diff options
| -rw-r--r-- | include/_ctype.h | 35 | ||||
| -rw-r--r-- | include/ctype.h | 42 | ||||
| -rw-r--r-- | lib/libc/locale/big5.c | 3 | ||||
| -rw-r--r-- | lib/libc/locale/euc.c | 3 | ||||
| -rw-r--r-- | lib/libc/locale/gb18030.c | 3 | ||||
| -rw-r--r-- | lib/libc/locale/gb2312.c | 3 | ||||
| -rw-r--r-- | lib/libc/locale/gbk.c | 3 | ||||
| -rw-r--r-- | lib/libc/locale/isctype.c | 38 | ||||
| -rw-r--r-- | lib/libc/locale/mskanji.c | 3 | ||||
| -rw-r--r-- | lib/libc/locale/none.c | 7 | ||||
| -rw-r--r-- | lib/libc/locale/setrunelocale.c | 5 | ||||
| -rw-r--r-- | lib/libc/locale/utf8.c | 8 | ||||
| -rw-r--r-- | share/mklocale/UTF-8.src | 3 |
13 files changed, 114 insertions, 42 deletions
diff --git a/include/_ctype.h b/include/_ctype.h index 1a42952a09dcd..e27ab24f6c199 100644 --- a/include/_ctype.h +++ b/include/_ctype.h @@ -87,6 +87,8 @@ __END_DECLS #define __inline #endif +extern int __mb_sb_limit; + /* * Use inline functions if we are allowed to and the compiler supports them. */ @@ -103,15 +105,28 @@ __maskrune(__ct_rune_t _c, unsigned long _f) } static __inline int +__sbmaskrune(__ct_rune_t _c, unsigned long _f) +{ + return (_c < 0 || _c >= __mb_sb_limit) ? 0 : + _CurrentRuneLocale->__runetype[_c] & _f; +} + +static __inline int __istype(__ct_rune_t _c, unsigned long _f) { return (!!__maskrune(_c, _f)); } static __inline int +__sbistype(__ct_rune_t _c, unsigned long _f) +{ + return (!!__sbmaskrune(_c, _f)); +} + +static __inline int __isctype(__ct_rune_t _c, unsigned long _f) { - return (_c < 0 || _c >= _CACHED_RUNES) ? 0 : + return (_c < 0 || _c >= __mb_sb_limit) ? 0 : !!(_DefaultRuneLocale.__runetype[_c] & _f); } @@ -123,12 +138,26 @@ __toupper(__ct_rune_t _c) } static __inline __ct_rune_t +__sbtoupper(__ct_rune_t _c) +{ + return (_c < 0 || _c >= __mb_sb_limit) ? _c : + _CurrentRuneLocale->__mapupper[_c]; +} + +static __inline __ct_rune_t __tolower(__ct_rune_t _c) { return (_c < 0 || _c >= _CACHED_RUNES) ? ___tolower(_c) : _CurrentRuneLocale->__maplower[_c]; } +static __inline __ct_rune_t +__sbtolower(__ct_rune_t _c) +{ + return (_c < 0 || _c >= __mb_sb_limit) ? _c : + _CurrentRuneLocale->__maplower[_c]; +} + static __inline int __wcwidth(__ct_rune_t _c) { @@ -146,10 +175,14 @@ __wcwidth(__ct_rune_t _c) __BEGIN_DECLS int __maskrune(__ct_rune_t, unsigned long); +int __sbmaskrune(__ct_rune_t, unsigned long); int __istype(__ct_rune_t, unsigned long); +int __sbistype(__ct_rune_t, unsigned long); int __isctype(__ct_rune_t, unsigned long); __ct_rune_t __toupper(__ct_rune_t); +__ct_rune_t __sbtoupper(__ct_rune_t); __ct_rune_t __tolower(__ct_rune_t); +__ct_rune_t __sbtolower(__ct_rune_t); int __wcwidth(__ct_rune_t); __END_DECLS #endif /* using inlines */ diff --git a/include/ctype.h b/include/ctype.h index 0825ff652c282..dfc89ccd5951a 100644 --- a/include/ctype.h +++ b/include/ctype.h @@ -86,19 +86,19 @@ int isspecial(int); #endif __END_DECLS -#define