/[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 1567 by ph10, Fri Jun 12 16:24:58 2015 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) 1997-2014 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] && \       UCHAR21TEST(p) == NLBLOCK->nl[0] && \
320       (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || UCHAR21TEST(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] && \       UCHAR21TEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] &&              \
334       (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || UCHAR21TEST(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    /* There is a proposed future special "UTF-21" mode, in which only the lowest
586    21 bits of a 32-bit character are interpreted as UTF, with the remaining 11
587    high-order bits available to the application for other uses. In preparation for
588    the future implementation of this mode, there are macros that load a data item
589    and, if in this special mode, mask it to 21 bits. These macros all have names
590    starting with UCHAR21. In all other modes, including the normal 32-bit
591    library, the macros all have the same simple definitions. When the new mode is
592    implemented, it is expected that these definitions will be varied appropriately
593    using #ifdef when compiling the library that supports the special mode. */
594    
595    #define UCHAR21(eptr)        (*(eptr))
596    #define UCHAR21TEST(eptr)    (*(eptr))
597    #define UCHAR21INC(eptr)     (*(eptr)++)
598    #define UCHAR21INCTEST(eptr) (*(eptr)++)
599    
600  /* 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
601  character. The macros for character handling generate simple sequences when  byte in 8-bit mode or a single short in 16-bit mode. The macros for character
602  used in character-mode, and more complicated ones for UTF characters.  handling generate simple sequences when used in the basic mode, and more
603  GETCHARLENTEST and other macros are not used when UTF is not supported,  complicated ones for UTF characters. GETCHARLENTEST and other macros are not
604  so they are not defined. To make sure they can never even appear when  used when UTF is not supported. To make sure they can never even appear when
605  UTF support is omitted, we don't even define them. */  UTF support is omitted, we don't even define them. */
606    
607  #ifndef SUPPORT_UTF  #ifndef SUPPORT_UTF
# Line 548  UTF support is omitted, we don't even de Line 622  UTF support is omitted, we don't even de
622    
623  #else   /* SUPPORT_UTF */  #else   /* SUPPORT_UTF */
624    
 #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  
   
625  /* Tests whether the code point needs extra characters to decode. */  /* Tests whether the code point needs extra characters to decode. */
626    
627  #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)  
628    
629  /* 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
630  advancing the pointer. */  advancing the pointer. */
# Line 595  advancing the pointer. */ Line 648  advancing the pointer. */
648            ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \            ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \
649      }      }
650    
 /* 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);  
   
651  /* 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
652  the pointer. */  the pointer. */
653    
# Line 643  the pointer. */ Line 682  the pointer. */
682        } \        } \
683      }      }
684    
685    #if defined COMPILE_PCRE8
686    
687    /* These macros were originally written in the form of loops that used data
688    from the tables whose names start with PRIV(utf8_table). They were rewritten by
689    a user so as not to use loops, because in some environments this gives a
690    significant performance advantage, and it seems never to do any harm. */
691    
692    /* Tells the biggest code point which can be encoded as a single character. */
693    
694    #define MAX_VALUE_FOR_SINGLE_CHAR 127
695    
696    /* Tests whether the code point needs extra characters to decode. */
697    
698    #define HAS_EXTRALEN(c) ((c) >= 0xc0)
699    
700    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
701    Otherwise it has an undefined behaviour. */
702    
703    #define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f])
704    
705    /* Returns TRUE, if the given character is not the first character
706    of a UTF sequence. */
707    
708    #define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80)
709    
710    /* Get the next UTF-8 character, not advancing the pointer. This is called when
711    we know we are in UTF-8 mode. */
712    
713    #define GETCHAR(c, eptr) \
714      c = *eptr; \
715      if (c >= 0xc0) GETUTF8(c, eptr);
716    
717    /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
718    pointer. */
719    
720    #define GETCHARTEST(c, eptr) \
721      c = *eptr; \
722      if (utf && c >= 0xc0) GETUTF8(c, eptr);
723    
724  /* 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
725  know we are in UTF-8 mode. */  know we are in UTF-8 mode. */
726    
# Line 722  because almost all calls are already wit Line 800  because almost all calls are already wit
800  #define ACROSSCHAR(condition, eptr, action) \  #define ACROSSCHAR(condition, eptr, action) \
801    while((condition) && ((eptr) & 0xc0) == 0x80) action    while((condition) && ((eptr) & 0xc0) == 0x80) action
802    
803  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
   
 #ifdef COMPILE_PCRE16  
804    
805  /* 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. */
806    
# Line 820  code. */ Line 896  code. */
896  #define ACROSSCHAR(condition, eptr, action) \  #define ACROSSCHAR(condition, eptr, action) \
897    if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action    if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action
898    
899  #endif  #elif defined COMPILE_PCRE32
900    
901  #endif /* COMPILE_PCRE8 */  /* These are trivial for the 32-bit library, since all UTF-32 characters fit
902    into one pcre_uchar unit. */
903    #define MAX_VALUE_FOR_SINGLE_CHAR (0x10ffffu)
904    #define HAS_EXTRALEN(c) (0)
905    #define GET_EXTRALEN(c) (0)
906    #define NOT_FIRSTCHAR(c) (0)
907    
908    /* Get the next UTF-32 character, not advancing the pointer. This is called when
909    we know we are in UTF-32 mode. */
910    
911    #define GETCHAR(c, eptr) \
912      c = *(eptr);
913    
914    /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the
915    pointer. */
916    
917    #define GETCHARTEST(c, eptr) \
918      c = *(eptr);
919    
920    /* Get the next UTF-32 character, advancing the pointer. This is called when we
921    know we are in UTF-32 mode. */
922    
923    #define GETCHARINC(c, eptr) \
924      c = *((eptr)++);
925    
926    /* Get the next character, testing for UTF-32 mode, and advancing the pointer.
927    This is called when we don't know if we are in UTF-32 mode. */
928    
929    #define GETCHARINCTEST(c, eptr) \
930      c = *((eptr)++);
931    
932    /* Get the next UTF-32 character, not advancing the pointer, not incrementing
933    length (since all UTF-32 is of length 1). This is called when we know we are in
934    UTF-32 mode. */
935    
936    #define GETCHARLEN(c, eptr, len) \
937      GETCHAR(c, eptr)
938    
939    /* Get the next UTF-32character, testing for UTF-32 mode, not advancing the
940    pointer, not incrementing the length (since all UTF-32 is of length 1).
941    This is called when we do not know if we are in UTF-32 mode. */
942    
943    #define GETCHARLENTEST(c, eptr, len) \
944      GETCHARTEST(c, eptr)
945    
946    /* If the pointer is not at the start of a character, move it back until
947    it is. This is called only in UTF-32 mode - we don't put a test within the
948    macro because almost all calls are already within a block of UTF-32 only
949    code.
950    These are all no-ops since all UTF-32 characters fit into one pcre_uchar. */
951    
952    #define BACKCHAR(eptr) do { } while (0)
953    
954    /* Same as above, just in the other direction. */
955    #define FORWARDCHAR(eptr) do { } while (0)
956    
957    /* Same as above, but it allows a fully customizable form. */
958    #define ACROSSCHAR(condition, eptr, action) do { } while (0)
959    
960    #else
961    #error Unsupported compiling mode
962    #endif /* COMPILE_PCRE[8|16|32] */
963    
964  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
965    
966    /* Tests for Unicode horizontal and vertical whitespace characters must check a
967    number of different values. Using a switch statement for this generates the
968    fastest code (no loop, no memory access), and there are several places in the
969    interpreter code where this happens. In order to ensure that all the case lists
970    remain in step, we use macros so that there is only one place where the lists
971    are defined.
