/[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 932 by ph10, Fri Feb 24 18:54:43 2012 UTC revision 1057 by chpe, Tue Oct 16 15:53:45 2012 UTC
# 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  #if !defined COMPILE_PCRE16 && ! defined COMPILE_PCRE32
57  #define COMPILE_PCRE8  #define COMPILE_PCRE8
58  #endif  #endif
59    
# Line 78  Until then we define it if SUPPORT_UTF i Line 78  Until then we define it if SUPPORT_UTF i
78  #define SUPPORT_UTF8 1  #define SUPPORT_UTF8 1
79  #endif  #endif
80    
81  /* 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"
82  script prevents both being selected, but not everybody uses "configure". */  script prevents both being selected, but not everybody uses "configure". */
83    
84  #if defined EBCDIC && defined SUPPORT_UTF  #if defined EBCDIC && defined SUPPORT_UTF
85  #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.
86  #endif  #endif
87    
88  /* 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 243  exactly 256 items. When the character is Line 243  exactly 256 items. When the character is
243  items, some check is needed before accessing these tables.  items, some check is needed before accessing these tables.
244  */  */
245    
246  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
247    
248  typedef unsigned char pcre_uchar;  typedef unsigned char pcre_uchar;
249  #define IN_UCHARS(x) (x)  #define IN_UCHARS(x) (x)
250  #define MAX_255(c) 1  #define MAX_255(c) 1
251  #define TABLE_GET(c, table, default) ((table)[c])  #define TABLE_GET(c, table, default) ((table)[c])
252    
253  #else  #elif defined COMPILE_PCRE16
254    
 #ifdef COMPILE_PCRE16  
255  #if USHRT_MAX != 65535  #if USHRT_MAX != 65535
256  /* This is a warning message. Change PCRE_UCHAR16 to a 16 bit data type in  /* This is a warning message. Change PCRE_UCHAR16 to a 16 bit data type in
257  pcre.h(.in) and disable (comment out) this message. */  pcre.h(.in) and disable (comment out) this message. */
# Line 260  pcre.h(.in) and disable (comment out) th Line 259  pcre.h(.in) and disable (comment out) th
259  #endif  #endif
260    
261  typedef pcre_uint16 pcre_uchar;  typedef pcre_uint16 pcre_uchar;
262  #define IN_UCHARS(x) ((x) << 1)  #define UCHAR_SHIFT (1)
263    #define IN_UCHARS(x) ((x) << UCHAR_SHIFT)
264  #define MAX_255(c) ((c) <= 255u)  #define MAX_255(c) ((c) <= 255u)
265  #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))  #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
266    
267    #elif defined COMPILE_PCRE32
268    
269    typedef pcre_uint32 pcre_uchar;
270    #define UCHAR_SHIFT (2)
271    #define IN_UCHARS(x) ((x) << UCHAR_SHIFT)
272    #define MAX_255(c) ((c) <= 255u)
273    #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
274    
275    /* Assert that pcre_uchar32 is a 32-bit type */
276    typedef int __assert_pcre_uchar32_size[sizeof(pcre_uchar) == 4 ? 1 : -1];
277    
278  #else  #else
279  #error Unsupported compiling mode  #error Unsupported compiling mode
280  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
   
 #endif /* COMPILE_PCRE8 */  
281    
282  /* 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
283  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 396  The macros are controlled by the value o Line 405  The macros are controlled by the value o
405  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
406  is automated on Unix systems via the "configure" command. */  is automated on Unix systems via the "configure" command. */
407    
408  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
409    
410  #if LINK_SIZE == 2  #if LINK_SIZE == 2
411    
# Line 441  is automated on Unix systems via the "co Line 450  is automated on Unix systems via the "co
450  #error LINK_SIZE must be either 2, 3, or 4  #error LINK_SIZE must be either 2, 3, or 4
451  #endif  #endif
452    
453  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
   
 #ifdef COMPILE_PCRE16  
454    
455  #if LINK_SIZE == 2  #if LINK_SIZE == 2
456    
457    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
458  #undef LINK_SIZE  #undef LINK_SIZE
459  #define LINK_SIZE 1  #define LINK_SIZE 1
460    
# Line 460  is automated on Unix systems via the "co Line 468  is automated on Unix systems via the "co
468    
469  #elif LINK_SIZE == 3 || LINK_SIZE == 4  #elif LINK_SIZE == 3 || LINK_SIZE == 4
470    
471    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
472  #undef LINK_SIZE  #undef LINK_SIZE
473  #define LINK_SIZE 2  #define LINK_SIZE 2
474    
# Line 477  is automated on Unix systems via the "co Line 486  is automated on Unix systems via the "co
486  #error LINK_SIZE must be either 2, 3, or 4  #error LINK_SIZE must be either 2, 3, or 4
487  #endif  #endif
488    
489    #elif defined COMPILE_PCRE32
490    
491    /* Only supported LINK_SIZE is 4 */
492    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
493    #undef LINK_SIZE
494    #define LINK_SIZE 1
495    
496    #define PUT(a,n,d)   \
497      (a[n] = (d))
498    
499    #define GET(a,n) \
500      (a[n])
501    
502    /* Keep it positive */
503    #define MAX_PATTERN_SIZE (1 << 30)
504    
