2 * msvcrt.dll mbcs functions
4 * Copyright 1999 Alexandre Julliard
5 * Copyright 2000 Jon Griffths
7 * This library is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * This library is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with this library; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
22 * Not currently binary compatible with win32. MSVCRT_mbctype must be
23 * populated correctly and the ismb* functions should reference it.
27 #include "wine/unicode.h"
28 #include "wine/debug.h"
30 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt
);
32 unsigned char MSVCRT_mbctype
[257] = { 0 };
33 static int g_mbcp_is_multibyte
= 0;
35 int MSVCRT___mb_cur_max
= 1;
37 /* It seems that the data about valid trail bytes is not available from kernel32
38 * so we have to store is here. The format is the same as for lead bytes in CPINFO */
39 struct cp_extra_info_t
42 BYTE TrailBytes
[MAX_LEADBYTES
];
45 static struct cp_extra_info_t g_cpextrainfo
[] =
47 {932, {0x40, 0x7e, 0x80, 0xfc, 0, 0}},
48 {936, {0x40, 0xfe, 0, 0}},
49 {949, {0x41, 0xfe, 0, 0}},
50 {950, {0x40, 0x7e, 0xa1, 0xfe, 0, 0}},
51 {20932, {1, 255, 0, 0}}, /* seems to give different results on different systems */
52 {0, {1, 255, 0, 0}} /* match all with FIXME */
55 static MSVCRT_wchar_t
msvcrt_mbc_to_wc(unsigned int ch
)
65 mbch
[0] = (ch
>> 8) & 0xff;
69 if (!MultiByteToWideChar(MSVCRT___lc_codepage
, 0, mbch
, n_chars
, &chW
, 1))
71 WARN("MultiByteToWideChar failed on %x\n", ch
);
77 static inline size_t u_strlen( const unsigned char *str
)
79 return strlen( (const char*) str
);
82 static inline unsigned char* u_strncat( unsigned char* dst
, const unsigned char* src
, size_t len
)
84 return (unsigned char*)strncat( (char*)dst
, (const char*)src
, len
);
87 static inline int u_strcmp( const unsigned char *s1
, const unsigned char *s2
)
89 return strcmp( (const char*)s1
, (const char*)s2
);
92 static inline int u_strcasecmp( const unsigned char *s1
, const unsigned char *s2
)
94 return strcasecmp( (const char*)s1
, (const char*)s2
);
97 static inline int u_strncmp( const unsigned char *s1
, const unsigned char *s2
, size_t len
)
99 return strncmp( (const char*)s1
, (const char*)s2
, len
);
102 static inline int u_strncasecmp( const unsigned char *s1
, const unsigned char *s2
, size_t len
)
104 return strncasecmp( (const char*)s1
, (const char*)s2
, len
);
107 static inline unsigned char *u_strchr( const unsigned char *s
, unsigned char x
)
109 return (unsigned char*) strchr( (const char*)s
, x
);
112 static inline unsigned char *u_strrchr( const unsigned char *s
, unsigned char x
)
114 return (unsigned char*) strrchr( (const char*)s
, x
);
117 static inline unsigned char *u_strtok( unsigned char *s
, const unsigned char *delim
)
119 return (unsigned char*) strtok( (char*)s
, (const char*)delim
);
122 static inline unsigned char *u__strset( unsigned char *s
, unsigned char c
)
124 return (unsigned char*) _strset( (char*)s
, c
);
127 static inline unsigned char *u__strnset( unsigned char *s
, unsigned char c
, size_t len
)
129 return (unsigned char*) _strnset( (char*)s
, c
, len
);
132 static inline size_t u_strcspn( const unsigned char *s
, const unsigned char *rej
)
134 return strcspn( (const char *)s
, (const char*)rej
);
137 /*********************************************************************
138 * __p__mbctype (MSVCRT.@)
140 unsigned char* CDECL
__p__mbctype(void)
142 return MSVCRT_mbctype
;
145 /*********************************************************************
146 * ___mb_cur_max_func(MSVCRT.@)
148 int* CDECL
MSVCRT____mb_cur_max_func(void)
150 return &MSVCRT___mb_cur_max
;
153 /*********************************************************************
