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"
29 #include "msvcrt/mbctype.h"
31 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt
);
33 unsigned char MSVCRT_mbctype
[257];
34 int MSVCRT___mb_cur_max
= 1;
35 extern int MSVCRT___lc_collate_cp
;
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}},
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 * __p___mb_cur_max(MSVCRT.@)
148 int* CDECL
__p___mb_cur_max(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_set_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
;
212 if (cpextra
->cp
== 0 || cpextra
->cp
== newcp
)
214 if (cpextra
->cp
== 0)
215 FIXME("trail bytes data not available for DBCS codepage %d - assuming all bytes\n", newcp
);
217 bytes
= cpextra
->TrailBytes
;
218 while (bytes
[0] || bytes
[1])
220 for (i
= bytes
[0]; i
<= bytes
[1]; i
++)
221 MSVCRT_mbctype
[i
+ 1] |= _M2
;
230 /* we can't use GetStringTypeA directly because we don't have a locale - only a code page
233 for (i
= 0; i
< 256; i
++)
234 if (!(MSVCRT_mbctype
[i
+ 1] & _M1
))
235 bufA
[charcount
++] = i
;
237 ret
= MultiByteToWideChar(newcp
, 0, bufA
, charcount
, bufW
, charcount
);
238 if (ret
!= charcount
)
239 ERR("MultiByteToWideChar of chars failed for cp %d, ret=%d (exp %d), error=%d\n", newcp
, ret
, charcount
, GetLastError());
241 GetStringTypeW(CT_CTYPE1
, bufW
, charcount
, chartypes
);
243 curr_type
= chartypes
;
244 for (i
= 0; i
< 256; i
++)
245 if (!(MSVCRT_mbctype
[i
+ 1] & _M1
))
247 if ((*curr_type
) & C1_UPPER
)
248 MSVCRT_mbctype
[i
+ 1] |= _SBUP
;
249 if ((*curr_type
) & C1_LOWER
)
250 MSVCRT_mbctype
[i
+ 1] |= _SBLOW
;
254 if (newcp
== 932) /* CP932 only - set _MP and _MS */
256 /* On Windows it's possible to calculate the _MP and _MS from CT_CTYPE1
257 * and CT_CTYPE3. But as of Wine 0.9.43 we return wrong values what makes
258 * it hard. As this is set only for codepage 932 we hardcode it what gives
259 * also faster execution.
261 for (i
= 161; i
<= 165; i
++)
262 MSVCRT_mbctype
[i
+ 1] |= _MP
;
263 for (i
= 166; i
<= 223; i
++)
264 MSVCRT_mbctype
[i
+ 1] |= _MS
;
267 MSVCRT___lc_collate_cp
= MSVCRT___lc_codepage
= newcp
;
268 TRACE("(%d) -> %d\n", cp
, MSVCRT___lc_codepage
);
272 /*********************************************************************
273 * _getmbcp (MSVCRT.@)
275 int CDECL
_getmbcp(void)
277 return MSVCRT___lc_codepage
;
280 /*********************************************************************
281 * _mbsnextc(MSVCRT.@)
283 unsigned int CDECL
_mbsnextc(const unsigned char* str
)
286 return *str
<< 8 | str
[1];
290 /*********************************************************************
291 * _mbctolower(MSVCRT.@)
293 unsigned int CDECL
_mbctolower(unsigned int c
)
295 if (MSVCRT_isleadbyte(c
))
297 FIXME("Handle MBC chars\n");
300 return tolower(c
); /* ASCII CP or SB char */
303 /*********************************************************************
304 * _mbctoupper(MSVCRT.@)
306 unsigned int CDECL
_mbctoupper(unsigned int c
)
308 if (MSVCRT_isleadbyte(c
))
310 FIXME("Handle MBC chars\n");
313 return toupper(c
); /* ASCII CP or SB char */