505  #else  #else
506  #error Unsupported compiling mode  #error Unsupported compiling mode
507  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
   
 #endif /* COMPILE_PCRE8 */  
508    
509  /* Convenience macro defined in terms of the others */  /* Convenience macro defined in terms of the others */
510    
# Line 492  is automated on Unix systems via the "co Line 515  is automated on Unix systems via the "co
515  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
516  capturing parenthesis numbers in back references. */  capturing parenthesis numbers in back references. */
517    
518  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
519    
520  #define IMM2_SIZE 2  #define IMM2_SIZE 2
521    
# Line 503  capturing parenthesis numbers in back re Line 526  capturing parenthesis numbers in back re
526  #define GET2(a,n) \  #define GET2(a,n) \
527    (((a)[n] << 8) | (a)[(n)+1])    (((a)[n] << 8) | (a)[(n)+1])
528    
529  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
530    
531  #ifdef COMPILE_PCRE16  #define IMM2_SIZE 1
532    
533    #define PUT2(a,n,d)   \
534       a[n] = d
535    
536    #define GET2(a,n) \
537       a[n]
538    
539    #elif defined COMPILE_PCRE32
540    
541  #define IMM2_SIZE 1  #define IMM2_SIZE 1
542    
# Line 517  capturing parenthesis numbers in back re Line 548  capturing parenthesis numbers in back re
548    
549  #else  #else
550  #error Unsupported compiling mode  #error Unsupported compiling mode
551  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
   
 #endif /* COMPILE_PCRE8 */  
552    
553  #define PUT2INC(a,n,d)  PUT2(a,n,d), a += IMM2_SIZE  #define PUT2INC(a,n,d)  PUT2(a,n,d), a += IMM2_SIZE
554    
555    /* The maximum length of a MARK name is currently one data unit; it may be
556    changed in future to be a fixed number of bytes or to depend on LINK_SIZE. */
557    
558    #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
559    #define MAX_MARK ((1u << 16) - 1)
560    #else
561    #define MAX_MARK ((1u << 8) - 1)
562    #endif
563    
564  /* 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
565  character. The macros for character handling generate simple sequences when  byte. The macros for character handling generate simple sequences when used in
566  used in character-mode, and more complicated ones for UTF characters.  character-mode, and more complicated ones for UTF characters. GETCHARLENTEST
567  GETCHARLENTEST and other macros are not used when UTF is not supported,  and other macros are not used when UTF is not supported, so they are not
568  so they are not defined. To make sure they can never even appear when  defined. To make sure they can never even appear when UTF support is omitted,
569  UTF support is omitted, we don't even define them. */  we don't even define them. */
570    
571  #ifndef SUPPORT_UTF  #ifndef SUPPORT_UTF
572    
# Line 548  UTF support is omitted, we don't even de Line 586  UTF support is omitted, we don't even de
586    
587  #else   /* SUPPORT_UTF */  #else   /* SUPPORT_UTF */
588    
589  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
590    
591  /* These macros were originally written in the form of loops that used data  /* These macros were originally written in the form of loops that used data
592  from the tables whose names start with PRIV(utf8_table). They were rewritten by  from the tables whose names start with PRIV(utf8_table). They were rewritten by
# Line 722  because almost all calls are already wit Line 760  because almost all calls are already wit
760  #define ACROSSCHAR(condition, eptr, action) \  #define ACROSSCHAR(condition, eptr, action) \
761    while((condition) && ((eptr) & 0xc0) == 0x80) action    while((condition) && ((eptr) & 0xc0) == 0x80) action
762    
763  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
   
 #ifdef COMPILE_PCRE16  
764    
765  /* 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. */
766    
# Line 820  code. */ Line 856  code. */
856  #define ACROSSCHAR(condition, eptr, action) \  #define ACROSSCHAR(condition, eptr, action) \
857    if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action    if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action
858    
859  #endif  #elif defined COMPILE_PCRE32
860    
861  #endif /* COMPILE_PCRE8 */  /* These are unnecessary for the 32-bit library */
862    #undef MAX_VALUE_FOR_SINGLE_CHAR
863    #undef HAS_EXTRALEN
864    #undef GET_EXTRALEN
865    #undef NOT_FIRSTCHAR
866    
867    /* Get the next UTF-32 character, not advancing the pointer. This is called when
868    we know we are in UTF-32 mode. */
869    
870    #define GETCHAR(c, eptr) \
871      c = *eptr;
872    
873    /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the
874    pointer. */
875    
876    #define GETCHARTEST(c, eptr) \
877      c = *eptr;
878    
879    /* Get the next UTF-32 character, advancing the pointer. This is called when we
880    know we are in UTF-32 mode. */
881    
882    #define GETCHARINC(c, eptr) \
883      c = *eptr++;
884    
885    /* Get the next character, testing for UTF-32 mode, and advancing the pointer.