154 * _setmbcp (MSVCRT.@)
156 int CDECL
_setmbcp(int cp
)
178 newcp
= MSVCRT___lc_codepage
;
181 newcp
= 20127; /* ASCII */
188 if (!GetCPInfo(newcp
, &cpi
))
190 WARN("Codepage %d not found\n", newcp
);
191 *MSVCRT__errno() = MSVCRT_EINVAL
;
195 /* setup the _mbctype */
196 memset(MSVCRT_mbctype
, 0, sizeof(MSVCRT_mbctype
));
198 bytes
= cpi
.LeadByte
;
199 while (bytes
[0] || bytes
[1])
201 for (i
= bytes
[0]; i
<= bytes
[1]; i
++)
202 MSVCRT_mbctype
[i
+ 1] |= _M1
;
206 if (cpi
.MaxCharSize
> 1)
208 /* trail bytes not available through kernel32 but stored in a structure in msvcrt */
209 struct cp_extra_info_t
*cpextra
= g_cpextrainfo
;
211 g_mbcp_is_multibyte
= 1;
214 if (cpextra
->cp
== 0 || cpextra
->cp
== newcp
)
216 if (cpextra
->cp
== 0)
217 FIXME("trail bytes data not available for DBCS codepage %d - assuming all bytes\n", newcp
);
219 bytes
= cpextra
->TrailBytes
;
220 while (bytes
[0] || bytes
[1])
222 for (i
= bytes
[0]; i
<= bytes
[1]; i
++)
223 MSVCRT_mbctype
[i
+ 1] |= _M2
;
232 g_mbcp_is_multibyte
= 0;
234 /* we can't use GetStringTypeA directly because we don't have a locale - only a code page
237 for (i
= 0; i
< 256; i
++)
238 if (!(MSVCRT_mbctype
[i
+ 1] & _M1
))
239 bufA
[charcount
++] = i
;
241 ret
= MultiByteToWideChar(newcp
, 0, bufA
, charcount
, bufW
, charcount
);
242 if (ret
!= charcount
)
243 ERR("MultiByteToWideChar of chars failed for cp %d, ret=%d (exp %d), error=%d\n", newcp
, ret
, charcount
, GetLastError());
245 GetStringTypeW(CT_CTYPE1
, bufW
, charcount
, chartypes
);
247 curr_type
= chartypes
;
248 for (i
= 0; i
< 256; i
++)
249 if (!(MSVCRT_mbctype
[i
+ 1] & _M1
))
251 if ((*curr_type
) & C1_UPPER
)
252 MSVCRT_mbctype
[i
+ 1] |= _SBUP
;
253 if ((*curr_type
) & C1_LOWER
)
254 MSVCRT_mbctype
[i
+ 1] |= _SBLOW
;
258 if (newcp
== 932) /* CP932 only - set _MP and _MS */
260 /* On Windows it's possible to calculate the _MP and _MS from CT_CTYPE1
261 * and CT_CTYPE3. But as of Wine 0.9.43 we return wrong values what makes
262 * it hard. As this is set only for codepage 932 we hardcode it what gives
263 * also faster execution.
265 for (i
= 161; i
<= 165; i
++)
266 MSVCRT_mbctype
[i
+ 1] |= _MP
;
267 for (i
= 166; i
<= 223; i
++)
268 MSVCRT_mbctype
[i
+ 1] |= _MS
;
271 MSVCRT___lc_collate_cp
= MSVCRT___lc_codepage
= newcp
;
272 TRACE("(%d) -> %d\n", cp
, MSVCRT___lc_codepage
);
276 /*********************************************************************
277 * _getmbcp (MSVCRT.@)
279 int CDECL
_getmbcp(void)
281 return MSVCRT___lc_codepage
;
284 /*********************************************************************
285 * _mbsnextc(MSVCRT.@)
287 unsigned int CDECL
_mbsnextc(const unsigned char* str
)
290 return *str
<< 8 | str
[1];
294 /*********************************************************************
295 * _mbctolower(MSVCRT.@)
297 unsigned int CDECL
_mbctolower(unsigned int c
)
299 if (MSVCRT_isleadbyte(c
))
301 FIXME("Handle MBC chars\n");
304 return tolower(c
); /* ASCII CP or SB char */
307 /*********************************************************************
308 * _mbctoupper(MSVCRT.@)
310 unsigned int CDECL
_mbctoupper(unsigned int c
)
312 if (MSVCRT_isleadbyte(c
))
314 FIXME("Handle MBC chars\n");
317 return toupper(c
); /* ASCII CP or SB char */
320 /*********************************************************************
323 unsigned char* CDECL
_mbsdec(const unsigned char* start
, const unsigned char* cur
)
325 if(MSVCRT___mb_cur_max
> 1)
326 return (unsigned char *)(_ismbstrail(start
,cur
-1) ? cur
- 2 : cur
-1);
328 return (unsigned char *)cur
- 1; /* ASCII CP or SB char */
331 /*********************************************************************
334 unsigned int CDECL
_mbclen(const unsigned char* str
)
