/[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 1083 by chpe, Tue Oct 16 15:55:24 2012 UTC revision 1359 by ph10, Tue Sep 3 10:10:59 2013 UTC
# Line 7  Line 7 
7  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
8    
9                         Written by Philip Hazel                         Written by Philip Hazel
10             Copyright (c) 1997-2012 University of Cambridge             Copyright (c) 1997-2013 University of Cambridge
11    
12  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
13  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  depending on the PRIV macro. */ Line 53  depending on 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  #if !defined COMPILE_PCRE16 && ! defined COMPILE_PCRE32  
57    #if !defined COMPILE_PCRE16 && !defined COMPILE_PCRE32
58  #define COMPILE_PCRE8  #define COMPILE_PCRE8
59  #endif  #endif
60    
# 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 272  typedef pcre_uint32 pcre_uchar; Line 287  typedef pcre_uint32 pcre_uchar;
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    
 /* Assert that pcre_uchar32 is a 32-bit type */  
 typedef int __assert_pcre_uchar32_size[sizeof(pcre_uchar) == 4 ? 1 : -1];  
   
290  #else  #else
291  #error Unsupported compiling mode  #error Unsupported compiling mode
292  #endif /* COMPILE_PCRE[8|16|32] */  #endif /* COMPILE_PCRE[8|16|32] */
# Line 304  start/end of string field names are. */ Line 316  start/end of string field names are. */
316         &(NLBLOCK->nllen), utf)) \         &(NLBLOCK->nllen), utf)) \
317      : \      : \
318      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
319       (p)[0] == NLBLOCK->nl[0] && \       RAWUCHARTEST(p) == NLBLOCK->nl[0] && \
320       (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || RAWUCHARTEST(p+1) == NLBLOCK->nl[1])       \
321      ) \      ) \
322    )    )
323    
# Line 318  start/end of string field names are. */ Line 330  start/end of string field names are. */
330         &(NLBLOCK->nllen), utf)) \         &(NLBLOCK->nllen), utf)) \
331      : \      : \
332      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
333       (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \       RAWUCHARTEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] &&              \
334       (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || RAWUCHARTEST(p - NLBLOCK->nllen + 1) == NLBLOCK->nl[1]) \
335      ) \      ) \
336    )    )
337    
# Line 344  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 523  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  #elif defined COMPILE_PCRE16  #elif defined COMPILE_PCRE16
551    
# Line 579  we don't even define them. */ Line 600  we don't even define them. */
600  #define GETCHARINC(c, eptr) c = *eptr++;  #define GETCHARINC(c, eptr) c = *eptr++;
601  #define GETCHARINCTEST(c, eptr) c = *eptr++;  #define GETCHARINCTEST(c, eptr) c = *eptr++;
602  #define GETCHARLEN(c, eptr, len) c = *eptr;  #define GETCHARLEN(c, eptr, len) c = *eptr;
603    #define RAWUCHAR(eptr) (*(eptr))
604    #define RAWUCHARINC(eptr) (*(eptr)++)
605    #define RAWUCHARTEST(eptr) (*(eptr))
606    #define RAWUCHARINCTEST(eptr) (*(eptr)++)
607  /* #define GETCHARLENTEST(c, eptr, len) */  /* #define GETCHARLENTEST(c, eptr, len) */
608  /* #define BACKCHAR(eptr) */  /* #define BACKCHAR(eptr) */
609  /* #define FORWARDCHAR(eptr) */  /* #define FORWARDCHAR(eptr) */
# Line 586  we don't even define them. */ Line 611  we don't even define them. */
611    
612  #else   /* SUPPORT_UTF */  #else   /* SUPPORT_UTF */
613    
 #if defined COMPILE_PCRE8  
   
 /* These macros were originally written in the form of loops that used data  
 from the tables whose names start with PRIV(utf8_table). They were rewritten by  
 a user so as not to use loops, because in some environments this gives a  
 significant performance advantage, and it seems never to do any harm. */  
   
 /* Tells the biggest code point which can be encoded as a single character. */  
   
 #define MAX_VALUE_FOR_SINGLE_CHAR 127  
   
614  /* Tests whether the code point needs extra characters to decode. */  /* Tests whether the code point needs extra characters to decode. */
615    
616  #define HAS_EXTRALEN(c) ((c) >= 0xc0)  #define HASUTF8EXTRALEN(c) ((c) >= 0xc0)
   
 /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.  
 Otherwise it has an undefined behaviour. */  
   
 #define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f])  
   
 /* Returns TRUE, if the given character is not the first character  
 of a UTF sequence. */  
   
 #define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80)  
617    
618  /* Base macro to pick up the remaining bytes of a UTF-8 character, not  /* Base macro to pick up the remaining bytes of a UTF-8 character, not
619  advancing the pointer. */  advancing the pointer. */
# Line 633  advancing the pointer. */ Line 637  advancing the pointer. */
637            ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \            ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \
638      }      }
639    
 /* Get the next UTF-8 character, not advancing the pointer. This is called when  
 we know we are in UTF-8 mode. */  
   
