256 items. When the character is
258  items, some check is needed before accessing these tables.  items, some check is needed before accessing these tables.
259  */  */
260    
261  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
262    
263  typedef unsigned char pcre_uchar;  typedef unsigned char pcre_uchar;
264  #define IN_UCHARS(x) (x)  #define IN_UCHARS(x) (x)
265  #define MAX_255(c) 1  #define MAX_255(c) 1
266  #define TABLE_GET(c, table, default) ((table)[c])  #define TABLE_GET(c, table, default) ((table)[c])
267    
268  #else  #elif defined COMPILE_PCRE16
269    
 #ifdef COMPILE_PCRE16  
270  #if USHRT_MAX != 65535  #if USHRT_MAX != 65535
271  /* This is a warning message. Change PCRE_SCHAR16 to a 16 bit data type in  /* This is a warning message. Change PCRE_UCHAR16 to a 16 bit data type in
272  pcre.h(.in) and disable (comment out) this message. */  pcre.h(.in) and disable (comment out) this message. */
273  #error Warning: PCRE_SCHAR16 is not a 16 bit data type.  #error Warning: PCRE_UCHAR16 is not a 16 bit data type.
274  #endif  #endif
275    
276  typedef pcre_uint16 pcre_uchar;  typedef pcre_uint16 pcre_uchar;
277  #define IN_UCHARS(x) ((x) << 1)  #define UCHAR_SHIFT (1)
278    #define IN_UCHARS(x) ((x) << UCHAR_SHIFT)
279    #define MAX_255(c) ((c) <= 255u)
280    #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
281    
282    #elif defined COMPILE_PCRE32
283    
284    typedef pcre_uint32 pcre_uchar;
285    #define UCHAR_SHIFT (2)
286    #define IN_UCHARS(x) ((x) << UCHAR_SHIFT)
287  #define MAX_255(c) ((c) <= 255u)  #define MAX_255(c) ((c) <= 255u)
288  #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))  #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
289    
290  #else  #else
291  #error Unsupported compiling mode  #error Unsupported compiling mode
292  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
   
 #endif /* COMPILE_PCRE8 */  
293    
294  /* This is an unsigned int value that no character can ever have. UTF-8  /* This is an unsigned int value that no character can ever have. UTF-8
295  characters only go up to 0x7fffffff (though Unicode doesn't go beyond  characters only go up to 0x7fffffff (though Unicode doesn't go beyond
# Line 295  start/end of string field names are. */ Line 316  start/end of string field names are. */
316         &(NLBLOCK->nllen), utf)) \         &(NLBLOCK->nllen), utf)) \
317      : \      : \
318      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
319       (p)[0] == NLBLOCK->nl[0] && \       RAWUCHARTEST(p) == NLBLOCK->nl[0] && \
320       (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || RAWUCHARTEST(p+1) == NLBLOCK->nl[1])       \
321      ) \      ) \
322    )    )
323    
# Line 309  start/end of string field names are. */ Line 330  start/end of string field names are. */
330         &(NLBLOCK->nllen), utf)) \         &(NLBLOCK->nllen), utf)) \
331      : \      : \
332      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
333       (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \       RAWUCHARTEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] &&              \
334       (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || RAWUCHARTEST(p - NLBLOCK->nllen + 1) == NLBLOCK->nl[1]) \
335      ) \      ) \
336    )    )
337    
# Line 335  values. */ Line 356  values. */
356  #include "pcre.h"  #include "pcre.h"
357  #include "ucp.h"  #include "ucp.h"
358    
359    #ifdef COMPILE_PCRE32
360    /* Assert that the public PCRE_UCHAR32 is a 32-bit type */
361    typedef int __assert_pcre_uchar32_size[sizeof(PCRE_UCHAR32) == 4 ? 1 : -1];
362    #endif
363    
364  /* When compiling for use with the Virtual Pascal compiler, these functions  /* When compiling for use with the Virtual Pascal compiler, these functions
365  need to have their names changed. PCRE must be compiled with the -DVPCOMPAT  need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
366  option on the command line. */  option on the command line. */
# Line 396  The macros are controlled by the value o Line 422  The macros are controlled by the value o
422  the config.h file, but can be overridden by using -D on the command line. This  the config.h file, but can be overridden by using -D on the command line. This
423  is automated on Unix systems via the "configure" command. */  is automated on Unix systems via the "configure" command. */
424    
425  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
426    
427  #if LINK_SIZE == 2  #if LINK_SIZE == 2
428    
# Line 441  is automated on Unix systems via the "co Line 467  is automated on Unix systems via the "co
467  #error LINK_SIZE must be either 2, 3, or 4  #error LINK_SIZE must be either 2, 3, or 4
468  #endif  #endif
469    
470  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
   
 #ifdef COMPILE_PCRE16  
471    
472  #if LINK_SIZE == 2  #if LINK_SIZE == 2
473    
474    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
475  #undef LINK_SIZE  #undef LINK_SIZE
476  #define LINK_SIZE 1  #define LINK_SIZE 1
477    
# Line 460  is automated on Unix systems via the "co Line 485  is automated on Unix systems via the "co
485    
486  #elif LINK_SIZE == 3 || LINK_SIZE == 4  #elif LINK_SIZE == 3 || LINK_SIZE == 4
487    
488    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
489  #undef LINK_SIZE  #undef LINK_SIZE
490  #define LINK_SIZE 2  #define LINK_SIZE 2
491    
# Line 477  is automated on Unix systems via the "co Line 503  is automated on Unix systems via the "co
503  #error LINK_SIZE must be either 2, 3, or 4  #error LINK_SIZE must be either 2, 3, or 4
504  #endif  #endif
505    
506    #elif defined COMPILE_PCRE32
507    
508    /* Only supported LINK_SIZE is 4 */
509    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
510    #undef LINK_SIZE
511    #define LINK_SIZE 1
512    
513    #define PUT(a,n,d)   \
514      (a[n] = (d))
515    
516    #define GET(a,n) \
517      (a[n])
518    
519    /* Keep it positive */
520    #define MAX_PATTERN_SIZE (1 << 30)
521    
522  #else  #else
523  #error Unsupported compiling mode  #error Unsupported compiling mode
524  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
   
 #endif /* COMPILE_PCRE8 */  
525    
526  /* Convenience macro defined in terms of the others */  /* Convenience macro defined in terms of the others */
527    
# Line 492  is automated on Unix systems via the "co Line 532  is automated on Unix systems via the "co
532  offsets changes. There are used for repeat counts and for other things such as  offsets changes. There are used for repeat counts and for other things such as
533  capturing parenthesis numbers in back references. */  capturing parenthesis numbers in back references. */
534    
535  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
536    
537  #define IMM2_SIZE 2  #define IMM2_SIZE 2
538    
# Line 500  capturing parenthesis numbers in back re Line 540  capturing parenthesis numbers in back re
540    a[n] = (d) >> 8; \    a[n] = (d) >> 8; \
541    a[(n)+1] = (d) & 255    a[(n)+1] = (d) & 255
542    
543    /* For reasons that I do not understand, the expression in this GET2 macro is
544    treated by gcc as a signed expression, even when a is declared as unsigned. It
545    seems that any kind of arithmetic results in a signed value. */
546    
547  #define GET2(a,n) \  #define GET2(a,n) \
548    (((a)[n] << 8) | (a)[(n)+1])    (unsigned int)(((a)[n] << 8) | (a)[(n)+1])
549    
550  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
551    
552  #ifdef COMPILE_PCRE16  #define IMM2_SIZE 1
553    
554    #define PUT2(a,n,d)   \
555       a[n] = d
556    
557    #define GET2(a,n) \
558       a[n]
559    
560    #elif defined COMPILE_PCRE32
561    
562  #define IMM2_SIZE 1  #define IMM2_SIZE 1
563    
# Line 517  capturing parenthesis numbers in back re Line 569  capturing parenthesis numbers in back re
569    
570  #else  #else
571  #error Unsupported compiling mode  #error Unsupported compiling mode
572  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
   
 #endif /* COMPILE_PCRE8 */  
573    
574  #define PUT2INC(a,n,d)  PUT2(a,n,d), a += IMM2_SIZE  #define PUT2INC(a,n,d)  PUT2(a,n,d), a += IMM2_SIZE
575    
576    /* The maximum length of a MARK name is currently one data unit; it may be
577    changed in future to be a fixed number of bytes or to depend on LINK_SIZE. */
578    
579    #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
580    #define MAX_MARK ((1u << 16) - 1)
581    #else
582    #define MAX_MARK ((1u << 8) - 1)
583    #endif
584    
585  /* When UTF encoding is being used, a character is no longer just a single  /* When UTF encoding is being used, a character is no longer just a single
586  character. The macros for character handling generate simple sequences when  byte. The macros for character handling generate simple sequences when used in
587  used in character-mode, and more complicated ones for UTF characters.  character-mode, and more complicated ones for UTF characters. GETCHARLENTEST
588  GETCHARLENTEST and other macros are not used when UTF is not supported,  and other macros are not used when UTF is not supported, so they are not
589  so they are not defined. To make sure they can never even appear when  defined. To make sure they can never even appear when UTF support is omitted,
590  UTF support is omitted, we don't even define them. */  we don't even define them. */
591    
592  #ifndef SUPPORT_UTF  #ifndef SUPPORT_UTF
593    
# Line 541  UTF support is omitted, we don't even de Line 600  UTF support is omitted, we don't even de
600  #define GETCHARINC(c, eptr) c = *eptr++;  #define GETCHARINC(c, eptr) c = *eptr++;
601  #define GETCHARINCTEST(c, eptr) c = *eptr++;  #define GETCHARINCTEST(c, eptr) c = *eptr++;
602  #define GETCHARLEN(c, eptr, len) c = *eptr;  #define GETCHARLEN(c, eptr, len) c = *eptr;
603    #define RAWUCHAR(eptr) (*(eptr))
604    #define RAWUCHARINC(eptr) (*(eptr)++)
605    #define RAWUCHARTEST(eptr) (*(eptr))
606    #define RAWUCHARINCTEST(eptr) (*(eptr)++)
607  /* #define GETCHARLENTEST(c, eptr, len) */  /* #define GETCHARLENTEST(c, eptr, len) */
608  /* #define BACKCHAR(eptr) */  /* #define BACKCHAR(eptr) */
609  /* #define FORWARDCHAR(eptr) */  /* #define FORWARDCHAR(eptr) */
# Line 548  UTF support is omitted, we don't even de Line 611  UTF support is omitted, we don't even de
611    
612  #else   /* SUPPORT_UTF */  #else   /* SUPPORT_UTF */
613    
 #ifdef COMPILE_PCRE8  
   
 /* These macros were originally written in the form of loops that used data  
 from the tables whose names start with PRIV(utf8_table). They were rewritten by  
 a user so as not to use loops, because in some environments this gives a  
 significant performance advantage, and it seems never to do any harm. */  
   
 /* Tells the biggest code point which can be encoded as a single character. */  
   
 #define MAX_VALUE_FOR_SINGLE_CHAR 127  
   
614  /* Tests whether the code point needs extra characters to decode. */  /* Tests whether the code point needs extra characters to decode. */
615    
616  #define HAS_EXTRALEN(c) ((c) >= 0xc0)  #define HASUTF8EXTRALEN(c) ((c) >= 0xc0)
   
 /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.  
 Otherwise it has an undefined behaviour. */  
   
 #define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f])  
   
 /* Returns TRUE, if the given character is not the first character  
 of a UTF sequence. */  
   
 #define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80)  
617    
618  /* Base macro to pick up the remaining bytes of a UTF-8 character, not  /* Base macro to pick up the remaining bytes of a UTF-8 character, not
619  advancing the pointer. */  advancing the pointer. */
# Line 595  advancing the pointer. */ Line 637  advancing the pointer. */
637            ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \            ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \
638      }      }
639    
 /* Get the next UTF-8 character, not advancing the pointer. This is called when  
 we know we are in UTF-8 mode. */  
   