336 return _ismbblead(*str
) ? 2 : 1;
339 /*********************************************************************
342 unsigned char* CDECL
_mbsinc(const unsigned char* str
)
344 return (unsigned char *)(str
+ _mbclen(str
));
347 /*********************************************************************
350 unsigned char* CDECL
_mbsninc(const unsigned char* str
, MSVCRT_size_t num
)
355 while (num
> 0 && *str
)
357 if (_ismbblead(*str
))
367 return (unsigned char*)str
;
370 /*********************************************************************
373 MSVCRT_size_t CDECL
_mbslen(const unsigned char* str
)
375 MSVCRT_size_t len
= 0;
378 if (_ismbblead(*str
))
381 if (!*str
) /* count only full chars */
390 /*********************************************************************
393 void CDECL
_mbccpy(unsigned char* dest
, const unsigned char* src
)
397 *++dest
= *++src
; /* MB char */
400 /*********************************************************************
403 * The parameter n is the number or characters to copy, not the size of
404 * the buffer. Use _mbsnbcpy for a function analogical to strncpy
406 unsigned char* CDECL
_mbsncpy(unsigned char* dst
, const unsigned char* src
, MSVCRT_size_t n
)
408 unsigned char* ret
= dst
;
411 if (g_mbcp_is_multibyte
)
416 if (_ismbblead(*src
))
436 if (!(*dst
++ = *src
++)) break;
439 while (n
--) *dst
++ = 0;
443 /*********************************************************************
444 * _mbsnbcpy_s(MSVCRT.@)
446 * Unlike _mbsnbcpy this function does not pad the rest of the dest
449 int CDECL
_mbsnbcpy_s(unsigned char* dst
, MSVCRT_size_t size
, const unsigned char* src
, MSVCRT_size_t n
)
451 MSVCRT_size_t pos
= 0;
453 if(!dst
|| size
== 0)
454 return MSVCRT_EINVAL
;
458 return MSVCRT_EINVAL
;
463 if(g_mbcp_is_multibyte
)
471 return MSVCRT_ERANGE
;
473 is_lead
= (!is_lead
&& _ismbblead(*src
));
478 if (is_lead
) /* if string ends with a lead, remove it */
489 return MSVCRT_ERANGE
;
502 return MSVCRT_ERANGE
;
508 /*********************************************************************
509 * _mbsnbcpy(MSVCRT.@)
511 * Like strncpy this function doesn't enforce the string to be
514 unsigned char* CDECL
_mbsnbcpy(unsigned char* dst
, const unsigned char* src
, MSVCRT_size_t n
)
516 unsigned char* ret
= dst
;
519 if(g_mbcp_is_multibyte
)
524 is_lead
= (!is_lead
&& _ismbblead(*src
));
529 if (is_lead
) /* if string ends with a lead, remove it */
537 if (!(*dst
++ = *src
++)) break;
540 while (n
--) *dst
++ = 0;
544 /*********************************************************************
547 int CDECL
_mbscmp(const unsigned char* str
, const unsigned char* cmp
)
549 if(MSVCRT___mb_cur_max
> 1)
551 unsigned int strc
, cmpc
;
554 return *cmp
? -1 : 0;
557 strc
= _mbsnextc(str
);
558 cmpc
= _mbsnextc(cmp
);
560 return strc
< cmpc
? -1 : 1;
561 str
+=(strc
> 255) ? 2 : 1;
562 cmp
+=(strc
> 255) ? 2 : 1; /* equal, use same increment */
565 return u_strcmp(str
, cmp
); /* ASCII CP */
568 /*********************************************************************
569 * _mbsicoll(MSVCRT.@)
570 * FIXME: handle locales.
572 int CDECL
_mbsicoll(const unsigned char* str
, const unsigned char* cmp
)
574 if(MSVCRT___mb_cur_max
> 1)
576 unsigned int strc
, cmpc
;
579 return *cmp
? -1 : 0;
582 strc
= _mbctolower(_mbsnextc(str
));
583 cmpc
= _mbctolower(_mbsnextc(cmp
));
585 return strc
< cmpc
? -1 : 1;
586 str
+=(strc
> 255) ? 2 : 1;
587 cmp
+=(strc
> 255) ? 2 : 1; /* equal, use same increment */
590 return u_strcasecmp(str
, cmp
); /* ASCII CP */
593 /*********************************************************************
595 * Performs a case-sensitive comparison according to the current code page
597 * _NLSCMPERROR if error
598 * FIXME: handle locales.