 #define GETCHAR(c, eptr) \  
   c = *eptr; \  
   if (c >= 0xc0) GETUTF8(c, eptr);  
   
 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the  
 pointer. */  
   
 #define GETCHARTEST(c, eptr) \  
   c = *eptr; \  
   if (utf && c >= 0xc0) GETUTF8(c, eptr);  
   
640  /* Base macro to pick up the remaining bytes of a UTF-8 character, advancing  /* Base macro to pick up the remaining bytes of a UTF-8 character, advancing
641  the pointer. */  the pointer. */
642    
# Line 681  the pointer. */ Line 671  the pointer. */
671        } \        } \
672      }      }
673    
674    #if defined COMPILE_PCRE8
675    
676    /* These macros were originally written in the form of loops that used data
677    from the tables whose names start with PRIV(utf8_table). They were rewritten by
678    a user so as not to use loops, because in some environments this gives a
679    significant performance advantage, and it seems never to do any harm. */
680    
681    /* Tells the biggest code point which can be encoded as a single character. */
682    
683    #define MAX_VALUE_FOR_SINGLE_CHAR 127
684    
685    /* Tests whether the code point needs extra characters to decode. */
686    
687    #define HAS_EXTRALEN(c) ((c) >= 0xc0)
688    
689    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
690    Otherwise it has an undefined behaviour. */
691    
692    #define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f])
693    
694    /* Returns TRUE, if the given character is not the first character
695    of a UTF sequence. */
696    
697    #define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80)
698    
699    /* Get the next UTF-8 character, not advancing the pointer. This is called when
700    we know we are in UTF-8 mode. */
701    
702    #define GETCHAR(c, eptr) \
703      c = *eptr; \
704      if (c >= 0xc0) GETUTF8(c, eptr);
705    
706    /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
707    pointer. */
708    
709    #define GETCHARTEST(c, eptr) \
710      c = *eptr; \
711      if (utf && c >= 0xc0) GETUTF8(c, eptr);
712    
713  /* Get the next UTF-8 character, advancing the pointer. This is called when we  /* Get the next UTF-8 character, advancing the pointer. This is called when we
714  know we are in UTF-8 mode. */  know we are in UTF-8 mode. */
715    
# Line 747  do not know if we are in UTF-8 mode. */ Line 776  do not know if we are in UTF-8 mode. */
776    c = *eptr; \    c = *eptr; \
777    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);
778    
779    /* Returns the next uchar, not advancing the pointer. This is called when
780    we know we are in UTF mode. */
781    
782    #define RAWUCHAR(eptr) \
783      (*(eptr))
784    
785    /* Returns the next uchar, advancing the pointer. This is called when
786    we know we are in UTF mode. */
787    
788    #define RAWUCHARINC(eptr) \
789      (*((eptr)++))
790    
791    /* Returns the next uchar, testing for UTF mode, and not advancing the
792    pointer. */
793    
794    #define RAWUCHARTEST(eptr) \
795      (*(eptr))
796    
797    /* Returns the next uchar, testing for UTF mode, advancing the
798    pointer. */
799    
800    #define RAWUCHARINCTEST(eptr) \
801      (*((eptr)++))
802    
803  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
804  it is. This is called only in UTF-8 mode - we don't put a test within the macro  it is. This is called only in UTF-8 mode - we don't put a test within the macro
805  because almost all calls are already within a block of UTF-8 only code. */  because almost all calls are already within a block of UTF-8 only code. */
# Line 842  we do not know if we are in UTF-16 mode. Line 895  we do not know if we are in UTF-16 mode.
895    c = *eptr; \    c = *eptr; \
896    if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);    if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
897    
898    /* Returns the next uchar, not advancing the pointer. This is called when
899    we know we are in UTF mode. */
900    
901    #define RAWUCHAR(eptr) \
902      (*(eptr))
903    
904    /* Returns the next uchar, advancing the pointer. This is called when
905    we know we are in UTF mode. */
906    
907    #define RAWUCHARINC(eptr) \
908      (*((eptr)++))
909    
910    /* Returns the next uchar, testing for UTF mode, and not advancing the
911    pointer. */
912    
913    #define RAWUCHARTEST(eptr) \
914      (*(eptr))
915    
916    /* Returns the next uchar, testing for UTF mode, advancing the
917    pointer. */
918    
919    #define RAWUCHARINCTEST(eptr) \
920      (*((eptr)++))
921    
922  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
923  it is. This is called only in UTF-16 mode - we don't put a test within the  it is. This is called only in UTF-16 mode - we don't put a test within the
924  macro because almost all calls are already within a block of UTF-16 only  macro because almost all calls are already within a block of UTF-16 only
# Line 858  code. */ Line 935  code. */
935    
936  #elif defined COMPILE_PCRE32  #elif defined COMPILE_PCRE32
937    
938  /* These are unnecessary for the 32-bit library */  /* These are trivial for the 32-bit library, since all UTF-32 characters fit
939  #undef MAX_VALUE_FOR_SINGLE_CHAR  into one pcre_uchar unit. */
940  #undef HAS_EXTRALEN  #define MAX_VALUE_FOR_SINGLE_CHAR (0x10ffffu)
941  #undef GET_EXTRALEN  #define HAS_EXTRALEN(c) (0)