886    This is called when we don't know if we are in UTF-32 mode. */
887    
888    #define GETCHARINCTEST(c, eptr) \
889      c = *eptr++;
890    
891    /* Get the next UTF-32 character, not advancing the pointer, not incrementing
892    length (since all UTF-32 is of length 1). This is called when we know we are in
893    UTF-32 mode. */
894    
895    #define GETCHARLEN(c, eptr, len) \
896      c = *eptr;
897    
898    /* Get the next UTF-832character, testing for UTF-32 mode, not advancing the
899    pointer, not incrementing the length (since all UTF-32 is of length 1).
900    This is called when we do not know if we are in UTF-32 mode. */
901    
902    #define GETCHARLENTEST(c, eptr, len) \
903      c = *eptr;
904    
905    /* If the pointer is not at the start of a character, move it back until
906    it is. This is called only in UTF-32 mode - we don't put a test within the
907    macro because almost all calls are already within a block of UTF-32 only
908    code. */
909    
910    #define BACKCHAR(eptr) do { } while (0)
911    
912    /* Same as above, just in the other direction. */
913    #define FORWARDCHAR(eptr) do { } while (0)
914    
915    /* Same as above, but it allows a fully customizable form. */
916    #define ACROSSCHAR(condition, eptr, action) do { } while (0)
917    
918    #else
919    #error Unsupported compiling mode
920    #endif /* COMPILE_PCRE[8|16|32] */
921    
922  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
923    
924    /* Tests for Unicode horizontal and vertical whitespace characters must check a
925    number of different values. Using a switch statement for this generates the
926    fastest code (no loop, no memory access), and there are several places in the
927    interpreter code where this happens. In order to ensure that all the case lists
928    remain in step, we use macros so that there is only one place where the lists
929    are defined.
930    
931    These values are also required as lists in pcre_compile.c when processing \h,
932    \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but
933    macros that define the values are here so that all the definitions are
934    together. The lists must be in ascending character order, terminated by
935    NOTACHAR (which is 0xffffffff).
936    
937    Any changes should ensure that the various macros are kept in step with each
938    other. NOTE: The values also appear in pcre_jit_compile.c. */
939    
940    /* ------ ASCII/Unicode environments ------ */
941    
942    #ifndef EBCDIC
943    
944    #define HSPACE_LIST \
945      CHAR_HT, CHAR_SPACE, 0xa0, \
946      0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \
947      0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \
948      NOTACHAR
949    
950    #define HSPACE_MULTIBYTE_CASES \
951      case 0x1680:  /* OGHAM SPACE MARK */ \
952      case 0x180e:  /* MONGOLIAN VOWEL SEPARATOR */ \
953      case 0x2000:  /* EN QUAD */ \
954      case 0x2001:  /* EM QUAD */ \
955      case 0x2002:  /* EN SPACE */ \
956      case 0x2003:  /* EM SPACE */ \
957      case 0x2004:  /* THREE-PER-EM SPACE */ \
958      case 0x2005:  /* FOUR-PER-EM SPACE */ \
959      case 0x2006:  /* SIX-PER-EM SPACE */ \
960      case 0x2007:  /* FIGURE SPACE */ \
961      case 0x2008:  /* PUNCTUATION SPACE */ \
962      case 0x2009:  /* THIN SPACE */ \
963      case 0x200A:  /* HAIR SPACE */ \
964      case 0x202f:  /* NARROW NO-BREAK SPACE */ \
965      case 0x205f:  /* MEDIUM MATHEMATICAL SPACE */ \
966      case 0x3000   /* IDEOGRAPHIC SPACE */
967    
968    #define HSPACE_BYTE_CASES \
969      case CHAR_HT: \
970      case CHAR_SPACE: \
971      case 0xa0     /* NBSP */
972    
973    #define HSPACE_CASES \
974      HSPACE_BYTE_CASES: \
975      HSPACE_MULTIBYTE_CASES
976    
977    #define VSPACE_LIST \
978      CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR
979    
980    #define VSPACE_MULTIBYTE_CASES \
981      case 0x2028:    /* LINE SEPARATOR */ \
982      case 0x2029     /* PARAGRAPH SEPARATOR */
983    
984    #define VSPACE_BYTE_CASES \
985      case CHAR_LF: \
986      case CHAR_VT: \
987      case CHAR_FF: \
988      case CHAR_CR: \
989      case CHAR_NEL
990    
991    #define VSPACE_CASES \
992      VSPACE_BYTE_CASES: \
993      VSPACE_MULTIBYTE_CASES
994    
995  /* In case there is no definition of offsetof() provided - though any proper  /* ------ EBCDIC environments ------ */
996  Standard C system should have one. */  
997    #else
998    #define HSPACE_LIST CHAR_HT, CHAR_SPACE
999    
1000    #define HSPACE_BYTE_CASES \
1001      case CHAR_HT: \