isalnum(c) __istype((c), _CTYPE_A|_CTYPE_D) -#define isalpha(c) __istype((c), _CTYPE_A) -#define iscntrl(c) __istype((c), _CTYPE_C) +#define isalnum(c) __sbistype((c), _CTYPE_A|_CTYPE_D) +#define isalpha(c) __sbistype((c), _CTYPE_A) +#define iscntrl(c) __sbistype((c), _CTYPE_C) #define isdigit(c) __isctype((c), _CTYPE_D) /* ANSI -- locale independent */ -#define isgraph(c) __istype((c), _CTYPE_G) -#define islower(c) __istype((c), _CTYPE_L) -#define isprint(c) __istype((c), _CTYPE_R) -#define ispunct(c) __istype((c), _CTYPE_P) -#define isspace(c) __istype((c), _CTYPE_S) -#define isupper(c) __istype((c), _CTYPE_U) +#define isgraph(c) __sbistype((c), _CTYPE_G) +#define islower(c) __sbistype((c), _CTYPE_L) +#define isprint(c) __sbistype((c), _CTYPE_R) +#define ispunct(c) __sbistype((c), _CTYPE_P) +#define isspace(c) __sbistype((c), _CTYPE_S) +#define isupper(c) __sbistype((c), _CTYPE_U) #define isxdigit(c) __isctype((c), _CTYPE_X) /* ANSI -- locale independent */ -#define tolower(c) __tolower(c) -#define toupper(c) __toupper(c) +#define tolower(c) __sbtolower(c) +#define toupper(c) __sbtoupper(c) #if __XSI_VISIBLE /* @@ -112,24 +112,24 @@ __END_DECLS * * XXX isascii() and toascii() should similarly be undocumented. */ -#define _tolower(c) __tolower(c) -#define _toupper(c) __toupper(c) +#define _tolower(c) __sbtolower(c) +#define _toupper(c) __sbtoupper(c) #define isascii(c) (((c) & ~0x7F) == 0) #define toascii(c) ((c) & 0x7F) #endif #if __ISO_C_VISIBLE >= 1999 -#define isblank(c) __istype((c), _CTYPE_B) +#define isblank(c) __sbistype((c), _CTYPE_B) #endif #if __BSD_VISIBLE -#define digittoint(c) __maskrune((c), 0xFF) -#define ishexnumber(c) __istype((c), _CTYPE_X) -#define isideogram(c) __istype((c), _CTYPE_I) -#define isnumber(c) __istype((c), _CTYPE_D) -#define isphonogram(c) __istype((c), _CTYPE_Q) -#define isrune(c) __istype((c), 0xFFFFFF00L) -#define isspecial(c) __istype((c), _CTYPE_T) +#define digittoint(c) __sbmaskrune((c), 0xFF) +#define ishexnumber(c) __sbistype((c), _CTYPE_X) +#define isideogram(c) __sbistype((c), _CTYPE_I) +#define isnumber(c) __sbistype((c), _CTYPE_D) +#define isphonogram(c) __sbistype((c), _CTYPE_Q) +#define isrune(c) __sbistype((c), 0xFFFFFF00L) +#define isspecial(c) __sbistype((c), _CTYPE_T) #endif #endif /* !_CTYPE_H_ */ diff --git a/lib/libc/locale/big5.c b/lib/libc/locale/big5.c index 44b9957b59789..19977d035dc51 100644 --- a/lib/libc/locale/big5.c +++ b/lib/libc/locale/big5.c @@ -49,6 +49,8 @@ __FBSDID("$FreeBSD$"); #include <wchar.h> #include "mblocal.h" +extern int __mb_sb_limit; + static size_t _BIG5_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict); static int _BIG5_mbsinit(const mbstate_t *); @@ -68,6 +70,7 @@ _BIG5_init(_RuneLocale *rl) __mbsinit = _BIG5_mbsinit; _CurrentRuneLocale = rl; __mb_cur_max = 2; + __mb_sb_limit = 128; return (0); } diff --git a/lib/libc/locale/euc.c b/lib/libc/locale/euc.c index b3b35edb5b6c4..188073e49e38d 