942  #undef NOT_FIRSTCHAR  #define GET_EXTRALEN(c) (0)
943    #define NOT_FIRSTCHAR(c) (0)
 #define UTF32_MASK (0x1ffffful)  
944    
945  /* Get the next UTF-32 character, not advancing the pointer. This is called when  /* Get the next UTF-32 character, not advancing the pointer. This is called when
946  we know we are in UTF-32 mode. */  we know we are in UTF-32 mode. */
947    
948  #define GETCHAR(c, eptr) \  #define GETCHAR(c, eptr) \
949    c = (*eptr) & UTF32_MASK;    c = *(eptr);
950    
951  /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the  /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the
952  pointer. */  pointer. */
953    
954  #define GETCHARTEST(c, eptr) \  #define GETCHARTEST(c, eptr) \
955    c = *eptr; \    c = *(eptr);
   if (utf) c &= UTF32_MASK;  
956    
957  /* Get the next UTF-32 character, advancing the pointer. This is called when we  /* Get the next UTF-32 character, advancing the pointer. This is called when we
958  know we are in UTF-32 mode. */  know we are in UTF-32 mode. */
959    
960  #define GETCHARINC(c, eptr) \  #define GETCHARINC(c, eptr) \
961    c = (*eptr++) & UTF32_MASK;    c = *((eptr)++);
962    
963  /* Get the next character, testing for UTF-32 mode, and advancing the pointer.  /* Get the next character, testing for UTF-32 mode, and advancing the pointer.
964  This is called when we don't know if we are in UTF-32 mode. */  This is called when we don't know if we are in UTF-32 mode. */
965    
966  #define GETCHARINCTEST(c, eptr) \  #define GETCHARINCTEST(c, eptr) \
967    c = *eptr++; \    c = *((eptr)++);
   if (utf) c &= UTF32_MASK;  
968    
969  /* Get the next UTF-32 character, not advancing the pointer, not incrementing  /* Get the next UTF-32 character, not advancing the pointer, not incrementing
970  length (since all UTF-32 is of length 1). This is called when we know we are in  length (since all UTF-32 is of length 1). This is called when we know we are in
# Line 906  This is called when we do not know if we Line 980  This is called when we do not know if we
980  #define GETCHARLENTEST(c, eptr, len) \  #define GETCHARLENTEST(c, eptr, len) \
981    GETCHARTEST(c, eptr)    GETCHARTEST(c, eptr)
982    
983    /* Returns the next uchar, not advancing the pointer. This is called when
984    we know we are in UTF mode. */
985    
986    #define RAWUCHAR(eptr) \
987      (*(eptr))
988    
989    /* Returns the next uchar, advancing the pointer. This is called when
990    we know we are in UTF mode. */
991    
992    #define RAWUCHARINC(eptr) \
993      (*((eptr)++))
994    
995    /* Returns the next uchar, testing for UTF mode, and not advancing the
996    pointer. */
997    
998    #define RAWUCHARTEST(eptr) \
999      (*(eptr))
1000    
1001    /* Returns the next uchar, testing for UTF mode, advancing the
1002    pointer. */
1003    
1004    #define RAWUCHARINCTEST(eptr) \
1005      (*((eptr)++))
1006    
1007  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
1008  it is. This is called only in UTF-32 mode - we don't put a test within the  it is. This is called only in UTF-32 mode - we don't put a test within the
1009  macro because almost all calls are already within a block of UTF-32 only  macro because almost all calls are already within a block of UTF-32 only
1010  code. */  code.
1011    These are all no-ops since all UTF-32 characters fit into one pcre_uchar. */
1012    
1013  #define BACKCHAR(eptr) do { } while (0)  #define BACKCHAR(eptr) do { } while (0)
1014    
# Line 935  are defined. Line 1034  are defined.
1034  These values are also required as lists in pcre_compile.c when processing \h,  These values are also required as lists in pcre_compile.c when processing \h,
1035  \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but  \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but
1036  macros that define the values are here so that all the definitions are  macros that define the values are here so that all the definitions are
1037  together. The lists must be in ascending character order, terminated by  together. The lists must be in ascending character order, terminated by
1038  NOTACHAR (which is 0xffffffff).  NOTACHAR (which is 0xffffffff).
1039    
1040  Any changes should ensure that the various macros are kept in step with each  Any changes should ensure that the various macros are kept in step with each
# Line 949  other. NOTE: The values also appear in p Line 1048  other. NOTE: The values also appear in p
1048    CHAR_HT, CHAR_SPACE, 0xa0, \    CHAR_HT, CHAR_SPACE, 0xa0, \
1049    0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \    0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \
1050    0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \    0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \
1051    NOTACHAR    NOTACHAR
1052    
1053  #define HSPACE_MULTIBYTE_CASES \  #define HSPACE_MULTIBYTE_CASES \
1054    case 0x1680:  /* OGHAM SPACE MARK */ \    case 0x1680:  /* OGHAM SPACE MARK */ \
# Line 973  other. NOTE: The values also appear in p Line 1072  other. NOTE: The values also appear in p
1072    case CHAR_HT: \    case CHAR_HT: \
1073    case CHAR_SPACE: \    case CHAR_SPACE: \
1074    case 0xa0     /* NBSP */    case 0xa0     /* NBSP */
1075    
1076  #define HSPACE_CASES \  #define HSPACE_CASES \
1077    HSPACE_BYTE_CASES: \    HSPACE_BYTE_CASES: \
1078    HSPACE_MULTIBYTE_CASES    HSPACE_MULTIBYTE_CASES
1079    