1002      case CHAR_SPACE
1003    
1004    #define HSPACE_CASES HSPACE_BYTE_CASES
1005    
1006    #ifdef EBCDIC_NL25
1007    #define VSPACE_LIST \
1008      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR
1009    #else
1010    #define VSPACE_LIST \
1011      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR
1012    #endif
1013    
1014    #define VSPACE_BYTE_CASES \
1015      case CHAR_LF: \
1016      case CHAR_VT: \
1017      case CHAR_FF: \
1018      case CHAR_CR: \
1019      case CHAR_NEL
1020    
1021    #define VSPACE_CASES VSPACE_BYTE_CASES
1022    #endif  /* EBCDIC */
1023    
1024    /* ------ End of whitespace macros ------ */
1025    
 #ifndef offsetof  
 #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))  
 #endif  
1026    
1027    
1028  /* Private flags containing information about the compiled regex. They used to  /* Private flags containing information about the compiled regex. They used to
# Line 841  are in a 16-bit flags word. From release Line 1031  are in a 16-bit flags word. From release
1031  the restrictions on partial matching have been lifted. It remains for backwards  the restrictions on partial matching have been lifted. It remains for backwards
1032  compatibility. */  compatibility. */
1033    
1034  #ifdef COMPILE_PCRE8  #define PCRE_MODE8         0x0001  /* compiled in 8 bit mode */
1035  #define PCRE_MODE          0x0001  /* compiled in 8 bit mode */  #define PCRE_MODE16        0x0002  /* compiled in 16 bit mode */
1036  #endif  #define PCRE_MODE32        0x0004  /* compiled in 32 bit mode */
 #ifdef COMPILE_PCRE16  
 #define PCRE_MODE          0x0002  /* compiled in 16 bit mode */  
 #endif  
1037  #define PCRE_FIRSTSET      0x0010  /* first_char is set */  #define PCRE_FIRSTSET      0x0010  /* first_char is set */
1038  #define PCRE_FCH_CASELESS  0x0020  /* caseless first char */  #define PCRE_FCH_CASELESS  0x0020  /* caseless first char */
1039  #define PCRE_REQCHSET      0x0040  /* req_byte is set */  #define PCRE_REQCHSET      0x0040  /* req_byte is set */
# Line 857  compatibility. */ Line 1044  compatibility. */
1044  #define PCRE_HASCRORLF     0x0800  /* explicit \r or \n in pattern */  #define PCRE_HASCRORLF     0x0800  /* explicit \r or \n in pattern */
1045  #define PCRE_HASTHEN       0x1000  /* pattern contains (*THEN) */  #define PCRE_HASTHEN       0x1000  /* pattern contains (*THEN) */
1046    
1047    #if defined COMPILE_PCRE8
1048    #define PCRE_MODE          PCRE_MODE8
1049    #elif defined COMPILE_PCRE16
1050    #define PCRE_MODE          PCRE_MODE16
1051    #elif defined COMPILE_PCRE32
1052    #define PCRE_MODE          PCRE_MODE32
1053    #endif
1054    #define PCRE_MODE_MASK     (PCRE_MODE8 | PCRE_MODE16 | PCRE_MODE32)
1055    
1056  /* Flags for the "extra" block produced by pcre_study(). */  /* Flags for the "extra" block produced by pcre_study(). */
1057    
1058  #define PCRE_STUDY_MAPPED  0x0001  /* a map of starting chars exists */  #define PCRE_STUDY_MAPPED  0x0001  /* a map of starting chars exists */
# Line 888  time, run time, or study time, respectiv Line 1084  time, run time, or study time, respectiv
1084    
1085  #define PUBLIC_STUDY_OPTIONS \  #define PUBLIC_STUDY_OPTIONS \
1086     (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \     (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \
1087      PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE)      PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED)
1088    
1089  /* Magic number to provide a small check against being handed junk. */  /* Magic number to provide a small check against being handed junk. */
1090    
# Line 940  macros to give the functions distinct na Line 1136  macros to give the functions distinct na
1136  #ifndef SUPPORT_UTF  #ifndef SUPPORT_UTF
1137    
1138  /* UTF-8 support is not enabled; use the platform-dependent character literals  /* UTF-8 support is not enabled; use the platform-dependent character literals
1139  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
1140    mode. Newline characters are problematic in EBCDIC. Though it has CR and LF
1141    characters, a common practice has been to use its NL (0x15) character as the
1142    line terminator in C-like processing environments. However, sometimes the LF
1143    (0x25) character is used instead, according to this Unicode document:
1144    
1145    http://unicode.org/standard/reports/tr13/tr13-5.html
1146    
1147    PCRE defaults EBCDIC NL to 0x15, but has a build-time option to select 0x25
1148    instead. Whichever is *not* chosen is defined as NEL.