316 /*********************************************************************
319 unsigned char* CDECL
_mbsdec(const unsigned char* start
, const unsigned char* cur
)
321 if(MSVCRT___mb_cur_max
> 1)
322 return (unsigned char *)(_ismbstrail(start
,cur
-1) ? cur
- 2 : cur
-1);
324 return (unsigned char *)cur
- 1; /* ASCII CP or SB char */
327 /*********************************************************************
330 unsigned char* CDECL
_mbsinc(const unsigned char* str
)
332 if(MSVCRT___mb_cur_max
> 1 && MSVCRT_isleadbyte(*str
))
333 return (unsigned char*)str
+ 2; /* MB char */
335 return (unsigned char*)str
+ 1; /* ASCII CP or SB char */
338 /*********************************************************************
341 unsigned char* CDECL
_mbsninc(const unsigned char* str
, MSVCRT_size_t num
)
345 if(MSVCRT___mb_cur_max
> 1)
349 return (unsigned char*)str
;
351 return (unsigned char*)str
+ num
; /* ASCII CP */
354 /*********************************************************************
357 unsigned int CDECL
_mbclen(const unsigned char* str
)
359 return _ismbblead(*str
) ? 2 : 1;
362 /*********************************************************************
365 MSVCRT_size_t CDECL
_mbslen(const unsigned char* str
)
367 MSVCRT_size_t len
= 0;
370 if (_ismbblead(*str
))
373 if (!*str
) /* count only full chars */
382 /*********************************************************************
385 void CDECL
_mbccpy(unsigned char* dest
, const unsigned char* src
)
388 if(MSVCRT___mb_cur_max
> 1 && MSVCRT_isleadbyte(*src
))
389 *++dest
= *++src
; /* MB char */
392 /*********************************************************************
395 unsigned char* CDECL
_mbsncpy(unsigned char* dst
, const unsigned char* src
, MSVCRT_size_t n
)
397 unsigned char* ret
= dst
;
400 if(MSVCRT___mb_cur_max
> 1)
406 if (MSVCRT_isleadbyte(*src
++))
415 if (!(*dst
++ = *src
++)) break;
418 while (n
--) *dst
++ = 0;
422 /*********************************************************************
423 * _mbsnbcpy(MSVCRT.@)
425 unsigned char* CDECL
_mbsnbcpy(unsigned char* dst
, const unsigned char* src
, MSVCRT_size_t n
)
427 unsigned char* ret
= dst
;
430 if(MSVCRT___mb_cur_max
> 1)
432 while (*src
&& (n
> 1))
436 if (MSVCRT_isleadbyte(*src
++))
442 if (*src
&& n
&& !MSVCRT_isleadbyte(*src
))
444 /* If the last character is a multi-byte character then
445 * we cannot copy it since we have only one byte left
456 if (!(*dst
++ = *src
++)) break;
459 while (n
--) *dst
++ = 0;
463 /*********************************************************************
466 int CDECL
_mbscmp(const unsigned char* str
, const unsigned char* cmp
)
468 if(MSVCRT___mb_cur_max
> 1)
470 unsigned int strc
, cmpc
;
473 return *cmp
? -1 : 0;
476 strc
= _mbsnextc(str
);
477 cmpc
= _mbsnextc(cmp
);
479 return strc
< cmpc
? -1 : 1;
480 str
+=(strc
> 255) ? 2 : 1;
481 cmp
+=(strc
> 255) ? 2 : 1; /* equal, use same increment */
484 return u_strcmp(str
, cmp
); /* ASCII CP */
487 /*********************************************************************
488 * _mbsicoll(MSVCRT.@)
489 * FIXME: handle locales.