1080  #define VSPACE_LIST \  #define VSPACE_LIST \
1081    CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR    CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR
1082    
1083  #define VSPACE_MULTIBYTE_CASES \  #define VSPACE_MULTIBYTE_CASES \
1084    case 0x2028:    /* LINE SEPARATOR */ \    case 0x2028:    /* LINE SEPARATOR */ \
# Line 1009  other. NOTE: The values also appear in p Line 1108  other. NOTE: The values also appear in p
1108    
1109  #ifdef EBCDIC_NL25  #ifdef EBCDIC_NL25
1110  #define VSPACE_LIST \  #define VSPACE_LIST \
1111    CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR    CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR
1112  #else  #else
1113  #define VSPACE_LIST \  #define VSPACE_LIST \
1114    CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR    CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR
1115  #endif  #endif
1116    
1117  #define VSPACE_BYTE_CASES \  #define VSPACE_BYTE_CASES \
1118    case CHAR_LF: \    case CHAR_LF: \
# Line 1030  other. NOTE: The values also appear in p Line 1129  other. NOTE: The values also appear in p
1129    
1130    
1131  /* Private flags containing information about the compiled regex. They used to  /* Private flags containing information about the compiled regex. They used to
1132  live at the top end of the options word, but that got almost full, so now they  live at the top end of the options word, but that got almost full, so they were
1133  are in a 16-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as  moved to a 16-bit flags word - which got almost full, so now they are in a
1134  the restrictions on partial matching have been lifted. It remains for backwards  32-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as the
1135    restrictions on partial matching have been lifted. It remains for backwards
1136  compatibility. */  compatibility. */
1137    
1138  #define PCRE_MODE8         0x0001  /* compiled in 8 bit mode */  #define PCRE_MODE8         0x00000001  /* compiled in 8 bit mode */
1139  #define PCRE_MODE16        0x0002  /* compiled in 16 bit mode */  #define PCRE_MODE16        0x00000002  /* compiled in 16 bit mode */
1140  #define PCRE_MODE32        0x0004  /* compiled in 32 bit mode */  #define PCRE_MODE32        0x00000004  /* compiled in 32 bit mode */
1141  #define PCRE_FIRSTSET      0x0010  /* first_char is set */  #define PCRE_FIRSTSET      0x00000010  /* first_char is set */
1142  #define PCRE_FCH_CASELESS  0x0020  /* caseless first char */  #define PCRE_FCH_CASELESS  0x00000020  /* caseless first char */
1143  #define PCRE_REQCHSET      0x0040  /* req_byte is set */  #define PCRE_REQCHSET      0x00000040  /* req_byte is set */
1144  #define PCRE_RCH_CASELESS  0x0080  /* caseless requested char */  #define PCRE_RCH_CASELESS  0x00000080  /* caseless requested char */
1145  #define PCRE_STARTLINE     0x0100  /* start after \n for multiline */  #define PCRE_STARTLINE     0x00000100  /* start after \n for multiline */
1146  #define PCRE_NOPARTIAL     0x0200  /* can't use partial with this regex */  #define PCRE_NOPARTIAL     0x00000200  /* can't use partial with this regex */
1147  #define PCRE_JCHANGED      0x0400  /* j option used in regex */  #define PCRE_JCHANGED      0x00000400  /* j option used in regex */
1148  #define PCRE_HASCRORLF     0x0800  /* explicit \r or \n in pattern */  #define PCRE_HASCRORLF     0x00000800  /* explicit \r or \n in pattern */
1149  #define PCRE_HASTHEN       0x1000  /* pattern contains (*THEN) */  #define PCRE_HASTHEN       0x00001000  /* pattern contains (*THEN) */
1150    #define PCRE_MLSET         0x00002000  /* match limit set by regex */
1151    #define PCRE_RLSET         0x00004000  /* recursion limit set by regex */
1152    #define PCRE_MATCH_EMPTY   0x00008000  /* pattern can match empty string */
1153    
1154  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
1155  #define PCRE_MODE          PCRE_MODE8  #define PCRE_MODE          PCRE_MODE8
# Line 1073  time, run time, or study time, respectiv Line 1176  time, run time, or study time, respectiv
1176     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
1177     PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \     PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
1178     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
1179     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE)     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE|PCRE_NEVER_UTF)
1180    
1181  #define PUBLIC_EXEC_OPTIONS \  #define PUBLIC_EXEC_OPTIONS \
1182    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \
# Line 1090  time, run time, or study time, respectiv Line 1193  time, run time, or study time, respectiv
1193     (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \     (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \
1194      PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED)      PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED)
1195    
1196    #define PUBLIC_JIT_EXEC_OPTIONS \
1197       (PCRE_NO_UTF8_CHECK|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|\
1198        PCRE_NOTEMPTY_ATSTART|PCRE_PARTIAL_SOFT|PCRE_PARTIAL_HARD)
1199    
1200  /* Magic number to provide a small check against being handed junk. */  /* Magic number to provide a small check against being handed junk. */
1201    
1202  #define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */  #define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */
# Line 1193  a positive value. */ Line 1300  a positive value. */
1300    
1301  /* The remaining definitions work in both environments. */  /* The remaining definitions work in both environments. */
1302    
1303    #define CHAR_NULL                   '\0'
1304  #define CHAR_HT                     '\t'  #define CHAR_HT                     '\t'
1305  #define CHAR_VT                     '\v'  #define CHAR_VT                     '\v'
1306  #define CHAR_FF                     '\f'  #define CHAR_FF                     '\f'
# Line 1432  a positive value. */ Line 1540  a positive value. */
1540  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"
1541  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"
1542  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"
1543  #ifdef COMPILE_PCRE8  #define STRING_UTF8_RIGHTPAR           "UTF8)"
1544  #define STRING_UTF_RIGHTPAR            "UTF8)"  #define STRING_UTF16_RIGHTPAR          "UTF16)"
1545  #endif  #define STRING_UTF32_RIGHTPAR          "UTF32)"
1546  #ifdef COMPILE_PCRE16  #define STRING_UTF_RIGHTPAR            "UTF)"
 #define STRING_UTF_RIGHTPAR            "UTF16)"  
 #endif  
 #ifdef COMPILE_PCRE32  
 #define STRING_UTF_RIGHTPAR            "UTF32)"  
 #endif  
1547  #define STRING_UCP_RIGHTPAR            "UCP)"  #define STRING_UCP_RIGHTPAR            "UCP)"
1548  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"
1549    #define STRING_LIMIT_MATCH_EQ          "LIMIT_MATCH="
1550    #define STRING_LIMIT_RECURSION_EQ      "LIMIT_RECURSION="
1551    
1552  #else  /* SUPPORT_UTF */  #else  /* SUPPORT_UTF */
1553    
# Line 1462  only. */ Line 1567  only. */
1567  #define CHAR_ESC                    '\033'  #define CHAR_ESC                    '\033'
1568  #define CHAR_DEL                    '\177'  #define CHAR_DEL                    '\177'
1569    
1570    #define CHAR_NULL                   '\0'
1571  #define CHAR_SPACE                  '\040'  #define CHAR_SPACE                  '\040'
1572  #define CHAR_EXCLAMATION_MARK       '\041'  #define CHAR_EXCLAMATION_MARK       '\041'
1573  #define CHAR_QUOTATION_MARK         '\042'  #define CHAR_QUOTATION_MARK         '\042'
# Line 1697  only. */ Line 1803  only. */
1803  #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
1804  #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
1805  #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
1806  #ifdef COMPILE_PCRE8  #define STRING_UTF8_RIGHTPAR           STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
1807  #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
1808  #endif  #define STRING_UTF32_RIGHTPAR          STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
1809  #ifdef COMPILE_PCRE16  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_RIGHT_PARENTHESIS
 #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS  
 #endif  
 #ifdef COMPILE_PCRE32  
 #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS  
 #endif  
1810  #define STRING_UCP_RIGHTPAR            STR_U STR_C STR_P STR_RIGHT_PARENTHESIS  #define STRING_UCP_RIGHTPAR            STR_U STR_C STR_P STR_RIGHT_PARENTHESIS
1811  #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_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
1812    #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
1813    #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
1814    
1815  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
1816    
# Line 1748  only. */ Line 1851  only. */
1851  #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 */
1852  #define PT_WORD       8    /* Word - L plus N plus underscore */  #define PT_WORD       8    /* Word - L plus N plus underscore */
1853  #define PT_CLIST      9    /* Pseudo-property: match character list */  #define PT_CLIST      9    /* Pseudo-property: match character list */
1854    #define PT_UCNC      10    /* Universal Character nameable character */
1855    
1856  /* 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
1857  contain characters with values greater than 255. */  contain characters with values greater than 255. */
# Line 2183  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 2287  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
2287         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
2288         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
2289         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
2290         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERRCOUNT };         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERR78, ERRCOUNT };
2291    
2292  /* JIT compiling modes. The function list is indexed by them. */  /* JIT compiling modes. The function list is indexed by them. */
2293  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
# Line 2193  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COM Line 2297  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COM
2297  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
2298  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
2299  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
2300  be well. For the case of compiled-on-4 and run-on-8, we include an extra  be well.
 pointer that is always NULL. For future-proofing, a few dummy fields were  
 originally included - even though you can never get this planning right - but  
 there is only one left now.  
   