1149    
1150    In both ASCII and EBCDIC environments, CHAR_NL and CHAR_LF are synonyms for the
1151    same code point. */
1152    
1153    #ifdef EBCDIC
1154    
1155    #ifndef EBCDIC_NL25
1156    #define CHAR_NL                     '\x15'
1157    #define CHAR_NEL                    '\x25'
1158    #define STR_NL                      "\x15"
1159    #define STR_NEL                     "\x25"
1160    #else
1161    #define CHAR_NL                     '\x25'
1162    #define CHAR_NEL                    '\x15'
1163    #define STR_NL                      "\x25"
1164    #define STR_NEL                     "\x15"
1165    #endif
1166    
1167    #define CHAR_LF                     CHAR_NL
1168    #define STR_LF                      STR_NL
1169    
1170    #define CHAR_ESC                    '\047'
1171    #define CHAR_DEL                    '\007'
1172    #define STR_ESC                     "\047"
1173    #define STR_DEL                     "\007"
1174    
1175    #else  /* Not EBCDIC */
1176    
1177    /* In ASCII/Unicode, linefeed is '\n' and we equate this to NL for
1178    compatibility. NEL is the Unicode newline character; make sure it is
1179    a positive value. */
1180    
1181    #define CHAR_LF                     '\n'
1182    #define CHAR_NL                     CHAR_LF
1183    #define CHAR_NEL                    ((unsigned char)'\x85')
1184    #define CHAR_ESC                    '\033'
1185    #define CHAR_DEL                    '\177'
1186    
1187    #define STR_LF                      "\n"
1188    #define STR_NL                      STR_LF
1189    #define STR_NEL                     "\x85"
1190    #define STR_ESC                     "\033"
1191    #define STR_DEL                     "\177"
1192    
1193    #endif  /* EBCDIC */
1194    
1195    /* The remaining definitions work in both environments. */
1196    
1197  #define CHAR_HT                     '\t'  #define CHAR_HT                     '\t'
1198  #define CHAR_VT                     '\v'  #define CHAR_VT                     '\v'
1199  #define CHAR_FF                     '\f'  #define CHAR_FF                     '\f'
1200  #define CHAR_CR                     '\r'  #define CHAR_CR                     '\r'
 #define CHAR_NL                     '\n'  
1201  #define CHAR_BS                     '\b'  #define CHAR_BS                     '\b'
1202  #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  
1203    
1204  #define CHAR_SPACE                  ' '  #define CHAR_SPACE                  ' '
1205  #define CHAR_EXCLAMATION_MARK       '!'  #define CHAR_EXCLAMATION_MARK       '!'
# Line 1057  so that PCRE works on both ASCII and EBC Line 1301  so that PCRE works on both ASCII and EBC
1301  #define STR_VT                      "\v"  #define STR_VT                      "\v"
1302  #define STR_FF                      "\f"  #define STR_FF                      "\f"
1303  #define STR_CR                      "\r"  #define STR_CR                      "\r"
 #define STR_NL                      "\n"  
1304  #define STR_BS                      "\b"  #define STR_BS                      "\b"
1305  #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  
1306    
1307  #define STR_SPACE                   " "  #define STR_SPACE                   " "
1308  #define STR_EXCLAMATION_MARK        "!"  #define STR_EXCLAMATION_MARK        "!"
# Line 1203  so that PCRE works on both ASCII and EBC Line 1439  so that PCRE works on both ASCII and EBC
1439  #ifdef COMPILE_PCRE16  #ifdef COMPILE_PCRE16
1440  #define STRING_UTF_RIGHTPAR            "UTF16)"  #define STRING_UTF_RIGHTPAR            "UTF16)"
1441  #endif  #endif
1442    #ifdef COMPILE_PCRE32
1443    #define STRING_UTF_RIGHTPAR            "UTF32)"
1444    #endif
1445  #define STRING_UCP_RIGHTPAR            "UCP)"  #define STRING_UCP_RIGHTPAR            "UCP)"
1446  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"
1447    
# Line 1216  only. */ Line 1455  only. */
1455  #define CHAR_VT                     '\013'  #define CHAR_VT                     '\013'
1456  #define CHAR_FF                     '\014'  #define CHAR_FF                     '\014'
1457  #define CHAR_CR                     '\015'  #define CHAR_CR                     '\015'
1458  #define CHAR_NL                     '\012'  #define CHAR_LF                     '\012'
1459    #define CHAR_NL                     CHAR_LF
1460    #define CHAR_NEL                    ((unsigned char)'\x85')
1461  #define CHAR_BS                     '\010'  #define CHAR_BS                     '\010'
1462  #define CHAR_BEL                    '\007'  #define CHAR_BEL                    '\007'
1463  #define CHAR_ESC                    '\033'  #define CHAR_ESC                    '\033'
# Line 1463  only. */ Line 1704  only. */
1704  #ifdef COMPILE_PCRE16  #ifdef COMPILE_PCRE16
1705  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS
1706  #endif  #endif
1707    #ifdef COMPILE_PCRE32
1708    #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
1709    #endif
1710  #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
1711  #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
1712    
# Line 1479  only. */ Line 1723  only. */
1723  #endif  #endif
1724    
1725  #ifndef ESC_n  #ifndef ESC_n
1726  #define ESC_n CHAR_NL  #define ESC_n CHAR_LF
1727  #endif  #endif
1728    
1729  #ifndef ESC_r  #ifndef ESC_r
# Line 1504  only. */ Line 1748  only. */
1748  #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */  #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */
1749  #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 */
1750  #define PT_WORD       8    /* Word - L plus N plus underscore */  #define PT_WORD       8    /* Word - L plus N plus underscore */
1751    #define PT_CLIST      9    /* Pseudo-property: match character list */
1752    
1753  /* 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
1754  contain characters with values greater than 255. */  contain characters with values greater than 255. */
# Line 1526  used for [^] in JavaScript compatibility Line 1771  used for [^] in JavaScript compatibility
1771  non-DOTALL mode, "." behaves like \N.  non-DOTALL mode, "." behaves like \N.