491 int CDECL
_mbsicoll(const unsigned char* str
, const unsigned char* cmp
)
493 if(MSVCRT___mb_cur_max
> 1)
495 unsigned int strc
, cmpc
;
498 return *cmp
? -1 : 0;
501 strc
= _mbctolower(_mbsnextc(str
));
502 cmpc
= _mbctolower(_mbsnextc(cmp
));
504 return strc
< cmpc
? -1 : 1;
505 str
+=(strc
> 255) ? 2 : 1;
506 cmp
+=(strc
> 255) ? 2 : 1; /* equal, use same increment */
509 return u_strcasecmp(str
, cmp
); /* ASCII CP */
512 /*********************************************************************
514 * Performs a case-sensitive comparison according to the current code page
516 * _NLSCMPERROR if error
517 * FIXME: handle locales.
519 int CDECL
_mbscoll(const unsigned char* str
, const unsigned char* cmp
)
521 if(MSVCRT___mb_cur_max
> 1)
523 unsigned int strc
, cmpc
;
526 return *cmp
? -1 : 0;
529 strc
= _mbsnextc(str
);
530 cmpc
= _mbsnextc(cmp
);
532 return strc
< cmpc
? -1 : 1;
533 str
+=(strc
> 255) ? 2 : 1;
534 cmp
+=(strc
> 255) ? 2 : 1; /* equal, use same increment */
537 return u_strcmp(str
, cmp
); /* ASCII CP */
541 /*********************************************************************
544 int CDECL
_mbsicmp(const unsigned char* str
, const unsigned char* cmp
)
546 if(MSVCRT___mb_cur_max
> 1)
548 unsigned int strc
, cmpc
;
551 return *cmp
? -1 : 0;
554 strc
= _mbctolower(_mbsnextc(str
));
555 cmpc
= _mbctolower(_mbsnextc(cmp
));
557 return strc
< cmpc
? -1 : 1;
558 str
+=(strc
> 255) ? 2 : 1;
559 cmp
+=(strc
> 255) ? 2 : 1; /* equal, use same increment */
562 return u_strcasecmp(str
, cmp
); /* ASCII CP */
565 /*********************************************************************
568 int CDECL
_mbsncmp(const unsigned char* str
, const unsigned char* cmp
, MSVCRT_size_t len
)
573 if(MSVCRT___mb_cur_max
> 1)
575 unsigned int strc
, cmpc
;
580 return *cmp
? -1 : 0;
583 strc
= _mbsnextc(str
);
584 cmpc
= _mbsnextc(cmp
);
586 return strc
< cmpc
? -1 : 1;
587 inc
=(strc
> 255) ? 2 : 1; /* Equal, use same increment */
591 return 0; /* Matched len chars */
593 return u_strncmp(str
, cmp
, len
); /* ASCII CP */
596 /*********************************************************************
597 * _mbsnbcmp(MSVCRT.@)
599 int CDECL
_mbsnbcmp(const unsigned char* str
, const unsigned char* cmp
, MSVCRT_size_t len
)
603 if(MSVCRT___mb_cur_max
> 1)
605 unsigned int strc
, cmpc
;
610 return *cmp
? -1 : 0;
613 if (MSVCRT_isleadbyte(*str
))
615 strc
=(len
>=2)?_mbsnextc(str
):0;
623 if (MSVCRT_isleadbyte(*cmp
))
624 cmpc
=(len
>=2)?_mbsnextc(cmp
):0;
628 return strc
< cmpc
? -1 : 1;
633 return 0; /* Matched len chars */
635 return u_strncmp(str
,cmp
,len
);
638 /*********************************************************************
639 * _mbsnicmp(MSVCRT.@)
641 * Compare two multibyte strings case insensitively to 'len' characters.