972    
973    These values are also required as lists in pcre_compile.c when processing \h,
974    \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but
975    macros that define the values are here so that all the definitions are
976    together. The lists must be in ascending character order, terminated by
977    NOTACHAR (which is 0xffffffff).
978    
979    Any changes should ensure that the various macros are kept in step with each
980    other. NOTE: The values also appear in pcre_jit_compile.c. */
981    
982    /* ------ ASCII/Unicode environments ------ */
983    
984    #ifndef EBCDIC
985    
986    #define HSPACE_LIST \
987      CHAR_HT, CHAR_SPACE, 0xa0, \
988      0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \
989      0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \
990      NOTACHAR
991    
992    #define HSPACE_MULTIBYTE_CASES \
993      case 0x1680:  /* OGHAM SPACE MARK */ \
994      case 0x180e:  /* MONGOLIAN VOWEL SEPARATOR */ \
995      case 0x2000:  /* EN QUAD */ \
996      case 0x2001:  /* EM QUAD */ \
997      case 0x2002:  /* EN SPACE */ \
998      case 0x2003:  /* EM SPACE */ \
999      case 0x2004:  /* THREE-PER-EM SPACE */ \
1000      case 0x2005:  /* FOUR-PER-EM SPACE */ \
1001      case 0x2006:  /* SIX-PER-EM SPACE */ \
1002      case 0x2007:  /* FIGURE SPACE */ \
1003      case 0x2008:  /* PUNCTUATION SPACE */ \
1004      case 0x2009:  /* THIN SPACE */ \
1005      case 0x200A:  /* HAIR SPACE */ \
1006      case 0x202f:  /* NARROW NO-BREAK SPACE */ \
1007      case 0x205f:  /* MEDIUM MATHEMATICAL SPACE */ \
1008      case 0x3000   /* IDEOGRAPHIC SPACE */
1009    
1010    #define HSPACE_BYTE_CASES \
1011      case CHAR_HT: \
1012      case CHAR_SPACE: \
1013      case 0xa0     /* NBSP */
1014    
1015    #define HSPACE_CASES \
1016      HSPACE_BYTE_CASES: \
1017      HSPACE_MULTIBYTE_CASES
1018    
1019    #define VSPACE_LIST \
1020      CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR
1021    
1022    #define VSPACE_MULTIBYTE_CASES \
1023      case 0x2028:    /* LINE SEPARATOR */ \
1024      case 0x2029     /* PARAGRAPH SEPARATOR */
1025    
1026    #define VSPACE_BYTE_CASES \
1027      case CHAR_LF: \
1028      case CHAR_VT: \
1029      case CHAR_FF: \
1030      case CHAR_CR: \
1031      case CHAR_NEL
1032    
1033    #define VSPACE_CASES \
1034      VSPACE_BYTE_CASES: \
1035      VSPACE_MULTIBYTE_CASES
1036    
1037    /* ------ EBCDIC environments ------ */
1038    
1039    #else
1040    #define HSPACE_LIST CHAR_HT, CHAR_SPACE
1041    
1042  /* In case there is no definition of offsetof() provided - though any proper  #define HSPACE_BYTE_CASES \
1043  Standard C system should have one. */    case CHAR_HT: \
1044      case CHAR_SPACE
1045    
1046  #ifndef offsetof  #define HSPACE_CASES HSPACE_BYTE_CASES
1047  #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))  
1048    #ifdef EBCDIC_NL25
1049    #define VSPACE_LIST \
1050      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR
1051    #else
1052    #define VSPACE_LIST \
1053      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR
1054  #endif  #endif
1055    
1056    #define VSPACE_BYTE_CASES \
1057      case CHAR_LF: \
1058      case CHAR_VT: \
1059      case CHAR_FF: \
1060      case CHAR_CR: \
1061      case CHAR_NEL
1062    
1063    #define VSPACE_CASES VSPACE_BYTE_CASES
1064    #endif  /* EBCDIC */
1065    
1066    /* ------ End of whitespace macros ------ */
1067    
1068    
1069    
1070  /* Private flags containing information about the compiled regex. They used to  /* Private flags containing information about the compiled regex. They used to
1071  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
1072  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
1073  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
1074    restrictions on partial matching have been lifted. It remains for backwards
1075  compatibility. */  compatibility. */
1076    
1077  #ifdef COMPILE_PCRE8  #define PCRE_MODE8         0x00000001  /* compiled in 8 bit mode */
1078  #define PCRE_MODE          0x0001  /* compiled in 8 bit mode */  #define PCRE_MODE16        0x00000002  /* compiled in 16 bit mode */
1079  #endif  #define PCRE_MODE32        0x00000004  /* compiled in 32 bit mode */
1080  #ifdef COMPILE_PCRE16  #define PCRE_FIRSTSET      0x00000010  /* first_char is set */
1081  #define PCRE_MODE          0x0002  /* compiled in 16 bit mode */  #define PCRE_FCH_CASELESS  0x00000020  /* caseless first char */
1082    #define PCRE_REQCHSET      0x00000040  /* req_byte is set */
1083    #define PCRE_RCH_CASELESS  0x00000080  /* caseless requested char */
1084    #define PCRE_STARTLINE     0x00000100  /* start after \n for multiline */
1085    #define PCRE_NOPARTIAL     0x00000200  /* can't use partial with this regex */
1086    #define PCRE_JCHANGED      0x00000400  /* j option used in regex */
1087    #define PCRE_HASCRORLF     0x00000800  /* explicit \r or \n in pattern */
1088    #define PCRE_HASTHEN       0x00001000  /* pattern contains (*THEN) */
1089    #define PCRE_MLSET         0x00002000  /* match limit set by regex */
1090    #define PCRE_RLSET         0x00004000  /* recursion limit set by regex */
1091    #define PCRE_MATCH_EMPTY   0x00008000  /* pattern can match empty string */
1092    
1093    #if defined COMPILE_PCRE8
1094    #define PCRE_MODE          PCRE_MODE8
1095    #elif defined COMPILE_PCRE16
1096    #define PCRE_MODE          PCRE_MODE16
1097    #elif defined COMPILE_PCRE32
1098    #define PCRE_MODE          PCRE_MODE32
1099  #endif  #endif
1100  #define PCRE_FIRSTSET      0x0010  /* first_char is set */  #define PCRE_MODE_MASK     (PCRE_MODE8 | PCRE_MODE16 | PCRE_MODE32)
 #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) */  
1101    
1102  /* Flags for the "extra" block produced by pcre_study(). */  /* Flags for the "extra" block produced by pcre_study(). */
1103    
# Line 871  time, run time, or study time, respectiv Line 1113  time, run time, or study time, respectiv
1113  #define PUBLIC_COMPILE_OPTIONS \  #define PUBLIC_COMPILE_OPTIONS \
1114    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
1115     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
1116     PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \     PCRE_NO_AUTO_CAPTURE|PCRE_NO_AUTO_POSSESS| \
1117       PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
1118     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
1119     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE)     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE|PCRE_NEVER_UTF)
1120    
1121  #define PUBLIC_EXEC_OPTIONS \  #define PUBLIC_EXEC_OPTIONS \
1122    (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 1130  time, run time, or study time, respectiv
1130     PCRE_NO_START_OPTIMIZE)     PCRE_NO_START_OPTIMIZE)
1131    
1132  #define PUBLIC_STUDY_OPTIONS \  #define PUBLIC_STUDY_OPTIONS \
1133     PCRE_STUDY_JIT_COMPILE     (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \
1134        PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED)
1135    
1136    #define PUBLIC_JIT_EXEC_OPTIONS \
1137       (PCRE_NO_UTF8_CHECK|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|\
1138        PCRE_NOTEMPTY_ATSTART|PCRE_PARTIAL_SOFT|PCRE_PARTIAL_HARD)
1139    
1140  /* Magic number to provide a small check against being handed junk. */  /* Magic number to provide a small check against being handed junk. */
1141    
# Line 898  in different endianness. */ Line 1146  in different endianness. */
1146    
1147  #define REVERSED_MAGIC_NUMBER  0x45524350UL   /* 'ERCP' */  #define REVERSED_MAGIC_NUMBER  0x45524350UL   /* 'ERCP' */
1148    
 /* Negative values for the firstchar and reqchar variables */  
   
 #define REQ_UNSET (-2)  
 #define REQ_NONE  (-1)  
   
1149  /* 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
1150  req_byte match. */  req_byte match. */
1151    
# Line 939  macros to give the functions distinct na Line 1182  macros to give the functions distinct na
1182  #ifndef SUPPORT_UTF  #ifndef SUPPORT_UTF
1183    
1184  /* UTF-8 support is not enabled; use the platform-dependent character literals  /* UTF-8 support is not enabled; use the platform-dependent character literals
1185  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
1186    mode. Newline characters are problematic in EBCDIC. Though it has CR and LF
1187    characters, a common practice has been to use its NL (0x15) character as the
1188    line terminator in C-like processing environments. However, sometimes the LF
1189    (0x25) character is used instead, according to this Unicode document:
1190    
1191    http://unicode.org/standard/reports/tr13/tr13-5.html
1192    
1193    PCRE defaults EBCDIC NL to 0x15, but has a build-time option to select 0x25
1194    instead. Whichever is *not* chosen is defined as NEL.