1772    
1773  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.
1774  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.
1775  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
1776  looked up from a table.  looked up from a table.
1777    
# Line 1568  enum { Line 1813  enum {
1813    OP_NOT_WORDCHAR,       /* 10 \W */    OP_NOT_WORDCHAR,       /* 10 \W */
1814    OP_WORDCHAR,           /* 11 \w */    OP_WORDCHAR,           /* 11 \w */
1815    
1816    OP_ANY,            /* 12 Match any character except newline */    OP_ANY,            /* 12 Match any character except newline (\N) */
1817    OP_ALLANY,         /* 13 Match any character */    OP_ALLANY,         /* 13 Match any character */
1818    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 */
1819    OP_NOTPROP,        /* 15 \P (not Unicode property) */    OP_NOTPROP,        /* 15 \P (not Unicode property) */
# Line 1579  enum { Line 1824  enum {
1824    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */
1825    OP_VSPACE,         /* 21 \v (vertical whitespace) */    OP_VSPACE,         /* 21 \v (vertical whitespace) */
1826    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */
1827    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) */
1828    OP_EOD,            /* 24 End of data: \z */    OP_EOD,            /* 24 End of data (\z) */
1829    
1830    OP_CIRC,           /* 25 Start of line - not multiline */    OP_CIRC,           /* 25 Start of line - not multiline */
1831    OP_CIRCM,          /* 26 Start of line - multiline */    OP_CIRCM,          /* 26 Start of line - multiline */
# Line 1940  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 2185  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
2185         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
2186         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
2187         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
2188         ERR70, ERR71, ERR72, ERR73, ERR74, ERRCOUNT };         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERRCOUNT };
2189    
2190  /* JIT compiling modes. The function list is indexed by them. */  /* JIT compiling modes. The function list is indexed by them. */
2191  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
# Line 1963  fields are present. Currently PCRE alway Line 2208  fields are present. Currently PCRE alway
2208  NOTE NOTE NOTE  NOTE NOTE NOTE
2209  */  */
2210    
2211  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2212  #define REAL_PCRE real_pcre  #define REAL_PCRE real_pcre
2213  #else  #elif defined COMPILE_PCRE16
2214  #define REAL_PCRE real_pcre16  #define REAL_PCRE real_pcre16
2215    #elif defined COMPILE_PCRE32
2216    #define REAL_PCRE real_pcre32
2217  #endif  #endif
2218    
2219  typedef struct REAL_PCRE {  /* It is necessary to fork the struct for 32 bit, since it needs to use
2220     * pcre_uchar for first_char and req_char. Can't put an ifdef inside the
2221     * typedef since pcretest needs access to  the struct of the 8-, 16-
2222     * and 32-bit variants. */
2223    
2224    typedef struct real_pcre8_or_16 {
2225    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2226    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2227    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
# Line 1985  typedef struct REAL_PCRE { Line 2237  typedef struct REAL_PCRE {
2237    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2238    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2239    const pcre_uint8 *nullpad;      /* NULL padding */    const pcre_uint8 *nullpad;      /* NULL padding */
2240  } REAL_PCRE;  } real_pcre8_or_16;
2241    
2242    typedef struct real_pcre8_or_16 real_pcre;
2243    typedef struct real_pcre8_or_16 real_pcre16;
2244    
2245    typedef struct real_pcre32 {
2246      pcre_uint32 magic_number;
2247      pcre_uint32 size;               /* Total that was malloced */
2248      pcre_uint32 options;            /* Public options */
2249      pcre_uint16 flags;              /* Private flags */
2250      pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2251      pcre_uint16 top_bracket;        /* Highest numbered group */
2252      pcre_uint16 top_backref;        /* Highest numbered back reference */
2253      pcre_uint32 first_char;         /* Starting character */
2254      pcre_uint32 req_char;           /* This character must be seen */
2255      pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2256      pcre_uint16 name_entry_size;    /* Size of any name items */
2257      pcre_uint16 name_count;         /* Number of name items */
2258      pcre_uint16 ref_count;          /* Reference count */
2259      pcre_uint16 dummy1;             /* for later expansion */
2260      pcre_uint16 dummy2;             /* for later expansion */
2261      const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2262      void *nullpad;                  /* for later expansion */
2263    } real_pcre32;
2264    
2265    /* Assert that the size of REAL_PCRE is divisible by 8 */
2266    typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];
2267    
2268    /* Needed in pcretest to access some fields in the real_pcre* structures
2269     * directly. They're unified for 8/16/32 bits since the structs only differ
2270     * after these fields; if that ever changes, need to fork those defines into
2271     * 8/16 and 32 bit versions. */