 #define GETCHAR(c, eptr) \  
   c = *eptr; \  
   if (c >= 0xc0) GETUTF8(c, eptr);  
   
 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the  
 pointer. */  
   
 #define GETCHARTEST(c, eptr) \  
   c = *eptr; \  
   if (utf && c >= 0xc0) GETUTF8(c, eptr);  
   
640  /* Base macro to pick up the remaining bytes of a UTF-8 character, advancing  /* Base macro to pick up the remaining bytes of a UTF-8 character, advancing
641  the pointer. */  the pointer. */
642    
# Line 643  the pointer. */ Line 671  the pointer. */
671        } \        } \
672      }      }
673    
674    #if defined COMPILE_PCRE8
675    
676    /* These macros were originally written in the form of loops that used data
677    from the tables whose names start with PRIV(utf8_table). They were rewritten by
678    a user so as not to use loops, because in some environments this gives a
679    significant performance advantage, and it seems never to do any harm. */
680    
681    /* Tells the biggest code point which can be encoded as a single character. */
682    
683    #define MAX_VALUE_FOR_SINGLE_CHAR 127
684    
685    /* Tests whether the code point needs extra characters to decode. */
686    
687    #define HAS_EXTRALEN(c) ((c) >= 0xc0)
688    
689    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
690    Otherwise it has an undefined behaviour. */
691    
692    #define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f])
693    
694    /* Returns TRUE, if the given character is not the first character
695    of a UTF sequence. */
696    
697    #define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80)
698    
699    /* Get the next UTF-8 character, not advancing the pointer. This is called when
700    we know we are in UTF-8 mode. */
701    
702    #define GETCHAR(c, eptr) \
703      c = *eptr; \
704      if (c >= 0xc0) GETUTF8(c, eptr);
705    
706    /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
707    pointer. */
708    
709    #define GETCHARTEST(c, eptr) \
710      c = *eptr; \
711      if (utf && c >= 0xc0) GETUTF8(c, eptr);
712    
713  /* Get the next UTF-8 character, advancing the pointer. This is called when we  /* Get the next UTF-8 character, advancing the pointer. This is called when we
714  know we are in UTF-8 mode. */  know we are in UTF-8 mode. */
715    
# Line 709  do not know if we are in UTF-8 mode. */ Line 776  do not know if we are in UTF-8 mode. */
776    c = *eptr; \    c = *eptr; \
777    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);
778    
779    /* Returns the next uchar, not advancing the pointer. This is called when
780    we know we are in UTF mode. */
781    
782    #define RAWUCHAR(eptr) \
783      (*(eptr))
784    
785    /* Returns the next uchar, advancing the pointer. This is called when
786    we know we are in UTF mode. */
787    
788    #define RAWUCHARINC(eptr) \
789      (*((eptr)++))
790    
791    /* Returns the next uchar, testing for UTF mode, and not advancing the
792    pointer. */
793    
794    #define RAWUCHARTEST(eptr) \
795      (*(eptr))
796    
797    /* Returns the next uchar, testing for UTF mode, advancing the
798    pointer. */
799    
800    #define RAWUCHARINCTEST(eptr) \
801      (*((eptr)++))
802    
803  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
804  it is. This is called only in UTF-8 mode - we don't put a test within the macro  it is. This is called only in UTF-8 mode - we don't put a test within the macro
805  because almost all calls are already within a block of UTF-8 only code. */  because almost all calls are already within a block of UTF-8 only code. */
# Line 722  because almost all calls are already wit Line 813  because almost all calls are already wit
813  #define ACROSSCHAR(condition, eptr, action) \  #define ACROSSCHAR(condition, eptr, action) \
814    while((condition) && ((eptr) & 0xc0) == 0x80) action    while((condition) && ((eptr) & 0xc0) == 0x80) action
815    
816  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
   
 #ifdef COMPILE_PCRE16  
817    
818  /* Tells the biggest code point which can be encoded as a single character. */  /* Tells the biggest code point which can be encoded as a single character. */
819    
# Line 806  we do not know if we are in UTF-16 mode. Line 895  we do not know if we are in UTF-16 mode.
895    c = *eptr; \    c = *eptr; \
896    if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);    if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
897    
898    /* Returns the next uchar, not advancing the pointer. This is called when
899    we know we are in UTF mode. */
900    
901    #define RAWUCHAR(eptr) \
902      (*(eptr))
903    
904    /* Returns the next uchar, advancing the pointer. This is called when
905    we know we are in UTF mode. */
906    
907    #define RAWUCHARINC(eptr) \
908      (*((eptr)++))
909    
910    /* Returns the next uchar, testing for UTF mode, and not advancing the
911    pointer. */
912    
913    #define RAWUCHARTEST(eptr) \
914      (*(eptr))
915    
916    /* Returns the next uchar, testing for UTF mode, advancing the
917    pointer. */
918    
919    #define RAWUCHARINCTEST(eptr) \
920      (*((eptr)++))
921    
922  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
923  it is. This is called only in UTF-16 mode - we don't put a test within the  it is. This is called only in UTF-16 mode - we don't put a test within the
924  macro because almost all calls are already within a block of UTF-16 only  macro because almost all calls are already within a block of UTF-16 only
# Line 820  code. */ Line 933  code. */
933  #define ACROSSCHAR(condition, eptr, action) \  #define ACROSSCHAR(condition, eptr, action) \
934    if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action    if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action
935    
936  #endif  #elif defined COMPILE_PCRE32
937    
938  #endif /* COMPILE_PCRE8 */  /* These are trivial for the 32-bit library, since all UTF-32 characters fit
939    into one pcre_uchar unit. */
940    #define MAX_VALUE_FOR_SINGLE_CHAR (0x10ffffu)
941    #define HAS_EXTRALEN(c) (0)
942    #define GET_EXTRALEN(c) (0)
943    #define NOT_FIRSTCHAR(c) (0)
944    
945    /* Get the next UTF-32 character, not advancing the pointer. This is called when
946    we know we are in UTF-32 mode. */
947    
948    #define GETCHAR(c, eptr) \
949      c = *(eptr);
950    
951    /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the
952    pointer. */
953    
954    #define GETCHARTEST(c, eptr) \
955      c = *(eptr);
956    
957    /* Get the next UTF-32 character, advancing the pointer. This is called when we
958    know we are in UTF-32 mode. */
959    
960    #define GETCHARINC(c, eptr) \
961      c = *((eptr)++);
962    
963    /* Get the next character, testing for UTF-32 mode, and advancing the pointer.
964    This is called when we don't know if we are in UTF-32 mode. */
965    
966    #define GETCHARINCTEST(c, eptr) \
967      c = *((eptr)++);
968    
969    /* Get the next UTF-32 character, not advancing the pointer, not incrementing
970    length (since all UTF-32 is of length 1). This is called when we know we are in
971    UTF-32 mode. */
972    
973    #define GETCHARLEN(c, eptr, len) \
974      GETCHAR(c, eptr)
975    
976    /* Get the next UTF-32character, testing for UTF-32 mode, not advancing the
977    pointer, not incrementing the length (since all UTF-32 is of length 1).
978    This is called when we do not know if we are in UTF-32 mode. */
979    
980    #define GETCHARLENTEST(c, eptr, len) \
981      GETCHARTEST(c, eptr)
982    
983    /* Returns the next uchar, not advancing the pointer. This is called when
984    we know we are in UTF mode. */
985    
986    #define RAWUCHAR(eptr) \
987      (*(eptr))
988    
989    /* Returns the next uchar, advancing the pointer. This is called when
990    we know we are in UTF mode. */
991    
992    #define RAWUCHARINC(eptr) \
993      (*((eptr)++))
994    
995    /* Returns the next uchar, testing for UTF mode, and not advancing the
996    pointer. */
997    
998    #define RAWUCHARTEST(eptr) \
999      (*(eptr))
1000    
1001    /* Returns the next uchar, testing for UTF mode, advancing the
1002    pointer. */
1003    
1004    #define RAWUCHARINCTEST(eptr) \
1005      (*((eptr)++))
1006    
1007    /* If the pointer is not at the start of a character, move it back until
1008    it is. This is called only in UTF-32 mode - we don't put a test within the
1009    macro because almost all calls are already within a block of UTF-32 only
1010    code.
1011    These are all no-ops since all UTF-32 characters fit into one pcre_uchar. */
1012    
1013    #define BACKCHAR(eptr) do { } while (0)
1014    
1015    /* Same as above, just in the other direction. */
1016    #define FORWARDCHAR(eptr) do { } while (0)
1017    
1018    /* Same as above, but it allows a fully customizable form. */
1019    #define ACROSSCHAR(condition, eptr, action) do { } while (0)
1020    
1021    #else
1022    #error Unsupported compiling mode
1023    #endif /* COMPILE_PCRE[8|16|32] */
1024    
1025  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
1026    
1027    /* Tests for Unicode horizontal and vertical whitespace characters must check a
1028    number of different values. Using a switch statement for this generates the
1029    fastest code (no loop, no memory access), and there are several places in the
1030    interpreter code where this happens. In order to ensure that all the case lists
1031    remain in step, we use macros so that there is only one place where the lists
1032    are defined.
1033    
1034    These values are also required as lists in pcre_compile.c when processing \h,
1035    \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but
1036    macros that define the values are here so that all the definitions are
1037    together. The lists must be in ascending character order, terminated by
1038    NOTACHAR (which is 0xffffffff).
1039    
1040    Any changes should ensure that the various macros are kept in step with each
1041    other. NOTE: The values also appear in pcre_jit_compile.c. */
1042    
1043    /* ------ ASCII/Unicode environments ------ */
1044    
1045    #ifndef EBCDIC
1046    
1047    #define HSPACE_LIST \
1048      CHAR_HT, CHAR_SPACE, 0xa0, \
1049      0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \
1050      0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \
1051      NOTACHAR
1052    
1053    #define HSPACE_MULTIBYTE_CASES \
1054      case 0x1680:  /* OGHAM SPACE MARK */ \
1055      case 0x180e:  /* MONGOLIAN VOWEL SEPARATOR */ \
1056      case 0x2000:  /* EN QUAD */ \
1057      case 0x2001:  /* EM QUAD */ \
1058      case 0x2002:  /* EN SPACE */ \
1059      case 0x2003:  /* EM SPACE */ \
1060      case 0x2004:  /* THREE-PER-EM SPACE */ \
1061      case 0x2005:  /* FOUR-PER-EM SPACE */ \
1062      case 0x2006:  /* SIX-PER-EM SPACE */ \
1063      case 0x2007:  /* FIGURE SPACE */ \
1064      case 0x2008:  /* PUNCTUATION SPACE */ \
1065      case 0x2009:  /* THIN SPACE */ \
1066      case 0x200A:  /* HAIR SPACE */ \
1067      case 0x202f:  /* NARROW NO-BREAK SPACE */ \
1068      case 0x205f:  /* MEDIUM MATHEMATICAL SPACE */ \
1069      case 0x3000   /* IDEOGRAPHIC SPACE */
1070    
1071    #define HSPACE_BYTE_CASES \
1072      case CHAR_HT: \
1073      case CHAR_SPACE: \
1074      case 0xa0     /* NBSP */
1075    
1076    #define HSPACE_CASES \
1077      HSPACE_BYTE_CASES: \
1078      HSPACE_MULTIBYTE_CASES
1079    
1080    #define VSPACE_LIST \
1081      CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR
1082    
1083    #define VSPACE_MULTIBYTE_CASES \
1084      case 0x2028:    /* LINE SEPARATOR */ \
1085      case 0x2029     /* PARAGRAPH SEPARATOR */
1086    
1087    #define VSPACE_BYTE_CASES \
1088      case CHAR_LF: \
1089      case CHAR_VT: \
1090      case CHAR_FF: \
1091      case CHAR_CR: \
1092      case CHAR_NEL
1093    
1094    #define VSPACE_CASES \
1095      VSPACE_BYTE_CASES: \
1096      VSPACE_MULTIBYTE_CASES
1097    
1098    /* ------ EBCDIC environments ------ */
1099    
1100    #else
1101    #define HSPACE_LIST CHAR_HT, CHAR_SPACE
1102    
1103    #define HSPACE_BYTE_CASES \
1104      case CHAR_HT: \
1105      case CHAR_SPACE
1106    
1107  /* In case there is no definition of offsetof() provided - though any proper  #define HSPACE_CASES HSPACE_BYTE_CASES
 Standard C system should have one. */  
1108    
1109  #ifndef offsetof  #ifdef EBCDIC_NL25
1110  #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))  #define VSPACE_LIST \
1111      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR
1112    #else
1113    #define VSPACE_LIST \
1114      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR
1115  #endif  #endif
1116    
1117    #define VSPACE_BYTE_CASES \
1118      case CHAR_LF: \
1119      case CHAR_VT: \
1120      case CHAR_FF: \
1121      case CHAR_CR: \
1122      case CHAR_NEL
1123    
1124    #define VSPACE_CASES VSPACE_BYTE_CASES
1125    #endif  /* EBCDIC */
1126    
1127    /* ------ End of whitespace macros ------ */
1128    
1129    
1130    
1131  /* Private flags containing information about the compiled regex. They used to  /* Private flags containing information about the compiled regex. They used to
1132  live at the top end of the options word, but that got almost full, so now they  live at the top end of the options word, but that got almost full, so they were
1133  are in a 16-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as  moved to a 16-bit flags word - which got almost full, so now they are in a
1134  the restrictions on partial matching have been lifted. It remains for backwards  32-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as the
1135    restrictions on partial matching have been lifted. It remains for backwards
1136  compatibility. */  compatibility. */
1137    
1138  #ifdef COMPILE_PCRE8  #define PCRE_MODE8         0x00000001  /* compiled in 8 bit mode */
1139  #define PCRE_MODE          0x0001  /* compiled in 8 bit mode */  #define PCRE_MODE16        0x00000002  /* compiled in 16 bit mode */
1140    #define PCRE_MODE32        0x00000004  /* compiled in 32 bit mode */
1141    #define PCRE_FIRSTSET      0x00000010  /* first_char is set */
1142    #define PCRE_FCH_CASELESS  0x00000020  /* caseless first char */
1143    #define PCRE_REQCHSET      0x00000040  /* req_byte is set */
1144    #define PCRE_RCH_CASELESS  0x00000080  /* caseless requested char */
1145    #define PCRE_STARTLINE     0x00000100  /* start after \n for multiline */
1146    #define PCRE_NOPARTIAL     0x00000200  /* can't use partial with this regex */
1147    #define PCRE_JCHANGED      0x00000400  /* j option used in regex */
1148    #define PCRE_HASCRORLF     0x00000800  /* explicit \r or \n in pattern */
1149    #define PCRE_HASTHEN       0x00001000  /* pattern contains (*THEN) */
1150    #define PCRE_MLSET         0x00002000  /* match limit set by regex */
1151    #define PCRE_RLSET         0x00004000  /* recursion limit set by regex */
1152    #define PCRE_MATCH_EMPTY   0x00008000  /* pattern can match empty string */
1153    
1154    #if defined COMPILE_PCRE8
1155    #define PCRE_MODE          PCRE_MODE8
1156    #elif defined COMPILE_PCRE16
1157    #define PCRE_MODE          PCRE_MODE16
1158    #elif defined COMPILE_PCRE32
1159    #define PCRE_MODE          PCRE_MODE32
1160  #endif  #endif
1161  #ifdef COMPILE_PCRE16  #define PCRE_MODE_MASK     (PCRE_MODE8 | PCRE_MODE16 | PCRE_MODE32)
 #define PCRE_MODE          0x0002  /* compiled in 16 bit mode */  
 #endif  
 #define PCRE_FIRSTSET      0x0010  /* first_char is set */  
 #define PCRE_FCH_CASELESS  0x0020  /* caseless first char */  
 #define PCRE_REQCHSET      0x0040  /* req_byte is set */  
 #define PCRE_RCH_CASELESS  0x0080  /* caseless requested char */  
 #define PCRE_STARTLINE     0x0100  /* start after \n for multiline */  
 #define PCRE_NOPARTIAL     0x0200  /* can't use partial with this regex */  
 #define PCRE_JCHANGED      0x0400  /* j option used in regex */  
 #define PCRE_HASCRORLF     0x0800  /* explicit \r or \n in pattern */  
 #define PCRE_HASTHEN       0x1000  /* pattern contains (*THEN) */  
1162    
1163  /* Flags for the "extra" block produced by pcre_study(). */  /* Flags for the "extra" block produced by pcre_study(). */
1164    
# Line 871  time, run time, or study time, respectiv Line 1174  time, run time, or study time, respectiv
1174  #define PUBLIC_COMPILE_OPTIONS \  #define PUBLIC_COMPILE_OPTIONS \
1175    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
1176     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
1177     PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \     PCRE_NO_AUTO_CAPTURE|PCRE_NO_AUTO_POSSESS| \
1178       PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
1179     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
1180     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE)     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE|PCRE_NEVER_UTF)
1181    
1182  #define PUBLIC_EXEC_OPTIONS \  #define PUBLIC_EXEC_OPTIONS \
1183    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \
# Line 887  time, run time, or study time, respectiv Line 1191  time, run time, or study time, respectiv
1191     PCRE_NO_START_OPTIMIZE)     PCRE_NO_START_OPTIMIZE)
1192    
1193  #define PUBLIC_STUDY_OPTIONS \  #define PUBLIC_STUDY_OPTIONS \
1194     PCRE_STUDY_JIT_COMPILE     (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \
1195        PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED)
1196    
1197    #define PUBLIC_JIT_EXEC_OPTIONS \
1198       (PCRE_NO_UTF8_CHECK|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|\
1199        PCRE_NOTEMPTY_ATSTART|PCRE_PARTIAL_SOFT|PCRE_PARTIAL_HARD)
1200    
1201  /* Magic number to provide a small check against being handed junk. */  /* Magic number to provide a small check against being handed junk. */
1202    
# Line 898  in different endianness. */ Line 1207  in different endianness. */
1207    
1208  #define REVERSED_MAGIC_NUMBER  0x45524350UL   /* 'ERCP' */  #define REVERSED_MAGIC_NUMBER  0x45524350UL   /* 'ERCP' */
1209    
 /* Negative values for the firstchar and reqchar variables */  
   