643 int CDECL
_mbsnicmp(const unsigned char* str
, const unsigned char* cmp
, MSVCRT_size_t len
)
645 /* FIXME: No tolower() for mb strings yet */
646 if(MSVCRT___mb_cur_max
> 1)
648 unsigned int strc
, cmpc
;
652 return *cmp
? -1 : 0;
655 strc
= _mbctolower(_mbsnextc(str
));
656 cmpc
= _mbctolower(_mbsnextc(cmp
));
658 return strc
< cmpc
? -1 : 1;
659 str
+=(strc
> 255) ? 2 : 1;
660 cmp
+=(strc
> 255) ? 2 : 1; /* Equal, use same increment */
662 return 0; /* Matched len chars */
664 return u_strncasecmp(str
, cmp
, len
); /* ASCII CP */
667 /*********************************************************************
668 * _mbsnbicmp(MSVCRT.@)
670 int CDECL
_mbsnbicmp(const unsigned char* str
, const unsigned char* cmp
, MSVCRT_size_t len
)
674 if(MSVCRT___mb_cur_max
> 1)
676 unsigned int strc
, cmpc
;
681 return *cmp
? -1 : 0;
684 if (MSVCRT_isleadbyte(*str
))
686 strc
=(len
>=2)?_mbsnextc(str
):0;
694 if (MSVCRT_isleadbyte(*cmp
))
695 cmpc
=(len
>=2)?_mbsnextc(cmp
):0;
698 strc
= _mbctolower(strc
);
699 cmpc
= _mbctolower(cmpc
);
701 return strc
< cmpc
? -1 : 1;
706 return 0; /* Matched len bytes */
708 return u_strncasecmp(str
,cmp
,len
);
711 /*********************************************************************
714 unsigned char * CDECL
_mbscat( unsigned char *dst
, const unsigned char *src
)
716 strcat( (char *)dst
, (const char *)src
);
720 /*********************************************************************
723 unsigned char* CDECL
_mbscpy( unsigned char *dst
, const unsigned char *src
)
725 strcpy( (char *)dst
, (const char *)src
);
729 /*********************************************************************
732 unsigned char * CDECL
_mbsstr(const unsigned char *haystack
, const unsigned char *needle
)
734 return (unsigned char *)strstr( (const char *)haystack
, (const char *)needle
);
737 /*********************************************************************
740 * Find a multibyte character in a multibyte string.
742 unsigned char* CDECL
_mbschr(const unsigned char* s
, unsigned int x
)
744 if(MSVCRT___mb_cur_max
> 1)
751 return (unsigned char*)s
;
754 s
+= c
> 255 ? 2 : 1;
757 return u_strchr(s
, x
); /* ASCII CP */
760 /*********************************************************************
763 unsigned char* CDECL
_mbsrchr(const unsigned char* s
, unsigned int x
)
765 if(MSVCRT___mb_cur_max
> 1)
768 unsigned char* match
=NULL
;
774 match
=(unsigned char*)s
;
777 s
+=(c
> 255) ? 2 : 1;
780 return u_strrchr(s
, x
);
783 /*********************************************************************
786 * Find and extract tokens from strings
788 unsigned char* CDECL
_mbstok(unsigned char *str
, const unsigned char *delim
)
790 thread_data_t
*data
= msvcrt_get_thread_data();
793 if(MSVCRT___mb_cur_max
> 1)
798 if (!(str
= data
->mbstok_next
)) return NULL
;
800 while ((c
= _mbsnextc(str
)) && _mbschr(delim
, c
)) {
801 str
+= c
> 255 ? 2 : 1;
803 if (!*str
) return NULL
;
805 while ((c
= _mbsnextc(str
)) && !_mbschr(delim
, c
)) {
806 str
+= c
> 255 ? 2 : 1;
810 if (c
> 255) *str
++ = 0;
812 data
->mbstok_next
= str
;
815 return u_strtok(str
, delim
); /* ASCII CP */
818 /*********************************************************************
821 int CDECL
MSVCRT_mbtowc(MSVCRT_wchar_t
*dst
, const char* str
, MSVCRT_size_t n
)
823 /* temp var needed because MultiByteToWideChar wants non NULL destination */
824 MSVCRT_wchar_t tmpdst
= '\0';
828 if(!MultiByteToWideChar(CP_ACP
, 0, str
, n
, &tmpdst
, 1))
832 /* return the number of bytes from src that have been used */
835 if(n
>= 2 && MSVCRT_isleadbyte(*str
) && str
[1])
840 /*********************************************************************
841 * _mbbtombc(MSVCRT.@)
843 unsigned int CDECL
_mbbtombc(unsigned int c
)
845 if(MSVCRT___mb_cur_max
> 1 &&
846 ((c
>= 0x20 && c
<=0x7e) ||(c
>= 0xa1 && c
<= 0xdf)))
848 /* FIXME: I can't get this function to return anything
849 * different from what I pass it...