100644 --- a/lib/libc/locale/euc.c +++ b/lib/libc/locale/euc.c @@ -49,6 +49,8 @@ __FBSDID("$FreeBSD$"); #include <wchar.h> #include "mblocal.h" +extern int __mb_sb_limit; + static size_t _EUC_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict); static int _EUC_mbsinit(const mbstate_t *); @@ -116,6 +118,7 @@ _EUC_init(_RuneLocale *rl) __mbrtowc = _EUC_mbrtowc; __wcrtomb = _EUC_wcrtomb; __mbsinit = _EUC_mbsinit; + __mb_sb_limit = 256; return (0); } diff --git a/lib/libc/locale/gb18030.c b/lib/libc/locale/gb18030.c index 3e43179d92137..1457d3e7540c3 100644 --- a/lib/libc/locale/gb18030.c +++ b/lib/libc/locale/gb18030.c @@ -39,6 +39,8 @@ __FBSDID("$FreeBSD$"); #include <wchar.h> #include "mblocal.h" +extern int __mb_sb_limit; + static size_t _GB18030_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict); static int _GB18030_mbsinit(const mbstate_t *); @@ -59,6 +61,7 @@ _GB18030_init(_RuneLocale *rl) __mbsinit = _GB18030_mbsinit; _CurrentRuneLocale = rl; __mb_cur_max = 4; + __mb_sb_limit = 128; return (0); } diff --git a/lib/libc/locale/gb2312.c b/lib/libc/locale/gb2312.c index 232dabae866e1..74a7bdc3be79a 100644 --- a/lib/libc/locale/gb2312.c +++ b/lib/libc/locale/gb2312.c @@ -35,6 +35,8 @@ __FBSDID("$FreeBSD$"); #include <wchar.h> #include "mblocal.h" +extern int __mb_sb_limit; + static size_t _GB2312_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict); static int _GB2312_mbsinit(const mbstate_t *); @@ -55,6 +57,7 @@ _GB2312_init(_RuneLocale *rl) __wcrtomb = _GB2312_wcrtomb; __mbsinit = _GB2312_mbsinit; __mb_cur_max = 2; + __mb_sb_limit = 128; return (0); } diff --git a/lib/libc/locale/gbk.c b/lib/libc/locale/gbk.c index 573fd08f34efa..1929861df9c39 100644 --- a/lib/libc/locale/gbk.c +++ b/lib/libc/locale/gbk.c @@ -46,6 +46,8 @@ __FBSDID("$FreeBSD$"); #include <wchar.h> #include "mblocal.h" +extern int __mb_sb_limit; + static size_t _GBK_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict); static int _GBK_mbsinit(const mbstate_t *); @@ -65,6 +67,7 @@ _GBK_init(_RuneLocale *rl) __mbsinit = _GBK_mbsinit; _CurrentRuneLocale = rl; __mb_cur_max = 2; + __mb_sb_limit = 128; return (0); } diff --git a/lib/libc/locale/isctype.c b/lib/libc/locale/isctype.c index d25772bec279e..e0e0db13f5b03 100644 --- a/lib/libc/locale/isctype.c +++ b/lib/libc/locale/isctype.c @@ -52,7 +52,7 @@ int digittoint(c) int c; { - return (__maskrune(c, 0xFF)); + return (__sbmaskrune(c, 0xFF)); } #undef isalnum @@ -60,7 +60,7 @@ int isalnum(c) int c; { - return (__istype(c, _CTYPE_A|_CTYPE_D)); + return (__sbistype(c, _CTYPE_A|_CTYPE_D)); } #undef isalpha @@ -68,7 +68,7 @@ int isalpha(c) int c; { - return (__istype(c, _CTYPE_A)); + return (__sbistype(c, _CTYPE_A)); } #undef isascii @@ -84,7 +84,7 @@ int isblank(c) int c; { - return (__istype(c, _CTYPE_B)); + return (__sbistype(c, _CTYPE_B)); } #undef iscntrl @@ -92,7 +92,7 @@ int iscntrl(c) int c; { - return (__istype(c, _CTYPE_C)); + return (__sbistype(c, _CTYPE_C)); } #undef isdigit @@ -108,7 +108,7 @@ int isgraph(c) int c; { - return (__istype(c, _CTYPE_G)); + return (__sbistype(c, _CTYPE_G)); } #undef ishexnumber @@ -116,7 +116,7 @@ int ishexnumber(c) int c; { - return (__istype(c, _CTYPE_X)); + return (__sbistype(c, _CTYPE_X)); } #undef isideogram @@ -124,7 +124,7 @@ int isideogram(c) int c; { - return (__istype(c, _CTYPE_I)); + return (__sbistype(c, _CTYPE_I)); } #undef islower @@ -132,7 +132,7 @@ int islower(c) int c; { - return (__istype(c, _CTYPE_L)); + return (__sbistype(c, _CTYPE_L)); } #undef isnumber @@ -140,7 +140,7 @@ int isnumber(c) int c; { - return (__istype(c, _CTYPE_D)); + return (__sbistype(c, _CTYPE_D)); } #undef isphonogram @@ -148,7 +148,7 @@ int isphonogram(c) int c; { - return (__istype(c, _CTYPE_Q)); + return (__sbistype(c, _CTYPE_Q)); } #undef isprint @@ -156,7 +156,7 @@ int isprint(c) int c; { - return (__istype(c, _CTYPE_R)); + return (__sbistype(c, _CTYPE_R)); } #undef ispunct @@ -164,7 +164,7 @@ int ispunct(c) int c; { - return (__istype(c, _CTYPE_P)); + return (__sbistype(c, _CTYPE_P)); } #undef isrune @@ -172,7 +172,7 @@ int isrune(c) int c; { - return (__istype(c, 0xFFFFFF00L)); + return (__sbistype(c, 0xFFFFFF00L)); } #undef isspace @@ -180,7 +180,7 @@ int isspace(c) int c; { - return (__istype(c, _CTYPE_S)); + return (__sbistype(c, _CTYPE_S)); } #undef isspecial @@ -188,7 +188,7 @@ int isspecial(c) int c; { - return (__istype(c, _CTYPE_T)); + return (__sbistype(c, _CTYPE_T)); } #undef isupper @@ -196,7 +196,7 @@ int isupper(c) int c; { - return (__istype(c, _CTYPE_U)); + return (__sbistype(c, _CTYPE_U)); } #undef isxdigit @@ -220,7 +220,7 @@ int tolower(c) int c; { - return (__tolower(c)); + return (__sbtolower(c)); } #undef toupper @@ -228,6 +228,6 @@ int toupper(c) int c; { - return (__toupper(c)); + return (__sbtoupper(c)); } diff --git a/lib/libc/locale/mskanji.c b/lib/libc/locale/mskanji.c index aba87e75ec86c..9ee91de28c155 100644 --- a/lib/libc/locale/mskanji.c +++ b/lib/libc/locale/mskanji.c @@ -47,6 +47,8 @@ __FBSDID("$FreeBSD$"); #include <wchar.h> #include "mblocal.h" +extern int __mb_sb_limit; + static size_t _MSKanji_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict); static int _MSKanji_mbsinit(const mbstate_t *); @@ -66,6 +68,7 @@ _MSKanji_init(_RuneLocale *rl) __mbsinit = _MSKanji_mbsinit; _CurrentRuneLocale = rl; __mb_cur_max = 2; + __mb_sb_limit = 256; return (0); } diff --git a/lib/libc/locale/none.c b/lib/libc/locale/none.c index 06b55b9071d50..cc84b94305e0c 100644 --- a/lib/libc/locale/none.c +++ b/lib/libc/locale/none.c @@ -62,6 +62,11 @@ static size_t _none_wcrtomb(char * __restrict, wchar_t, static size_t _none_wcsnrtombs(char * __restrict, const wchar_t ** __restrict, size_t, size_t, mbstate_t * __restrict); +/* setup defaults */ + +int __mb_cur_max = 1; +int __mb_sb_limit = 256; /* Expected to be <= _CACHED_RUNES */ + int _none_init(_RuneLocale *rl) { @@ -73,6 +78,7 @@ _none_init(_RuneLocale *rl) __wcsnrtombs = _none_wcsnrtombs; _CurrentRuneLocale = rl; __mb_cur_max = 1; + __mb_sb_limit = 256; return(0); } @@ -180,7 +186,6 @@ _none_wcsnrtombs(char * __restrict dst, const wchar_t ** __restrict src, /* setup defaults */ -int __mb_cur_max = 1; size_t (*__mbrtowc)(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict) = _none_mbrtowc; int (*__mbsinit)(const mbstate_t *) = _none_mbsinit; diff --git a/lib/libc/locale/setrunelocale.c b/lib/libc/locale/setrunelocale.c index 398bfc5bb46db..73cd0e78c1944 100644 --- a/lib/libc/locale/setrunelocale.c +++ b/lib/libc/locale/setrunelocale.c @@ -49,6 +49,8 @@ __FBSDID("$FreeBSD$"); #include "mblocal.h" #include "setlocale.h" +extern int __mb_sb_limit; + extern _RuneLocale *_Read_RuneMagi(FILE *); static int __setrunelocale(const char *); @@ -63,6 +65,7 @@ __setrunelocale(const char *encoding) static char ctype_encoding[ENCODING_LEN + 1]; static _RuneLocale *CachedRuneLocale; static int Cached__mb_cur_max; + static int Cached__mb_sb_limit; static size_t (*Cached__mbrtowc)(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict); static size_t (*Cached__wcrtomb)(char * __restrict, wchar_t, @@ -89,6 +92,7 @@ __setrunelocale(const char *encoding) strcmp(encoding, ctype_encoding) == 0) { _CurrentRuneLocale = CachedRuneLocale; __mb_cur_max = Cached__mb_cur_max; + __mb_sb_limit = Cached__mb_sb_limit; __mbrtowc = Cached__mbrtowc; __mbsinit = Cached__mbsinit; __mbsnrtowcs = Cached__mbsnrtowcs; @@ -151,6 +155,7 @@ __setrunelocale(const char *encoding) } CachedRuneLocale = _CurrentRuneLocale; Cached__mb_cur_max = __mb_cur_max; + Cached__mb_sb_limit = __mb_sb_limit; Cached__mbrtowc = __mbrtowc; Cached__mbsinit = __mbsinit; Cached__mbsnrtowcs = __mbsnrtowcs; diff --git a/lib/libc/locale/utf8.c b/lib/libc/locale/utf8.c index e467fc0533904..0ccadd3e8d3bf 100644 --- a/lib/libc/locale/utf8.c +++ b/lib/libc/locale/utf8.c @@ -35,6 +35,8 @@ __FBSDID("$FreeBSD$"); #include <wchar.h> #include "mblocal.h" +extern int __mb_sb_limit; + static size_t _UTF8_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t, mbstate_t * __restrict); static int _UTF8_mbsinit(const mbstate_t *); @@ -63,6 +65,12 @@ _UTF8_init(_RuneLocale *rl) __wcsnrtombs = _UTF8_wcsnrtombs; _CurrentRuneLocale = rl; __mb_cur_max = 6; + /* + * UCS-4 encoding used as the internal representation, so + * slots 0x0080-0x00FF are occuped and must be excluded + * from the single byte ctype by setting the limit. + */ + __mb_sb_limit = 128; return (0); } diff --git a/share/mklocale/UTF-8.src b/share/mklocale/UTF-8.src index f8ffd81a7bbd0..73a7cc79ff677 100644 --- a/share/mklocale/UTF-8.src +++ b/share/mklocale/UTF-8.src @@ -3,6 +3,9 @@ * * Generated by Hye-Shik Chang <perky@FreeBSD.org> * + * WARNING: UCS-4 used here as the internal representation, not UTF-8. + * See lib/libc/locale/utf8.c. + * * $FreeBSD$ */ |