600 int CDECL
_mbscoll(const unsigned char* str
, const unsigned char* cmp
)
602 if(MSVCRT___mb_cur_max
> 1)
604 unsigned int strc
, cmpc
;
607 return *cmp
? -1 : 0;
610 strc
= _mbsnextc(str
);
611 cmpc
= _mbsnextc(cmp
);
613 return strc
< cmpc
? -1 : 1;
614 str
+=(strc
> 255) ? 2 : 1;
615 cmp
+=(strc
> 255) ? 2 : 1; /* equal, use same increment */
618 return u_strcmp(str
, cmp
); /* ASCII CP */
622 /*********************************************************************
625 int CDECL
_mbsicmp(const unsigned char* str
, const unsigned char* cmp
)
627 if(MSVCRT___mb_cur_max
> 1)
629 unsigned int strc
, cmpc
;
632 return *cmp
? -1 : 0;
635 strc
= _mbctolower(_mbsnextc(str
));
636 cmpc
= _mbctolower(_mbsnextc(cmp
));
638 return strc
< cmpc
? -1 : 1;
639 str
+=(strc
> 255) ? 2 : 1;
640 cmp
+=(strc
> 255) ? 2 : 1; /* equal, use same increment */
643 return u_strcasecmp(str
, cmp
); /* ASCII CP */
646 /*********************************************************************
649 int CDECL
_mbsncmp(const unsigned char* str
, const unsigned char* cmp
, MSVCRT_size_t len
)
654 if(MSVCRT___mb_cur_max
> 1)
656 unsigned int strc
, cmpc
;
661 return *cmp
? -1 : 0;
664 strc
= _mbsnextc(str
);
665 cmpc
= _mbsnextc(cmp
);
667 return strc
< cmpc
? -1 : 1;
668 inc
=(strc
> 255) ? 2 : 1; /* Equal, use same increment */
672 return 0; /* Matched len chars */
674 return u_strncmp(str
, cmp
, len
); /* ASCII CP */
677 /*********************************************************************
678 * _mbsnbcmp(MSVCRT.@)
680 int CDECL
_mbsnbcmp(const unsigned char* str
, const unsigned char* cmp
, MSVCRT_size_t len
)
684 if(MSVCRT___mb_cur_max
> 1)
686 unsigned int strc
, cmpc
;
691 return *cmp
? -1 : 0;
694 if (MSVCRT_isleadbyte(*str
))
696 strc
=(len
>=2)?_mbsnextc(str
):0;
704 if (MSVCRT_isleadbyte(*cmp
))
705 cmpc
=(len
>=2)?_mbsnextc(cmp
):0;
709 return strc
< cmpc
? -1 : 1;
714 return 0; /* Matched len chars */
716 return u_strncmp(str
,cmp
,len
);
719 /*********************************************************************
720 * _mbsnicmp(MSVCRT.@)
722 * Compare two multibyte strings case insensitively to 'len' characters.
724 int CDECL
_mbsnicmp(const unsigned char* str
, const unsigned char* cmp
, MSVCRT_size_t len
)
726 /* FIXME: No tolower() for mb strings yet */
727 if(MSVCRT___mb_cur_max
> 1)
729 unsigned int strc
, cmpc
;
733 return *cmp
? -1 : 0;
736 strc
= _mbctolower(_mbsnextc(str
));
737 cmpc
= _mbctolower(_mbsnextc(cmp
));
739 return strc
< cmpc
? -1 : 1;
740 str
+=(strc
> 255) ? 2 : 1;
741 cmp
+=(strc
> 255) ? 2 : 1; /* Equal, use same increment */
743 return 0; /* Matched len chars */
745 return u_strncasecmp(str
, cmp
, len
); /* ASCII CP */
748 /*********************************************************************
749 * _mbsnbicmp(MSVCRT.@)
751 int CDECL
_mbsnbicmp(const unsigned char* str
, const unsigned char* cmp
, MSVCRT_size_t len
)
755 if(MSVCRT___mb_cur_max
> 1)
757 unsigned int strc
, cmpc
;
762 return *cmp
? -1 : 0;
765 if (MSVCRT_isleadbyte(*str
))
767 strc
=(len
>=2)?_mbsnextc(str
):0;
775 if (MSVCRT_isleadbyte(*cmp
))
776 cmpc
=(len
>=2)?_mbsnextc(cmp
):0;
779 strc
= _mbctolower(strc
);
780 cmpc
= _mbctolower(cmpc
);
782 return strc
< cmpc
? -1 : 1;
787 return 0; /* Matched len bytes */
789 return u_strncasecmp(str
,cmp
,len
);
792 /*********************************************************************
795 unsigned char * CDECL
_mbscat( unsigned char *dst
, const unsigned char *src
)
797 strcat( (char *)dst
, (const char *)src
);
801 /*********************************************************************
804 unsigned char* CDECL
_mbscpy( unsigned char *dst
, const unsigned char *src
)
806 strcpy( (char *)dst
, (const char *)src
);
810 /*********************************************************************
813 unsigned char * CDECL
_mbsstr(const unsigned char *haystack
, const unsigned char *needle
)
815 return (unsigned char *)strstr( (const char *)haystack
, (const char *)needle
);
818 /*********************************************************************
821 * Find a multibyte character in a multibyte string.