852 return c
; /* ASCII CP or no MB char */
855 /*********************************************************************
858 int CDECL
_mbbtype(unsigned char c
, int type
)
862 if ((c
>= 0x20 && c
<= 0x7e) || (c
>= 0xa1 && c
<= 0xdf))
864 else if ((c
>= 0x40 && c
<= 0x7e) || (c
>= 0x80 && c
<= 0xfc))
871 if ((c
>= 0x20 && c
<= 0x7e) || (c
>= 0xa1 && c
<= 0xdf))
873 else if ((c
>= 0x81 && c
<= 0x9f) || (c
>= 0xe0 && c
<= 0xfc))
880 /*********************************************************************
881 * _ismbbkana(MSVCRT.@)
883 int CDECL
_ismbbkana(unsigned int c
)
885 /* FIXME: use lc_ctype when supported, not lc_all */
886 if(MSVCRT___lc_codepage
== 932)
888 /* Japanese/Katakana, CP 932 */
889 return (c
>= 0xa1 && c
<= 0xdf);
894 /*********************************************************************
895 * _ismbcdigit(MSVCRT.@)
897 int CDECL
_ismbcdigit(unsigned int ch
)
899 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
900 return (get_char_typeW( wch
) & C1_DIGIT
);
903 /*********************************************************************
904 * _ismbcgraph(MSVCRT.@)
906 int CDECL
_ismbcgraph(unsigned int ch
)
908 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
909 return (get_char_typeW( wch
) & (C1_UPPER
| C1_LOWER
| C1_DIGIT
| C1_PUNCT
| C1_ALPHA
));
912 /*********************************************************************
913 * _ismbcalpha (MSVCRT.@)
915 int CDECL
_ismbcalpha(unsigned int ch
)
917 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
918 return (get_char_typeW( wch
) & C1_ALPHA
);
921 /*********************************************************************
922 * _ismbclower (MSVCRT.@)
924 int CDECL
_ismbclower(unsigned int ch
)
926 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
927 return (get_char_typeW( wch
) & C1_UPPER
);
930 /*********************************************************************
931 * _ismbcupper (MSVCRT.@)
933 int CDECL
_ismbcupper(unsigned int ch
)
935 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
936 return (get_char_typeW( wch
) & C1_LOWER
);
939 /*********************************************************************
940 * _ismbcsymbol(MSVCRT.@)
942 int CDECL
_ismbcsymbol(unsigned int ch
)
944 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
946 if (!GetStringTypeW(CT_CTYPE3
, &wch
, 1, &ctype
))
948 WARN("GetStringTypeW failed on %x\n", ch
);
951 return ((ctype
& C3_SYMBOL
) != 0);
954 /*********************************************************************
955 * _ismbcalnum (MSVCRT.@)
957 int CDECL
_ismbcalnum(unsigned int ch
)
959 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
960 return (get_char_typeW( wch
) & (C1_ALPHA
| C1_DIGIT
));
963 /*********************************************************************
964 * _ismbcspace (MSVCRT.@)
966 int CDECL
_ismbcspace(unsigned int ch
)
968 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
969 return (get_char_typeW( wch
) & C1_SPACE
);
972 /*********************************************************************
973 * _ismbcprint (MSVCRT.@)
975 int CDECL
_ismbcprint(unsigned int ch
)
977 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
978 return (get_char_typeW( wch
) & (C1_UPPER
| C1_LOWER
| C1_DIGIT
| C1_PUNCT
| C1_ALPHA
| C1_SPACE
));
981 /*********************************************************************
982 * _ismbcpunct(MSVCRT.@)
984 int CDECL
_ismbcpunct(unsigned int ch
)
986 MSVCRT_wchar_t wch