2272    #define REAL_PCRE_MAGIC(re)     (((REAL_PCRE*)re)->magic_number)
2273    #define REAL_PCRE_SIZE(re)      (((REAL_PCRE*)re)->size)
2274    #define REAL_PCRE_OPTIONS(re)   (((REAL_PCRE*)re)->options)
2275    #define REAL_PCRE_FLAGS(re)     (((REAL_PCRE*)re)->flags)
2276    
2277  /* 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
2278  remark (see NOTE above) about extending this structure applies. */  remark (see NOTE above) about extending this structure applies. */
# Line 2036  typedef struct compile_data { Line 2323  typedef struct compile_data {
2323    int  external_flags;              /* External flag bits to be set */    int  external_flags;              /* External flag bits to be set */
2324    int  req_varyopt;                 /* "After variable item" flag for reqbyte */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2325    BOOL had_accept;                  /* (*ACCEPT) encountered */    BOOL had_accept;                  /* (*ACCEPT) encountered */
2326      BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
2327    BOOL check_lookbehind;            /* Lookbehinds need later checking */    BOOL check_lookbehind;            /* Lookbehinds need later checking */
2328    int  nltype;                      /* Newline type */    int  nltype;                      /* Newline type */
2329    int  nllen;                       /* Newline string length */    int  nllen;                       /* Newline string length */
# Line 2130  typedef struct match_data { Line 2418  typedef struct match_data {
2418    const  pcre_uchar *mark;        /* Mark pointer to pass back on success */    const  pcre_uchar *mark;        /* Mark pointer to pass back on success */
2419    const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */    const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */
2420    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 */
2421    #ifdef NO_RECURSE
2422      void  *match_frames_base;       /* For remembering malloc'd frames */
2423    #endif
2424  } match_data;  } match_data;
2425    
2426  /* 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 2186  total length. */ Line 2477  total length. */
2477    
2478  /* Internal function and data prefixes. */  /* Internal function and data prefixes. */
2479    
2480  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2481  #ifndef PUBL  #ifndef PUBL
2482  #define PUBL(name) pcre_##name  #define PUBL(name) pcre_##name
2483  #endif  #endif
2484  #ifndef PRIV  #ifndef PRIV
2485  #define PRIV(name) _pcre_##name  #define PRIV(name) _pcre_##name
2486  #endif  #endif
2487  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
2488  #ifndef PUBL  #ifndef PUBL
2489  #define PUBL(name) pcre16_##name  #define PUBL(name) pcre16_##name
2490  #endif  #endif
2491  #ifndef PRIV  #ifndef PRIV
2492  #define PRIV(name) _pcre16_##name  #define PRIV(name) _pcre16_##name
2493  #endif  #endif
2494    #elif defined COMPILE_PCRE32
2495    #ifndef PUBL
2496    #define PUBL(name) pcre32_##name
2497    #endif
2498    #ifndef PRIV
2499    #define PRIV(name) _pcre32_##name
2500    #endif
2501  #else  #else
2502  #error Unsupported compiling mode  #error Unsupported compiling mode
2503  #endif /* COMPILE_PCRE16 */  #endif /* COMPILE_PCRE[8|16|32] */
 #endif /* COMPILE_PCRE8 */  
2504    
2505  /* 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
2506  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 2224  but are not part of the PCRE public API. Line 2520  but are not part of the PCRE public API.
2520  pcre_tables.c module. */  pcre_tables.c module. */
2521    
2522  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
   
2523  extern const int            PRIV(utf8_table1)[];  extern const int            PRIV(utf8_table1)[];
2524  extern const int            PRIV(utf8_table1_size);  extern const int            PRIV(utf8_table1_size);
2525  extern const int            PRIV(utf8_table2)[];  extern const int            PRIV(utf8_table2)[];
2526  extern const int            PRIV(utf8_table3)[];  extern const int            PRIV(utf8_table3)[];
2527  extern const pcre_uint8     PRIV(utf8_table4)[];  extern const pcre_uint8     PRIV(utf8_table4)[];
   
2528  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
2529    
2530  extern const char           PRIV(utt_names)[];  extern const char           PRIV(utt_names)[];
2531  extern const ucp_type_table PRIV(utt)[];  extern const ucp_type_table PRIV(utt)[];
2532  extern const int            PRIV(utt_size);  extern const int            PRIV(utt_size);
2533    
2534    extern const pcre_uint8     PRIV(OP_lengths)[];
2535  extern const pcre_uint8     PRIV(default_tables)[];  extern const pcre_uint8     PRIV(default_tables)[];
2536    
2537  extern const pcre_uint8     PRIV(OP_lengths)[];  extern const pcre_uint32    PRIV(hspace_list)[];
2538    extern const pcre_uint32    PRIV(vspace_list)[];
2539    
2540    
2541  /* 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 2247  one of the exported public functions. Th Line 2543  one of the exported public functions. Th
2543  sense, but are not part of the PCRE public API. */  sense, but are not part of the PCRE public API. */
2544    
2545  /* String comparison functions. */  /* String comparison functions. */
2546  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2547    
2548  #define STRCMP_UC_UC(str1, str2) \  #define STRCMP_UC_UC(str1, str2) \
2549    strcmp((char *)(str1), (char *)(str2))    strcmp((char *)(str1), (char *)(str2))