 NOTE NOTE NOTE:  
 Because people can now save and re-use compiled patterns, any additions to this  
 structure should be made at the end, and something earlier (e.g. a new  
 flag in the options or one of the dummy fields) should indicate that the new  
 fields are present. Currently PCRE always sets the dummy fields to zero.  
 NOTE NOTE NOTE  
 */  
2301    
2302  #if defined COMPILE_PCRE8  The size of the structure must be a multiple of 8 bytes. For the case of
2303  #define REAL_PCRE real_pcre  compiled-on-4 and run-on-8, we include an extra pointer that is always NULL so
2304  #elif defined COMPILE_PCRE16  that there are an even number of pointers which therefore are a multiple of 8
2305  #define REAL_PCRE real_pcre16  bytes.
2306  #elif defined COMPILE_PCRE32  
2307  #define REAL_PCRE real_pcre32  It is necessary to fork the struct for the 32 bit library, since it needs to
2308  #endif  use pcre_uint32 for first_char and req_char. We can't put an ifdef inside the
2309    typedef because pcretest needs access to the struct of the 8-, 16- and 32-bit
2310  /* It is necessary to fork the struct for 32 bit, since it needs to use  variants.
2311   * pcre_uchar for first_char and req_char. Can't put an ifdef inside the  
2312   * typedef since pcretest needs access to  the struct of the 8-, 16-  *** WARNING ***
2313   * and 32-bit variants. */  When new fields are added to these structures, remember to adjust the code in
2314    pcre_byte_order.c that is concerned with swapping the byte order of the fields
2315    when a compiled regex is reloaded on a host with different endianness.
2316    *** WARNING ***
2317    There is also similar byte-flipping code in pcretest.c, which is used for
2318    testing the byte-flipping features. It must also be kept in step.
2319    *** WARNING ***
2320    */
2321    
2322  typedef struct real_pcre8_or_16 {  typedef struct real_pcre8_or_16 {
2323    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2324    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2325    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2326    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2327      pcre_uint32 limit_match;        /* Limit set from regex */
2328      pcre_uint32 limit_recursion;    /* Limit set from regex */
2329      pcre_uint16 first_char;         /* Starting character */
2330      pcre_uint16 req_char;           /* This character must be seen */
2331    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2332    pcre_uint16 top_bracket;        /* Highest numbered group */    pcre_uint16 top_bracket;        /* Highest numbered group */
2333    pcre_uint16 top_backref;        /* Highest numbered back reference */    pcre_uint16 top_backref;        /* Highest numbered back reference */
   pcre_uint16 first_char;         /* Starting character */  
   pcre_uint16 req_char;           /* This character must be seen */  
2334    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2335    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2336    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2337    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2338      pcre_uint16 dummy1;             /* To ensure size is a multiple of 8 */
2339      pcre_uint16 dummy2;             /* To ensure size is a multiple of 8 */
2340      pcre_uint16 dummy3;             /* To ensure size is a multiple of 8 */
2341    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2342    const pcre_uint8 *nullpad;      /* NULL padding */    void             *nullpad;      /* NULL padding */
2343  } real_pcre8_or_16;  } real_pcre8_or_16;
2344    
2345  typedef struct real_pcre8_or_16 real_pcre;  typedef struct real_pcre8_or_16 real_pcre;
# Line 2244  typedef struct real_pcre32 { Line 2349  typedef struct real_pcre32 {
2349    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2350    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2351    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2352    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2353      pcre_uint32 limit_match;        /* Limit set from regex */
2354      pcre_uint32 limit_recursion;    /* Limit set from regex */
2355      pcre_uint32 first_char;         /* Starting character */
2356      pcre_uint32 req_char;           /* This character must be seen */
2357    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2358    pcre_uint16 top_bracket;        /* Highest numbered group */    pcre_uint16 top_bracket;        /* Highest numbered group */
2359    pcre_uint16 top_backref;        /* Highest numbered back reference */    pcre_uint16 top_backref;        /* Highest numbered back reference */
   pcre_uint32 first_char;         /* Starting character */  
   pcre_uint32 req_char;           /* This character must be seen */  
2360    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2361    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2362    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2363    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2364    pcre_uint16 dummy1;             /* for later expansion */    pcre_uint16 dummy;              /* To ensure size is a multiple of 8 */
   pcre_uint16 dummy2;             /* for later expansion */  
2365    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2366    void *nullpad;                  /* for later expansion */    void             *nullpad;      /* NULL padding */
2367  } real_pcre32;  } real_pcre32;
2368    
2369    #if defined COMPILE_PCRE8
2370    #define REAL_PCRE real_pcre
2371    #elif defined COMPILE_PCRE16
2372    #define REAL_PCRE real_pcre16
2373    #elif defined COMPILE_PCRE32
2374    #define REAL_PCRE real_pcre32
2375    #endif
2376    
2377  /* Assert that the size of REAL_PCRE is divisible by 8 */  /* Assert that the size of REAL_PCRE is divisible by 8 */