823 unsigned char* CDECL
_mbschr(const unsigned char* s
, unsigned int x
)
825 if(MSVCRT___mb_cur_max
> 1)
832 return (unsigned char*)s
;
835 s
+= c
> 255 ? 2 : 1;
838 return u_strchr(s
, x
); /* ASCII CP */
841 /*********************************************************************
844 unsigned char* CDECL
_mbsrchr(const unsigned char* s
, unsigned int x
)
846 if(MSVCRT___mb_cur_max
> 1)
849 unsigned char* match
=NULL
;
855 match
=(unsigned char*)s
;
858 s
+=(c
> 255) ? 2 : 1;
861 return u_strrchr(s
, x
);
864 /*********************************************************************
867 * Find and extract tokens from strings
869 unsigned char* CDECL
_mbstok(unsigned char *str
, const unsigned char *delim
)
871 thread_data_t
*data
= msvcrt_get_thread_data();
874 if(MSVCRT___mb_cur_max
> 1)
879 if (!(str
= data
->mbstok_next
)) return NULL
;
881 while ((c
= _mbsnextc(str
)) && _mbschr(delim
, c
)) {
882 str
+= c
> 255 ? 2 : 1;
884 if (!*str
) return NULL
;
886 while ((c
= _mbsnextc(str
)) && !_mbschr(delim
, c
)) {
887 str
+= c
> 255 ? 2 : 1;
891 if (c
> 255) *str
++ = 0;
893 data
->mbstok_next
= str
;
896 return u_strtok(str
, delim
); /* ASCII CP */
899 /*********************************************************************
902 int CDECL
MSVCRT_mbtowc(MSVCRT_wchar_t
*dst
, const char* str
, MSVCRT_size_t n
)
904 /* temp var needed because MultiByteToWideChar wants non NULL destination */
905 MSVCRT_wchar_t tmpdst
= '\0';
909 if(!MultiByteToWideChar(CP_ACP
, 0, str
, n
, &tmpdst
, 1))
913 /* return the number of bytes from src that have been used */
916 if(n
>= 2 && MSVCRT_isleadbyte(*str
) && str
[1])
921 /*********************************************************************
922 * _mbbtombc(MSVCRT.@)
924 unsigned int CDECL
_mbbtombc(unsigned int c
)
926 if(MSVCRT___mb_cur_max
> 1 &&
927 ((c
>= 0x20 && c
<=0x7e) ||(c
>= 0xa1 && c
<= 0xdf)))
929 /* FIXME: I can't get this function to return anything
930 * different from what I pass it...
933 return c
; /* ASCII CP or no MB char */
936 /*********************************************************************
939 int CDECL
_mbbtype(unsigned char c
, int type
)
943 if ((c
>= 0x20 && c
<= 0x7e) || (c
>= 0xa1 && c
<= 0xdf))
945 else if ((c
>= 0x40 && c
<= 0x7e) || (c
>= 0x80 && c
<= 0xfc))
952 if ((c
>= 0x20 && c
<= 0x7e) || (c
>= 0xa1 && c
<= 0xdf))
954 else if ((c
>= 0x81 && c
<= 0x9f) || (c
>= 0xe0 && c
<= 0xfc))
961 /*********************************************************************
962 * _ismbbkana(MSVCRT.@)
964 int CDECL
_ismbbkana(unsigned int c
)
966 /* FIXME: use lc_ctype when supported, not lc_all */
967 if(MSVCRT___lc_codepage
== 932)
969 /* Japanese/Katakana, CP 932 */
970 return (c
>= 0xa1 && c
<= 0xdf);
975 /*********************************************************************
976 * _ismbcdigit(MSVCRT.@)
978 int CDECL
_ismbcdigit(unsigned int ch
)
980 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
981 return (get_char_typeW( wch
) & C1_DIGIT
);
984 /*********************************************************************
985 * _ismbcgraph(MSVCRT.@)
987 int CDECL
_ismbcgraph(unsigned int ch
)
989 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
990 return (get_char_typeW( wch
) & (C1_UPPER
| C1_LOWER
| C1_DIGIT
| C1_PUNCT
| C1_ALPHA
));
993 /*********************************************************************
994 * _ismbcalpha (MSVCRT.@)
996 int CDECL
_ismbcalpha(unsigned int ch
)
998 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
999 return (get_char_typeW( wch
) & C1_ALPHA
);
1002 /*********************************************************************
1003 * _ismbclower (MSVCRT.@)
1005 int CDECL
_ismbclower(unsigned int ch
)
1007 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
1008 return (get_char_typeW( wch
) & C1_UPPER
);
1011 /*********************************************************************