 #define REQ_UNSET (-2)  
 #define REQ_NONE  (-1)  
   
1210  /* The maximum remaining length of subject we are prepared to search for a  /* The maximum remaining length of subject we are prepared to search for a
1211  req_byte match. */  req_byte match. */
1212    
# Line 939  macros to give the functions distinct na Line 1243  macros to give the functions distinct na
1243  #ifndef SUPPORT_UTF  #ifndef SUPPORT_UTF
1244    
1245  /* UTF-8 support is not enabled; use the platform-dependent character literals  /* UTF-8 support is not enabled; use the platform-dependent character literals
1246  so that PCRE works on both ASCII and EBCDIC platforms, in non-UTF-mode only. */  so that PCRE works in both ASCII and EBCDIC environments, but only in non-UTF
1247    mode. Newline characters are problematic in EBCDIC. Though it has CR and LF
1248    characters, a common practice has been to use its NL (0x15) character as the
1249    line terminator in C-like processing environments. However, sometimes the LF
1250    (0x25) character is used instead, according to this Unicode document:
1251    
1252    http://unicode.org/standard/reports/tr13/tr13-5.html
1253    
1254    PCRE defaults EBCDIC NL to 0x15, but has a build-time option to select 0x25
1255    instead. Whichever is *not* chosen is defined as NEL.
1256    
1257    In both ASCII and EBCDIC environments, CHAR_NL and CHAR_LF are synonyms for the
1258    same code point. */
1259    
1260    #ifdef EBCDIC
1261    
1262    #ifndef EBCDIC_NL25
1263    #define CHAR_NL                     '\x15'
1264    #define CHAR_NEL                    '\x25'
1265    #define STR_NL                      "\x15"
1266    #define STR_NEL                     "\x25"
1267    #else
1268    #define CHAR_NL                     '\x25'
1269    #define CHAR_NEL                    '\x15'
1270    #define STR_NL                      "\x25"
1271    #define STR_NEL                     "\x15"
1272    #endif
1273    
1274    #define CHAR_LF                     CHAR_NL
1275    #define STR_LF                      STR_NL
1276    
1277    #define CHAR_ESC                    '\047'
1278    #define CHAR_DEL                    '\007'
1279    #define STR_ESC                     "\047"
1280    #define STR_DEL                     "\007"
1281    
1282    #else  /* Not EBCDIC */
1283    
1284    /* In ASCII/Unicode, linefeed is '\n' and we equate this to NL for
1285    compatibility. NEL is the Unicode newline character; make sure it is
1286    a positive value. */
1287    
1288    #define CHAR_LF                     '\n'
1289    #define CHAR_NL                     CHAR_LF
1290    #define CHAR_NEL                    ((unsigned char)'\x85')
1291    #define CHAR_ESC                    '\033'
1292    #define CHAR_DEL                    '\177'
1293    
1294    #define STR_LF                      "\n"
1295    #define STR_NL                      STR_LF
1296    #define STR_NEL                     "\x85"
1297    #define STR_ESC                     "\033"
1298    #define STR_DEL                     "\177"
1299    
1300    #endif  /* EBCDIC */
1301    
1302    /* The remaining definitions work in both environments. */
1303    
1304    #define CHAR_NULL                   '\0'
1305  #define CHAR_HT                     '\t'  #define CHAR_HT                     '\t'
1306  #define CHAR_VT                     '\v'  #define CHAR_VT                     '\v'
1307  #define CHAR_FF                     '\f'  #define CHAR_FF                     '\f'
1308  #define CHAR_CR                     '\r'  #define CHAR_CR                     '\r'
 #define CHAR_NL                     '\n'  
1309  #define CHAR_BS                     '\b'  #define CHAR_BS                     '\b'
1310  #define CHAR_BEL                    '\a'  #define CHAR_BEL                    '\a'
 #ifdef EBCDIC  
 #define CHAR_ESC                    '\047'  
 #define CHAR_DEL                    '\007'  
 #else  
 #define CHAR_ESC                    '\033'  
 #define CHAR_DEL                    '\177'  
 #endif  
1311    
1312  #define CHAR_SPACE                  ' '  #define CHAR_SPACE                  ' '
1313  #define CHAR_EXCLAMATION_MARK       '!'  #define CHAR_EXCLAMATION_MARK       '!'
# Line 1056  so that PCRE works on both ASCII and EBC Line 1409  so that PCRE works on both ASCII and EBC
1409  #define STR_VT                      "\v"  #define STR_VT                      "\v"
1410  #define STR_FF                      "\f"  #define STR_FF                      "\f"
1411  #define STR_CR                      "\r"  #define STR_CR                      "\r"
 #define STR_NL                      "\n"  
1412  #define STR_BS                      "\b"  #define STR_BS                      "\b"
1413  #define STR_BEL                     "\a"  #define STR_BEL                     "\a"
 #ifdef EBCDIC  
 #define STR_ESC                     "\047"  
 #define STR_DEL                     "\007"  
 #else  
 #define STR_ESC                     "\033"  
 #define STR_DEL                     "\177"  
 #endif  
1414    
1415  #define STR_SPACE                   " "  #define STR_SPACE                   " "
1416  #define STR_EXCLAMATION_MARK        "!"  #define STR_EXCLAMATION_MARK        "!"
# Line 1189  so that PCRE works on both ASCII and EBC Line 1534  so that PCRE works on both ASCII and EBC
1534    
1535  #define STRING_DEFINE               "DEFINE"  #define STRING_DEFINE               "DEFINE"
1536    
1537  #define STRING_CR_RIGHTPAR             "CR)"  #define STRING_CR_RIGHTPAR              "CR)"
1538  #define STRING_LF_RIGHTPAR             "LF)"  #define STRING_LF_RIGHTPAR              "LF)"
1539  #define STRING_CRLF_RIGHTPAR           "CRLF)"  #define STRING_CRLF_RIGHTPAR            "CRLF)"
1540  #define STRING_ANY_RIGHTPAR            "ANY)"  #define STRING_ANY_RIGHTPAR             "ANY)"
1541  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"  #define STRING_ANYCRLF_RIGHTPAR         "ANYCRLF)"
1542  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"  #define STRING_BSR_ANYCRLF_RIGHTPAR     "BSR_ANYCRLF)"
1543  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"  #define STRING_BSR_UNICODE_RIGHTPAR     "BSR_UNICODE)"
1544  #ifdef COMPILE_PCRE8  #define STRING_UTF8_RIGHTPAR            "UTF8)"
1545  #define STRING_UTF_RIGHTPAR            "UTF8)"  #define STRING_UTF16_RIGHTPAR           "UTF16)"
1546  #endif  #define STRING_UTF32_RIGHTPAR           "UTF32)"
1547  #ifdef COMPILE_PCRE16  #define STRING_UTF_RIGHTPAR             "UTF)"
1548  #define STRING_UTF_RIGHTPAR            "UTF16)"  #define STRING_UCP_RIGHTPAR             "UCP)"
1549  #endif  #define STRING_NO_AUTO_POSSESS_RIGHTPAR "NO_AUTO_POSSESS)"
1550  #define STRING_UCP_RIGHTPAR            "UCP)"  #define STRING_NO_START_OPT_RIGHTPAR    "NO_START_OPT)"
1551  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"  #define STRING_LIMIT_MATCH_EQ           "LIMIT_MATCH="
1552    #define STRING_LIMIT_RECURSION_EQ       "LIMIT_RECURSION="
1553    
1554  #else  /* SUPPORT_UTF */  #else  /* SUPPORT_UTF */
1555    
# Line 1215  only. */ Line 1561  only. */
1561  #define CHAR_VT                     '\013'  #define CHAR_VT                     '\013'
1562  #define CHAR_FF                     '\014'  #define CHAR_FF                     '\014'
1563  #define CHAR_CR                     '\015'  #define CHAR_CR                     '\015'
1564  #define CHAR_NL                     '\012'  #define CHAR_LF                     '\012'
1565    #define CHAR_NL                     CHAR_LF
1566    #define CHAR_NEL                    ((unsigned char)'\x85')
1567  #define CHAR_BS                     '\010'  #define CHAR_BS                     '\010'
1568  #define CHAR_BEL                    '\007'  #define CHAR_BEL                    '\007'
1569  #define CHAR_ESC                    '\033'  #define CHAR_ESC                    '\033'
1570  #define CHAR_DEL                    '\177'  #define CHAR_DEL                    '\177'
1571    
1572    #define CHAR_NULL                   '\0'
1573  #define CHAR_SPACE                  '\040'  #define CHAR_SPACE                  '\040'
1574  #define CHAR_EXCLAMATION_MARK       '\041'  #define CHAR_EXCLAMATION_MARK       '\041'
1575  #define CHAR_QUOTATION_MARK         '\042'  #define CHAR_QUOTATION_MARK         '\042'
# Line 1448  only. */ Line 1797  only. */
1797  #define STRING_xdigit               STR_x STR_d STR_i STR_g STR_i STR_t  #define STRING_xdigit               STR_x STR_d STR_i STR_g STR_i STR_t
1798    
1799  #define STRING_DEFINE               STR_D STR_E STR_F STR_I STR_N STR_E  #define STRING_DEFINE               STR_D STR_E STR_F STR_I STR_N STR_E
1800    #define STRING_WEIRD_STARTWORD      STR_LEFT_SQUARE_BRACKET STR_COLON STR_LESS_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1801    #define STRING_WEIRD_ENDWORD        STR_LEFT_SQUARE_BRACKET STR_COLON STR_GREATER_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1802    
1803  #define STRING_CR_RIGHTPAR             STR_C STR_R STR_RIGHT_PARENTHESIS  #define STRING_CR_RIGHTPAR              STR_C STR_R STR_RIGHT_PARENTHESIS
1804  #define STRING_LF_RIGHTPAR             STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_LF_RIGHTPAR              STR_L STR_F STR_RIGHT_PARENTHESIS
1805  #define STRING_CRLF_RIGHTPAR           STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_CRLF_RIGHTPAR            STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1806  #define STRING_ANY_RIGHTPAR            STR_A STR_N STR_Y STR_RIGHT_PARENTHESIS  #define STRING_ANY_RIGHTPAR             STR_A STR_N STR_Y STR_RIGHT_PARENTHESIS
1807  #define STRING_ANYCRLF_RIGHTPAR        STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_ANYCRLF_RIGHTPAR         STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1808  #define STRING_BSR_ANYCRLF_RIGHTPAR    STR_B STR_S STR_R STR_UNDERSCORE STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_BSR_ANYCRLF_RIGHTPAR     STR_B STR_S STR_R STR_UNDERSCORE STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1809  #define STRING_BSR_UNICODE_RIGHTPAR    STR_B STR_S STR_R STR_UNDERSCORE STR_U STR_N STR_I STR_C STR_O STR_D STR_E STR_RIGHT_PARENTHESIS  #define STRING_BSR_UNICODE_RIGHTPAR     STR_B STR_S STR_R STR_UNDERSCORE STR_U STR_N STR_I STR_C STR_O STR_D STR_E STR_RIGHT_PARENTHESIS
1810  #ifdef COMPILE_PCRE8  #define STRING_UTF8_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
1811  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS  #define STRING_UTF16_RIGHTPAR           STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS
1812  #endif  #define STRING_UTF32_RIGHTPAR           STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
1813  #ifdef COMPILE_PCRE16  #define STRING_UTF_RIGHTPAR             STR_U STR_T STR_F STR_RIGHT_PARENTHESIS
1814  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS  #define STRING_UCP_RIGHTPAR             STR_U STR_C STR_P STR_RIGHT_PARENTHESIS
1815  #endif  #define STRING_NO_AUTO_POSSESS_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_A STR_U STR_T STR_O STR_UNDERSCORE STR_P STR_O STR_S STR_S STR_E STR_S STR_S STR_RIGHT_PARENTHESIS
1816  #define STRING_UCP_RIGHTPAR            STR_U STR_C STR_P STR_RIGHT_PARENTHESIS  #define STRING_NO_START_OPT_RIGHTPAR    STR_N STR_O STR_UNDERSCORE STR_S STR_T STR_A STR_R STR_T STR_UNDERSCORE STR_O STR_P STR_T STR_RIGHT_PARENTHESIS
1817  #define STRING_NO_START_OPT_RIGHTPAR   STR_N STR_O STR_UNDERSCORE STR_S STR_T STR_A STR_R STR_T STR_UNDERSCORE STR_O STR_P STR_T STR_RIGHT_PARENTHESIS  #define STRING_LIMIT_MATCH_EQ           STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_M STR_A STR_T STR_C STR_H STR_EQUALS_SIGN
1818    #define STRING_LIMIT_RECURSION_EQ       STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_R STR_E STR_C STR_U STR_R STR_S STR_I STR_O STR_N STR_EQUALS_SIGN
1819    
1820  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
1821    
# Line 1478  only. */ Line 1830  only. */
1830  #endif  #endif
1831    
1832  #ifndef ESC_n  #ifndef ESC_n
1833  #define ESC_n CHAR_NL  #define ESC_n CHAR_LF
1834  #endif  #endif
1835    
1836  #ifndef ESC_r  #ifndef ESC_r