1195    
1196    In both ASCII and EBCDIC environments, CHAR_NL and CHAR_LF are synonyms for the
1197    same code point. */
1198    
1199    #ifdef EBCDIC
1200    
1201    #ifndef EBCDIC_NL25
1202    #define CHAR_NL                     '\x15'
1203    #define CHAR_NEL                    '\x25'
1204    #define STR_NL                      "\x15"
1205    #define STR_NEL                     "\x25"
1206    #else
1207    #define CHAR_NL                     '\x25'
1208    #define CHAR_NEL                    '\x15'
1209    #define STR_NL                      "\x25"
1210    #define STR_NEL                     "\x15"
1211    #endif
1212    
1213    #define CHAR_LF                     CHAR_NL
1214    #define STR_LF                      STR_NL
1215    
1216    #define CHAR_ESC                    '\047'
1217    #define CHAR_DEL                    '\007'
1218    #define STR_ESC                     "\047"
1219    #define STR_DEL                     "\007"
1220    
1221    #else  /* Not EBCDIC */
1222    
1223    /* In ASCII/Unicode, linefeed is '\n' and we equate this to NL for
1224    compatibility. NEL is the Unicode newline character; make sure it is
1225    a positive value. */
1226    
1227    #define CHAR_LF                     '\n'
1228    #define CHAR_NL                     CHAR_LF
1229    #define CHAR_NEL                    ((unsigned char)'\x85')
1230    #define CHAR_ESC                    '\033'
1231    #define CHAR_DEL                    '\177'
1232    
1233    #define STR_LF                      "\n"
1234    #define STR_NL                      STR_LF
1235    #define STR_NEL                     "\x85"
1236    #define STR_ESC                     "\033"
1237    #define STR_DEL                     "\177"
1238    
1239    #endif  /* EBCDIC */
1240    
1241    /* The remaining definitions work in both environments. */
1242    
1243    #define CHAR_NULL                   '\0'
1244  #define CHAR_HT                     '\t'  #define CHAR_HT                     '\t'
1245  #define CHAR_VT                     '\v'  #define CHAR_VT                     '\v'
1246  #define CHAR_FF                     '\f'  #define CHAR_FF                     '\f'
1247  #define CHAR_CR                     '\r'  #define CHAR_CR                     '\r'
 #define CHAR_NL                     '\n'  
1248  #define CHAR_BS                     '\b'  #define CHAR_BS                     '\b'
1249  #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  
1250    
1251  #define CHAR_SPACE                  ' '  #define CHAR_SPACE                  ' '
1252  #define CHAR_EXCLAMATION_MARK       '!'  #define CHAR_EXCLAMATION_MARK       '!'
# Line 1056  so that PCRE works on both ASCII and EBC Line 1348  so that PCRE works on both ASCII and EBC
1348  #define STR_VT                      "\v"  #define STR_VT                      "\v"
1349  #define STR_FF                      "\f"  #define STR_FF                      "\f"
1350  #define STR_CR                      "\r"  #define STR_CR                      "\r"
 #define STR_NL                      "\n"  
1351  #define STR_BS                      "\b"  #define STR_BS                      "\b"
1352  #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  
1353    
1354  #define STR_SPACE                   " "  #define STR_SPACE                   " "
1355  #define STR_EXCLAMATION_MARK        "!"  #define STR_EXCLAMATION_MARK        "!"
# Line 1188  so that PCRE works on both ASCII and EBC Line 1472  so that PCRE works on both ASCII and EBC
1472  #define STRING_xdigit               "xdigit"  #define STRING_xdigit               "xdigit"
1473    
1474  #define STRING_DEFINE               "DEFINE"  #define STRING_DEFINE               "DEFINE"
1475    #define STRING_WEIRD_STARTWORD      "[:<:]]"
1476    #define STRING_WEIRD_ENDWORD        "[:>:]]"
1477    
1478  #define STRING_CR_RIGHTPAR             "CR)"  #define STRING_CR_RIGHTPAR              "CR)"
1479  #define STRING_LF_RIGHTPAR             "LF)"  #define STRING_LF_RIGHTPAR              "LF)"
1480  #define STRING_CRLF_RIGHTPAR           "CRLF)"  #define STRING_CRLF_RIGHTPAR            "CRLF)"
1481  #define STRING_ANY_RIGHTPAR            "ANY)"  #define STRING_ANY_RIGHTPAR             "ANY)"
1482  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"  #define STRING_ANYCRLF_RIGHTPAR         "ANYCRLF)"
1483  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"  #define STRING_BSR_ANYCRLF_RIGHTPAR     "BSR_ANYCRLF)"
1484  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"  #define STRING_BSR_UNICODE_RIGHTPAR     "BSR_UNICODE)"
1485  #ifdef COMPILE_PCRE8  #define STRING_UTF8_RIGHTPAR            "UTF8)"
1486  #define STRING_UTF_RIGHTPAR            "UTF8)"  #define STRING_UTF16_RIGHTPAR           "UTF16)"
1487  #endif  #define STRING_UTF32_RIGHTPAR           "UTF32)"
1488  #ifdef COMPILE_PCRE16  #define STRING_UTF_RIGHTPAR             "UTF)"
1489  #define STRING_UTF_RIGHTPAR            "UTF16)"  #define STRING_UCP_RIGHTPAR             "UCP)"
1490  #endif  #define STRING_NO_AUTO_POSSESS_RIGHTPAR "NO_AUTO_POSSESS)"
1491  #define STRING_UCP_RIGHTPAR            "UCP)"  #define STRING_NO_START_OPT_RIGHTPAR    "NO_START_OPT)"
1492  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"  #define STRING_LIMIT_MATCH_EQ           "LIMIT_MATCH="
1493    #define STRING_LIMIT_RECURSION_EQ       "LIMIT_RECURSION="
1494    
1495  #else  /* SUPPORT_UTF */  #else  /* SUPPORT_UTF */
1496    
# Line 1215  only. */ Line 1502  only. */
1502  #define CHAR_VT                     '\013'  #define CHAR_VT                     '\013'
1503  #define CHAR_FF                     '\014'  #define CHAR_FF                     '\014'
1504  #define CHAR_CR                     '\015'  #define CHAR_CR                     '\015'
1505  #define CHAR_NL                     '\012'  #define CHAR_LF                     '\012'
1506    #define CHAR_NL                     CHAR_LF
1507    #define CHAR_NEL                    ((unsigned char)'\x85')
1508  #define CHAR_BS                     '\010'  #define CHAR_BS                     '\010'
1509  #define CHAR_BEL                    '\007'  #define CHAR_BEL                    '\007'
1510  #define CHAR_ESC                    '\033'  #define CHAR_ESC                    '\033'
1511  #define CHAR_DEL                    '\177'  #define CHAR_DEL                    '\177'
1512    
1513    #define CHAR_NULL                   '\0'
1514  #define CHAR_SPACE                  '\040'  #define CHAR_SPACE                  '\040'
1515  #define CHAR_EXCLAMATION_MARK       '\041'  #define CHAR_EXCLAMATION_MARK       '\041'
1516  #define CHAR_QUOTATION_MARK         '\042'  #define CHAR_QUOTATION_MARK         '\042'
# Line 1448  only. */ Line 1738  only. */
1738  #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
1739    
1740  #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
1741    #define STRING_WEIRD_STARTWORD      STR_LEFT_SQUARE_BRACKET STR_COLON STR_LESS_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1742    #define STRING_WEIRD_ENDWORD        STR_LEFT_SQUARE_BRACKET STR_COLON STR_GREATER_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1743    
1744  #define STRING_CR_RIGHTPAR             STR_C STR_R STR_RIGHT_PARENTHESIS  #define STRING_CR_RIGHTPAR              STR_C STR_R STR_RIGHT_PARENTHESIS
1745  #define STRING_LF_RIGHTPAR             STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_LF_RIGHTPAR              STR_L STR_F STR_RIGHT_PARENTHESIS
1746  #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
1747  #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
1748  #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
1749  #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
1750  #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
1751  #ifdef COMPILE_PCRE8  #define STRING_UTF8_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
1752  #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
1753  #endif  #define STRING_UTF32_RIGHTPAR           STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
1754  #ifdef COMPILE_PCRE16  #define STRING_UTF_RIGHTPAR             STR_U STR_T STR_F STR_RIGHT_PARENTHESIS
1755  #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
1756  #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
1757  #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
1758  #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
1759    #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
1760    
1761  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
1762    
1763  /* Escape items that are just an encoding of a particular data value. */  /* Escape items that are just an encoding of a particular data value. */
1764    
1765    #ifndef ESC_a
1766    #define ESC_a CHAR_BEL
1767    #endif
1768    
1769  #ifndef ESC_e  #ifndef ESC_e
1770  #define ESC_e CHAR_ESC  #define ESC_e CHAR_ESC
1771  #endif  #endif
# Line 1478  only. */ Line 1775  only. */
1775  #endif  #endif
1776    
1777  #ifndef ESC_n  #ifndef ESC_n
1778  #define ESC_n CHAR_NL  #define ESC_n CHAR_LF
1779  #endif  #endif
1780    
1781  #ifndef ESC_r  #ifndef ESC_r
# Line 1503  only. */ Line 1800  only. */
1800  #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */  #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */
1801  #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 */
1802  #define PT_WORD       8    /* Word - L plus N plus underscore */  #define PT_WORD       8    /* Word - L plus N plus underscore */
1803    #define PT_CLIST      9    /* Pseudo-property: match character list */
1804    #define PT_UCNC      10    /* Universal Character nameable character */
1805    #define PT_TABSIZE   11    /* Size of square table for autopossessify tests */
1806    
1807    /* The following special properties are used only in XCLASS items, when POSIX
1808    classes are specified and PCRE_UCP is set - in other words, for Unicode
1809    handling of these classes. They are not available via the \p or \P escapes like
1810    those in the above list, and so they do not take part in the autopossessifying
1811    table. */
1812    
1813    #define PT_PXGRAPH   11    /* [:graph:] - characters that mark the paper */
1814    #define PT_PXPRINT   12    /* [:print:] - [:graph:] plus non-control spaces */
1815    #define PT_PXPUNCT   13    /* [:punct:] - punctuation characters */
1816    
1817  /* 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
1818  contain characters with values greater than 255. */  contain characters with values greater than 255. */
1819    
1820  #define XCL_NOT    0x01    /* Flag: this is a negative class */  #define XCL_NOT       0x01    /* Flag: this is a negative class */
1821  #define XCL_MAP    0x02    /* Flag: a 32-byte map is present */  #define XCL_MAP       0x02    /* Flag: a 32-byte map is present */
1822    #define XCL_HASPROP   0x04    /* Flag: property checks are present. */
1823    
1824  #define XCL_END       0    /* Marks end of individual items */  #define XCL_END       0    /* Marks end of individual items */
1825  #define XCL_SINGLE    1    /* Single item (one multibyte char) follows */  #define XCL_SINGLE    1    /* Single item (one multibyte char) follows */
# Line 1517  contain characters with values greater t Line 1828  contain characters with values greater t
1828  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */
1829    
1830  /* 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
1831  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
1832  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
1833  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
1834  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
1835  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
1836  non-DOTALL mode, "." behaves like \N.  non-DOTALL mode, "." behaves like \N.