1012 * _ismbcupper (MSVCRT.@)
1014 int CDECL
_ismbcupper(unsigned int ch
)
1016 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
1017 return (get_char_typeW( wch
) & C1_LOWER
);
1020 /*********************************************************************
1021 * _ismbcsymbol(MSVCRT.@)
1023 int CDECL
_ismbcsymbol(unsigned int ch
)
1025 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
1027 if (!GetStringTypeW(CT_CTYPE3
, &wch
, 1, &ctype
))
1029 WARN("GetStringTypeW failed on %x\n", ch
);
1032 return ((ctype
& C3_SYMBOL
) != 0);
1035 /*********************************************************************
1036 * _ismbcalnum (MSVCRT.@)
1038 int CDECL
_ismbcalnum(unsigned int ch
)
1040 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
1041 return (get_char_typeW( wch
) & (C1_ALPHA
| C1_DIGIT
));
1044 /*********************************************************************
1045 * _ismbcspace (MSVCRT.@)
1047 int CDECL
_ismbcspace(unsigned int ch
)
1049 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
1050 return (get_char_typeW( wch
) & C1_SPACE
);
1053 /*********************************************************************
1054 * _ismbcprint (MSVCRT.@)
1056 int CDECL
_ismbcprint(unsigned int ch
)
1058 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
1059 return (get_char_typeW( wch
) & (C1_UPPER
| C1_LOWER
| C1_DIGIT
| C1_PUNCT
| C1_ALPHA
| C1_SPACE
));
1062 /*********************************************************************
1063 * _ismbcpunct(MSVCRT.@)
1065 int CDECL
_ismbcpunct(unsigned int ch
)
1067 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
1068 return (get_char_typeW( wch
) & C1_PUNCT
);
1071 /*********************************************************************
1072 * _ismbchira(MSVCRT.@)
1074 int CDECL
_ismbchira(unsigned int c
)
1076 /* FIXME: use lc_ctype when supported, not lc_all */
1077 if(MSVCRT___lc_codepage
== 932)
1079 /* Japanese/Hiragana, CP 932 */
1080 return (c
>= 0x829f && c
<= 0x82f1);
1085 /*********************************************************************
1086 * _ismbckata(MSVCRT.@)
1088 int CDECL
_ismbckata(unsigned int c
)
1090 /* FIXME: use lc_ctype when supported, not lc_all */
1091 if(MSVCRT___lc_codepage
== 932)
1094 return _ismbbkana(c
);
1095 /* Japanese/Katakana, CP 932 */
1096 return (c
>= 0x8340 && c
<= 0x8396 && c
!= 0x837f);
1101 /*********************************************************************
1102 * _ismbblead(MSVCRT.@)
1104 int CDECL
_ismbblead(unsigned int c
)
1106 return (MSVCRT_mbctype
[(c
&0xff) + 1] & _M1
) != 0;
1110 /*********************************************************************
1111 * _ismbbtrail(MSVCRT.@)
1113 int CDECL
_ismbbtrail(unsigned int c
)
1115 return (MSVCRT_mbctype
[(c
&0xff) + 1] & _M2
) != 0;
1118 /*********************************************************************
1119 * _ismbslead(MSVCRT.@)
1121 int CDECL
_ismbslead(const unsigned char* start
, const unsigned char* str
)
1125 if(!g_mbcp_is_multibyte
)
1128 /* Lead bytes can also be trail bytes so we need to analyse the string
1130 while (start
<= str
)
1134 lead
= !lead
&& _ismbblead(*start
);
1138 return lead
? -1 : 0;
1141 /*********************************************************************
1142 * _ismbstrail(MSVCRT.@)
1144 int CDECL
_ismbstrail(const unsigned char* start
, const unsigned char* str
)
1146 /* Note: this function doesn't check _ismbbtrail */
1147 if ((str
> start
) && _ismbslead(start
, str
-1))
1153 /*********************************************************************
1154 * _mbsbtype (MSVCRT.@)
1156 int CDECL
_mbsbtype(const unsigned char *str
, MSVCRT_size_t count
)
1159 const unsigned char *end
= str
+ count
;
1160 int mbcp
= g_mbcp_is_multibyte
;
1162 /* Lead bytes can also be trail bytes so we need to analyse the string.