= msvcrt_mbc_to_wc( ch
);
987 return (get_char_typeW( wch
) & C1_PUNCT
);
990 /*********************************************************************
991 * _ismbchira(MSVCRT.@)
993 int CDECL
_ismbchira(unsigned int c
)
995 /* FIXME: use lc_ctype when supported, not lc_all */
996 if(MSVCRT___lc_codepage
== 932)
998 /* Japanese/Hiragana, CP 932 */
999 return (c
>= 0x829f && c
<= 0x82f1);
1004 /*********************************************************************
1005 * _ismbckata(MSVCRT.@)
1007 int CDECL
_ismbckata(unsigned int c
)
1009 /* FIXME: use lc_ctype when supported, not lc_all */
1010 if(MSVCRT___lc_codepage
== 932)
1013 return _ismbbkana(c
);
1014 /* Japanese/Katakana, CP 932 */
1015 return (c
>= 0x8340 && c
<= 0x8396 && c
!= 0x837f);
1020 /*********************************************************************
1021 * _ismbblead(MSVCRT.@)
1023 int CDECL
_ismbblead(unsigned int c
)
1025 return (MSVCRT_mbctype
[(c
&0xff) + 1] & _M1
) != 0;
1029 /*********************************************************************
1030 * _ismbbtrail(MSVCRT.@)
1032 int CDECL
_ismbbtrail(unsigned int c
)
1034 return (MSVCRT_mbctype
[(c
&0xff) + 1] & _M2
) != 0;
1037 /*********************************************************************
1038 * _ismbslead(MSVCRT.@)
1040 int CDECL
_ismbslead(const unsigned char* start
, const unsigned char* str
)
1042 /* Lead bytes can also be trail bytes if caller messed up
1043 * iterating through the string...
1045 if(MSVCRT___mb_cur_max
> 1)
1048 start
+= MSVCRT_isleadbyte(*str
) ? 2 : 1;
1051 return MSVCRT_isleadbyte(*str
);
1053 return 0; /* Must have been a trail, we skipped it */
1056 /*********************************************************************
1057 * _ismbstrail(MSVCRT.@)
1059 int CDECL
_ismbstrail(const unsigned char* start
, const unsigned char* str
)
1061 /* Must not be a lead, and must be preceded by one */
1062 return !_ismbslead(start
, str
) && MSVCRT_isleadbyte(str
[-1]);
1065 /*********************************************************************
1068 unsigned char* CDECL
_mbsset(unsigned char* str
, unsigned int c
)
1070 unsigned char* ret
= str
;
1072 if(MSVCRT___mb_cur_max
== 1 || c
< 256)
1073 return u__strset(str
, c
); /* ASCII CP or SB char */
1075 c
&= 0xffff; /* Strip high bits */
1077 while(str
[0] && str
[1])
1083 str
[0] = '\0'; /* FIXME: OK to shorten? */
1088 /*********************************************************************
1089 * _mbsnbset(MSVCRT.@)
1091 unsigned char* CDECL
_mbsnbset(unsigned char *str
, unsigned int c
, MSVCRT_size_t len
)
1093 unsigned char *ret
= str
;
1098 if(MSVCRT___mb_cur_max
== 1 || c
< 256)
1099 return u__strnset(str
, c
, len
); /* ASCII CP or SB char */
1101 c
&= 0xffff; /* Strip high bits */
1103 while(str
[0] && str
[1] && (len
> 1))
1111 /* as per msdn pad with a blank character */
1118 /*********************************************************************
1119 * _mbsnset(MSVCRT.@)
1121 unsigned char* CDECL
_mbsnset(unsigned char* str
, unsigned int c
, MSVCRT_size_t len
)
1123 unsigned char *ret
= str
;
1128 if(MSVCRT___mb_cur_max
== 1 || c
< 256)
1129 return u__strnset(str
, c
, len
); /* ASCII CP or SB char */
1131 c
&= 0xffff; /* Strip high bits */
1133 while(str
[0] && str
[1] && len
--)
1139 str
[0] = '\0'; /* FIXME: OK to shorten? */
1144 /*********************************************************************
1145 * _mbsnccnt(MSVCRT.@)
1146 * 'c' is for 'character'.