2378  typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];  typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];
2379    
# Line 2293  typedef struct open_capitem { Line 2407  typedef struct open_capitem {
2407    pcre_uint16 flag;             /* Set TRUE if recursive back ref */    pcre_uint16 flag;             /* Set TRUE if recursive back ref */
2408  } open_capitem;  } open_capitem;
2409    
2410    /* Structure for building a list of named groups during the first pass of
2411    compiling. */
2412    
2413    typedef struct named_group {
2414      const pcre_uchar  *name;          /* Points to the name in the pattern */
2415      int                length;        /* Length of the name */
2416      pcre_uint32        number;        /* Group number */
2417    } named_group;
2418    
2419  /* Structure for passing "static" information around between the functions  /* Structure for passing "static" information around between the functions
2420  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
2421    
# Line 2305  typedef struct compile_data { Line 2428  typedef struct compile_data {
2428    const pcre_uchar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
2429    const pcre_uchar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
2430    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 */  
2431    pcre_uchar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
2432      open_capitem *open_caps;          /* Chain of open capture items */
2433      named_group *named_groups;        /* Points to vector in pre-compile */
2434    pcre_uchar *name_table;           /* The name/number table */    pcre_uchar *name_table;           /* The name/number table */
2435    int  names_found;                 /* Number of entries so far */    int  names_found;                 /* Number of entries so far */
2436    int  name_entry_size;             /* Size of each entry */    int  name_entry_size;             /* Size of each entry */
2437      int  named_group_list_size;       /* Number of entries in the list */
2438    int  workspace_size;              /* Size of workspace */    int  workspace_size;              /* Size of workspace */
2439    int  bracount;                    /* Count of capturing parens as we compile */    unsigned int bracount;            /* Count of capturing parens as we compile */
2440    int  final_bracount;              /* Saved value after first pass */    int  final_bracount;              /* Saved value after first pass */
2441    int  max_lookbehind;              /* Maximum lookbehind (characters) */    int  max_lookbehind;              /* Maximum lookbehind (characters) */
2442    int  top_backref;                 /* Maximum back reference */    int  top_backref;                 /* Maximum back reference */
2443    unsigned int backref_map;         /* Bitmap of low back refs */    unsigned int backref_map;         /* Bitmap of low back refs */
2444    int  assert_depth;                /* Depth of nested assertions */    int  assert_depth;                /* Depth of nested assertions */
2445    int  external_options;            /* External (initial) options */    pcre_uint32 external_options;     /* External (initial) options */
2446    int  external_flags;              /* External flag bits to be set */    pcre_uint32 external_flags;       /* External flag bits to be set */
2447    int  req_varyopt;                 /* "After variable item" flag for reqbyte */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2448    BOOL had_accept;                  /* (*ACCEPT) encountered */    BOOL had_accept;                  /* (*ACCEPT) encountered */
2449    BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */    BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
# Line 2341  call within the pattern; used by pcre_ex Line 2466  call within the pattern; used by pcre_ex
2466    
2467  typedef struct recursion_info {  typedef struct recursion_info {
2468    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
2469    int group_num;                  /* Number of group that was called */    unsigned int group_num;         /* Number of group that was called */
2470    int *offset_save;               /* Pointer to start of saved offsets */    int *offset_save;               /* Pointer to start of saved offsets */
2471    int saved_max;                  /* Number of saved offsets */    int saved_max;                  /* Number of saved offsets */
2472      int saved_capture_last;         /* Last capture number */
2473    PCRE_PUCHAR subject_position;   /* Position at start of recursion */    PCRE_PUCHAR subject_position;   /* Position at start of recursion */
2474  } recursion_info;  } recursion_info;
2475    
# Line 2380  typedef struct match_data { Line 2506  typedef struct match_data {
2506    int    nllen;                   /* Newline string length */    int    nllen;                   /* Newline string length */
2507    int    name_count;              /* Number of names in name table */    int    name_count;              /* Number of names in name table */
2508    int    name_entry_size;         /* Size of entry in names table */    int    name_entry_size;         /* Size of entry in names table */
2509      unsigned int skip_arg_count;    /* For counting SKIP_ARGs */
2510      unsigned int ignore_skip_arg;   /* For re-run when SKIP arg name not found */
2511    pcre_uchar *name_table;         /* Table of names */    pcre_uchar *name_table;         /* Table of names */
2512    pcre_uchar nl[4];               /* Newline string when fixed */    pcre_uchar nl[4];               /* Newline string when fixed */
2513    const  pcre_uint8 *lcc;         /* Points to lower casing table */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
2514    const  pcre_uint8 *fcc;         /* Points to case-flipping table */    const  pcre_uint8 *fcc;         /* Points to case-flipping table */
2515    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 */  