# Line 1503  only. */ Line 1855  only. */
1855  #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */  #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */
1856  #define PT_PXSPACE    7    /* POSIX space - Z plus 9,10,11,12,13 */  #define PT_PXSPACE    7    /* POSIX space - Z plus 9,10,11,12,13 */
1857  #define PT_WORD       8    /* Word - L plus N plus underscore */  #define PT_WORD       8    /* Word - L plus N plus underscore */
1858    #define PT_CLIST      9    /* Pseudo-property: match character list */
1859    #define PT_UCNC      10    /* Universal Character nameable character */
1860    #define PT_TABSIZE   11    /* Size of square table for autopossessify tests */
1861    
1862    /* The following special properties are used only in XCLASS items, when POSIX
1863    classes are specified and PCRE_UCP is set - in other words, for Unicode
1864    handling of these classes. They are not available via the \p or \P escapes like
1865    those in the above list, and so they do not take part in the autopossessifying
1866    table. */
1867    
1868    #define PT_PXGRAPH   11    /* [:graph:] - characters that mark the paper */
1869    #define PT_PXPRINT   12    /* [:print:] - [:graph:] plus non-control spaces */
1870    #define PT_PXPUNCT   13    /* [:punct:] - punctuation characters */
1871    
1872  /* Flag bits and data types for the extended class (OP_XCLASS) for classes that  /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
1873  contain characters with values greater than 255. */  contain characters with values greater than 255. */
# Line 1517  contain characters with values greater t Line 1882  contain characters with values greater t
1882  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */
1883    
1884  /* These are escaped items that aren't just an encoding of a particular data  /* These are escaped items that aren't just an encoding of a particular data
1885  value such as \n. They must have non-zero values, as check_escape() returns  value such as \n. They must have non-zero values, as check_escape() returns 0
1886  their negation. Also, they must appear in the same order as in the opcode  for a data character.  Also, they must appear in the same order as in the
1887  definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it  opcode definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it
1888  corresponds to "." in DOTALL mode rather than an escape sequence. It is also  corresponds to "." in DOTALL mode rather than an escape sequence. It is also
1889  used for [^] in JavaScript compatibility mode, and for \C in non-utf mode. In  used for [^] in JavaScript compatibility mode, and for \C in non-utf mode. In
1890  non-DOTALL mode, "." behaves like \N.  non-DOTALL mode, "." behaves like \N.
1891    
1892  The special values ESC_DU, ESC_du, etc. are used instead of ESC_D, ESC_d, etc.  The special values ESC_DU, ESC_du, etc. are used instead of ESC_D, ESC_d, etc.
1893  when PCRE_UCP is set, when replacement of \d etc by \p sequences is required.  when PCRE_UCP is set and replacement of \d etc by \p sequences is required.
1894  They must be contiguous, and remain in order so that the replacements can be  They must be contiguous, and remain in order so that the replacements can be
1895  looked up from a table.  looked up from a table.
1896    
1897  The final escape must be ESC_REF as subsequent values are used for  Negative numbers are used to encode a backreference (\1, \2, \3, etc.) in
1898  backreferences (\1, \2, \3, etc). There are two tests in the code for an escape  check_escape(). There are two tests in the code for an escape
1899  greater than ESC_b and less than ESC_Z to detect the types that may be  greater than ESC_b and less than ESC_Z to detect the types that may be
1900  repeated. These are the types that consume characters. If any new escapes are  repeated. These are the types that consume characters. If any new escapes are
1901  put in between that don't consume a character, that code will have to change.  put in between that don't consume a character, that code will have to change.
# Line 1540  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E Line 1905  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E
1905         ESC_W, ESC_w, ESC_N, ESC_dum, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H,         ESC_W, ESC_w, ESC_N, ESC_dum, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H,
1906         ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z,         ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z,
1907         ESC_E, ESC_Q, ESC_g, ESC_k,         ESC_E, ESC_Q, ESC_g, ESC_k,
1908         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu,         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu };
1909         ESC_REF };  
1910    
1911  /* Opcode table: Starting from 1 (i.e. after OP_END), the values up to  /********************** Opcode definitions ******************/
 OP_EOD must correspond in order to the list of escapes immediately above.  
1912    
1913  *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions  /****** NOTE NOTE NOTE ******
1914  that follow must also be updated to match. There are also tables called  
1915  "coptable" and "poptable" in pcre_dfa_exec.c that must be updated. */  Starting from 1 (i.e. after OP_END), the values up to OP_EOD must correspond in
1916    order to the list of escapes immediately above. Furthermore, values up to
1917    OP_DOLLM must not be changed without adjusting the table called autoposstab in
1918    pcre_compile.c
1919    
1920    Whenever this list is updated, the two macro definitions that follow must be
1921    updated to match. The possessification table called "opcode_possessify" in
1922    pcre_compile.c must also be updated, and also the tables called "coptable"
1923    and "poptable" in pcre_dfa_exec.c.
1924    
1925    ****** NOTE NOTE NOTE ******/
1926    
1927    
1928    /* The values between FIRST_AUTOTAB_OP and LAST_AUTOTAB_RIGHT_OP, inclusive,
1929    are used in a table for deciding whether a repeated character type can be
1930    auto-possessified. */
1931    
1932    #define FIRST_AUTOTAB_OP       OP_NOT_DIGIT
1933    #define LAST_AUTOTAB_LEFT_OP   OP_EXTUNI
1934    #define LAST_AUTOTAB_RIGHT_OP  OP_DOLLM
1935    
1936  enum {  enum {
1937    OP_END,            /* 0 End of pattern */    OP_END,            /* 0 End of pattern */
# Line 1567  enum { Line 1950  enum {
1950    OP_NOT_WORDCHAR,       /* 10 \W */    OP_NOT_WORDCHAR,       /* 10 \W */
1951    OP_WORDCHAR,           /* 11 \w */    OP_WORDCHAR,           /* 11 \w */
1952    
1953    OP_ANY,            /* 12 Match any character except newline */    OP_ANY,            /* 12 Match any character except newline (\N) */
1954    OP_ALLANY,         /* 13 Match any character */    OP_ALLANY,         /* 13 Match any character */
1955    OP_ANYBYTE,        /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */    OP_ANYBYTE,        /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */
1956    OP_NOTPROP,        /* 15 \P (not Unicode property) */    OP_NOTPROP,        /* 15 \P (not Unicode property) */
# Line 1578  enum { Line 1961  enum {
1961    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */
1962    OP_VSPACE,         /* 21 \v (vertical whitespace) */    OP_VSPACE,         /* 21 \v (vertical whitespace) */
1963    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */
1964    OP_EODN,           /* 23 End of data or \n at end of data: \Z. */    OP_EODN,           /* 23 End of data or \n at end of data (\Z) */
1965    OP_EOD,            /* 24 End of data: \z */    OP_EOD,            /* 24 End of data (\z) */
1966    
1967      /* Line end assertions */
1968    
1969      OP_DOLL,           /* 25 End of line - not multiline */
1970      OP_DOLLM,          /* 26 End of line - multiline */
1971      OP_CIRC,           /* 27 Start of line - not multiline */
1972      OP_CIRCM,          /* 28 Start of line - multiline */
1973    
1974      /* Single characters; caseful must precede the caseless ones */
1975    
   OP_CIRC,           /* 25 Start of line - not multiline */  
   OP_CIRCM,          /* 26 Start of line - multiline */  
   OP_DOLL,           /* 27 End of line - not multiline */  
   OP_DOLLM,          /* 28 End of line - multiline */  
1976    OP_CHAR,           /* 29 Match one character, casefully */    OP_CHAR,           /* 29 Match one character, casefully */
1977    OP_CHARI,          /* 30 Match one character, caselessly */    OP_CHARI,          /* 30 Match one character, caselessly */
1978    OP_NOT,            /* 31 Match one character, not the given one, casefully */    OP_NOT,            /* 31 Match one character, not the given one, casefully */
# Line 1593  enum { Line 1981  enum {
1981    /* The following sets of 13 opcodes must always be kept in step because    /* The following sets of 13 opcodes must always be kept in step because
1982    the offset from the first one is used to generate the others. */    the offset from the first one is used to generate the others. */
1983    
1984    /**** Single characters, caseful, must precede the caseless ones ****/    /* Repeated characters; caseful must precede the caseless ones */
1985    
1986    OP_STAR,           /* 33 The maximizing and minimizing versions of */    OP_STAR,           /* 33 The maximizing and minimizing versions of */
1987    OP_MINSTAR,        /* 34 these six opcodes must come in pairs, with */    OP_MINSTAR,        /* 34 these six opcodes must come in pairs, with */
# Line 1611  enum { Line 1999  enum {
1999    OP_POSQUERY,       /* 44 Posesssified query, caseful */    OP_POSQUERY,       /* 44 Posesssified query, caseful */
2000    OP_POSUPTO,        /* 45 Possessified upto, caseful */    OP_POSUPTO,        /* 45 Possessified upto, caseful */
2001    
2002    /**** Single characters, caseless, must follow the caseful ones */    /* Repeated characters; caseless must follow the caseful ones */
2003    
2004    OP_STARI,          /* 46 */    OP_STARI,          /* 46 */
2005    OP_MINSTARI,       /* 47 */    OP_MINSTARI,       /* 47 */
# Line 1629  enum { Line 2017  enum {
2017    OP_POSQUERYI,      /* 57 Posesssified query, caseless */    OP_POSQUERYI,      /* 57 Posesssified query, caseless */
2018    OP_POSUPTOI,       /* 58 Possessified upto, caseless */    OP_POSUPTOI,       /* 58 Possessified upto, caseless */
2019    
2020    /**** The negated ones must follow the non-negated ones, and match them ****/    /* The negated ones must follow the non-negated ones, and match them */
2021    /**** Negated single character, caseful; must precede the caseless ones ****/    /* Negated repeated character, caseful; must precede the caseless ones */
2022    
2023    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */
2024    OP_NOTMINSTAR,     /* 60 these six opcodes must come in pairs, with */    OP_NOTMINSTAR,     /* 60 these six opcodes must come in pairs, with */
# Line 1648  enum { Line 2036  enum {
2036    OP_NOTPOSQUERY,    /* 70 */    OP_NOTPOSQUERY,    /* 70 */
2037    OP_NOTPOSUPTO,     /* 71 */    OP_NOTPOSUPTO,     /* 71 */
2038    
2039    /**** Negated single character, caseless; must follow the caseful ones ****/    /* Negated repeated character, caseless; must follow the caseful ones */
2040    
2041    OP_NOTSTARI,       /* 72 */    OP_NOTSTARI,       /* 72 */
2042    OP_NOTMINSTARI,    /* 73 */    OP_NOTMINSTARI,    /* 73 */
# Line 1666  enum { Line 2054  enum {
2054    OP_NOTPOSQUERYI,   /* 83 */    OP_NOTPOSQUERYI,   /* 83 */
2055    OP_NOTPOSUPTOI,    /* 84 */    OP_NOTPOSUPTOI,    /* 84 */
2056    
2057    /**** Character types ****/    /* Character types */
2058    
2059    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */
2060    OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */    OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */
# Line 1697  enum { Line 2085  enum {
2085    OP_CRRANGE,        /* 104 These are different to the three sets above. */    OP_CRRANGE,        /* 104 These are different to the three sets above. */
2086    OP_CRMINRANGE,     /* 105 */    OP_CRMINRANGE,     /* 105 */