1837    
1838  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.
1839  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.
1840  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
1841  looked up from a table.  looked up from a table.
1842    
1843  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
1844  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
1845  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
1846  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
1847  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 1851  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E
1851         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,
1852         ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z,         ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z,
1853         ESC_E, ESC_Q, ESC_g, ESC_k,         ESC_E, ESC_Q, ESC_g, ESC_k,
1854         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu,         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu };
1855         ESC_REF };  
1856    
1857    /********************** Opcode definitions ******************/
1858    
1859    /****** NOTE NOTE NOTE ******
1860    
1861    Starting from 1 (i.e. after OP_END), the values up to OP_EOD must correspond in
1862    order to the list of escapes immediately above. Furthermore, values up to
1863    OP_DOLLM must not be changed without adjusting the table called autoposstab in
1864    pcre_compile.c
1865    
1866    Whenever this list is updated, the two macro definitions that follow must be
1867    updated to match. The possessification table called "opcode_possessify" in
1868    pcre_compile.c must also be updated, and also the tables called "coptable"
1869    and "poptable" in pcre_dfa_exec.c.
1870    
1871    ****** NOTE NOTE NOTE ******/
1872    
1873    
1874  /* Opcode table: Starting from 1 (i.e. after OP_END), the values up to  /* The values between FIRST_AUTOTAB_OP and LAST_AUTOTAB_RIGHT_OP, inclusive,
1875  OP_EOD must correspond in order to the list of escapes immediately above.  are used in a table for deciding whether a repeated character type can be
1876    auto-possessified. */
1877    
1878  *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions  #define FIRST_AUTOTAB_OP       OP_NOT_DIGIT
1879  that follow must also be updated to match. There are also tables called  #define LAST_AUTOTAB_LEFT_OP   OP_EXTUNI
1880  "coptable" and "poptable" in pcre_dfa_exec.c that must be updated. */  #define LAST_AUTOTAB_RIGHT_OP  OP_DOLLM
1881    
1882  enum {  enum {
1883    OP_END,            /* 0 End of pattern */    OP_END,            /* 0 End of pattern */
# Line 1567  enum { Line 1896  enum {
1896    OP_NOT_WORDCHAR,       /* 10 \W */    OP_NOT_WORDCHAR,       /* 10 \W */
1897    OP_WORDCHAR,           /* 11 \w */    OP_WORDCHAR,           /* 11 \w */
1898    
1899    OP_ANY,            /* 12 Match any character except newline */    OP_ANY,            /* 12 Match any character except newline (\N) */
1900    OP_ALLANY,         /* 13 Match any character */    OP_ALLANY,         /* 13 Match any character */
1901    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 */
1902    OP_NOTPROP,        /* 15 \P (not Unicode property) */    OP_NOTPROP,        /* 15 \P (not Unicode property) */
# Line 1578  enum { Line 1907  enum {
1907    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */
1908    OP_VSPACE,         /* 21 \v (vertical whitespace) */    OP_VSPACE,         /* 21 \v (vertical whitespace) */
1909    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */
1910    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) */
1911    OP_EOD,            /* 24 End of data: \z */    OP_EOD,            /* 24 End of data (\z) */
1912    
1913      /* Line end assertions */
1914    
1915      OP_DOLL,           /* 25 End of line - not multiline */
1916      OP_DOLLM,          /* 26 End of line - multiline */
1917      OP_CIRC,           /* 27 Start of line - not multiline */
1918      OP_CIRCM,          /* 28 Start of line - multiline */
1919    
1920      /* Single characters; caseful must precede the caseless ones */
1921    
   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 */  
1922    OP_CHAR,           /* 29 Match one character, casefully */    OP_CHAR,           /* 29 Match one character, casefully */
1923    OP_CHARI,          /* 30 Match one character, caselessly */    OP_CHARI,          /* 30 Match one character, caselessly */
1924    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 1927  enum {
1927    /* 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
1928    the offset from the first one is used to generate the others. */    the offset from the first one is used to generate the others. */
1929    
1930    /**** Single characters, caseful, must precede the caseless ones ****/    /* Repeated characters; caseful must precede the caseless ones */
1931    
1932    OP_STAR,           /* 33 The maximizing and minimizing versions of */    OP_STAR,           /* 33 The maximizing and minimizing versions of */
1933    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 1945  enum {
1945    OP_POSQUERY,       /* 44 Posesssified query, caseful */    OP_POSQUERY,       /* 44 Posesssified query, caseful */
1946    OP_POSUPTO,        /* 45 Possessified upto, caseful */    OP_POSUPTO,        /* 45 Possessified upto, caseful */
1947    
1948    /**** Single characters, caseless, must follow the caseful ones */    /* Repeated characters; caseless must follow the caseful ones */
1949    
1950    OP_STARI,          /* 46 */    OP_STARI,          /* 46 */
1951    OP_MINSTARI,       /* 47 */    OP_MINSTARI,       /* 47 */
# Line 1629  enum { Line 1963  enum {
1963    OP_POSQUERYI,      /* 57 Posesssified query, caseless */    OP_POSQUERYI,      /* 57 Posesssified query, caseless */
1964    OP_POSUPTOI,       /* 58 Possessified upto, caseless */    OP_POSUPTOI,       /* 58 Possessified upto, caseless */
1965    
1966    /**** The negated ones must follow the non-negated ones, and match them ****/    /* The negated ones must follow the non-negated ones, and match them */
1967    /**** Negated single character, caseful; must precede the caseless ones ****/    /* Negated repeated character, caseful; must precede the caseless ones */
1968    
1969    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */
1970    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 1982  enum {
1982    OP_NOTPOSQUERY,    /* 70 */    OP_NOTPOSQUERY,    /* 70 */
1983    OP_NOTPOSUPTO,     /* 71 */    OP_NOTPOSUPTO,     /* 71 */
1984    
1985    /**** Negated single character, caseless; must follow the caseful ones ****/    /* Negated repeated character, caseless; must follow the caseful ones */
1986    
1987    OP_NOTSTARI,       /* 72 */    OP_NOTSTARI,       /* 72 */
1988    OP_NOTMINSTARI,    /* 73 */    OP_NOTMINSTARI,    /* 73 */
# Line 1666  enum { Line 2000  enum {
2000    OP_NOTPOSQUERYI,   /* 83 */    OP_NOTPOSQUERYI,   /* 83 */
2001    OP_NOTPOSUPTOI,    /* 84 */    OP_NOTPOSUPTOI,    /* 84 */
2002    
2003    /**** Character types ****/    /* Character types */
2004    
2005    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */
2006    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 2031  enum {
2031    OP_CRRANGE,        /* 104 These are different to the three sets above. */    OP_CRRANGE,        /* 104 These are different to the three sets above. */
2032    OP_CRMINRANGE,     /* 105 */    OP_CRMINRANGE,     /* 105 */
2033    
2034      OP_CRPOSSTAR,      /* 106 Possessified versions */
2035      OP_CRPOSPLUS,      /* 107 */
2036      OP_CRPOSQUERY,     /* 108 */
2037      OP_CRPOSRANGE,     /* 109 */
2038    
2039    /* End of quantifier opcodes */    /* End of quantifier opcodes */
2040    
2041    OP_CLASS,          /* 106 Match a character class, chars < 256 only */    OP_CLASS,          /* 110 Match a character class, chars < 256 only */
2042    OP_NCLASS,         /* 107 Same, but the bitmap was created from a negative    OP_NCLASS,         /* 111 Same, but the bitmap was created from a negative
2043                                class - the difference is relevant only when a                                class - the difference is relevant only when a
2044                                character > 255 is encountered. */                                character > 255 is encountered. */
2045    OP_XCLASS,         /* 108 Extended class for handling > 255 chars within the    OP_XCLASS,         /* 112 Extended class for handling > 255 chars within the
2046                                class. This does both positive and negative. */                                class. This does both positive and negative. */
2047    OP_REF,            /* 109 Match a back reference, casefully */    OP_REF,            /* 113 Match a back reference, casefully */
2048    OP_REFI,           /* 110 Match a back reference, caselessly */    OP_REFI,           /* 114 Match a back reference, caselessly */
2049    OP_RECURSE,        /* 111 Match a numbered subpattern (possibly recursive) */    OP_DNREF,          /* 115 Match a duplicate name backref, casefully */
2050    OP_CALLOUT,        /* 112 Call out to external function if provided */    OP_DNREFI,         /* 116 Match a duplicate name backref, caselessly */
2051      OP_RECURSE,        /* 117 Match a numbered subpattern (possibly recursive) */
2052    OP_ALT,            /* 113 Start of alternation */    OP_CALLOUT,        /* 118 Call out to external function if provided */
2053    OP_KET,            /* 114 End of group that doesn't have an unbounded repeat */  
2054    OP_KETRMAX,        /* 115 These two must remain together and in this */    OP_ALT,            /* 119 Start of alternation */
2055    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 */
2056    OP_KETRPOS,        /* 117 Possessive unlimited repeat. */    OP_KETRMAX,        /* 121 These two must remain together and in this */
2057      OP_KETRMIN,        /* 122 order. They are for groups the repeat for ever. */
2058      OP_KETRPOS,        /* 123 Possessive unlimited repeat. */
2059    
2060    /* 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
2061    asserts must remain in order. */    asserts must remain in order. */
2062    