1148 MSVCRT_size_t CDECL
_mbsnccnt(const unsigned char* str
, MSVCRT_size_t len
)
1151 if(MSVCRT___mb_cur_max
> 1)
1154 while(*str
&& len
-- > 0)
1156 if(MSVCRT_isleadbyte(*str
))
1169 return min(ret
, len
); /* ASCII CP */
1172 /*********************************************************************
1173 * _mbsnbcnt(MSVCRT.@)
1174 * 'b' is for byte count.
1176 MSVCRT_size_t CDECL
_mbsnbcnt(const unsigned char* str
, MSVCRT_size_t len
)
1179 if(MSVCRT___mb_cur_max
> 1)
1181 const unsigned char* xstr
= str
;
1182 while(*xstr
&& len
-- > 0)
1184 if (MSVCRT_isleadbyte(*xstr
++))
1190 return min(ret
, len
); /* ASCII CP */
1193 /*********************************************************************
1194 * _mbsnbcat(MSVCRT.@)
1196 unsigned char* CDECL
_mbsnbcat(unsigned char* dst
, const unsigned char* src
, MSVCRT_size_t len
)
1198 if(MSVCRT___mb_cur_max
> 1)
1200 unsigned char *res
= dst
;
1202 if (MSVCRT_isleadbyte(*dst
++)) {
1206 /* as per msdn overwrite the lead byte in front of '\0' */
1212 while (*src
&& len
--) *dst
++ = *src
++;
1216 return u_strncat(dst
, src
, len
); /* ASCII CP */
1220 /*********************************************************************
1221 * _mbsncat(MSVCRT.@)
1223 unsigned char* CDECL
_mbsncat(unsigned char* dst
, const unsigned char* src
, MSVCRT_size_t len
)
1225 if(MSVCRT___mb_cur_max
> 1)
1227 unsigned char *res
= dst
;
1230 if (MSVCRT_isleadbyte(*dst
++))
1233 while (*src
&& len
--)
1236 if(MSVCRT_isleadbyte(*src
++))
1242 return u_strncat(dst
, src
, len
); /* ASCII CP */
1246 /*********************************************************************
1249 unsigned char* CDECL
_mbslwr(unsigned char* s
)
1251 unsigned char *ret
= s
;
1254 if (MSVCRT___mb_cur_max
> 1)
1259 c
= _mbctolower(_mbsnextc(s
));
1260 /* Note that I assume that the size of the character is unchanged */
1269 else for ( ; *s
; s
++) *s
= tolower(*s
);
1274 /*********************************************************************
1277 unsigned char* CDECL
_mbsupr(unsigned char* s
)
1279 unsigned char *ret
= s
;
1282 if (MSVCRT___mb_cur_max
> 1)
1287 c
= _mbctoupper(_mbsnextc(s
));
1288 /* Note that I assume that the size of the character is unchanged */
1297 else for ( ; *s
; s
++) *s
= toupper(*s
);
1302 /*********************************************************************
1303 * _mbsspn (MSVCRT.@)
1305 MSVCRT_size_t CDECL
_mbsspn(const unsigned char* string
, const unsigned char* set
)
1307 const unsigned char *p
, *q
;
1309 for (p
= string
; *p
; p
++)
1311 if (MSVCRT_isleadbyte(*p
))
1313 for (q
= set
; *q
; q
++)
1317 if ((*p
== *q
) && (p
[1] == q
[1]))
1321 if (!q
[0] || !q
[1]) break;
1325 for (q
= set
; *q
; q
++)