2087    
2088      OP_CRPOSSTAR,      /* 106 Possessified versions */
2089      OP_CRPOSPLUS,      /* 107 */
2090      OP_CRPOSQUERY,     /* 108 */
2091      OP_CRPOSRANGE,     /* 109 */
2092    
2093    /* End of quantifier opcodes */    /* End of quantifier opcodes */
2094    
2095    OP_CLASS,          /* 106 Match a character class, chars < 256 only */    OP_CLASS,          /* 110 Match a character class, chars < 256 only */
2096    OP_NCLASS,         /* 107 Same, but the bitmap was created from a negative    OP_NCLASS,         /* 111 Same, but the bitmap was created from a negative
2097                                class - the difference is relevant only when a                                class - the difference is relevant only when a
2098                                character > 255 is encountered. */                                character > 255 is encountered. */
2099    OP_XCLASS,         /* 108 Extended class for handling > 255 chars within the    OP_XCLASS,         /* 112 Extended class for handling > 255 chars within the
2100                                class. This does both positive and negative. */                                class. This does both positive and negative. */
2101    OP_REF,            /* 109 Match a back reference, casefully */    OP_REF,            /* 113 Match a back reference, casefully */
2102    OP_REFI,           /* 110 Match a back reference, caselessly */    OP_REFI,           /* 114 Match a back reference, caselessly */
2103    OP_RECURSE,        /* 111 Match a numbered subpattern (possibly recursive) */    OP_DNREF,          /* 115 Match a duplicate name backref, casefully */
2104    OP_CALLOUT,        /* 112 Call out to external function if provided */    OP_DNREFI,         /* 116 Match a duplicate name backref, caselessly */
2105      OP_RECURSE,        /* 117 Match a numbered subpattern (possibly recursive) */
2106    OP_ALT,            /* 113 Start of alternation */    OP_CALLOUT,        /* 118 Call out to external function if provided */
2107    OP_KET,            /* 114 End of group that doesn't have an unbounded repeat */  
2108    OP_KETRMAX,        /* 115 These two must remain together and in this */    OP_ALT,            /* 119 Start of alternation */
2109    OP_KETRMIN,        /* 116 order. They are for groups the repeat for ever. */    OP_KET,            /* 120 End of group that doesn't have an unbounded repeat */
2110    OP_KETRPOS,        /* 117 Possessive unlimited repeat. */    OP_KETRMAX,        /* 121 These two must remain together and in this */
2111      OP_KETRMIN,        /* 122 order. They are for groups the repeat for ever. */
2112      OP_KETRPOS,        /* 123 Possessive unlimited repeat. */
2113    
2114    /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four    /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four
2115    asserts must remain in order. */    asserts must remain in order. */
2116    
2117    OP_REVERSE,        /* 118 Move pointer back - used in lookbehind assertions */    OP_REVERSE,        /* 124 Move pointer back - used in lookbehind assertions */
2118    OP_ASSERT,         /* 119 Positive lookahead */    OP_ASSERT,         /* 125 Positive lookahead */
2119    OP_ASSERT_NOT,     /* 120 Negative lookahead */    OP_ASSERT_NOT,     /* 126 Negative lookahead */
2120    OP_ASSERTBACK,     /* 121 Positive lookbehind */    OP_ASSERTBACK,     /* 127 Positive lookbehind */
2121    OP_ASSERTBACK_NOT, /* 122 Negative lookbehind */    OP_ASSERTBACK_NOT, /* 128 Negative lookbehind */
2122    
2123    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately
2124    after the assertions, with ONCE first, as there's a test for >= ONCE for a    after the assertions, with ONCE first, as there's a test for >= ONCE for a
2125    subpattern that isn't an assertion. The POS versions must immediately follow    subpattern that isn't an assertion. The POS versions must immediately follow
2126    the non-POS versions in each case. */    the non-POS versions in each case. */
2127    
2128    OP_ONCE,           /* 123 Atomic group, contains captures */    OP_ONCE,           /* 129 Atomic group, contains captures */
2129    OP_ONCE_NC,        /* 124 Atomic group containing no captures */    OP_ONCE_NC,        /* 130 Atomic group containing no captures */
2130    OP_BRA,            /* 125 Start of non-capturing bracket */    OP_BRA,            /* 131 Start of non-capturing bracket */
2131    OP_BRAPOS,         /* 126 Ditto, with unlimited, possessive repeat */    OP_BRAPOS,         /* 132 Ditto, with unlimited, possessive repeat */
2132    OP_CBRA,           /* 127 Start of capturing bracket */    OP_CBRA,           /* 133 Start of capturing bracket */
2133    OP_CBRAPOS,        /* 128 Ditto, with unlimited, possessive repeat */    OP_CBRAPOS,        /* 134 Ditto, with unlimited, possessive repeat */
2134    OP_COND,           /* 129 Conditional group */    OP_COND,           /* 135 Conditional group */
2135    
2136    /* These five must follow the previous five, in the same order. There's a    /* These five must follow the previous five, in the same order. There's a
2137    check for >= SBRA to distinguish the two sets. */    check for >= SBRA to distinguish the two sets. */
2138    
2139    OP_SBRA,           /* 130 Start of non-capturing bracket, check empty  */    OP_SBRA,           /* 136 Start of non-capturing bracket, check empty  */
2140    OP_SBRAPOS,        /* 131 Ditto, with unlimited, possessive repeat */    OP_SBRAPOS,        /* 137 Ditto, with unlimited, possessive repeat */
2141    OP_SCBRA,          /* 132 Start of capturing bracket, check empty */    OP_SCBRA,          /* 138 Start of capturing bracket, check empty */
2142    OP_SCBRAPOS,       /* 133 Ditto, with unlimited, possessive repeat */    OP_SCBRAPOS,       /* 139 Ditto, with unlimited, possessive repeat */
2143    OP_SCOND,          /* 134 Conditional group, check empty */    OP_SCOND,          /* 140 Conditional group, check empty */
2144    
2145    /* The next two pairs must (respectively) be kept together. */    /* The next two pairs must (respectively) be kept together. */
2146    
2147    OP_CREF,           /* 135 Used to hold a capture number as condition */    OP_CREF,           /* 141 Used to hold a capture number as condition */
2148    OP_NCREF,          /* 136 Same, but generated by a name reference*/    OP_DNCREF,         /* 142 Used to point to duplicate names as a condition */
2149    OP_RREF,           /* 137 Used to hold a recursion number as condition */    OP_RREF,           /* 143 Used to hold a recursion number as condition */
2150    OP_NRREF,          /* 138 Same, but generated by a name reference*/    OP_DNRREF,         /* 144 Used to point to duplicate names as a condition */
2151    OP_DEF,            /* 139 The DEFINE condition */    OP_DEF,            /* 145 The DEFINE condition */
2152    
2153    OP_BRAZERO,        /* 140 These two must remain together and in this */    OP_BRAZERO,        /* 146 These two must remain together and in this */
2154    OP_BRAMINZERO,     /* 141 order. */    OP_BRAMINZERO,     /* 147 order. */
2155    OP_BRAPOSZERO,     /* 142 */    OP_BRAPOSZERO,     /* 148 */
2156    
2157    /* These are backtracking control verbs */    /* These are backtracking control verbs */
2158    
2159    OP_MARK,           /* 143 always has an argument */    OP_MARK,           /* 149 always has an argument */
2160    OP_PRUNE,          /* 144 */    OP_PRUNE,          /* 150 */
2161    OP_PRUNE_ARG,      /* 145 same, but with argument */    OP_PRUNE_ARG,      /* 151 same, but with argument */
2162    OP_SKIP,           /* 146 */    OP_SKIP,           /* 152 */
2163    OP_SKIP_ARG,       /* 147 same, but with argument */    OP_SKIP_ARG,       /* 153 same, but with argument */
2164    OP_THEN,           /* 148 */    OP_THEN,           /* 154 */
2165    OP_THEN_ARG,       /* 149 same, but with argument */    OP_THEN_ARG,       /* 155 same, but with argument */
2166    OP_COMMIT,         /* 150 */    OP_COMMIT,         /* 156 */
2167    
2168    /* These are forced failure and success verbs */    /* These are forced failure and success verbs */
2169    
2170    OP_FAIL,           /* 151 */    OP_FAIL,           /* 157 */
2171    OP_ACCEPT,         /* 152 */    OP_ACCEPT,         /* 158 */
2172    OP_ASSERT_ACCEPT,  /* 153 Used inside assertions */    OP_ASSERT_ACCEPT,  /* 159 Used inside assertions */
2173    OP_CLOSE,          /* 154 Used before OP_ACCEPT to close open captures */    OP_CLOSE,          /* 160 Used before OP_ACCEPT to close open captures */
2174    
2175    /* This is used to skip a subpattern with a {0} quantifier */    /* This is used to skip a subpattern with a {0} quantifier */
2176    
2177    OP_SKIPZERO,       /* 155 */    OP_SKIPZERO,       /* 161 */
2178    
2179    /* This is not an opcode, but is used to check that tables indexed by opcode    /* This is not an opcode, but is used to check that tables indexed by opcode
2180    are the correct length, in order to catch updating errors - there have been    are the correct length, in order to catch updating errors - there have been
# Line 1790  enum { Line 2185  enum {
2185    
2186  /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro  /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro
2187  definitions that follow must also be updated to match. There are also tables  definitions that follow must also be updated to match. There are also tables
2188  called "coptable" and "poptable" in pcre_dfa_exec.c that must be updated. */  called "opcode_possessify" in pcre_compile.c and "coptable" and "poptable" in
2189    pcre_dfa_exec.c that must be updated. */
2190    
2191    
2192  /* This macro defines textual names for all the opcodes. These are used only  /* This macro defines textual names for all the opcodes. These are used only
# Line 1803  some cases doesn't actually use these na Line 2199  some cases doesn't actually use these na
2199    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \
2200    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \
2201    "extuni",  "\\Z", "\\z",                                        \    "extuni",  "\\Z", "\\z",                                        \
2202    "^", "^", "$", "$", "char", "chari", "not", "noti",             \    "$", "$", "^", "^", "char", "chari", "not", "noti",             \
2203    "*", "*?", "+", "+?", "?", "??",                                \    "*", "*?", "+", "+?", "?", "??",                                \
2204    "{", "{", "{",                                                  \    "{", "{", "{",                                                  \
2205    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
# Line 1819  some cases doesn't actually use these na Line 2215  some cases doesn't actually use these na
2215    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
2216    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2217    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \
2218    "class", "nclass", "xclass", "Ref", "Refi",                     \    "*+","++", "?+", "{",                                           \
2219      "class", "nclass", "xclass", "Ref", "Refi", "DnRef", "DnRefi",  \
2220    "Recurse", "Callout",                                           \    "Recurse", "Callout",                                           \
2221    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \
2222    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \
# Line 1828  some cases doesn't actually use these na Line 2225  some cases doesn't actually use these na
2225    "Cond",                                                         \    "Cond",                                                         \
2226    "SBra", "SBraPos", "SCBra", "SCBraPos",                         \    "SBra", "SBraPos", "SCBra", "SCBraPos",                         \
2227    "SCond",                                                        \    "SCond",                                                        \