2063    OP_REVERSE,        /* 118 Move pointer back - used in lookbehind assertions */    OP_REVERSE,        /* 124 Move pointer back - used in lookbehind assertions */
2064    OP_ASSERT,         /* 119 Positive lookahead */    OP_ASSERT,         /* 125 Positive lookahead */
2065    OP_ASSERT_NOT,     /* 120 Negative lookahead */    OP_ASSERT_NOT,     /* 126 Negative lookahead */
2066    OP_ASSERTBACK,     /* 121 Positive lookbehind */    OP_ASSERTBACK,     /* 127 Positive lookbehind */
2067    OP_ASSERTBACK_NOT, /* 122 Negative lookbehind */    OP_ASSERTBACK_NOT, /* 128 Negative lookbehind */
2068    
2069    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately
2070    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
2071    subpattern that isn't an assertion. The POS versions must immediately follow    subpattern that isn't an assertion. The POS versions must immediately follow
2072    the non-POS versions in each case. */    the non-POS versions in each case. */
2073    
2074    OP_ONCE,           /* 123 Atomic group, contains captures */    OP_ONCE,           /* 129 Atomic group, contains captures */
2075    OP_ONCE_NC,        /* 124 Atomic group containing no captures */    OP_ONCE_NC,        /* 130 Atomic group containing no captures */
2076    OP_BRA,            /* 125 Start of non-capturing bracket */    OP_BRA,            /* 131 Start of non-capturing bracket */
2077    OP_BRAPOS,         /* 126 Ditto, with unlimited, possessive repeat */    OP_BRAPOS,         /* 132 Ditto, with unlimited, possessive repeat */
2078    OP_CBRA,           /* 127 Start of capturing bracket */    OP_CBRA,           /* 133 Start of capturing bracket */
2079    OP_CBRAPOS,        /* 128 Ditto, with unlimited, possessive repeat */    OP_CBRAPOS,        /* 134 Ditto, with unlimited, possessive repeat */
2080    OP_COND,           /* 129 Conditional group */    OP_COND,           /* 135 Conditional group */
2081    
2082    /* 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
2083    check for >= SBRA to distinguish the two sets. */    check for >= SBRA to distinguish the two sets. */
2084    
2085    OP_SBRA,           /* 130 Start of non-capturing bracket, check empty  */    OP_SBRA,           /* 136 Start of non-capturing bracket, check empty  */
2086    OP_SBRAPOS,        /* 131 Ditto, with unlimited, possessive repeat */    OP_SBRAPOS,        /* 137 Ditto, with unlimited, possessive repeat */
2087    OP_SCBRA,          /* 132 Start of capturing bracket, check empty */    OP_SCBRA,          /* 138 Start of capturing bracket, check empty */
2088    OP_SCBRAPOS,       /* 133 Ditto, with unlimited, possessive repeat */    OP_SCBRAPOS,       /* 139 Ditto, with unlimited, possessive repeat */
2089    OP_SCOND,          /* 134 Conditional group, check empty */    OP_SCOND,          /* 140 Conditional group, check empty */
2090    
2091    /* The next two pairs must (respectively) be kept together. */    /* The next two pairs must (respectively) be kept together. */
2092    
2093    OP_CREF,           /* 135 Used to hold a capture number as condition */    OP_CREF,           /* 141 Used to hold a capture number as condition */
2094    OP_NCREF,          /* 136 Same, but generated by a name reference*/    OP_DNCREF,         /* 142 Used to point to duplicate names as a condition */
2095    OP_RREF,           /* 137 Used to hold a recursion number as condition */    OP_RREF,           /* 143 Used to hold a recursion number as condition */
2096    OP_NRREF,          /* 138 Same, but generated by a name reference*/    OP_DNRREF,         /* 144 Used to point to duplicate names as a condition */
2097    OP_DEF,            /* 139 The DEFINE condition */    OP_DEF,            /* 145 The DEFINE condition */
2098    
2099    OP_BRAZERO,        /* 140 These two must remain together and in this */    OP_BRAZERO,        /* 146 These two must remain together and in this */
2100    OP_BRAMINZERO,     /* 141 order. */    OP_BRAMINZERO,     /* 147 order. */
2101    OP_BRAPOSZERO,     /* 142 */    OP_BRAPOSZERO,     /* 148 */
2102    
2103    /* These are backtracking control verbs */    /* These are backtracking control verbs */
2104    
2105    OP_MARK,           /* 143 always has an argument */    OP_MARK,           /* 149 always has an argument */
2106    OP_PRUNE,          /* 144 */    OP_PRUNE,          /* 150 */
2107    OP_PRUNE_ARG,      /* 145 same, but with argument */    OP_PRUNE_ARG,      /* 151 same, but with argument */
2108    OP_SKIP,           /* 146 */    OP_SKIP,           /* 152 */
2109    OP_SKIP_ARG,       /* 147 same, but with argument */    OP_SKIP_ARG,       /* 153 same, but with argument */
2110    OP_THEN,           /* 148 */    OP_THEN,           /* 154 */
2111    OP_THEN_ARG,       /* 149 same, but with argument */    OP_THEN_ARG,       /* 155 same, but with argument */
2112    OP_COMMIT,         /* 150 */    OP_COMMIT,         /* 156 */
2113    
2114    /* These are forced failure and success verbs */    /* These are forced failure and success verbs */
2115    
2116    OP_FAIL,           /* 151 */    OP_FAIL,           /* 157 */
2117    OP_ACCEPT,         /* 152 */    OP_ACCEPT,         /* 158 */
2118    OP_ASSERT_ACCEPT,  /* 153 Used inside assertions */    OP_ASSERT_ACCEPT,  /* 159 Used inside assertions */
2119    OP_CLOSE,          /* 154 Used before OP_ACCEPT to close open captures */    OP_CLOSE,          /* 160 Used before OP_ACCEPT to close open captures */
2120    
2121    /* This is used to skip a subpattern with a {0} quantifier */    /* This is used to skip a subpattern with a {0} quantifier */
2122    
2123    OP_SKIPZERO,       /* 155 */    OP_SKIPZERO,       /* 161 */
2124    
2125    /* 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
2126    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 2131  enum {
2131    
2132  /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro  /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro
2133  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
2134  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
2135    pcre_dfa_exec.c that must be updated. */
2136    
2137    
2138  /* 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 2145  some cases doesn't actually use these na
2145    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \
2146    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \
2147    "extuni",  "\\Z", "\\z",                                        \    "extuni",  "\\Z", "\\z",                                        \
2148    "^", "^", "$", "$", "char", "chari", "not", "noti",             \    "$", "$", "^", "^", "char", "chari", "not", "noti",             \
2149    "*", "*?", "+", "+?", "?", "??",                                \    "*", "*?", "+", "+?", "?", "??",                                \
2150    "{", "{", "{",                                                  \    "{", "{", "{",                                                  \
2151    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
# Line 1819  some cases doesn't actually use these na Line 2161  some cases doesn't actually use these na
2161    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
2162    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2163    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \
2164    "class", "nclass", "xclass", "Ref", "Refi",                     \    "*+","++", "?+", "{",                                           \
2165      "class", "nclass", "xclass", "Ref", "Refi", "DnRef", "DnRefi",  \
2166    "Recurse", "Callout",                                           \    "Recurse", "Callout",                                           \
2167    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \
2168    "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 2171  some cases doesn't actually use these na
2171    "Cond",                                                         \    "Cond",                                                         \
2172    "SBra", "SBraPos", "SCBra", "SCBraPos",                         \    "SBra", "SBraPos", "SCBra", "SCBraPos",                         \
2173    "SCond",                                                        \    "SCond",                                                        \
2174    "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def",   \    "Cond ref", "Cond dnref", "Cond rec", "Cond dnrec", "Cond def", \
2175    "Brazero", "Braminzero", "Braposzero",                          \    "Brazero", "Braminzero", "Braposzero",                          \
2176    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \
2177    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \
# Line 1853  in UTF-8 mode. The code that uses this t Line 2196  in UTF-8 mode. The code that uses this t
2196    3, 3,                          /* \P, \p                                 */ \    3, 3,                          /* \P, \p                                 */ \
2197    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \
2198    1,                             /* \X                                     */ \    1,                             /* \X                                     */ \
2199    1, 1, 1, 1, 1, 1,              /* \Z, \z, ^, ^M, $, $M                   */ \    1, 1, 1, 1, 1, 1,              /* \Z, \z, $, $M ^, ^M                    */ \
2200    2,                             /* Char  - the minimum length             */ \    2,                             /* Char  - the minimum length             */ \
2201    2,                             /* Chari  - the minimum length            */ \    2,                             /* Chari  - the minimum length            */ \
2202    2,                             /* not                                    */ \    2,                             /* not                                    */ \
# Line 1884  in UTF-8 mode. The code that uses this t Line 2227  in UTF-8 mode. The code that uses this t
2227    /* Character class & ref repeats                                         */ \    /* Character class & ref repeats                                         */ \
2228    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \
2229    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \
2230      1, 1, 1, 1+2*IMM2_SIZE,        /* Possessive *+, ++, ?+, CRPOSRANGE      */ \