1334 /*********************************************************************
1335 * _mbsspnp (MSVCRT.@)
1337 const unsigned char* CDECL
_mbsspnp(const unsigned char* string
, const unsigned char* set
)
1339 const unsigned char *p
, *q
;
1341 for (p
= string
; *p
; p
++)
1343 if (MSVCRT_isleadbyte(*p
))
1345 for (q
= set
; *q
; q
++)
1349 if ((*p
== *q
) && (p
[1] == q
[1]))
1353 if (!q
[0] || !q
[1]) break;
1357 for (q
= set
; *q
; q
++)
1368 /*********************************************************************
1369 * _mbscspn(MSVCRT.@)
1371 MSVCRT_size_t CDECL
_mbscspn(const unsigned char* str
, const unsigned char* cmp
)
1373 if (MSVCRT___mb_cur_max
> 1)
1374 FIXME("don't handle double character case\n");
1375 return u_strcspn(str
, cmp
);
1378 /*********************************************************************
1379 * _mbsrev (MSVCRT.@)
1381 unsigned char* CDECL
_mbsrev(unsigned char* str
)
1383 int i
, len
= _mbslen(str
);
1384 unsigned char *p
, *temp
=MSVCRT_malloc(len
*2);
1389 /* unpack multibyte string to temp buffer */
1391 for(i
=0; i
<len
; i
++)
1393 if (MSVCRT_isleadbyte(*p
))
1405 /* repack it in the reverse order */
1407 for(i
=len
-1; i
>=0; i
--)
1409 if(MSVCRT_isleadbyte(temp
[i
*2]))
1425 /*********************************************************************
1426 * _mbspbrk (MSVCRT.@)
1428 unsigned char* CDECL
_mbspbrk(const unsigned char* str
, const unsigned char* accept
)
1430 const unsigned char* p
;
1434 for(p
= accept
; *p
; p
+= (MSVCRT_isleadbyte(*p
)?2:1) )
1437 if( !MSVCRT_isleadbyte(*p
) || ( *(p
+1) == *(str
+1) ) )
1438 return (unsigned char*)str
;
1440 str
+= (MSVCRT_isleadbyte(*str
)?2:1);
1447 * Functions depending on locale codepage
1450 /*********************************************************************
1453 * Unlike most of the multibyte string functions this function uses
1454 * the locale codepage, not the codepage set by _setmbcp
1456 int CDECL
MSVCRT_mblen(const char* str
, MSVCRT_size_t size
)
1458 if (str
&& *str
&& size
)
1460 if(MSVCRT___mb_cur_max
== 1)
1461 return 1; /* ASCII CP */
1463 return !MSVCRT_isleadbyte(*str
) ? 1 : (size
>1 ? 2 : -1);
1468 /*********************************************************************
1469 * _mbstrlen(MSVCRT.@)
1471 * Unlike most of the multibyte string functions this function uses
1472 * the locale codepage, not the codepage set by _setmbcp
1474 MSVCRT_size_t CDECL
_mbstrlen(const char* str
)
1476 if(MSVCRT___mb_cur_max
> 1)
1478 MSVCRT_size_t len
= 0;
1481 /* FIXME: According to the documentation we are supposed to test for
1482 * multi-byte character validity. Whatever that means
1484 str
+= MSVCRT_isleadbyte(*str
) ? 2 : 1;
1489 return strlen(str
); /* ASCII CP */