1163 * Also we must return _MBC_ILLEGAL for chars past the end of the string
1165 while (str
< end
) /* Note: we skip the last byte - will check after the loop */
1168 return _MBC_ILLEGAL
;
1169 lead
= mbcp
&& !lead
&& _ismbblead(*str
);
1174 if (_ismbbtrail(*str
))
1177 return _MBC_ILLEGAL
;
1179 if (_ismbblead(*str
))
1185 /*********************************************************************
1188 unsigned char* CDECL
_mbsset(unsigned char* str
, unsigned int c
)
1190 unsigned char* ret
= str
;
1192 if(MSVCRT___mb_cur_max
== 1 || c
< 256)
1193 return u__strset(str
, c
); /* ASCII CP or SB char */
1195 c
&= 0xffff; /* Strip high bits */
1197 while(str
[0] && str
[1])
1203 str
[0] = '\0'; /* FIXME: OK to shorten? */
1208 /*********************************************************************
1209 * _mbsnbset(MSVCRT.@)
1211 unsigned char* CDECL
_mbsnbset(unsigned char *str
, unsigned int c
, MSVCRT_size_t len
)
1213 unsigned char *ret
= str
;
1218 if(MSVCRT___mb_cur_max
== 1 || c
< 256)
1219 return u__strnset(str
, c
, len
); /* ASCII CP or SB char */
1221 c
&= 0xffff; /* Strip high bits */
1223 while(str
[0] && str
[1] && (len
> 1))
1231 /* as per msdn pad with a blank character */
1238 /*********************************************************************
1239 * _mbsnset(MSVCRT.@)
1241 unsigned char* CDECL
_mbsnset(unsigned char* str
, unsigned int c
, MSVCRT_size_t len
)
1243 unsigned char *ret
= str
;
1248 if(MSVCRT___mb_cur_max
== 1 || c
< 256)
1249 return u__strnset(str
, c
, len
); /* ASCII CP or SB char */
1251 c
&= 0xffff; /* Strip high bits */
1253 while(str
[0] && str
[1] && len
--)
1259 str
[0] = '\0'; /* FIXME: OK to shorten? */
1264 /*********************************************************************
1265 * _mbsnccnt(MSVCRT.@)
1266 * 'c' is for 'character'.
1268 MSVCRT_size_t CDECL
_mbsnccnt(const unsigned char* str
, MSVCRT_size_t len
)
1271 if(MSVCRT___mb_cur_max
> 1)
1274 while(*str
&& len
-- > 0)
1276 if(MSVCRT_isleadbyte(*str
))
1289 return min(ret
, len
); /* ASCII CP */
1292 /*********************************************************************
1293 * _mbsnbcnt(MSVCRT.@)
1294 * 'b' is for byte count.
1296 MSVCRT_size_t CDECL
_mbsnbcnt(const unsigned char* str
, MSVCRT_size_t len
)
1299 if(MSVCRT___mb_cur_max
> 1)
1301 const unsigned char* xstr
= str
;
1302 while(*xstr
&& len
-- > 0)
1304 if (MSVCRT_isleadbyte(*xstr
++))
1310 return min(ret
, len
); /* ASCII CP */
1313 /*********************************************************************
1314 * _mbsnbcat(MSVCRT.@)
1316 unsigned char* CDECL
_mbsnbcat(unsigned char* dst
, const unsigned char* src
, MSVCRT_size_t len
)
1318 if(MSVCRT___mb_cur_max
> 1)
1320 unsigned char *res
= dst
;
1322 if (MSVCRT_isleadbyte(*dst
++)) {
1326 /* as per msdn overwrite the lead byte in front of '\0' */
1332 while (*src
&& len
--) *dst
++ = *src
++;
1336 return u_strncat(dst
, src
, len
); /* ASCII CP */
1340 /*********************************************************************
1341 * _mbsncat(MSVCRT.@)
1343 unsigned char* CDECL
_mbsncat(unsigned char* dst
, const unsigned char* src
, MSVCRT_size_t len
)
1345 if(MSVCRT___mb_cur_max
> 1)
1347 unsigned char *res
= dst
;
1350 if (MSVCRT_isleadbyte(*dst
++))
1353 while (*src
&& len
--)
1356 if(MSVCRT_isleadbyte(*src
++))
1362 return u_strncat(dst
, src
, len
); /* ASCII CP */
1366 /*********************************************************************
1369 unsigned char* CDECL
_mbslwr(unsigned char* s
)
1371 unsigned char *ret
= s
;
1374 if (MSVCRT___mb_cur_max
> 1)
1379 c
= _mbctolower(_mbsnextc(s
));