2228    "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def",   \    "Cond ref", "Cond dnref", "Cond rec", "Cond dnrec", "Cond def", \
2229    "Brazero", "Braminzero", "Braposzero",                          \    "Brazero", "Braminzero", "Braposzero",                          \
2230    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \
2231    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \
# Line 1853  in UTF-8 mode. The code that uses this t Line 2250  in UTF-8 mode. The code that uses this t
2250    3, 3,                          /* \P, \p                                 */ \    3, 3,                          /* \P, \p                                 */ \
2251    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \
2252    1,                             /* \X                                     */ \    1,                             /* \X                                     */ \
2253    1, 1, 1, 1, 1, 1,              /* \Z, \z, ^, ^M, $, $M                   */ \    1, 1, 1, 1, 1, 1,              /* \Z, \z, $, $M ^, ^M                    */ \
2254    2,                             /* Char  - the minimum length             */ \    2,                             /* Char  - the minimum length             */ \
2255    2,                             /* Chari  - the minimum length            */ \    2,                             /* Chari  - the minimum length            */ \
2256    2,                             /* not                                    */ \    2,                             /* not                                    */ \
# Line 1884  in UTF-8 mode. The code that uses this t Line 2281  in UTF-8 mode. The code that uses this t
2281    /* Character class & ref repeats                                         */ \    /* Character class & ref repeats                                         */ \
2282    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \
2283    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \
2284      1, 1, 1, 1+2*IMM2_SIZE,        /* Possessive *+, ++, ?+, CRPOSRANGE      */ \
2285    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \
2286    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \
2287    0,                             /* XCLASS - variable length               */ \    0,                             /* XCLASS - variable length               */ \
2288    1+IMM2_SIZE,                   /* REF                                    */ \    1+IMM2_SIZE,                   /* REF                                    */ \
2289    1+IMM2_SIZE,                   /* REFI                                   */ \    1+IMM2_SIZE,                   /* REFI                                   */ \
2290      1+2*IMM2_SIZE,                 /* DNREF                                  */ \
2291      1+2*IMM2_SIZE,                 /* DNREFI                                 */ \
2292    1+LINK_SIZE,                   /* RECURSE                                */ \    1+LINK_SIZE,                   /* RECURSE                                */ \
2293    2+2*LINK_SIZE,                 /* CALLOUT                                */ \    2+2*LINK_SIZE,                 /* CALLOUT                                */ \
2294    1+LINK_SIZE,                   /* Alt                                    */ \    1+LINK_SIZE,                   /* Alt                                    */ \
# Line 1913  in UTF-8 mode. The code that uses this t Line 2313  in UTF-8 mode. The code that uses this t
2313    1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \    1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \
2314    1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \    1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \
2315    1+LINK_SIZE,                   /* SCOND                                  */ \    1+LINK_SIZE,                   /* SCOND                                  */ \
2316    1+IMM2_SIZE, 1+IMM2_SIZE,      /* CREF, NCREF                            */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* CREF, DNCREF                           */ \
2317    1+IMM2_SIZE, 1+IMM2_SIZE,      /* RREF, NRREF                            */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* RREF, DNRREF                           */ \
2318    1,                             /* DEF                                    */ \    1,                             /* DEF                                    */ \
2319    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \
2320    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \
# Line 1923  in UTF-8 mode. The code that uses this t Line 2323  in UTF-8 mode. The code that uses this t
2323    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \
2324    1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */    1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */
2325    
2326  /* A magic value for OP_RREF and OP_NRREF to indicate the "any recursion"  /* A magic value for OP_RREF to indicate the "any recursion" condition. */
 condition. */  
2327    
2328  #define RREF_ANY  0xffff  #define RREF_ANY  0xffff
2329    
# Line 1939  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 2338  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
2338         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
2339         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
2340         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
2341         ERR70, ERR71, ERR72, ERR73, ERR74, ERRCOUNT };         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERR78, ERR79,
2342           ERR80, ERR81, ERR82, ERR83, ERR84, ERRCOUNT };
2343    
2344    /* JIT compiling modes. The function list is indexed by them. */
2345    
2346    enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
2347           JIT_NUMBER_OF_COMPILE_MODES };
2348    
2349  /* The real format of the start of the pcre block; the index of names and the  /* The real format of the start of the pcre block; the index of names and the
2350  code vector run on as long as necessary after the end. We store an explicit  code vector run on as long as necessary after the end. We store an explicit
2351  offset to the name table so that if a regex is compiled on one host, saved, and  offset to the name table so that if a regex is compiled on one host, saved, and
2352  then run on another where the size of pointers is different, all might still  then run on another where the size of pointers is different, all might still
2353  be well. For the case of compiled-on-4 and run-on-8, we include an extra  be well.
2354  pointer that is always NULL. For future-proofing, a few dummy fields were  
2355  originally included - even though you can never get this planning right - but  The size of the structure must be a multiple of 8 bytes. For the case of
2356  there is only one left now.  compiled-on-4 and run-on-8, we include an extra pointer that is always NULL so
2357    that there are an even number of pointers which therefore are a multiple of 8
2358  NOTE NOTE NOTE:  bytes.
2359  Because people can now save and re-use compiled patterns, any additions to this  
2360  structure should be made at the end, and something earlier (e.g. a new  It is necessary to fork the struct for the 32 bit library, since it needs to
2361  flag in the options or one of the dummy fields) should indicate that the new  use pcre_uint32 for first_char and req_char. We can't put an ifdef inside the
2362  fields are present. Currently PCRE always sets the dummy fields to zero.  typedef because pcretest needs access to the struct of the 8-, 16- and 32-bit
2363  NOTE NOTE NOTE  variants.
2364    
2365    *** WARNING ***
2366    When new fields are added to these structures, remember to adjust the code in
2367    pcre_byte_order.c that is concerned with swapping the byte order of the fields
2368    when a compiled regex is reloaded on a host with different endianness.
2369    *** WARNING ***
2370    There is also similar byte-flipping code in pcretest.c, which is used for
2371    testing the byte-flipping features. It must also be kept in step.
2372    *** WARNING ***
2373  */  */
2374    
2375  typedef struct real_pcre {  typedef struct real_pcre8_or_16 {
2376    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2377    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2378    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2379    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2380    pcre_uint16 dummy1;             /* For future use */    pcre_uint32 limit_match;        /* Limit set from regex */
2381    pcre_uint16 top_bracket;    pcre_uint32 limit_recursion;    /* Limit set from regex */
   pcre_uint16 top_backref;  
2382    pcre_uint16 first_char;         /* Starting character */    pcre_uint16 first_char;         /* Starting character */
2383    pcre_uint16 req_char;           /* This character must be seen */    pcre_uint16 req_char;           /* This character must be seen */
2384      pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2385      pcre_uint16 top_bracket;        /* Highest numbered group */
2386      pcre_uint16 top_backref;        /* Highest numbered back reference */
2387    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2388    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2389    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2390    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2391      pcre_uint16 dummy1;             /* To ensure size is a multiple of 8 */
2392      pcre_uint16 dummy2;             /* To ensure size is a multiple of 8 */
2393      pcre_uint16 dummy3;             /* To ensure size is a multiple of 8 */
2394      const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2395      void             *nullpad;      /* NULL padding */
2396    } real_pcre8_or_16;
2397    
2398    typedef struct real_pcre8_or_16 real_pcre;
2399    typedef struct real_pcre8_or_16 real_pcre16;
2400    
2401    typedef struct real_pcre32 {
2402      pcre_uint32 magic_number;
2403      pcre_uint32 size;               /* Total that was malloced */
2404      pcre_uint32 options;            /* Public options */
2405      pcre_uint32 flags;              /* Private flags */
2406      pcre_uint32 limit_match;        /* Limit set from regex */
2407      pcre_uint32 limit_recursion;    /* Limit set from regex */
2408      pcre_uint32 first_char;         /* Starting character */
2409      pcre_uint32 req_char;           /* This character must be seen */
2410      pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2411      pcre_uint16 top_bracket;        /* Highest numbered group */
2412      pcre_uint16 top_backref;        /* Highest numbered back reference */
2413      pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2414      pcre_uint16 name_entry_size;    /* Size of any name items */
2415      pcre_uint16 name_count;         /* Number of name items */
2416      pcre_uint16 ref_count;          /* Reference count */
2417      pcre_uint16 dummy;              /* To ensure size is a multiple of 8 */
2418    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2419    const pcre_uint8 *nullpad;      /* NULL padding */    void             *nullpad;      /* NULL padding */
2420  } real_pcre;  } real_pcre32;
2421    
2422    #if defined COMPILE_PCRE8
2423    #define REAL_PCRE real_pcre
2424    #elif defined COMPILE_PCRE16
2425    #define REAL_PCRE real_pcre16
2426    #elif defined COMPILE_PCRE32
2427    #define REAL_PCRE real_pcre32
2428    #endif
2429    
2430    /* Assert that the size of REAL_PCRE is divisible by 8 */
2431    typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];
2432    
2433    /* Needed in pcretest to access some fields in the real_pcre* structures
2434     * directly. They're unified for 8/16/32 bits since the structs only differ
2435     * after these fields; if that ever changes, need to fork those defines into
2436     * 8/16 and 32 bit versions. */
2437    #define REAL_PCRE_MAGIC(re)     (((REAL_PCRE*)re)->magic_number)
2438    #define REAL_PCRE_SIZE(re)      (((REAL_PCRE*)re)->size)
2439    #define REAL_PCRE_OPTIONS(re)   (((REAL_PCRE*)re)->options)