2231    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \
2232    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \
2233    0,                             /* XCLASS - variable length               */ \    0,                             /* XCLASS - variable length               */ \
2234    1+IMM2_SIZE,                   /* REF                                    */ \    1+IMM2_SIZE,                   /* REF                                    */ \
2235    1+IMM2_SIZE,                   /* REFI                                   */ \    1+IMM2_SIZE,                   /* REFI                                   */ \
2236      1+2*IMM2_SIZE,                 /* DNREF                                  */ \
2237      1+2*IMM2_SIZE,                 /* DNREFI                                 */ \
2238    1+LINK_SIZE,                   /* RECURSE                                */ \    1+LINK_SIZE,                   /* RECURSE                                */ \
2239    2+2*LINK_SIZE,                 /* CALLOUT                                */ \    2+2*LINK_SIZE,                 /* CALLOUT                                */ \
2240    1+LINK_SIZE,                   /* Alt                                    */ \    1+LINK_SIZE,                   /* Alt                                    */ \
# Line 1913  in UTF-8 mode. The code that uses this t Line 2259  in UTF-8 mode. The code that uses this t
2259    1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \    1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \
2260    1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \    1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \
2261    1+LINK_SIZE,                   /* SCOND                                  */ \    1+LINK_SIZE,                   /* SCOND                                  */ \
2262    1+IMM2_SIZE, 1+IMM2_SIZE,      /* CREF, NCREF                            */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* CREF, DNCREF                           */ \
2263    1+IMM2_SIZE, 1+IMM2_SIZE,      /* RREF, NRREF                            */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* RREF, DNRREF                           */ \
2264    1,                             /* DEF                                    */ \    1,                             /* DEF                                    */ \
2265    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \
2266    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 2269  in UTF-8 mode. The code that uses this t
2269    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \
2270    1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */    1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */
2271    
2272  /* 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. */  
2273    
2274  #define RREF_ANY  0xffff  #define RREF_ANY  0xffff
2275    
# Line 1939  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 2284  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
2284         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
2285         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
2286         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
2287         ERR70, ERR71, ERR72, ERR73, ERR74, ERRCOUNT };         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERR78, ERR79,
2288           ERR80, ERR81, ERR82, ERR83, ERR84, ERR85, ERR86, ERRCOUNT };
2289    
2290    /* JIT compiling modes. The function list is indexed by them. */
2291    
2292    enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
2293           JIT_NUMBER_OF_COMPILE_MODES };
2294    
2295  /* 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
2296  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
2297  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
2298  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
2299  be well. For the case of compiled-on-4 and run-on-8, we include an extra  be well.
2300  pointer that is always NULL. For future-proofing, a few dummy fields were  
2301  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
2302  there is only one left now.  compiled-on-4 and run-on-8, we include an extra pointer that is always NULL so
2303    that there are an even number of pointers which therefore are a multiple of 8
2304  NOTE NOTE NOTE:  bytes.
2305  Because people can now save and re-use compiled patterns, any additions to this  
2306  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
2307  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
2308  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
2309  NOTE NOTE NOTE  variants.
2310    
2311    *** WARNING ***
2312    When new fields are added to these structures, remember to adjust the code in
2313    pcre_byte_order.c that is concerned with swapping the byte order of the fields
2314    when a compiled regex is reloaded on a host with different endianness.
2315    *** WARNING ***
2316    There is also similar byte-flipping code in pcretest.c, which is used for
2317    testing the byte-flipping features. It must also be kept in step.
2318    *** WARNING ***
2319  */  */
2320    
2321  typedef struct real_pcre {  typedef struct real_pcre8_or_16 {
2322    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2323    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2324    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2325    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2326    pcre_uint16 dummy1;             /* For future use */    pcre_uint32 limit_match;        /* Limit set from regex */
2327    pcre_uint16 top_bracket;    pcre_uint32 limit_recursion;    /* Limit set from regex */
   pcre_uint16 top_backref;  
2328    pcre_uint16 first_char;         /* Starting character */    pcre_uint16 first_char;         /* Starting character */
2329    pcre_uint16 req_char;           /* This character must be seen */    pcre_uint16 req_char;           /* This character must be seen */
2330      pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2331      pcre_uint16 top_bracket;        /* Highest numbered group */
2332      pcre_uint16 top_backref;        /* Highest numbered back reference */
2333    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2334    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2335    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2336    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2337      pcre_uint16 dummy1;             /* To ensure size is a multiple of 8 */
2338      pcre_uint16 dummy2;             /* To ensure size is a multiple of 8 */
2339      pcre_uint16 dummy3;             /* To ensure size is a multiple of 8 */
2340      const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2341      void             *nullpad;      /* NULL padding */
2342    } real_pcre8_or_16;
2343    
2344    typedef struct real_pcre8_or_16 real_pcre;
2345    typedef struct real_pcre8_or_16 real_pcre16;
2346    
2347    typedef struct real_pcre32 {
2348      pcre_uint32 magic_number;
2349      pcre_uint32 size;               /* Total that was malloced */
2350      pcre_uint32 options;            /* Public options */
2351      pcre_uint32 flags;              /* Private flags */
2352      pcre_uint32 limit_match;        /* Limit set from regex */
2353      pcre_uint32 limit_recursion;    /* Limit set from regex */
2354      pcre_uint32 first_char;         /* Starting character */
2355      pcre_uint32 req_char;           /* This character must be seen */
2356      pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2357      pcre_uint16 top_bracket;        /* Highest numbered group */
2358      pcre_uint16 top_backref;        /* Highest numbered back reference */
2359      pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2360      pcre_uint16 name_entry_size;    /* Size of any name items */
2361      pcre_uint16 name_count;         /* Number of name items */
2362      pcre_uint16 ref_count;          /* Reference count */
2363      pcre_uint16 dummy;              /* To ensure size is a multiple of 8 */
2364    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2365    const pcre_uint8 *nullpad;      /* NULL padding */    void             *nullpad;      /* NULL padding */
2366  } real_pcre;  } real_pcre32;
2367    
2368    #if defined COMPILE_PCRE8
2369    #define REAL_PCRE real_pcre
2370    #elif defined COMPILE_PCRE16
2371    #define REAL_PCRE real_pcre16
2372    #elif defined COMPILE_PCRE32
2373    #define REAL_PCRE real_pcre32
2374    #endif
2375    
2376    /* Assert that the size of REAL_PCRE is divisible by 8 */
2377    typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];
2378    
2379    /* Needed in pcretest to access some fields in the real_pcre* structures
2380     * directly. They're unified for 8/16/32 bits since the structs only differ
2381     * after these fields; if that ever changes, need to fork those defines into
2382     * 8/16 and 32 bit versions. */
2383    #define REAL_PCRE_MAGIC(re)     (((REAL_PCRE*)re)->magic_number)
2384    #define REAL_PCRE_SIZE(re)      (((REAL_PCRE*)re)->size)
2385    #define REAL_PCRE_OPTIONS(re)   (((REAL_PCRE*)re)->options)
2386    #define REAL_PCRE_FLAGS(re)     (((REAL_PCRE*)re)->flags)
2387    
2388  /* 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
2389  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 2406  typedef struct open_capitem {
2406    pcre_uint16 flag;             /* Set TRUE if recursive back ref */    pcre_uint16 flag;             /* Set TRUE if recursive back ref */
2407  } open_capitem;  } open_capitem;
2408    
2409    /* Structure for building a list of named groups during the first pass of
2410    compiling. */
2411    
2412    typedef struct named_group {
2413      const pcre_uchar  *name;          /* Points to the name in the pattern */
2414      int                length;        /* Length of the name */
2415      pcre_uint32        number;        /* Group number */
2416    } named_group;
2417    
2418  /* Structure for passing "static" information around between the functions  /* Structure for passing "static" information around between the functions
2419  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
2420    
# Line 2010  typedef struct compile_data { Line 2427  typedef struct compile_data {
2427    const pcre_uchar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