1380 /* Note that I assume that the size of the character is unchanged */
1389 else for ( ; *s
; s
++) *s
= tolower(*s
);
1394 /*********************************************************************
1397 unsigned char* CDECL
_mbsupr(unsigned char* s
)
1399 unsigned char *ret
= s
;
1402 if (MSVCRT___mb_cur_max
> 1)
1407 c
= _mbctoupper(_mbsnextc(s
));
1408 /* Note that I assume that the size of the character is unchanged */
1417 else for ( ; *s
; s
++) *s
= toupper(*s
);
1422 /*********************************************************************
1423 * _mbsspn (MSVCRT.@)
1425 MSVCRT_size_t CDECL
_mbsspn(const unsigned char* string
, const unsigned char* set
)
1427 const unsigned char *p
, *q
;
1429 for (p
= string
; *p
; p
++)
1431 if (MSVCRT_isleadbyte(*p
))
1433 for (q
= set
; *q
; q
++)
1437 if ((*p
== *q
) && (p
[1] == q
[1]))
1441 if (!q
[0] || !q
[1]) break;
1445 for (q
= set
; *q
; q
++)
1454 /*********************************************************************
1455 * _mbsspnp (MSVCRT.@)
1457 unsigned char* CDECL
_mbsspnp(const unsigned char* string
, const unsigned char* set
)
1459 const unsigned char *p
, *q
;
1461 for (p
= string
; *p
; p
++)
1463 if (MSVCRT_isleadbyte(*p
))
1465 for (q
= set
; *q
; q
++)
1469 if ((*p
== *q
) && (p
[1] == q
[1]))
1473 if (!q
[0] || !q
[1]) break;
1477 for (q
= set
; *q
; q
++)
1485 return (unsigned char *)p
;
1488 /*********************************************************************
1489 * _mbscspn(MSVCRT.@)
1491 MSVCRT_size_t CDECL
_mbscspn(const unsigned char* str
, const unsigned char* cmp
)
1493 if (MSVCRT___mb_cur_max
> 1)
1494 FIXME("don't handle double character case\n");
1495 return u_strcspn(str
, cmp
);
1498 /*********************************************************************
1499 * _mbsrev (MSVCRT.@)
1501 unsigned char* CDECL
_mbsrev(unsigned char* str
)
1503 int i
, len
= _mbslen(str
);
1504 unsigned char *p
, *temp
=MSVCRT_malloc(len
*2);
1509 /* unpack multibyte string to temp buffer */
1511 for(i
=0; i
<len
; i
++)
1513 if (MSVCRT_isleadbyte(*p
))
1525 /* repack it in the reverse order */
1527 for(i
=len
-1; i
>=0; i
--)
1529 if(MSVCRT_isleadbyte(temp
[i
*2]))
1545 /*********************************************************************
1546 * _mbspbrk (MSVCRT.@)
1548 unsigned char* CDECL
_mbspbrk(const unsigned char* str
, const unsigned char* accept
)
1550 const unsigned char* p
;
1554 for(p
= accept
; *p
; p
+= (MSVCRT_isleadbyte(*p
)?2:1) )
1557 if( !MSVCRT_isleadbyte(*p
) || ( *(p
+1) == *(str
+1) ) )
1558 return (unsigned char*)str
;
1560 str
+= (MSVCRT_isleadbyte(*str
)?2:1);
1567 * Functions depending on locale codepage
1570 /*********************************************************************
1573 * Unlike most of the multibyte string functions this function uses
1574 * the locale codepage, not the codepage set by _setmbcp
1576 int CDECL
MSVCRT_mblen(const char* str
, MSVCRT_size_t size
)
1578 if (str
&& *str
&& size
)
1580 if(MSVCRT___mb_cur_max
== 1)
1581 return 1; /* ASCII CP */
1583 return !MSVCRT_isleadbyte(*str
) ? 1 : (size
>1 ? 2 : -1);
1588 /*********************************************************************
1589 * _mbstrlen(MSVCRT.@)
1591 * Unlike most of the multibyte string functions this function uses
1592 * the locale codepage, not the codepage set by _setmbcp
1594 MSVCRT_size_t CDECL
_mbstrlen(const char* str
)
1596 if(MSVCRT___mb_cur_max
> 1)
1598 MSVCRT_size_t len
= 0;
1601 /* FIXME: According to the documentation we are supposed to test for
1602 * multi-byte character validity. Whatever that means
1604 str
+= MSVCRT_isleadbyte(*str
) ? 2 : 1;
1609 return strlen(str
); /* ASCII CP */