2440    #define REAL_PCRE_FLAGS(re)     (((REAL_PCRE*)re)->flags)
2441    
2442  /* The format of the block used to store data from pcre_study(). The same  /* The format of the block used to store data from pcre_study(). The same
2443  remark (see NOTE above) about extending this structure applies. */  remark (see NOTE above) about extending this structure applies. */
# Line 1998  typedef struct open_capitem { Line 2460  typedef struct open_capitem {
2460    pcre_uint16 flag;             /* Set TRUE if recursive back ref */    pcre_uint16 flag;             /* Set TRUE if recursive back ref */
2461  } open_capitem;  } open_capitem;
2462    
2463    /* Structure for building a list of named groups during the first pass of
2464    compiling. */
2465    
2466    typedef struct named_group {
2467      const pcre_uchar  *name;          /* Points to the name in the pattern */
2468      int                length;        /* Length of the name */
2469      pcre_uint32        number;        /* Group number */
2470    } named_group;
2471    
2472  /* Structure for passing "static" information around between the functions  /* Structure for passing "static" information around between the functions
2473  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
2474    
# Line 2010  typedef struct compile_data { Line 2481  typedef struct compile_data {
2481    const pcre_uchar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
2482    const pcre_uchar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
2483    const pcre_uchar *end_pattern;    /* The end of the pattern */    const pcre_uchar *end_pattern;    /* The end of the pattern */
   open_capitem *open_caps;          /* Chain of open capture items */  
2484    pcre_uchar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
2485      open_capitem *open_caps;          /* Chain of open capture items */
2486      named_group *named_groups;        /* Points to vector in pre-compile */
2487    pcre_uchar *name_table;           /* The name/number table */    pcre_uchar *name_table;           /* The name/number table */
2488    int  names_found;                 /* Number of entries so far */    int  names_found;                 /* Number of entries so far */
2489    int  name_entry_size;             /* Size of each entry */    int  name_entry_size;             /* Size of each entry */
2490      int  named_group_list_size;       /* Number of entries in the list */
2491    int  workspace_size;              /* Size of workspace */    int  workspace_size;              /* Size of workspace */
2492    int  bracount;                    /* Count of capturing parens as we compile */    unsigned int bracount;            /* Count of capturing parens as we compile */
2493    int  final_bracount;              /* Saved value after first pass */    int  final_bracount;              /* Saved value after first pass */
2494      int  max_lookbehind;              /* Maximum lookbehind (characters) */
2495    int  top_backref;                 /* Maximum back reference */    int  top_backref;                 /* Maximum back reference */
2496    unsigned int backref_map;         /* Bitmap of low back refs */    unsigned int backref_map;         /* Bitmap of low back refs */
2497      unsigned int namedrefcount;       /* Number of backreferences by name */
2498      int  parens_depth;                /* Depth of nested parentheses */
2499    int  assert_depth;                /* Depth of nested assertions */    int  assert_depth;                /* Depth of nested assertions */
2500    int  external_options;            /* External (initial) options */    pcre_uint32 external_options;     /* External (initial) options */
2501    int  external_flags;              /* External flag bits to be set */    pcre_uint32 external_flags;       /* External flag bits to be set */
2502    int  req_varyopt;                 /* "After variable item" flag for reqbyte */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2503    BOOL had_accept;                  /* (*ACCEPT) encountered */    BOOL had_accept;                  /* (*ACCEPT) encountered */
2504      BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
2505    BOOL check_lookbehind;            /* Lookbehinds need later checking */    BOOL check_lookbehind;            /* Lookbehinds need later checking */
2506      BOOL dupnames;                    /* Duplicate names exist */
2507    int  nltype;                      /* Newline type */    int  nltype;                      /* Newline type */
2508    int  nllen;                       /* Newline string length */    int  nllen;                       /* Newline string length */
2509    pcre_uchar nl[4];                 /* Newline string when fixed length */    pcre_uchar nl[4];                 /* Newline string when fixed length */
# Line 2044  call within the pattern; used by pcre_ex Line 2522  call within the pattern; used by pcre_ex
2522    
2523  typedef struct recursion_info {  typedef struct recursion_info {
2524    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
2525    int group_num;                  /* Number of group that was called */    unsigned int group_num;         /* Number of group that was called */
2526    int *offset_save;               /* Pointer to start of saved offsets */    int *offset_save;               /* Pointer to start of saved offsets */
2527    int saved_max;                  /* Number of saved offsets */    int saved_max;                  /* Number of saved offsets */
2528      int saved_capture_last;         /* Last capture number */
2529    PCRE_PUCHAR subject_position;   /* Position at start of recursion */    PCRE_PUCHAR subject_position;   /* Position at start of recursion */
2530  } recursion_info;  } recursion_info;
2531    
# Line 2083  typedef struct match_data { Line 2562  typedef struct match_data {
2562    int    nllen;                   /* Newline string length */    int    nllen;                   /* Newline string length */
2563    int    name_count;              /* Number of names in name table */    int    name_count;              /* Number of names in name table */
2564    int    name_entry_size;         /* Size of entry in names table */    int    name_entry_size;         /* Size of entry in names table */
2565      unsigned int skip_arg_count;    /* For counting SKIP_ARGs */
2566      unsigned int ignore_skip_arg;   /* For re-run when SKIP arg name not found */
2567    pcre_uchar *name_table;         /* Table of names */    pcre_uchar *name_table;         /* Table of names */
2568    pcre_uchar nl[4];               /* Newline string when fixed */    pcre_uchar nl[4];               /* Newline string when fixed */
2569    const  pcre_uint8 *lcc;         /* Points to lower casing table */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
2570    const  pcre_uint8 *fcc;         /* Points to case-flipping table */    const  pcre_uint8 *fcc;         /* Points to case-flipping table */
2571    const  pcre_uint8 *ctypes;      /* Points to table of type maps */    const  pcre_uint8 *ctypes;      /* Points to table of type maps */
   BOOL   offset_overflow;         /* Set if too many extractions */  
2572    BOOL   notbol;                  /* NOTBOL flag */    BOOL   notbol;                  /* NOTBOL flag */
2573    BOOL   noteol;                  /* NOTEOL flag */    BOOL   noteol;                  /* NOTEOL flag */
2574    BOOL   utf;                     /* UTF-8 / UTF-16 flag */    BOOL   utf;                     /* UTF-8 / UTF-16 flag */
# Line 2100  typedef struct match_data { Line 2580  typedef struct match_data {
2580    BOOL   hitend;                  /* Hit the end of the subject at some point */    BOOL   hitend;                  /* Hit the end of the subject at some point */
2581    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
2582    BOOL   hasthen;                 /* Pattern contains (*THEN) */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
   BOOL   ignore_skip_arg;         /* For re-run when SKIP name not found */  
2583    const  pcre_uchar *start_code;  /* For use when recursing */    const  pcre_uchar *start_code;  /* For use when recursing */
2584    PCRE_PUCHAR start_subject;      /* Start of the subject string */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
2585    PCRE_PUCHAR end_subject;        /* End of the subject string */    PCRE_PUCHAR end_subject;        /* End of the subject string */
# Line 2109  typedef struct match_data { Line 2588  typedef struct match_data {
2588    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */
2589    int    partial;                 /* PARTIAL options */    int    partial;                 /* PARTIAL options */
2590    int    end_offset_top;          /* Highwater mark at end of match */    int    end_offset_top;          /* Highwater mark at end of match */
2591    int    capture_last;            /* Most recent capture number */    pcre_int32 capture_last;        /* Most recent capture number + overflow flag */
2592    int    start_offset;            /* The start offset value */    int    start_offset;            /* The start offset value */
2593    int    match_function_type;     /* Set for certain special calls of MATCH() */    int    match_function_type;     /* Set for certain special calls of MATCH() */
2594    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */
# Line 2119  typedef struct match_data { Line 2598  typedef struct match_data {
2598    const  pcre_uchar *mark;        /* Mark pointer to pass back on success */    const  pcre_uchar *mark;        /* Mark pointer to pass back on success */
2599    const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */    const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */
2600    const  pcre_uchar *once_target; /* Where to back up to for atomic groups */    const  pcre_uchar *once_target; /* Where to back up to for atomic groups */
2601    #ifdef NO_RECURSE
2602      void  *match_frames_base;       /* For remembering malloc'd frames */
2603    #endif
2604  } match_data;  } match_data;
2605    
2606  /* A similar structure is used for the same purpose by the DFA matching  /* A similar structure is used for the same purpose by the DFA matching
# Line 2173  total length. */ Line 2655  total length. */
2655  #define ctypes_offset (cbits_offset + cbit_length)  #define ctypes_offset (cbits_offset + cbit_length)
2656  #define tables_length (ctypes_offset + 256)  #define tables_length (ctypes_offset + 256)
2657    
2658  /* Internal function prefix */  /* Internal function and data prefixes. */
2659    
2660  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2661    #ifndef PUBL
2662  #define PUBL(name) pcre_##name  #define PUBL(name) pcre_##name
2663    #endif
2664    #ifndef PRIV
2665  #define PRIV(name) _pcre_##name  #define PRIV(name) _pcre_##name
2666  #else  #endif
2667  #ifdef COMPILE_PCRE16  #elif defined COMPILE_PCRE16
2668    #ifndef PUBL
2669  #define PUBL(name) pcre16_##name  #define PUBL(name) pcre16_##name
2670    #endif
2671    #ifndef PRIV
2672  #define PRIV(name) _pcre16_##name  #define PRIV(name) _pcre16_##name
2673    #endif
2674    #elif defined COMPILE_PCRE32
2675    #ifndef PUBL
2676    #define PUBL(name) pcre32_##name
2677    #endif
2678    #ifndef PRIV
2679    #define PRIV(name) _pcre32_##name
2680    #endif
2681  #else  #else
2682  #error Unsupported compiling mode  #error Unsupported compiling mode
2683  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
 #endif /* COMPILE_PCRE8 */  
2684    
2685  /* Layout of the UCP type table that translates property names into types and  /* Layout of the UCP type table that translates property names into types and
2686  codes. Each entry used to point directly to a name, but to reduce the number of  codes. Each entry used to point directly to a name, but to reduce the number of
# Line 2205  but are not part of the PCRE public API. Line 2700  but are not part of the PCRE public API.
2700  pcre_tables.c module. */  pcre_tables.c module. */
2701    
2702  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
   