2516    BOOL   notbol;                  /* NOTBOL flag */    BOOL   notbol;                  /* NOTBOL flag */
2517    BOOL   noteol;                  /* NOTEOL flag */    BOOL   noteol;                  /* NOTEOL flag */
2518    BOOL   utf;                     /* UTF-8 / UTF-16 flag */    BOOL   utf;                     /* UTF-8 / UTF-16 flag */
# Line 2397  typedef struct match_data { Line 2524  typedef struct match_data {
2524    BOOL   hitend;                  /* Hit the end of the subject at some point */    BOOL   hitend;                  /* Hit the end of the subject at some point */
2525    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
2526    BOOL   hasthen;                 /* Pattern contains (*THEN) */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
   BOOL   ignore_skip_arg;         /* For re-run when SKIP name not found */  
2527    const  pcre_uchar *start_code;  /* For use when recursing */    const  pcre_uchar *start_code;  /* For use when recursing */
2528    PCRE_PUCHAR start_subject;      /* Start of the subject string */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
2529    PCRE_PUCHAR end_subject;        /* End of the subject string */    PCRE_PUCHAR end_subject;        /* End of the subject string */
# Line 2406  typedef struct match_data { Line 2532  typedef struct match_data {
2532    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */
2533    int    partial;                 /* PARTIAL options */    int    partial;                 /* PARTIAL options */
2534    int    end_offset_top;          /* Highwater mark at end of match */    int    end_offset_top;          /* Highwater mark at end of match */
2535    int    capture_last;            /* Most recent capture number */    pcre_int32 capture_last;        /* Most recent capture number + overflow flag */
2536    int    start_offset;            /* The start offset value */    int    start_offset;            /* The start offset value */
2537    int    match_function_type;     /* Set for certain special calls of MATCH() */    int    match_function_type;     /* Set for certain special calls of MATCH() */
2538    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */
# Line 2577  extern unsigned int      PRIV(strlen_uc) Line 2703  extern unsigned int      PRIV(strlen_uc)
2703    
2704  #endif /* COMPILE_PCRE[8|16|32] */  #endif /* COMPILE_PCRE[8|16|32] */
2705    
2706    #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16
2707    
2708    #define STRCMP_UC_UC_TEST(str1, str2) STRCMP_UC_UC(str1, str2)
2709    #define STRCMP_UC_C8_TEST(str1, str2) STRCMP_UC_C8(str1, str2)
2710    
2711    #elif defined COMPILE_PCRE32
2712    
2713    extern int               PRIV(strcmp_uc_uc_utf)(const pcre_uchar *,
2714                               const pcre_uchar *);
2715    extern int               PRIV(strcmp_uc_c8_utf)(const pcre_uchar *,
2716                               const char *);
2717    
2718    #define STRCMP_UC_UC_TEST(str1, str2) \
2719      (utf ? PRIV(strcmp_uc_uc_utf)((str1), (str2)) : PRIV(strcmp_uc_uc)((str1), (str2)))
2720    #define STRCMP_UC_C8_TEST(str1, str2) \
2721      (utf ? PRIV(strcmp_uc_c8_utf)((str1), (str2)) : PRIV(strcmp_uc_c8)((str1), (str2)))
2722    
2723    #endif /* COMPILE_PCRE[8|16|32] */
2724    
2725  extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);  extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);
2726  extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,  extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2727                             int *, BOOL);                             int *, BOOL);
2728  extern int               PRIV(ord2utf)(pcre_uint32, pcre_uchar *);  extern unsigned int      PRIV(ord2utf)(pcre_uint32, pcre_uchar *);
2729  extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);  extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);
2730  extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,  extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2731                             int *, BOOL);                             int *, BOOL);
# Line 2589  extern BOOL              PRIV(xclass)(pc Line 2734  extern BOOL              PRIV(xclass)(pc
2734  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2735  extern void              PRIV(jit_compile)(const REAL_PCRE *,  extern void              PRIV(jit_compile)(const REAL_PCRE *,
2736                             PUBL(extra) *, int);                             PUBL(extra) *, int);
2737  extern int               PRIV(jit_exec)(const REAL_PCRE *, const PUBL(extra) *,  extern int               PRIV(jit_exec)(const PUBL(extra) *,
2738                             const pcre_uchar *, int, int, int, int *, int);                             const pcre_uchar *, int, int, int, int *, int);
2739  extern void              PRIV(jit_free)(void *);  extern void              PRIV(jit_free)(void *);
2740  extern int               PRIV(jit_get_size)(void *);  extern int               PRIV(jit_get_size)(void *);
# Line 2610  extern const pcre_uint32 PRIV(ucd_casele Line 2755  extern const pcre_uint32 PRIV(ucd_casele
2755  extern const ucd_record  PRIV(ucd_records)[];  extern const ucd_record  PRIV(ucd_records)[];
2756  extern const pcre_uint8  PRIV(ucd_stage1)[];  extern const pcre_uint8  PRIV(ucd_stage1)[];
2757  extern const pcre_uint16 PRIV(ucd_stage2)[];  extern const pcre_uint16 PRIV(ucd_stage2)[];
2758  extern const int         PRIV(ucp_gentype)[];  extern const pcre_uint32 PRIV(ucp_gentype)[];
2759  extern const pcre_uint32 PRIV(ucp_gbtable)[];  extern const pcre_uint32 PRIV(ucp_gbtable)[];
2760  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2761  extern const int         PRIV(ucp_typerange)[];  extern const int         PRIV(ucp_typerange)[];

Legend:
Removed from v.1083  
changed lines
  Added in v.1359

  ViewVC Help
Powered by ViewVC 1.1.5