2428    const pcre_uchar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
2429    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 */  
2430    pcre_uchar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
2431      open_capitem *open_caps;          /* Chain of open capture items */
2432      named_group *named_groups;        /* Points to vector in pre-compile */
2433    pcre_uchar *name_table;           /* The name/number table */    pcre_uchar *name_table;           /* The name/number table */
2434    int  names_found;                 /* Number of entries so far */    int  names_found;                 /* Number of entries so far */
2435    int  name_entry_size;             /* Size of each entry */    int  name_entry_size;             /* Size of each entry */
2436      int  named_group_list_size;       /* Number of entries in the list */
2437    int  workspace_size;              /* Size of workspace */    int  workspace_size;              /* Size of workspace */
2438    int  bracount;                    /* Count of capturing parens as we compile */    unsigned int bracount;            /* Count of capturing parens as we compile */
2439    int  final_bracount;              /* Saved value after first pass */    int  final_bracount;              /* Saved value after first pass */
2440      int  max_lookbehind;              /* Maximum lookbehind (characters) */
2441    int  top_backref;                 /* Maximum back reference */    int  top_backref;                 /* Maximum back reference */
2442    unsigned int backref_map;         /* Bitmap of low back refs */    unsigned int backref_map;         /* Bitmap of low back refs */
2443      unsigned int namedrefcount;       /* Number of backreferences by name */
2444      int  parens_depth;                /* Depth of nested parentheses */
2445    int  assert_depth;                /* Depth of nested assertions */    int  assert_depth;                /* Depth of nested assertions */
2446    int  external_options;            /* External (initial) options */    pcre_uint32 external_options;     /* External (initial) options */
2447    int  external_flags;              /* External flag bits to be set */    pcre_uint32 external_flags;       /* External flag bits to be set */
2448    int  req_varyopt;                 /* "After variable item" flag for reqbyte */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2449    BOOL had_accept;                  /* (*ACCEPT) encountered */    BOOL had_accept;                  /* (*ACCEPT) encountered */
2450      BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
2451    BOOL check_lookbehind;            /* Lookbehinds need later checking */    BOOL check_lookbehind;            /* Lookbehinds need later checking */
2452      BOOL dupnames;                    /* Duplicate names exist */
2453      BOOL iscondassert;                /* Next assert is a condition */
2454    int  nltype;                      /* Newline type */    int  nltype;                      /* Newline type */
2455    int  nllen;                       /* Newline string length */    int  nllen;                       /* Newline string length */
2456    pcre_uchar nl[4];                 /* Newline string when fixed length */    pcre_uchar nl[4];                 /* Newline string when fixed length */
# Line 2039  typedef struct branch_chain { Line 2464  typedef struct branch_chain {
2464    pcre_uchar *current_branch;    pcre_uchar *current_branch;
2465  } branch_chain;  } branch_chain;
2466    
2467    /* Structure for mutual recursion detection. */
2468    
2469    typedef struct recurse_check {
2470      struct recurse_check *prev;
2471      const pcre_uchar *group;
2472    } recurse_check;
2473    
2474  /* Structure for items in a linked list that represents an explicit recursive  /* Structure for items in a linked list that represents an explicit recursive
2475  call within the pattern; used by pcre_exec(). */  call within the pattern; used by pcre_exec(). */
2476    
2477  typedef struct recursion_info {  typedef struct recursion_info {
2478    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
2479    int group_num;                  /* Number of group that was called */    unsigned int group_num;         /* Number of group that was called */
2480    int *offset_save;               /* Pointer to start of saved offsets */    int *offset_save;               /* Pointer to start of saved offsets */
2481    int saved_max;                  /* Number of saved offsets */    int saved_max;                  /* Number of saved offsets */
2482      int saved_capture_last;         /* Last capture number */
2483    PCRE_PUCHAR subject_position;   /* Position at start of recursion */    PCRE_PUCHAR subject_position;   /* Position at start of recursion */
2484  } recursion_info;  } recursion_info;
2485    
# Line 2083  typedef struct match_data { Line 2516  typedef struct match_data {
2516    int    nllen;                   /* Newline string length */    int    nllen;                   /* Newline string length */
2517    int    name_count;              /* Number of names in name table */    int    name_count;              /* Number of names in name table */
2518    int    name_entry_size;         /* Size of entry in names table */    int    name_entry_size;         /* Size of entry in names table */
2519      unsigned int skip_arg_count;    /* For counting SKIP_ARGs */
2520      unsigned int ignore_skip_arg;   /* For re-run when SKIP arg name not found */
2521    pcre_uchar *name_table;         /* Table of names */    pcre_uchar *name_table;         /* Table of names */
2522    pcre_uchar nl[4];               /* Newline string when fixed */    pcre_uchar nl[4];               /* Newline string when fixed */
2523    const  pcre_uint8 *lcc;         /* Points to lower casing table */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
2524    const  pcre_uint8 *fcc;         /* Points to case-flipping table */    const  pcre_uint8 *fcc;         /* Points to case-flipping table */
2525    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 */  
2526    BOOL   notbol;                  /* NOTBOL flag */    BOOL   notbol;                  /* NOTBOL flag */
2527    BOOL   noteol;                  /* NOTEOL flag */    BOOL   noteol;                  /* NOTEOL flag */
2528    BOOL   utf;                     /* UTF-8 / UTF-16 flag */    BOOL   utf;                     /* UTF-8 / UTF-16 flag */
# Line 2100  typedef struct match_data { Line 2534  typedef struct match_data {
2534    BOOL   hitend;                  /* Hit the end of the subject at some point */    BOOL   hitend;                  /* Hit the end of the subject at some point */
2535    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
2536    BOOL   hasthen;                 /* Pattern contains (*THEN) */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
   BOOL   ignore_skip_arg;         /* For re-run when SKIP name not found */  
2537    const  pcre_uchar *start_code;  /* For use when recursing */    const  pcre_uchar *start_code;  /* For use when recursing */
2538    PCRE_PUCHAR start_subject;      /* Start of the subject string */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
2539    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 2542  typedef struct match_data {
2542    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */
2543    int    partial;                 /* PARTIAL options */    int    partial;                 /* PARTIAL options */
2544    int    end_offset_top;          /* Highwater mark at end of match */    int    end_offset_top;          /* Highwater mark at end of match */
2545    int    capture_last;            /* Most recent capture number */    pcre_int32 capture_last;        /* Most recent capture number + overflow flag */
2546    int    start_offset;            /* The start offset value */    int    start_offset;            /* The start offset value */
2547    int    match_function_type;     /* Set for certain special calls of MATCH() */    int    match_function_type;     /* Set for certain special calls of MATCH() */
2548    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */
# Line 2119  typedef struct match_data { Line 2552  typedef struct match_data {
2552    const  pcre_uchar *mark;        /* Mark pointer to pass back on success */    const  pcre_uchar *mark;        /* Mark pointer to pass back on success */
2553    const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */    const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */
2554    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 */
2555    #ifdef NO_RECURSE
2556      void  *match_frames_base;       /* For remembering malloc'd frames */
2557    #endif
2558  } match_data;  } match_data;
2559    
2560  /* 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 2609  total length. */
2609  #define ctypes_offset (cbits_offset + cbit_length)  #define ctypes_offset (cbits_offset + cbit_length)
2610  #define tables_length (ctypes_offset + 256)  #define tables_length (ctypes_offset + 256)
2611    
2612  /* Internal function prefix */  /* Internal function and data prefixes. */
2613    
2614  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2615    #ifndef PUBL
2616  #define PUBL(name) pcre_##name  #define PUBL(name) pcre_##name
2617    #endif
2618    #ifndef PRIV
2619  #define PRIV(name) _pcre_##name  #define PRIV(name) _pcre_##name
2620  #else  #endif
2621  #ifdef COMPILE_PCRE16  #elif defined COMPILE_PCRE16
2622    #ifndef PUBL
2623  #define PUBL(name) pcre16_##name  #define PUBL(name) pcre16_##name
2624    #endif
2625    #ifndef PRIV
2626  #define PRIV(name) _pcre16_##name  #define PRIV(name) _pcre16_##name
2627    #endif
2628    #elif defined COMPILE_PCRE32
2629    #ifndef PUBL
2630    #define PUBL(name) pcre32_##name
2631    #endif
2632    #ifndef PRIV
2633    #define PRIV(name) _pcre32_##name
2634    #endif
2635  #else  #else
2636  #error Unsupported compiling mode  #error Unsupported compiling mode
2637  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
 #endif /* COMPILE_PCRE8 */  
2638    
2639  /* 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
2640  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 2654  but are not part of the PCRE public API.