# Line 2259  sense, but are not part of the PCRE publ Line 2555  sense, but are not part of the PCRE publ
2555    strncmp((char *)(str1), (str2), (num))    strncmp((char *)(str1), (str2), (num))
2556  #define STRLEN_UC(str) strlen((const char *)str)  #define STRLEN_UC(str) strlen((const char *)str)
2557    
2558  #else  #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2559    
2560  extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,  extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,
2561                             const pcre_uchar *);                             const pcre_uchar *);
# Line 2281  extern unsigned int      PRIV(strlen_uc) Line 2577  extern unsigned int      PRIV(strlen_uc)
2577    PRIV(strncmp_uc_c8)((str1), (str2), (num))    PRIV(strncmp_uc_c8)((str1), (str2), (num))
2578  #define STRLEN_UC(str) PRIV(strlen_uc)(str)  #define STRLEN_UC(str) PRIV(strlen_uc)(str)
2579    
2580  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE[8|16|32] */
2581    
2582  extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);  extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);
2583  extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,  extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
# Line 2290  extern int               PRIV(ord2utf)(p Line 2586  extern int               PRIV(ord2utf)(p
2586  extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);  extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);
2587  extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,  extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2588                             int *, BOOL);                             int *, BOOL);
2589  extern BOOL              PRIV(xclass)(int, const pcre_uchar *, BOOL);  extern BOOL              PRIV(xclass)(pcre_uint32, const pcre_uchar *, BOOL);
2590    
2591  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2592  extern void              PRIV(jit_compile)(const REAL_PCRE *, PUBL(extra) *, int);  extern void              PRIV(jit_compile)(const REAL_PCRE *,
2593  extern int               PRIV(jit_exec)(const REAL_PCRE *, void *,                             PUBL(extra) *, int);
2594                             const pcre_uchar *, int, int, int, int, int *, int,  extern int               PRIV(jit_exec)(const REAL_PCRE *, const PUBL(extra) *,
2595                             pcre_uchar **);                             const pcre_uchar *, int, int, int, int *, int);
2596  extern void              PRIV(jit_free)(void *);  extern void              PRIV(jit_free)(void *);
2597  extern int               PRIV(jit_get_size)(void *);  extern int               PRIV(jit_get_size)(void *);
2598  extern const char*       PRIV(jit_get_target)(void);  extern const char*       PRIV(jit_get_target)(void);
# Line 2305  extern const char*       PRIV(jit_get_ta Line 2601  extern const char*       PRIV(jit_get_ta
2601  /* Unicode character database (UCD) */  /* Unicode character database (UCD) */
2602    
2603  typedef struct {  typedef struct {
2604    pcre_uint8 script;    pcre_uint8 script;     /* ucp_Arabic, etc. */
2605    pcre_uint8 chartype;    pcre_uint8 chartype;   /* ucp_Cc, etc. (general categories) */
2606    pcre_int32 other_case;    pcre_uint8 gbprop;     /* ucp_gbControl, etc. (grapheme break property) */
2607      pcre_uint8 caseset;    /* offset to multichar other cases or zero */
2608      pcre_int32 other_case; /* offset to other case, or zero if none */
2609  } ucd_record;  } ucd_record;
2610    
2611    extern const pcre_uint32 PRIV(ucd_caseless_sets)[];
2612  extern const ucd_record  PRIV(ucd_records)[];  extern const ucd_record  PRIV(ucd_records)[];
2613  extern const pcre_uint8  PRIV(ucd_stage1)[];  extern const pcre_uint8  PRIV(ucd_stage1)[];
2614  extern const pcre_uint16 PRIV(ucd_stage2)[];  extern const pcre_uint16 PRIV(ucd_stage2)[];
2615  extern const int         PRIV(ucp_gentype)[];  extern const int         PRIV(ucp_gentype)[];
2616    extern const pcre_uint32 PRIV(ucp_gbtable)[];
2617  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2618  extern const int         PRIV(ucp_typerange)[];  extern const int         PRIV(ucp_typerange)[];
2619  #endif  #endif
# Line 2326  extern const int         PRIV(ucp_typera Line 2626  extern const int         PRIV(ucp_typera
2626          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(ch) / UCD_BLOCK_SIZE] * \          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(ch) / UCD_BLOCK_SIZE] * \
2627          UCD_BLOCK_SIZE + (ch) % UCD_BLOCK_SIZE])          UCD_BLOCK_SIZE + (ch) % UCD_BLOCK_SIZE])
2628    
2629  #define UCD_CHARTYPE(ch)  GET_UCD(ch)->chartype  #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
2630  #define UCD_SCRIPT(ch)    GET_UCD(ch)->script  #define UCD_SCRIPT(ch)      GET_UCD(ch)->script
2631  #define UCD_CATEGORY(ch)  PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]  #define UCD_CATEGORY(ch)    PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]
2632  #define UCD_OTHERCASE(ch) (ch + GET_UCD(ch)->other_case)  #define UCD_GRAPHBREAK(ch)  GET_UCD(ch)->gbprop
2633    #define UCD_CASESET(ch)     GET_UCD(ch)->caseset
2634    #define UCD_OTHERCASE(ch)   (ch + GET_UCD(ch)->other_case)
2635    
2636  #endif /* SUPPORT_UCP */  #endif /* SUPPORT_UCP */
2637    

Legend:
Removed from v.932  
changed lines
  Added in v.1057

  ViewVC Help
Powered by ViewVC 1.1.5