2703  extern const int            PRIV(utf8_table1)[];  extern const int            PRIV(utf8_table1)[];
2704  extern const int            PRIV(utf8_table1_size);  extern const int            PRIV(utf8_table1_size);
2705  extern const int            PRIV(utf8_table2)[];  extern const int            PRIV(utf8_table2)[];
2706  extern const int            PRIV(utf8_table3)[];  extern const int            PRIV(utf8_table3)[];
2707  extern const pcre_uint8     PRIV(utf8_table4)[];  extern const pcre_uint8     PRIV(utf8_table4)[];
   
2708  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
2709    
2710  extern const char           PRIV(utt_names)[];  extern const char           PRIV(utt_names)[];
2711  extern const ucp_type_table PRIV(utt)[];  extern const ucp_type_table PRIV(utt)[];
2712  extern const int            PRIV(utt_size);  extern const int            PRIV(utt_size);
2713    
2714    extern const pcre_uint8     PRIV(OP_lengths)[];
2715  extern const pcre_uint8     PRIV(default_tables)[];  extern const pcre_uint8     PRIV(default_tables)[];
2716    
2717  extern const pcre_uint8     PRIV(OP_lengths)[];  extern const pcre_uint32    PRIV(hspace_list)[];
2718    extern const pcre_uint32    PRIV(vspace_list)[];
2719    
2720    
2721  /* Internal shared functions. These are functions that are used by more than  /* Internal shared functions. These are functions that are used by more than
# Line 2228  one of the exported public functions. Th Line 2723  one of the exported public functions. Th
2723  sense, but are not part of the PCRE public API. */  sense, but are not part of the PCRE public API. */
2724    
2725  /* String comparison functions. */  /* String comparison functions. */
2726  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2727    
2728  #define STRCMP_UC_UC(str1, str2) \  #define STRCMP_UC_UC(str1, str2) \
2729    strcmp((char *)(str1), (char *)(str2))    strcmp((char *)(str1), (char *)(str2))
# Line 2240  sense, but are not part of the PCRE publ Line 2735  sense, but are not part of the PCRE publ
2735    strncmp((char *)(str1), (str2), (num))    strncmp((char *)(str1), (str2), (num))
2736  #define STRLEN_UC(str) strlen((const char *)str)  #define STRLEN_UC(str) strlen((const char *)str)
2737    
2738  #else  #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2739    
2740  extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,  extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,
2741                             const pcre_uchar *);                             const pcre_uchar *);
# Line 2262  extern unsigned int      PRIV(strlen_uc) Line 2757  extern unsigned int      PRIV(strlen_uc)
2757    PRIV(strncmp_uc_c8)((str1), (str2), (num))    PRIV(strncmp_uc_c8)((str1), (str2), (num))
2758  #define STRLEN_UC(str) PRIV(strlen_uc)(str)  #define STRLEN_UC(str) PRIV(strlen_uc)(str)
2759    
2760  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE[8|16|32] */
2761    
2762    #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16
2763    
2764    #define STRCMP_UC_UC_TEST(str1, str2) STRCMP_UC_UC(str1, str2)
2765    #define STRCMP_UC_C8_TEST(str1, str2) STRCMP_UC_C8(str1, str2)
2766    
2767    #elif defined COMPILE_PCRE32
2768    
2769    extern int               PRIV(strcmp_uc_uc_utf)(const pcre_uchar *,
2770                               const pcre_uchar *);
2771    extern int               PRIV(strcmp_uc_c8_utf)(const pcre_uchar *,
2772                               const char *);
2773    
2774    #define STRCMP_UC_UC_TEST(str1, str2) \
2775      (utf ? PRIV(strcmp_uc_uc_utf)((str1), (str2)) : PRIV(strcmp_uc_uc)((str1), (str2)))
2776    #define STRCMP_UC_C8_TEST(str1, str2) \
2777      (utf ? PRIV(strcmp_uc_c8_utf)((str1), (str2)) : PRIV(strcmp_uc_c8)((str1), (str2)))
2778    
2779    #endif /* COMPILE_PCRE[8|16|32] */
2780    
2781  extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);  extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);
2782  extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,  extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2783                             int *, BOOL);                             int *, BOOL);
2784  extern int               PRIV(ord2utf)(pcre_uint32, pcre_uchar *);  extern unsigned int      PRIV(ord2utf)(pcre_uint32, pcre_uchar *);
2785  extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);  extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);
2786  extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,  extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2787                             int *, BOOL);                             int *, BOOL);
2788  extern BOOL              PRIV(xclass)(int, const pcre_uchar *, BOOL);  extern BOOL              PRIV(xclass)(pcre_uint32, const pcre_uchar *, BOOL);
2789    
2790  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2791  extern void              PRIV(jit_compile)(const real_pcre *, PUBL(extra) *);  extern void              PRIV(jit_compile)(const REAL_PCRE *,
2792  extern int               PRIV(jit_exec)(const real_pcre *, void *,                             PUBL(extra) *, int);
2793                             const pcre_uchar *, int, int, int, int, int *, int);  extern int               PRIV(jit_exec)(const PUBL(extra) *,
2794                               const pcre_uchar *, int, int, int, int *, int);
2795  extern void              PRIV(jit_free)(void *);  extern void              PRIV(jit_free)(void *);
2796  extern int               PRIV(jit_get_size)(void *);  extern int               PRIV(jit_get_size)(void *);
2797    extern const char*       PRIV(jit_get_target)(void);
2798  #endif  #endif
2799    
2800  /* Unicode character database (UCD) */  /* Unicode character database (UCD) */
2801    
2802  typedef struct {  typedef struct {
2803    pcre_uint8 script;    pcre_uint8 script;     /* ucp_Arabic, etc. */
2804    pcre_uint8 chartype;    pcre_uint8 chartype;   /* ucp_Cc, etc. (general categories) */
2805    pcre_int32 other_case;    pcre_uint8 gbprop;     /* ucp_gbControl, etc. (grapheme break property) */
2806      pcre_uint8 caseset;    /* offset to multichar other cases or zero */
2807      pcre_int32 other_case; /* offset to other case, or zero if none */
2808  } ucd_record;  } ucd_record;
2809    
2810    extern const pcre_uint32 PRIV(ucd_caseless_sets)[];
2811  extern const ucd_record  PRIV(ucd_records)[];  extern const ucd_record  PRIV(ucd_records)[];
2812  extern const pcre_uint8  PRIV(ucd_stage1)[];  extern const pcre_uint8  PRIV(ucd_stage1)[];
2813  extern const pcre_uint16 PRIV(ucd_stage2)[];  extern const pcre_uint16 PRIV(ucd_stage2)[];
2814  extern const int         PRIV(ucp_gentype)[];  extern const pcre_uint32 PRIV(ucp_gentype)[];
2815    extern const pcre_uint32 PRIV(ucp_gbtable)[];
2816  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2817  extern const int         PRIV(ucp_typerange)[];  extern const int         PRIV(ucp_typerange)[];
2818  #endif  #endif
# Line 2302  extern const int         PRIV(ucp_typera Line 2822  extern const int         PRIV(ucp_typera
2822    
2823  #define UCD_BLOCK_SIZE 128  #define UCD_BLOCK_SIZE 128
2824  #define GET_UCD(ch) (PRIV(ucd_records) + \  #define GET_UCD(ch) (PRIV(ucd_records) + \
2825          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(ch) / UCD_BLOCK_SIZE] * \          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \
2826          UCD_BLOCK_SIZE + (ch) % UCD_BLOCK_SIZE])          UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE])
2827    
2828  #define UCD_CHARTYPE(ch)  GET_UCD(ch)->chartype  #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
2829  #define UCD_SCRIPT(ch)    GET_UCD(ch)->script  #define UCD_SCRIPT(ch)      GET_UCD(ch)->script
2830  #define UCD_CATEGORY(ch)  PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]  #define UCD_CATEGORY(ch)    PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]
2831  #define UCD_OTHERCASE(ch) (ch + GET_UCD(ch)->other_case)  #define UCD_GRAPHBREAK(ch)  GET_UCD(ch)->gbprop
2832    #define UCD_CASESET(ch)     GET_UCD(ch)->caseset
2833    #define UCD_OTHERCASE(ch)   ((pcre_uint32)((int)ch + (int)(GET_UCD(ch)->other_case)))
2834    
2835  #endif /* SUPPORT_UCP */  #endif /* SUPPORT_UCP */
2836    

Legend:
Removed from v.850  
changed lines
  Added in v.1408

  ViewVC Help
Powered by ViewVC 1.1.5