2654  pcre_tables.c module. */  pcre_tables.c module. */
2655    
2656  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
   
2657  extern const int            PRIV(utf8_table1)[];  extern const int            PRIV(utf8_table1)[];
2658  extern const int            PRIV(utf8_table1_size);  extern const int            PRIV(utf8_table1_size);
2659  extern const int            PRIV(utf8_table2)[];  extern const int            PRIV(utf8_table2)[];
2660  extern const int            PRIV(utf8_table3)[];  extern const int            PRIV(utf8_table3)[];
2661  extern const pcre_uint8     PRIV(utf8_table4)[];  extern const pcre_uint8     PRIV(utf8_table4)[];
   
2662  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
2663    
2664  extern const char           PRIV(utt_names)[];  extern const char           PRIV(utt_names)[];
2665  extern const ucp_type_table PRIV(utt)[];  extern const ucp_type_table PRIV(utt)[];
2666  extern const int            PRIV(utt_size);  extern const int            PRIV(utt_size);
2667    
2668    extern const pcre_uint8     PRIV(OP_lengths)[];
2669  extern const pcre_uint8     PRIV(default_tables)[];  extern const pcre_uint8     PRIV(default_tables)[];
2670    
2671  extern const pcre_uint8     PRIV(OP_lengths)[];  extern const pcre_uint32    PRIV(hspace_list)[];
2672    extern const pcre_uint32    PRIV(vspace_list)[];
2673    
2674    
2675  /* 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 2677  one of the exported public functions. Th
2677  sense, but are not part of the PCRE public API. */  sense, but are not part of the PCRE public API. */
2678    
2679  /* String comparison functions. */  /* String comparison functions. */
2680  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2681    
2682  #define STRCMP_UC_UC(str1, str2) \  #define STRCMP_UC_UC(str1, str2) \
2683    strcmp((char *)(str1), (char *)(str2))    strcmp((char *)(str1), (char *)(str2))
# Line 2240  sense, but are not part of the PCRE publ Line 2689  sense, but are not part of the PCRE publ
2689    strncmp((char *)(str1), (str2), (num))    strncmp((char *)(str1), (str2), (num))
2690  #define STRLEN_UC(str) strlen((const char *)str)  #define STRLEN_UC(str) strlen((const char *)str)
2691    
2692  #else  #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2693    
2694  extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,  extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,
2695                             const pcre_uchar *);                             const pcre_uchar *);
# Line 2262  extern unsigned int      PRIV(strlen_uc) Line 2711  extern unsigned int      PRIV(strlen_uc)
2711    PRIV(strncmp_uc_c8)((str1), (str2), (num))    PRIV(strncmp_uc_c8)((str1), (str2), (num))
2712  #define STRLEN_UC(str) PRIV(strlen_uc)(str)  #define STRLEN_UC(str) PRIV(strlen_uc)(str)
2713    
2714  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE[8|16|32] */
2715    
2716    #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16
2717    
2718    #define STRCMP_UC_UC_TEST(str1, str2) STRCMP_UC_UC(str1, str2)
2719    #define STRCMP_UC_C8_TEST(str1, str2) STRCMP_UC_C8(str1, str2)
2720    
2721    #elif defined COMPILE_PCRE32
2722    
2723    extern int               PRIV(strcmp_uc_uc_utf)(const pcre_uchar *,
2724                               const pcre_uchar *);
2725    extern int               PRIV(strcmp_uc_c8_utf)(const pcre_uchar *,
2726                               const char *);
2727    
2728    #define STRCMP_UC_UC_TEST(str1, str2) \
2729      (utf ? PRIV(strcmp_uc_uc_utf)((str1), (str2)) : PRIV(strcmp_uc_uc)((str1), (str2)))
2730    #define STRCMP_UC_C8_TEST(str1, str2) \
2731      (utf ? PRIV(strcmp_uc_c8_utf)((str1), (str2)) : PRIV(strcmp_uc_c8)((str1), (str2)))
2732    
2733    #endif /* COMPILE_PCRE[8|16|32] */
2734    
2735  extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);  extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);
2736  extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,  extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2737                             int *, BOOL);                             int *, BOOL);
2738  extern int               PRIV(ord2utf)(pcre_uint32, pcre_uchar *);  extern unsigned int      PRIV(ord2utf)(pcre_uint32, pcre_uchar *);
2739  extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);  extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);
2740  extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,  extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2741                             int *, BOOL);                             int *, BOOL);
2742  extern BOOL              PRIV(xclass)(int, const pcre_uchar *, BOOL);  extern BOOL              PRIV(xclass)(pcre_uint32, const pcre_uchar *, BOOL);
2743    
2744  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2745  extern void              PRIV(jit_compile)(const real_pcre *, PUBL(extra) *);  extern void              PRIV(jit_compile)(const REAL_PCRE *,
2746  extern int               PRIV(jit_exec)(const real_pcre *, void *,                             PUBL(extra) *, int);
2747                             const pcre_uchar *, int, int, int, int, int *, int);  extern int               PRIV(jit_exec)(const PUBL(extra) *,
2748                               const pcre_uchar *, int, int, int, int *, int);
2749  extern void              PRIV(jit_free)(void *);  extern void              PRIV(jit_free)(void *);
2750  extern int               PRIV(jit_get_size)(void *);  extern int               PRIV(jit_get_size)(void *);
2751    extern const char*       PRIV(jit_get_target)(void);
2752  #endif  #endif
2753    
2754  /* Unicode character database (UCD) */  /* Unicode character database (UCD) */
2755    
2756  typedef struct {  typedef struct {
2757    pcre_uint8 script;    pcre_uint8 script;     /* ucp_Arabic, etc. */
2758    pcre_uint8 chartype;    pcre_uint8 chartype;   /* ucp_Cc, etc. (general categories) */
2759    pcre_int32 other_case;    pcre_uint8 gbprop;     /* ucp_gbControl, etc. (grapheme break property) */
2760      pcre_uint8 caseset;    /* offset to multichar other cases or zero */
2761      pcre_int32 other_case; /* offset to other case, or zero if none */
2762  } ucd_record;  } ucd_record;
2763    
2764    extern const pcre_uint32 PRIV(ucd_caseless_sets)[];
2765  extern const ucd_record  PRIV(ucd_records)[];  extern const ucd_record  PRIV(ucd_records)[];
2766  extern const pcre_uint8  PRIV(ucd_stage1)[];  extern const pcre_uint8  PRIV(ucd_stage1)[];
2767  extern const pcre_uint16 PRIV(ucd_stage2)[];  extern const pcre_uint16 PRIV(ucd_stage2)[];
2768  extern const int         PRIV(ucp_gentype)[];  extern const pcre_uint32 PRIV(ucp_gentype)[];
2769    extern const pcre_uint32 PRIV(ucp_gbtable)[];
2770  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2771  extern const int         PRIV(ucp_typerange)[];  extern const int         PRIV(ucp_typerange)[];
2772  #endif  #endif
# Line 2302  extern const int         PRIV(ucp_typera Line 2776  extern const int         PRIV(ucp_typera
2776    
2777  #define UCD_BLOCK_SIZE 128  #define UCD_BLOCK_SIZE 128
2778  #define GET_UCD(ch) (PRIV(ucd_records) + \  #define GET_UCD(ch) (PRIV(ucd_records) + \
2779          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(ch) / UCD_BLOCK_SIZE] * \          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \
2780          UCD_BLOCK_SIZE + (ch) % UCD_BLOCK_SIZE])          UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE])
2781    
2782  #define UCD_CHARTYPE(ch)  GET_UCD(ch)->chartype  #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
2783  #define UCD_SCRIPT(ch)    GET_UCD(ch)->script  #define UCD_SCRIPT(ch)      GET_UCD(ch)->script
2784  #define UCD_CATEGORY(ch)  PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]  #define UCD_CATEGORY(ch)    PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]
2785  #define UCD_OTHERCASE(ch) (ch + GET_UCD(ch)->other_case)  #define UCD_GRAPHBREAK(ch)  GET_UCD(ch)->gbprop
2786    #define UCD_CASESET(ch)     GET_UCD(ch)->caseset
2787    #define UCD_OTHERCASE(ch)   ((pcre_uint32)((int)ch + (int)(GET_UCD(ch)->other_case)))
2788    
2789  #endif /* SUPPORT_UCP */  #endif /* SUPPORT_UCP */
2790    

Legend:
Removed from v.850  
changed lines
  Added in v.1567

  ViewVC Help
Powered by ViewVC 1.1.5